Sponsors and Participating organizations

GLOBAL OBSERVATION OF FOREST AND LAND

COVER DYNAMICS

MONGOLIAN GEOSCIENCE AND REMOTE SENSING SOCIETY

NATIONAL UNIVERSITY OF MONGOLIA

MONGOLIAN UNIVERSITY OF SCIENCE AND TECHNOLOGY

SCHOOL OF MATERIALS SCIENCE

INTERNATIONAL INSTITUTE FOR GEO-INFORMATION SCIENCE

AND EARTH OBSERVATION (ITC), NETHERLAND

CEReS, CHIBA UNIVERSITY, JAPAN

RESEARCH INSTITUTE OF EMERGENCY SERVICES

SUN ERDENE TOUR, MONGOLIA

Proceedings of The 2nd International Conference On Land Cover /Land Use
Study Using Remote Sensing And Geographic Information System And The
GOFC-GOLD Regional Capacity Building Meeting In Mongolia

June 08-09, 2006

Organized by Mongolian Remote Sensing Society and GOFC-GOLD

(Editors)
N. Tugjsuren and R. Tsolmon
 Welcome Message to the participants of
The 2nd International Conference on Land cover / Land use study using Remote Sensing
and Geographic Information System and the GOFC-GOLD regional capacity building
meeting in Mongolia

Distinguished International Scientists, Conference participants, ladies and gentlemen,

On behalf of the Government of Mongolia, the Members of the Mongolian Parliament, I would
like to welcome all of you to this event, and express our gratitude to you for coming to Mongolia
to support and contribute to this event.

Two years ago, we held the very successful First National Conference on Land Cover Study
Using Remote sensing and GIS, sponsored by the Mongolian Geosciences and Remote Sensing
Society and the ITC from the Netherlands and Center Environmental Remote Sensing Center
Chiba University in Japan. Since then, there have been many remote sensing initiatives in
Mongolia. One of the major events has been the establishment of the NUM-ITC-UNESCO
Laboratory for Remote Sensing and GIS, at the National University of Mongolia, funded by
UNESCO and the ITC.

As a result, this laboratory became a centre for geoscientists, researchers, police-makers, students,
and volunteers, all of whom have found a need for the information provided by remote sensing
and geographical information systems to advance the technical requirements of their given field.
The types of scientific studies concerning the remote sensing and applied branches of scientific,
social and economical sectors of Mongolia have increased within this joint laboratory, with
excellent hardware and modern Remote Sensing/GIS Software.

I know that the GOFC-GOLD, supported by FAO, UNESCO, UNEP, ICSU and WMO is a most
powerful, multifaceted international strategy to bring the Earth’s land cover under continuous
observations. It has a vision to share information, data and knowledge. It consists of a
coordinated program of activities to ensure that earth observation and other data are used
effectively for global monitoring of terrestrial resources and the study of global change. It
operates through a network of participants implementing coordinated demonstration and
operational projects.

I would like to underline that the objectives of conference and discussion topics are indeed, of
immense importance and actual problems for Mongolia and a main goal of GOFC-GOLD is
contemporary forest dynamics. For instance, I would like to share with you my views about
Mongolia’s forest capacity. The forest area of Mongolia is only 10 million hectors, in other
words, 6.5% of total territory of country. This little land cover is faced with a large amount of
damage. In 1999-2000 deforestation rate decreased by 0.72% each year, but between 2000-2005,
this rate increased intensively by 0.77% each year. For example, in 2005, the consequence of 180
forest fires destroyed more than 450 thousands hectors of forest.

Another reason for deforestation is the insect impact. On the one hand, results of your detailed
studies and successive cooperation are very important for detection and determination of how
many territories are impacted by insect. On the other hand, the remote sensing method can make
an important contribution to study the impact of insects on forests, monitor the spread of
infestations, etc.

In the last few years, the frequencies and amplitudes of drought and dzud calamities has
intensified. Between 1999-2000, 10.4% (3.4 million of heads) of total livestock sector, and in
2000-2001 another 15.2% (4.8 million) of livestock died from the dzud, which has decimated for
the Mongolian livestock farming system. The desertification and dust storm issues have also
intensified in total area of Mongolia, particularly in Gobi Desert. Thousands of lakes, rivers,
springs and wells have dried. I can list so many difficulties concerning ‘Mongolian’ calamities
which have occurred in recent times in our country.

As a result of this destructive environmental change, the next chain of events have effected
Mongolia’s demography: the herders have become poor, they have migrated to urban areas, so,
urban life is faced with many new difficulties and poverty persists. Poor people influence also
further contribute to deforestation in consequence with the large wood uses for fuel. In order to
overcome all these difficulties, Mongolia must set forth upon multilateral and regional
cooperation with other countries to aid with Mongolia’s many environment and development
issues.

I would like to reassure you that the results of this conference, and your future cooperation in
Mongolia will be supported by Mongolian Parliament, Government of Mongolia and myself.

I wish you all very fruitful discussions and a successful conference, plus, I hope you will all enjoy
some of our Mongolian culture during your stay in our country.

Thank You.

Mongolian Parliament Member, B. Munkhtuya

Contents

Challenges in the mapping and monitoring of forests using remote sensing

John R. Townshend ------------------------------------------------------------------------------------------------- 1

Ground survey of the Mongolian rangelands on plant onset trend

T. Chuluun, J. Sanjid, I. Tuvshintogtoh, B. Oyun, B. Tserenchunat and Dennis Ojima ------------------- 5

Trends in Northern Eurasia’s land cover derived on SPOT-VGT time series analysis 1998-2005
M. Herold, C. Hɶttich, C. Schmullius, and S. A. Bartalev ----------------------------------------------------- 6

Application of Multitemporal Optical and SAR Data for Different Forest Studies
Damdinsuren Amarsaikhan ---------------------------------------------------------------------------------------- 7

Brief description on land cover and current activities in Korea

Nam-Sun, Oh, Hong-Yeon Cho, and Geun-Sang Lee --------------------------------------------------------- 12

Glacier change estimation using Landsat TM data

M. Erdenetuya, P.Khishigsuren, M. Otgontugs -----------------------------------------------------------------18

Forage Monitoring Technology to Improve Risk Management Decision Making by Herders in the Gobi
Region of Mongolia
J. P. Angerer, L.Bolor-Erdene, D. Tsogoo, M.Urgamal and S. Granville-Ross ---------------------------- 22

Monitoring environmental degradation in Mongolia with NPP and rainfall data.

M. Tyburski, R. Tsolmon, D. Sodnomragchaa, R. Harris1 and A. Warren ---------------------------------- 23

Projects and Initiatives addressing Environmental Impact Studies in Northern Mongolia and the Lake
Baikal Region
K. Frotscher, C.C. Schmullius ------------------------------------------------------------------------------------- 31

Digital Asia -Information Network for Sustainable Future

Hiromichi Fukui ----------------------------------------------------------------------------------------------------- 32

Annual variation of aerosol optical thickness derived from PAR observation in Mongolia
T. Takamura, T. Karasuyama, N. Tugjsuren, G. Batsukh, and H. Takenaka ------------------------------- 33

Determination Of The Photosynthetically Active Radiation For Vegetation Growth Period Of The
Mongolian Grain Farm Region
Tugjsuren Nasurt ---------------------------------------------------------------------------------------------------- 37

Coherent time in cloud analysis using 95GHz FM-CW cloud profiling radar
Y. Nakanishi, t. Takano, k. Akita, h. Kubo, y. Kawamura, h. Kumagai,
T. Takamura and t. Nakajima ---------------------------------------------------------------------------- --------- 41

Land cover mapping of mongolia

Sh. Munkhtuya ------------------------------------------------------------------------------------------------------- 45

Development of a Better Atmosphere and Soil Resistant Vegetation Index for Forestry Monitoring in
Taiwan
G. Dashnyam, G. R. Liu, C. K. Liang, T. H. Kuo C. W. Lan, T. H. Lin, Y. C. Chen --------------------- 49

Sharing ground truth data for land cover mapping – GLCNMO

Ryutaro Tateishi ----------------------------------------------------------------------------------------------------- 52

Ecosystem changes mapping for Eastern Shore of Lake Hovsgol from satellite imagery and
GIS a case study
B.Gantsetseg --------------------------------------------------------------------------------------------------------- 53

GIS application on micro relief development

Altangerel, B., Schwanghart, W. & Walther, M. --------------------------------------------------------------- 56

Vegetation mapping of the Great Gobi A strictly protected area

A.Tsolmon & H. von Wehrden ----------------------------------------------------------------------------------- 57

Climate change impact on rangeland productivity in the Eurasian steppe

Dennis S. Ojima and Togtohyn Chuluun ------------------------------------------------------------------------ 61

Classification of Multitemporal InSAR Data for Land Cover Mapping in Selenga River Basin, Mongolia
Damdinsuren Amarsaikhan ---------------------------------------------------------------------------------------- 65

Urban Land Cover Change Studies Using Multitemporal RS Images

D.Amarsaikhan, M.Ganzorig and B.Nergui --------------------------------------------------------------------- 69

Surface Water Pollution Of The Ulaanbaatar City

(determined by BOD, dissolved O2, NH4+, NO2-, NO3-, PO4-3, Cr+6, COD) (between 1996-2004)
Ch.Gonchigsumlaa, O.Altansukh -------------------------------------------------------------------------------- 76

Analysis on the Land Use in General and Neighborhood Commercial Areas of Ulaanbaatar city
using RS and GIS
B.Chinbat, D.Amarsaikhan and Tae-Heon Moon --------------------------------------------------------------- 78

The Heat Island Experiment over the Western Taiwan Plain with MODIS satellite and concurrent
helicopter-borne IR imager data
Chia Wei Lan*, Gin-Rong Liu, Tsung-Hua Kuo, Kun-Wei Lin, Tang-Huang Lin, Ming-Chang Hsu,
Yen-Ju Chen, Chia-Chi Liu ---------------------------------------------------------------------------------------- 85

Global vegetation continuous field tree cover products and Siberian forest types
M. Herold , D. Knorr, K. Kornhaə, A. Shvidenko , O. Cartus and C. Schmullius ------------------------ 89

Possible use of fuzzy-knowledge for improving the geographic boundary representation

Sang-Jun Kim, Ju-Whan Kang, and Soung-Yong Yun -------------------------------------------------------- 90

Diagnoses for the Drought and Dzud Frequencies in Mongolia

T. Ulaanbaatar ------------------------------------------------------------------------------------------------------- 99
Evaluation of atmospheric optical characteristics In the mongolian territory
G. Batsukh, B. Daariimaa, T. Narangarav----------------------------------------------------------------------- 109
Landscape unit map of uvs nuur and adjacent areas
Michael walther------------------------------------------------------------------------------------------------------ 110
Some initial results of using The 6p forest inventory method in Mongolia
Sh. Tsogtbayar------------------------------------------------------------------------------------------------------- 111
Pasture land classification using remote sensing data
D.Narangerel, N.Monkhoo, B.Suvdantsetseg, B.Batzorig, A. Saruulzaya ---------------------------------- 112
Use of Caesium-137 for the agriculture soil degradation in Central part of Mongolia
O.Batkhishig, N.Enkhbat, B.Burmaa-----------------------------------------------------------------------------113

The Mongolian forest characteristic and ecological changes.

G.Tsedendash -------------------------------------------------------------------------------------------------------- 114

A method to estimate soil moisture using l-band synthetic aperture radar data
Javzandulam Tsend-Ayush, Josaphat Tetuko S.S, Ryutaro Tateishi, Tsolmon Renchin ---------------- 116

Moving into international geographic information standards

Bolorchuluun.Ch; Battsengel.V ----------------------------------------------------------------------------------- 120

LUCC and Terrestrial Study based on RS, GIS and Ecological Observation & Proposal for
International Collaboration
Jiyuan Liu Lin Zhen Yunfeng Hu Qian Zhang --------------------------------------------------------------- 127

Remote Sensing Parameterization of Land Surface Heat Fluxes over Arid and Semi-arid
region in Mongolia
Jadamba Batbayar , Nas-Urt Tugjsuren ------------------------------------------------------------------------- 129

A Possibility of Cooperation in Detection of Water and Heat Losses of District Heating System in
Ulaanbaatar
Ulaanbaatar T., Legden M. and Danbayar ----------------------------------------------------------------------- 135

Monitoring of Saxaul forest in Gobi of Mongolia

B. Suvdantsetseg, Y. Aruinzul and D.Narantuya --------------------------------------------------------------- 139

Dust and Sandstorm Monitoring of Mongolia Using NOAA AVHRR data

L.Ochirkhuyag, R.Tsolmon; S.Khudulmur; J.Sumyasuren; L.Natsagdorj; D.Jugder --------------------- 144
 Challenges in the mapping and monitoring of forests using remote sensing.
John R. Townshend
Department of Geography, University of Maryland College Park, Maryland 20742
USA

Abstract
The need for improved observations of the Earth’s changing forests is essential for many
different types of stake-holders. Remote sensing data from satellites can now contribute greatly to
the monitoring of forests. But there are many challenges in ensuring that long term operational
monitoring is achieved. These are largely non-technical and relate to long term funding
commitments and ensuring appropriate distribution of products. Of crucial importance in
increasing the uptake of remote sensing data is eliminating charging for data from government-
funded satellites wherever possible. This would help most users and also would help stimulate
service providers relying on remote sensing.

I. INTRODUCTION
The world’s forests continue to change rapidly primarily as a result of anthropogenic changes both
direct and indirect. The need for reliable information on forests has never been stronger. They play a
major role in climate change through their impact on the carbon cycle both through sequestration and
through release of CO2 especially during forest fires. Forests also often are locations of high biodiversity.
They help reduce floods, supply drinking water and prevent erosion. Forests are themselves key natural
resources in developing and developed countries. Consequently there are many stake-holders for reliable
observations, including the global change science community, those concerned with international
environmental agreements as well as natural resource managers.

II. CONTRIBUTION OF REMOTE SENSING SATELLITES

National capabilities to monitor forest cover vary greatly. Remote sensing data are often
unavailable because of high costs or inadequate satellite acquisition strategies. Comparisons between
countries are often thwarted because of different definitions and protocols. The Forest Resource
Assessment (FRA) of FAO uses a tree canopy cover limit of only 10% but in practice many forest agencies
use a threshold of 35-40% and may include harvestable lands whether these are actual or potential forests.
There are many types of satellites with the potential to contribute to the monitoring of forests.
Moderate spatial resolution sensors (250m-1km) such as NASA’s Moderate Resolution Imaging
Spectrometer (MODIS) typically provide a daily overview of the whole Earth. There are many fine
resolution systems in orbit such as Landsat’s Enhanced Thematic Mapper (ETM+), SPOT’s HRVIR (High
Resolution Visible Infrared), the AVNIR (Advanced Visible and Near Infrared Radiometer) instrument of
the Indian Remote Sensing Satellite Systems, the Japanese ASTER (Advanced Spaceborne Thermal
Emission and Reflection Radiometer) or the CBERS (Chinese Brazilian Earth Resource Satellite) of China
and Brazil. These sensors provide much more detail globally but less frequently than moderate resolution
sensors - usually only once every approximately16 days. Ultrafine resolution images (<5m resolution) are
usually from commercial satellites, though the Japanese Advance Land Observing Satellite (ALOS) now
has this capability; these provide a very detailed look though only for sample areas with relatively limited
coverage.
Satellite data have been increasingly used for monitoring forests though there is a paucity of
internationally agreed protocols and standards for extracting the information in the creation of information
products. Results from satellite based analyses often show substantial disagreements compared with
statistics from international agencies. For example work by Defries et al (2002) combining coarse and fine
resolution remotely sensed data for the tropics indicated that the satellite-derived estimates of forest change
agree with FAO estimates in Latin America and Tropical Asia for the 1990s, but are substantially lower for
the 1980s. The net rate of tropical forest clearing increased approximately 10 percent from the 1980s to
90s, most notably in Southeast Asia, in contrast to an 11 percent reduction reported by FAO. Hence for
this period the rate of tropical deforestation is increasing and not decreasing. In terms of carbon this
suggests that the net mean annual carbon fluxes from tropical deforestation and regrowth are 0.6 (.4-1.0)
and 1.0 (.5-1.4) Gt/yr for the 1980s and 1990s respectively.

-1-
 III. CHALLENGES IN MOVING TOWARDS OPERATIONAL SYSTEMS FOR THE
LAND COVER
A key question for many who wish to utilize products from remotely sensed data is whether they
can be assured that they will continue to be available in the future. If not then it can be risky for
operational agencies to adopt remote sensing. Any survey of remote sensing instruments shows that there
are currently many missions in orbit. But many of these have no long term continuity planned; capacity to
generate products in the ground segment is often limited and poor data policies may militate against wide
use. Overall for land remote sensing cooperation in the use of satellite assets is weak compared with
weather satellites.
For moderate resolution remote sensing there are now quite hopeful indications that long-term
observations are assured through the present MODIS instrument and the successor VIIRS (Visible/Infrared
Imager/Radiometer Suite) instrument which will be on the next generation of meteorological polar orbiters
called NPOESS (National Polar-orbiting Operational Environmental Satellite System). The NPOESS
Preparatory Project (NPP) has a VIIRS instrument and this should fill in any gap between MODIS and the
first VIIRS on NPOESS.
For fine Landsat-class instruments the situation is much less certain at least in the near to medium
term. The ETM + of Landsat 7 has major deficiencies due to the failure of its Scan Line Corrector mirror
and currently there is heavy reliance on the ancient TM of Landsat 5, whose data are only available in sight
of receiving stations. Efforts in many countries are on-going to use other fine resolution assets or increase
receiving station capabilities for Landsat 5 data. Although many other systems similar to Landsat exist
(outlined in section 2), there is no coherent plan to try and duplicate the acquisition strategy of Landsat 7
and consequently the current availability of this class of data is much poorer now than in the early years of
this decade. Unfortunately this situation may persist for another 5 years.
It is to be hoped in the longer term that nations such as the US, EU, China and India will work
together to create an integrated constellation of fine resolution satellites better able to meet the needs of the
many users of this class of data. The latter should include more frequent imaging that that achieved by
Landsat. Landsat 5 and Landsat 7 working together provided an 8 day frequency of imaging.
Coordination between agencies in choice of orbits of future Landsat-class missions could result in
frequencies of once every four or even two days and ultimately in daily imaging. Alternatively increasing
the swath width of future sensors could also significantly increase the frequency of imaging.
Recently considerable advances have been made in the quality of data available from radar for
land cover monitoring, by use of these sensors in an interferometric mode notably in L-band. Regular
global coverage of products suitable for land cover monitoring is not feasible as yet, if only because of
computational challenges, but such data could have a significant complementary role in the cloudier parts
of the world, where optical remote sensing is rarely capable of allowing regular acquisition of high quality
data. Identification of such areas, followed by their systematic imaging by radars could make an important
contribution to operational global monitoring of forests.
In Table 1 there is an attempt to identify the main challenges hindering the regular supply of the
main categories of remote sensing data needed for forest monitoring. Three types of challenge are
identified:
Technical challenges: These are challenges associated with the design and construction of sensors
to meet the needs of users.
Continuity challenges: These challenges are those in obtaining committed resources for
systematic long-term observations into the foreseeable future. Only sensors on meteorological
platforms currently fall into this category. To ensure full operational status there should be
back-up missions available.
Distribution challenges: Even though missions may be in orbit this does not mean that users will
have ready access to data suitable for there needs. For example there may be restrictions due to
under-scoping the ground segment in relation to user demands for data; costs and other
aspects of data policy may greatly hinder use; formats of products may make it difficult for
users readily to integrate their data with other data sets.
What Table 1 implies is that the challenges are now largely not technical. Sensors can be built
satisfying many of our needs (Townshend et al 2004). The only major exception is for canopy lidars,
which are needed to provide information on the vertical structure of canopies but these challenges likely

-2-
will be overcome in the next few years. Although the VIIRS has some deficiencies relative to MODIS and
the data processing system has still to be completed, it appears that all of the continuity and distribution
challenges are being overcome for MODIS-class instruments.
For all the other types of sensors there remain significant continuity and distribution challenges not
withstanding the number of such missions in orbit or being planned. Many systems may appear to be
sufficient to meet user needs but there are often significant difficulties in users obtaining the data or
ensuring that the data are acquired when and where users need them. Furthermore countries rarely will
commit to long term funding to ensure that there are no gaps in the record. It is to be hoped that
burgeoning organizations such as the Group on Earth Observations (GEO) can improve cooperation and
coordination between countries to overcome some of these challenges.

Technical Continuity Distribution

Challenge Challenge Challenges
Landsat- s s
No Y Y
class
Thermal IR No Y Y
(<120m)
MODIS- No No No
class
Radar No Y Y
Canopy Y Y Y
Lidar
Table 1 Characterization of remote sensing capabilities in terms of technical challenges, continuity
challenges and distribution challenges.

IV. IMPORTANCE OF CHARGING POLICIES

Charging for data undoubtedly inhibits applications of remote sensing. The US is unique in not
charging for the vast majority of its remote sensing from government funded satellites However Landsat
data, essential for forest and land cover change, are usually charged for, though once purchased they can be
copied without restriction. In contrast for comparable data from other sources it may be impossible to
purchase them: one simply buys the rights to use the data and in such circumstances normally one is not
allowed to copy the data and distribute it to others.
In the case of government-financed remote sensing systems there is little justification for charging
for data. Tax-payers have already paid for the satellites and sensors and the cost of the data is a marginal
fraction of the total cost of the system. Furthermore there is little evidence after 30 plus years that
terrestrial remote sensing can be commercially viable except possibly for ultra-fine resolution systems and
even for these systems it is only massive government data purchases that makes them commercially viable.
There is however likely to be a growing commercial future for services based on remote sensing
data, and this commercial sector would be stimulated by not charging for data. Hence it is fair to conclude
that charging for data reduces the economic benefits of remote sensing. It is true that scientists are
sometimes charged less for data or even may be charged zero, but there is no logical reason why this class
of user should be charged less than those in developing countries concerned with alleviating poverty or
even those in the commercial sector aiming to bring economic development to a region.

V. CONCLUSIONS
Forests are changing rapidly and their ability to provide services is declining in many areas. There
is therefore a crucial need for reliable monitoring of forest cover change. Remote sensing capabilities to
carry out this task have improved substantially in recent years, but there are many deficiencies in
operational capabilities especially those concerned with maintaining observational continuity and ensuring
products are distributed to users. The likely gap in Landsat coverage is a major blow to our ability to
monitor the Earth and improved coordination is needed in the short and longer terms to remedy current
deficiencies. There is a vital need to maintain open data policies and to eliminate any charging for data
from government-funded systems.

-3-
REFERENCES
1. DeFries, R., Houghton, R.A., Hansen, M., Field, C., Skole, D., Townshend, J.R.G. 2002 Carbon
Emissions from Tropical Land Use Change Based on Satellite Observations for the 1980s and 90s,
Proceeding National Academy of Sciences, 99, 14256-14261.
2. Townshend, J.R.G., Justice, C.O., Skole, D.L., Belward, A., A. Janetos, Gunawan, I., Goldammer,
J., Lee, B. (2004) Meeting The Goals Of GOFC: an evaluation of progress and steps for the future.
In Gutman, G. et al (Eds.) Land Change Science, Kluwer Academic Publishers, Dordrecht,
Netherlands, pp 31-52.

-4-
 Ground survey of the Mongolian rangelands on plant onset trend
T. Chuluun1, J. Sanjid2, I. Tuvshintogtoh2, B. Oyun3, B. Tserenchunat1 and Dennis Ojima4
1
National University of Mongolia, phone 976 9911-1301, Email: chuluun@nrel.colostate.edu
2
Institute of Botany, Mongolian Academy of Sciences
3
Department of Ecology & Environmental Sciences, Inner Mongolia University
235 West University Road, Hohhot,Inner Mongolia, P.R. China 010021
4
Natural Resource Ecology Laboratory, Colorado State University, Fort Colllins, Colorado 80523, USA

Plant onset trends of grassland ecosystems of the Mongolian Steppe have been analyzed
using a long-term RS data identifying the zones with delayed or advanced plant onset trends in the
Mongolian rangelands (Ellis et al., 2002). The region of advanced green-up covers much of the
eastern steppe, extending south from the forest zone to near the northern edge of the Gobi desert.
This is the region with the highest rainfall in the Mongolian steppe, with precipitation ranging from
200 mm per annum in the south, to more than 400 mm near the forest zone. The delayed green-up
zone forms band along the boundary area of the dry steppe and the Gobi desert steppe and covers
the desert steppes located in the southern slopes of high mountains such as Altai, Hangai and Khan-
Khohii. Mean annual rainfall in most of the delayed green-up zone is 100-200mm. The delayed
green-up in dry ecosystems seen here could be linked to lower photosynthetic rates, lower CO2
uptake and reduced primary production rates. These changes would certainly viewed as negative
for the herbivores and human that depend on high latitude grassland environment for their
sustenance and support.

A ground survey was conducted to better understand factors impacting on plant onset trends in the
Mongolian rangeland ecosystems. Generally, land use was a major factor rather than climate
change to impact on vulnerability of rangelands in Inner Mongolia, which was opposite in
Mongolia, being climate change a major factor influencing on vulnerability of rangelands.
However, there were areas where both climate change and land use inter-acted causing serious land
degradation in Mongolia with signs of desertification. More extensive field survey was done for
rangelands in Mongolia during 2001-2003 to better understand how different types of ecosystems
and landscapes respond differently to climate change. The soil moisture availability in the spring
time was a main factor for the green-up trends of the rangelands. The eastern steppe, grasslands on
the northern slopes of the Altai and Khan Khohii Mountains and some oasis or sandy lands, where
the soil moisture was available in early spring, had an advanced green-up trends. Extensive band
along the boundary area of the Gobi and steppe, and desert ecosystems located in the southern
slopes of the Altai and Khan Khohii Mountains had delayed green-up trends making these
rangelands vulnerable in spring time. The grasslands located in the dry end of the steppe region
seemed more vulnerable to spring drying trend than the desert steppe located in the northern edge of
the Gobi due to adaptive capacity of the Gobi plants to drought. Thus, the Mongolian rangeland
ecosystems vulnerable to climate change have been identified. Further field study is necessary to
better understand climate change effects on functional groups of ecosystems and defining adaptive
rangeland management strategies in these vulnerable regions.

Key words: Climate change effects on grassland ecosystems, ground validation and rangeland
vulnerability

Literature
1. Ellis, J., K. Price, R. Boone, Yu. Fangfang, T. Chuluun and Yu. Mei. 2002. Integrated
assessment of climate change effects on vegetation in Mongollia and Inner Mongolia. In:
Togtohyn Chuluun and Dennis Ojima (eds.) Symposium Proceedings “Change and
Sustainability of Pastoral Land Use Systems in Temperate and Central Asia”, pp. 26-34.

-5-
Trends in Northern Eurasia’s land cover derived on SPOT-VGT time series
analysis 1998-2005
M. Herold1, C. Hɶttich1, C. Schmullius1, and S. A. Bartalev2
(1)
Friedrich-Schiller-University Jena, Department of Earth Observation
(2)
Russian Academy of Sciences, Space Research Institute, Boreal Ecosystems Monitoring Laboratory

The Boreal and Tundra ecosystems of the mid-high latitudes provide sensitive
indicators of environmental impacts both of climate change and human activities. A
number of studies have emphasized changes and trends in Eurasia due to drivers of natural
and human induced land cover dynamics, in particular for the period 1982-1999. The
investigations of this study focus on the very recent years of 1998-2005 and cover the
whole boreal ecosystems of Northern Eurasia with its geographical dimensions of 42°N to
75°N and 5°E to 180°E. This northern hemispheric belt includes to its most extent the
national territory of Russia but also partly its abutter nations, the Scandinavian Nations
and the central and south-European nations, as well as, Mongolia. The study focus on
linear trends from 1998-2005 using inter-annual and inter-seasonal trends in the SPOT-
VGT mosaics.

Significant trends could be detected in the NDVI and NDWI time series from 1998 -2005.
The trends differ by season (spring, summer, autumn), land cover type and latitude. The
spring trends show significant positive NDVI regression slopes and strong negative NDWI
slopes over the evergreen needleleaf-, needleleaf/broadleaf-, and mixed forests of the
Russian and Scandinavian boreal zone, which indicates an onset of the vegetation green-
up dates (NDVI trends) over eight years linked with earlier snowmelt (NDWI trends).
Most affected are the forests in high altitudes like the Ural region, the central Siberian
forests and the forests of middle Sweden. Similar vegetation dynamics can be exposed in
the fall. Positive NDVI slopes over nearly all vegetation classes indicate a longer durance
of the vegetation period. Contrary trends were detected in the tundra. The tundra
ecosystems of the northern Eurasia latitudes seemed to be affected by trends of negative
NDVI and positive NDWI slopes. The significant trends could be acquired in the prostrate
shrub tundra followed by sedge and shrub tundra. This may be explained by earlier
snowmelt from higher temperature anomalies since the last eight years. The estimates over
these mainly climate controlled processes can be consolidated by analyzing the surface
temperature anomalies from 1998 to 2005. In comparison with the base from 1951 to 1998,
positive surface temperature anomalies are observed particular in spring and fall season.
The analysis of the summer NDVI trends with GLC2000 cropland classes on Oblast level
in Russia point at hot spots of agricultural land changes in the south-western part of Russia.
Comparisons with multi-temporal Landsat data showed natural succession on former
agricultural land is in particular located at remote areas, along river lines, forests and along
the state and Oblast borders.

-6-
 Application of Multitemporal Optical and SAR Data for Different Forest
Studies
Damdinsuren AMARSAIKHAN

Institute of Informatics and RS, Mongolian Academy of Sciences

Faculty of Geography and Geology, National University of Mongolia

Abstract

The aim of this study is to demonstrate different applications of the passive and active
sensor data for different forest studies in Mongolia. For this purpose, Landsat TM data of
1988, SPOT XS image of 1997, ERS-1/2 tandem pass SAR images of 1997, ERS-2 SAR
and JERS-1 SAR intensity images of 1997, and other thematic information are used and
different digital image processing techniques are applied. The results indicated that the
integrated use of optical and microwave data can be successfully used for different forest
studies as well as for differentiation between the fuzzy boundaries of different forest and
vegetation classes.

Key words: Optical, SAR, Multitemporal, Forest change, Reflectance, Backscatter

1. Introduction

Optical remote sensing (RS) data sets taken from different Earth observation satellites such as
Landsat and SPOT have been successfully used for forest monitoring and management since the operation
of the first Landsat launched in 1972. SAR images taken from space platforms have been widely used for
different forest applications since the launch of the ERS-1/2, JERS-1 and RADARSAT satellites. The
combined application of data sets from both sources can provide unique information for different forest
studies, because passive sensor images will represent spectral variations of the top layer of the forest
classes, whereas microwave data with its penetrating capabilities can provide some additional information
about forest canopy [3].

The aim of this research is to demonstrate different applications of the passive and active sensor d
ata for different forest studies in Mongolia that is a) to conduct a forest change study using multite
mporal optical RS images, b) to create a forest biomass map using SAR images and c) to analyze t
he boundary between fuzzy classes: grass-herb and young forest using both optical and SAR imag
es. As a test site, Bogdkhan Mountain situated in central part of Mongolia, near the city of Ulaanb
aatar has been selected. To reach the final goals, different RS and GIS techniques have been appli
ed.

2. Study area and data sources

As a test site Bogdkhan Mountain situated in central part of Mongolia, near the city of Ulaanbaatar has
been selected. The mountain is a protected area and has a territory of 41651ha, of which 55% is covered by
forest. The mountain has 588 species of high plants, which are related to 256 genuses of 70 families. 135
species such as carex, artemisa, oxytropis that relate to 11 main genuses comprise 22.9% of all species
distributed on the mountain. Cedar and larch dominate in the forest cover but pine, birch, spruce and poplar
are also occur [1].

The data used consisted of Landsat TM data from summer of 1988, SPOT XS image of 19 June
1997, ERS-1/2 tandem pass SAR single look complex (SLC) images acquired on 10 and 11 October 1997,
ERS-2 SAR intensity image of 25 September 1997 and JERS-1 SAR intensity image of April 1997. In
addition, a topographic map of 1984, scale 1:50,000 and a forest taxonomy map, scale 1:100,000 were
available.

-7-
 Figure 1. a) Forest taxonomy map of Bogdkhan Mountain, b) The selected part of the
study area, c) Landsat TM image of 1988, d) SPOT XS image of 1997, e) Classified
image of Landsat TM, f) Classified image of SPOT XS.

3. Forest change study using multitemporal optical RS images

In the test area, most of the mixed forests that represent fuzzy boundaries among different forest
classes were situated in the central and western parts of the mountain. Therefore, for the analysis, the areas
situated in these parts have been considered (figure 1a,b). Initially, the optical images (i.e. Landsat TM
data of 1988 and SPOT XS image of 1997) were thoroughly analyzed in terms of brightness and geometric
distortion and the images were of a good quality. Then, the SPOT XS and Landsat TM images were
successively georeferenced to a UTM map projection using a topographic map of the study area, scale
1:50,000. The ground control points (GCP) were selected on clearly delineated sites and in total 9 regularly

-8-
distributed points were chosen. For the actual transformation, a second order transformation and nearest
neighbour resampling approach [6] have been applied and the related root mean square (RMS) errors were
0.68 pixel, and 0.76 pixel, respectively.

In order to demonstrate the forest changes, the selected multitemporal optical images were classified using
the traditional statistical maximum likelihood classification (MLC) [7,9]. For the actual classification green,
red and near infrared bands of the images were used and the images were classified into just two classes:
forest and non-forest. The original SPOT XS and Landsat TM images and the results of the MLC are shown
in figure 1c-f. As seen from the figure 1, different local changes had occurred in the mountain within a 9
year period.

4. Creation of a forest biomass map using SAR images

In the present study, for the creation of a forest biomass map of Bogdkhan Mountain, ERS-1/2 tandem pass
SAR images and JERS-1 SAR intensity image have been used. Initially, we had to extract coherence and
amplitude images from the ERS-1/2 data sets and for this purpose, the techniques used in Amarsaikhan and
Sato (2004) have been applied.

Figure 2. Forest biomass map of Bogdkhan Mountain

In general, the coherence is a measure of the variance of the phase difference of the imaged surface in the
time between the two SAR data acquisitions. The coherence values range between 0 and 1. If some land
surface changes had occurred in a target area between the two image acquisition periods, then coherence is
low and if no changes had occurred, then the coherence is high [10]. In general, the coherence over a dense
forest and shrub will be the lowest, while for the bare soil, the coherence will be the highest. Based on this
characteristics of coherence, it is possible to define different forest volumes which are directly related with
forest biomass. To create a forest biomass map, initially the SAR images were successively georeferenced to
a UTM map projection using a topographic map of the study area, scale 1:50,000. Then, the combined SAR
bands were classified using the MLC defining such biomass classes as very high, high, moderate and low
(figure 2). It was not possible to define the amount of biomass in the areas affected by radar layover.

-9-
5. Analysis of the boundary between fuzzy classes using ERS-2 SAR image

In some areas of the forest classes represented on the optical images, the boundary between fuzzy classes:
grass-herb and young forest could not be distinguished due to their similar spectral characteristics.
However, these two classes might be distinguished on the SAR image because they have different structure
that can cause different backscatter return. These two fuzzy classes have the following backscattering
properties [2,5,8].

From forest canopy, at different radar wavelengths, volume scattering derived from multiple-path
reflections from leaves, twigs, branches and trunks can be expected. However, in case of the ERS-2 SAR
data with its VV polarization only volume scattering from the top layer can be expected, because the
wavelength is too short to penetrate the forest canopy. The backscatter will also be influenced by the local
incidence angle as well as the underlying topography. In total, the forest area will behave as a diffuse
reflector due to volume scattering although some other scattering might also be expected depending upon
the height and geometry of the trees. As a result, the area will have brighter appearance on the radar image.

Grass-herb will behave as a mixture of grass and soil and the backscatter will depend upon the volume and
characteristics of either of them. In C-band frequency, such a class will have components of both diffuse
and specular reflection depending on the plant characteristics and incident angle. The backscattering of soil
will also depend on different surface and system parameters. Specifically, the backscatter from a soil layer
is very much dependent on the moisture content and the higher the water content the more reflection is
expected. However, in this mountain area, the moisture content cannot be high enough to cause high
reflection, and the reflection from the soil will most probably be dominated by specular reflection. As a
result, the backscatter from this class will not be as high as in the case of volume scattering, thus resulting
in lower to middle brightness.

a b

c d

Figure 3. (a) The original speckle suppressed image. (b) SFCC image (7x7 variance=R,
5x5 edge enhancement=G, and 5x5 mean euclidian distance=B). (c) A subset from
(a) indicating some of the selected sites. (d) A subset from (b).

- 10 -
Initial visual inspection of the speckle suppressed SAR image gave some distinctions between different
features, but further interpretation highly required local knowledge about the sites and backscatter
properties. To improve the image interpretation and increase the tonal discrimination between different
forest and non-forest types, a synthetic false colour composite (SFCC) image has been created. To create
such an image, at first the gammamap filtered SAR image was filtered by the use of different texture
analysis and high pass filters of different sizes. Then, the results of three different filter operations were
assigned to the red, green and blue (RGB) colours, respectively. The best colour image to represent the
tonal variations was obtained by the combination of the results of 7x7 variance, 5x5 edge enhancement,
and 5x5 mean euclidian distance filters. Figure 3 shows the comparison between the original speckle
suppressed image and the created SFCC image. As seen from figure 3, despite the radar layover and
foreshortening effect, the SFCC demonstrates more tonal variations between the two fuzzy classes: young
forest and natural vegetation, which were not distinguishable on the optical images. The radar image in
these areas shows brighter colour if there is a forest due to volume scattering. The tonal variations change
to darker colour when there are classes that cause less volume scattering.

6. Conclusions

The aim of this study was to demonstrate different applications of the optical and microwave data sets for
different forest studies in Mongolia. Within the framework of the study a) a forest change study using
multitemporal optical RS images, b) creation of a forest biomass map using SAR images and c) analysis of
the boundary between fuzzy classes: grass-herb and young forest using both optical and SAR images, were
carried out.

Overall, the study demonstrated that the integrated use of optical and microwave data can be successfully
used for different forest studies as well as for differentiation between the fuzzy boundaries of different
forest and vegetation classes.

References

1. Adyasuren, Ts., Shiirevdamba, Ts., and Darin, B., 1998, Ecosystems Atlas of Bogdkhan Mountain,
Ulaanbaatar, Mongolia, pp40.
2. Amarsaikhan, D., and Ganzorig, M., 1999, Interpretation and comparison of AirSAR quad-polarised
radar images. Proceedings of the 20th Asian Conference on RS, Hong Kong, China, 22-26 November
1999, 695-700.
3. Amarsaikhan, D., Ganzorig, M., Batbayar, G., Narangerel, D., and Tumentsetseg, Sh., 2004, An
integrated approach of optical and SAR images for forest change study, Asian Journal of
Geoinformatics, No.3, 2004, pp.27-33.
4. Amarsaikhan, D. and Sato, M., 2004, Integration of RS and GIS for sustainable forest management,
International Boreal Forest Research Association Conference, Alaska, USA, 3-6 May 2004.
5. Amarsaikhan, D., Ganzorig, M. and V.Battsengel, 2005, Knowledge Acquisition in C-band and L-
band Radar Frequencies, CD-ROM Proceedings of the Asian Conference on RS, Hanoi, Vietnam,
pp.DTP4-1_1-4.
6. ERDAS, 1999, Field guide, Fifth Edition, ERDAS, Inc. Atlanta, Georgia.
7. Mather, P.M., 1999, Computer Processing of Remotely-Sensed Images: An Introduction, 2nd edition
(Wiley, John & Sons).
8. Richards, J.A., Milne, A.K., and Forster, B.C., 1987, Remote sensing with synthetic aperture radar.
Centre for Remote Sensing, UNSW, Sydney, Australia.
9. Richards, J.A., 1993, Remote Sensing Digital Image Analysis-An Introduction, 2nd edition (Berlin:
Springer-Verlag).
10. Weydahl, D.J., 2001. Analysis of ERS SAR coherence images acquired over vegetated areas and
urban features. International Journal of Remote Sensing, 22, 2811-2830.

- 11 -
 Brief description on land cover and current activities in Korea
Nam-Sun, Oh1, Hong-Yeon Cho2, and Geun-Sang Lee3
1
Department of Ocean Civil Engineering, Mokpo National Maritime University, Mokpo, S. Korea 530-729, Tel) +82
061-240-7078, Fax) +82 061-240-7284, Email: sun@mmu.ac.kr
2
Korea Ocean Research and Development Institute (KORDI), Ansan, S. Korea 425-600,
Tel) +82 031-400-6318, Fax) +82 031-408-5823, Email: hycho@kordi.re.kr
3
Korea Water Resource Corporation (KOWACO), Daejeon, S. Korea 305-391
Tel) +82-42-860-0354, Fax) +82-42-860-0319, Email: ilovegod@kowaco.or.kr

Abstract
This paper describes brief description on land cover and current activities in Korea. In 1998, the
Korean Ministry of Environment (MOE) constructed the land cover map on the basis of the satellite
data in an attempt to support main environmental policy formulation such as non-point source
evaluation, environmental impact assessment and riparian buffer zone establishment. The MOE
have constructed three different levels of the land cover map (coarse, medium, and fine) for various
purposes. Use of the land cover map has been increasing in broader areas, such as hydrology,
agriculture, engineering, and atmosphere as well as environment.

1. Introduction

The Korean community recently stands on the brink of a new era of environmental and water
resources management with the advent of the remotely sensed data handled by broad geographic
information system (GIS). Traditionally environmental problems have been concentrated on the control of
point source but non-point source, which is hard to manage systematically, is becoming a main interest to
scientist.
The pollution resulting from non-point source is involved with both natural characteristics and artificial
activities. Thus, the impact of non-point source on social and economic activities is considerable and
complicated. In order to deal with the complicated problems of pollution process induced by non-point
source, broad information on surface characteristics needs to be collected and quantified. Comprehensive,
precise, and rapid analysis is simultaneously necessary.
In 1998, the Korean Ministry of Environment (MOE) constructed the land cover map on the basis of the
satellite data in an attempt to support main environmental policy formulation such as non-point source
evaluation, environmental impact assessment and riparian buffer zone establishment.
Old fashioned documental information is converted into the digitally formatted information such
as satellite photograph, land cover map, and natural ecology system. The remotely sensed data has no
restriction in temporal and spatial scale and then data availability is very flexible.
In typical fashion, the land cover map constructed by MOE is classified into specific classes and the land
cover map, along with the soil texture map and DEM, is easily accessible to people without special
knowledge.
More people in broader areas, such as hydrology, agriculture, engineering, and atmosphere as well as
environment, are using the land surface map and the MOE is trying to upgrade the land surface data.

Table 1. Change in environmental management in Korea. In the 80’s, environmental management

in Korea was focused on local and point source but the trend was changed into national and non-
point source after 2000. Accordingly the RS and GIS skill was becoming more important.

- 12 -
 Period 1980's 1990's After 2000
- regulation of
- daily environmental - nature-friendly environment
pollutant source
management management
environmental - pollution level
- regulation of pollutant - regulation in total maximum
management measurement
source daily load (TMDL)
- passive
- passive evaluation - active evaluation
evaluation
data - point source data - plane source data - spatial source data
- regional
- local - national
coverage - statistical analysis,
- simple source - mapping, GIS tool
comprehensive report
- COBOL - RDBMS - RS
Tools
- FORTRAN - EXCEL - GIS

2. Mapping the land cover (MOE activities)

The MOE started mapping the land cover in 1998 and activities are continued, finishing the fifth
stage in 2005. At the first stage, the MOE constructed the coarse land cover map (hereafter named coarse
level: 7 classes) in an attempt to grasp the land use/cover characteristics nationwide, which is calibrated to
the 215 points of the ground samples. The activities are becoming more specified and functionalized in
need of practical application. At the second stage in 2000, the MOE classified the land cover into 23
classes (hereafter named medium level) and 48 classes (hereafter named fine level) as shown in table 3 to
more effectively retrieve necessary information for environmental management. Especially, for Seoul
cosmopolitan area, medium level of the land cover map was constructed using 5m resolution of IRS image.
6 areas were also selected for intensive analysis and fine level of the land cover map was constructed. The
guideline to making the fine level of the land cover map is given to local governments so that they can
make the local land cover map for their own purpose. All the geographic information constructed by the
MOE is freely available to people on the internet (http://egis.me.go.kr/egis/intro.asp).
At the third stage in 2002, the project was focused on the urban area. The IKONOS image was used to
characterize the urban cover such as building and road. During this stage, the standardized classification
was suggested on the basis of the precedent results to make the classification processes systematic. The
output of the project was tested with water quality modeling in the small basin called Gyeongan-Cheon and
examined the accuracy and usefulness of the land cover classification.
At the fourth stage in 2003, the SPOT image (2.5m resolution), which is monitored since 2002, was used.
Some information was not open to public because of its security but the secured information was available
with permission by the MOE (called infranet service).
At the fifth stage in 2005, medium level (23 classes) of the land cover map was constructed on a
nationwide scale. Subsequently applicability of the data was investigated in various areas: water quality,
flood forecasting, drought monitoring, and soil loss estimation etc. Relative comparison of the existing
methods was performed to improve the quality of the land cover information.
The followings show the chronicle of the land cover map in Korea.

Table 2. Land cover map construction by a year

budget($) Activities
- mapping coarse level of the land cover
࡮South Korea area(90's), 1:50000 scale, 238 sheets
1st stage 77,000
- digital elevation map construction
(1998.11~1999.11)
࡮South Korea area(90's), 1:50000 scale, 238 sheets
- sampling points (215 points)

- 13 -
 - mapping coarse level of the land cover
࡮South Korea area(80's), 1:50000 scale, 238 sheets
࡮North Korea area(80's), 1:50000 scale, 249 sheets
࡮North Korea area(90's), 1:50000 scale, 249 sheets
- mapping medium level of the land cover
࡮Seoul metropolitan area, 1:25000 scale, 121sheets
2nd stage 2,300,000
(2000.12~2001.6) - mapping fine level of the land cover
࡮6 test areas, 1:5000, 7 sheets
- aerial photographs scanning and manufacture
࡮aerial photographs scanning, 20,569 sheets
࡮aerial photographs manufacture, South Korea
- sampling points (200 points)
- constructing internet service system

- mapping fine level of the land cover map

࡮Han river and Geum river basin, 1:25000
3rd stage ࡮Standardized and guideline suggested
896,000
(2002.2~2002.12) - case study
࡮water quality modeling to a Gyeongan-Cheon basin.
- enforcement of internet service system

- mapping medium level of the land cover map

࡮Nakdong rive basin, 1:25000
4th stage
950,000 - broadening internet service system
(2003.2~2003.12)
- construction of intranet service system
- case study.
- mapping medium level of the land cover map
5th stage ࡮Yeongsan river basin and Jeju island, 1:25000
1,190,000
(2004.6~2005.4) - construction of aerial photographs
࡮South Korea, 1:25000, 101 sheets

3. Mapping processes

As mentioned earlier, the land cover map can be used for various areas but it is true that the MOE
performed the mapping processes with emphasis on environmental management such as non-point source,
watershed management, and basis for modeling approach. The followings briefly describe the mapping
processes;

ƒ The overseas classification systems (USGS and/or CORINE) were thoroughly reviewed and
each class is selected on the basis of these systems. Then, each name was adjusted to Korean
environment in which surface conditions are more diverse and individual patches are relatively
small.
ƒ Some peculiar items like the bushes in high mountain region were excluded in the list. The
classification system also considered seasonal variation such as vegetation fraction, rice
farming area, and tidal flat.
ƒ The classification system was made in such a way that coarse level of the land cover
classification includes several medium levels of the land cover classifications (see table 3). In
that case you can track coarse level of the land cover classification from medium level of the
land cover classification.
ƒ Coarse level of the land cover was structured for prompt response, while medium level of the
land cover system was structured for assisting monitoring or providing a basis for modeling
setting in national-wide management plans.

- 14 -
Table 3. Coarse and medium level of the land cover classification

A large-scale classification A middle-scale classification

Class Code Class Code
Residential district. 110
Industrial area. 120
Commercial area. 130
Urban and dry area 100
Recreation facilities 140
Transport facilities 150
Public facilities 160
A rice field 210
A dry field 220
Agriculture region 200 Glass culture 230
An orchard 240
Other culture 250
non-conifer forest 310
Forest region 300 A coniferous forest 320
Mixed forest 330
A nature grassland 410
Grassland 400 A golf course 420
Other grassland 430
Inland wetlands 510
Wetlands 500
Coastal wetlands 520
A mining region 610
Barren area 600
Other barren area 620
A inland water 710
Water area 700
A marine water 720

Fig.1 presents the land cover map constructed by the MOE. The left panel shows the SPOT 5 derived
satellite image and the right panel shows the corresponding land cover map. The symbols for each land
cover and location are marked under the map.

- 15 -
 Fig 1. A sample of land cover map
Fig.2 and 3 shows coarse and medium level of the land cover map for Andong lake basin which is located
in South Korea. The table 4 and 5 shows vegetation fraction which corresponds to Fig.2 and Fig.3

Fig 2. Land cover map(coarse) Fig 3. Land cover map(medium)

Table 4. Analysis of land cover map of Fig.2 (coarse)

- 16 -
 Land cover Area(ᓚ) percentage(%) Remarks
Urban area 22.532 1.417
Rice field 43.750 2.750
Dry field 145.194 9.128
Forest 1,307.775 82.215 Fraction of farm land
Grassland 6.544 0.411 (11.878%)

Wet land 9.813 0.617

Unused area 16.467 1.035
Water area 38.587 2.427
Total 1,590.662 100.000

Table 5 Analysis of land cover map of Fig.2 (medium)

Class Code Area(ᓚ) Occupation ratio(%)

Residential district. 11 12.251 0.770

Industrial area 12 0.870 0.055
Commercial area 13 1.574 0.099
Recreation facilities 14 0.377 0.024
Transport facilities 15 6.503 0.409
Public facilities 16 0.957 0.060
A rice field 21 43.750 2.752
A dry field 22 123.763 7.784
Glass culture 23 0.277 0.017
An orchard 24 18.687 1.175
Other culture 25 2.467 0.155
non-conifer forest 31 371.797 23.383
A coniferous forest 32 681.280 42.848
Mixed forest 33 254.698 16.019
A nature grassland 41 2.904 0.183
A golf course 42 0.040 0.002
Other grassland 43 3.600 0.226
Inland wetlands 51 9.813 0.617
A mining region 61 2.835 0.178
Other barren 62 13.632 0.857
Water area 71 38.587 2.427
Total 1590.662 100.000

- 17 -
 Glacier change estimation using Landsat TM data
M. Erdenetuya* (PhD), P.Khishigsuren**, M. Otgontugs*
* National Remote Sensing Center/Information and Computer Center
** Agency of Land Affairs, Geodesy and Cartography
m_erdenetuya@yahoo.com, khishigsuren25@yahoo.com, motgoo@yahoo.com

1. Introduction
By the geographical position and “ecotone” formation whole territory of Mongolia was selected as
main part of study areas in Northeastern Asia. In Mongolia located Altai, Khangai and Khuvsgul high
mountains and formulated permanent snow and ice. As mentioned the glacier is most important land cover
type to keep the freshwater resources and as indicator of the climatic temporal variability.
Since the middle of the last century, the global climate is changing drastically and as a result the
current climate experiences more frequent extremes in Mongolia causing big losses amongst animal and
land degradation. Also the National Program on Climate Change mentioned that for every increase of 3
degrees in air temperature, there will be a 10 percent reduction in carbon (C) and nitrogen (N) contents of
the plant-soil ecosystem, a 3-10 percent reduction in pasture vegetation and a biomass reduction of 21.5
percent1. Nowadays, the air temperature has already increased by 1.9 degrees2. The global warming factor
could strongly influence to melt the permanent snow and ice on top of higher mountain system in Western
Mongolia at the same time to decrease fresh water resources.
In order to estimate glacier area we have used the Landsat TM data from different period and
applied NDSI (normalized difference snow index) calculation method and maximum likelihood
classification approach. Also the SRTM/DEM data have been used for 3 dimensional processing and for
calculation of the glaciers area.

2. Data and Methods

There have been used Uvs lake hydrology data of 1970-2002 and meteorological data observed at
Ulaangom and other stations, located in the basin.
For estimation of glacier area dynamics have been used Landsat of 3 different time period data for
glacier massifs as Kharkhiraa, Turgen Tsambagarav, Munkhkhairkhan and Sair Mts. The Landsat TM and
ETM+ scenes were selected from 141-144 paths and 26-27 rows and obtained on following dates.
x 16 September 1990 (used for only Munkhkhairkhan Mt. Noted with* in Table 1)
x 25 June 1992
x 10 Sep., 2000 /Kadota and Davaa, 2003/
x 08 August 2002

For mapping the glacier area from the satellite image we have applied several approaches such as,
x To identify spectral characteristics of glacier in each Landsat band
x To apply bands combination method for glacier extraction
x To apply both supervised and unsupervised classification methods
x To calculated normalized difference snow index
x To analyze three dimensional view of images
x To compare calculated areas

In order to distinguish snow from similarly bright soil, rock and cloud we have calculated NDSI (normalized
difference snow index) by following formulae:

(TM 2  TM 5) (1)
NDSI
(TM 2  TM 5)

In Fig. 1 shows Landsat TM data fragments of 2 mountains in different periods.

1
National Program on Climate Change, 2000
2
L. Natsagdorj – Assessment of Climate factors to Mongolian pasture degradation, 2006

- 18 -
 A. B.
Fig. 1. Landsat TM, ETM+ images of glaciers in 1990 and 2002
(A – Tsambagarav, B – Sair mountain)

3. Results and discussions

On the background of the paleoclimate data, more precise climatic chronology has been given by the
research on reconstruction of climate data with tree ring indices.
One (Khalzan Khamar) of the tree ring chronology sites were selected in the Altai Mountains. It has
permanent snow fields, ice and permafrost and located near the timberline where temperature appears to be
limiting factor for growth. Similar fluctuations have been derived from the tree-ring width indices record,
taken from the sample of the Turgen Mountain. Reconstructed with 5 year moving average of tree ring
width indices (r=0.60) of the Turgen and instrumental, annual temperature at Ulaangom station were well
correlated and show that highest temperatures are in last decades.
Melting of the ice masses in the mountains, the end of pediment formation on the lower mountain slopes
and the slow regeneration of plant cover owing to rising temperatures and increased precipitation all lead
to a relative quick rise in lake level, since the basins filled up with both rainwater and melt water from the
rapidly melting glaciers. Therefore, pretty good relationship exists between 5 year moving average of tree
ring width indices and annual average of water level of the Uvs lake for the period of 1970-2002.
Reconstructed with 5 year moving average of tree ring width indices (r=0.76) of the Turgen and
observed water level of the Uvs lake were well correlated and show that highest water levels are observed
in the last decade.
It is obvious that dynamics of the water balance elements of the Uvs Lake were following water level
fluctuations. However, it is possible to estimate water balance elements in last 40 years.

Spectral characteristic of glacier

The maximum, minimum and mean spectral values of each of land cover class were calculated from
high resolution Landsat images, based on that spectral reflectance of each land cover types is different in
each wave length of electro-magnetic radiation [6].
On Landsat ETM+ data the glacier spectral values accounted as 255, 145-255, 191-255, 116-217, 20-31
and 3-18 in each spectral bands 1-5 and 7 respectively. The calculated spectral values of land cover classes
used as a reference value for glacier classification.
Also for identification of glacier have been used band combination method and the glacier was extracted
in each 3,2,1 and 4,3,2 and 5,4,3 combinations of bands as showed in Fig. 2.

Fig. 2 Landsat ETM+ data combination

(Tsambagarav mountain)

- 19 -
 For extraction of glacier area we have used the NDSI (normalized difference snow index) calculation to
distinguish snow from similarly bright soil, rock and cloud from entire images.

Fig. 3 NDSI images of Kharkhiraa and Sutai mountains

Also the SRTM/DEM data have used for 3 dimensional processing and identification of the glaciers area.
In next step of processing we will integrate DEM data to calculate real or volume area of glaciers.

A.

Fig. 4 Three
Dimensional
Landsat images
(A –
Kharkhiraa, B –
Tsambagarav

B.

Reason of reduction of evaporation from water surface area of the lake can be the decrease in water
temperature due to the increase in melt water, draining to the Uvs Lake primarily in the form of
underground flow. Retreat of Kharkhiraa and Turgen glaciers is drastically increasing since 1940s.
Kharkhiraa, Turgen, Kharkhiraa, Tsambagarav and Tavanbogd glacier areas were 50.13, 43.02, 105.09 and
88.88 sq. km, estimated from topographic map, scaled as 1:100 000 and compiled in 1940s [3]. Areas of
the Kharkhiraa, Turgen, Munkhkhairkhan, Tsambagarav and Sair glaciers were decreasing by 45.5, 33.7,
25.8, 21.4 and 42.5 percent since 1992 till 2002, respectively (Table 1).

Table 1. Changes in glacier areas

1940s 8 Aug.
Glacier massif topo map 25 June 1992 10 Sep 2000 2002
Kharkhiraa 43.02 57.37 36.08 31.29
Turgen 50.13 51.03 34.74 33.83
Munkhkhairkhan - 36.96* - 27.42
Tsambagarav 105.09 90.98 74.8 71.52
Sair - 11.51 - 6.62

- 20 -
 Retreat of glaciers is intensified in last decades and many of glacier peaks got ice free especially in very
dry year 2002. We have analyzed Landsat TM, ETM+ data from 1992, 2002 and compared their massif
changes (Fig. 5).

Fig. 5 Comparison of Turgen and Tsambagarav mountains

glacier massif in 1992 (red) and 2002 (blue)

4. Concluding remarks
Comprehensive investigation focusing on glacier mass balance, ground and surface water interaction,
dating and the environment changes are desired in the near future.
Compilation of glacier inventory is important issue, using remote sensing and ground observation data,
especially vertical air photographs, which are basic information for development of hydro-climate-glacier
and integrated water resource management studies.
For satellite data application on glacier mapping we still need fresher (2005 and 2006) and higher
resolution satellite (ASTER, IKONOS) imageries.

References
[1]. Davaa G. Dashdeleg N. Tseveendorj N., The dynamics of water balance elements of the Uvs lake,
Proceedings of International symposium on “Global change- Uvs lake”, Ulaanbaater, Mongolia, 1991,
pp.18-19
[2]. Erdenetuya M. and Khudulmur S. Glacier assessment using Landsat satellite data. Proceeding of First
National Conference on Remote Sensing and GIS Applications. Ulaanbaatar, Mongolia, May 2005,
pp103-106 (in Mongolian).
[3]. Grunert J. F. Lehnkuhl, Walter M., Paleoclimatic evolution of the Uvs Nuur basin and adjacent areas
(Western Mongolia), Quaternary International 65/66 (2000), pp. 171-192.
[4]. Kadota T. and Davaa G. A preliminary study on Glaciers in Mongolia, proceedings of International
workshop “Terrestrial Change in Mongolia”, Japan, 2003, published in Mongolia, Ulaanbaatar, 2004,
pp.
[5]. Lovilius N.V., Davaajamts T. and Gunin P.D., 1992. Dendroindications of forest growth conditions
in Mongolia and possibilities of forecasting (in Russian), Russian Academy of Sciences, Puchino,
Moscow, pp. 32-49.
[6]. Munkhtuya Sh., (2004): Remote sensing methodology and technology for land cover classification.
Dissertation, UB, Mongolia.

- 21 -
 Forage Monitoring Technology to Improve Risk Management Decision
Making by Herders in the Gobi Region of Mongolia
J. P. Angerer1, L.Bolor-Erdene2, D. Tsogoo2 , M.Urgamal2 and S. Granville-Ross2
1
-Dept of Rangeland Ecology & Management, Texas A&M University, College Station, Texas, USA;
jangerer@cnrit.tamu.edu
2
-Mercy Corps, P.O. Box 761, Ulaanbaatar - 49, Mongolia ; foragestaff@mercycorps.org.mn; 976-11-461145

Due to increased frequency and coverage area of drought and "zud" disaster in
Mongolia in recent years, the pastureland vegetation, its structure and composition has
considerably been changed and the natural and ecological balance distorted that directly
affects the livelihood of the herders. Therefore, the predefinition of present and future
pasture conditions, distribution of relevant information to the herders to raise their
awareness, undertaking appropriate preparatory and organizational actions and consulting
are urgently needed. Therefore we are workong on developing risk management
technologies to provide drought and winter disaster early warning to improve rural
business in the livestock sector of the Gobi region.
Our objectives are developing a regional forage monitoring system that provides
near-real time spatial and temporal assessment of current and forecasted forage conditions
and a communication infrastructure to provides herders with forage condition information
to assist in making timely and specific management decisions.
We established 246 monitoring sites were established in the Gobi region to
parameterize the forage production model and to enable mapping of available forage.
Vegetation parameter collection has been completed at each monitoring site. Grazingland
communities being monitored range from mountain steppe to desert grasslands.
Across the monitoring sites, 390 unique species of plants were encountered. Of
these, 258 were forbs, 69 were grass or grass-like, and 65 were shrubs
Communication protocols are being developed to provide herders, administrative, and
government personnel in the region with current forage conditions and 90-day forecasts on
a 16-day cycle.

- 22 -
 Monitoring environmental degradation in Mongolia with NPP and rainfall
data.
M. Tyburski1,2, R. Tsolmon2, D. Sodnomragchaa3, R. Harris1 and A. Warren1.
1
Department of GeographyUniversity College London,London, UK
2
NUM-ITC-UNESCO Remote Sensing and GIS Laboratory, National University of Mongolia
Ulaanbaatar, Mongolia
3
Mongolian Research Institute of Emergency, Ulaanbaatar, Mongolia

Abstract
Environmental degradation was assessed with net primary production and precipitation
estimates for Mongolia during the period 1982-2000. Umnugobi aimag was the most
degraded aimag during the study period and future studies should use high resolution
satellite remote sensing imagery to determine why degradation occurred in this aimag. We
suggest future research on environmental degradation use a greater suite climate variables
than just rainfall data.

Introduction

Mongolia became a signatory to the United Nations Convention to Combat Desertification in 1994. 90%
of Mongolia is considered vulnerable to desertification pressures (MNEM, 1997). Desertification, i.e.
environmental degradation in arid environments (UNCCD, 1994), can be defined as a reduction in the
biological productivity of the land (Reynolds & Stafford-Smith, 2002). 76.5% of Mongolia is used for
pasture land and pasture yields have decreased by 19-24% during the period 1977-2002 (UNCCD, 2002).
It is therefore necessary to determine whether the environmental degradation of Mongolian pasture land is
attributable to human activity or climate change. In this study, we assess environmental degradation in
Mongolia with satellite remote sensing imagery and modeled precipitation data during the period 1982-
2000 in an effort to deduce where Mongolian pasture land has been most likely degraded by anthropogenic
forces.

Remote sensing of environmental degradation

Prince (2002) reviews various methods used by satellite remote sensing to monitor environmental
degradation. Of these approaches, the Rain Use Efficiency (RUE) model (Prince et al., 1998; Nicholson et
al. 1998; Diouf & Lambin, 2001; Yu et al. 2004; Hein & De Ridder, 2006) is the most useful method for
monitoring large-scale environmental degradation in environments with less than 1000 mm of annual
rainfall. However, the RUE model has been criticized for the incorrect assumption that in the absence of
degradation, the RUE amount should be constant over time (Hein & De Ridder, 2006). Prince (2002) also
proposed a DNPP model to quantify environmental degradation in environments receiving greater than
1000 mm of annual rainfall. At finer spatial scales, other methods have been developed to accommodate
for factors such as soil type, elevation, distance from water source, etc. (Stafford-Smith & Pickup, 1993;
Pickup,1996; Pickup et al., 1998;Wessels et al., 2004), which contribute to the spatial variability of
biological productivity in the context of a specific locality. Therefore, monitoring environmental
degradation by satellite remote sensing imagery is scale dependent (Prince, 2002) and requires more
detailed parameterization at increasing spatial resolution.

Materials and Methods

Data
Net Primary Production (NPP) is the biomass amount for a given time (Geider et al., 2001), and we use
annual amounts of NPP to measure biological productivity. NPP estimates were derived from the Global
Production Efficiency Model (GLO-PEM). GLO-PEM is a terrestrial ecosystem model that is driven by
reflectance data from satellite remote sensing imagery (Prince & Goward, 1995). The annual NPP data
derived from GLO-PEM were downloaded online from the University of Maryland’s Global Land Cover

- 23 -
Facility (downloadable at: http://glcf.umiacs.umd.edu/index.shtml). Cao et al. (2004) performed the
GLO-PEM modeling procedures and described in detail the inter-annual NPP change for the GLO-PEM
data set we used in this study. The GLO-PEM data was modeled with NDVI estimates from the NOAA
AVHRR Pathfinder data set, which provides global coverage at 8km2 pixel resolution in 10-day increments
during the period 1982-2000. The University of East Anglia’s Climate Research Unit’s (CRU) global 0.5
degree precipitation data (Mitchell & Jones, 2005) were used to calculate annual rainfall estimates for
Mongolia. Vegetation zone (MNRSCICC ,1999) and soil data (CWG, 2004) were used to aggregate
vegetative productivity in Mongolia. Annual and inter-annual biological variability for different
vegetation zones were reported in past field-based case studies (Fernandez-Gimenez & Allen-Diaz, 1999;
Fernandez-Gimenez & Allen-Diaz, 2001). Mongolian nature reserves were used to control for natural
amounts of biological productivity. IUCN Protected Areas boundaries (WDPA Consortium, 2004) were
downloaded from the University of Maryland’s Global Land Cover Facility (downloadable at:
http://glcf.umiacs.umd.edu/index.shtml). Nature reserves were only sampled after the reserve was
established (Table 1).

Quantitative Methods
First, change analysis by a linear regression method (y = a+bx) was used to quantify NPP change during
the period 1982-2000. Second, we calculated the annual Rain Use Efficiency (RUE = NPP / Rainfall)
between 1982-2000 and performed a linear regression on the annual RUE estimates to quantify RUE
change. Third, an aggregated method was applied to quantify environmental degradation. Mongolia was
aggregated by vegetation zone (Figure 1) and soil variability (Figure 2). Nature reserves were used as
controls (Table 1). We assumed nature reserves would have the least amount of environmental
degradation related to anthropogenic agency, and biological variability would thus function under natural
conditions. Mean annual NPP and rainfall estimates were sampled for individual soil polygons within each
vegetation zone inside nature reserves and we then quantified the relationship by regression analyses. Soil
polygons with less than five pixels sampled were discarded due to statistical and pixel uncertainties
(Young & Harris, 2005), which may cause discrepancies between mean annual NPP amounts. The
function of the NPP and rainfall relationship was used to calculate a Potential NPP (PNPP) amount from
the CRU annual precipitation estimates falling in each vegetation zone.
The annual Difference in NPP (DNPP) between the Actual NPP estimates (ANPP), derived from
the GLO-PEM NPP data, and modeled PNPP amounts were calculated for each vegetation zone. Linear
regression was used to determine DNPP change during the period 1982-2000. Since we used different
functions for individual vegetation zones to quantify PNPP, the DNPP amounts for different vegetation
zones were assumed to be quantifiably incomparable. Relative DNPP amounts were quantified so that
DNPP amounts in different vegetation zones could be then comparable throughout Mongolia. To calculate
relative DNPP amounts over time, we use the linear function y = a + bx, where relative DNPP change =
(bx/a) * (a/abs(a)). (a/abs(a)) was used for the possibility of a negative y-intercept. Also, we reclassify a
for DNPP amounts between -1 and 0 to equal -1 and DNPP amounts between 0 and 1 to equal 1. We
reclassify so that DNPP amounts for a between -1 and 1 are not confounded by small relative DNPP
amounts. For example, if pixel "1" has bx at 20 and a at .5 and pixel "2" has bx at 20 and a at 1, pixel "1"
will have a relative DNPP amount of 40 and pixel "2" will be 20. This is a misleading result for
environmental degradation.

Results/Discussion
Figure 3 shows the results for regression analyses between mean annual NPP and rainfall for
individual soil polygons of different vegetation zones in nature reserves. In the desert vegetation zone
(Figure 3a), there was a weak relationship in terms of the coefficient of determination (r2 = 0.28), but there
was an obvious linear pattern in the annual NPP and precipitation relationship. This pattern suggests a
relationship between annual NPP and precipitation, but other environmental factors, such as potential
evapotranspiration or plant biology, may also limit NPP in deserts. There was a stronger relationship
between annual NPP and precipitation in the desert steppe (Figure 3b) than the desert vegetation zone in
terms of coefficient of determination (r2 = 0.56), and there was a linear pattern between the NPP and
rainfall relationship. This pattern suggests that annual rainfall was a dominant factor for determining
annual NPP in the desert steppe vegetation zone. In the steppe vegetation zone (Figure 3c), there was a
significant relationship in terms of the coefficient of determination (r2 = 0.71), a higher a-value compared
to desert and desert steppe vegetation zones, suggesting the annual NPP amount is greater in the steppe,

- 24 -
and there was a linear pattern between the annual NPP and rainfall relationship. In the mountain (forest)
steppe (Figure 3d), there was a weaker relationship in terms of the coefficient of determination (r2 = 0.39)
when compared to desert steppe and steppe vegetation zones, but there was a linear pattern between the
annual NPP and rainfall relationship and a much higher a-value compared with desert, desert steppe and
steppe vegetation zones. There was no relationship or pattern between annual NPP and rainfall in either
the taiga or alpine vegetation zones (Figures 3e & d).
Figure 4 illustrates the results for NPP change, RUE change and relative DNPP change in Mongola.
Table 2 depicts the results for the mean amount of annual NPP, RUE and relative DNPP change per aimag.
Arkhangai aimag had the greatest mean amount of negative NPP change. Umnugobi aimag had the
greatest mean amount of both negative RUE change and relative DNPP change.

Conclusion
x NPP change is not a good indicator for environmental degradation by human agency in Mongolia,
because it does not account for climate variability. There is no relationship between rainfall and
NPP in Alpine and Taiga vegetation zones, and therefore, the RUE model is not useful for
monitoring environmental degradation in Alpine and Taiga vegetation zones in Mongolia. More
detailed, high resolution land use and land cover change assessments are required to understand
the local context (Warren, 2002) of environmental degradation in the Omnogovi aimag during the
period 1982-2000.
x In theory, the implementation of the RUE model to monitor environmental degradation in
environments receiving less than 1000 mm of annual rainfall is confounded by biological and soil
constraints within different vegetation zones, and perhaps other climatic variables, such as
radiation and temperature. Nemani et al. (2003) use a mix of precipitation, radiation and
temperature data to determine dominant global constraints on NPP change. We suggest future
environmental degradation work, both in Mongolia and in other countries, to focus upon the
relationships between NPP and precipitation, radiation and temperature in nature reserves.
Potential NPP should then be modeled with the function derived from multivariate analyses
between NPP and precipitation, radiation and temperature.

References
1. Cao M, Prince SD, Small J,Goetz SJ (2004) Remotely Sensed Interannual Variations and Trends
in Terrestrial Net Primary Productivity 1981–2000. Ecosystems, 7, 233–242

2. (CWG) Cryosol Working Group. 2004. Northern and Mid-Latitude Soil Database, Version 1. Data
set. Available on-line [http://www.daac.ornl.gov] from Oak Ridge National Laboratory Distributed
Active Archive Center, Oak Ridge, Tennessee, U.S.A.

3. Diouf A, Lambin E (2001) Monitoring land-cover changes in semiarid regions: Remote sensing
data and field observations in the Ferlo, Senegal. Journal of Arid Environments, 48, 129-148.

4. Fernandez-Gimenez ME & Allen-Diaz, B (2001) Vegetation change along gradients from water
sources in three grazed Mongolian ecosystems. Plant Ecology, 157, 101-118.

5. Fernandez-Gimenez ME & Allen-Diaz, B (1999) Testing a non-equilibrium model of rangeland

vegetation dynamics in Mongolia. Journal of Applied Ecology, 36, 871-885.

6. Geider RJ, Delucia EH, Falkowski PG, et al. (2001) Primary productivity of planet earth:
biological determinants and physical constraints in terrestrial and aquatic habitats. Global
Change Biology, 7, 849-882.

7. Hein L, De Ridder N (2006) Desertification in the Sahel: a reinterpretation. Global Change

Biology, 12, 1-8.

8. Mitchell T, Jones D (2005) An improved method of constructing a database of monthly climate

observations and associated high-resolution grids. International Journal of Climatology, 25, 693-
712.

- 25 -
9. (MNEM) Ministry for Nature and the Environment of Mongolia (1997) National Plan of Action to
Combat
10. Desertification in Mongolia. Ulaanbaatar, Mongolia.

11. (MNRSCICC) Mongolian National Remote Sensing Center/Information and Computer Center
(1999). Ulaanbaatar, Mongolia.

12. Nemani RR, Keeling CD, Hashimoto H, Jolly WH, Piper SC, Tucker CJ, Myneni RB, Running
SW (2003) Climate-Driven Increases in Global Terrestrial Net Primary Production from 1982 to
1999. Science, 300, 1560-1563.

13. Nicholson SE, Tucker CJ, Ba MB (1998) Desertification, drought, and surface vegetation: An
example from the West African Sahel. Bulletin of the American Meteorological Society, 79, 1-15.

14. Pickup G (1996) Estimating the effects of land degradation and rainfall variation on productivity in
rangelands: An approach using remote sensing and models of grazing and herbage dynamics.
Journal of Applied Ecology, 33, 819-832.

15. Pickup G, Bastin GN, Chewings VH (1998) Identifying trends in land degradation in non-
equilibrium rangelands. Journal of Applied Ecology, 35, 365-377.

16. Prince SD (2002). Spatial and temporal scales of measurement of desertification. In: Global
desertification: Do humans create deserts? (eds. Stafford-Smith M, Reynolds JF), pp. 23-40.
Dahlem University Press, Berlin.

17. Prince SD, Brown de Colstoun E, Kravitz L (1998) Evidence from rain use efficiencies does not
support extensive Sahelian desertification. Global Change Biology, 4, 359-374.

18. Prince SD, Goward SN (1995) Global primary production: A remote sensing approach. Journal of
Biogeography, 22, 815-835.

19. Reynolds JF, Stafford Smith M (2002) Do humans create deserts? In: Global desertification: Do
humans create deserts? (eds. Stafford-Smith M, Reynolds JF), pp. 1-22. Dahlem University Press,
Berlin.

20. Stafford Smith DM, Pick-up G (1993). Out of Africa, looking in: Understanding vegetation
change. In: Range Ecology at Disequilibrium: New Models of Natural Variabilty and Pastoral
Adaptation in African Savannas (eds. Behnke Jr RH, Scoones I, Kerven C), pp. 196-244.
Overseas Development Institute and international Institute for Environment and Development,
London.

21. Tucker CJ, Dregne HE, Newcomb WW (1991) Expansion and contraction of the Sahara desert
from 1980-1990. Science, 253, 299-301.

22. UNCCD (2002) Mongolian National Report to United Nations Convention to Combat
Desertification.

23. UNCCD (1994) United Nations Convention To Combat Desertification In Those Countries
Experiencing Serious Drought And/Or Desertification. United Nations General Assembly, New
York.

24. Warren A (2002) Land degradation is contextual. Land Degradation and Development, 13, 449-
459.

- 26 -
25. WDPA Consortium. World Database on Protected Areas" 2004 . Copyright World Conservation
Union (IUCN) and UNEP-World Conservation Monitoring Centre (UNEP-WCMC), 2004.

26. Wessels KJ, Prince SD, Frost PE, van Zyl D. (2004) Assessing the effects of human-induced land
degradation in the former homelands of northern South Africa with a 1 km AVHRR NDVI time-
series. Remote Sensing of Environment, 91, 47–67.

27. Young S, Harris R (2005) Changing patterns of global-scale vegetation

photosynthesis, 1982-1999, International Journal of Remote Sensing, 26,
4537-4563.

28. Yu F, Price KP, Ellis J; Feddema JJ, Shi P (2004) Interannual variations of the grassland
boundaries bordering the eastern edges of the Gobi Desert in central Asia. International Journal
of Remote Sensing, 25, 327-346.

- 27 -
Table 1 List of nature reserves sampled. D. Steppe is Desert Steppe and M.F Steppe is Mountain (forest) Steppe.
Nature Reserve Date Est. Sampled Veg. Sample Nature Reserve Date Est. Sampled Veg. Sample
Khorgo Terkh Zagaan Nuur 1/1/1965 1982-2000 Alpine Bogdkhan Mountain 1/1/1978 1982-2000 M.F. Steppe
Khasagt Khairkhan 1/1/1965 1982-2000 Alpine Mongol Daguur 1/1/1992 1992-2000 M.F. Steppe
Khukh Serkhyn Nuruu 1/1/1977 1982-2000 Alpine Eastern Mongolian Steppe 1/1/1992 1992-2000 M.F. Steppe
Khovsgol Lake 1/1/1992 1992-2000 Alpine Nomrog 1/1/1992 1992-2000 M.F. Steppe
Khovsgol Lake 1/1/1992 1992-2000 Alpine Otgontenger 1/1/1992 1992-2000 M.F. Steppe
Khan Khentee 1/1/1992 1992-2000 Alpine Ugtam Mountain 1/1/1993 1993-2000 M.F. Steppe
Khan Khentee 1/1/1992 1992-2000 Alpine Uvs Nuur Basin 1/1/1993 1993-2000 M.F. Steppe
Otgontenger 1/1/1992 1992-2000 Alpine Uvs Nuur Basin 1/1/1993 1993-2000 M.F. Steppe
Uvs Nuur Basin 1/1/1993 1993-2000 Alpine Sharga-Mankhan 1/1/1993 1993-2000 M.F. Steppe
Khangai nuruu 1/1/1996 1996-2000 Alpine Uvs Nuur Basin 1/1/1993 1993-2000 M.F. Steppe
Great Gobi 1/1/1975 1982-2000 Desert Gorkhi-Terelj 1/1/1995 1995-2000 M.F. Steppe
Great Gobi 1/1/1975 1982-2000 Desert Alag Khairkhan 1/1/1996 1996-2000 M.F. Steppe
Sharga-Mankhan 1/1/1993 1993-2000 Desert Burkhan Buudai 1/1/1996 1996-2000 M.F. Steppe
Gobi Gurvansaikhan 1/1/1993 1993-2000 Desert Khangai nuruu 1/1/1996 1996-2000 M.F. Steppe
Eej Khairkhan 1/1/1995 1995-2000 Desert Batkhaan 1/1/1957 1982-2000 Steppe
Erglyn Zoo 1/1/1996 1996-2000 Desert Bogdkhan Mountain 1/1/1978 1982-2000 Steppe
Small Gobi 1/1/1996 1996-2000 Desert Eastern Mongolian Steppe 1/1/1992 1992-2000 Steppe
Small Gobi 1/1/1996 1996-2000 Desert Mongol Daguur 1/1/1992 1992-2000 Steppe
Great Gobi 1/1/1975 1982-2000 D. Steppe Sharga-Mankhan 1/1/1993 1993-2000 Steppe
Sharga-Mankhan 1/1/1993 1993-2000 D. Steppe Ganga Lake 1/1/1993 1993-2000 Steppe
Uvs Nuur Basin 1/1/1993 1993-2000 D. Steppe Sharga-Mankhan 1/1/1993 1993-2000 Steppe
Uvs Nuur Basin 1/1/1993 1993-2000 D. Steppe Hustain Nuruu 1/1/1993 1993-2000 Steppe
Sharga-Mankhan 1/1/1993 1993-2000 D. Steppe Khorgo Terkh Zagaan Nuur 1/1/1965 1982-2000 Taiga
Gobi Gurvansaikhan 1/1/1993 1993-2000 D. Steppe Khovsgol Lake 1/1/1992 1992-2000 Taiga
Erglyn Zoo 1/1/1996 1996-2000 D. Steppe Khan Khentee 1/1/1992 1992-2000 Taiga
Ikh nart 1/1/1996 1996-2000 D. Steppe Gorkhi-Terelj 1/1/1995 1995-2000 Taiga
Small Gobi 1/1/1996 1996-2000 D. Steppe
Zagiin us 1/1/1996 1996-2000 D. Steppe
Khorgo Terkh Zagaan Nuur 1/1/1965 1982-2000 M.F. Steppe
Khukh Serkhyn Nuruu 1/1/1977 1982-2000 M.F. Steppe

28
Table 2 Mean change per aimag.
Aimag Mean NPP change Mean RUE change Mean relative DNPP change
Arkhangai -30.0271 -0.0015 -1.0186
Bayan-Ulgii 29.2072 -0.1908 -0.4444
Bayankhongor 9.0231 -0.2195 -2.2930
Bulgan -3.2581 0.2197 0.6145
Dornod -2.3570 0.2250 2.4493
Dornogobi -10.7844 0.1108 0.8510
Dundgobi -6.2522 -0.0201 -0.5537
Zavkhan -8.4938 0.0128 -0.6901
Gobi-Altai 4.7270 -0.2419 -1.1372
Khentii -11.7224 0.3226 3.1339
Khovd 15.1625 -0.0885 0.0539
Khuvsgul 2.2329 0.1209 -1.3943
Umnugobi 0.6552 -0.2861 -3.1386
Uvurkhangai -8.7405 0.0063 -0.5003
Selenge 11.7548 0.1172 1.0649
Sukhbaatar -6.7816 0.3384 2.0418
Tuv -18.6570 0.1047 0.3996
Uvs -1.9268 0.0583 -0.4796

Vegetation Zone
Desert
Desert Steppe
Steppe
Forest Steppe
Taiga
Alpine
Lake

Figure 1 Vegetation zones of Mongolia. Figure 2 Soil aggregation for Mongolia.

a) b)
400 900
800
350
700
300
600
250
500
NPP
NPP

200 400
150 300
100 200 y = 1.4964x + 0.8198
y = 0.6265x + 64.419
50
R2 = 0.2786 100 R2 = 0.5567
0 0
0 50 100 150 200 250 0 100 200 300 400 500
Rainfall (mm) Rainfall (mm)

c) d)
1200
900
800 1000
700
600 800
NPP

500
NPP

600
400
300 400
200 y = 1.7902x + 88.021 y = 1.3778x + 229.95
200
100 R2 = 0.7131 R2 = 0.3871
0 0
0 100 200 300 400 500 0 100 200 300 400 500 600

Rainfall (mm) Rainfall (mm)

e) f)

- 29 -
 800 580
560
700 540
520
600
500
NPP

NPP
480
500
460
400 y = -0.13x + 595.89 440 y = 0.0363x + 486.23
R2 = 0.0245 420
R2 = 0.0066
300 400
0 100 200 300 400 500 600 700 200 250 300 350 400 450 500 550 600
Rainfall (mm) Rainfall (mm)

Figure 3 The relationship between mean annual NPP (gCm-2) and rainfall for individual soil aggregates in the a)
desert (n = 1065), b) desert steppe (n = 293), c) steppe (n = 88), d) mountain (forest) steppe (n = 243), e) alpine (n =
135), f) taiga (n = 150) vegetation zones.

a) b)

576
7.5

-738
-5.5
Aimag boundary
Aimag boundary

c)

201

-96
Aimag boundary

Figure 4 a) NPP change (gCm-2yr-1) between 1982-2000. b) RUE Change (gCmm-2yr-1) between 1982-2000. c)
Relative DNPP change (gCm-2yr-1) between 1982-2000.

- 30 -
Projects and Initiatives addressing Environmental Impact Studies in Northern
Mongolia and the Lake Baikal Region

K. Frotscher & C.C. Schmullius,

Friedrich-Schiller-University Jena, Germany

Fast-growing economies and worldwide growing consumer demands have a considerable

impact on natural resources and thus on the way Earth Science data community addresses
the acquisition, storage and analyses of spatial data. Forest resources of Mongolia and the
Lake Baikal region came in the foreground since China’s exports of wood products have
been fast growing over the last decade and hence their imports of timber. Furthermore the
region is rich in many types of mineral resources which are of interest to international
investors since bullish market indices and proximity to major metal markets in China and
Japan. Under these circumstances an issue that must be addressed is the monitoring of the
region and the subsequent analyses pertaining to environmental impacts. The University of
Jena carries out several investigations using multidimensional satellite data. Reliable and
up-to-date information on land surface characteristics and changes are therefore required
by decision makers in order to fulfil several international and national treaties and for its
own policy. One of the essential components is the use of open standards as recommended
by the World Wide Web Consortium (W3C®). Open standards enable easy integration
between systems and support the easy retrieval of geospatial information in a distributed
environment.

The FAO-funded Technical Cooperation Programme Mongolia for example used the full
capacity of area-wide and mostly free available Remote Sensing data to support the
development of participatory forestry. Moreover technology consultancy and the provision
of advanced training on the use of Image Processing/GIS encouraged capacity building.
The SELENGA initiative provides a generic methodology using Space-based Earth
Observation data to overcome information gaps for evaluating processes as controlled by
the Selenga River and other large and inaccessible watersheds. These are relevant for
assessing the hydrological balance in the catchment area and understanding related
sediment mass and pollutant transfer to the Lake Baikal. The Irkutsk Regional Information
System for Environmental Protection (IRIS) will assess the current status and dynamics of
the Region’s forestry environment, influenced by man-made changes and anthropogenic
impact arising from pollution sources and other negative anthropogenic drivers located in
the region and in adjacent areas. The ‘GMES (Global Monitoring for Environment and
Security) Service Element Forest Monitoring’ provides another powerful tool for effective
forest monitoring and inventory at regional scale. Based on open standard technologies the
Siberian Earth System Science Cluster (SIB-ESS-C) will be developed as a spatial data
infrastructure for remote sensing product generation, data dissemination and scientific data
analysis.

- 31 -
 Digital Asia --- Information Network for Sustainable Future
Hiromichi Fukui
Professor, Faculty of Policy Management, Research Director, Global Security Research Institute, Keio University
Secretary General of GIS Association, Japan 5322 Endoh, Fujisawa, Kanagawa, 252-8520 Japan
Tel. +81-466-49-3497,Fax. +81-466-49-1334, E-mail : hfukui@sfc.keio.ac.jp
Abstract
Our current status of environmental issues on a global scale clearly indicates that we can
no longer consider the capacity of the natural system that provides foundation to human activities,
such as water, atmosphere, soil to be infinite, nor can we ignore the scope of human activities or
their speed of expansion as negligible. We should always consciously maintain an overall
perspective, always zooming in on various aspects while zooming right back to the wide overview
of the whole picture. We have tried to look at the concepts and technologies of the Digital Earth
that would enable this approach.Digital Earth is a virtual representation of the planet, encompassing
all its systems, and life forms, including human societies. It is designed as a multi-dimensional,
multi-scale, multi-temporal, and multi-layer information facility. The Digital Earth vision
incorporates a computerized Earth, as its interface, whereby a corresponding virtual body of
knowledge, or global encyclopedia of the real Earth and its digital representation for understanding
the oneness of the Earth and its relevant phenomena.
An overview of the scope and breadth of these innovative activities in Japan, such as the project
“Digital Asia (Digital Asia Research Centre for Strategic Design, Keio University as Academic
Frontier Project, Matching Fund Subsidy from MEXT(2004-2009))” is showcased. Digital Asia is
an initiative to provide people and communities with easy access to geo-spatial information and
events over the Internet through open sharing of GIS & Remote Sensing Data and News among all
the countries of Asia. Digital Asia will form the Digital Asia Node Network (DANN) to bring
together all participating people and agencies, and to provide a place where they can obtain useful
information for developing their applications and demonstration systems. Digital Earth technology
play key roles in economic and social sustainable development, environmental protection, disaster
mitigation, conservation of natural resources and to improve humankind’s standard of living.
Through seamless visualization of information ranging from the global to the local level and an
easily understandable overall representation of global issues, we can expect a formation of
“knowledge of the global community” based upon shared sympathy of the majority of the global
population. The potential of the Digital Earth as a media to nurture a tangible sense of being part of
one connected planet and one humankind is enormous. We have strong expectations that building
such world will further enrich us and realize concepts such as “global citizenship” and “global
society.”

Key words: GIS, Digital Earth, Media Browser, Spatial Data Infrastructures, Interoperability
biographical note

Dr. Hiromichi FUKUI has graduated from Nagoya University in 1980. He holds a Doctor of
Science in Earth Sciences from Nagoya University in 1987. He is a Professor of Faculty of Policy
Management, the Graduate School of Media and Governance and Research Director of Global
Security Research Institute of Keio University. Before joined Keio University in 1996, Prof Fukui
worked as a chief scientist in the Research Institute of Japanese Bank. He has extensive
experience in technical assistance overseas as an expert of GIS for JICA and World Bank.
His current research interests include regional planning, ecological development and global
environment issues with emphasis on spatial information sciences. He has served on the secretary
general of GIS Association and ISDE-Japan. He also served on a member of the board of directors
in Center for Environment Information Sciences, senior scientist of JAXA and on the guest
professor of Chinese Academy of Science. His specializations are in Geo-spatial Informatics and
Environmentlogy.

- 32 -
 Annual variation of aerosol optical thickness derived
from PAR observation in Mongolia
T. Takamura1, T. Karasuyama2, N. Tugjsuren3, G. Batsukh4, and H. Takenaka2

(1) Center for Environmental Remote Sensing, Chiba University; takamura@faculty.chiba-u.jp

(2) Graduate school of Science and Technology, Chiba University; Takenaka_ceres@graduate.chiba-u.jp
(3) S\chool of Materials Science, Mongolian University of Science and Technology; Tugjsurn@must.edu.mn
(4) School of Physics and Electronics, National University of Mongolia; batsukh@num.edu.mn

Abstract
The PAR(Photo-synthetic Active Radiation) has been observed in Mongolian grassland for a long
time. It is useful for not only vegetation but also atmospheric research, especially for aerosols,
because of its major part of solar radiation. In this study, a method to estimate optical thickness of
aerosol(AOT) has been developed with an error analysis and then the AOT has been derived from
the PAR data in Mongolia during a period of 1985 to 2000.
There are two remarkable features shown in the analytical results; the AOT in 1992 is
temporarily increasing possibly due to the effect of Pinatubo eruption(June 1991), and after 1997
the AOT is gradually increasing and has a clear seasonal trend compared before 1995.

Introduction
PAR(photosynthetic Active Radiation) regime is an efficient part of the solar radiation for all the
vegetation through the energy of growth. So it is one of the most important parameters for vegetation
research as well as water and has been observed in Mongolian grassland for a long time(Tugjsuren and
Takamura, 2001).
PAR ranges over about 400 nm to 700 nm where is sensitive to photosynthetic activity and also most
strongest region in the solar spectrum. So the intensity in the PAR region is strongly dependent on an
atmospheric condition, such as cloud, aerosol and so on. On the other hand, the PAR region includes no
strong absorption bands for the standard atmospheric constituents, which means the “visible atmospheric
window”. Therefore, the PAR data can reflect aerosol information under clear sky conditions.
The PAR in the Mongolian territory has been observed at the top of the building of the National
University of Mongolia in Ulaanbataar for a long time. A pyranometer and pyrheliometer with two kinds
of sharp cut-off glass filters, 380 nm and 710 nm, have been working. Then the PAR data have been
derived from the difference between data of both filters.
In this study, the aerosol optical thickness(AOT) in the Mongolian grassland is estimated from the
observed PAR data.

Methods
The PAR region has two absorption bands of ozone and water vapor, where ozone absorption called
Chappuis band extends widely from about 500 nm to 700 nm piled up with very weak water vapor
absorption bands. Water vapor is the most remarkable and effective species in the solar radiation, but it is
negligible in the PAR region so that an aerosol optical thickness can be made possible to estimate without
knowledge of water vapor content in the atmosphere. That is one of advantages on estimating AOT from
PAR data. Also, an ozone amount has weak change in time and space that the influence of ozone might be
estimated using monthly mean data from TOMS.
When aerosol is lightly loaded in the atmosphere, the extinction of the direct solar radiation can be
assumed to be single-scattered. Therefore, the direct PAR is as follows,

cos T
I 0 O u exp mW O dO ,
710
FDirectPAR
R2 ³
380
(1)

whereWO is the optical thickness of the atmosphere at wavelengthO, m the relative air-mass which is
roughly proportional to the inverse of cosine of the solar zenith angle TR the relative distance in

- 33 -
astronomical unit between the Earth and the Sun. The variable ,RO shows the input solar flux at the top of
the atmosphere and FdirectPAR the direct PAR irradiance received at the surface.
The optical thickness of the atmosphere consists of three components, air molecule scattering(WRayleigKOҏ),
aerosol extinction(WaerosolO) and gas absorption, which has several species, but in the PAR region ozone is
the most effective(WozoneO) and water vapor is negligible, as described,

WO W Rayleigh.O  W ozone.O  W aerosol .O , (2)

where WRayleigKOandҏWozone̤Oҏcan be estimated from an atmospheric pressure and TOMS data, but the
wavelength dependence of aerosols is due to their size distribution and refractive index. So in the analysis
the Angstrom’s relationship is assumed,

D
§ O ·
W aerosol .O E¨ ¸ , (3)
© 0 .5 ¹
where is the optical thickness of aerosols at 500nm.
The direct radiation of PAR can be
easily simulated using Eq.1 with an
assumption of variables Dand E in Eq.
3. In the present analysis, the variables
are assumed to be 0, 0.5, 1.0, and 1.5 as
Dand to be 0, 0.05, 0.1, 0.2, 0.3, 0.5,
0.8, 1.0, 1.5, 2.0 as E As these
simulated results are corresponding to
the observed direct PAR FdirectPAR.obs, the
most suitable value of mWcan be
estimated using a polynomial equation
derived from the calculated results.
Figure 1 shows an example of May 8,
Fig.1 A relationship between FdirectPAR and mW simulated
2000 in Ulaanbataar, Mongolia. The
using aerosol parameters, D and E, in Angstrom’s relation
variation in each calculated point is due to
shown in Eq.3.
an effect of wavelength dependence (D) of
aerosol optical thickness(AOT), so that the
approximated curve gives some errors on estimating an AOT. The pattern of the curve is a function of
solar zenith angle T.
In the figure, the effect of different aerosol types(size distribution, refractive index) becomes surely
smaller with a smaller AOT.
In the calculation, the ozone comes from monthly mean data of TOMS provided by NASA. In the
database, it lacks during a period from May 1993 to July 1996, so the monthly mean with other years is
used.

RESULTS AND DISCUSSION

The direct PAR data used in this study have been
collected at Ulaanbattar(47.9N in Latitude, 106.9E in
longitude, 1160m MSL) during a period of January
1985 to December 2000. The original data are shown in
Fig. 2, where the year 1995 has no data due to an
instrumental trouble and also a period of a latter half of
1992 to the end of 1994 has sparse data.
Figure 3 shows the annual trend of AOT(500nm)
derived from the direct PAR corresponding to Fig.2.
Fig.2 Direct PAR irradiance observed at Ulaanbattar
The red dots in Fig.3 represent monthly mean data of
during a period of 1995 to 2000. (unit:W/m2)
AOT. It should be noted that it is difficult to

- 34 -
discriminate a very thin cloud such as thin cirrus, so these aerosol data might include an effect of thin
cloud. The bias to AOT by this effect, however, should be small because observations have been
performed basically under sky-clear conditions with eye-watching. The annual trend of AOT shows clear
features as follows;
1. The AOT is temporarily increasing and
decreasing with a peak period of 1992.
The increase in this feature is due to the volcano
dust of Mt. Pinatubo, the Philippines, which erupted
on June 15, 1991. After eruption, the effect of the
volcano dust were gradually extended to the
northern hemisphere and observed in the same
period at many places(e.g., Guasta et al. 1994;
Russell et al. 1993). The aerosols coming from the
Mt. Pinatubo has been staying mainly in the upper
troposphere and the lower stratosphere with an AOT
of about 0.2(Takamura et al. 1994). The excess of Fig.3 Aerosol optical thickness(500nm) derived from
AOT is also the similar, as shown in Fig.3. the direct PAR observed at Ulaanbataar.
2. After 1997, the AOT is gradually increasing.
Figure 3 also shows a trend of increase in AOT after 1997 and a clear seasonal variation. It is curious why
the trend is changed. One of possible reasons might reflect the Mongolian economical situation, but there
is no clear evidence in the viewpoint.

Fig.4 Seasonal variation of AOT at Ulaanbataar during a period of 1985 to

2000. After 1997, it is clear that an increase in AOT is remarkable in
summer season.

Figure 4 shows a seasonal variation of AOT. Except for the effect of Mt. Pinatubo, an increase in AOT
after 1997 is distinguished in summer season, compared with the trend before 1991. Takamura et al.(1984)
showed an seasonal trend of AOT observed at Sendai, Japan. According to their results, the AOT is bigger
in summer than in winter. They have simulated an increase in AOT utilizing a growth theory of aerosol
particles dependent on relative humidity. Simultaneously this simulation can give a change of optical
properties such as complex refractive index of aerosols. They suggested that the summer aerosols grow up
with adsorption of water vapor due to higher RH because they have lighter absorptive index of refraction.
Ulaanbataar is in the dry and highland area inside the Asian continent, so it is hard to explain the seasonal
trend using an effect of relative humidity, but there is some possibility that the aerosols during these years
have changed in chemical quality of aerosols. When the optical characteristics such as single scattering
albedo or complex refractive index will be made clear, these might be also clear.

- 35 -
ERROR ESTIMATION

This method has several error sources in measurement and analysis. In this analysis, two main sources
are anticipated, one of which is due to an assumption of wavelength dependence of aerosol, and the other
due to ozone absorption.
An assumption of the wavelength dependence of Table 1 Error caused by different wavelength dependence
aerosols such as Angstrom’s relation may of AOT
introduce some uncertainty into AOT in the
analysis. However, there is no information on
FdirectPAR
2
W500.aerosol 'W500.aerosol Percentage
error(%)
(W/m )
this dependence only from the PAR observation. 318.7 0.02 0.00 0
Therefore, possible ranges of parameters and 205.7 0.34 0.02 5.9
have been assumed. The error can be estimated 100.4 0.87 0.06 6.9
with the extreme cases of parameters far from the
mean value in the assumed ranges.
We try three cases of FdirectPAP , 100.4, 205.7 and 318.7 W/m2 corresponding to Fig. 1. The simple results
are summarized in Table 1. The error due to this assumption is dependent on AOT itself and shows less
than 10 % even for the heavy aerosol loadings.
The effect of ozone has been also examined. The maximum and minimum value of monthly mean of
ozone amount during a period from Jan. 1985 to Dec. 2000 is 449DU and 289DU, respectively, and the
averaged monthly mean is 350.0DU with a standard deviation of 36.8DU. So we put an average ozone
amount with plus 50% and minus 50% for error estimation. The resultant approximation equations for
two extreme cases are almost the same with negligible errors. Therefore, the effect of ozone amount to
AOT is negligible.

SUMMARY
A PAR radiation has been observed for long time in Ulaanbataar, Mongolia. It is very useful not only for
vegetation research but also for atmospheric research of aerosols, because the spectrum region of PAR has
negligible water vapor absorption and other minor gases except for ozone. The key of this method is not
necessary to consider water vapor amount in the atmosphere.
The PAR data during a period of Jan. 1985 to Dec. 2000 are analyzed to estimate an AOT with a
correction of ozone using TOMS data. As a result, an annual trend of AOT shows interesting features, one
of which is an effect of Mt. Pinatubo's eruption and the other is an increase trend after 1997. Also in the
latter case, there is a clear seasonal variation, which shows that AOT is increasing in summer and smaller
in winter. This trend is not shown before 1994 and may reflect a change of aerosol situation in Mongolia.
It suggests that the air quality should be examined by many kinds of way such as physical and chemical
methods.
References
1. Yamauchi, Toyotaro, 1995: Statistical Analysis of Atmospheric Turbidity over Japan: The
Influence of Three Volcanic Eruptions. J. Meteor. Soc. Jpn., 73, 91-103.
2. Guasta, M. Del, M. Morandi, L.Stefanutti, B.Stein, and J.P.Wolf, 1994: Derivation of Mount
Pinatubo Stratospheric Aerosol Mean Size Distribution by Means of a Multiwavelength Lidar.
Appl. Opt., 33, 5690-5697.
3. Russell, P.B., J.M. Livingston, E.G. Dutton, R.F. Pueschel, J.A. Reagan, T.E. Defoor, M.A. Box,
D. Allen, P. Pilewskie, B.M. Herman, S.A. Kinne, and D.J. Hofmann, 1993: Pinatubo and Pre-
Pinatubo Optical-Depth spectra: Mauna Loa Measurements, Comparisons, Inferred Particle Size
Distributions, Radiative Effects, and Relationship to Lidar data. J. Geophys. Res., 98, 22969-
22985.
4. Takamura, T., Y. Sasano, and T. Hayasaka, 1994: Tropospheric aerosol optical properties derived
from lidar, sun photometer and optical particle counter measurements. Appl. Opt., 33(30), 7132-
7140.
5. .Takamura, T., M.Tanaka, T.Nakajima,1984: Effects of Atmospheric Humidity on the Refractive
Index and the Size Distribution of Aerosols as Estimated from Light Scattering Measurements. J.
Meteor. Soc. Japan, 62(4), 573-582.
6. Tugjsuren, Nas-Urt, and Tamio TAKAMURA, 2001: Investigation for Photosynthetically Active
Radiation Regime in the Mongolian Grain Farm Region, J. Agric. Meteorol., 57(4), 201-207.
7. TOMS ozone data: http://toms.gsfc.nasa.gov/ozone/ozoneother.html

- 36 -
 Determination of the Photosynthetically Active Radiation for Vegetation
Growth Period of the Mongolian Grain Farm Region
By Tugjsuren Nasurt

Mongolian University of Science and Technology,Ulaanbaatar, Mongolia

E-mail:tugjsurn@must.edu.mn

AbstractReality, worldwide routine network for the measurement of photosyenthetically active

radiation (PAR) is not yet and PAR is often calculated as a nearly constant ratio of the broadband
solar radiation. This study focuses on the PAR to crop vegetation period of Mongolian grain farm
region and presents and discusses some specific aspects related to the wheat phenological phases.

Photosynthetically active radiation (PAR) contributes significantly in comprehensive studies of

radiation climate, remote sensing of vegetation, radiation regimes of plant canopy and
photosynthesis. The PAR radiation covering both photon and energy terms lies between 400 and
700 nm or 380–700 nm in the solar spectrum . However, today's more commonly accepted spectral
interval 400–700 nm does not cause misunderstanding.

1. Introduction

Photosynthetically active radiation (PAR) contributes significantly in comprehensive

studies of radiation climate, remote sensing of vegetation, radiation regimes of plant canopy and
photosynthesis. The PAR radiation covering both photon and energy terms lies between 400 and
700 nm or 380–700 nm in the solar spectrum . However, today's more commonly accepted
spectral interval 400–700 nm does not cause misunderstanding. Most published experimental
results use measured values of global PAR (QP) and global solar radiation (Q) to determining the
PAR fraction of the broadband solar radiation. In the literature, there exist three distinct
measuring techniques for determining PAR: (i) spectrally, by integrating spectral irradiance
distribution measurements over the waveband 400–700 nm; (ii) indirectly, through combined
filtered data; and (iii) directly, via spectral PAR measurements (400–700 nm) by means of
quantum sensor (QP). Published values for the PAR fraction of global irradiance are around 0.45
or 2 E J-1 for photon efficiency. Nevertheless, the range of the PAR fraction suggests the
desirability for recalibration accounting for local climatic differences. Thus, the present analysis
aims to quantify temporal variations of the PAR efficiency for various atmospheric conditions at
Ugtaal, Mongolia (48°3'N, 105°25'E, 1160 m above the sea level), and to identify reasons for such
variations. Fraction of photosynthetically active radiation is defined as the fraction of
photosynthetically active radiation absorbed by a plant canopy in vegetation period. It excludes
the fraction of incident PAR reflected from the canopy and the fraction absorbed by the soil
surface or the combination of forest floor and understory, but includes the portion of PAR which is
reflected by the soil and absorbed by the canopy on the way back to space.

2. Measurement and data processing

In this study, we used investigative data at the Ugtaal (48°3'N, 105°25'E, 1160 m above the sea
level), which is located in the Mongolian grain farm region. Data were collected at the Ugtaal for every
hour in the daytime during 1986-1996. In order to measure the PAR received at the Earth’s surface, a
thermo-electric actinometer (AT-50) was mounted. This instrument has two filters (BS-8, KS-19), which
can selectively pass only PAR. The direct solar radiation in the full wavelength and the direct PAR can be
obtained by the filter operation, manually. We simultaneously measured global and diffuse radiation using
a thermo-electric pyronometer (M-80M) with a black and white type sensor. With this instrument, the
global and diffuse radiation can be observed with and without a shadow disk for direct solar flux.
Measurements of photosynthetically active radiation at Ugtaal, combined with simultaneous
meteorological parameters such as soil and atmospheric temperature, soil and atmospheric humidity, wind

- 37 -
 speed and its direction, cloud cover or sunshine duration.We were determined QP at Ugtaal for every
phenological phase of wheat ( wheat sort -Buriad-34 ) development during 11 years (1986-1996).

3. Results and discussion

Global solar radiation is one of the most essential factors for vegetation. Results of the
investigation show that a mean value of integrated solar radiation over all wavelengths at Ugtaal for 11
years, is 4.4-4.7 GJ/m 2 for the whole year, and 2.3-2.7GJ/(m 2 yr) for the vegetation period. The study of
PAR in Mongolia shows that the average ratio of yearly global solar radiation to global PAR in
Ulaanbaatar and Ugtaal are 0.469 and 0.472, respectively (N.Tugjsuren 1996, 2004). The ratio (QP/Q)
exhibits seasonal dependence, high values in spring-summer (0.48-0.47) and lower and more variable
values (0.44) in winter. The higher proportion of PAR irradiance occurred in May (0.49), while the
respective lower proportions occurred in December and January (0.436). The analysis of hourly values
also reveals significant diurnal variation of the ratio during daylight hours. The sky clearness and
brightness indices and path length caused substantial changes in the PAR fraction. The ratio of PAR
dependence on vegetation period is included in table 1.

A number of researchers have proposed various strategies that need to be adopted in order to
develop sustainable food production systems in the country and the same time reduce the vulnerability of
small farmers to crop failures. The search for drought-adopted species of short cycle, well adapted to the
arid and semi-arid environments is also at the top of the research agenda of Mongolian agricultural
researchers. The plants’ phenological development and water and PAR requirements are important
attributes that need to be considered when evaluating their suitability for a given climatic environment.
Besides phenology and crop water requirements, we need to be tested on eco-physiological properties such
as responses to high temperature and water stress. The response of annual crops to water stress depends
upon the developmental stage of the plants at the time water becomes limiting a factor.
The analyses of hourly or daily measurements of PAR in period May 1986 to October 2006 for
grain farm region of Mongolia, are briefly summarized as follows (Fig.1): The mean values of QPAR and
phase duration of derivation and tillering stages of Buryad-34 wheat has been found 93.6 MJ/m 2 and 123.9
MJ/m 2 and 9 and 13 days, respectively. Then, QPAR and phase duration for stem production, heading and
flowering stages are 138.7 MJ/m 2 , 119.8 MJ/m 2 and 83.7 MJ/m 2 and 15, 15, 11 days,
respectively. The mean values of QPAR and phase duration of this sort of wheat for milk, wax and full
maturity has been found 101.0, 87.2 and 96.0 MJ/m 2 and 14, 12 and 14 days, respectively. It should be
pointed out here that the differences in days to maturity between were 34-53 days.

Table.1.
productio
Derivatio

Floweing

maturity

maturity

maturity
Tillering

Heading
Yiuld

Stem

Milk

Wax

Full
n

Y T Qp T Qp T Qp T Qp T Qp T Qp T Qp T Qp
1986 21.1 7 100.9 10 95.2 14 133.4 20 104.6 12 114 14 88.1 9 78.6 18 146.4
1987 13.5 11 106.9 10 97.9 13 117.3 13 117.3 4 31.4 16 125.6 12 146.7 6 40.7
1988 7.3 8 78.3 14 138.8 20 194.7 13 124.2 9 82.6 10 87.1 8 69.7 10 67.7
1989 6.9 8 69.4 14 138.7 18 111.8 14 118.9 6 48.5 14 110.2 16 88.1 20 125.5
1990 6 12 113.2 16 135.6 17 143.8 16 118.6 14 94.9 17 113.2 16 79 20 79
1991 4 6 61 17 161 16 132.7 14 119.6 8 62.2 13 97.8 19 112.2 13 87.6
1992 14.4 9 106.9 11 135 14 182.2 16 126.9 14 104.2 14 69.6 9 99.4 14 90.4
1993 14.6 7 100.7 13 109.4 14 114.8 14 129.4 14 118.6 15 113.2 10 32.2 14 169.5
1994 10 11 99 14 112 14 138 13 118.9 13 118.9 14 101 13 84 12 98.8
1995 5 12 100 14 115 14 118 15 120 14 91 14 95 12 82 12 95.5
10.3 9 93.6 13 123.9 15 138.7 15 119.8 11 83.7 14 101 12 87.2 14 96

- 38 -
 QPAR and dermination QPAR and tillering period

120 200
100
QPAR (MJ/m2)

QPAR(MJ/m2)
150
80 Yield Yield
60 Veg. period 100 Veg.period
40 QPAR QPAR
50
20
0 0
1985 1990 1995 2000 1985 1990 1995 2000
YEAR YEAR

QPAR and steam production QPAR and Heading

250 140
120
200
QPAR(MJ/m2)

QPAR(MJ/m2)
100
150 Yield Yield
80
Veg.period Veg.period
100 60
QPAR QPAR
40
50
20
0 0
1985 1990 1995 2000 1985 1990 1995 2000
YEAR YEAR

QPAR and flowering QPAR and milk maturity

140 140
120 120
QPAR (MJ/m2)

QPAR(MJ/m2)

100 100
Yield Yield
80 80
Veg. period Veg. period
60 60
QPAR QPAR
40 40
20 20
0 0
1985 1990 1995 2000 1985 1990 1995 2000
YEAR YEAR

QPAR and wax maturity QPAR and full maturity

160 200
140
QPAR (MJ/m2)

QPAR (MJ/m2)

120 150
100 Yield Yield
80 Veg. period 100 Veg. period
60 QPAR QPAR
40 50
20
0 0
1985 1990 1995 2000 1985 1990 1995 2000
YEAR YEAR

Figure1. Photosynthetically active radiation for vegetation period of Ugtaal, Mongolia

- 39 -
 CONCLUTION

1. The mean values of QPAR and phase duration of derivation and tillering stages of
Buryad-34 wheat has been found 93.6 MJ/m 2 and 123.9 MJ/m 2 and 9 and 13 days,
respectively.
2. QPAR and phase duration for stem production, heading and flowering stages are 138.7
MJ/m 2 , 119.8 MJ/m 2 and 83.7 MJ/m 2 and 15, 15, 11 days, respectively.
3. The mean values of QPAR and phase duration of this sort of wheat for milk, wax and full
maturity has been found 101.0, 87.2 and 96.0 MJ/m 2 and 14, 12 and 14 days, respectively.
It should be pointed out here that the differences in days to maturity between were 34-53
days.
4. The agricultural advisory service will in future play an extremely important role in
bringing about the necessary modification of the cropping systems.
5. We need to investigate the possibilities to use of satellite remote sensing techniques for
crop monitoring and yield forecasting. The satellite data are integrated with agro-
meteorological models to improve crop monitoring and yield forecasting.

References

1. G.Papaioannou, G.Nikolidakis, D.Asimakopoulos, D.Retalis, 1996: Photosynthetically

active radiation in Athens, Agric.For.Meteorol. 81(1996), pp287-298
2. D.Rijks, J.M.Terres, P.Vossen, 1998: Agrometeorogical applications for regional crop
monitoring and production assessment, Joint Research Centre, European Commission.
3. Tugjsuren N., Takamura T, 2001: Investigation for photosynthetically active radiation regime
in the Mongolian grain farm region, J. Agric. Meteorology, 57(4), 201-207.
4. Tugjsuren N, 1996: Investigation of solar radiation regime of dry and cool zone crop-growing
region of Mongolia, Science and Technology Report, Science and Technology Information
Center, Ulaanbaatar, Mongolia 75 pp, (In Mongolian).
5. Tugjsuren N, 2003: Latitudinal distribution of solar radiation under clear and cloudy
conditions on the territory of Mongolia, Proceedings of the CEReS International Symposium
on Remote Sensing , ‘Monitoring of Environmental Change in Asia’, Desember 16-17,
CEReS, Chiba University, Japan, pp 129-133
6. Tugjsuren N, 2003: Aerosol pollution of the atmosphere and its sources features in Mongolia,
Proceedings of the CEReS International Symposium on Remote Sensing , ‘Monitoring of
Environmental Change in Asia’, Desember 16-17, CEReS, Chiba University, Japan, pp 153-
159.
7. Tugjsuren N, 2004: Investigation of photosynthetically active radiation in Mongolia,
Proceedings of First International Workshop on Land Cover Study of Mongolia using Remote
Sensing /GIS, 8-10 June 2004, p 45-55

- 40 -
 Coherent time in cloud analysis using 95GHz FM-CW cloud profiling radar
Y. Nakanishi(1), T. Takano(1), K. Akita(1), H. Kubo(1), Y. Kawamura(3), H.
Kumagai(4), T. Takamura(1,5) And T. Nakajima(6)

(1) Graduate School of Science and Technology, Chiba University

(2) Center for Frontier Electronics and Photonics, Chiba University
(3) Faculty of Engineering, Chiba University
(4) National Institute of Information and Communications Technology
(5) Center for Environmental Remote Sensing, Chiba University
(6) Center for Climate System Research, The University of Tokyo

Abstract
A low-power and high-sensitivity cloud profiling radar with a frequency-modulated
continuous wave (FM-CW) of 95 GHz has been developed in order to survey an internal
cloud structure by ground-based observations. Millimeter wavelength at 95 GHz is more
suitable for cloud profiling than a traditional one with 35 GHz because of its higher
sensitivity for cloud particles. An FM-CW radar has several advantages to a pulse radar,
such as a lower output of power, easy operation, low cost and so on. The radar
developed with a 500mW output power has almost the same performance as the pulse
radar with 1.6 kW output power of the National Institute of Information and
Communications Technology. Doppler function is also applicable for a vertical motion
of cloud particles. 
In the previous study, Doppler sensitivity has not reached to the expected level even
if the signal sensitivity has been raised theoretically by a signal averaging technique.
The reason might be turbulence inside the cloud, which means that coherency reflects
the droplet motion of cloud. Therefore, coherent time should be considered when
averaging reflected signal for an improvement of Doppler sensitivity. A dependency of
length of the coherent time on internal motion of cloud has been discussed in the present
study. 


1. Introduction

Cloud in the atmosphere plays an important role for the Earth climate through radiation budget. As
described in the ISCCP report, however, cloud behaviors in each GCM gives some different aspects due to
unknown cloud processes in formation. These are key issues for climate studies.

We have many tools for cloud research, such as satellite- and ground-based instruments. At the ground,
cloud observation is so limited in space but can be performed by many methods. A radar system is a
powerful tool in order to make internal structure of cloud clear. Traditional radar has a frequency of
35GHz, which is very sensitive to rain drops and bigger ice crystals, but insensitive to cloud particles,
especially water clouds. In these cases, new radar system with higher frequency of 95GHz has been
developed.

FM-CW radar is one of the most powerful tools for cloud research, which has continuously transmitted
and frequency-modulated wave of 94GHz. Table 1 Specific features of RADAR
The power emitted from an antenna can be Type pulse Frequency-modulated
reflected by cloud and then received by Peak power High (1600W) Low (0.5W)
another antenna with a Doppler effect. The Instrumental Large Small(solid state
cloud features such as cloud density and size made)
particle speed can be derived from the
frequency shifts caused by its traveling time Except antenna
and Doppler shift. FM-CW radar has a big
Portability Hard(Heavy) Easy(relatively light)
advantage rather than traditional pulse radar
Cost Relatively high Relatively low
because of its lower power, smaller size
with semiconductor-components and lower cost performance. These features can make it possible to have
observation at many sites.

- 41 -
2. Basic concept of FM-CW radar and its signal analysis
Basic concept of FM-CW in signal processing radar is different from pulse radar. It requires
conversion of time-frequency domain to distance from the emitted source, while pulse radar is simple
conversion of reflected time to distance of target. Therefore, its signal analysis is a little more
complicated for FM-CW radar signal than that for pulse radar.
The principle and a photograph of an FM-CW radar are shown in Figs.1 and 2. The signal frequency is
modulated in the range of
f0 ҙҡҟҏF. Transmitted signal from one of the antennas is reflected by cloud particles, returns, and is received
by the other antenna with a delay time of t relative to the original transmitted signal. Mixing the
transmitted and received frequencies, beat frequencies fb are observed in the spectra, which are caused by
ensemble of clouds particles:

fb = 4 F r / (c Tm) (1),

where r is the height of the clouds,

Tm is the modulation interval, and c is the
light velocity. When theobjects move in the
line of sight, the frequencies of reflected
signals change by fd :
fd = -2 ( f0 / c ) ( dr / dt ) (2).
Spectra of the obtained beat frequency, fb ,
corresponds to the cloud ranging
profiles.

Fig.1. Principle of a FM-

CW Radar

3. Observation of clouds
Using the developed millimeter-wave FM-CW radar at 95 GHz, we observed clouds on Mirai, a
Japanese scientific research vessel shown in Figs.3 and 4, in the Arctic Ocean and southwest Pacific Ocean
in 2004 and in the northern half of the Pacific Ocean in 2005 (Fig.5). Fig.6 shows an example of observed
cloud profiles in the Arctic Ocean on September 3 to 6, 2004 with our FM-CW radar at 95 GHz. Thin
clouds at 5-10km height were detected as well as low-height clouds and precipitation as shown in Fig.6.

- 42 -
Fig.2 Developped FM-CW radar at 95GHz Fig.3 Japanese research vessel Mirai

Fig.4 Container of the FM-CW radar on Mirai Fig.5 Cruise track of Mirai in 2004 and 2005

4. Coherency of signal from clouds and simulation of coherent integration

Received signal of the FM-CW radar would be an ensemble of scattered waves from cloud particles.
The modulation interval Tm is 1 msec and single frequency spectrum corresponding to a cloud profile is
obtained each 1 msec. There are two ways to integrate signals in random noise for improving S/N ratio:
coherent integration and incoherent integration. If the signal from clouds is coherent in a longer period
than 1 msec, coherent integration is more effective to improve S/N ratio by factor 2 than incoherent
integration. If the coherent time of scattered wave from clouds is, however, shorter or comparable to 1
msec, incoherent integration should be done to avoid reduction of S/N ratio.
Fig.7 shows a result of computer simulation of improvement in S/N ratio of observed signal. The
input signal is 100 times smaller in amplitude than a random white noise Figs.7(a),(b). Assuming that
the coherency of the signal is maintained in the period, coherent integration shown in Fig.7(c) present
much better S/N ratio than the case of incoherent integration shown in Fig.7(d).

5. Conclusion
The result of simulation shows that coherent integration is fairly effective if the signal is coherent in
the period of integration. In order to realize coherent integration, however, it is very important to know
how long coherency is maintained in clouds and precipitations. Investigation of coherent time in various
clouds would be interesting on the standpoint of microphysics of clouds. We, therefore, will measure
dependence of S/N ratio on coherent integration period in the near future.

References
1. Toshiaki TAKANO, Ken-ichi AKITA, Hiroshi KUBO,  Youhei KAWAMURA, Hiroshi
KUMAGAI , Tamio TAKAMURA, Yuji NAKANISHI, and Teruyuki NAKAJIMA,
“Observations of Clouds with the Newly Developed Cloud Profiling FM-CW Radar at 95GHz”,
The International Society of Optical Engineering, Symposium on Remote Sensing, Bruges
(Belgium), Sept.19, Vol.5979, No.07, 2005.

- 43 -
 20

15

10

0
00:00 00:00 00:00 00:00 24:00

Tim e[UT ]

Fig.6 Example of data obtained in the Arctic Ocean using the millimeter-wave FM-CW radar.
䎰䏄 䏊 䏑䏌 䏗䏘 䏇 䏈 䎰 䏄 䏊 䏑䏌 䏗 䏘 䏇䏈
䎚䎙䎑䎓䎛 䎚䎙䎑䎓䎛
䎚䎕䎑䎘䎓 䎚䎕䎑䎘䎓
䎚䎓䎑䎓䎓 䎚䎓䎑䎓䎓
䎙䎚䎑䎘䎓 䎙䎚䎑䎘䎓
䎙䎘䎑䎓䎓
䎙䎘䎑䎓䎓
䎙䎕䎑䎘䎓
䎙䎕䎑䎘䎓
䎙䎓䎑䎓䎓
䎙䎓䎑䎓䎓
䎘䎚䎑䎘䎓
䎘䎚䎑䎘䎓
䎘䎘䎑䎓䎓
䎘䎘䎑䎓䎓
䎘䎕䎑䎘䎓
䎘䎓䎑䎓䎓
䎘䎕䎑䎘䎓
䎗䎚䎑䎘䎓 䎘䎓䎑䎓䎓
䎗䎘䎑䎓䎓 䎗䎚䎑䎘䎓
䎗䎕䎑䎘䎓 䎗䎘䎑䎓䎓
䎗䎓䎑䎓䎓 䎗䎕䎑䎘䎓
䎖䎚䎑䎘䎓 䎗䎓䎑䎓䎓
䎖䎘䎑䎓䎓 䎖䎚䎑䎘䎓
䎖䎕䎑䎘䎓 䎖䎘䎑䎓䎓
䎖䎓䎑䎓䎓 䎖䎕䎑䎘䎓
䎕䎚䎑䎘䎓 䎖䎓䎑䎓䎓
䎕䎘䎑䎓䎓 䎕䎚䎑䎘䎓
䎕䎕䎑䎘䎓 䎕䎘䎑䎓䎓
䎕䎓䎑䎓䎓
䎕䎕䎑䎘䎓
䎔䎚䎑䎘䎓
䎕䎓䎑䎓䎓
䎔䎘䎑䎓䎓
䎔䎚䎑䎘䎓
䎔䎕䎑䎘䎓
䎔䎘䎑䎓䎓
䎔䎓䎑䎓䎓
䎔䎕䎑䎘䎓
䎚䎑䎘䎓
䎘䎑䎓䎓 䎔䎓䎑䎓䎓
䎕䎑䎘䎓 䎚䎑䎘䎓
䎓䎑䎓䎓 䎘䎑䎓䎓

(a) single spectrum of a signal

䎓䎑䎓 䎘䎑䎓 䎔䎓䎑䎓 䎔䎘䎑䎓 䎕䎓䎑䎓 䎕䎘䎑䎓 䎖䎓䎑䎓 䎖䎘䎑䎓 䎗䎓䎑䎓 䎗䎘䎑䎓 䎘䎓䎑䎓 䎘䎘䎑䎓 䎙䎓䎑䎓 䎙䎘䎑䎓 䎚䎓䎑䎓 䎚䎘䎑䎓 䎛䎓䎑䎓 䎛䎘䎑䎓 䎜䎓䎑䎓 䎜䎘䎑䎓 䎔䎓䎕䎑䎗

(b) spectrum of random white noise whose

䎕䎑䎘䎓
䎓䎑䎓䎓
䎓䎑䎓 䎘䎑䎓 䎔䎓䎑䎓 䎔䎘䎑䎓 䎕䎓䎑䎓 䎕䎘䎑䎓 䎖䎓䎑䎓 䎖䎘䎑䎓 䎗䎓䎑䎓 䎗䎘䎑䎓 䎘䎓䎑䎓 䎘䎘䎑䎓 䎙䎓䎑䎓 䎙䎘䎑䎓 䎚䎓䎑䎓 䎚䎘䎑䎓 䎛䎓䎑䎓 䎛䎘䎑䎓 䎜䎓䎑䎓 䎜䎘䎑䎓 䎔䎓䎕䎑䎗

amplitude is 100 times larger than the signal.

䎰 䏄 䏊 䏑䏌 䏗䏘䏇䏈 䎕䎓䎑䎓䎓 䎰 䏄 䏊 䏑䏌 䏗䏘䏇䏈
䎕䎓䎑䎓䎓

䎔䎜䎑䎓䎓 䎔䎜䎑䎓䎓

䎔䎛䎑䎓䎓 䎔䎛䎑䎓䎓

䎔䎚䎑䎓䎓 䎔䎚䎑䎓䎓

䎔䎙䎑䎓䎓 䎔䎙䎑䎓䎓

䎔䎘䎑䎓䎓
䎔䎘䎑䎓䎓
䎔䎗䎑䎓䎓
䎔䎗䎑䎓䎓
䎔䎖䎑䎓䎓
䎔䎖䎑䎓䎓
䎔䎕䎑䎓䎓
䎔䎕䎑䎓䎓
䎔䎔䎑䎓䎓
䎔䎔䎑䎓䎓
䎔䎓䎑䎓䎓
䎔䎓䎑䎓䎓
䎜䎑䎓䎓
䎜䎑䎓䎓
䎛䎑䎓䎓
䎛䎑䎓䎓
䎚䎑䎓䎓
䎚䎑䎓䎓
䎙䎑䎓䎓
䎙䎑䎓䎓
䎘䎑䎓䎓
䎘䎑䎓䎓
䎗䎑䎓䎓
䎗䎑䎓䎓
䎖䎑䎓䎓
䎖䎑䎓䎓
䎕䎑䎓䎓
䎕䎑䎓䎓
䎔䎑䎓䎓

(c) 50 times coherent integration of the

䎔䎑䎓䎓

䎓䎑䎓䎓
䎓䎑䎓 䎘䎑䎓 䎔䎓䎑䎓
(d) 50 times incoherent integration of the
䎔䎘䎑䎓 䎕䎓䎑䎓 䎕䎘䎑䎓 䎖䎓䎑䎓 䎖䎘䎑䎓 䎗䎓䎑䎓 䎗䎘䎑䎓 䎘䎓䎑䎓 䎘䎘䎑䎓 䎙䎓䎑䎓 䎙䎘䎑䎓 䎚䎓䎑䎓 䎚䎘䎑䎓 䎛䎓䎑䎓 䎛䎘䎑䎓 䎜䎓䎑䎓 䎜䎘䎑䎓 䎔䎓䎕䎑䎗
䎓䎑䎓䎓
䎓䎑䎓 䎘䎑䎓 䎔䎓䎑䎓 䎔䎘䎑䎓 䎕䎓䎑䎓 䎕䎘䎑䎓 䎖䎓䎑䎓 䎖䎘䎑䎓 䎗䎓䎑䎓 䎗䎘䎑䎓 䎘䎓䎑䎓 䎘䎘䎑䎓 䎙䎓䎑䎓 䎙䎘䎑䎓 䎚䎓䎑䎓 䎚䎘䎑䎓 䎛䎓䎑䎓 䎛䎘䎑䎓 䎜䎓䎑䎓 䎜䎘䎑䎓 䎔䎓䎕䎑䎗

signal and noise. The signal appears signal and noise. The signal cannot be
in the spectrum above noise. seen in the spectrum.
Fig.7 Simulation of coherent integration of received data.

- 44 -
 LAND COVER MAPPING OF MONGOLIA

Sh. Munkhtuya
GIS Specialist/ IT
Asia Gold Mongolia LLC
Email: munkhtuya@asiagold.com

Abstract:

The goal of this research is to develop a land cover classification system for Mongolia, a method for
integrated information processing techniques which would produce a land cover map for multi-class area,
that meets international standards using Landsat thematic mapper information.

To achieve this goal the following specific objectives were set:

1. To study Mongolian and International land cover classification system and to evaluate it,
2. To study the possibility of a land cover classification system using spectral parameters measured
by satellite sensors,
3. To create a database incorporating satellitte imagery, thematic and topographic maps, field
measurement and linking them by spatial and temporal scales and by contents,
4. To develop a methodology and information processing technique for digital classification and land
cover mapping,
5. To develop a methodology for assessment of land cover change,
6. To produce a multi-class land cover map at a scale of 1:50,000 for a selected area.

This research goal and objectives have been achieved succesfully. The developed system and methodology
have been tested for 11 different scenes of Landsat satellite. 1:50,000 scale land cover maps have been
produced for 4.8 mln hectare of main crop area of Mongolia.
This research study was done at the Information and Computer Center of the National Agency for
Meteorology, Hydrology and Environmental Monitoring (NAMHEM) of the Ministry of Nature and
Environment (MNE) of Mongolia, ERDAS IMAGINE image processing system and ArcINFO geographic
information system based on an UNIX operating platform.

Practical value of this research: The developed system is a theoretical and methdological base for real
time land cover mapping of Mongolia using Landsat TM data. It would be used for 1:50,000 scale land
cover mapping for the remaining area of Mongolia. Land cover information is useful in practical
implementation of new system for land monitoirng and management.

Also it can be used for thematic mapping, such as vegetation, soil, forest, water and land use, for
collecting of reference information for modeling of ecosystem.
Study Area
The study areas for land cover mapping were selected based on different landscape types of
Mongolia. It includes different ranges of land cover, such as Khangai mountains, Taiga upland, Gobi and
Steppe area of Mongolia, which covered 12 Landsat scenes (Figure 1.1).

- 45 -
Data Used

RS Data:
x 12 scenes of Landsat ETM from 1989 to 2000
GIS Data:
1. Main geographic features like administrative boundaries, lakes, roads, rivers, and relief (scale
1:500,000)
2. Land use map of Selenge aimag (scale 1:500,000)
3. Ecosystem map of Selenge region (scale 1:500,000)
4. Fire map, which produced from NOAA AVHRR data of 2000
5. Parcel map (scale 1:100,000) (Figure 1.5)
6. Desertification and degradation map (scale 1:1,000,000)
Statistical Data:
- Local agronomist data (Figure 1.6)
- Parcel data from LMA, MAF and National Census

2. CLASSIFICATION APPROACH PROPOSED FOR LAND COVER MAPPING

Concerning the topic land cover mapping initial analysis came to the result that a part of the mapping
carried out by supervised classification. Specific classes and issues had to be mapped/refined by a
following interpretation of the data. Based on the available data (in a first step data of 1989 were used) a
general work approach has been worked out which is replicable for 2000 data sets, re-using training areas,
adapted techniques and applications.

The main working steps are as follows:

x Geocoding of satellite data
x Preparation of material for ground survey and execution (carried out for Selenge)
x Develop land cover classification and interpretation keys
x Supervised classification using existing information and ground survey results
x Post-processing and filtering of systematic errors
x Interpretation and refinement of classification results
x Generalize land cover classes into big classes
x Generate land cover change map

3. RESULTS

3.1 Land cover classification system

Depending on the data used in this research work, the second level land cover classes were classified. 7
land cover classes and 38 subclasses were developed over Mongolia by comparing international and
Mongolian previous land cover classification systems.

Mongolian land cover classification system

1. Grassland 5. Wetland
1.1 Mountain tundra 5.1 Swamp
1.2 Mountain meadow steppe 5.2 Clay and solonchacks
1.3 Mountain steppe 5.3 Bulrush
1.4 Valley steppe 6. Water body
1.5 Plain steppe 6.1 Salt lake
1.6 Burnt steppe 6.2 Fresh lake
1.7 River valley meadow 6.3 Pond
1.8 Sparce vegetated land 6.4 River
2. Cultivated land 6.5 Water basin
2.1 Agricultural land 7. Settlement
2.2 Fallow 7.1 Urban area

- 46 -
 2.3 Abandonned land 7.2 Road, communication lines
3. Forest 7.3 Constructions out side from city
3.1 Coniferous forest
3.2 Decidous forest
3.3 Sparce forest
3.4 Shrub land
3.5 Mixed forest
3.6 Burnt forest
3.7 Diseased forest
4. Bare land
4.1 Rocks
4.2 Glacier
4.3 Dried land
4.4 Desert
4.5 Sand
4.6 Developing sand
4.7 Stones
4.8 Dry valley
4.9 Mining

3.2 Land cover maps

- 47 -
4. CONCLUSION
1. Development of land cover classification system and its implementation technology is theoretical
base of land cover mapping.
2. Developed system, methodology and technology can be used for land cover classificaton of other
areas of Mongolia in a 1:50,000 scale using Landsat imagery.
3. Land cover information is useful for generating basic information of thematic mapping, such as
grassland, soil, forest, water and land use, bio-physics, and ecosystem modeling in real situation,
by a large area and in a short time.
4. It can be used for land cover suitability assessment, land cover database generation.
5. Spectral values of land cover types, such as forest, bare land, meadow, water and wetland give
high contrast from each other in the Landsat image bands of 4, 5, 7 but spectral values of grassland
cover are more complex.
6. Land cover monitoring study has been started due to the land cover classification having been
done in a certain time interval and to determine its change detection.

Land cover classification for the whole territory of Mongolia is very useful information for other
thematic study and to implement this goal we need to process a total of 110 Landsat scenes to cover the
entire area of Mongolia.

- 48 -
 Development of a Better Atmosphere and Soil Resistant Vegetation Index for
Forestry Monitoring in Taiwan

G. Dashnyam1 *, G. R. Liu2, C. K. Liang1, T. H. Kuo2 C. W. Lan1, T. H. Lin2, Y. C. Chen1

1
Institute of Atmospheric Physics, National CentralUniversity, Chung-Li, Taiwan 320
2
Center for Space and Remote Sensing Research,National Central University, Chung-Li,
Taiwan 320
*corresponding author: gerelmaa@atm.ncu.edu.tw

Forests are considered as an important sink in capturing atmospheric CO2. The Tokyo Protocol
places special emphasis on how much forest areas a country owns. With the rising concern over
global warming, the monitoring and preservation of forest areas have become an increasingly
crucial task for the entire human race. The main purpose of this study is to develop a more accurate
forest area assessment via satellite data. It is well known that the effects from atmospheric radiation
can degrade satellite image quality, rendering inaccurate interpretation of the data. For example,
although the NDVI index can provide surface vegetation information, the atmospheric effect can
cause significant errors. With the launch of the more advanced MODIS sensor, which is equipped
with more spectral bands, more accurate vegetation indices have been devised. In this paper, the
EVI, AFRI, and SARVI vegetation indices have been chosen to compare their atmosphere and soil
resistant capabilities in finding the most suitable index for forestry monitoring in Taiwan.
Keywords: Forest, Satellite, MODIS, Vegetation index

1. Introduction
Vegetation is considered one of the most important components of the ecosystems. It influences the
planet’s energy balance, climate, hydrological and bio-geochemical cycles, and so forth. Knowledge about
the vegetation species, distribution patterns, and changes in the phonological cycle can give relevant
information about the climatic and geological characteristics of an area. Therefore, understanding in the
vegetation distribution and its changes is very crucial. This is where satellite remote sensing plays a vital
role in its ability to generate spectral indices of the vegetation indicator from their multiple spectral bands.
The main objective of the study is to estimate various vegetation indices over the Taiwan region via
satellite data, where an optimized vegetation index is chosen.

2. Vegetation Index (VI)

Several vegetation indexes----- Normalized Difference Vegetation Index (NDVI), Enhanced
Vegetation Index (EVI), Soil and Atmosphere Resistant Vegetation Index (SARVI), Aerosol Free
Vegetation Index (AFRI), and the Atmospherically Resistant Vegetation Index (ARVI)-----were
investigated in this study. Among these indexes, the NDVI is the most well-known indicator for vegetation
mapping. However, it has already been in use for decades. Currently, there are several other remotely
sensed channels that can be utilized in reducing the atmospheric influence and soil effects. For example,
the EVI adopts the blue ray spectrum, which can better correct the atmospheric haze, while the SARVI has
been clamed for its capability to correct both the soil and atmospheric noise. In addition, the AFRI
vegetation index owns the capability to resist the existence of man made smoke (Rouse et al., 1973; French
et al., 1997; Kaufman and Tanre, 1992; Kaufman et al., 1997; Santer, 1999).

3. Data Description
Data from the MODerate resolution Imaging Spectroradiometer (MODIS) onboard the Terra and
Aqua satellites are mainly employed in the study. The satellites scan the Earth twice a day. The first scan
occurs in the morning, and the second scan is conducted in the afternoon. The same location in scanned
every sixteen days. The MODIS sensor has 36 spectral bands, situated between 0.4 to 14.2 microns. It has

- 49 -
a maximum spatial resolution of 250m for bands 1-2, 500m for bands 3-5 and 1000m for bands 6-29,
respectively.
For our present study, the MODIS-Terra satellite bands 1 (0.62-0.67 um), 2 (0.84-0.87 um), 3 (0.45-
0.47 um), and 7 (2.10-2.15 um) with 1 km resolutions were used. The MODIS-Terra level 1B data were
collected between November - December 2004 and January - May 2005. This particular time period was
chosen, where the maximum amount of cloud free images with a near nadir view could be obtained. Also,
the data from SPOT VI were used in this study.

4. Analysis and Result

In order to obtain an accurate vegetation computation, the geometrical correction is very important,
especially for regions such as Taiwan, because of its complex landuse/landcover distribution, mountainous
terrain and fine agriculture farming. Fig. 1 is an example demonstrating the vegetation computation errors
caused by misregistration. Due to this aspect, a precise MODIS data geo-correction procedure was
developed. This was done by conducting an image matching process to a reference SPOT geo-corrected
image set, along with a subsequent fine resampling procedure. Results show a great improvement in the
sup-pixel registration accuracy.
As this study seeks to find a vegetation index that is more atmospherically resistant, the various
computed vegetation indexes from SPOT-4 VEGETATION data were compared with the ground measured
aerosol optical depth (AOD) of South Korea’s Anmyon station, one of the Aerosol Robotic Network
(AERONET) observation sites.
From the slope of Fig. 2, we see that the NDVI index value goes down steeply with an increasing
AOD. This is consistent with the fact that a higher AOD causes the suspended particles in the atmosphere
to scatter the red band more. This increased scattering may produce a larger reflectance in the red band
than the NIR for the sensor to detect, and thus render the value of the NDVI index to drop. As for the
scatter plot of the AFRI index with the AOD, the regression line slopes a little upward with an increasing
AOD (not shown). This indicates that the SWIR can still be slightly affected from an increasing AOD, but
not as much as the red band. Finally, for the ARVI index (Fig. 3), when gamma equals 1.3, the regression
lines tilts up a little; when gamma equals 1, it becomes nearly horizontal; and when gamma equals 0.7, it
slightly edges down. The more horizontal regression line becomes, the more likely it is not so easily
affected by the AOD. At this point, the lower correlation coefficients indicate that that model is less
sensitive to the AOD variation. In other words, the model, which has the lowest correlation relationship, is
the one we want. Therefore, it seems logical to assume that the AFRI index and the ARVI index with a
gamma value of 1.0 would be the most capable of overcoming the influence of aerosols.
Following a similar procedure used by Kaufman et al (1997) in defining AFRI2:1, an alternative EVI
(EVI SWIR) was developed by replacing the blue ray with the short wave infrared band. The composite
EVI and EVI SWIR images are shown in Fig. 4. The EVI values are very high in the western plain areas,
which appears to be a forestry area. Yet, this is not true. The EVI SWIR does not show such high values in
these regions. This demonstrates a more reasonable landuse/landcover distribution. Subsequent detailed
comparisons also supported that the EVI SWIR was better than the EVI.

5. Summary
Through comparisons of the three vegetation indexes, it appears that the AFRI and ARVI index when
Ȗ=1, seems to be the most capable in reducing the atmospheric effects. In contrast, an over-correction
appears to occur when Ȗ=1.3, and an under-correction seems to occur when Ȗ=0.7. In addition, the value of
0.7 demonstrated the lowest fluctuation throughout the changes in the AOD. This contradicts to the fact
that the gamma value of 1 is supposedly the most capable of decreasing the atmospheric influence.
Consequently, more research should be done regarding the value of 0.7. Furthermore, it is also suggested
that additional gamma values ranging from 0.7 and 1.3 should be tested in the future.
Meanwhile, the addition of the blue ray or short wave channels could provide additional information
in evaluating the atmospheric condition (AOD, turbidity, or smoke). It could help reduce the noise, when
computing the vegetation index. This can be seen from the consideration of the short wave data utilized in

- 50 -
this study. Basically, any single vegetation index is unable to provide a perfect delineation of the
vegetation canopy distribution. Needless to say, a better atmospheric and soil resistant vegetation index
could be established by compositing the different vegetation index’s advantages where our analysis.

References
1. French, A.N., Schmugge, T. J. and Kustas, W.P., 2000: Discrimination of Senescent Vegetation
Using Thermal Emissivity Contrast. Remote Sensing Environ., 74, 249~254.
2. Kaufman, Y. J. and Tanre, D., 1992: Atmospherically Resistant Vegetation Index (ARVI) for
EOS-MODIS. IEEEE Trans. Geosci. Remote Sensing., 30 (2), 261~270.
3. Kaufman, Y. J., Wald, A. E., Remer, L. A., Gao, B. C., Li, R. R. and Flynn, L., 1997: The MODIS
2.1ȝm Channel---Correlation with Visible Reflectance for Use in Remote Sensing of Aerosol. IEEEE
Trans. Geosci. Remote Sensing., 35 (2), 1286~1298.
4. Rouse, J.W., Haas, R.H., Schell, J.A. and Deering, D.W., 1973: Monitoring vegetation systems in
the great plains with ERTS. Third ERTS Symposium, Goddard Space Flight Center, Washington, DC.
NASA SP-351, 390~317.
5. Santer, R., V. Carrere, Ph. Dubuisson and J. C. Roger, 1999: Atmospheric corrections over land
for MERIS. International

0,25

0,2

0,15

0,1

0,05

0
Red_ref NIR_ref NDVI
-0,05
05Nov05 09Nov04_reference 09Nov05_actual

Fig.1 Change in the NDVI values due Fig. 2 Comparison of the SWIR reflectance
misregistration. and AOD-500nm.

Fig. 2 Comparison of the NDVI reflectance Fig.4 Monthly composite EVI (left panel)
and AOD-500nm. and EVI SWIR (right panel) for May 2005.

- 51 -
Sharing ground truth data for land cover mapping – GLCNMO

Ryutaro Tateishi
CEReS, Chiba Univeristy, 1-33 Yayoi-cho Inage-ku Chiba 263-8522 Japan
E-mail: <mailto:tateishi@faculty.chiba-u.jp>tateishi@faculty.chiba-u.jp

Ground truth data in land cover mapping are important as training data and validation
data. The quality of mapping result is highly dependent on training data, and validation is
an essential step in the mapping. However collection of ground truth data needs much
work. Land cover ground truth data for global area are being prepared by national mapping
organizations with the initiative of Chiba University in the Global Mapping project
(<http://www.iscgm.org/>http://www.iscgm.org/). These data will be used for the Global
Land Cover mapping by National Mapping Organizations (GLCNMO), and also they will
be available to anyone in order to promote sharing ground truth by any organizations and
projects.

- 52 -
 Ecosystem changes mapping for Eastern Shore of Lake Hovsgol from satellite
imagery and GIS a case study
B.Gantsetseg
Institute of Geoecology, Baruun Selbe-15,Ulaanbaatar-211218, Mongolia
Fax :976-11-321862, E-mail : bgantsetseg@yahoo.com

ABSTRACT. Analysis of the watersheds on the eastern-shore of Hovsgol Lake in

relation to soil, pasture and vegetation was done using remote sensing techniques
through the use of Arcview, ENVI (the environment for visualizing images) and
ERDAS IMAGINE and ArcGIS geographic information systems software. Then,
depending on the priority of the factors to be considered for alignment, proper
weight ages were selected for different features. The Turag valley has the largest
catchment area (22635.1 ha) and also has the longest river in the study area. The
area of intensively burned forest area was estimated to cover 2903.6 ha using a
supervised classification map. The Permafrost and forest pest insect distribution
maps are shown for the Eastern part of Lake Hovsgol, prepared by building a
model operation in ArcGIS (based on topographic parameters of solar radiation,
elevation, wetness index, slope, NDVI and landcover).

KEY WORDS : Landsat-TM, ASTER, Supervised classification,

INTRODUCTION
Watershed studies are essentially the collection of information on a wide range of parameters of
static and dynamic processes related to the landscape, hydrology, vegetation, land use, soils,
geomorphology, drainage condition, climate etc.
Remote sensing provides valuable data over vast areas in a short time about natural resources, meteorology
and environment leading to better resource management and accelerating national development. The data
collected through aerial and satellite sensing is either visually interpreted or analyzed on the computer for
specific resource applications. Normally both visual and computer techniques are employed to tackle the
complex problems involving more disciplines. The main advantages of automatic analysis are
classification of many resource themes covering a large area in a short time.
While, the remote sensing data is operationally being utilized for mapping various resources, the need is to
step ahead towards integrating these resource maps with other resource related information and socio-
economic data for generating action plans (Rao et al.,1994).
A Geographic Information System is a specific information system applied to geographic data and is
mainly referred to as a system of hardware, software and procedures designed to support the capture,
management, manipulation, analysis, modeling and display of spatially referenced data for solving
complex planning and management problems (Burroughs, 1986). The GIS will have to be the workhorse of
integrated database system, as both spatial and non-spatial data for habitat suitability need to be handled.
More intensely developed approach to information integration: attribute information is associated with
point, line and polygons as spatial entities that describe features occurring in the real world.

MATERIALS AND METHODS

Materials
The project study area includes six watershed valleys along the eastern shore of Hovsgol Lake, an area that
is mountainous and mostly covered by taiga forest.
Landsat-7 imaging bands 4(R), 7(G) and 2 (B), as well as digital data dated 28 June 2000 and ASTER
imaging (with 14 channels of digital data dated 27 August 2001) were used to prepare forest cover type
and landcover maps on a scale of 1:100,000. Topographis maps, scale 1:100,000, sheets M47-22, M47-34,
M47-46, M47-23, M47-35 and M47-47 produced by Mongolian Government Survey Agency, were used as
base map for the sampling design and generating a Digital Elevation Model.

- 53 -
Methodology
In the first step Landsat TM and ASTER images were used for visual interpretation and pre-classification
in order to create thematic maps of forest covering and density. The next was image processing which
included image classification, interpretation and distribution analysis. The comparison between spectral
signatures of different important cover types such as burnt forest, unburnt forest and of image from
ASTER was performed.
Finally, GIS was conducted to investigate the relationship between different research fields and the
environmental factors. The research flowchart is shown in figure 1. Fieldwork carried out during July-
September 2004-2005.
The ASTER image was classified by using Supervised Classification techniques and maximum Likelihood
Classifier.

RELEVANT PARAMETERS USED TO RELATIONSHIP FIELD STUDY

SATELLITE IMAGE Ancillary data

TOPOSHEETS

C T S D F
Digitize LANDSAT ASTER L e o i o
I r c a r
M r I T e
Contour A e A o s
Top point T L m
GEOMETRIC CORRECTION s t
E t
D d e
DEM D i A a c
A n T t o
LANDSAT ASTER T s A a l
A e o
Filter (DX, DY)] c g
NDVI t y
FCC
Air tem, solar Forest age,
Slope Aspect
Forest covering
rad, precip. etc fire
and density map Grasshopper Diatom
distribution distribution
Vegetation map
Influenced
Family location
factor for Influenced and income
forest pest Influenced Influenced factors on
insect factor forest factor on permafrost and
pest insect forest pest
distribution permafrost insect
Relationship
distribution between
distribution distribution
research fields

FLOW CHART FOR METHODOLOGY ON GIS MAPPING

Basic data Thematic maps Relationship between Distribution map
research fields

DEM Family
Remote sensing location
data
Family Permafrost
Vegetation, Distribution distribution
income for
geomorphology, terrestrial
Analog data landscape insects

(Toposheet) Distribution Forest pest

for diatom insects’
Hydrology distribution

Ancillary
data
Climate data
(Field data)

Figure 1. Methodology
- 54 - flow chart
RESULTS AND DISCUSSIONS

The ASTER image 2001 was classified to the specific three classes: intensive burnt, low burnt and
unburnt forest. The intensive burnt areas cover 0.4% or 440.50 ha and low burnt area covers 20% or
23010.9 ha of the study area. The unburnt forests cover 12.3% or 13522 ha of the total area.
The toposheet and field data for pasture type, grazing area and vegetation dominant species were used to
prepare pasture map on a scale of 1:100,000 for study area.

The permafrost distribution map prepared based on topographic parameters of solar radiation,
elevation, wetness index, ndvi and landcover. The permafrost distribution values, as active
permafrost layer low in steppe and grassland areas and the active layer ground temperatures
fell and permafrost was present, particularly on the north facing slopes of the mountains.
These altered parameter values showed that on the north facing slopes of mountains and in
riparian zones were much more widespread.

The forest pest insect distribution maps prepared by building model operation in arcgis (based
on topographic parameters of solar radiation, elevation, wetness index, ndvi and landcover, air
temperature and slope). Gypsy moth lays their eggs in rock outcrops located in warm and
sheltered south facing slopes, so their larvae distribution is obviously near the rock is shown in
the distribution map. The sample data of defoliated area by gypsy moth is overlapping to high
distribution area identified by influence factors. It is allowing to conclude us that slope is the
relatively higher influence factor to the distribution of gypsy moth.

REFERENCES

1. Annual report 2004., Institute of Geoecology of MAS Project “The Dynamics of Biodiversity Loss
and Permafrost Melt in Lake Hovsgol National Park, Mongolia”
2. Annual report 2002., Institute of Geoecology of MAS Project “The Dynamics of Biodiversity Loss
and Permafrost Melt in Lake Hovsgol National Park, Mongolia”
3. Richard D.Hunter 2004 “Climatologically aided mapping of daily precipitation and temperature”,
Journal of applied meteorology, 44
4. Michael N.Demes 2002., “GIS modeling in Aster”
5. David O’Sulvivain, David J.Unwin 2003., “Geographic Information Abalysis”
6. Paul A.Longley, Michael F.Goodchild, David J.Maguire, David W.Rhind 2004., “Geographic
Information Systems and Science”
7. Eva Solbjord Flo Heggem 2005, “Mountain permafrost distribution and ground surface
temperature variability in Southern Norway and Northern Mongolia-spatial modeling and
validation”
8. Burrough, P.A. 1988. Principles of Geographical Information Systems for Land Resources
Assesment. Clarendon Press, Oxford.
9. Manual for ArcGIS software

- 55 -
 GIS application on micro relief development
Altangerel, B1., Schwanghart, W2. & Walther, M.1

1. MOLARE Research Centre, NUM, mgl_agi@yahoo.co.uk

2. Free University of Berlin, schwanghart@geog.fu-berlin.de

Research area;
The research area is about 1 km south of Ugii Nuur. Ugii Nuur is located 350km west to
Ulaanbaatar. There could be observed 3 closed depressions located at one line next to each other.
These kinds of form are rarely to be found in Mongolia’s nature. Local nomads called the area as
“Gurvan togoo”.
The geographical coordinates E 102o 46i 46ii
N 47o 43i 53ii
Nowadays especially erosion processes are heet flushes. That means in summer time severe rain
showers happen in short periods of sometimes only one hour.

Field measurement;
1. mapping the area focused on the surface development.
2. measurement by differential GPS. Highest elevation error possibility is 20 cm in total
more than 2300 points were measured during the field work.
3. save all data as vector data on notebook.
4. make TIN calculation
5. convert into the Digital Elevation Model (DEM)

Result:
Visualisation of the depression in a 3D model from DEM file.

Hypothesis based on the result:

What is the genesis of the forms? The depressions could be interpreted as Pingo relicts in a former
lake depression of Ugii Nuur. This is evidence for a large extension.

Conclusion (based on additional geomorphological field work):

1. depressions are Pingo relicts
2. Paleoclimate condition. Caused a higher lake level to be proved by a higher shoreline.
3. Paleotemperature must be less than -7’C mean annual temperature.
Bay of the former lake was 100 m higher than present lake level.

Reference:
1. Strahler, A. & A. Strahler (2002): Physical Geography.- John Wiley and sons, New York,
USA.
2. Lillisand, M. & W. Kiefer (2004): Remote sensing and image interpretation.- John Wiley
and Sons, New York, USA.
3. Tsegmid, Sh. (1969): Physical Geography of Mongolia.- National Publishing House,
Ulaanbaatar, Mongolia. [in Mongolian]

- 56 -
 Vegetation mapping of the Great Gobi A strictly protected area
A. Tsolmon1 & H. von Wehrden2,3
1)
Institute of Botany, Mongolian Academy of Sciences, Mongolia
2)
Research Institute of Wildlife Ecology, University of Veterinary Medicine, Savoyen Strasse 1, A-1160 Vienna,
Austria 3) Institute of Geobotany, Martin-Luther University of Halle-Wittenberg, Germany

Spatially explicit information on the distribution of habitats and vegetation types is a key
requirement for resource management. The Outer Mongolian Republic has designated some of the world's
largest nature reserves in the last years, yet some protected areas are situated in remote semi-desert regions
with limited baseline data available. Here, we present a recent study on remote-sensing based vegetation
mapping in the probably driest reserve in Mongolia (HIJMANS et al. 2005), the Great Gobi A special
protected area situated in the deserts und semi-deserts of the Transaltay Gobi.
Field-work followed a Braun-Blanquet approach (MUELLER-DOMBOIS & ELLENBERG 1974) with
deliberate selection of study sites; sampling was guided by visual interpretation of unsupervised
classifications from Landsat satellite data (CAMPBELL 1996). A similar approach was already successfully
applied in other protected areas in southern Mongolia (VON WEHRDEN & TUNGALAG 2004; VON
WEHRDEN et al. in print). Relevɣs were classified by phytosociological table work following the general
framework provided by HILBIG (1995, 2000). This classification was compared to the Russian
classification scheme based on dominant species (RAýHKOVSKAYA & VOLKOVA 1977), to cluster analysis
(PODANI et al. 2000), and to a COCKTAIL classification (BRUELHEIDE 2000). Indirect and direct gradient
analysis techniques (standard DCA and CCA, see HILL & GAUCH 1980; JONGMAN et al. 1995; TITEUX et
al. 2004) were used to infer ecological relationships among relevɣs and species; secondary data for the
ordination included climatic data from a public domain climate model, digital elevation data and remote
sensing information in addition to survey data gathered during fieldwork (VON WEHRDEN & WESCHE
2005).
Phytosociologically classified plots served as ground checks for a supervised classification of
Landsat 7 data in order to create a vegetation map (see VON WEHRDEN et al. in print on further details of
the method). Prior to classification, plots had to be manually enlarged as relevɣs were not large enough for
the employed classification algorithm. Isoclass unsupervised classifications of the raster data were the first
approach to identify homogenous areas within the Landsat scenes. As vegetation cover in the study area is
mostly below 20 % several NDVI transformations were tested to assess primary productivity and
vegetation cover. However, tasselled cap transformations proved to be a better tool for individual
enlargement of the plot data.
Enlarged plots were finally implemented into a maximum likelihood classification, results of
which were smoothed using a nearest neighbour 7x7 pixel mean filter. Classification accuracy was
assessed by cross-validation with an independent data set. The final map was implemented in a GIS and
compared to maps produced with the same methods for the neighbouring Great Gobi B strictly protected
area (VON WEHRDEN & TUNGALAG 2004) and the Gobi Gurvan Saikhan national park (VON WEHRDEN et
al. in print).
Sound spatially information on the distribution of habitats and vegetation types is a necessity for
resource management. The Outer Mongolian Republic has designated some of the world's largest nature
reserves in the last years and decades, yet some protected areas are situated in remote semi-desert regions
with limited baseline data available (see map 1). Here, we present a recent study on remote-sensing based
vegetation mapping of the driest reserve in Mongolia (based on data from HIJMANS et al. 2005), the Great
Gobi A strictly protected area situated in the deserts und semi-deserts of the Transaltay Gobi. Sampling
was undertaken during summer 2004 and contained two surveys within the working area (see map 2).
Field-work followed a Braun-Blanquet approach (MUELLER-DOMBOIS & ELLENBERG 1974) with
deliberate selection of study sites; sampling was guided by visual interpretation of unsupervised
classifications and RGB-transformations from Landsat satellite data (CAMPBELL 1996). A similar approach
was already successfully applied in other protected areas in southern Mongolia (VON WEHRDEN &
TUNGALAG 2004; VON WEHRDEN et al. in print). Relevɣs were initially classified by phytosociological
table work following the general framework provided by HILBIG (1995, 2000). This classification was
compared with the Russian classification scheme based on dominant species (RACHKOVSKAYA &
VOLKOVA 1977), with cluster analyses (PODANI et al. 2000), and with a COCKTAIL classification

- 57 -
(BRUELHEIDE 2000). Indirect and direct gradient analysis techniques (standard DCA and CCA, see HILL
1974; GAUCH 1994) were used to infer ecological relationships among relevɣs and species; secondary data
for the ordination included climatic data from a public domain climate model, digital elevation data and
remote sensing information in addition to survey data gathered during fieldwork (VON WEHRDEN &
WESCHE 2005). The environmental parameters of each community were analyzed using Boxplots; the
results were partly used as stratification key for the raster data classification. The phytosociological system
is currently in preparation for publication (VON WEHRDEN et al. in prep.). By now eight vegetation units
were derived; however some 13 contained sub-association or sub-communities. Altogether 18 units were
classified, out of which 13 were included in the preliminary classification. Compared to vegetation maps
from other Gobi regions this is seemingly lower (VON WEHRDEN & TUNGALAG 2004; VON WEHRDEN et al.
in print). This might be due to the aridity of the here presented region (see table 1).
Phytosociologically classified plots served as ground checks for a supervised classification of
Landsat 7 data in order to create a vegetation map (see VON WEHRDEN et al. in print on further details of
the method). Prior to classification, plots had to be enlarged as most relevɣs were too small for the
employed classification algorithm. Isoclass unsupervised classifications of the raster data were the first and
initial approach to identify homogenous areas within the Landsat scenes. As vegetation cover in the study
area is mostly below 20 % several NDVI transformations were tested to assess primary productivity and
vegetation cover. However, tasselled cap transformations proved to be a better tool for individual
enlargement of the plot data. To asses the mapping possibilities based on the available ground truth data,
spectral separability was checked; units which were closely related regarding their spectral values were
therefore merged or excluded. However this was only the case regarding some few extrazonal communities,
which contained comparable vegetation cover and structure.
The training data was finally implemented into a maximum likelihood classification; the results
were smoothed using a nearest neighbour 7x7 pixel mean filter. Classification accuracy was assessed by
cross-validation with an independent data set. Thus kappa statistics were compiled and accuracy checks
performed, which yielded around 91 %. Deserts and semi-desert ecosystems are more homogenous, and
therefore showed a higher accuracy. The dry steppe units occurred mainly at higher elevations on more
heterogenous hills and mountains, and therefore showed an overall lower accuracy. The final map was
implemented into a GIS and compared to maps produced with the same methods for the neighbouring
Great Gobi B strictly protected area (VON WEHRDEN & TUNGALAG 2004) and the Gobi Gurvan Saikhan
National Park (VON WEHRDEN et al. in print).
For future investigation we want to compare the results with available public-domain climate
datasets (HIJMANS et al. 2005) and thus examine the climatic variance of these communities. Using
MODIS timelines we plan to identify NDVI variability for each mapping unit. This approach was already
widely applied in central Asian ecosystems, yet on a broader resolution regarding vegetation units and
working area (e.g. YU et al. 2003; YU et al. 2004). The NDVI timelines will be correlated with rainfall
estimates. These will be linked with Khulan observations obtained from collared animals using ARCOR. A
combination with other animal observation (such as Ibex, wild Camel, Argali, Gobi Bear, Gazelle etc.)
would be desirable to understand the habitat conditions of these endangered animals. The final goal of this
project is to give baseline data for better protection of the fragile Gobi ecosystem; some applications are
currently tested in some protected areas of the region (visit www.takhi.org; KACZENSKY et al. in prep.).

Acknowledgements: Fieldwork would not have been possible without the support of the staff of
the Great Gobi A strictly protected area office. The project was supported by the DAAD (German
Academic Exchange Service) and the Austrian Science Foundation (FWF project P18624).
Tab. 1: Variance of the precipitation of the protected areas within southern Mongolia. All values represent
millimetres of precipitation per year, assessed using data from HIJMANS et al. 2005.

AREANAME MIN MAX RANGE MEAN STD

Great Gobi "B" 69 177 108 96 17
Great Gobi "A" 33 124 91 54 11
Gobi Gurvan Saykhan 39 222 183 103 35
Small Gobi "A" 63 115 52 84 8
Small Gobi "B" 113 173 60 139 14

- 58 -
 Map 1: Protected areas within southern Mongolia.

Map 2: Route of our surveys; the thin black outline surrounds the protected area; the thick black outline
shows the border to China.

References

1. BRUELHEIDE, H. 2000: A new measure of fidelity and its application to defining species groups. -
Journal of Vegetation Science 11: 167-178.
2. CAMPBELL, J. B. 1996: Introduction to Remote Sensing. - New York.
3. GAUCH, H. G. 1994: Multivariate analysis in community ecology. - Cambridge.
4. HIJMANS, R. J.; CAMERON, S. E.; PARRA, J. L.; JONES, P. G. & JARVIS, A. 2005: Very high
resolution interpolated climate surfaces for global land areas. - Int. J. Climatol. 25: 1965-1978.
5. HILBIG, W. 1995: The vegetation of Mongolia. - Amsterdam.

- 59 -
6. HILBIG, W. 2000: Kommentierte ɖbersicht ɶber die Pflanzengesellschaften und ihre hɰheren
Syntaxa in der Mongolei. - Feddes Repert. 111: 75-120.
7. HILL, M. O. 1974: Correspondence analysis: a neglected multivariate method. - Journal of Applied
Statistics 23: 340-354.
8. KACZENSKY, P.; WALZER, C. & GANBAATAR, O. in prep.: Niche separation of the two native
Asian equids: the Przewalski's horse and the Asiatic wild ass in the Gobi areas of SW Mongolia.
9. MUELLER-DOMBOIS, D. & ELLENBERG, H. 1974: Aims and methods of vegetation ecology. - New
York, London, Sydney, Toronto.
10. PODANI, J.; CSONTOS, P. & TAMȻS, J. 2000: Additive trees in the analysis of community data. -
Community ecology 1.
11. RACHKOVSKAYA, E. I. & VOLKOVA, E. A. 1977: Rastitelnost Zaaltayskoy Gobi. - Biol. Res. prir.
Uslov MNR 7: 46-74.
12. VON WEHRDEN, H. & TUNGALAG, R. 2004: Mapping of vegetation units of the Great Gobi B
National Park. - Final report for the Przewalski Horse Project. - Univ. Marburg, Fac. of Geography.
13. VON WEHRDEN, H. & WESCHE, K. 2005: Mapping Khulan habitats - a GIS-based approach. -
Erforsch. biol. Ress. Mongolei 10.
14. VON WEHRDEN, H.; WESCHE, K. & HILBIG, W. in prep.: Plant communities of the Mongolian
Transaltay Gobi. - Feddes Repertorium.
15. VON WEHRDEN, H.; WESCHE, K.; REUDENBACH, C. & MIEHE, G. in print: Mapping of large-scale
vegetation pattern in southern Mongolian semi-deserts - an application of LANDSAT 7 data. -
Erdkunde.
16. YU, F.; PRICE, K. P.; ELLIS, J.; FEDDEMA, J. J. & SHI, P. 2004: Interannual variations of the
grassland boundaries bordering the eastern edges of the Gobi Desert in central Asia. - International
Journal of Remote Sensing 25: 327-346.
17. YU, F. F.; PRICE, K. P.; ELLIS, J. & SHI, P. J. 2003: Response of seasonal vegetation development
to climatic variations in eastern central Asia. - Remote Sensing of Environment 87: 42-54.

- 60 -
 Climate change impact on rangeland productivity in the Eurasian steppe
Dennis S. OJIMA1 and Togtohyn CHULUUN2
1-Natural Resource Ecology Laboratory, Colorado State University, Fort Collins, Colorado 80523-1499, U.S.A.
2- Mongolian Academy of Sciences, Ulaanbaatar - 46, Mongolia

Abstract

Dramatic changes have occurred in pastoral systems of Mongolia, China, Central Asia and
Russia over the past several decades. Integrated assessment of these changes on the environment
and quality of life is essential for sustainability of the region. Integrated assessment entails
determining the interactions and impacts of various management strategies on the environment and
human systems. Recently, an evaluation of the pastoral systems has been conducted in the region.
Pastoral systems, where humans depend on livestock, exist largely in arid or semi-arid ecosystems
where climate is highly variable. Thus, in many ways, pastoral systems are adapted to climatic
variability. It is plausible to assume direct connection between climate variability, ecosystem
dynamics and nomadic land use system in Mongolia. Interactions between ecosystems and
nomadic land use systems co-shaped them in mutual adaptive ways for hundreds of years, thus
making both the Mongolian rangeland ecosystem and nomadic pastoral system resilient and
sustainable.
We also recognize the pervasive role of demographic, political and economic driving
forces on pastoral exploitation. The general trend involves greater intensification of resource
exploitation at the expense of traditional patterns of range utilization. This set of drivers is
orthogonal to the described climate drivers. Thus we expect climate-land use interactions to be
modified by socio-economic forces. Nevertheless, the complex relationship between climate
variability and pastoral exploitation patterns will still form the environmental framework for overall
patterns of land use change. Integration of knowledge and delivery of this knowledge to scientists,
policy makers and land users is critical for regionally sustainable development.

Introduction

In the semiarid regions of the Mongolian steppe, nomadic pastoralism has been the dominant
agronomic activity for many centuries. Recent changes in cultural, political and economic factors have caused
changes in how the pastoral systems operate within the region. These systems encompass a range of grazing
patterns (i.e., frequency, intensity of grazing and the types of animals), and have incorporated new breeding
stocks that are potentially not suitable for certain climate regimes (e.g., drought conditions of the Gobi desert,
cold hardiness against severe winter storms in the Mongolian steppe region). These changes in pastoral
management have altered the nomadic patterns of the region.
Pastoral systems, where humans depend on livestock, exist largely in arid or semi-arid ecosystems
where climate is highly variable. Thus, in many ways, the historical pastoral livestock systems are intimately
adapted to climatic variability. There is a direct relationship between climate variability and the spatial scale
of pastoral exploitation. Extensive nomadic systems are found in the most variable regions; less extensive,
more intensive modes of livestock management occur in less variable grazing lands. Climate change in
drylands can thus be expected to have important implications for the dynamics and viability of pastoral people,
and their exploitation patterns on land cover and land cover change.
We also recognize the pervasive role of demographic, political and economic driving forces on
pastoral exploitation. The general trend involves greater intensification of resource exploitation at the expense
of traditional patterns of extensive range utilization. This set of social factors tend to over-ride the climate
drivers during periods of economic change. Thus we expect climate-land use-land cover relationships to be
modified by the socio-economic forces. Recent political and economic changes (i.e., in the past 50 years) in
land use management have resulted in a more sedentary livestock management system. These changes have
led to more intensive stocking rates in localized areas and change in the breeds of animals used. More recent
changes in the social-economic setting have forced new changes in pastoral management due to relaxation of
central government controls and the implementation of a more "free-enterprise" system. What will result from
these recent changes is unclear, and the effect on the human and natural resources of these arid and semi-arid
regions needs to be determined.

- 61 -
Climate Trends
In the last 60 years, the mean annual air temperature increased by 1.560C, due to winter warming
(Mongolia National Action Program on Climate Change 2000). Changes in warming are more pronounced in
the high mountains and mountain valley, and less in the Gobi desert and the steppe. There is a slight increase
in the annual precipitation in the last 60 years (Natsagdorj 2000). During 1940-1998, the annual precipitation
increased by 6%, while summer precipitation increased by 11% and spring precipitation decreased by 17%,
with the main spring dryness happening in May.
The frequency of extreme events such as drought, flood, dust storm, thunderstorm, heavy snow, and
flash flooding, has increased over the past 30 years (Natsagdorj 2000). In addition, the forest and steppe fire
frequency is increasing because of the extremely dry springs. The economic losses of these extreme events is
estimated to be 1 to 3 billion Togrog (1 to 3 million US$) per year. Economic losses due to drought range
annually from 5 to 7 million Togrog (5 to 7 thousand US$). Livestock losses from 1999-2000 zud (the
Mongolian term used for severe for livestock winter condition) were 2.4 million, and for the winter of 2001
was over 1.3 million livestock lost by February 25, 2001.
It is likely that the boundary zone between the Gobi desert and steppe is already affected by global
warming and land use impact. Analysis of onset of green-up, an indicator of spring thaw and the initiation of
the growing season, during the 1982 to 1991 time period indicates that large portions of eastern Mongolia and
Inner Mongolia are experiencing earlier green-up (Yu et al. 1999), the authors suggest that this is associated
with warming of winter and spring temperatures during this decade. This region of advanced green-up is
dominated by Meadow Steppe and relatively mesic areas of typical steppe. There are also large portions of the
desert steppe and dry areas of the typical steppe, where there is a strong trend towards delayed green-up. The
old herders from central regions of Mongolia also complained that the plant productivity decreased from one-
third to one-half during their lifetime (Ellis and Chuluun 1993).
According to the GCM scenarios, the annual mean temperature in Mongolia is projected to increase by
about 1.8-2.80C in the first quarter of the 21st century (Mongolia National Action Program on Climate Change
2000). This increase is projected to double during 2025-2050. An increase in total precipitation by 20-40%
can be projected in the first half of new century, but precipitation is expected to decline from 2040 to 2070.
Increased plant productivity is expected until the 2040’s, when less favorable conditions will follow.

Land Use Effects

Covering nearly 120 million hectares, Inner Mongolia accounts for 12.3% of China’s total area
(Grasslands and Grassland Sciences in Northern China 1992). In 1989, Inner Mongolia had 86.7 million
hectares of grassland, of which 18.7 million hectares were deemed “unusable” and another 29.9 million
hectares were considered “deteriorated” or “seriously deteriorated”, leaving only 38.1 million hectares both
usable and in good condition. Compared to 1965, the area of the region’s grasslands is said to have
decreased by 6.2 million hectares, deteriorated grasslands have increased by 28.7 million hectares, and
total grass production has dropped by 30%. Currently, Inner Mongolia has about 8 million hectares of
cropland (Enkhee 2000). Over 80% of land in Mongolia (123 million hectares) is rangelands (Tserendash
2000). Arable lands occupy only less than 1% of total agricultural lands (1.35 million hectares) in
Mongolia.
A comparative study of culture and environment in Inner Asia conducted by Humphrey and Sneath
(1999) found that pasture degradation was associated with the loss of mobility in pastoral systems. Pasture
degradation was most severe at the research sites from Buyatia and Chita Oblast’ (Russia), where the
sedentarisation level was the highest, compared to other research sites from Mongolia, Tuva (Russia),
Inner Mongolia and Xinjiang (China). Thus, they concluded that mobile pastoralism still remains a viable
and useful technique in the modern age.
Our simulation studies using CENTURY (Chuluun & Ojima 1996; Ojima et al. 1998, Chuluun & Ojima
1999) confirmed that the Mongolian grasslands could lose significant amounts of carbon in continuous
year-long or summer heavy grazing systems. The effect of different seasonal and year-long grazing
treatments on soil carbon levels was simulated for 50 years. Summer or year-long heavy grazing for 50
years resulted in the largest loss of total soil carbon relative to other seasonal grazing scenarios. Heavy
summer grazing resulted in a 15% soil carbon loss. The influence of spring grazing on soil organic carbon
levels was slightly greater than winter and fall grazing. A total soil organic carbon level decrease of about
25% was observed in the heavy grazed Xilingole research site treatments compared to control treatments.
However, a decrease in soil organic carbon was not observed in the heavy or overgrazed grassland sites in
Mongolia relative to control sites. Thus, we have to be cautious when we quantify carbon loss because of

- 62 -
overgrazing.
Not all land use changes were negative for grassland health and carbon storage. The crop sector has
been collapsing in Mongolia during the transition to a market economy since 1990. In 1995, sown areas have
decreased by half, and crop production decreased by about one third of the 1990 levels. The growing of plant
fodder and cereals other than wheat has practically been eliminated by 1995. This agricultural failure
decreased food security of the country. Livestock reduction in Central Asia and cropping industry failure in
Mongolia could potentially contribute to a restoration of degraded pastures and eroded arable lands.
Restoration would have the added benefit of increasing carbon storage and consequently these lands can serve
as potential sinks for atmospheric carbon.

Regional implications

Ellis and lee (1999) analyzed rain use efficiency of different vegetation communities of the
lake balkash basin in southeastern kazakhstan, between 430 and 470 north latitude, using
remote-sensed avhrr-ndvi data. Site ndvi data were compared to annual rainfall figures from
the closest weather station. Improved rainfall use efficiency by typical steppe and shrub
steppe vegetation communities (indication of grassland recovery) for last decade has been
observed, and this change has coincided with a shift from intensive year-long grazing, to very
light or no grazing pressure for those grasslands.

analysis of rangeland recovery in central asia indicates that as growing season increases, there
is more carbon uptake into soils, suggesting that there is potential for carbon sequestration.
These improved grazing management practices may prove useful for the mongolian steppe
situation, which is undergoing large increases in livestock numbers. Implementation of
improved grazing practices can lead to sustainable carbon storage in these ecosystems, and
offset some of the negative impacts of climate warming by conserving soil moisture in the
rangelands.

Summary

Dramatic changes in land use have occurred in this region of East Asia during the last several
decades. The extent of grassland conversion into croplands and grassland degradation is large due to
increased human population and political reforms of pastoral systems. Rangeland ecosystems of this
region are vulnerable to environmental and political shocks, and response of pastoral systems to these
shocks have varied across the region. Indeed, this part of the world has experienced many perturbations in
the recent decades, such as, the collapse of the livestock sector in some states of central Asia, expansion of
livestock in China and Mongolia, intensification of cropland conversion, multiple zud of 1999-2001 in
Mongolia, and recent intensive dust storm events.
However, this region has great potential for rangeland improvements and carbon sequestration, if
appropriate land management practices are adopted. Traditional pasture management with greater mobility
should be incorporated in rural development policies. For instance, traditional pastoral networks should be
encouraged with new land reform policy in Mongolia. The traditional pastoral networks emerged as
dissipative structures during the past in areas of limited natural resources (water and soil organic matter)
and highly variable environment conditions. These strategies improved the resilience and sustainability of
grazing of grazing lands in Mongolia and surrounding pastoral regions in Eurasia.

Literature Cited

1. Chuluun, T. and Ojima, DS. 1996. Rangeland management. U.S. Country Studies Program
Mongolia’s Study Team, Mongolia’s Country Studies Report on Climate Change: Mitigation
Analysis, Ulaanbaatar, Mongolia, vol. 4: 57-78.
2. Chuluun, T. and Ojima, DS. 1999. Climate and grazing sensitivity of the Mongolian rangeland
ecosystem. Proceedings of the VI International Rangeland Congress on “People and Rangelands:
Building the Future”, Townsville, Queensland, Australia, 1999, vol. 2: 877-878.

- 63 -
3. Ellis, J. and Chuluun, T. 1993. Cross-country survey of climate, ecology and land use among
Mongolian pastoralists. Report to Project on Policy Alternatives for Livestock Development
(PALD) in Mongolia. Institute of Development Studies at the University of Sussex, UK.
4. Ellis, J. and Lee, RY. 1999. Collapse of the Kazakhstan livestock sector: Catastrophic
convergence of ecological degradation, economic transition and climate change. Report on
“Impacts of privatization on livestock and rangeland management in semiarid Central Asia”,
Overseas Development Institute, London, 1999.
5. Enkhee, J. 2000. The Mongolian tradition of legal culture and the grassland management in Inner
Mongolia today, Proceedings of the International Symposium on “Nomads and use of pastures
today”, 2000, 194-199.
6. Humphrey, C. and Sneath, D. 1999. The End of Nomadism?: Society, State and the Environment
in Inner Asia, Durham: Duke University Press, 1999.
7. Batjargal, Z., Dagvadorj, D. and Batima, R. (eds.). Mongolia National Action Programme on
Climate Change. 2000. JEMR Publishing, Ulaanbaatar.
8. Natsagdorj, L. (2000). Climate Change. In Climate change and its impacts in Mongolia, ed. R.
Batima and D. Dagvadorj, JEMR Publishing, Ulaanbaatar.
9. Ojima, D.S., Xiangming, X., Chuluun, T., Zhang, XS. 1998. Asian grassland biogeochemistry:
factors affecting past and future dynamics of Asian grasslands. In Asian Change in the context of
global climate change, ed. James N. Galloway and Jerry M. Melillo, IGBP publication series,
Cambridge: Cambridge University Press 3: 128-144.
10. Tserendash, S. 2000. Pasture resource – utilization and management in Mongolia. Proceedings
of the International Symposium on “Nomads and use of pastures today”. Pp: 141-143.
11. Yu, F., Price, KP., Lee, RY., and Ellis, J. 1999. Use of time series of AVHRR NDVI composite
images to monitor grassland dynamics in Inner Mongolia, China. Proceeding of ASPRS’99.

- 64 -
 Classification of Multitemporal InSAR Data for Land Cover Mapping in
Selenga River Basin, Mongolia
Damdinsuren AMARSAIKHAN

Institute of Informatics and RS, Mongolian Academy of Sciences

&
Faculty of Geography and Geology, National University of Mongolia

Abstract
The aim of this study is to evaluate different features extracted from the multitemporal
spaceborne interferometric synthetic aperture radar (InSAR) data sets for a rural land cover
mapping. For the actual land cover classification, the traditional statistical maximum likelihood
classification and neural network method are performed and the results are compared. Overall, the
research indicated that the multitemporal InSAR data sets have a valuable contribution to efficient
land cover mapping.

Keywords: InSAR, Multitemporal, Classification, Accuracy

1. Introduction

At present, InSAR data sets are being widely used for land cover/use and other resources mapping.
Unlike the traditional single frequency and single polarisation SAR, the InSAR uses both the amplitude
and phase information from a pair of single look complex (SLC) SAR images. From this pair of SLC
images, different SAR products such as amplitude and coherence images as well as a digital elevation
model can be generated. These derived products or their enhanced features combined with other data sets
can be successfully used for different classifications to increase the performance of the applied decision
rules [2,4].

In the present study, we wanted to discriminate rural land cover types in Mongolia using the
features derived from multitemporal InSAR data sets. As a test site, the Selenga River Basin, Northern
Mongolia has been selected. The area represents a forest-steppe ecosystem and is characterized by fertile
for agriculture chestnut soil. In the area, such classes as forest, agricultural fields, swampy area, natural
vegetation, soil and water were available.

For the discrimination of the selected classes, the traditional statistical maximum likelihood
(MLH) and neural network (NN) classifications have been applied and compared. The actual
classifications have been performed using a) the original InSAR products, b) the SAR derivative features
and c) the results of a principal component analysis (PCA), and the final results were compared. Overall,
the research indicated that the multitemporal InSAR data sets have a valuable contribution to the efficient
land cover mapping.

2. Data sources

As data sources, multitemporal interferometric ERS-1/2 SLC SAR data with a spatial resolution of 25m
acquired with 1 day repeat pass interval on 16 and 17 October 1997, and 8 and 9 August 1998 were used.
In addition, for ground truth checking topographic maps of 1984, scales 1:100,000 and 1:200,000 as well
as soil and vegetation maps, scale 1:200,000 were available.

3. Derivation of the InSAR coherence and amplitude images

The coherence and amplitude images have been derived as follows [1,4]:

ƒ Initially, 200 ground control points (GCP) regularly distributed over the images were automaticall

- 65 -
 y defined using the satellite orbit parameters and the two SLC images were co-registered with 0.1p
ixel accuracy. Then, a course registration followed by a fine registration was performed.
ƒ Coherence has been calculated using 10x3 size window and the coherence image was generated.
ƒ From the complex images, amplitude images were generated.
ƒ The preliminary SLC images were converted from the slant range onto a flat ellipsoid surface.
ƒ The true size (5800x5800) SAR images were generated using image undersampling applying 3x3 s
ize low pass filter.

4. Derivation of the texture features

To derive texture features occurrence and co-occurrence measures were applied to the coherence and
average amplitude images of the multitemporal ERS-1/2 data sets. The occurrence measures use the
number of occurrences of each grey level within the processing window for the texture calculations, while
the co-occurrence measures use a grey-tone spatial dependence matrix to calculate texture values [5]. By
applying these measures, initially 36 features have been obtained, but after thorough checking of each
individual feature only 12 features, including the results of the data range, mean and variance filters
applied to the original SAR products were selected.

5. Principal component (PC) images

To reduce the dimensionality of the dataset, the PCA [7,8] has been performed to the extracted
SAR features. For the PCA 16 features, including the multitemporal ERS-1/2 coherence and average
amplitude images, and 12 texture features were used. As it was seen from the PCA, PC1 which contained
49.68% of the overall variance was dominated by the variance (loadings of 0.72) of the ERS amplitude
image of 1997 looked very similar to the original variance dominating image and did not contain useful
information, while PC2 which contained 21.82% of the overall variance was also dominated by the
variance (loadings of 0.63) of the ERS amplitude image of 1997 and did not contain much information.
Moreover, PC3 and PC4 which were dominated by the variances of the ERS coherence images of 1997 and
1998 also did not contain useful information. However, PC5, PC6 and PC7 which were dominated by the
variances of the mean filtered coherence image of 1998, data range filtered coherence image of 1998 and
data range filtered amplitude image of 1997, respectively, contained just 1.6% of the overall data variance,
but contained very useful information. These PC images delineated clear spectral views of the selected
classes of objects. Therefore, these PC images have been selected for further analysis and the other PCs
were rejected.

6. Classification of the InSAR features

As it is known, before classification of SAR images, the speckle noise of the SAR images should be
reduced, because, the reduction of the speckle increases the spatial homogeneity of the classes which in
turn improves the classification accuracy. In this study, to reduce the speckle of the selected features a 5x5
size frost filter has been applied [5].
After the speckle suppression, from the features, 2-3 areas of interest (AOI) representing the six selected
classes have been selected using a polygon-based approach. Then, training samples were selected on the
basis of these AOIs. The separability of the training signatures was firstly checked on the feature space
images and then evaluated using JM distance [8]. Then the samples which demonstrated the greatest
separability were chosen to form the final signatures. For the classification, the following feature
combinations have been used:

1. The coherence and amplitude images of InSAR products (4 bands),

2. 12 features selected from the occurrence and co-occurrence measures,
3. The PC5, PC6 and PC7 derived from the PCA.

For each of these feature combinations, the statistical MLH classification and NN method have been
applied and the results were compared [7,8].

- 66 -
 Figure 1. The original SAR image of the test area (a) and the results of the MLH classification
((b) original InSAR products, (c) 12 features, (d) PCs). (dark green-forest, brown-agricultural
fields, pink-swampy area,
green-natural vegetation, cyan-soil and blue-water).

For the accuracy assessment of the final classification results, the overall performance [4] has been used.
As ground truth information, for each class several regions containing the purest pixels have been selected.
In all cases, the performance of the MLH classification was better than the performance of the NN method.
The overall classification accuracies of the selected classes in the selected features are shown in table 1.
As seen from table 1, the performances of the classifications using the 12 features were higher than any
other combinations. The original SAR image of the test area and the results of the MLH classification are
shown in Figure 1.

Table 1. The overall classification accuracy of the classified features.

- 67 -
 Feature combinations Overall accuracy of Overall accuracy of
MLH method (%) NN method (%)

The original InSAR 81.02 76.16

products
12 features 82.29 77.86
The PCs 78.54 71.47

7. Conclusions

The aim of this research was to evaluate different features extracted from the multitemporal
spaceborne InSAR data sets for a rural land cover mapping. For the classification of the individual and
extracted features, the statistical MLH and NN classifications were used and the results were compared by
measuring the overall accuracy. In all cases, the performance of the MLH classification was better than the
performance of the NN method. Overall, the study indicated that the multitemporal InSAR data sets could
be efficiently used for a land cover mapping.

References

1. Amarsaikhan, D. and M.Ganzorig, 2001, Application of spectral and scattering knowledge for
interpretation of RS images, Journal of Informatics, Ulaanbaatar, Mongolia, pp. 87-95.
2. Amarsaikhan, D. and Sato, M., 2003, Feature extraction and multisource image classification,
Proceedings of the Asian Conference on Remote Sensing and International Symposium on Remote
Sensing, pp.597-600, Busan, Korea, November 2003.
3. Amarsaikhan, D., and Douglas, T., 2004, Data fusion and multisource data classification,
International Journal of Remote Sensing, No.17, Vol.25, pp.3529-3539.
4. D.Amarsaikhan, M.Ganzorig, M.Sato, 2005, Application of Multitemporal Interferometric SAR Data
for Land Cover Mapping in Mongolia, CD-ROM Proceedings of the Asian Conference on RS, Hanoi,
Vietnam.
5. ENVI, 1999, User's Guide, Research Systems, USA.
6. ERDAS, 1999. Field guide, Fifth Edition, ERDAS, Inc. Atlanta, Georgia.
nd
7. Mather, P.M., 1999. Computer Processing of Remotely-Sensed Images: An Introduction, 2 edition
(Wiley, John & Sons).
nd
8. Richards, J.A., 1993. Remote Sensing Digital Image Analysis-An Introduction, 2 edition (Berlin:
Springer-Verlag).

- 68 -
 Urban Land Cover Change Studies Using Multitemporal RS Images
D.Amarsaikhan, M.Ganzorig and B.Nergui
Institute of Informatics and RS, Mongolian Academy of Sciences

Abstrac
The aim of this study is to compare the changes in the main urban land cover classes of
Ulaanbaatar city, Mongolia occurred during a centralized economy with the changes occurred
during a market economy and describe the socio-economic reasons for the changes. For this
purpose, multitemporal remote sensing (RS) and cartographic data sets as well as census data are
used. To extract the reliable urban land cover information from the selected RS data sets, a refined
parametric classification algorithm that uses spatial thresholds defined from the local and contextual
knowledge is constructed. Overall, the study indicated that during the centralized economy
significant changes occurred in the ger area of the city, while during the market economy the
changes occurred in both ger and building areas.

Keywords: RS; Urban change; Building area; Ger area; Classes

1. Introduction
At present, Mongolia is facing a problem with the urban expansion and the growth of
population in the main cities. In general, much of Mongolia's urban growth has taken place since the
middle of 1970s, because, at that time, the government encouraged migration to urban areas, specifically to
Ulaanbaatar in the belief that this would increase the industrialization and productivity in the country. To
accommodate the growing population in the capital city, the government mainly constructed high-rise
apartment blocks [5]. However, they were far to satisfy the demands of the growing population. Therefore,
rural people when migrated to Ulaanbaatar, usually used gers (Mongolian national dwelling) for their
accommodation and built them up usually in urban fringes.
In 1990, Mongolia entered the market economy and it totally changed the lives in the society. For
the isolated rural people, it had become very difficult to reach the central market. Meanwhile, almost
everything started to centralize in the capital city and Ulaanbaatar had become the dream for many rural
people. Therefore, many rural families officially and unofficially moved to Ulaanbaatar. As a result, the
population of Ulaanbaatar had been significantly increased and the city area had significantly expanded.
To analyze such changes, the urban planners need detailed regularly updated maps, however, there are no
such maps. In this case, an updated map generated through an information extraction procedure from RS
data with an acceptable resolution can give them an impression about the changes in the city area where
some planning actions are considered [1,2].
In general, it should be interesting to study the urban growth in Ulaanbaatar city comparing the
growths occurred during the main industrialization period, that is the period in between 1969 and 1990
when Mongolia had socialist economy with the changes occurred during a certain period of the market
economy. In this research, we wanted to study the urban growth in the capital city considering six classes
such as building area, ger area, forest, grassland, soil and water. For the final urban change analysis, the
changes occurred in Ulaanbaatar area in between 1969 and 1990 were compared with the changes occurred
in between 1990 and 2001 using a topographic map as well as multitemporal RS images, and the socio-
economic reasons for the changes were described. To extract the reliable urban land cover information
from the selected RS data sets, a refined parametric classification algorithm that uses spatial thresholds
defined from the local and contextual knowledge was used.

2. Test site and data sources

As a test site Ulaanbaatar, the capital city of Mongolia has been selected. Ulaanbaatar is situated in
the central part of Mongolia, on the Tuul River, at an average height of 1350m above sea level and
currently has nearly 1 million inhabitants [10]. The selected part of the capital city is about 28kmx20km
and in the selected area, such land cover classes as building area, ger area, forest, grass land, soil and water
can be identified.

- 69 -
 As the data sources, Landsat TM data of 10 September 1990 with a spatial resolution of 30m,
SPOT PAN data of May 1990 with a spatial resolution of 10m, and Landsat ETM (+) data of 31 August
2001 were used. The ETM data consisted of a panchromatic band resampled to a pixel resolution of 14m
and 7 multispectral bands resampled to a pixel resolution of 28m. In addition, topographic maps of 1969
and 1984, scale 1:50,000, and a general urban planning map were available. Figure 1 shows recent look of
the test area in the ETM image and some examples of its land cover.

Figure 1. ETM image of Ulaanbaatar area (Red=band5, Green= band4, Blue= panchromatic
band). The size of the displayed area is about 28kmx20km.

3. Radiometric correction and georeferencing of the multi-sensor images

At the beginning, all the available images were thoroughly analyzed in terms of radiometric quality
and geometric distortion. For the SPOT PAN data of 1990 some line dropouts occurred, while the
panchromatic band of the ETM data had some radiometric noise. In order to correct the error of the SPOT
PAN data, averaging of the upper and lower lines of the lines that had to be corrected was applied, whereas
for the panchromatic band of the ETM data, a 3x3 size average filtering [8,9] was applied. In case of the
multispectral images, band 1 (i.e., blue band) of both TM and ETM had noise from the atmosphere and it was
difficult to correlate their radiometric values with other bands. Therefore, they were excluded from the
analysis. Moreover, as band 6 of TM and bands 61 and 62 of ETM are thermal bands, they were excluded
from the further analysis, too.
In order to merge data sets with different spatial resolutions, a high geometric accuracy and good
geometric correlation between the data sets are needed. Initially, the panchromatic image of the ETM data
was georeferenced to a UTM map projection using a topographic map of 1984, scale 1:50,000. The ground
control points (GCP) have been selected on well defined cross sections of roads, streets and other clearly
delineated sites. In total, 16 regularly distributed points were selected. For the transformation, a second
order transformation and nearest neighbour resampling approach [7] have been applied and the related root
mean square (RMS) error was 0.92 pixel. In order to georeference other images, 16 more regularly
distributed GCPs were selected on each image comparing the locations of the selected points with other
information such as the already georeferenced panchromatic image of the ETM and the selected
topographic maps. Then, the images were successively georeferenced to the UTM map projection using the

- 70 -
topographic map of 1984. For the actual transformations, a second order transformation and nearest
neighbour resampling approach were applied. The RMS errors of the image transformations were 0.98 pixel
for the SPOT PAN, 0.65 pixel for the TM and 0.79 pixel for the ETM, respectively. In each case of the
georeferencing, an image was resampled to a pixel resolution of 14m.
4. The refined classification method

Over the years, multispectral RS data sets have been widely used for urban land cover mapping at
a regional scale and for the generation of land cover information, diverse classification methods have been
applied. The early methods mainly involved supervised and unsupervised methods and hence, many
techniques have been developed [3,4]. In this study, to extract the reliable urban land cover information
from the selected RS data sets, a refined parametric classification algorithm that uses spatial thresholds
defined from the local and contextual knowledge has been used.
Generally, in the classification process, it is desirable to include only the features in which the
signatures of the selected classes are highly separable from each other in a multidimensional feature space
[6,12]. In the present study, as the features, infrared bands of TM and ETM as well as panchromatic bands
were used. To define the sites for the training signature selection, initially, from the images, several areas
of interest (AOI) have been selected for each available class using the local and contextual knowledge. The
local knowledge has been defined from the historical GIS data sets (i.e. topographic and urban planning
maps), while the contextual knowledge was defined on the basis of the spectral variations of the land
surface features as well as the texture information delineated on the colour images. The separability of the
selected training signatures was firstly checked in feature space and then evaluated using JM distance [12]
and it revealed that high statistical overlaps exist between the classes: building area and ger area. Then, the
samples which demonstrated the best possible separability were chosen to form the final signatures. The
final signatures included about 126-498 pixels.

For the actual classification, the Mahalanobis distance classifier (MDC) has been used. The MDC is a
parametric method, in which the criterion to determine the class membership of a pixel is the minimum
Mahalanobis distance between the pixel and the class centre [7].

Initially, in order to check the performance of the standard method, the selected PC features were classified
using the MDC. However, on the classified images there were different mixed classes between the classes:
building area and ger area. This should be evident, because the previous signature analysis indicated that
the signature distributions of these classes had significant overlaps in the multidimensional feature space.

To separate these statistically overlapping classes, different spatial thresholds determined on the basis of
the local and contextual knowledge have been used. The local knowledge was based on the knowledge
about the site as well as the historical GIS data sets, whereas the contextual knowledge was based on the
spectral and textural variations of the selected classes in different parts of the colour images. To determine
the initial spatial thresholds, firstly the appropriate polygon boundaries related to the selected classes were
defined from the historical GIS data sets. As these polygon boundaries represented old information, it was
necessary to update them. For this purpose, new polygon boundaries were defined from the PC images on
the basis of the contextual knowledge (i.e., defining class boundaries in relation to its neighbourhood) and
added to the initial spatial thresholds. The results of the classifications using the defined spatial thresholds
are shown in figure 2a,b. For the accuracy assessment of the classification results, the overall performance
has been used [12]. As ground truth information, different AOIs containing the purest pixels have been
selected. The confusion matrices produced for the refined parametric classification method showed overall
accuracy of 91.98% for the 1990 data sets, while for the 2001 data sets it was 92.89%, meanwhile
indicating an accuracy of more than 90% for each of the selected classes. This means that these
classification results can be reliably used for further analysis.

5. Urban change analysis

In this study, we wanted to compare the general changes occurred in the main land cover classes (i.e.,
building area, ger area, forest, grass land, soil and water) of Ulaanbaatar area during the centralized
economy (i.e., in between 1969 and 1990) with the changes occurred during a market economy (i.e., in

- 71 -
between 1990 and 2001) using multitemporal RS images, and explain the socio-economic reasons for the
changes.
Initially, to create the primary historical GIS data, the classes: building area, ger area, forest and
water were digitized in a UTM map projection from a topographic map of 1969, scale 1:50,000. On this
topographic map, the areas related to the grass land were not delineated. Therefore, it was not possible to
distinguish between grassland areas and soil classes. The apartments, residential houses, industrial
buildings and all other building areas were included in the building area class, because on the RS images it
was not possible to distinguish among these classes, due to their very similar spectral characteristics.
To define the total area of each class, the vector map has been rasterized with a pixel resolution of
14m and then a number of pixels falling into each of the classes was calculated. Likewise, from the
classified PC images, the total number of pixels falling into each of the classes has been calculated. The
total areas related to each class defined from the digitized map as well as classified multitemporal RS
images are shown in table 1. Although, we have census data of Ulaanbaatar city, it is not possible to
directly relate it to the current analysis, because our study area does not cover all the areas from where the
final census data is collected. However, as the test area covers the majority of the area belonging to the
capital city, it is possible to use the census data for a comparison of the general population increase with
the actual urban expansion process.

Table 1. The total areas for each class in different years, evaluated from multitemporal GIS and RS
data sets.

Land cover classes The total areas for each class in different years (ha)

1969 1990 2001

Building area 2498.92 3996.71 4812.64
Ger area 978.86 2719.65 4304.26
Forest 11896.12 11075.94 10324.26
Grass land N/A 11626.26 9153.98
Soil N/A 26273.23 27359.04
Water 1078.84 716.40 454.01
Total 56408.19 56408.19 56408.19

As seen from table 1, in 1969 in Ulaanbaatar city, the building area and ger area covered 2498.92ha and
978.86ha, respectively, whereas in 1990 these two urban classes covered 3996.71ha and 2719.65ha,
respectively. Moreover, it is seen that the forest and water resources had been reduced. The available
census data indicated that in 1969 the population of the capital city was 267,400, while in 1990 it had
become 574,900. As seen, within the 21 year period of the centralized economy, the building areas were
increased by only 59.94%, whereas the ger areas were increased by more than 2 times. Meanwhile, the
population had been increased by more than 2 times. These changes are related with the following:

1) Due to the industrialization process, many people came from different parts of the country seeking for
better lives.
2) During the centralized economy, the government constructed mostly high-rise apartment blocks with
many stories to accommodate the growing population. These residential blocks, though could contain
families of many ger districts, occupied much smaller land parcels than the ger parcels. For example,
one apartment block with five stories which could accommodate 60 families occupied just about
1000sq.m, while one family living in a ger district occupied usually 600-800sq.m, because the gers in
urban areas are surrounded by fences.

- 72 -
Figure 2. Comparison of the classification results for the selected classes. (a) Classified image
using 1990 data sets, (b) Classified image using 2001 data sets.

- 73 -
3) During the centralized economy, many people came from rural sites with their gers to the city.
Although, the government constructed high-rise apartment blocks, they could not satisfy the growing
demands of the increasing population. Therefore, those people who could not get apartments usually
used their gers as dwelling houses. Therefore, the ger districts had such a significant expansion.

Furthermore, as seen from table 1, within the 11 year period since the country entered the market
economy, the building area and ger area had been increased by 20.41% and 58.26%, respectively.
Meanwhile, forest, grass land and water resources had been reduced, expanding the areas of bare soil. The
previous study, conducted by Amarsaikhan et al. (2001) had found that within a 7 year period of a market
economy, in the central part of Ulaanbaatar city the building area and ger area had been increased by 48%
and 50%, respectively. However, the current study covering much larger area than the previous study has
revealed that over recent years, the ger areas in the capital city have been significantly expanded,
specifically in the urban fringes. The reasons for these changes are:

1) Due to the market economy many people moved to the areas with good infrastructures, thus increasing
the population. For example, census data indicated that in 2001, the population of the capital city had
become 812,500.
2) Since the country entered the market economy, the government has not constructed new residential
apartments. The apartment prices of the existing construction companies are very high, so, most
ordinary people cannot afford them. Therefore, most people when moved to Ulaanbaatar built up their
gers surrounded by fences, thus significantly expanding ger districts.
3) During the market economy, most of the constructed buildings are western cottage style houses or
buildings with few stories, not like the former apartment blocks. Therefore, they occupied larger areas
than the high-rise apartments and also contributed to the expansion of the city areas.
4) At present in Mongolia there is boom to own land. Because, the new land law issued in 2002 [11]
gives a distinct right to land owners and they can use their land for receiving bank loans. Therefore, the
interests of people to own land parcels have greatly increased.

6. Conclusions

The overall idea of the research was to compare the changes in the main urban land cover classes
of Ulaanbaatar city, Mongolia occurred during a centralized economy with the changes occurred during a
market economy and describe the socio-economic reasons for the changes using multitemporal RS data
sets.

To extract the reliable urban land cover information from the available RS data sets, a refined parametric
classification algorithm that uses spatial thresholds defined from the local and contextual knowledge was
constructed. As seen from the classification results, the spatial thresholds defined from the local and
contextual knowledge could significantly improve the performance of the classification and for the
accurate classification, proper spatial thresholds should be applied.

As seen from the urban land cover change analysis, during the centralized economy significant changes
occurred in the ger area of the city, while during the market economy the changes occurred in both areas.
Moreover, as seen from the analysis, during all this time, the natural resources such as forest and water had
been reduced.

Because of the spatial resolution of the used RS data sets, the results of this study could be used for a
decision-making process at a regional scale. For detailed analysis, large scale maps and very high
resolution RS images covering all areas should be used.

- 74 -
 References

1. AMARSAIKHAN, D. GANZORIG, M. and SAANDAR, M., 2001, Urban Change Study Using RS and GIS,
Asian Journal of Geoinformatics, December 2001, No2, Vol2. pp. 73-79.

2. AMARSAIKHAN, D. and SATO, M., 2003, The role of high resolution satellite images for urban area
mapping in Mongolia. In ‘Reviewed Papers’ part of Proceedings of the Computers for Urban Planning
and Urban Management (CUPUM)’03 International Conference, Sendai, Japan, pp.1-12, May 2003.

3. AMARSAIKHAN, D. and DOUGLAS, T., 2004, Data fusion and multisource data classification,
International Journal of Remote Sensing, 17, pp. 3529-3539.

4. AMARSAIKHAN, D. and SATO, M., 2004, Validation of the Pi-SAR data for land cover mapping,
Journal of the Remote Sensing Society of Japan, No.2, Vol.24, pp. 133-139.

5. AMARSAIKHAN, D., GANZORIG, M. and Moon, T.H., 2005, Application of multitemporal RS and GIS
data for urban change studies, Proceedings of the Korean GIS Conference, Busan, Korea, pp.190-215.

6. BORAK, J. S. and STRAHLER, A. H., 1999, Feature selection and land cover classification of a MODIS-
like data set for a semiarid environment. International Journal of Remote Sensing, 20, pp. 919-938.

7. ERDAS, 1999, Field guide, Fifth Edition, ERDAS, Inc. Atlanta, Georgia.

8. GONZALEZ, R. C. and WOODS, R. E., 2002, Digital Image Processing, Second Edition, Upper Saddle
River, (New Jersey: Prentice-Hall).

9. MATHER, P.M., 1999, Computer Processing of Remotely-Sensed Images: An Introduction, Second

Edition, (Wiley, John & Sons).

10. MONGOLIAN STATISTICAL YEAR BOOK, 2006, National Statistical Office of Mongolia, Ulaanbaatar,
Mongolia.

11. NEW LAND LAW OF MONGOLIA, 2002, New Land Law of Mongolia, Ulaanbaatar, Mongolia.

12. RICHARDS, J.A., and JIA, X., 1999, Remote Sensing Digital Image Analysis-An Introduction, Third
Edition, (Berlin: Springer-Verlag).

- 75 -
 SURFACE WATER POLLUTION OF THE ULAANBAATAR CITY

(determined by BOD, dissolved O2, NH4+, NO2-, NO3-, PO4-3, Cr+6, COD)
(between 1996-2004)
(1) Professor, Ch.Gonchigsumlaa, (2) MSc, O.Altansukh
National University of Mongolia, Faculty of Earth Sciences,
Department of Geoecology-Land Management, altansukh22@yahoo.com

Key words:
Chemical analysis, surface water, water pollution, water quality, permissible limit, standard of water
quality

Introduction:
Chemical component of river water, is running near by Ulaanbaatar city (Tuul, Terelj, Uliastai,
Selbe and Dund rivers), divided two sections of water quality when compared with Mongolian standard of
water quality.
1. Section with fresh water - in permissible limit
2. Section with polluted water - out permissible limit

Method:
 Used the chemical analysis results of river water that did in the Central Laboratory of
Environmental Monitoring.
 Compared with Mongolian standard of water quality
 Drew by ArsView 3.3

Result:
1) Section with fresh water - in permissible limit
 Tuul-Uubulan
 Tuul-Nalaih
 Tuul-Bayanzurh
 Tuul-Zaisan
 Tuul-Sonsgolon
 Tuul-Upper Songino
 Terelj-Terelj
 Uliastai-Ulaanbaatar
 Selbe-Ulaanbaatar

2) Section with polluted water - out permissible limit

 Tuul-Lower Songino
 Tuul-Shuvuun fabric
 Tuul-Khadanhyasaa
 Tuul-Altanbulag
 Dund-Ulaanbaatar

Chemical component of river water increased downstream until city center. Near by Ulaanbaatar city, river
water quality is decreasing when it is running through the city. There is having several reasons:
1) Industrial waste water is pouring into river
2) Increasing number and density of population in city center
3) Decreasing natural water cleaning capacity near city
In addition, there is having a biggest factor of water pollution. This is a central sewage treatment station of
city. This station is source of section with polluted water.

Suggestion:

- 76 -
  Immediately, to change technique and technology of central sewage treatment station.
 To improve industries treatment station.

Surface water pollution of the Ulaanbaatar city

BOD, P, dissolved O2, NO-3, NO-2, NH+4, COD, Cr+6
(between 1996-2004)
N

W E

ãîë
èéí
ý ëæ
Òý ð
Á óðä

S
ãî
ë
Íà

ë
ð

í ãî
èé
í èé

Ìàíäà ëû
ãîë

ë Íàð èéíèé ãîë

ãî
é
éíè
îë

à ðè
Í
Õ óéí ã

Ø
àâ
àð
ò
ãî
ë

ë
ãî
îë
ýã

í
á

èé
ýë

æ
Ñ

ýë
ð
Òý

Ó ëè
à ñò
àéí ãîë
Á
àð
ó
óí
ãî

ë
ãîë

í
óó
Ç

ë
ë
ãî

í ãî í
óé à éò û
Õ íäã
Õà
Òî
ë
ãî

éò
ûí
ë
îë ãî
ã

ÿí
áà
éä
Õî

Õà÷óóð ãîë
Ñý ëá

ã îë
í
ý

àÿ
äá
ãî ë

Òý
Óð ð ý
ë
Ø

æè
æè
ð í
Òàõèë

é
è ãî ë
í í ãîë
àò û
é

ãî ë óã

ãîë

Á
òûí

í
ðò û
îë
àðã

íã

àâà
ãîë
îë

Ñ ýë

é
õè

Ø
àéí

Øèëæ èð
õ ãî

ãý

ãîë
Áèë

àñ ò
ë

Óëè
Çóóí ñàëàà ã îë
Áàð

#
óó
íñ

îë
àëà

íã

Òýðýëæ - Òýðýëæ ãî

ë
à ãîë

é
õè óë
ã Òó
ý

Áèë

èéí ãîë
Òýðý ëæ

ðõè
Ñý ëáý ãî ë

ãî
Ãà÷óóðò ûí ã îë

Àð
í ãîë

Òî ëã îéòûí ã îë

ë
Óëèàñò àéí ãîë

ãî
Òàõèëò ûí ãîë

ûí
àÿ

ä
Á

àí
õó

Óëààíáààòàð
Èõ

Ãà÷óóðò

Ñýëá ý - Óëààíá ààòàð Óëèàñòàé - Ó ëààíáààòàð

ã îë

#
Ä óíä ãîë
# #
Ò ó óë

ãîë
Äóíä ãîë - Óëààíá ààòàð

Òóóë
ë î ë
ãî

Ò ëã

# óó
ãîë
íä
Äó

éí
äè

#
ýí

Òóóë - Çàé ñàí

ðõè
Ò îë
Ä

óó ë Òóóë - Á àÿíçóðõ è
ã îë éí

ãî
ã
î

ð
âî
ë

Òóóë - Ñ îíñã îëîí Î

Òóóë - Ñ îíã èíû äî îä

#
# î ãí è
Ñ í î
Òóóë - Ñ îíã èíû äýýä
è
Òóóë - Í àëàéõ
û í ã îð õ
àò
ìá ã îë
ë

#
ó
Á

ó
#
Òó
Ò

óó
ë ãî
ë

ë Òóóë - Óóá óëàí

ãî
í
èé

ë
àíã

î
ã
í
ñ

û
àé

ç õ
îð õóð
Òî ð
Îâ

Õîë í ãî ë

Ýë
ñò

ãî
èé
ë

ë
ãî

é
Òóóë - Ø óâóóí ôàáð èê íè
ë ãý
ãî Ò óð Ýëñò èé
é
íè í ãî
ãý

#
ë
Ò óð Íàëàé õ
Òóóë - Õ àäàíõÿñàà

# óë ãî ë

# Research point
ûí ã îë

Òó
Ý ëñòã îë
Í àëàéõ

Òóóë - À ëòàíáóëàã
Clean water - in permissible limit
Polluted water - out permissible limit
Á
óõ

# ýã
ãî
ë

ã îë
Ò óóë

Unstudying river
Studying river
Tehnogen area
Territory of the ulaanbaatar city

- 77 -
 Analysis on the Land Use in General and Neighborhood Commercial Areas of
Ulaanbaatar city using RS and GIS3
B.Chinbat1, D.Amarsaikhan1,2 and Tae-Heon Moon3

1
Faculty of Geography, National University of Mongolia, Ulaanbaatar, Mongolia
E-mail: chin_bat2002@yahoo.com
2
Institute of Informatics and RS, Mongolian Academy of Sciences
av.Enkhtaivan-54B, Ulaanbaatar-51, Mongolia
E-mail: amar64@arvis.ac.mn
3
ERI, Urban Engineering Major, Construction Division, Gyeongsang National University
900 Gazwa Jinju City, Gyeongnam 660- 701 Korea
E-mail: thmoon@nongae.gsnu.ac.kr
Abstract
The aim of this study is to analyze the land use of general and neighborhood commercial areas of
Ulaanbaatar city using RS and GIS techniques. For this purpose, two urban land use sites
representing these areas have been selected in different locations. For the analysis, historical GIS
data as well as SPOT 5 and Quickbird images of 2002 have been used. The analysis was carried out
using the ArcGIS and Erdas Imagine 8.6 installed in a PC environment and to reach the final goal,
different RS and GIS techniques have been applied.

Keywords: Land use, commercial area, RS, GIS, analysis

1. Introduction

In recent years, Ulaanbaatar, the capital city of Mongolia has faced different urban development
problems, similar to many cities in developing countries. In the city, various problems had been
accumulated during the communist era and they have been accelerated by the reforms of the entire political
and economic systems, unregulated market development and the rapid population growth caused mainly by
migration from rural areas.
Since the transition to a market economy, the Ulaanbaatar city has experienced much more
developments, which resulted in changes of the spatial and functional structures of the city and the most
significant changes have been the increase of commercial functions in the city centre and inner city area;
the expansion of the urbanized areas along with the growth of formal and informal ger-settlements; the
formation of satellite nodes with clusters of commercial functions, and the residential suburbanization in
the outer city by single family houses [5].
At present, in the country there are missing urban-oriented research activities, based on the modern
urban geographical theory and methodologies, because the research on detailed urban studies, including
functional and spatial differentiation of urban areas is a relatively new research direction in urban
geography of Mongolia. One of the fundamental problems in urban study could be the research on how
different urban features with various profiles and duties can be spatially located better coping with each
other in order to be ecologically, economically and socially efficient and satisfy the requirements of the
sustainable development [4,5].
In this study, we wanted to analyze the land use of general and neighborhood commercial areas of
Ulaanbaatar city using RS and GIS techniques. For this purpose, 2 urban land use sites representing these
areas have been selected and for the analysis, historical GIS data as well as SPOT 5 and Quickbird images
of 2002 have been used. The analysis was carried out using the ArcGIS and Erdas Imagine 8.6 installed in
a PC environment.

2. The selected sites and data sources

In this study, as a general commercial area III-IV microdistrict located in the north western part,
whereas as a neighborhood commercial area Sansar microdistrict located in the central part of Ulaanbaatar
city have been selected. The locations of these sites represented in a SPOT 5 image of 2002 are shown in
Figure 1.

3
Paper to be published in Proceedings of 2nd International Conference on Land cover /Land use study using RS and GIS

- 78 -
 As the RS data sources, multispectral SPOT 5 image of 2002 resampled to a pixel
resolution of 4m and Quickbird image of 2002 with a spatial resolution of 70cm have been used. In
addition, a topographic map of 1984, scale 1:50,000 and a topographic map of 2000, scale 1:10,000 as well
as a general urban planning map were available. The digital forms of these maps were considered as
historical GIS data sets.

Figure 1. SPOT 5 image of Ulaanbaatar city (Red=band 1, Green=band 3, Blue=ban 2).

1-General commercial area (III-IV microdistrict),
2-Neighborhood commercial area (Sansar microdistrict).

3. Georeferencing of the SPOT 5 and Quickbird images

In order to georeference the Quickbird image to a Gauss-Kruger map projection, a topographic
map of 2000, scale 1:10,000 has been used. The ground control points (GCP) have been selected on well
defined cross sections of roads, streets and building corners and in total, 12 and 9 regularly distributed
points were selected for the III-IV microdistrict and Sansar microdistrict, respectively. For the
transformation, a linear transformation and nearest neighbour resampling approach [6,8] have been applied
and the related root mean square (RMS) errors were 1.56 pixel and 1.79 pixel, accordingly. Likewise, two
subsets selected from the multispectral SPOT 5 image have been georeferenced to a Gauss-Kruger map
projection [3] using the same topographic map of the test area. For the transformation the same number of
GCPs has been used and the related RMS errors were 1.21 pixel and 0.98 pixel, accordingly. In each case
of the georeferencing, an image was resampled to a pixel resolution of 70cm.

4. Image fusion
In the present study, in order to enhance the spectral and spatial variations of different land
use classes as well as to merge the images with different spatial resolutions, two image fusion techniques
such as Brovey transform and intensity–hue–saturation (IHS) transformation have been used and compared.
After applying corrections, data with different spatial resolutions can easily be integrated. The image
fusion is the integration of different digital images in order to create a new image and obtain more
information than can be separately derived from any of them [2,9]. In the case of this study, the
panchromatic image provides more spatial information due to its higher spatial resolution, while the
multispectral images provide the information about the spectral variations of the urban classes. Image
fusion can be performed at pixel, feature and decision levels [1,9]. In this study, the fusion has been
performed at a pixel level. Before applying the fusion techniques, a 5x5 size high pass filtering [7,10] has
been applied to the panchromatic images in order to enhance the edges.
Brovey transform: In this method, multispectral images with a lower spatial resolution are
integrated with an image with a higher spatial resolution, thus creating spectrally and spatially enhanced color
images [11]. To create spectrally and spatially enhanced color (RGB) images, the sum normalized
multispectral bands are multiplied by the image with a higher spatial resolution as shown below:

- 79 -
 R=(B1/B1+B2+B3)*B4
G=(B2/B1+B2+B3)*B4
B=(B3/B1+B2+B3)*B4

where B1,B2 and B3 - multispectral bands, and B4 - band with a higher spatial resolution.
In the present study, for the Brovey transform, the bands of SPOT 5 were considered as multispectral bands,
while Quickbird image was considered as higher spatial resolution band.

IHS transformation: The IHS method is the most widely used data fusion technique. This method
assumes that the H and S components contain the spectral information, while the I component represents
the spatial information [1,8]. A detailed review of this approach is given in Mather (1999). For the IHS
transformation, the RGB image created by green and near infrared bands of the SPOT 5 data as well as
panchromatic band of Quickbird data have been used and the panchromatic band was considered as the I.
When the IHS image was transformed back to the RGB colour space, contrast stretching has been
performed to the I and S channels.
In order to obtain a reliable color image that can illustrate the spectral and spatial variations of the
selected land use classes, different band combinations have been used and compared. Although, the images
created by the Brovey transform contained some shadows that were present on the panchromatic images,
they still illustrated good results in terms of separation of the available land use classes. The images created
by the IHS method contained less shadow effects, however, it was very difficult to analyze the final
images, because they contained too much color variations. Therefore, for the interpretation of the selected
land use classes, for both test sites, the images created by the Brovey transform have been used. The fused
images of SPOT 5 and Quickbird are shown in Figure 2.

- 80 -
 Figure 2. The fused images of SPOT 5 and Quickbird data.
a) and c) Brovey transformed images, b) and d) IHS transformed images.

- 81 -
5. Land use analysis of general and neighborhood commercial areas

At the beginning, from the Brovey transformed images of both test sites, the available land
use types have been digitized using the ArcGIS system (Figure 3).
The general commercial region consists of the trade and service street stretched about 2 km along
the Ayush avenue, which separates the III and IV microdistricts. As seen from the results of the
interpretation, in this region the high rise and middle rise residential houses occupy 15,2 hectares (ha) or
49,9%; general educational schools, crɢches and kindergartens occupy 6,39ha or 21%, trade and services
areas occupy 3,75ha or 12,4%; research institutions, schools occupy 1,3ha or 4,3%; companies and banks
occupy 0,85ha or 2.8% of the total area, thus forming the prevailing portion of the land use. These
microdistricts which were constructed in 1970s and 1980s consist of several residential apartment blocks
and neighborhoods together with schools, kindergartens, trade and services centers. The average radius of
the apartment or neighborhood was planned to be 400-500 meters and their trade and services centers to be
located neighboring each other in the Ayush avenue. During the process of the transition period which
started in 1990, on the bases of the old apartment blocks, trade and services buildings there were centered
many new types of land use such as supermarkets, garment shops, electronics, cosmetics diversified
groceries, department stores, banks, companies as well as universities and institutes, hotels and restaurants.
As results of this, this street started to play the role of a general commercial or commercial center with own
local character. Moreover, inside of the residential blocks of the microdistricts there were established and
opened a great number of offices, hotels, restaurants, night clubs, private universities and institutes,
supermarkets, which diminished the environmental value and quality of the given residential area. In
addition, one of the negative phenomena occurring in the current land use of the residential areas is that
there are being constructed too many low-rise buildings designated for auto garages and other construction
facilities of cheaper value and quality. This deteriorates the friendly environment and living conditions of
the people inside of these areas.

Neighborhood commercial zone refers to commercial area within the residential area. As seen from the
results of the interpretation, in this microdistrict, the middle rise apartments occupy 1,9ha or 42,2%; high-
rise residential houses occupy 0,61ha or 13,5%; trade and services facilities occupy 0,234ha or 5,2%;
general educational schools, crɢche and kindergartens occupy 0,69ha or 15,3% of the total area. The
service radius of the given service center is 500 m. The services center was planned from the beginning
and constructed with a supermarket, grocery shop for consumer goods, bookshop and communal services
center. During the transition period, all of these including the sewing, hairdressing, photographic, postal,
cinema and dry cleaning services as well as TV repair shop were privatized. After the privatization, the
profile and designation of significant part of these services had been altered and changed to night clubs,
entertainment places, hotels and so on. Such a change deteriorates the surrounding environment and makes
too much noise, thus worsening the living conditions of people living in this region.

6. Conclusions

The aim of this study was to analyze the land use of general and neighborhood commercial areas of
Ulaanbaatar city using RS and GIS techniques. For this purpose, two urban land use sites representing
these areas have been selected in different locations. For the analysis, two urban land use sites were
selected and historical GIS data as well as SPOT 5 and Quickbird images of 2002 were used.

As seen from the analysis, since the irreversible transfer of the country into the market economy,
commercialization became the most important process in these regions and it has influences on the changes
of spatial and functional structures. Also, it is seen that there are being formed new types of land use which
might deteriorate the residential zones of the urban population. Therefore, thorough urban planning and
management based on modern theory and methodologies are urgently needed.

- 82 -
Figure 3. The land use types interpreted from the Brovey transformed images.
a) General commercial area, b) Neighborhood commercial area.

- 83 -
References

1. Amarsaikhan, D., and Douglas, T., 2004, Data fusion and multisource data classification,
International Journal of Remote Sensing, No.17, Vol.25, pp.3529-3539.
2. Amarsaikhan, D., Ganzorig, M. and Moon, T.H., 2005, Application of multitemporal RS and GIS
data for urban change studies, Proceedings of the Korean GIS Conference, Busan, Korea, pp.190-
215.
3. ArcGIS, User’s guide, ESRI, Inc. Atlanta, Georgia, USA.
4. Chinbat, B., 2005, On a new land use classification and zoning scheme of Ulaanbaatar, Mongolia,
Proceedings of Mongolia-Korea Conference in Urban Planning, Ulaanbaatar, Mongolia.
5. Chinbat, B., Bayantur and Amarsaikhan, D., 2006, Investigation of the internal structure changes of
Ulaanbaatar city using RS and GIS, Paper presented at the ISPRS Mid-term Symposium, ITC,
Enschede, The Netherlands.
6. ERDAS, 1999, Field guide, Fifth Edition, ERDAS, Inc. Atlanta, Georgia.
7. GONZALEZ, R. C. and WOODS, R. E., 2002, Digital Image Processing, Second Edition, Upper Saddle
River, (New Jersey: Prentice-Hall).
8. MATHER, P.M., 1999, Computer Processing of Remotely-Sensed Images: An Introduction, Second
Edition, (Wiley, John & Sons).
9. POHL, C., and VAN GENDEREN, J.L., 1998, Multisensor image fusion in remote sensing: concepts,
methods and applications. International Journal of Remote Sensing, 19, 823-854.
10. RICHARDS, J.A., and JIA, X., 1999, Remote Sensing Digital Image Analysis-An Introduction, Third
Edition, (Berlin: Springer-Verlag).
11. VRABEL, J., 1996, Multispectral imagery band sharpening study. Photogrammetric Engineering and
Remote Sensing, 62, 1075-1083.

- 84 -
 The Heat Island Experiment over the Western Taiwan Plain with MODIS
satellite and concurrent helicopter-borne IR imager data

Chia Wei Lan*, Gin-Rong Liu, Tsung-Hua Kuo, Kun-Wei Lin, Tang-Huang Lin, Ming-Chang Hsu, Yen-Ju
Chen, Chia-Chi Liu
Center for Space and Remote Sensing Research, National Central University, Jhong-Li, Taiwan
320, Corresponding author: Chia Wei Lan, 93621002@cc.ncu.edu.tw

Abstract
Previous studies have shown that the urban air temperature can be 4 к hotter than the
surrounding rural areas, which is usually referred to as the “heat island effect”. The effect is mainly
induced by a combination of human activities and the different types of landcover/landuse.
Different types of landcover/landuse render different specific heat values of the land surface. The
higher temperature produced from the effect could have a serious impact on the local, regional, and
even global environment. It could also affect the public’s health, and waste a country’s energy
resources. Furthermore, the heat island effect is a very important factor in global change studies,
because it can substantially influence weather and climate systems.
With the rising concern over the impact from the heat island effect, a series of observations were
conducted over western Taiwan. Concurrent datasets from the satellite-borne MODIS sensor and
helicopter-borne thermal IR imager were both employed. In addition, the ground truths from
surface stations and automatic weather recorders were also utilized during the same time period.
The sampling data in this experiment covered an area of more than 100 km2.
Based upon these comprehensive observations, a heat island pattern of Taiwan’s western plain
was delineated. The relationship between the temperature increase rate and landcover types was
also investigated. The result shows that the surface canopy and the air humidity together play an
important role in the surface temperature increase. In this mid-term report, the procedure in the
ground truth observation, along with the study’s preliminary results is essentially explained.

Keywords: Heat island, MODIS, Landcover/Landuse

1. Introduction
A growing amount of evidences obtained from ground truth and satellites show that the influence
from human activities can seriously affect the earth’s climate and biosphere (Gallo et al., 1999; Hung et al.,
2006). For example, Taiwan nearly has the highest population density over the world. More than
20,000,000 people live in an island, which covers an area of about 36000 km*2. Unfortunately, more than
2/3 area of this island is mountainous terrain, rendering most of the population to be situated within the
western plains of the island. The urbanization evidently changes the pattern of landuse/landcover types and
causes the so-called heat island effect (Hoard, 1833, Kato et al., 2005; Rosenzweig et al., 2005, Streutker,
2003).
Therefore, a heat island observation experiment covering a large area was conducted for the first time
in Taiwan to understand the urbanization influence in the heat island effect on Taiwan’s western plains.
Observations were conducted by the MODIS satellite, coupled with ground truths collected from weather
stations and helicopter-borne thermal IR Imager observations.

2. Data
The data used in this study are shown as follows: Helicopter-borne thermal IR imager (ThermoVision
A40M), MODIS, and weather station data. The first two platforms supplied the thermal IR images for the
surface temperature retrieval, while the weather stations provided the surface temperature, humidity, and
wind speeds.

- 85 -
 The concurrent observations were conducted on 2006/02/22. Most of the IR images were taken between
the altitude of 600 to 800 meters.

3. Result
Since several of the different landcover types could generally appear during a FOV scanning
procedure, the statistical medium values gathered from this analysis could be considered as the nominal
temperature of the single scanning observation.
The IR images acquired by the helicopter could be converted into surface temperature maps. The maps
clearly showed that the surface temperature was strongly dependent with the corresponding landcover
types. In order to obtain a comprehensive picture of the heat distribution in this large area, the MODIS data
was utilized. From the thermal bands in the MODIS platform, a large land surface temperature (LST) map
could be obtained (Figure 1).
The temperature map clearly revealed that the left part of the test area obviously had a higher
temperature distribution than the right portion, and the temperature values decreased as the vegetation
canopy increased. The temperature values varied from 18 to 36 °C.
The highest values occurred in the eastern parts of Pu-Tzu, while the lowest values were seen in the
eastern part (Ta-Hu and Chang-Nao-Liao) of the observation area. This probably was caused by the low
cloud layers and higher landscape. Once again, we used the MODIS red and near IR bands to calculate the
NDVI map, which was a simple but efficient indicator for finding the dense or isolated vegetation areas
(Figure 2). The contrast between Figure 1 and 2 demonstrates that the main surface heating was caused
from deforestation, reduction of vegetation areas, and increment of artificial surfaces.
After obtaining the surface temperature and vegetation data, the relationships between the MODIS
LST, IR Imager LST and station measured temperatures for the sampling locations can be seen in Figure 3.
The line plots showed a fine consistency between the three land surface temperatures.
It is reasonable that the results showed a high correlation between the temperature and humidity (not
shown). However, the temperature showed a null relationship with the wind speed. More detailed analysis
regarding these parameters and the roles they play in the heating process and flux mechanism should be
investigated in the future. Needless to say, the (decreasing) change of the vegetation canopy (with a
correlation coefficient, R=0.93) plays a very important role in the growth of heat islands (Figure 4).

4. Summary
This study has so far successfully conducted one helicopter observation during a very clear sky
condition. Basically, the thermal temperature derived in the MODIS and IR Imager data acquired in
February of 2006 were seen as a dataset representing a winter heat island pattern. More detailed analysis is
currently undergoing. To understand the seasonal influence, another helicopter observation is being
planned around this June or July to obtain a second dataset in delineating the summer heat island pattern.
In addition, the temperature difference from this summer data could aid in the investigation of the
variations between the atmospheric parameters.
Based upon our current analysis, some conclusions can be drawn out. Both from the helicopter and
satellite observations, the heat island pattern is obvious in the western Taiwan Plains. The roles of the
surface canopy will be detailed further for investigation. The influence of the atmospheric effect in the
MODIS and IR Imager observations must be computed or modeled from the radiation transfer models.
Meanwhile, the influence of the FOV resolution to the mean surface temperature should be seriously
treated. If possible, an accurate model taking into account the influence of aerosols, artificial surfaces, and
surface waterforms is recommended to compute the heat flux, where a better understanding of the sources
and sinks of the heat island heating process can be obtained.

Acknowledgements

- 86 -
 We would like to gratitude to the Academia Sinica in supporting the grans of theme project “Heat island
effect over Taiwan’s western plain and its impact on climate changes”, project number is AS-93-TP-A04.
In addition, we appreciate the help of staffs in Academia Sinica and Remote Sensing Center in NCU.

References
1. Gallo, K., T. W. Owen, 1999, Satellite-based adjustments for the urban heat island temperature
bias. Journal of Applied Meteorology, 38, 806-813.
2. Howard, L., 1833, The climate of London, Vols. I-III, Harvey and Dorton.
3. Hung, T. D. Uchihama, S. Ochi, Y. Yasuoka, 2006, Assessment with satellite data of the urban
heat island effects in Asian mega cities. International Journal of Applied Earth Observation and
Geoinformation. 8, 34-48.
4. Kato, S., Y. Yamaguchi, 2005, Analysis of urban heat-island effect using ASTER and ETM+ data:
separation of anthropogenic heat discharge and natural heat radiation from sensible heat flux. 2005,
99, 44-54.
5. Rosenzweig, C., W. D. Solecki, L. Parshall, M. Chopping, G. Pope, R. Goldberg, 2005,
Characterizing the urban heat island in current and future climates in New Jersey. Environmental
Hazards, 6, 51-63.
6. Streutker, D. R., 2003, Satellite-measured growth of the urban heat island of Houston, Texas.
Remote Sensing Environment, 85, 282-289.

- 87 -
Figure 1 The land surface temperature map derived by Figure 2 Same as Figure 1, but an NDVI map.
concurrent observations from MODIS thermal
IR bands.

Temperature & NDVI

NDVI to MODIS-LST
40 0.5
40
Temperature (degree)

35 0.4
30 35
0.3
LSTΰdegreeα

25 30
NDVI

0.2
20 25
0.1
15
10 0 20
5 -0.1 15
0 -0.2 10 y = -24.067x + 32.696
5
uo n
CY u
N o

zu

2
Pu ao
N ou

La u
Ta u

Y ou
iao S
Ch H -Ko

K Hsi
o
ia

-H

R = 0.8804
P

N G
i-T

Li
-T
g- -K

-K
ei-
-L
o

u-
-
an is
ao
A

MODIS-Lst 0
IR-avg
AT _1m
-0.1 0.1 0.3 0.5
Station NDVI NDVI
Figure 3 Relationship between the different surface Figure 4 The relationship between the MODIS
temperatures with the MODIS-derived NDVI measured surface temperature with the NDVI
index. values.

- 88 -
 Global vegetation continuous field tree cover products and Siberian forest
types

M. Herold (1), D. Knorr (1), K. Kornhaə (1), A. Shvidenko (2), O. Cartus (1) and C. Schmullius (1)
(1) Department of Geoinformatics and Remote Sensing, Friedrich Schiller University, Jena, Germany
(daniela.knorr@uni-jena.de)
(2) International Institute for Applied Systems Analysis, Laxenburg, Austria

Global vegetation continuous fields (VCF) products offer a different perspective on land
surface characteristics than traditional discrete classifications. By presenting each pixel as a
percent coverage, spatial heterogeneity may be better represented with significant advantages for
vegetation modelling. A common VCF product representing the percentage of tree canopy cover
per pixel was generated from monthly MODIS composites. Currently, this product is available for
2001 with a spatial resolution of 500m. The following four years with a resolution of 250 m are
expected to be available in the near future
(http://glcfapp.umiacs.umd.edu:8080/esdi/index.jsp?productID=20). Despite the obvious potential
of such global datasets, there is only limited validation and insufficient comparability with
traditional forest inventory data. Thus, this study aims at building understanding and confidence
in VCF tree cover products by a systematic comparison and analysis with large scale boreal forest
inventory information.

In the framework of the EU funded project SIBERIA-II, the International Institute for Applied
Systems Analysis (IIASA) provided a GIS based vector database of 74 test sites at scale 1:50.000
with detailed forest inventory data as well as an aggregated vegetation data base at scale 1:1 Mio.
The latter covers the entire 3 Mio km² SIBERIA-II study region in Central Siberia. This provides
unique database to verify the VCF tree cover product with detailed inventory information and for
a large region covering the full range of boreal forest types.
Empirical attempts to find statistical relations between the VCF tree cover, representing
canopy closure, and forest parameters are promising and show both potentials and obvious
limitations of the datasets. The best linear regressions could be found between the VCF tree cover
and tree height (R² = 0.44) and growing stock (R² = 0.35) including all data available. However,
the regressions strongly vary for different tree species and eco-regions of this large study area.
Prominent correlations between VCF tree cover and growing stock could be found for larch (R² =
0.53). Multivariate geographically weighted regressions showed even higher correlations and
reveal interesting spatial patterns. 63% of large residuals found in the multivariate geographically
weighted regression coincide with forest disturbances detected by satellite remote sensing. Further
investigations provided initial understanding of the link between ERS-SAR coherence, VCF tree
cover, and forest inventory information.

- 89 -
 Possible use of fuzzy-knowledge for improving the geographic boundary
representation
Sang-Jun Kim1, Ju-Whan Kang2, and Soung-Yong Yun3
1
Department of Civil Engineering, Kyungwon University, Seongnam, S. Korea 461-701
Tel) +82 031-756-0248, Fax) +82 031-756-0248, Email) sj282kim@kyungwon.ac.kr

2
Department of Civil Engineering, Mokpo National University, Mokpo, S. Korea 534-729
Tel) +82 061-450-2473, Fax) +82 061-452-6468, Email) jwkang@mokpo.ac.kr

3
Department of Civil Engineering, Ansan College of Technology, Ansan, S. Korea 425-080
Tel) +82 031-490-6077, Fax) +82 031-490-6075, Email) yun8306@ansantc.ac.kr

Abstract

The polygon boundaries on the digital map of land surface characteristic are conventionally
represented as a sharp change and this leads to discrepancy between real world phenomena and the
information presented by boundaries on digital map. As an alternative to the sharp-change
representation commonly seen in crisp thematic map, this study presents a fuzzy-based method to
represent the possible gradual transition along the geographic boundary in the numerical modeling
approach. To test acceptability of the newly suggested method, the Revised Universal Soil Loss
Equation (RUSLE) was performed with emphasis on the soil erodibilty factor (K). The new
approach was facilitated at a small basin in Korea and the results of the RUSLE using the
conventional sharp edged boundaries and the fuzzy boundaries are tested against field
measurement. The results show the fuzzy representation of geographic boundary provided better
performance.

Keywords: Fuzzy, GIS, Soil Loss, RUSLE

1. Introduction

In typical fashion, the digital map is represented as a categorical format and polygon boundaries
delineate and thereby distinguish areas with different surface characteristics. Conventionally the polygon
boundaries on the digital map are represented as a sharp change, which results in discrepancy between real
world conditions and the information presented by boundaries on map (Burrough, 1986; 1992). Each zone
with an abrupt line is a cadastral map where abrupt boundary definition is required to differentiate land
parcels that have unique property (Hunter and Williamson, 1990). It is especially true for soil properties
because boundaries of soil properties are continuously changing and rarely sharp or crisp. In reality,
localized partial change, gradual transition, and other non-sharp changes coexist in soil properties
(Burrough, 1986; Burrough and Frank, 1996).
The fuzzy representation of geographic boundary is an advanced geographic boundary
representation which enhances the expressive properties of polygons.
In this study, a new approach is attempted to improve the representation of geographic boundary in the
RUSLE model, examining probable impact of the representation of geographic boundary on soil erosion.
The model results show the fuzzy boundary provides better performance and the impact of the fuzzy
representation on the RUSLE model is considerable.

2. Numerical Experiment

The RUSLE is facilitated at the Yongdam basin, southern part of Geum river. The center of the
basin is 127° 33ƍ E, 35° 45ƍ N, which is about 200km south of the capital of Korea and it covers about
575.21 ໊ . Its annual average temperature and humidity are approximately 14ଇ and 74%, respectively
and its annual average precipitation (=1,238mm) is slightly lower than the Korean national average
(=1,283mm).

- 90 -
The RUSLE model computes soil erosion using six major factors which describe land surface
characteristics (e.g. soil erodibility, topography, and land use and management) and meteorological
conditions (e.g. rainfall erosovity). The cell-based representations of map features used in the RUSLE offer
analytical capabilities for continuous data and allow fast processing of map layer (Fernandez et al., 2003).
The mean annual gross soil erosion is calculated on the cell basis using the combination of the product of
six factors as follows;

A Ru K u Lu S uC u P (1)

where A denotes the average soil loss due to water erosion (in ton ˜ ha 1 ˜ yr 1 ). R denotes the rainfall and
runoff factor (in MJ ˜ mm ˜ ha 1 ˜ yr 1 ). The soil erodibility factor, K reflects the ease with which the soil
is detached by splash and surface flow.
The RUSLE describes topographic effect by means of the L- and S- factor, which accounts for the effect of
slope length and slope gradient on erosion, respectively. The C-factor reflects the effects of cropping and
management practices on erosion rates. The P-factor is a reflection of soil loss due to the flow pattern
change, gradient, direction of surface runoff, and reduction of runoff rate resulting from variable
cultivation and particular support pratices (Renard and Foster, 1983). The L, S, C, and P are all
dimensionless. The amount of soil loss generation is calculated on a yearly basis for the spatial resolution
of 22m in this study.
It is hard to directly measure soil erosion on a basin scale in an effective way and the measurement of the
basin sediment yield was used for the study. The basin sediment yield can be defined as the quantity of
sediments which is routed to the basin outlet for a certain time of period. Considering that only some of the
eroded soils are routed to the basin outlet, knowing the ratio between the basin sediment yield at the basin
outlet and soil erosion over the basin, which is called sediment delivery ratio (SDR), is important for the
decision makers but the SDR is involved in numerous uncertainties including temporal discontinuity and
spatial variability (Wolman, 1977; Walling, 1983). The RUSLE calculates soil loss forced by rainfall but
doesn’t take the sediment yield into account. To generate the sediment yield at the outlet, an equation (6)
for SDR, which is an empirical equation derived form the filed experimental data carried out in Korea
(KICT, 1992), was introduced as follows;

YJ ( R u K u L u S u C u P) u SDR (2)

where YJ denotes the unit sediment yield and

0.577
SDR 152.581u Aba sin (3)

where, SDR is the sediment delivery ratio and Aba sin is the basin area. The SDR physically means the ratio
of the sediment routed to the outlet over the basin (both overland and channel).

Figure 1 shows spatial distribution of the RUSLE factors and table 1 shows basic statistics of the six
factors. The rainfall erosivity factor, R is in the range of 413-563. The mean values of the L- and S- factor
are 2.695 and 3.864, respectively, while the standard deviation (SD) of the L- and S- factor are 1.904 and
3.307, respectively. The SD of L- and S- factor is relatively large and it might be a direct reflection of high
topographic variation. The average C- factor was evaluated as 0.028, while the P-factor factor was
estimated as 0.759.

3. Theory

Three different types of models are, in general, used to effectively represent geographic
boundaries in GIS; abrupt change (Type I), large change (Type II), and gradual change (Type III)
(Vincent, 1991; Wang and Hall, 1996).

- 91 -
The membership function of a set defines how the 'grade of membership' of an individual with an
attribute value of x is determined. The membership function converts attribute values x to
membership function values ( MFx ). For conventional crisp sets of the polygon boundaries on the
digital map, the membership function can be represented as follow;

MFx 0 for x  b1
MFx 1 for b2 t x t b1 (4)
MFx 0 for x ! b2
where b1 and b2 define the exact upper and lower limits of the set.
For fuzzy sets, the limits b1 , b2 define the central concept of the set. The fuzzy membership
function (FMF) defines how the possibility of membership varies continuously from 0 (for
individuals that are completely outside the set) to 1 (for objects that within the central concept).
The attribute value at the point where 'the grade of membership = 0.5' is called the 'crossover
point’ (Burrough, 1992).
Rather than the binary membership conditions of classic set theory (1 or 0), a fuzzy membership
condition allows more realistic modeling of geographic properties with high spatial within-class
variability, whereby membership grades accommodate the extreme classical case, as well as all
other possibilities in between (Wang and Hall, 1996). The following fuzzy classification models,
which are suitable for soil property data, are an extension version of Kandel (1998). It is a simple
model and a general symmetric bell-form FMF.

1
FMFx for 0 d x d P
[1  {( x  b) / d }2 ]

FMFx 1 for x ! P (5)

The parameter b defines the value of the attributes x at the central concept of the standard index
of the set. The form of the membership function and the position of the crossover points can be
easily changed by changing the value of the dispersion index, d. The parameter d gives the width
of the bell curve at the crossover point, which defines the transition zone around the central core
of the set in the same units as the central concept. Accordingly the shape of FMF strongly
depends on the parameters b and d and these parameters affect the fuzzified distribution of the
K-factor and soil erosion eventually.

4. Application of fuzzy boundary to Soil Loss

We made two simulation scenarios; one is for the conventional representation of sharp change and
the other is for the fuzzy representation of within-class variability described in equation (7).
Hence, the only source that would differentiate the amount of soil loss generation among the two
simulation scenarios is the different description of geographic boundaries. In typical fashion, a
value of soil erodibility factor, K is assigned for each grid cell (22m) according to the soil type-
soil erodibility conversion.
Each soil type has its own sand %, clay %, and silt % from the sampling test (KICT, 1992) and
then, K-factors were retrieved from the Erickson’s triangle diagram (Erickson, 1997).
Consequently the basin is divided into several patches that are homogeneous in terms of soil
properties and it is called the conventional representation of sharp change (Type I in figure 2).
Each soil patch is assigned one soil type (one K-factor) and is distinguished by another boundary
of soil type. However, the K factor has different values depending in part on how to specify soil
type in the RUSLE boundary cell and it can be explicitly controlled as mentioned earlier. To make

- 92 -
the soil boundary of soil map more realistic, the simple FMF is then used and the image of figure
3 shows a detailed fuzzy-knowledge based boundary description with 500m Euclidian distance. In
Korea, soil samples were taken at 1km spacing to calibrate the 2-D digital soil texture map. The
boundary area near which soil characteristics varies was arbitrarily bisected to discern different
soil texture for each category and these uncertainties have an influence on soil erosion estimation
in the RUSLE. It is, hence, assumed that the Euclidian distance of 500m is covered by the fuzzy
representation from the boundary on the basis of ground sampling spacing (1km) (Ahamed et al.,
2000) and the fuzzy distance is independent of the soil type, the tillage, and topography. In other
words, the FMF to 500m spacing on both directions of the soil boundary is considered to calculate
soil erodibility factor, K as shown in figure 3. The application of the fuzzy approach to the
RUSLE is as same as that of the conventional approach but the fuzzified K-factors (sharply
changed K-factors for the conventional approach as in type I of figure 2) only in the boundary
area.
Figure 4(a) shows the 2-D imagery map for the soil erodibility, K of the conventional sharp
change, while figure 4(b) shows the 2-D imagery map for the soil erodibility, K as a result of the
fuzzy representation. The fuzzy-knowledge based boundary in figure 4(b) (blurry area) is called
the fuzzy representation of geographic boundary (Type II in figure 2) and then it is regarded as
reproducing more real condition in relative terms. Table 1 presents basic statistics of the value of
K-factor for both methods. With the fuzzy representation, the mean value of the K-factors is
slightly high and the standard deviation is low, while the maximum/minimum values keep
constant.
Table 2 presents basic statistics of annual soil erosion calculated by the RUSLE for each method.
The mean (standard deviation) of soil loss for the fuzzy representation (figure 4(d)) is higher
(lower) by 3.1(4.4)% and the maximum loss for the fuzzy representation is higher by 1.8% than
that of the conventional boundaries (figure 4(C)).
The total soil erosion simulated by the RUSLE for the fuzzy boundary is 1474781 ton / yr , while
1429339 ton / yr for the conventional boundary. The sediment yield is given by SDR (3.9% by
equation 7) as 57516 and 55744 ton / yr for the fuzzy and conventional boundary, respectively.
Consequently the RUSLE sediment yield for the fuzzy representation of geographic boundary
results in lower error (2.9%) comparing to field measurement (unit sediment yield=103
ton / km 2 / yr (KICT, 1992)) as shown in table 3. It implies the fuzzy boundary shows better
performance, ignoring the selection of model, the quality of geospatial data, measurement
accuracy, and the basin characteristics.

5. Summary and conclusions

A fuzzy representation of geographic boundary is presumably better description of soil properties

in that it includes within-class variability concept, which can not be properly described by
membership in a single set of sharp change. The results of the RUSLE using the conventional
sharp edged boundaries and the fuzzy boundaries are tested against field measurement. The
primary conclusions of the study are as follows;

ƒ With the fuzzy representation, the mean value of the K-factors is slightly high and the
standard deviation is low, while the maximum/minimum values keep constant.
ƒ Accordingly, the mean (standard deviation) of soil loss for the fuzzy representation is
higher (lower) by 3.1(4.4)% and the maximum loss for the fuzzy representation is
higher by 1.8%.
ƒ The sediment yield is 57516 and 55744 ton for the fuzzy and conventional boundary,
respectively. The fuzzy representation of geographic boundary in the RUSLE results
in lower error (2.9%) and shows better performance.

- 93 -
To some degrees, error may be different for every new basin of interest, but the method used
herein should work anywhere. More realistic description of geographic boundary such as the
fuzzy representation is desirable in dealing with soil properties of the soil loss model.

References

1.Ahamed, T.R., Gopal Rao, K., Murthy, J.S.R. 2000. GIS-based fuzzy membership model for
crop-land suitability analysis. Agricultural Systems 63, pp. 75-95.
2.Ahamed, T.R., Gopal Rao, K., Murthy, J.S.R. 2000. Fuzzy class membership approach to
soil loss modeling. Agricultural Systems 63, pp. 97-110.
3. Burrough, P.A., 1986. Principles of geographical information systems for land resource
assessment. Clarenden Press, Oxford, UK
4.Burrough, P.A., 1992. Fuzzy classification methods for determining land suitability from
soil profile observations and topography. Journal of Soil Science 43, 193-210
5.Burrough, P.A., and Frank, A.U., 1996. Geographic Objects with Indeterminate Boundaries.
Taylor & Francis, pp.3-28.
6. Erickson, A.J., 1997. Aids for estimating soil erodibility – K value class and soil loss
tolerance. U.S. Department of Agriculture, Soil Conservation Services, Salt Lake City of
Utah.
7. Fernandez, C., Wu, J.Q., McCool, D.K., and Stockle, C.O., 2003. Estimating water
erosion and sediment yield with GIS, RUSLE, and SEDD. Journal of Soil Water
Conservation, 58, 128–136.
8.Hunter, G.J., and Williamson, I.P. 1990. The development of a historical digital cadastral
database. International Journal of Geographic Information Systems, 4,169-180.
9. Korea Institute of Construction Technology ( KICT), 1992. The development of selection
standard for calculation method of unit sediment yield in river. KICT 89-WR-113
Research Paper (In Korean)
10. Renard, K.G., and Foster, G..R., 1983. Soil Conservation-Principles of erosion by water,
In: Dregne, H.E., Willies, W.O. (Eds.), Dryland Agriculture, American Society of
Agronomy. Soil Science Society of America, Madison, WO, USA, pp 155-176.
11. Vincent, P. 1991. Modeling binary maps using ARC/INFO and GLIM. In Proceedings of
the European Conference on Geographic Information Systems (EGIS 90), (Utrecht: EGIS
Foundation), pp. 1108-1116.
12. Walling, D.E., 1983. The sediment delivery problem. Journal of Hydrology, 65, 209-237.
13. Walsh, SJ, 1989. User considerations in landscape characterization. In the accuracy of
spatial databases, Taylor and Francis, London, UK, pp. 35-44
14. Wang F and Hall GB, 1996. Fuzzy representation of geographical boundaries in GIS.
International Journal of Geographic Information System, 10(5), 573-590
15. Wolman, M.G., 1977. Changing needs and opportunities in the sediment field. Water
Resources Research, 13, 50-59.

- 94 -
 Table 1. RUSLE factors and soil erodibility factor, K between the conventional sharp change
(Type I) and fuzzy representation (Type II) used for the study.
Min Max Mean S.D.
R 413.420 562.571 459.286 24.099
K Conventional 0.100 0.390 0.251 0.055
Fuzzy 0.100 0.390 0.255 0.054
L 0.072 9.372 2.695 1.904
S 0.049 14.927 3.864 3.307
C 0.000 0.500 0.028 0.079
P 0.100 1.000 0.759 0.266

Table 2. Basis statistics of soil erosion for each method.

Min Max Mean SD
( ton / ha / yr ) ( ton / ha / yr ) ( ton / ha / yr ) ( ton / ha / yr )

Conventional 0.000 4931.244 24.849 59.374

Fuzzy 0.100 5021.634 25.639 56.742
Table 3. Comparison of annual soil erosion calculated by both methods. The measured unit
sediment yield is 103 ton / km 2 / yr (KICT, 1992).
Total erosion SDR Sed. Yield Measurement Error
( ton / yr ) (%) ( ton / yr ) ( ton / yr ) (%)
Conventional 1429339 3.9 55744 59246 5.91
Fuzzy 1474781 57516 2.92

- 95 -
Illustration captions

Figure 1: Spatial distribution of the RUSLE factors: (a) R-factor. (b) K-factor, (c) L-factor, (d) S-
factor, (e) C-factor, and (f) P-factor

- 96 -
Figure 2: Three different types of models used to effectively represent geographic boundaries in GIS; abrupt change
(Type I), large change (Type II), and gradual change (Type III)

Figure 3: Detailed fuzzy-knowledge based boundary description with 500m Euclidian distance

- 97 -
Figure 4: 2-D imagery map of soil erodibility factor (K) and soil erosion calculated by the RUSLE. The (a) and (c) is
for the conventional representation of sharp change, while the (b) and (d) is for the fuzzy representation of
geographic boundary.

- 98 -
Diagnoses for the Drought and Dzud Frequencies in Mongolia
T. Ulaanbaatar
Geophysical Department, National University of Mongolia, tarzad@yahoo.com

Abstract. The largest natural disasters are the drought and dzud in Mongolia. The studies
concerning drought and dzud in Mongolia indicate that the amplitudes and frequencies drought and
dzud increased during last 60 years due to the Global Warming. The paper shows some reasons and
consequences of these disasters such as desertification, and dust storm based on space and time
spectra of their frequencies. A possibility to mitigate the drought, desertification and dust storm
within the southwest region of Gobi Desert by regional cooperation is also presented.

Background

In Mongolia, the agricultural sector employed 48.5% of the labor force and contributed 26.0% to
GDP as of 2001. Approximately 90% of agricultural production came from the animal products sub-sector.
In 2001, animal products (mainly cashmere, wool, leather and fur) accounted for about 52.3% of all
exports. However, the calamity of ‘Dzud’ (disaster due to harsh winter or weather) occurred continuously
in the period 1999 and 2000, causing loss of about 10% of the livestock that died from hunger and frost.
The total number of livestock that died from this calamity reached 5.75 million heads, and more than
12000 herders lost their livestock caused by the dzud and as a consequence, agricultural output dropped by
about 30%. [The Study…, 2006]
At the end of 1999 the number of livestock is highest since 1918 or 33568.9 million heads, but
during the 1999-2000 years dzud is destroyed 3391.1 million (10.4%). At the winter of 2001 (after the
large summer drought of 2000) 4758.8 million (15.2%) heads of livestock died by dzud. [Batima P. and
others, 2005]
Moreover, dzud is becoming the main factor of increasing the poverty in the countryside
as well as intensifying the mass migration of rural people to the urban areas [Sarantuya, G., 2005]
In order to mitigate the damage from Dzud, the Government of Mongolia decided to
implement “The National Programme to Protect the Livestock from Drought and Dzud”.
Our study concerning with the Drought and Dzud, and their consequences consists of all
the territory of Mongolia and period 1943 to 2005.
In this paper we attempt to explore answers to next questions. What is the precursor to begin the
drought and dzud? What natural disasters are dominant? Are there any connections between the natural
disasters? Are there any peculiarities or oscillations to occur the drought and dzud in time and space
(within limits of Mongolian territory)? Are there any way to mitigate the drought and dzud?

Dzud

The term “Dzud” is unknown in world scientific community. Its description is also not accepted
commonly. Some have written that this disaster is due to harsh winter or weather (The Study…, 2006),
some said dzud have some variations, in example, there are “white dzud”, “black dzud”, “stormy dzud”,
“cold dzud”, “joint dzud” and “foot dzud” (Lesson of 1999-2000…, 2000) and others determine only the
natural disaster by death-rate of men. Sarantuya G., [2005] wrote that dzud is determined by losses of
livestock in Mongolia.
The longer anomalous climatic condition, which continues by season or extremely hot or cold
months and seasons play of immense negative role in the agricultural life of Mongolia. In the world the
understanding of harsh and severe winter and extremely cold snowy periods is familiar, however they do
not know this kind of calamity, which dominates in deserts and steppe with continental climate and
influences directly on the nomadic livestock farming as Mongolia.
We would like to underline that the “Dzud and drought are like a type of “mass extinction”
occurred during the Earth’s history.

- 99 -
 The frequencies of dzud are shown by dominant aimags (Figure 1 and Figure 2). We summarize
that as follow:
Ÿ During last 60 years Zavkhan was 29 years (48.3%) under the Dzud, except for 24 of 60
years with good climate conditions (40%),
Ÿ In 1944-1972, 13 times in Zavkhan (Northeast territory of Western Mongolia ), 12 times in
Tuv (Central Mongolia) and, Dornod and Sukhbaatar (Eastern Mongolia). So, the Dzud
focuses abovementioned territories, but Gobi Desert enters into this calamity,
Ÿ In 1972-1980, Zavkhan and Tuv aimag are still the hot points of Dzud, but Sukhbaatar
participates more intensively,
Ÿ Dzud increases intensively since 1992 by intervals 1992-1994, 1995-1998 and 1999-2002

35
Number of Dzud during last 60 years by Aimags
30
Number of Dzud

25

20

15

10
5 Aimags
0
ho i
ya obi
Se obi

Bu i i

K ii

H tar
Kh j
ha d
ob e

n
st i

D or

ul
j

v
nk ob

n
vs
kh vd
a
Ea lta

ga

t
lg

vu rno

Tu
G lng

Ar lga

ha
en
ng

sg
ng

U
ya l G
U

Su ho
Ba G
G

i-A

C han

ba

vk
uv
n-
le

o
th

Ba tra

Za
rk
k
u

en
So

Figure 1. The Dzud frequency by Aimags

The Figure 2 shows that the dzud increases in 1943-1945, which is so-called Large Monkey-Year
Dzud, 1951-1957, 1965-1968, 1974-1977, 1980-1987, 1998-2002 and decreases in 1948-1952, 1967-1975,
1976-1986, 1987-1992 and 1999-2001.

20

18

16

14

12

10

8
Ⱥɣɦɝɢɣɧ ɬɨɨ

0
1943- 1947- 1950- 1952- 1954- 1956- 1958- 1960- 1962- 1964- 1966- 1968- 1970- 1972- 1974- 1976- 1978- 1980- 1982- 1984- 1986- 1988- 1990- 1992- 1994- 1996- 1998- 2000-
1944 1948 1951 1953 1955 1957 1959 1961 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001

Figure 2. The frequencies of Dzud during last 60 years

The precursor and existence of dzud does not only depend upon the winter meteorological
conditions, but mainly on the behavior of the previous summer. (Sarantuya G., 2005)

- 100 -
 Drought

Drought causes losses of livestock, and also very seriously affects environment: land degradation,
desertification, fire and water deterioration. 68% of Mongolia is drought prone, most affected by drought is
central and southern regions, in which every year drought over 40-50 days occur once 2-3 years.
(Bayasgalan, M., 2005)
The PDSI (Palmer Drought Severity Index) trend for the last 62 years from 1940 to 2002 is shown
in Figure 2. PDSI is averaged value over the whole territory of Mongolia for growing season, from June to
August. Periods of PDSI increasing approximately are 1944-1959, 1980-1994, of decreasing 1959-1980,
1994-2002. According to the Palmer’s drought severity classification severe drought occurred 7 times
in1944, 1978, 1980 and consecutively 1999-2002, moderate 10 times in 1942,1946, 1947, 1951, 1972,
1981, 1982, 1989, 1998 and slight 8 times in 1941, 1948, 1949, 1952, 1957, 1968, 1979 and 1997. Since
1994 drought intensity has been increased significantly. (Bayasgalan, M., 2005) (Figure 3)

Figure 3. PDSI changes for 1940-2003

Figure 4. Map of drought frequency in Mongolia (Erdenetuya, M., ICC, Ministry of

Natural Environment)

- 101 -
 Figure 5. Desertified area in Mongolia by average NDVI

Connection of Drought and Dzud

Comparing Figure 2, Figure 3, Figure 5 and Figure 6, we can see easily that high possibility of
occurrence of very dangerous dzud, when very dry summer follows harsh winter.

Table 1. Frequency of Drought and Dzud

rough 0 8 5 3 8 3 0 6 7 4 4 7
t

zud 9 0 0 0 4 0 4 57 92 14
http://www.fao.org/ag/agl/swlwpnr/reports/y_ea/z_mn/mntb262.htm

221
201
181
161
141
Frequency

121 Dzud
101 Drought
81
61
41
21
1
19 8 8 19 8 9 19 9 0 19 9 1 19 9 2 19 9 3 19 9 4 19 9 5 19 9 6 19 9 7 19 9 8 19 9 9 2000 2001 2002 2003

Figure 6. The drought and dzud frequencies in Mongolia during last 16 years

- 102 -
 Based on Table 1 and Figure 6, we can conclude that drought follows dzud by 6 months to one
year later and there is whatever oscillation between the drought and dzud. The drought does not occur after
dzud without 1999-2001 years.

Consequences of Drought and Dzud

Contemporary Earth’s climate behavior is multiplicity, which includes all the characters of
Warming and Glaciation: floods, land degradation, forest and steppe fire, desertification, deforestation,
drought, rainfall, sandstorm, snowstorm, duststorm, eliminations in water resource, heavy cold and hot
condition, environment pollutions and so on.
The impacts of these calamities differs by space and time, the score of damage in each country and
nations.
We must select and group the natural disasters by the feedback mechanisms and chains. For
example, the Central Asian Arid Climate Zone as a group consisting of main types of natural calamity:
drought, Dzud. This group has next types of calamity:
1. Decreasing the water resources
2. “Mass Extinction” in biodiversity and losses of livestock
3. Wild Fire
4. Land degradation
5. Desertification
6. Deforestation
7. Poverty
8. Mass migration to urban area
9. Air and water pollutions
10. Dust Storm and so on.
From this list we conclude that at first, consequences of Global Warming are the decreasing of
water resources, drought, wild fire, land degradation, deforestation and dzud.
The consequences of drought and dzud are “mass extinction”, poverty, mass migration to urban
area and so on.
In this paper we could limit our study by some important types of this group and explore their
reasons.

Exploration of Reasons

Abovementioned group calamity must have whatever order and connection for each other
and reason for their existences. To mitigate negative impact of this group we need to determine the first
reason, second reason, third and so on. There is two large reasons: The Global Warming and deficit of
water resource.

The Global Warming

The largest reason is, of course, the man-made Global Warming or Global Change. It goes by
natural and anthropogenic way itself.
According to our study [Ulaanbaatar T., 1994; 1998a; 1998b; 1999] the natural global warming
has some gradual increases which may tell that they are indivisible steps each other and follow in next
order:
1. Stabilizing effect of water vapor on thermal storage near the earth surface (Insolation-Water
vapor positive feedback),
2. Decreasing the albedo of snow cover and ice sheets off, (Insolation-Albedo feedback
and Ice/snow- Albedo negative feedback)
3. Decrease of daily and yearly temperature range and, temperature contrast between the equator

- 103 -
and poles, (Feedback on Isotherms of Heliogeothermosphere -Global average temperature of planet),
4. Wave length of terrestrial radiation becomes shorter than previous climate epoch by blocking
influences of water vapor and clouds, finally, wet and warm climate covers throughout the earth surface,
(Negative Feedback on Increasing of air temperature-Shorting of wave length of terrestrial radiation)
5. Continuous cloudiness exists in equatorial regions, (Positive feedback on Insolation-
Cloudiness)
6. Disturbance in underdeveloped atmospheric global circulations due to the continuous cloudiness,
(Negative feedback on Insolation-Cloudiness)
7. Decreasing the permafrost off, (Negative Feedback on flow of substance in Lithosphere-
Permafrost)
8. Increasing the atmospheric CO2 content in consequence of volcanic and tectonic activities,
(Positive Feedback of Volcanic activity-Warming)
9. Rising the isotherms of heliogeothermosphere due to rising of sea level i.e.
transgression, (Positive Feedback on Isotherms of heliogeothermosphere -Global average temperature of
planet).
If the Earth’s climate system naturally remains without the human-influences as like above, this
World Climate SuperCirculation (so-called by Ulaanbaatar T.) would continue some hundred millions of
years. Unfortunately, new man-made feedbacks in all of these steps can effect powerfully and most quickly
than natural variability into the climate system. Today, the Earth’s climate goes already at first 3-4 steps.
On a word, human hazardous activities can influence some thousand-fold beyond the volcanic activities
producting carbondioxide. If we cannot immediately impact in mitigation of our dangerous activities and
climate change, present climate system becomes incontrollable critically.
As results, in latest few years a multitude worst climate event are occurring throughout the Earth.
Situated deep in the interior of Asia and unpenetrated by the air currents from the oceans,
Mongolia has conspicuous continental climate, with highly changeable temperature, sharp difference in
temperature between day and night, abundant sunshine, intense evaporation and little precipitation.
The Gobi Desert of Mongolia is very sensitive to Global Warming.

The Water Resources

x Surface Water

More than half the country is covered by permafrost, which makes construction, road building, and
mining difficult. All rivers and freshwater lakes freeze over in the winter, and smaller streams commonly
freeze to the bottom.
In consequence of global warming in whole territory of Mongolia more than 400 hundreds of
rivers and lakes are depleted during the last 30 years. 37% of the territory of Gobi and Great Gobi
Protected Areas have no any sources of water supply, which have caused the elimination of hundreds of
oases, springs, rivers and wells.

x The Ground Water

Demands of water in Mongolia are mainly met from the ground water sources. But in course of
last 16 years the ground water table lowers down due to the drying of surface water and decreasing of
precipitation.
The level of lowering down of ground water varies by 2-3 meters in the Shallow Well and Shaft
Well, and 1-2 m in Production Well, but there dominates no water in Traditional Well in the Mountain
Steppe. The level of lowering down in Mountain regions is more than in the Deserts and Steppes.
The lowering increases also in the Gobi Desert. We have experienced the lowering down of Gobi
ground water table, when T.Ulaanbaatar participated in JICA water exploration expedition in Dundgobi,
Umnugobi and Dornogobi, and when met herdsmen.

- 104 -
 x The Precipitation

Main sources of precipitation in Mongolia dominates as follow:

o The Arctic Oscillation
The arctic cycle depends mainly on the precipitations of whole territory, specially, Northwest of
Western Mongolia, Eastern and Central Mongolia.
o The west trade-wind. All year, the west and northwest wind stream dominates in Mongolia. The
northwest trade-wind generates from Arctic Circulation, and slight west stream into the Gobi
Desert is from the Caspian Sea and the Mediterranean Sea. (Figure 7)
o The Pacific Circulation, influence of which is slightly on a very little zone of Khalhin Gol River in
Eastern Mongolia. (Figure 8)
There is a short rainy season in July and August during which most of the yearly rain falls. Around
67-78 percent of all precipitation falls during these three summer months.

http://www.wunderground.com/global/Region/AS/WindSpeed.html

Figure 7. Trade-wind map of Asia

- 105 -
 Figure 8. 1961-1990 Mean annual precipitation of Mongolia

From Figure 4, Figure 5 and Figure 8 we can see the west region of Gobi Desert has a very little
annual precipitation and very large frequency of drought.
Now we would like to show a possibility to improve the ecosystem of the southwest Gobi Desert.
The regional intensive cooperation can mitigate only some local problems concerning
desertification, drought and dust storm of Gobi Desert.

Can We Mitigate Dust Storm Generating from Gobi Desert?

According to our study the main reason of the drought in Mongolia is the lowering down of
ground water table, and next reason is the decreasing the humidity (precipitation) on and near the earth
surface. Very sensitive vegetation system of Gobi Desert irrigates on the one hand, by ground water, on the
other hand, by the humidity and rare precipitation.
The vegetation system of southwest Gobi Desert irrigates only on the one hand by ground water,
and on the other hand, seasonal humidity (precipitation) delivering from territory of XinJing Uigur
Autonomous Region, China by west trade-wind (Figure 7). This type of humidity and precipitation is very
little, however, very useful for the life of sensitive and thirsty vegetation of southwest Gobi Desert.
The level of ground water table in this region also lowers down.
Gong S.L. and others [2004] wrote that “depending on the degree of desertification, newly
formed deserts covered 15% to 19% of the original desert areas and would generate more dust storm,
ranging from 10%-40%, under the same meteorological conditions for spring 2001. …There are two major
non-Chinese dust sources: Mongolian Gobi over north of China and Kazakhstan/Uzbekstan deserts in the
northwest. The non-Chinese deserts (mainly from Mongolia) contribute about 30-60% surface dust
concentrations in Northeast China and 40-50% in Korea and Japan”.
http://gsa.confex.com/gsa/inqu/finalprogram/abstract_54086.htm
The desertification is not only Chinese problem, it is also problem of Mongolia, furthermore Korea
and Japan. For this reason, if we contribute in the framework of the artificial precipitation project of NW
China, we can mitigate the desertification and dust storm generating from Mongolia
Qiang, Mingrui, and Chen, Fahu (2003) reconstructed the dust history in Northwestern China for
the last 2000 years at a 5-10 year resolution and showed quite clearly an increased trend in dust storm
events over the last 20 years.
http://gsa.confex.com/gsa/inqu/finalprogram/abstract_54086.htm

“Northwest China's Xinjiang Uygur Autonomous Region plans to explore air water resources
through artificial efforts to ease the shortage of surface water and improve the local environment.
Meteorologist Zhang Jianxin said that the region has been technically prepared to launch the
artificial precipitation project. Currently, two-thirds of Xinjiang's land area and more than 12 million of its
population are threatened by desertification, which spreads by 350 square kilometers annually.
The air in Xinjiang is rich in water vapor, which totals 1.38 trillion cubic meters annually, Zhang
said, adding that only 17.6 percent of the total amount transforms into rain and snow on yearly basis.
Zhang said, if one percent more of the vapor is transformed into rain and snow through artificial
technologies, an additional 1.4 billion cubic meters of water could be utilized by the region annually. That
amount is equal to one-sixth of the region's current runoff volume.
Xinjiang first launched artificial precipitation in the 1960s and has achieved encouraging results.”
www.hartford-hwp.com/archives/55/469.html
“Zhang and his Lanzhou Arid Climate Research Institute are now engaging themselves into an
ambitious "heaven water seeking program," a part of the Ministry of Science and Technologies' West
Development Scientific and Technological Program Package.
Started at the end of 2004, the program is to research the cloud resources in the air above the
Qilianshan Mountain in the province and provide solutions on exploiting the air water resources.

- 106 -
 "If the program goes smoothly, it will bring 370 million cubic meters more water to the Qilianshan
Mountain and the continental rivers of the nearby Hexi Corridor in Gansu," Zhang said. "This will greatly
improve the environment of the areas and create a profit of 600 million yuan (about 72.6 million US
dollars)."
"China defines the water conservancy holding more than 100 million cubic meters of water to be
large-scale reservoirs and the program will bring nearly four large-scale reservoirs to the province," said
Senior Engineer Qi Xinhui from the provincial flood control and drought relief headquarters, which is
believed to be one of the biggest beneficiaries of the program.
As a matter of fact, the "heaven water seeking program" is not only carried out in Gansu, but also
in the neighboring Shanxi Province and Ningxia Hui Autonomous Region.
In these regions, the per capita water resources is below 1,000tons per person, less than half of the
water-needy China's average.China's per captia water resource possession is 2,200 tons, only one quarter of
the world's average.
Liu Chunzhen from the Ministry of Water Resources' informationcenter said the continuing
drought and unplanned water use worsensthe water shortage in China's northwest.
To ease the situation, local governments have already started to apply the artificial precipitation.
The statistics from Gansu artificial precipitation office said the province has established more than
300 operation sites for artificial rainfall and snowfall and will perform more than 1000 operations every
year with the help of rockets, cannons and airplanes.
Gansu has benefited about one billion cubic meters of precipitation from 2004 to now thanks to the
artificial operations.
Zhang said the development of the cloud water resources will bring the provinces "some invisible
and long-term benefits."
http://news.xinhuanet.com/english/2005-05/25/content_3002822.htm

Gong S.L. and others [2004] wrote that some regions that are mainly in northern China close to the
extensive desertification areas could reach reduction by 40% to 50%.
There are many advantages in NW China Region to Launch Artificial Precipitation Project, but
some disadvantages too. We think that more than 300 operation sites for artificial rainfall and snowfall and
more than 1000 operations every year with the help of rockets, cannons and airplanes can influence
powerfully on expansions of desertification and increasing of the drought frequency in Gobi Desert.

Conclusion

1. Frequencies of dzud and drought increase due to the Global Warming.

Approximately every 60-year an extremely harsh dzud may be occur
throughout the country. 30-years oscillation is indicated.
2. Drought follows dzud by 0.4-0.7 years late of phase.
3. Drought is reason of the hazard of Dzud, difficulties of livestock farming system, poverty and
other negative phenomena.
4. The drought does not occur generally after dzud without 1999-2001 years.
5. There is a right way to mitigate the frequencies and amplitudes of drought, desertification and
dust storm in Gobi Desert: To expand the NW China Region to Launch Artificial Precipitation
Project to southwest Gobi Desert and to intensity the regional cooperation.

Reference

1. Batima P., Oyun R., Erdenetuya M., Erdenetsetseg B. and T. Ganbaatar, Affects of Livestock
Farming
2. System by Climate Change, UNEP, START and AIACC, Ulaanbaatar, 2005
3. Bayasgalan, M., Drought monitoring, Proceedings of First International Conference on Studies on
4. Mongolian Environmental Issues Using Remote Sensing and Geographical Information System,
5. 99-103 pp, Chiba University, Japan, 2005
6. Bayasgalan M. “Drought monitoring in Mongolia”, Dissertation, Ulaanbaatar, 2006.

- 107 -
7. Erdenetuya M., Remote sensing methodology and technology for pasture vegetation assessment,
8. Dissertation, Ulaanbaatar, 2004.
9. Gong, S. L., Zhang, X. Y., Zhao, T.L., and L. A. Barrie, Sensitivity of Asian dust storm to natural
and
10. anthropogenic factors, Geophysical Research Letters, Vol., 31, L07210,
doi:10.1029/2004GL019502, 2004
11. Sarantuya, G., Studies of dzud occurring in Mongolia and possibilities of assessment, National
University
12. of Mongolia, Ulaanbaatar, 2005
13. The Study for Improvement Plan of Livestock Farming System in Rural Area, Draft Final Report,
14. (Mongolian counterpart person is Ulaanbaatar T.), Japan International Cooperation Agency
15. (JICA), Ministry of Food and Agriculture Mongolia (MFA), Pacific Consultants International and
16. Mitsui Mineral Development Engineering Co. Ltd, January, 2006.
17. Ulaanbaatar T., Radiative properties of water vapor, Symposium: Radiative effects of water
vapour on
18. climate, IAMAS, XXII General Assambly, IUGG, Birmingham, UK, Vol. A, 260, 18 July-2
19. August, 1999,
20. Ulaanbaatar T., Mathematical modeling for the thermal regime of the Earth’s surface and
cryosphere, Ph.D
21. Thesis, Mong. Univ.of Technology, Ministry of Science and Education, Ulaanbaatar, 1994.
22. Ulaanbaatar, T., Climate Supercirculation of the Earth, Scientific Translation, 3(132), 246-264,
Nat. Univ.
23. Mong., Ulaanbaatar, 1998a,
24. Ulaanbaatar, T., Feedbacks in the World Climate Supercirculation and their interacts, Scientific
25. Translation, N6(147), 124-139, Nat. Univ. Mong, Ulaanbaatar, 1998b.

- 108 -
 EVALUATION OF ATMOSPHERIC OPTICAL CHARACTERISTICS
IN THE MONGOLIAN TERRITORY

G. Batsukh1, B. Daariimaa2, T. Narangarav3

Department of Geophysics, NUM
batsukh@num.edu.mn

Abstract: In this paper, we have determined the yearly and long-term changes of atmospheric
aerosol optical thickness in photosynthetically active spectral region (380-710nm) and atmospheric
optical thickness in photosynthetically active (380-710nm), infra red (>710nm) and biologically
active (<510nm) spectral regions in Ulaanbaatar. We have used hourly measurements of direct solar
irradiance, obtained by AT-50 actinometer with glass filters in Ulaanbaatar (1980-2004), Ugtaal
(1987-1996), Sainshand (1993-1997) and Darkhan (1992-1998), where the population and climate
are different.
The aerosol optical thickness and atmospheric optical thickness in region of photosynthetically
active radiation were increased sharply in long-term change, especially, during the 1982-1984 and
1991-1993. It is related to the change of aerosol optical thickness for upper atmosphere. And, we
have shown how the integral coefficient of transparency expresses the change of aerosol optical
thickness in long-term change.
x When the Sun’s altitude increases, the integral coefficient of atmospheric transparency decreases,
and the aerosol optical thickness and atmospheric optical thickness increase. This relationship is
same for all seasons.
x We have shown that long-term change of atmospheric optical characteristics change the same at
any altitude of Sun and in any season.
x The atmospheric optical characteristics are greater in summer than in winter, because the Sun’s
altitude is greater in summer than in winter.

- 109 -
 LANDSCAPE UNIT MAP OF UVS NUUR AND ADJACENT AREAS

Michael Walther
National University of Mongolia
MOLARE Research Centre, Director
mwaltherub@web.de

Uvs Nuur Basin is located around 1.200 km west of Ulaanbaatar (Mongolia) in the western province of
Uvs Aimak (Capital: Ulangom). The lake covers a surface of 3.350 sqkm, has a diameter of around 75 km
and a maximum depth of 22 m. The endorehic lake has an inlet in the eastern part of Tes Gol.

Based of geomorphologic field studies in the late nineties different landscape units are subdivided into:
1. recent lake surface
2. paleo lake surface
3. paleo shorelines
4. pediments
5. alluvial fans (different old stages)
6. fluvial meadow bottoms
7. fluvial high flood beds
8. rivers
9. tectonic lines
10. dunes (different old stages)
11. basement (mountain ranges)

More information is published in:

1. Grunert, J., F. Lehmkuhl & M. Walther 2000: Paleoclimatic evolution of the Uvs Nuur basin and
adjacent areas (Western Mongolia). - Quaternary International, 65/66: 171 - 192.
2. Grunert, J, & D. Dasch 2000: Binnendɶnen im nɰrdlichen Zentralasien. - in Walther, M. et al.
(Eds) 2000: Proceedings of the Congress Mongolia 2000, Freie University of Berlin, Berliner
Geogr. Abh., 205: 2 - 8; Berlin.
3. Klein, M. 2001: Binnendɶnen im nɰrdlichen Zentralasien. - Mainer Geogr. Studien, 47: 182 S.;
Mainz
4. Klinge, M. 2001: Glazialgeomorphologische Untersuchungen im Mongolischen Altai als Beitrag
zur jungquartɞren Landschafts- und Klimageschichte der Westmongolei. - Aachener Geogr.
Arbeiten, 35:125 S.; Aachen.
5. Naumann, S. & M. Walther 2000: Mid-Holocene lake level fluctuations of Bayan Nuur (North-
Western Mongolia). - in: Miehe, G. & Y. Zhang (Eds): Environmental Changes in High Asia;
Marburger Geogr. Schriften, 135: 15 - 27; Marburg.
6. Walther, M. 1999: Befunde zur jungquartɞren Klimaentwicklung rekonstruiert am Beispiel der
Seespiegelstɞnde des Uvs Nuur Beckens (NW-Mongolei). - Die Erde, 130: 131 - 150; Berlin.
7. Walther, M. et al. (Eds) 2000: Proceedings of the Congress Mongolia 2000, Freie University of
Berlin, Berliner Geogr. Abh., 205: 2 - 8; Berlin.

- 110 -
 SOME INITIAL RESULTS OF USING
THE 6P FOREST INVENTORY METHOD IN MONGOLIA

Sh. Tsogtbayar
tsogtbayar@biology.num.edu.mn
Department of Forestry, Faculty of Biology, NUM

The 6P forest inventory method was used in Mongolia to determine stand characteristics for a 900
ha site in Mongonmorit soum, Tov province of Mongolia. The data was compared to data
collected by traditional inventory methods used in Mongolia, to make an evaluation of the 6P
method. The research was conducted in cooperation with German professionals working under
the natural resource sustainable development project framework of German Technical
Cooperation (GTZ).

One hundred and fifty (150) plots were laid out on a satellite image of 1:50000 scale. The ground
distance between plots was 100 m. In the field, each plot was located, and the latitude, longitude
and altitude for each plot were measured at its center. From the center, a 12.62 m radius circle
was constructed and the six nearest trees to the center were selected. For each selected tree the
following characteristics were recorded: species type, DBH (diameter at breast height), and
distance from center. The height of the tallest of the six trees was also measured. In addition to the
selected tree measurements, natural regeneration was described, presence or absence of previous
forest fire recorded, and site quality was measured. After measurements and records for a given
plot was completed, the next plot center 100 m away was identified by GPS and the procedure
was repeated. In total 150 plots in the 900 ha site were selected and described in this manner.
The data from the 6P method were compared to data acquired through traditional inventory
methods used in Mongolia. The results were summarized on a per hectare basis. From an analysis
of the data, the following results can be concluded. First, Basal area increment and resource per
hectare will increase as distance from the sixth selected tree to the center decreases. Second,
measurement of the tallest trees in the sample plot is automatically increasing resource per hectare.
Due to this study, we focused to produce most comfortable methods for Mongolian forest
inventory study.

- 111 -
 PASTURE LAND CLASSIFICATION USING REMOTE SENSING DATA

D.Narangerel
Informatics & RS Institute, Mongolian Academy of Sciences,
narangerel@arvis.ac.mn
N.Monkhoo
Agency for Land Administration Geodesy and Cartography, Ministry of Construction & Urban
development,
B.Suvdantsetseg
“NUM-ITC-UNESCO” laboratory for Remote Sensing/GIS , The National University of Mongolia,
E-mail: suvd16@num.edu.mn
A.Saruulzaya
Institute of Geography Mongolian Academy of Science
B.Batzorig
Ministry of Food and Agriculture of Mongolia

Abstract:
Pasture land classification is one of the emerging issues in Mongolia.
Since 2002 when land privatization and privatization of pasture land and surroundings of
Winter, summer & autumn camps become reality; the Mongolian Government needs to have and
to implement advanced techniques for assessment of natural and land surface resources & their
potential.

Current paper describes methods of pasture lands classification system used at present and
application of new advanced technologies such as high precision Remote sensing ASTER data
and global land cover classification system, which can be applicable for Mongolia.

- 112 -
 Use of Caesium-137 for the agriculture soil degradation in
Central part of Mongolia
O.Batkhishig1 , N.Enkhbat2 and B.Burmaa3
1
Institute of Geography, MAS
2
Nuclear Research Centre, NUM
3
Ministry of Food and Agriculture

Land degradation and soil erosion becoming very serious problem in Mongolia. According
to statistics over 90% of total cropping land is eroded and the yield loss on eroded area comprises of
24-50%. Due to heavy migration of rural habitants to the central market area, over 70% of total
pastureland is overgrazed. According to the National Statistics the number of lifestock in 2005
reached upto 30 million head. The another factor affected to the soil erosion and pasture
degradation is continuous drought and poor farming practices. Soil degradation appears thus as a
major limitation for increased food production. It is then essential to document the magnitude of the
problem and develop conservation approaches adapted to the Mongolian agri-environmental
conditions.
Use of Caesium-137 radionuclide of for solving soil erosion, land degradation problems
very important for assessment, monitoring and land degradation process and designing effective
soil conservation measures. Our study region situated in Erdenesant somon Tuv aimag. Soil
erosion is a prime cause of loss of productivity of land. Compared 30 years cultivated field and non
cultivated field down to 0-5, 5-10, 10-15, 15-20, 20-25 cm depths. This soil is by Mongolian soil
classification Dark Kastanozem soil. Thickness of humus horizon 30 cm, soil reaction neutral pH
ranges 6.65-7.13, humus content varied 7.206 – 2.217 %. Result of long cultivation soil becoming
sandy, soil fertility declining. Soil humus content in the upper 5 cm of non-cultivated soil is
7.206 %, but in agriculture fields soil humus content falling down to the 3.186 % or more than 2
times.
Caesium-137 radionuclide measurement of soil erosion fields show following results. In
the non-cultivated fields Caesium-137 has a 3.1, 2.1, 1.4 kBq/m-2, but in the 30 years cultivated
fields this value ranges 1.5, 1.5, 1.6, 0.7 kBq/m-2. Result of cultivation soil erosion rate nearly 2
times increased. Mostly by wind deflation. Upper 10 cm most changed by erosion.
The results of this study showed that soil erosion in the Erdenesant Soum of Central aimag
of Mongolia is accelerating, should be an immediate concern to conservationists and development
planners at all levels.
Further research focused involve to satellite images for the soil erosion and degradation
combining radionuclide techniques.

Key words: Caesium-137, Soil erosion, degradation, Mongolia, use of radionuclide.

E-mail address: batkhishig@gmail.com; Enkhbat_N@yahoo.com

- 113 -
 The Mongolian forest characteristic and ecological changes.
G.Tsedendash1
1
Institute of Botany, Mongolian Academy of Sciences, Ulaanbaatar, Mongolia

Mongolian forest located in taiga of Eastern Sibiria-Southern Baikal, between the Central-Asian
steppe and desert as mountain forest and the most southern part of southern spruce ( Larix) forest.
However, totally square of Mongolian forest almost 10 percent consists of Mongolian total
territory, its ecological efficient is the most important for ecosystems.
Totally square forest of Mongolia is 17515.2 ha and the total resource is 1334.6 million cub.
Selenge river’s water collecting square has 73% of forest land and has differences by its vegetation,
regeneration and dynamics, which separated it into four zones: tundra, taiga, pseudo-taiga, sub-taiga. The
lower belt of forest situated at 650m ( Tujiin nars), but higher belt situated at to 2600m elevation
( Khangain nuruu).
On the basis of above-mentioned differences of forest and vegetation, total forest territory
separated into the three provinces, ten regions ( Karta lesov MNR, 1983, Tsedendash,1996).
Over 2500 species of vascular plants grow in Mongolia, ¼ of them grow in forest zone and about
1/3 of medicinal plants existed in Mongolia.
Anthropogenic impact to forest ecosystem during the last 100 years shows that 40% affected by
human activities, and regenerated 12% by birch and aspen forest, 1.6% changed by ecosystem, other 9.2%
is lost its forest characteristics (Methodi otsenki, 1990).
Comparison of 1974 years forest covered territory with 2004 years, shows decrease number by
1187.6 thousand ha. In spite of this open woodland and felled area has increased by 10-15 times during this
period.
Forest water is high efficiency. Khentii Nuruu’s larch forest crown keeps rain water 19.2% in itself
and 80.8% to soil. Larch forest with forbs litter is 2-3cm2 , resource is 9.5-16.8 thousand ha (Lesa
MNR,1983).
The melted snow stream of felled area 90-200 times more than virgin forest and 20-50 times of
summer hibernal stream. The maximum temperature of 2m height forest is 29.40C, logged forest 29.70C,
minimum temperature 3.40C, -5.90C (Lesa MNR).
The soil surface temperature of felled area reached in Khangai 60.00C, in Khuvsgul 53.50C.
The air humidity in July ( 13 PM of daytime) in forest 76%, in felled area 69%, which indicated
mild impact to the climate.
Soil litter of taiga forest ( 3-5cm), resource9-23 thousand ha, humidity capacity 6-11mm. Water
permeability of soil is 2.0-9.9mm/min, soil surface stream of all precipitation is not more than 3% and
erosion coefficient 1*10-5-12*10-5.
Annually 600-1000t/km2 soil has erosion in felled area. The soil surface stream of burnt larch
forest affected 12 years before has 2.2-3.4mm, erosion coefficient 74*10-4-100*10-4, annually soil erosion
1.52-9.62 t/km2.
By vegetation coefficient related to natural forest-felled area after a year 24.1%, after 12 years -
22.9%, after 30 years-18.9%.
If the grass layer of forest resource is 2.1cent/ha, but 30 after logging is 11.9cent/ha, moss from
23.3cent/ha into 3.5cent/ha. The temperature regime of felled area forest has changed and affected to the
eternal frost, which follows mire and growing of willow stand, birch, elm forest moss.
The mire will be changed into the bunchgrass steppe or condition for growing birch forest. The
process from coniferous forest to larch forest change is common.
The resource of forestry is already used and started to cutting in green area, stricter lines and even
affected the trees not-reached Y class of growing. In 1996-1997, lost forest by fire are over 5 million ha
forest, approximately annually 500 thousand ha in vary intensive forest fire. Forest fire separated into the
three intensifications and has mixed specifics like crown fire, surface fire, which follows different impact.
The grass layer, seedlings, young growth are burnt in intensive fire of surface. Some scientists
have noticed that surface fire has positive impact to the regeneration of forest with rich vegetation.

- 114 -
 Comparison of years between 1988 in 1981, the shows forest land reducing by 11.2 thousand ha or
25%, pine forest land by 10.4 thousand ha ( 14.3%), aspen forest 0.3 thousand ha ( 30%) increasing in
Shariin gol region.
After 13 years, felled area of burnt forest have no any signs of regeneration. From Tujiin nars
research, the forest has Cerambycidae, Scolytidae in the second summer of burnt region. These larva’s are
wintering in trees, in autumn of the second year.
Therefore, impossible to have standard cuttings after the second year of fire affected forest. The
burnt tree even its not alive still efficient in decreasing of wind speed, shading someway the soil surface
and influencing in losing the humidity.
The forest main issue is to study the forest regeneration conformity, the reforestation suited in its
specifics. It is considered that more over 2-3 thousand of 6-10 years young growth is satisfied, not satisfied
if lower than a thousand.
From research in Khangai nuruu, the first year we have calculated 451 thousand seedlings in 1 ha,
135 thousand in the second year, decreased into 38 thousand in the 3-rd year and recovering work is started
from 1972 in Mongolia.
During this period the reforestation is done in 105 thousand ha land. The reforestation is provided
only in 30% or over 30 thousand ha in felled area.
A long-term, cyclical drying of climate is causing a slow northerly retreat of its forest. For instance,
forest distribution certified by trees of Khentii , Khangain nuruu, Gobi Altai nuruu’s different forests.
Conclusion.
x Forest ecosystem changes of Mongolia, deforestation of eternal frost taiga
x Reforestation of felled areas, non-decrease of vitality coniferous forest
x Needs of appropriate method, research suited in climate and landscape of Mongolia
different from European technology
x Selection and putting into operation of data exchange system of training-monitoring-
union-emergency agency for disaster preparedness.

- 115 -
 A METHOD TO ESTIMATE SOIL MOISTURE USING L-BAND
SYNTHETIC APERTURE RADAR DATA
JAVZANDULAM TSEND-AYUSH*, JOSAPHAT TETUKO S.S**, RYUTARO TATEISHI**, TSOLMON
RENCHIN***
*Department of Geoinformatic Engineering, Inha University, 253, Yonghyun-Dong, Nam-Gu, Incheon, 402-751,
ROK
**Center for Environmental Remote Sensing (CEReS), Chiba University, 1-33 Yayoi-cho, Inage-ku, Chiba, 263-8522,
Japan
*** Geophysical department, Mongolian National University, Ulaanbaatar, Mongolia

Tel: 82-32-874-7660 Fax: 81-43-290-3857

E-mail: javzaa@inha.ac.kr

Abstract.
This study focused on development of a method to estimate the soil moisture from l-band synthetic
aperture radar (sar) data in arid area. To develop a method relationship between backscattering
coefficient of remotely sensed data and soil moisture, and relationship between dielectric constant
properties and soil moisture were investigated. Based on the dielectric constant properties of soil
samples, the relationship between the backscattering coefficient and soil moisture was obtained by
calculation using the developed a multi-layer modeling analysis. Then by using this model, the soil
moisture had been estimated by means of derived backscattering coefficient from japanese earth
resources satellite-1 (jers-1) sar data.

Key words: Soil moisture, Dielectric constant, and JERS-1 SAR

Introduction
Soil moisture is an important component of the hydrological cycle. The use of remote sensing to measure
soil moisture has been researched on for the last 20 years, with the use of both passive and active
microwave instruments (Ulaby et al., 1986). Quantitative measurements of soil moisture in the surface
layer of soil have been most successful using passive remote sensing in the microwave region. SAR data
have great potential for terrestrial observations, as has been demonstrated by L-band SAR on Seasat
(Ulaby et al., 1983).
The objective of this research was to develop a method to retrieve soil moisture in arid and semi arid
area of Mongolia using SAR data. For this purpose, the relationship between radar response and dielectric
constant, and the relationship between dielectric constant and soil moisture are investigated. Based on the
dielectric properties of the collected samples, the relationship between backscattering coefficient and soil
moisture was estimated by developing a multi-layer modeling analysis.

1. Study area and ground measurement data

Our experiment was conducted on arid area of Mongolia, which is located in the southern part (103o00’E
to 104o00’E, 43o20’N to 44o20N) of country .
A generally arid climate prevails in the study area, dominated by low precipitation, with average rainfall
less 50mm per year.

Field survey was conducted in July 2003 to collect soil samples to measure soil moisture. Soil samples
were collected from 12 points in the large homogeneous sparse vegetated arid area for dielectric constant
measurements.
The soil moisture content may be expressed by weight as the ratio of the weight of water present to the dry
weight of the soil sample. To determine the ratio for a particular soil sample, the sample is weighed before
and after drying it at temperatures of 100o - 110oC. Water content of the sample is calculated as:

- 116 -
 mwi  mdi
Ki (1)
mdi  mb
where Ki is soil water content of ith sample , mwi , mdi and mb are weights of ith wet soil sample and
box, weight of ith dry soil sample and box, and empty box, respectively.

2. Analysis method
Based on enhanced analysis method, developed by Tetuko et al. (2003b), a model of scattered waves
Z total

air

Eo Ti air
st
(Z1 ) 1 soil
[1
ZL1
ZC1
[1

[i [i i th soil
i th soil ZCi (Z i )
trapped ZLi
waves limestone rock
f (Z f ) limestone rock
ZL

(b)
Fig 1 . (a) Two-dimensional analysis model that is composed of multi-layer of media; infinite length of air,
thickness [i of ith layer of topsoil, and infinite depth of limestone rocks.
(b) The equivalent circuit of the model.

from different amount of soil moisture is considered.

Fig 1a shows a two-dimensional model of analysis that is composed of multi-layer of media; infinite
length of air, thickness [i of ith layer of soil, and infinite depth of rocks. In this research, to simply the
analysis, the impact of surface roughness on the scattered waves was considered because the average of
roughness is too smallest than the JERS-1’s wavelength. The incident wave was assumed to be a plane
wave with an incident angle T i .
The equivalent circuit of the model used in this analysis is shown in fig 1.(b), where the
effective series impedance of the ith layer of soil, the parallel impedance of ith layer of soil, and total
input impedance are ZCi , Z Li and Ztotal respectively. To simplify the analysis, the parallel impedance of
soils ( Z Li ) is neglected and assumed as zero, and the bedrock layer is assumed to be an infinitely deep
perfect conductor, consequently, Zf is zero. Based on transmission line theory method, the total input
impedance Ztotal is derived from fig 1.(b) and determined by
f
Z total ¦Z
i 1
i (2)

Z Li  Z Ci tan J Ci[ i
Zi Z Ci (3)
Z Ci  Z Li tan J Ci[ i

where J Ci and [i are propagation constant and thickness of ith layer of soil, respectively.

117
 By considering the propagation of the wave transmitted from the air to the soils. Referring to this
figure, the propagation constant J Ci is derived from Maxwell’s equation as:
2S
J Ci j H ri Pri cosTti (4)
Oi

where H i , Pri and Oi are complex dielectric constant, complex specific permeability,
transmission angle, and wavelength of ith layer of soil respectively. j is equal to  1 . The wave
impedance ZCi of transmitted wave in soil is obtained from component of electromagnetic field
perpendicular to the propagation axis.
E yti P ri
ZCi Z0 cosTti (5)
H zti H ri

§ P0 ·
where Z0 ¨ ¸ is the wave impedance in air or free space (=120 S ohms). Based on Snell’s law,
¨ H0 ¸
© ¹
the relationship between
incidence angle T i and transmission angle T ti at the boundary yield
sin T i
H ri P ri (6)
sin T ti

By substituting (4) to (6) into (3), wave impedance Ztotal of incident wave in ith layer of soil
Z0 § 2S[ i ·
becomes Zi H ri P ri  sin 2 T i tan ¨¨ j H ri P ri  sin 2 T i ¸
¸ (7)
H ri © Oi ¹
Furthermore, the total input impedance is obtained by substituting (7) into (2), therefore the
reflection coefficient becomes:
Z total  Z 0 cos T i
* (8)
Z total  Z 0 cos T i

Then the backscattering coefficient V is defined as:

o

V0 20 log( * ) (9)

The other hand let relationship between dielectric constant H ri and soil moisture Ki of ith soil is
defined to
H ri f (K i ) (10)

where f (Ki ) is the correlation function of dielectric constant and soil moisture. By substituting
the (10) into (7), hence the relationship between backscattering coefficient V o and soil moisture of soil
can be obtained from equations (7) to (9).
Dielectric constants H ri of ith soil were measured experimentally using dielectric probe kit
HP58070B in frequency range from 0.3 to 3.0 GHz.

3. Results and discussion

In order to develop a method to retrieve soil moisture, firstly, the relationship between dielectric
constant and soil moisture was investigated. Several studies have been conducted over the past few
decades to study the relationship between dielectric constant and soil moisture (Ulaby and Batlivala,

118
1976; Ulaby et al., 1978). In this study, the relationship between dielectric constant and soil moisture
was obtained by using measured dielectric constant and soil moisture of the soil samples. Figure 3 shows
statistical relationship between the dielectric constant and soil moisture. The correlation coefficient of
H
real part of dielectric constant and soil moisture is 0.76. Dielectric constants r of soil were measured
experimentally using dielectric probe kit HP58070B in frequency range from 0.3 to 3.0 GHz. The
equation (10) describes empirical relationship between dielectric constant and soil moisture
H 'r 2.68e 0.1K
H rcc 0.004K 2  0.01K  0.1 (10)
where H r , H r ,K is real part of dielectric constant, imaginary part of dielectric constant and soil moisture
c cc
respectively.
We considered the average thickness of soil layer on the bedrock in study area as 45cm. And by
[
substituting (11) and =45cm into (8), the relationship between backscattering coefficient ( V ) and
o

soil moisture was obtained. The equation (13) describes the relationship backscattering coefficient and
soil moisture.
K 1.81e 0.23V
0

(12)

where K , V
o
are soil moisture and backscatterin coefficient respectively. The backscattering coefficient
( V ) were calculated using the equation V =20logI-68.2dB (Shimada 2002), where I is the pixel
o o

intensity of JERS-1 SAR data. Employing the developed method, soil moisture was estimated using
JERS-1 SAR L-band data in study area. It was found that the percent soil moisture ranged between 4.5%
and 9%.

Estimated soil moisture with 12 points of ground measured soil moisture of July 2003. The
correlation coefficient was 0.94 and standard error of moisture percentage was 1.66.

References
1. Tetuko S.S., J., Tateishi, R., and Tateuchi, N., 2003a, Estimation of burnt coal seam thickness in
central Borneo using a JERS-1 SAR image. International Journal of Remote Sensing, 24, 879-884.
2. Tetuko S.S., J., Tateishi, R., and Tateuchi, N., 2003b, A physical method to analyze scattered waves
from burnt seam and its application to estimate thickness of fire scars in central Borneo using L-
Band SAR data. International Journal of Remote Sensing, 24, 3119-3136.
3. Ulaby, F. T., and Batlivala, P. P. (1976), Optimum radar parameters for mapping soil moisture. IEEE
transactions on Geoscience and Remote Sensing 14, 81-93

119
 Moving into international geographic information standards

Bolorchuluun.Ch; Battsengel

School of Earth Sciences, National University of Mongolia,

Email: bolorchuluun@num.edu.mn; vbattsengel@num.edu.mn

Key words: standards, GIS, geospatial information, ISO standards

Absract
The purpose of this paper is to provide an introduction to standards and an overview
of the standardization process in terms of the development, implementation, and deployment of
international standards for spatial data infrastructures (SDIs) such as GSDI.
The introduction identifies the multiple functions of standards. Within the context of the
International Organization for Standardization (ISO) Technical Committee 211, Geographic
Information/Geomatics, this introduction describes the functionality of ISO standards beyond
being just technical solutions. These functions include: standards serving as compromises, as
forms of technology transfer, as democratic and as research mechanisms. It also discusses the
consensus process and its implications for standardization.
The necessity for worldwide standardization of Geographic Information is well known.
Standardization is important for the production and use of geographic data, the GIS industry
and the application of their products. The development of national and international
Geographic Information Systems has been proceeding rapidly for many years.
GIS standards, the relationship between the GIS standards infrastructure and spatial data
infrastructures are also discussed.
This paper concludes with a brief review of the status ISO/TC 211.
In the Geo ICT environment, access to different datasets and grid computing becomes more
applicable using the Internet for cross-border datasets. However, still, in many applications,
one suffers from the possibility the transfer meaningful data to each other's applications.

Introduction
The information technology environment has become far more integrated. Web Services, E-
Government, Federated Architectures and Grid Computing, have benefited government and businesses,
helping them to gain access to a substantial amount of data, including geospatial data.
Mongolia is in transition from a centrally planned economy to a market oriented one. It is a
historically unique process without any given rules and scenarios. Transformation of standardization is one
example of deep changes, which generally have had to be made, reflecting quite different understanding of
social and economic systems. The depth of these changes is evident when we realise existing differences
between the concept of standardization and the function of standards in planned and market economies.
 The characteristic features of the earlier approach to standardization were:
- mandatory standards,
- strict state control of standardization
- financing of standardization from the state budget
One of the consequences of adopting free market principles is a need to transform the way that
standards are developed and adopted within the society. This requires fundamental decisions by
government which lead towards:
- a movement from mandatory to voluntary standards
- adaptation of the national system to conditions relevant to a market economy
- harmonization of national standards with the international standards
We are going to solve these problems step by step.
Standards are often deceptive because they serve different functions. Generally, standards are
perceived as consensus accepted technical solutions. Within the International Organization for
Standardization (ISO), consensus does not necessarily imply unanimity or approval by majority. The
notion of consensus, within this context, refers to the absence of sustained objection. Closer scrutiny
reveals that standards are more likely to be political compromises that may have significant roles and
implications in the management, policy, and financial considerations of governments, industry, and user

- 120 -
communities. In this regard, the approved standard is less than likely to be a superior technical solution.
Standards frequently serve as forms of technology transfer between advanced and emerging countries. The
traditional technology lag existing between developed and emerging countries is disappearing because
emerging countries are now joining technical committees within standardization organizations as
participating members or as observers.

Who wants geographic information standards?

The answer is that all businesses that produce, distribute, or utilize spatial information, either alone
or in conjunction with non-spatial information, benefit from spatial standards. These range from
geographic information, decision support, data mining, data warehousing, to modelling and simulation.
Application areas include – but are not limited to – automated mapping, geo-engineering, computer-aided
drafting and design, entertainment, modelling, and simulation. These broad categories span the planning,
design, construction, operation, and maintenance of facilities and their supporting infrastructure such as
communications, transportation, and utilities.
A common way to describe the market is to divide it up into segments: the traditional geographic
information systems (GIS) market, business support systems (BSS), and personal productivity (PP). Let's
look at each:
_ GIS (geographic information systems):
– Spatial information (contributes the most value)
– Traditional market for spatial technology
_ BSS (business support systems):
– Spatial information (does not contribute the most value)
– Spatial technology embedded in business applications
_ PP (personal productivity):
– Users want to communicate by use of maps/geographic information
– Follows the office suite market
A new emerging market is location based mobile services (LBMS), shown left.
Many industry sectors within the marketplace will benefit significantly from interoperable access
to spatial information and services, including such areas as the travel and tourist industries, the mapping
and routing industries, communications, utilities, transportation, national defense, agriculture, disaster
management and public safety, location/mobile services, inventory management, real and synthetic
environmental modelling and gaming, and the emerging needs of electronic commerce for spatial
information.
Achieving more interoperability requires a proactive coordination of spatial standards at both the
abstract and implementation levels.
Proactive cooperation among spatial standards activities should also help to use available resources more
efficiently by minimizing technical overlap, wherever this occurs. Such coordination and cooperation
should lead to more market-relevant spatial standards, and could serve as a useful roadmap for all
interested parties.

Mongolian standards evolution

The standardization agency of Mongolia was initially established in 1953 under the name "Price
and Standardization Bureau". Since that time its structure and responsibilities have been changed several
times. However, it has been consistently engaged in nation-wide management and coordination of
standardization work. Today these functions have been assumed by Mongolian National Center for
Standardization and Metrology (MNCSM).
MNCSM approves and publishes all Mongolian standards and represents Mongolia in
International Organization of Standardization. MNCSM is a member of ISO since 1979. And today it is a
P-participating member of 47 ISO technical committees and subcommittees as well as O-observer member
of 69 ISO/TC and SCs.
Nowadays, MNCSM has 25 Technical committees. One of them is a standardization technical committee
on Information technology, which is open to everybody who is interested to take part. However, members
of technical committees unable to participate in international activities due to pure financial resources and
well-trained experts.
We have got a significant result in harmonization of our National standards with international ones.
We have adopted ISO Directives into Mongolian national standardization system. These are:

- 121 -
 x Part 1. Procedure for technical work. MNS1.1: 99
x Part 2. Methodology of standards work. MNS1.2: 99
x Part 3. Drafting and presentation of standards. MNS1.3: 99
For the promotion of alignment of the MNS (national standards) with international ones, MNCSM
successfully implemented the concept of standards harmonized to ISO/EIC Directives. There are over 3800
Mongolian standards. The percentage of harmonized national standards with international ones is reached
18.4% in all Mongolian standards for last three years. It meant that the percentage increased 3.4 times
more than in 1996 and at the end of 2002 it expected to be reached 35 percents.
MNCSM now has 20 Technical Committees, through which all interested parties can participate in
standardization on concerned fields. Membership in committees is voluntary and it is open to all who are
interested to take part in their work at their own expenses. Each technical committee may establish
subcommittee (SC) and working group (WG) to cover different aspects of its work. Each technical
committee has a Secretary who employed by MNCSM.
Mongolian National Center for Standardization and Metrology (MNCSM) is a governmental
regulatory agency and national standards body responsible for the coordination and management of the
MSTQ activities in Mongolia. The preparation, application and promotion of National standards are set out
in the Mongolian law on “Standardization and Quality Certification” adopted in 1994.
The Mongolian script character set is a coding proposal of Mongolian scripts which includes Mongolian,
Todo, Sibe and Manchu letters, punctuation marks, digits and control characters. The written languages,
Todo, Sibe and Manchu, all share Mongolian letters. Many Mongolian characters have different forms
according to their position in the word (initial, middle or final). According to the relevant principles of
ISO/IEC 10646, only one of those presentation forms has to be encoded. This form is named as basic
character. Due to the official use of Cyrillic character as a writing media of Mongolian language there were
approved the standards for Cyrillic characters based on Microsoft Code page 866 and 1251 respectively for
MS DOS and WINDOWS.
Historically, Mongolia had been and keeping traditional tabby for protection environmental since
including water, pasture-lands, plans and fauns. Mongolia has adopted and ensured the implementation of
several Environmental Laws. These laws, however, did not consider natural processes as ecosystem as a
whole. Instead, they focused primarily on securing right to possess natural resources to the relevant
Ministers.
The entire has been mapped geologically at a scale of 1:1,500,00, much of the country to
1:1,000,000 and 1:50,000, and selected areas to larger scales. Gravity, aeromagnetic and geochemical data
are also available.
For the promotion of alignment of the national standards with international ones, MNCSM
successfully implemented the concept of national standardization system standards harmonized to ISO
Directives. MNCSM sets a priority in IT standardization in order to:
- Facilitate international, regional and national trade,
- Achieve mutual understanding in intellectual, scientific, technical and economic fields,
- Implement and transfer advanced technology and technique,
- Computerize information exchange
- Support mutual understanding and cooperation between governmental authorities, non-governmental
organizations, manufacturing companies and businessman of industry and commerce.

Standardization of geographic information

The International Standards Organization (ISO), the Technical Electro-technical Commission
(IEC), and the Telephone Consultative Committee (CCITT), all located in Geneva, are responsible for the
standardization at the international level. In ISO, the national standards bodies of some 120 countries
cooperate in activities that aim to facilitate the international exchange of goods and services by creating
uniform standards with global validity, and to stimulate cooperation in the scientific, technical and
economic fields across national frontiers.
Since 1994, DGIWG and IHO have played a special role in the development of geospatial
standards. More recently the OGC, which has now entered into a cooperative agreement with ISO/TC211,
has a significant impact on the standards formulating process. The results of this effective cooperation are
ISO Standards.
FIG takes special note of the ISO activities in order to transfer knowledge about ISO/TC211
standards to its members for practical use. A survey conducted by FIG shows that there is a general lack of

- 122 -
knowledge and practice of the official standardization. Many other useful links to ISO have been started by
a special task force on standards, established at the FIG congress in 1998 in Brighton.
Since ISO/TC211 was established, in 1994 (Secretariat NTS, Norway) this committee has been
steadily increasing. There are now 33 P-(participating) members and 19 O-(observer) members.
The ISO 19100 is a series of standards for defining, describing, and managing geographic
information. This standard defines the architectural framework of the ISO 19100 series of standards and
sets forth the principles by which this standardization takes place. Standardization of geographic
information can best be served by a set of standards that integrates a detailed description of the concepts of
geographic information with the concepts of information technology. A goal of this standardization effort
is to facilitate interoperability of geographic information systems, including interoperability in distributed
computing environments. Figure 1 depicts this approach.
Geographic Information

•Spatial reference Framework and Reference Model

•Temporal reference
•Spatial properties Reference Model, Overview
•Spatial operations Conceptual schema language,
•Topology Terminology, Conformance and testing
•Quality
•………... Geographic Information
Services
•
Positioning services
Portrayal
Services
Encoding
Data Profiles &
Administration Data Models
& Operators Functional
Cataloguing Standards
Spatial schema
Spatial Reference
Temporal Schema
Descriptive Reference
Information Technology Spatial Operators
Quality
Rules for
Quality Evaluation
Application Schema
•Open Systems Environment (OSE) Procedures
•Information Technology Services Metadata
•Open Distributed Processing (ODP)
•Conceptual Schema Languages (CSL)
•………………………….

Figure 1 — Integration of geographic information and information technology

The ISO 19100 series of geographic information standards establishes a structured set of standards
for information concerning objects or phenomena that are directly or indirectly associated with a location
relative to the Earth. This standard specifies methods, tools and services for management of geographic
information, including the definition, acquisition, analysis, access, presentation, and transfer of such data
in digital/electronic form between different users, systems and locations. In figure 1, the ISO 19100 series
of geographic information standards can be grouped into five major areas, each of which incorporate
information technology concepts to standardize geographic information. These major areas describe the:
- The framework for the ISO 19100 series of geographic information standards including ISO 19101,
Geographic information  Reference model. The framework and reference model cover the more
general aspects of the ISO 19100 series of standards. The reference model identifies all components
involved and defines how they fit together. It relates the different aspects of the ISO 19100 series of
standards together and provides a common basis for communication.
- Geographic information services define the encoding of information in transfer formats and the
methodology for presentation of geographic information that is based on cartography and the old
traditions of standardized visualisations.
- Data administration is concerned with the description of quality principles and quality evaluation
procedures for geographic information datasets. Data administration also includes the description of the
data itself, or metadata, together with feature catalogues. This area also covers the spatial referencing
of geographical objects - either directly through coordinates, or more indirectly by use of, for instance,
area codes like postal or zip codes, addresses, etc.
- Data models and operators are concerned with the underlying geometry of the globe and how
geographic features and their spatial characteristics may be modelled. This area defines important
spatial characteristics and how these are related to each other.
- Profiles and functional standards consider the technique of profiling. Profiling consists of
putting together “packages/subsets” of the total set of standards to fit individual application

- 123 -
 areas or users. This supports rapid implementation and penetration in the user environments
due to the comprehensiveness of the total set of standards. Equally important is the task of
“absorbing” existing de facto standards from the commercial sector and harmonizing them
with profiles of the emerging ISO standards.
ISO/TC211 Working Groups and Projects are 31 projects handled by 5 working groups and 2 project
teams work directly under ISO/TC211 (ISO/TC211-N1042, 2001).
Working group 1-Framework and reference model
19101 Geographic information – Reference model
19102 Geographic information – Overview
19103 Geographic information – Conceptual schema language
19104 Geographic information – Terminology
19105 Geographic information – Conformance and testing IS
19121 Geographic information – Imagery and gridded data TR
19124 Geographic information – Imagery and gridded data components Stage 0
Working group 2 – Geospatial data models and operators
19107 Geographic information – Spatial schema
19108 Geographic information – Temporal schema
19109 Geographic information – Rules for application schema
19123 Geographic information – Schema for coverage geometry and functions
19110 Geographic information – Feature cataloguing methodology
19111 Geographic information – Spatial referencing by coordinates
19112 Geographic information – Spatial referencing by geographic identifiers
19113 Geographic information – Quality principles
19114 Geographic information – Quality evaluation procedures
19115 Geographic information – Metadata
19126 Geographic information – Profile – FACC Data Dictionary NP
19127 Geographic information – Geodetic codes and parameters NP
Working group 4 – Geospatial services
19116 Geographic information – Positioning services
19117 Geographic information – Portrayal
19118 Geographic information – Encoding
19119 Geographic information – Services
19125 Geographic information – Simple feature access–Part 1: Common architecture
19125-2 Geographic information – Simple feature access–Part 2: SQL options
19125-3 Geographic information – Simple feature access – Part 3: COM/OLE options
Working group 5 – Profiles and functional standards
19106 Geographic information – Profiles Final 2.CD
19120 Geographic information – Functional standards 2.CD
19120 Amd. 1Geographic information – Functional standards – Technical amendment NP
Project directly under the ISO/TC211
19122 Geographic information/Geomatics –Qualification and Certification of personnel NP
19128 Geographic information - Web Map server interface
The major clauses of the Reference model are Conceptual modelling, the Domain reference model
(clause 8), the Architectural reference model, and Profiles. These clauses detailed in ISO 19101 are related
to the major areas of the ISO 19100 series of geographic information standards. These relationships are
summarized in figure 2, and explained in the paragraphs that follow.

- 124 -
 Conceptual modelling Architectural reference model
(Clause 7) (Clause 9)

Geographic information
services

•Positioning services
•Portrayal
•Services
•Encoding
Data
administration Data models Profiles &
& operators functional
•Cataloguing standards
•Reference by coord. •Spatial schema
•Reference by geo. id. •Temporal schema
•Quality •Spatial operators
•Quality evaluation •Rules for application
procedures schema
•Metadata
Domain reference model Profiles
(Clause 8) (Clause 10)

Figure 2 — Relationship of the Reference model to other standards

in the ISO 19100 series of geographic information standards

The strategic direction of an international deployment

of geographic information standards
The mandate for ISO/TC 211 – that at least which is implied – is to develop an integrated set of
standards for geographic information.
Equally important, if not more so, is the unstated strategic direction of the international
deployment of such standards. Accordingly, the strategic directions for ISO/TC 211 can be viewed in
terms of development, deployment, and the underlying coordination/consensus process that integrates both
these phases in successful standardization.
Over and above standardization of traditional geographic functionalities, innovative, new, and
unknown technology and application domains present challenges that transcend the established process of
Geographic standardization.
In terms of development, the major issues include: the technical development of standards, the
organizations developing geographic or related standards, the priorities within standards, standards and
interoperability testing, and the speed of developing technical specifications. As to deployment, the key
issues are: the implementation of standards, standards education/training, and the user communities
supporting ISO/TC 211
standards.
Inherently present and all-pervasive throughout the standards-development process, the
deployment of standards and their coordination/ consensus process are considerations for the implementers
and users of geographic standards: such items as data transfer standards that are implemented by vendors,
or data cataloguing standards implemented by data producers, or metadata standards implemented by
vendors, data producers, and general users of geographic information.
Implementers and user requirements need to be considered in conjunction with the standards development,
deployment, the process of integrating such requirements.
Traditionally, geographic information was produced and used by the geographic community.

Conclusion
In this age of globalization, the human communication, it can only be achieved through shared
efforts and information technology that effort to the needs of tomorrow’s ever more interdependent and
ever more technologically advanced world. As we entered a new century, we need to build a shared
information and ensure a shared growth benefits through integrating developing countries or countries of
economies in transition.
We need to establish the Mongolian Geospatail Standardization System. However the size of the
task of considering, adopting and publishing Mongolian geographic information standards cannot be
underestimated. MNCSM recognises it has much to do and that it needs to do all it can to speed up the
process.

- 125 -
 The geospatial information community is a mature and well developed community, which over an
extended period has proven its worth in developing tools to make geospatial data accessible from multiple
technologies and software vendors. At the same time organisations have started to develop their policies
and data management practices to ensure they are ready for the new demands and potentials of
interoperability.
Increasingly, geographic information is being created and used by everyone else, especially people
in the business community.

References
1. The SDI Cookbook, version 1.0, GSDI, www.gsdi.org
2. International Organization for Standardization, www.iso.ch
3. ISO/TC 211 Geographic information/Geomatics, www.statkart.no/isotc211/
4. Joint Steering Group on
5. [AFNOR, 1992], Association Franɡaise de Normalisation, see: http://www.afnor.fr
6. [CEN, 1992], Comitɣ Europɣen de Normalisation CEN/TC 287 - Geographic Information;
7. see: http://www.cenorm.be
8. [EU,2003], European Union, framework programmes; see:http://europa.eu.int/pol/rd/index_en.htm
9. Eurogeographics, 2003], a network of National European Mapping Agencies (NMAs),
10. see: http://www.eurogeographics.org
11. [EUROGI, 1993], European Umbrella Organisation for Geographic Information
see:http://www.eurogi.org
12. [DIGEST, 2003], DIgital Geographic Information Exchange Standard;
13. See:http://www.digest.org of the Digital Geographic Information Working Group - DGIWG
14. [INSPIRE, 2002], Data Policy and Legal Issues Position Paper, 2002, see: http://inspire.jrc.it
15. [INSPIRE, 2006], Infrastructure for Spatial Information in Europe Initiative; see:
http://inspire.jrc.it
16. [ISO, 2006], International Organisation for Standardisation; see: http://www.isotc211.org
17. [SNV, 2003], (Schweizerische Normen-Vereinigung, see: http://www.snv.ch

- 126 -
LUCC and Terrestrial Study based on RS, GIS and Ecological Observation &
Proposal for International Collaboration
Jiyuan Liu Lin Zhen Yunfeng Hu Qian Zhang
(Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, PRC)

Land Use/Cover Changes (LUCC) is a core element leading to terrestrial ecosystem dynamics and
socio-economic consequences, and the relevant research has been the focus of many institutions and
scientists worldwide. The Institute of Geographic Sciences and Natural Resources Research (IGSNRR) is a
leading institute in China in the field of LUCC and terrestrial research through application of sophisticated
technology and method such as RS, GIS and ground based ecological observations. Over the decades, the
Institute has contributed significantly to theoretical development and practical application of LUCC and
terrestrial study, and ultimately to sustainable management of land systems and ecosystems of the country
in specific and the surrounding Central/North-east Asian countries in general. This paper summary the
major achievements made in those fields, and propose points for future collaboration with international
partners.

Major research achievements:

ƒ LUCC study based on RS and GIS: both spatial (eg. TM/ETM㧘MODIS) and attribute data
covering land use, environmental and social and economic conditions were gathered for
exploration of LUCC dynamics and major drivers of China over the past 20 years. Land surface
key parameters (eg. NDVI, LAI, EVI, fPAR) inversion algorithm and software, land surface
temperature, Leaf Area Index and annual changes, and tree distribution of China have been
developed through a long term based research and monitoring. The database of LUCC and
environmental changes of the entire country has been established, and the LUCC zones were
identified for improved management.
ƒ Terrestrial ecosystem observation and research: ChinaFLUX (Chinese Terrestrial Ecosystem Flux
Research Network) and CERN (Chinese Ecosystem Research Network) covering typical terrestrial
ecosystems with powerful long-term quantitative observation functions have been established.
There are now 36 and 8 countrywide stations respectively. The networks are providing strong basis
for quantifying carbon budget, estimating terrestrial ecosystem NPP (eg. Cropland, forest through
developing and using CEVSA model), simulating carbon cycle, assessing the effects of increasing
atmospheric CO2 and climate variability, and land use changes through integration of both top-
down and bottom-up approaches. Integrated use of process-based ecosystem model (CEVSA
Model) and data from weather stations, FLUX and natural resource surveys, and RS model (eg.
GLO-PEM model) driven by Remote Sensing data are the main features of the network function.
Integrated Ecosystem Assessment of Western China (MAWEC) is a successful example of
terrestrial ecosystem assessment via application of all those means. It was found, for example, that
the NPP in western China was approximately 70% of the total amount in China in twenty years,
and policy options for improvement of ecosystem services of the region have been made and
adopted by the government.
ƒ Ground-based observation for research of impact of climate changes on ecosystem services, land
degradation and permafrost changes. It was visualized, for instance, the permafrost changes in the
form of digital map by means of GIS analysis in Tibet. The research is to be expanded to Mongolia
through using PulseEKKOPRO ground-penetrating radar.
ƒ Urbanization monitoring: mapped spatio-temporal changes of urban land expansion in China using
high-resolution Landsat Thematic Mapper and Enhanced Thematic Mapper data since the 1980s. It
was found that China’s urban expansion had a high spatial and temporal variability. Case studies
of the thirteen large cities showed that urban expansion has been driven by demographic changes,
economic growth, and changes in land use policies and regulations.
ƒ Forest fire monitoring: large scale forest fire maps were developed using MODIS data, and have
been used for the fire monitoring and forecasting.
ƒ Powerful platform and support: Data-Sharing Network of China Earth System Science and GIS
software-SuperMap, self-developed GIS software with powerful functions are strong basis for

- 127 -
 conducting relevant researches.

International collaboration and proposal for further research:

The IGSNRR has been keeping close contact with international partners in the field of LUCC and
terrestrial ecosystem research under the framework of GOFC-GOLD. Major contributions and plans are
listed, but not limited as follows:

ƒ Strengthening capacity building: the IGSNRR has already kicked-off GIS training program for the
scientists from the Institute of Geography, Mongolian Academy of Sciences, and scheduled to
offer training courses for researchers from Russia, Kazakstan, and North Korea. Apart from
international activities, the Institute is focusing on domestic needs and strengthening capacity
building for economically backward western regions. The future training will cover the fields of
land system, terrestrial ecosystem assessment, carbon cycling, water cycling, socio-economic
systems, urban and settlement development, etc.
ƒ Enhancing collaboration with economic powers such as Japan and South Korea to establish
working groups (WG) in the East Asian Region. The WG will take a leading role in the sustainable
development of the Central/North-east Asian nations. A network and data sharing system among
the countries will be established.
ƒ Promoting joint research activities: the IGSNRR has been implementing joint research projects
with Mongolia and Russia in the field of ecosystem and environmental changes and sustainable
development. Those activities will be further expanded to other surrounding Central/North-east
Asian countries following the framework of GOFC-GOLD.

- 128 -
 REMOTE SENSING PARAMETERIZATION OF LAND SURFACE HEAT
FLUXES OVER ARID AND SEMI-ARID REGION IN MONGOLIA

Jadamba Batbayar , Nas-Urt Tugjsuren

Mongolian University of Science and Technology

E-mail: tugjsurn@must.edu.mn

Abstract
Remote sensing in the thermal infra-red spectrum is essential for the estimation surface
temperature and the components of the surface energy balance. The intent of this study is to
evaluate the applicability of the remotely sensed surface energy balance method to a regional land
surface heat fluxes distribution for arid and semi arid surface with homogeneous area in Mongolia.
The Dundgovi region is selected to apply surface energy balance method for heat fluxes distribution
in arid and semi-arid region of Mongolia. The surface heat fluxes derived from Landsat+ETM in
the morning overpass satellite, and computed on a pixel-by-pixel basis on 10.August.1999.

Key words: Net radiation, Soil heat flux, Sensible heat flux, Latent heat flux
Introduction
The study on the energy exchanges between the land surface and atmosphere is important to
understand for arid and semi-arid regions environmental changes. Remote Sensing from different satellite
sensors offers the possibility to derive regional distribution of land surface heat fluxes. Many new remote
sensing algorithms to estimate surface energy balance components have been proposed during the last 10
years. An overview of the most common one is presented in Kustas W. P, et al. (1994), Moran M. S, et al.
(1994, 1995). Remote sensing of surface temperature (i.e. radiative surface temperature) together with
some ground-based data has been widely used in conjunction with simple one-dimensional models to
estimate components of the energy balance equation from field to regional scales (Jackson, R, D. 1985).
Briefly, the method uses the energy balance equation with an estimate of the surface temperature from an
IR sensor; wind speed, air temperature and incoming solar radiation come from ground measurements. The
purpose of this study is to estimate regional land surface heat fluxes distribution for large arid and semi-
arid surface with homogeneous area condition. The method is based on Surface Energy Balance and
Normalized Difference Vegetation Index (NDVI) as well as surface parameters. The parameters will be
derived from LANSAT+ETM satellites data, taken on 10.August, 1999 supplemented by ground
observations.

THE DATA SET AND STUDY AREA

The Landsat-ETM satellite images on 10 August 1999 were evaluated for Land Surface Heat Fluxes
distribution in Dundgovi of Mongolia. These images overpass time was 10:00 AM of Landsat-ETM
path/row 131/28 on local time. These images had favorable weather conditions without clouds in study
area. Data from ground measurement area were available to assist the calculation of the Land Surface Heat
Fluxes in the locations of study area. (Lat: 45.49N, long: 106.12E). Table 1.summarizes the meteorological
ground conditions at the study area of satellites overpass time on 10.August, 1999.

Table1. Ground measurement data with selected satellites overpass in Mandalgovi (lat: 45.49N, long:
106.12E) of Mongolia on 10. August, 1999.

Time Cloud Wind Air Air Relative Ground Incoming Incoming

(0-10) Speed pressure temperature humidity temperature solar longwave
(m/s) (hp) (oC) (oC) radiation IR (Wm-2)
10 0 9.4 848.8 18.65 23.9 27.1 701.6 281.9

The study area is Dundgovi aimag (province) of Mongolia. The coordinates are 44000cN and
46000cN, 103000cE and 109000cE, and the area is 78 thousand km2 an arid and semi-arid region.
Annual mean precipitation is about 150-250 mm and 85-90 percent of the annual precipitation falls

- 129 -
 as rain during the summer of which 50-60% in July and August. Cloudy skies are infrequent, and
4,467 hours of sunshine per year.

METHOD
The surface energy balance method is an image-processing algorithm which calculates
energy exchanges at the earth’s surface using digital image data collected by Landsat or other
remote sensing satellites by measuring visible, near-infrared and thermal infrared radiation.
Instantaneous net radiation values were computed according to incoming and reflected solar and
thermal radiation. Rn is computed for each pixel using albedo and transmittances computed from
short wave bands and long wave emission computed from the thermal band and some ground data.
(See Eq 1.) Soil heat flux (G) is predicted using vegetation indices computed from combinations of
bands and net radiation. Sensible heat fluxes (H) is calculated from several factors: surface
temperature and wind speed measurements from ground data, and estimated surface roughness and
surface –to-air temperature differences predicted from vegetation indices. LE is latent heat flux,
which is the energy used to evaporate water. All computations were made pixel by pixel.

Net radiation
Usually, net radiation can be estimated by:
4
Rn (1  D ) Rs  HG (H a Ta  Ts4 ) (1)
Where Rs is the incoming short-wave solar radiation, D the surface short-wave albedo, H the surface
emissivity, G the Stefan–Boltzman constant, and ea the effective atmospheric emissivity, with
(Brutsaert, 1975)
1/ 7
§e ·
HD 1.24¨¨ a ¸¸ (2)
© Ta ¹
Where eD is the atmospheric vapour pressure.
The surface emissivity is calculated as a weighted average between bare soil and vegetation.
Where Hv is the emissivity of vegetation assumed to be 0.95, Hs, the emissivity of bare soil assumed
to be 0.85 and fv the fractional vegetation cover.

H H v f v  H s 1  f v 
H v | 0.95 (3)
H s | 0.85
.
The fv vegetation coverage in a pixel, is yielded by a relationship between f and NDVI (Gutman
1998):
NDVI  NDVI min
f (4)
NDVI max  NDVI min

Where NDVImax and NDVImin are the maximum and minimum NDVI in a whole growth of
vegetation.

Soil heat flux

The soil heat flux can be empirically estimated using the net radiation and LAI or a satellite derived
vegetation index, such as the normalized difference vegetation index, NDVI. In this case the Moran
et al. (1989) model was adopted as:

G Rn >0.58 exp 2.13NDVI @ (5)

Where NDVI=(NIR_RED)/(NIR+RED), and NIR and RED are the near-infrared and red reflectance’s,
respectively.
Sensible heat flux

- 130 -
 The sensible heat flux can be written as:
H UC p (Ts  Ta ) / ra (6)
Where UCp is the thermal capacity of the atmosphere, (ra) aerodynamic resistance and (Ta) is the air
temperature at the reference height where meteorological measurements are available. The land
surface temperature (Ts) can be derived from a remotely sensed radiometric surface temperature.

Latent Heat flux

The regional latent heat flux is derived as the residual of the energy balance theorem for the land
surface, i.e.,
LE Rn  H  G (7)
Where Rn is the net radiation, H is the sensible heat flux, G is the soil heat flux, and LE is the latent
heat flux (W/m2).

Results

The main output of Surface energy balance method is the partitioning of energy balance. The
figure 1 shows the results of the different land surface heat fluxes (Rn, G, H, LE) distribution
for the study area and surface variables (NDVI, LST) have been calculated using Landsat-
ETM data with ground observations over the homogeneous areas of Mandalgovi on
10.August, 1999. The distribution range of land surface heat fluxes over the study area is
shown in Table2.

Net radiation (Rn) Soil heat flux (G)

Wm-2 Wm-2
751 874 240 258

- 131 -
 Sensible heat flux (H) Latent heat flux (LE)

Wm-2 Wm-2
212 467 136 368

NDVI Co
-0.15 0.3 25 40

Vegetation indices (NDVI) Land surface temperature (LST)

Figure 1. Maps of land surface fluxes and surface variables for the arid and semi
arid region of Dundgovi aimag. Land surface fluxes calculated from Landsat-ETM+
satellite data on 10.August, 1999. Lines are illustrated administrative boundary of
soum level.
Table 2. The distribution of land surface heat fluxes over study area
Surface heat Range of Dundgovi area Mandalgovi (station)
fuxes (Wm-2) (Wm-2)
-2
Rn (Wm ) 751 - 874 836.92
Go (Wm-2) 240 - 258 250.71
H (Wm-2) 212 - 467 339.81
LE (Wm-2) 136 - 368 246.4

The map for vegetation indices (NDVI) shows high values 0.1-0.3 in the northern part of the
Dundgovi, which in vegetated area with low values of land surface temperature 25-28Co.
And sparsely vegetated and bare soil regions have low values of vegetation indices –0.15-
0.01 related with high land surface temperature 29-40 Co.

Table 3. Shows comparison of ground measurement data with overpass satellites data in
Mandalgovi of Dundgivi (lat: 45.49N, long: 106.12E) on 10.august, 1999
Ground Incoming solar Incoming
Time temperature ( ) radiation (Wm-2) longwave IR, (Wm-2)

- 132 -
 10 AM 27.13 701.568 282.934
Ground
measurement
10 AM 30 647.00 304.98
Landsat-ETM
The some results are presented in Table 3, which illustrated comparison of satellite data with ground
measurement data. But satellite estimated surface heat fluxes (net radiation flux Rn, soil heat flux G,
sensible heat flux H, latent heat flux LE) are did not made comparison, due to insufficient of ground
measured data.

Conclusions

The surface energy balance method has been applied for regional distributions of land surface
variables (land surface temperature and vegetation index) and land surface heat fluxes (net radiation
Rn, soil heat flux G, sensible heat flux H, latent heat flux LE). The land surface temperature,
incoming solar radiation and incoming long wave radiation derived from Landsat-ETM satellite was
approximately 80% near with ground observations. It’s concluded satellite data is better convenient
to the estimation of the distribution of land surface fluxes and land surface variables for large scale
of real time observations.
The study has demonstrated that surface energy balance method can be determined successfully
from satellites. In future studies, more attention should be paid to the validation of satellite estimation and
ground measurements of land surface variables and land surface heat fluxes for arid and semi-arid region
of Mongolia.

References

1. J.Batbayar, N.Tugjsuren, (2005). Net radiation estimation using MODIS-TERRA data

for clear sky days over homogeneous region of Mongolia, Proceeding of the CEReS
International Symposium, Chiba , Japan, 2005,p 206-213.
2. Jackson, R. D., Pinter, Jr., P.J., and Reginato, R. J., (1985). Net radiation calculated
from multispectral and ground station meteorological data, Agric. For. Meteorol., 35,
153-164.
3. Kustas, W. P., Moran, M. S., Humes, K. S., Stannard, D. I., Pinter, P. J. Jr., Hipps, L. E., Swiatek,
E., & Goodrich, D. C. (1994). Surface energy balance estimates at local and regional scales
using optical remote sensing from an aircraft platform and atmospheric data collected over
semiarid rangelands. Water Resources Research, 30 (5), 1241– 1259.
4. Moran, M. S., Kustas, W. P., Vidal, A., Stannard, D. I., Blanford, J. H., & Nichols, W. D.
(1994). Use of ground-based remotely sensed data for surface energy balance evaluation of a
semiarid rangeland. Water Resources Research, 30 (5), 1339–1349.
5. Moran, M. S., Jackson, R. D., Clarke, T. R., Qi, J., Cabot, F., Thome, K. J., & Markham, B. L.
(1995). Reflectance factor retrieval from Landsat TM and SPOT HRV data for bright and dark
targets. Remote Sensing of Environment, 52, 218– 230.
6. Stewart, J. B., Kustas, W. B., Humes, K. S., Nichols, W. D., Moran, M. S., & de Bruin, H. A. R.
(1994). Sensible heat flux — radiometric surface temperature relationship for 8 semi-arid sites.
Journal of Applied Meteorology, 33, 1110 – 1117.
7. Tugjsuren N, Batbayar J (2005): Satellite detection of the atmospheric aerosol for some region
of Mongolia, First National Conference on RS and GIS applications, May 02-03 Ulaanbaatar,
2005, p 77-82
8. Wukelic, G. E., Gibbons, E. E., Martucci, L. M., & Foote, H. P. (1989). Radiometric
calibration of Landsat Thematic Mapper thermal Band. Remote Sensing of
Environment, 28, 327–339.

- 133 -
A Possibility of Cooperation in Detection of Water and Heat Losses of District
Heating System in Ulaanbaatar
Ulaanbaatar1 T., Legden2 M. and Danbayar1

1
Geophysical Department, National University of Mongolia, tarzad@yahoo.com;
2
NewCom Group

Abstract. The heat and water losses in the underground pipelines and structures of District Heating
System in Ulaanbaatar are very large. The detection method of location and intensity of losses and
underground repair and maintenance of leakage are also carried out by multitude difficulties. In this
paper a new initiative to cooperate for detection of heat and water losses in District Heating
Network by remote sensing indicators as high resolution satellite IR images, airborne thermal
methods and in-situ is shown.

Introduction.

The pipelines and structures of District Heating System are established in 1970, 1983 and 1987. For this
reason, the pipelines of the over ground and under ground are exceedingly aged. The water and heat losses
become very large.
10% of total heat energy production losses whatever way, but more than 50% of this loss is in the
underground pipelines.

Figure 1. Additional water (million tons) in 1991-1996

62% of water losses was in the distribution, transmossion and branch piplelines.
In results of project implementation of Asian Development Bank and DANIDA the water losses decreased
in 1999-2003. (Figure 2)
4600

4400

4200

4000

3800
1998 1999 2000 2001 2002 2003 2004

Figure 2. Additional water (million tons) in 1999-2003

Now we would like to show by layouts

- 134 -
 Determination of difficulties and problems

Main goals of UB District Heating

Company is

Difficulties in detection of
defects
R i tl ti

x Increasing the expenditure of water

E ceeding the additional ater (losses of

Problems:
x To improve the marketing of heat transport economic
compensation and compensate the heat distribution
x To evaluate correctly the losses of heat energy
producing from power plants within the distribution
d i i i li

Losses by norm are Today 15-20%

50-100 times

Large part of heat and electric energy producing

Electric Power Plants expends to pump and to heat the
additional water

To new technology
To best technology

Remote Sensing
- 135 -
 Conclusion

For decreasing the water and heat losses, and detection of location and intensity of their leakages the time
serious high resolution images of ASTER and IKONOS need.

Reference

1. Ulaanbaatar, T., Tsolmon, R., and M. Erdenetuya, Detection of heat and water leakages in
Ulaanbaatar District Heating Network by Remote Sensing, First National Conference on
Remote Sensing and Geographical Information System Applications, Mongolian Geoscience
and Remote Sensing Society, National University of Mongolia, Ulaanbaatar, 2 May, 2005.
2. www.worldbank.org/html/fpd/esmap/pdfs/247-01_1.pdf
3. www.adb.org/printer-friendly.asp?fn=/Documents/News/1996/nr1996145.asp
4. www-wds.worldbank.org/servlet/.../Rendered/INDEX/multi0page.txt

- 136 -
 Heat Losses

Resident Building
District Heating

Influencing Factors: Heat Influencing factors:

From losses
x Length of pipelines insulation x Condition of exploitation
x Insulation by in pipeline system
of the
x Environment x Densitization of pipelines
pipelines
water losses x Deadline of exploitation

Real heat losses: Additional water since 1994 in

By the Russian x In central pipelines 5-30% District Heating System has
norms of 1999 increased 2 times and 300 By the Russian
1.1. Branch pipeline 10- repairs each year norms of 1999

In result of change of pipelines by

pre-isolated pipelines the losses will
decrease x Changes by siliphon compensator
¾ 2-5 times. x Partial compensation of density

- 137 -
 Monitoring of Saxaul forest in Gobi of Mongolia
B. Suvdantsetseg1, Y. Aruinzul 2 D.Narantuya 3
1
“NUM-ITC-UNESCO” laboratory for Remote Sensing/GIS , The National University of Mongolia,
E-mail: suvd16@num.edu.mn
2
“State Specialized Inspection Agency, E-mail: y_aruinzul@yahoo.com
3
. Ministry of Mongolian Nature and Environment E-mail: dnarantuya@yahoo.com

Abstract
The paper aimed at determining the relative proportions of saxaul forest cover in the Gobi
region of Mongolia. A Linear Mixing Model was applied in the study area to map the saxual forest.
Three endmembers (saxaul forest, vegetation, soil) were identified based on the image and a
constrained least-square solution was used to unmix the image. The reflective channels of SPOT-4/
VEGETATION satellite over years 1998-2003 were used for the analysis and monitoring for Saxaul
forest. The output results area a contribution to research on vegetation cover change in southern
region of Mongolia
Key words: Saxaul forest, Linear Mixing model, desertification

Introduction
The main objective of the study is to monitor and map the saxaul forest in Mongolia. This has
never before been completed with satellite remote sensing imagry. The saxaul forest is dominated by an
endemic brush type plant and is found in the deserts of Mongolia ( figure 1). It is presumed that many
hectars of saxaul forest are disappearing every year due to logging and its gathering for firewood. The
saxaul forest grows in an area within the Gobi of about 1650 km wide, and about 360 km in diameter from
north to south. There are about 39 subprovinces and 7 provinces where the Saxaul forest is presumed to
grow in Mongolia.

Few scientists have ever studied the saxaul forest. The first scientific studies of the Saxual forest
were completed in the 1940’s and only random notes were taken in previous years. Two censuses of the
saxual forest were completed during the years 1960 and 1997. Between 1961 to 1968, J.Gal (ref)
completed several studies on the saxual forest, where he described the natural history of the saxual forest,
such as its area, growth patterns etc., as well as it’s use by local inhabitants. J Gal (ref) also completed
other research on the fund of zag and its enemies and diseases.

The importance of the saxaul forest to Mongolia is that it stabilizes active sand dunes and reduces
the effect of sand storms. The saxual forest does this because of its root size and depth, which holds most
of the soil moisture in arid and hyper-arid environments. The sparsely structered spatial pattern of
individuals within saxual forest stands also contributes to the saxual forest being a natural defense against
sand storms. If the saxual forest is decreasing in area due to logging and gathering for firewood, it would
be one reason why sand storms are becoming worse in Mongolia. It is therefore necessray to monitor the
saxual forest over large areas in Mongolia. In this study, we attempt to map and monitor the area of the
saxual forest with satellite remote sensing techniques.
Remote sensing is defined as the science and art of obtaining information about an object, or phenomenon
througth the analysis of data acquired by a device that is not in contact with the object, area or
phenomenon under investigation (Lillesand 1994). Land cover classifications are among the most
important applications of remote sensing; however, classifying vegetation cover is problematic because
there is no standardised approach for classifying and mapping different land cover types. This is related to
many factors, such as the spatial heterogeneity of different vegetation structures, vegetation classification
and plant species idenification, plant geometry and biomass. The linear mixing model (LMM) approach is
one of the most often used methods for handling the mixed pixel problem.

- 138 -
 Figure 1. Saxaul forest in the Gobi

Study area
Figure 2 shows the seven Gobi region provinces where the saxaul forest is presumed to be located, and
these provinces are generally in the east south and west south parts of the Mongolia (41° 30’ -49° N and
90°43/- 113°E). These provinces border China to the south, west and east, and have a total area of 639.000
km2. The basic economy is livestock ranching. The topography of these provinces consists mainly of a
plateaus and mountain ranges.

Figure2 The location of the study area

The climate of these province is characterized by short dry, summers and a long cold winter season.
The plant diversity varies between populus diversifolia, saxaul, tamarisk, hippobopac, peplars, willows are
along the humid areas and larch, curgana pygmaya, sympegma and bramble. Wild fauna consist of wild
sheep, ibex, deer, mountains-funas, snow-leopard, lynx, corsa, wild cat, wolf, fox, ermin and marmot.
There are also wild-camel, wild-horse, saiga, tatarica Mongolia, black tailed antilope, and gobi-bear in the
semi-desert.

Methodology

Several techniques (Smith et al. 1985, Shimabukuro 1987, Adams et al. 1989) have been developed to
solve the mixture problems in a number of fine spatial resolution data sets from Multispectral Scanner
System(MSS); Thematic Mapper (TM) data (Adams and Adams 1984, Shimabukuro 1987); and AVIRIS
(Airborne Visible/Infrared Imaging Spectrometer) data (Gillespie et al.1990). All of the above techniques
produced similar results (Shimabukuro 1987) and their uses are usually dictated by the investigators
personal preference.
As there are the same number of equations as there are unknowns, proportion was solved directly as
opposed to least-squares (Shimabukuro and Smith 1991) methods to minimize the error. Quarmby et al.

- 139 -
(1992) applied linear mixture modeling to AVHRR data to estimate crop coverage. The mixture model
(e.g., Settle and Drake 1993) was used to separate out the green vegetation component from other
components. The resulting land cover map was then used as a boundary condition in biosphere models to
estimate the exchange of radiation between the boosher-atmosphere, as well as sensible and latent heat
parameters and surface roughness (Dickinson 1995, Sellers et al.1997) . Studies by Cross (Cross et.al
1991) showed the model produced image outputs in which pixel intensities indicated the proportion of
forest cover per square kilometer. Unmixing had already been applied to coarse resolution data in a
number of studies, especially for vegetation monitoring. While some were based on the first two channels
of NOAA-AVHRR (Quarmby et al. 1992, Hlavka and Spanner 1995) others used the reflective part of the
third channel as well (Holben and Shimabukuro 1993, Shimabukuro et al. 1994). The first four AVHRR
channels were used by Cross et al. (1991) for unmixing and was able to differentiate tropical forest from
non-forest, with satisfactory results compared with TM images. More recent studies (Bastin 1997, DeFries
et al. 1997) reflect the ongoing interest in sub-pixel analysis using coarse resolution satellite imagery.
The linear mixture model based on Optimization Method (R.Tsolmon 2003) was applied in this study to
the saxaul forest monitoring using SPOT/VEGETATION -4 10 day composite, 1-10 August , 1 km2
resolution data from 1998 to 2003. The saxaul percentage images was derived (Figure 3) using spectral
bands of the VEGETATION sensor. For this study, three bands were used, band1-reflectance at 0,50-
0,59µm, band2-reflectance at 0,61-0,68µm, band3-reflectance at 0,79-0,89µm.

Linear Mixing Model

The Linear Mixing Model approach assumes that the spectrum measured by a sensor is a linear
combination of the spectra of all components within the pixel. For solving the LMM, Lagrange method and
optimization technique were used. This method was developed for assumed n components in a pixel. The
matematic model of LMM can be expressed as shown in equations (1-2).
R1 a11 x1  a12 x 2  "  a1n x n  e1
R2 a 21 x1  a 22 x 2  "  a 2 n x n  e2
! !!
Rm a m1 x1  a m 2 x 2  "  a mn x n  em
m m n
f ( x) ¦ e2i
i 1
¦ ( Ri  ¦ aij x j ) 2 o min,
i 1 j 1
(1)

subject to:
x1  x 2  x3 1,
(2)
x1 t 0, x 2 t 0, x3 t 0! x n t 0 ,

where
Ri is measured satellite sensor response for a pixel in spectral band i
ai j is spectral response of mixture component, j, for spectral band i
x j is proportion of mixture component, j, for a pixel
ei is the error term for spectral band i

- 140 -
 1998 1999 0%

20%

40%
2000 2001

60%

2002 2003 80%

Figure 3 The saxaul percentage images over years 1998-2003 100%

Conclution and Discussion

In the derived fraction images, each pixel is associated with percentage values from 0 to 100 for Saxaul.
The study focused on applying LMM as a sub pixel classification. The saxaul forest is unique brush plant,
therefore the saxual forest should be taken into consideration to investigate and protect as a tool for
combating desertification. The linear mixing model based on the Optimization Method, which we have
concentrated on, may be applicable to many forest types, and leads to a particularly simple mathematical
description of the generation for signal of a given mixture. Application of the method to higher resolution
data to test its potential was hindered in this present study, by the lack of multi–temporal high resolution
data set.
Further work should test the robustness of the approach adopted here, when applied to large areas by
using multi-temporal data to detect saxaul forest changes resulting from desertification in Gobi. We would
expect mixtures of vegetation types to be present within a pixel, even with high-resolution data from
Landsat TM and the SPOT High Resolution Visible Imaging System(HRV).

VI. ACKNOWLEDGMENTS
The authors are also grateful to SPOT/VEGETATION for supplying data.
We thank our colleague Matthew Tyburski for comments and help in English.

1.1. References
1. Adams, J. B., and Smith, M. O, 1986, Spectral mixture modelling: a new analysis of rock and soil
types on the Viking Lander 1 Site. Journal of Geophysical Research, 91, 8098–8112
2. Adams,J.B.,and Adams,J.D.,1984, Geologic mapping using Landsat MSS and TM
images:removing vegetation by modeling spectral mixtures.Proceedings of the Third Thematic
Conference on Remote Sensing for Experimental Geology,Colorado
Springs,Colorado,(Michigan:ERIM), pp.615-622.
3. Bastin, L., 1997, Comparison of fuzzy c-means classification, linear mixture modelling and MLC
probabilities as tools for unmixing coarse pixels. International Journal of Remote Sensing, 18,
3629–3648.
4. Cross, A.M., Settle, J.J., Drake, N.A., and Paivinen, R.T.M., 1991, Subpixel measurement of
tropical forest cover using AVHRR data. International Journal of Remote Sensing, 6, 1159-1177.

- 141 -
5. Cross, A. M., Settle, J. J., Drake, N. A., and Paivinen, R. T. M., 1991, Subpixel measurement of
tropical forest cover using AVHRR data. International Journal of Remote Sensing, 12, 1119–1129.
6. DeFries, R. S., Hansen, M., Steininger, M., Dubayah, R., Sohlberg, R., and Townshend, J., 1997,
Subpixel forest cover in central Africa from multisensor, multitemporal data. Remote Sensing of
Environment, 60, 228–246.
7. DeFries, R. S., and Townshend, J. R. G., 1994, NDVI-derived land cover classification at a global
scale. International Journal of Remote Sensing, 15, 3567–3586.
8. Gillespie, A.R.,Smith,M.O.,Adams,J.B.,Willis,S.C.,Fischer,A.F.III,and Sabol, D.E., 1990,
Interpretation of residuals images: spectral mixture analysis of AVIRIS images, Owens Valley,
California. Proceedings of the Airborne Science Workshop: AVIRIS, JPL, Pasadena,CA, (JPL
Publication 90-54),pp.243-270.
9. Hlavka, C. A., and Spanner, M. A., 1995, Unmixing AVHR imagery to assess clearcuts and forest
regrowth in Oregon. IEEE Transactions on Geoscience and Remote Sensing, 33, 788–795.
10. Holben, B. N., and Shimabukuro, Y. E., 1993, Linear mixing model applied to coarse spatial
resolution data from multispectral satellite sensors. International Journal of Remote Sensing, 14,
2231–2240.
11. Quarmby, N.A., Townshend, J.R.G., Settle, J.J., White, K.H., Milnes, M., Hindle,T.L., and Silleos,
N., 1992, Linear mixture modeling applied to AVHRR data for crop estimation. International
Journal of Remote Sensing, 13, 415-425.
12. Shimabukuro, Y.E., and Smith, J.A., 1991, The least-squares mixing models to generate fraction
images derived from remote sensing multispectral data. I.E.E.E. Transactions on Geoscience and
Remote Sensing, GE-29, 16-20
13. Shimabukuro, Y. E., Holben, B. N., and Tucker, C. J., 1994, Fraction images derived from NOAA
AVHRR data for studying the deforestation of the Brazilian Amazon. International Journal of
Remote Sensing, 15, 517–520.
14. Smith,M.O.,Johnson, P.E., and Adams, J.B., 1985, Quantitative determination of mineral types and
abundances from reflectance spectra using principal component analysis. Journal of Geophysical
Research,90, 792-804.
15. Shimabukuro, Y.E., 1987, Shade images derived from linear mixing models of multispectral
measurements of forested areas. Ph.D. Dissertation, Colorado State University, Fort Collins,CO.
16. Tsolmon R, 2003 “ Methodology to Estimate Coverage and Biomass of Boreal Forests using
Satellite Data” Ph.D. Dissertation, Center Environmental Remote Sensing, Chiba University,
Japan
17. Vasiliev,O.V., 1996,Optimization Methods, 1 st edition (Atlanta: World Federation Publishers).

- 142 -
 Dust and Sandstorm Monitoring of Mongolia
Using NOAA AVHRR data
L.Ochirkhuyag1; R.Tsolmon2; S.Khudulmur3; J.Sumyasuren3; L.Natsagdorj4; D.Jugder4
1
The Wildlife Conservation Society Mongolia Program
Amar str-3, Internom bookstore building 3rd floor, Ulaanbaatar, Mongolia
olkhamjav@wcs.org; lochir@yahoo.com;
2
Laboratory of Remote sensing NUM-ITC-UNESCO, National University of Mongolia
Ikh surguuliin gudamj, Ulaanbaatar 210646a, Mongolia
tsolmon@num.edu.mn;
3
Information and Computer Center, National Remote Sensing Center
Juulchiny street-5, Ulaanbaatar 210646, Mongolia
mtt@magicnet.mn, sumya@yahoo.com
4
Institutes of Meteorology and Hydrology
Juulchiny street-5, Ulaanbaatar 210646, Mongolia
natsag03@yahoo.com; jugderd@yahoo.com

Abstract: Dust and sandstorms are a common meteorological phenomenon in the Gobi desert of
Mongolia like in the Taklamakan Desert of northwest China, the Sahara Desert of northern Africa
and other arid and semiarid regions. The brightness temperature channels 4, 5 of the NOAA
AVHRR data and meteorological station data were used for the dust and sandstorm mapping in the
Gobi area of Mongolia. The NOAA/AVHRR thermal infrared bands difference with combination
GIS layers were carried out for the mapping this study. The result shows that the dust and
sandstorm map can be achieved from the thermal bands of the satellite.

Introduction

Each year from March to May, it is observed that the dust and sandstorm, which occurred in the
Gobi desert of Mongolia and the Taklamakan Desert of northwest China, flies to the north pacific
archipelago and the west coast of America in addition to East Asia area. The dust and sandstorm study is
useful in the application of meteorological field. For example, the air turbidity increases with dust storm
occurrence and hence affects the radiation budget. It is also important to assess dust storm impact on
climate changes in Asia as well as the world.

Mongolia is landlocked country, has continental arid climate with four different seasons. When the
phenomenon of dust and sandstorms continue over several days in the Gobi Desert, Mongolian nomadic
herders refer to it as the “Ugalz”. Dust and sandstorms disrupt human life and economic activities and
result in soil erosion. Dust storms have some negative consequences; they delay and reduce pasture yields
for livestock, sands move and affect roads, settlements and villages, which can become enshroud with
sands (L.Natsagdorj and D.Jugder). One study showed that more than 70% of the pastureland area of
Mongolia is under desertification, 22.1% is strongly overgrazed, and sand movement covers the 7.9
million-hectare pastureland area (Jigjidsuren and Oyuntsetseg, 1998).

The distribution of the number of days with dust storms is shown in Fig. 1. The number of days
with dust storm is less than 5 days over the Khangai, Khuvusgul and Khentei mountainous areas of
Mongolia, 10-17 days over the area of Great Lakes, and 20-37 days over the desert and the semi-desert
areas in Mongolia.

- 143 -
 Fig.1. Number of days with dust storms observed in Mongolia.

The highest frequency of dust storms is over the three areas in the Gobi Desert in Mongolia, such as the
south side of the Altai Mountain and around Ulaan-nuur Lake and Zamiin-Uud. It should be noted that the
distribution of dust storms has coincided well with the distribution of strong wind (L.Natsagdorj, 1982;
D.Jugder, 1999) and soil conditions.

Previously, we mentioned how to obtain the number of dusty days. The distribution of dusty days is
shown in Fig. 2. We can see in the figure that the number of dusty days is less than 10 over the Khangai,
the Khentei and the Khuvusgul mountainous areas and 61 to 124 over the Great Lakes hollow and the Gobi
Desert area. It is 91 to 120 days over the Gobi on the south side of the Altai Mountain, and about 80 days
surrounding the Arts Bogd Mountain. The highest occurrence of dusty days is around the Mongol Els area.

Fig.2. Number of dusty days observed in Mongolia.

We attempt to monitor dust and sandstorm using NOAA satellite data. For this purpose, we used brightness
temperature differences in the thermal infrared bands in NOAA/AVHRR sensor data, the 2006 spring
season are observed and received at the NOAA data receiving station of National Remote Sensing Center
(NRSC).

Approach

The approach based on the split-window method. First we have done pre- processing for the raw
NOAA/AVHRR data and Level 3 data. The NOAA/AVHRR has five bands:
x Band 1, visible (VIS, 0.59-0.68mm)
x Band 2, near-infrared (NIR, 0.73-1.10mm)
x Band 3, mid-infrared (MIR, 3.55- 3.93mm)
x Bands 4 and 5, thermal-infrared (TIR, 10.3- 11.3mm and 11.5-12.5mm). (Fig.3)

- 144 -
 Fig.3. NOAA/AVHRR bands

Secondly, the geometric correction was applied using the PCI Geomatica software. Third the difference of
the thermal bands (brightness temperature) was calculated. This algorithm is based on the split-window
method. (Kinoshita, 1997)
Band _ difference = ch4-ch5 (1)

The last procedure was that the derived map (Fig.4) using equation 1 was overlied with GIS vectors with
meteorological station data. The dust and sandstorm is described by yellow and orange color (Fig.4).

Result and discussion

In spring, dust and sandstorm occurs frequently in the Gobi Desert, Southern Mongolia. From the given
approach we produced a dust and sandstorm map for the period March.6 to March.9, 2006. (Fig.4)
A GIS layer on dust and sandstorm (Fig.4) describes the weather component such as wind speed
and direction at the meteorological station. Mapping of the dust and sandstorm with GIS layers is effective
for dust and sandstorm monitoring.
The basic properties from thermal bands for NOAA/AVHRR were used for the detection of the
dust and sandstorm . Further quantitative studies using NOAA/AVHRR are important for the comparative
study with other satellites (MODIS, TOMS etc) and ground observation data such as Lidar and numerical
simulation based on the meteorological model.

a) March.6, 2006 b) March.7, 2006

c) March.8, 2006 d) March.9, 2006

Fig.4. Dust and sandstorm map of March.6-9, 2006

- 145 -
Reference

1. Jigjidsuren and S. Oyuntsetseg, 1998: Pastureland utilization problems and ecosystem. Ecological
sustainable development. Ulaanbaatar, No.2, p.206-212.
2. Naoko Iino and Kisei Kinoshita, Properties of AVHRR images Asia dust in April 1997. Satellite
Imagery of Asian Dust Events. Sep,2001, Kagoshima. p.23-28
3. L.Natsagdorj, 1982: Atmospheric circulation and dangerous weather phenomenon over the
territory of Mongolia. Publication of Hydro-Meteorological Research Institute of Mongolia, No. 6,
Ulaanbaatar, p. 300.
4. L.Natsagdorj and D. Jugder, 1992: Statistics method for prediction of dust storm over the Gobi
and steppe area in Mongolia in spring. Scientific report, Ulaanbaatar, p. 83.
5. L. Natsagdorj and D. Jugder, 1992: Dust storm in the Mongolian Gobi. Proceeding of the
Symposium on Global Change and the Gobi Desert, Ulaanbaatar, p25-40.
6. L.Natsagdorj and D. Jugder, 1993: Dust Storms in Gobian Zone of Mongolia. The First PRC-
Mongolia Workshop on Climate Change in Arid and Semi-arid Region over the Central Asia, May
8-11, 1993, p99-104, Beijing.
7. J.Sumyasuren, Large Area Characterization of Dust Aerosols Properties usiong Remote Sensing
Satellite Data. Post-graduate diploam thesis, Apr, 2005, Ahmedabad

- 146 -