Вы находитесь на странице: 1из 12

Physics and Chemistry of the Earth 35 (2010) 730741

Contents lists available at ScienceDirect

Physics and Chemistry of the Earth


journal homepage: www.elsevier.com/locate/pce

Integrated assessment of land use and cover changes in the Malagarasi


river catchment in Tanzania
J.J. Kashaigili *, A.M. Majaliwa
Department of Forest Mensuration and Management, Sokoine University of Agriculture, P.O. Box 3013, Chuo Kikuu, Morogoro, Tanzania

a r t i c l e

i n f o

Article history:
Received 4 February 2010
Received in revised form 12 July 2010
Accepted 24 July 2010
Available online 6 August 2010
Keywords:
Land use land cover changes
Malagarasi river catchment
Muyovozi wetland
Perceptions
Remote sensing

a b s t r a c t
Malagarasi river catchment represents one of the largest and most signicant transboundary natural ecosystems in Africa. The catchment constitutes about one third of the catchment area of Lake Tanganyika
and contains ecosystems of both national and international importance (i.e. Muyovozi Wetland Ramsar
site). It has been increasingly said that increased anthropogenic activities have had negative impacts on
the Muyovozi wetland in particular and other catchment resources. Nevertheless, these beliefs are little
supported by quantitative data. A study on the dynamics of land use and cover in the Malagarasi river
catchment therefore investigated long-term and seasonal changes that have occurred as a result of
human activities in the area for the periods between 1984 and 2001. Landsat TM and ETM+ images were
used to locate and quantify the changes. Perceptions of local people on historical changes and drivers for
the changes were also collected and integrated in the assessment. The study revealed a signicant change
in land use and cover within a period of 18 year. Between 1984 and 2001, the woodland and wetland vegetation covers declined by 0.09% and 2.51% per year. Areas with settlements and cultivation increased by
1.05% annually while bushed grassland increased at 1.93% annually. The perceived principal drivers for
the changes were found to include re, cultivation along rivers and lake shores, overgrazing, poor law
enforcement, insufcient knowledge on environmental issues, increasing poverty, deforestation and population growth. The human population growth rate stands at 4.8% against a national gure of 2.9%. The
most perceived environmental problems include drying of streams and rivers, change in rainfall, loss of
soil fertility, soil erosion and reduced crop yield. The study concludes that, there has been signicant
changes in land use and cover in the catchment and these require concerted actions to reverse the
changes. The study highlights the importance of integrating remote sensing and local knowledge in
understanding the dynamics catchment resources and generating information that could be used to overcome the catchment management problems.
2010 Elsevier Ltd. All rights reserved.

1. Introduction
Human activities in many parts of the world have greatly changed the natural land cover. Large tracts of natural vegetation cover
have been converted into croplands or deserts, and natural wetlands
have been drained and lled in order to feed and shelter expanding
population (WCED, 1987; Monela and Solberg, 1998). The concerns
about land use and land cover change globally emerged due to realization that changes of the land surface inuences climate and impact on ecosystem goods and services (Lambin et al., 2003). It is
however important to realize the differences between the two terminologies. The terms land use and land cover are not synonymous
and literature draws attention to their proper use in studies. According to Di Gregorio and Jansen (2000), land cover describes the physical states of the land surface including cropland, forest, wetlands,
* Corresponding author. Tel.: +255 754207117; fax: +255 232604648.
E-mail address: jkashaigili@yahoo.co.uk (J.J. Kashaigili).
1474-7065/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.pce.2010.07.030

pastures roads and urban areas, whereas land use relates to the manner in which these biophysical assets are used by humans (Cihlar
and Jansen, 2001). Since use depends largely on the land characteristics, there is a close relationship between land cover and land use.
However, land cover observation does not automatically mean land
use, because land cover and land use, though interrelated, are not
identical. The land use choices made will vary in space and time
and so will the resulting land cover (Cihlar and Jansen, 2001). One
of the important land use changes is that the worlds forest, grasslands and woodlands have declined and cropped land areas have expanded (Skole and Tucker, 1993; Slayback, 2003).
To understand the magnitude, and pattern of change, carrying
out land use and land cover change analysis using remotely sensed
data is imperative. According to Lambin (1997), land use and land
cover change analysis is an important tool to assess global change
at various spatialtemporal scales. Lopez et al. (2001) afrmed that
it reects the dimension of human activities on a given environment.
According to Zhou et al. (2008) land cover change often reects the

J.J. Kashaigili, A.M. Majaliwa / Physics and Chemistry of the Earth 35 (2010) 730741

most signicant impact on the environment due to human activities


or natural forces and that remote sensing can be an appropriate tool
for getting wide impression on land cover change. It is now widely
accepted that information generated from remotely sensed data is
useful for planning, and decision making. For example, Dewan and
Yamaguchi (2009) quantied the patterns of land use and land cover
change for the last 45 year for Dhaka Metropolitan that formed valuable resources for urban planners and decision makers to devise sustainable land use and environmental planning. According to
Kennedy et al. (2009), for the resource manager, a particular attraction of satellite remote sensing technology is the ability to provide
consistent measurements of landscape condition, allowing detection of both abrupt changes and slow trends over time. Detection
and characterization of change in key resource attributes allows resource managers to monitor landscape dynamics over large areas,
including those areas where access is difcult or hazardous, and
facilitates extrapolation of expensive ground measurements or strategic deployment of more expensive resources for monitoring or
management (Li et al., 2003; Schuck et al., 2003). In addition, longterm change detection results can provide insight into the stressors
and drivers of change, potentially allowing for management strategies targeted toward cause rather than simply the symptoms of the
cause (Kennedy et al., 2009). According to MacLeod and Congalton
(1998), change detection on land cover focuses mainly on four aspects, namely; (a) detecting if a change has occurred, (b) identifying
the nature of the change, (c) measuring the areal extent of the
change, and (d) assessing the spatial pattern of the change. With
the growing use of remote sensing, some studies have embarked
on assessing and improving the methods for change detection (e.g.
Fraser et al., 2009, 2005), while others are looking at the accuracy
(e.g. Stehman et al., 2009; Foody, 2002; Zhang and Foody, 2009).
The assessment of spatial patterns of land cover changes over a
long period using images of multi-temporal coverage is now possible considering the accumulation of remotely sensed images over
the past decades; as such making it possible to generate an understanding of the drivers for the changes. Tanzanias ecosystems have
been affected by agriculture, pasture, logging, charcoal making and
mining (Kidegesho, 2001; Ogungo and Njuguna, 2004). Like many
other developing countries, most of the population in Tanzania
lives in rural areas and depends directly on the land for their livelihoods. This rural population is causing resource degradation
brought about by the decrease in the area under natural vegetation
and its conversion into other types of land use and land cover that
are human-managed systems.
This paper presents an assessment of land use and land cover
changes in the Malagarasi river catchment in Tanzania. Malagarasi
river catchment represents one of the largest and most signicant
transboundary natural ecosystems in Africa. The catchment constitutes about one third of the catchment area of Lake Tanganyika and
contains ecosystems of both national and international importance
(i.e. Muyovozi Wetland Ramsar site) and game reserves (Moyowosi/
Kigozi and Ugalla Game Reserves). It has been increasingly said that
increased anthropogenic activities have had negative impacts on the
Muyovozi wetland in particular and other catchment resources.
Nevertheless, these beliefs are little supported by quantitative data.
Therefore this study investigated long-term and seasonal changes
that have occurred as a result of human activities in the area for
the periods between 1984 and 2002.

731

ern monsoon winds as well as the southeast and northeast trade


winds. The area receives bimodal rains, namely November to
December (short rains) and April to May (main rainy season).
The average annual rainfall is about 8001000 mm/year. Temperature ranges from 12 C to 20 C in July/August and to as high as
32 C to 35 C in August and September (main dry season). Altitude
ranges between 800 and 1600 m above mean sea level with an
average of 1200 m asl.
A large part of the Malagarasi Catchment is occupied by the
wetland (MalagarasiMuyovozi Ramsar site) and game reserves
(Moyowosi/Kigozi and Ugalla Game Reserves). The major land uses
are nature and game reserves conservation, recreational and tourist hunting including other licensed activities that are shing and
honey gathering. Other land uses include cultivation for subsistence (food crops-primarily cassava) and tobacco growing (for
commercial purpose). In the surrounding areas (general lands),
the principal land use are settlement, subsistence cultivation,
shing and seasonal grazing (Jones and Hill, 1994; NEMC/IUCN,
1994; Lyaluu, 1996). Woodlands and grasslands are the dominant
land cover within the catchment.
Kigoma District has one of the fastest growing population in
Tanzania. According to the 2002 census, the population of Kigoma
District was 490,816 of which 225,002 are males. The average
household size is 6.8 persons against the national average of 4.9
persons. The annual average population growth for the period of
19982002 was 4.1% against the national average of 2.9%. This high
growth has been attributed to the inux of refugees from DRC,
Burundi and Rwanda (URT, 2003).
2.2. Methods
The study applied contemporary techniques namely remote
sensing and GIS to assess the land use and land cover changes in
the study area. Furthermore, interviews using a checklist, semistructured questionnaire and focus group discussion were conducted to get perceptions about the changes in the catchment.
2.3. Socio-economic data collection and analysis

2. Materials and methods

Socio-economic data were collected from four purposefully selected villages. These include Malagarasi, Kasisi, Ilalangulu and
Mtegowanoti. The selection of households for interviews was
based on a simple random sampling technique. A total of 120
household (i.e. 30 households from each village) were interviewed.
A semi-structured questionnaire with both open-ended and closeended questions was used to elicit information from the community. In addition, guiding questions were asked in focus group discussion and key informants interviews to capture in-depth
understanding of historical resources use pattern in the area. Visual observations through transect walk were made along selected
routes to ground truthing preliminary assumptions and maps.
The transect walks (1 km long and 500 m apart) were used to identify the various human activities and different land use/cover at
specic sample points. The waypoints were marked using a global
positioning system (GPS) and used in map verication exercise, in
which classes in imagery base map were correlated with actual
ground data. Descriptive statistics were summarized and tabulation was employed to report all quantitative information. Frequencies and percentage were calculated to facilitate the drawing up of
inferences related to socio-economic ndings.

2.1. Study areas

2.4. Remotely sensed data, processing and change detection

The Malagarasi river catchment lies within longitudes 315S


610S and latitudes 3040E3230E (Fig. 1). The climate of the area
is tropical in nature characterized by the interaction of the south-

2.4.1. Image selection and acquisition


The images used in the study are summarized in Table 1. The
target was images acquired during the dry season (JulyOctober)

732

J.J. Kashaigili, A.M. Majaliwa / Physics and Chemistry of the Earth 35 (2010) 730741

Fig. 1. A map of study area. The inset map (upper left) shows the location of Malagarasi river catchment in relation to Tanzania.

with minimum cloud cover. Nevertheless, the required images for


the entire catchment were not readily available. As a consequence,
three images acquired during the transition period from wet season
to dry season were used. As such, these might have implications of
the change detection due to seasonal effect. Nevertheless, detailed
training of the site had to be done and GIS tools such as area of interest (AOI) were applied using visual analysis, reference data and local
knowledge to minimize the seasonality effect.
2.4.2. Image pre-processing
The methods for the images analysis combined both visual and
digital image processing. The processing involved image rectication/georeferencing and co-registration and image enhancement.

Prior to image processing, images layers/bands were imported


and layer stacked to full scene. All image processing and subsequent image analysis were carried out using ERDAS Imagine Software Version 9.2.
2.4.3. Image rectication
Image rectication was carried to correct for distortions or degradation resulting from the image acquisition process. To ensure
accurate identication of temporal changes and geometric compatibility with other sources of information, the image were coded to
the co-ordinate and mapping system of the national topographic
maps, i.e. UTM co-ordinate zone 36 South, Spheroid Clarke 1880,
Datum Arc 1960, based on a previous georeferenced Landsat image

J.J. Kashaigili, A.M. Majaliwa / Physics and Chemistry of the Earth 35 (2010) 730741
Table 1
Remotely sensed data used in the analysis of land use/cover change.
Image
Landsat
Landsat
Landsat
Landsat
Landsat
Landsat
Landsat
Landsat
Landsat
Landsat

TM
ETM+
TM
ETM+
TM
ETM+
TM
ETM+
TM
ETM+

Path/row

Acquisition date

Season

171/63
171/63
171/64
171/64
172/63
172/63
172/64
172/64
170/64
170/64

13th June 1984


16th May 2000
13th June 1984
22nd May 2002
12th July 1986
01st October 2001
12th July 1986
01st October 2001
31st August 1986
18th January 2000

Dry
Wet
Dry
Wet
Dry
Dry
Dry
Dry
Dry
Wet

Note: TM = thematic mapper; ETM+ = enhanced thematic mapper plus.

of 3rd August and 4th September 1994. Since the available satellite
images had been already corrected for radiometric distortions and
had no apparent noise, the created sub-scene was only subjected
to geometric correction. The geometric correction allows to compensate for various distortion introduced by several factors including earth rotation effects, panoramic distortion (with eld of view
of some sensor), curvature of the earth, atmospheric refraction, relief
displacement, variations in platform altitude, attitude and velocity
and panoramic effects related to the imaging geometry (Lillesand
and Kiefer, 1987; Richards, 1993). At least 60 well distributed
ground control points (GCPs) were used in the rectication process.
The root mean square error (RMSE) varied from 0.3 to 0.5 pixels. A
rst order polynomial t was applied and all the data were resampled to a spatial resolution of 30 m using nearest neighbour method.
2.4.4. Image enhancement
Enhancement usually reinforces the visual interpretation of the
images (Kashaigili, 2006). In order to reinforce visual interpretability of images, a colour composite (Landsat TM bands 4, 5 and 3)
was prepared and its contrast was stretched using Gaussian distribution function. The 3  3 high pass lters was applied to the colour composite to further enhance visual interpretation of linear
features, e.g. rivers and vegetation features.
2.4.5. Ground truthing
Ground truthing was done in order to verify and modify land
covers described in the preliminary image interpretation. GPS
was used to locate sampled land cover observations while digital
camera recorded photos on physical features about the areas. All
sampled GPS points were booked as way points on a booking paper
and photograph numbered. Key informants were also involved to
give some information on land cover and land use particularly
for the past years. The exercise was done during the dry season
to enable access to all the areas which is not possible during the
wet season.
2.4.6. Image classication
Supervised classication, using Maximum Likelihood Classier
(MLC), was utilized. Supervised classication process involved
selection of training sites on the image, which represent specic
land classes to be mapped. Training sites are sites of pixels that
represent specic land classes to be mapped (ERDAS, 1999). They
are pixels that represent what is recognized as a discernable pattern, or potential land cover class. The training sites were generated by on-screen digitizing of selected areas for each land cover
class identied on colour composite. Training was an iterative process, whereby the selected training pixels were evaluated by performing an estimated classication (ALARM command). The
image alarm performs a quick pre-classication of the image
data and indicates where potential confusion among classes may
occur. Basically it is a visual tool that gives an overview of where

733

the classes will be assigned in the image and whether additional


classes are required. Based on the inspection of results, training
samples were rened until a satisfactory result was obtained.
The objective was to produce thematic classes that resemble or
can be related to actual land cover types on the earths surface.
The advantage of digital image classication is that it can provide
efcient, consistent and repeatable routines for mapping large
areas (Kashaigili, 2006). Visual interpretation involved the use of
image characteristics such as texture, pattern and colour to translate image into land covers. The enhanced image colour composite
was used in this operation.
2.4.7. Post-processing of classied images
Classied images were recoded to respective classes (i.e. wetland vegetation, forest, water, bareland, woodland, bushland,
grassland, settlement, cloud and cloud shadow). Following the
recoding, images were ltered using a 3  3 majority-neighbourhood lter. The classied images were ltered in order to eliminate
patches smaller than a specied value and replace them with the
value that is most common among the neighbouring pixels. A mosaic operation was performed to multiple classied images to produce one map for the entire study area (Fig. 2). The image
mosaicking involved the joining of geo-referenced images together. The input images contained the same map and projection
information with the same number of layers. After mosaicking,
sub-setting was performed in order to breaking out a portion of a
large image le into one or more smaller les. Often, image les
contain areas much larger than a particular study area. In these
cases, it is helpful to reduce the size of the image le to include
only the area of interest (AOI). This not only eliminates the extraneous data in the le, but it speeds up processing due to the smaller amount of data to process (ERDAS, 1999).
2.5. Change detection
Change detection is a very common and powerful application of
satellite based remote sensing. Change detection analysis entails
ndings the type, amount and location of land use changes that
are taking place (Yeh et al., 1996). Various algorithms are available
for change detection analysis and they can be grouped into two
categories namely (a) pixel-to-pixel comparison of multi-temporal
images before image classication, and (b) post-classication comparison (Jensen, 1996). In this study, a post-classication comparison method was used to asses land use and cover changes. It is the
most common approach (Jensen, 1996; Mundia and Aniya, 2006)
for comparing data from different sources and dates. The advantage of post-classication comparison is that it bypasses the difculties associated with the analysis of images acquired at
different times of the year and/or by different sensors (Yuan
et al., 2005; Coppin et al., 2004; Alphan, 2003). The method has
been found to be the most suitable for detecting land cover
changes (Wickware and Howarth, 1981); as this enables estimation of the amount, location, and nature of change. The only pitfall
is that the accuracy of the change maps depends on the accuracy of
individual classications and subject to error propagation (Yuan
et al., 2005; Zhang et al., 2002). The approach identies changes
by comparing independently classied multi-date images on
pixel-by-pixel basis using a change detection matrix (Yuan and
Elvidge, 1998). The matrix produces a thematic layer that contains
a separate class for every coincidence of classes in multi-date
dataset.
2.6. Assessment of the rate of cover change
The estimation for the rate of change for the different covers
was computed based on the following formula:

734

J.J. Kashaigili, A.M. Majaliwa / Physics and Chemistry of the Earth 35 (2010) 730741

Fig. 2. A mosaic map of different image scenes used in land cover change.

Fig. 3. Land use and land cover map 1984/1986.

J.J. Kashaigili, A.M. Majaliwa / Physics and Chemistry of the Earth 35 (2010) 730741

% Cover change

Areai

year x

Pn

 Areai

year x1

i1 Areai year x

 100%

735

3. Results and discussion


3.1. Land cover maps

Annual rate of change

Areai

% Annual rate of change

year x

 Areai
tyears

year x1

Areai year x  Areai year


Areai year x  t years

 100%

x1

 100%
3

where Areaiyearx = area of cover i at the rst date, Areaiyearx+1 = area


P
of cover i at the second date, ni1 Areaiyearx = total cover area at the
rst and tyears = period in years between the rst and second scene
acquisition dates.

The land cover maps for 1984/1986 and 2000/2002 are presented in Figs. 3 and 4 respectively. Generally, the maps show
the variation in cover coverage between the two periods under
consideration.
3.2. Change detection results
Table 2 presents the coverage of each land cover/use class in the
1980s and 2000s including the area and percentage area change
between the two periods for the Malagarasi river catchment while

Fig. 4. Land use and land cover map 2000/2002.

736

J.J. Kashaigili, A.M. Majaliwa / Physics and Chemistry of the Earth 35 (2010) 730741

Table 2
Cover area, changed area and the rate of change between 1984/1986 and 2000/2002.
Cover class

1984/1986
Cover
area (ha)

2000/2002
% Cover
coverage

Cover
area (ha)

Change area (ha)

% Change

Annual rate of
change (ha/year)

% Annual rate of
change (%/year)

30636.1
65647.4
28924.4
293390.9
487257.5
164403.8
81035.6
651988.6
25727.2
64931.8

+0.4
0.8
+0.4
+3.6
+6.0
2.0
+1.0
8.1
+0.32
0.80

1702.0
3647.1
1606.9
16299.5
27069.9
9133.5
4502.0
36221.6

3.2
0.1
3.0
1.9
1.9
2.5
1.1
4.6

% Cover
coverage

FO
WD
W
BS
GR
WET
SET + C
BR
CL
CLS

52535.7
4054761.7
54548.0
851002.5
1402296.7
363505.1
429120.0
789515.0
14099.8
81917.1

0.7
50.1
0.7
10.5
17.3
4.5
5.3
9.8
0.2
1.0

83171.8
3989114.2
83472.3
1144393.3
1889554.2
199101.3
510155.5
137526.4
39827.0
16985.3

1.0
49.3
1.0
14.1
23.4
2.5
6.3
1.7
0.5
0.2

Total

8093301.0

100.0

8093301.0

100.0

Note: FO = forest; WD = woodland; W = water; BS = bushland; GR = grassland; WET = wetland vegetation/marsh; SET + C = settlement/cultivation; BR = bareland/burnscars;
CL = clouds; CLS = cloud shadow.

Figs. 57 present the change in woodland, forest and wetland vegetation covers respectively. Considering the subset area of
80,93,301 ha, the results (Table 2) indicate that in the year 1984/
1986 the woodland covered 50.1% of the area followed by grassland 17.3%, bushland 10.5%, bareland 9.8%, settlement/cultivation
5.3%, wetland/marsh 4.5%, water 0.7% and forest 0.6% while the
remaining are areas appeared covered by clouds and cloud shadow. The area under forest which occupied 52535.7 ha (0.6%) in
1984/1986, increased to 83171.8 ha (1.0%) in 2000/2002, indicating an increase in forested area of about 0.4%. Bushland increased
from 10.5% to 14.1%, while grassland increased from 17.3% to 23.4%
and settlement/cultivation increased from 5.3% to 6.3% indicating
an increase of +3.6%, +6.1% and +1.0% respectively between the
two periods under consideration. The wetland/marsh area declined
by 164403.8 ha (2.0% of the subset area), while the woodland
area declined by 65647.4 ha (0.8%) between the two periods.
The total area of water which occupied 54547.95 ha in 1980s, increased to 83472.3 ha (+0.4%) in 2000s, indicating an increase of
about 28924.4 ha.
As revealed from Table 2, the forest cover increased at a rate of
+1702.0 ha/year (+3.2%/year) over an average period of 18 year
(i.e.1984/1986 and 2000/2002) assuming a linear increase. The
woodlands and wetland vegetation/marsh decreased consistently
at a rate of 3647.1 ha/year (0.1%/year) and 9133.5 ha/year
(2.5%/year) respectively. It is possible that the increase in forest
cover is attributable to reduced forest disturbance following the
designation of MalagarasiMuyovozi Ramsar site in 2000 including
regeneration (Fig. 6). Bushland and grassland cover increased at a
rate of 16299.5 ha/year (+1.9%/year) and 27069.9 ha/year (+1.9%/
year) respectively. This rapid increase might be due to clear felling
of trees mainly woodlands for rewood, and increased settlement
and agricultural activities (tobacco farms and subsistence farming).
This has also been accentuated by local people during the interviews and about 55% of the respondent reported cutting trees. Fire
burning has been a serious problem in recent years. It is clear from
Table 2 that the woodland area decreased consistently over the
18 year while the settlement and cultivation increased at a rate
of +4502 ha/year (1.1%/year). The expansion in settlement and cultivated areas reects on the land use transformation in the
catchment.
3.3. Land covers transformation/ow
Table 3 presents the land covers transition matrix between land
cover classes in 1984/1986 and 2000/2002. The numbers in brackets indicates the cover area which remained unchanged between
1984/1986 and 2000/2002, while others indicate the ow of covers

or covers that changed to another cover category. It is important to


note that all land cover categories changed but with varying
magnitudes. For example, 272514.5 ha of woodland was transformed to bushland, 844317.1 ha to grassland, 147483.1 ha to
settlement and cultivation, 36154.7 ha to bareland/burnt area,
while 7905.0 ha and 47427.9 ha appeared to be in water and wetland respectively, and 10949.1 ha were covered by clouds while
5263.5 ha were under cloud shadow.
Table 4 presents the detected changes in selected cover for the
period of 1984/1986 and 2000/2002 deduced from the change
detection matrix. The arrow in each row indicates a conversion
from. . .to. As presented in Table 4; 52579.2 ha (1.3%) of total area
under woodland in 1984/1986 changed to forest in 2000/2002,
1116831.6 ha (27.5%) changed to bushland and grassland,
47427.9 ha (1.2%) changed to wetland vegetation, 147483.1 ha
(3.6%) to settlement and cultivation; and 36154.71 ha (0.7%) changed to bareland/burnscars. There has also been signicant change
of 155799.4 ha (42.9%) from wetland vegetation/swamp to bushland and grassland, implying encroachment of wetland vegetation
by other swamp vegetations species and 5073.4 ha (1.4%) to settlement/cultivation. The increased area under dry season farming
along the wetlands, river and lakes banks which has also been revealed during interviews and conrmed during ground truthing.
Table 5 presents a summary on changed and unchanged cover
areas between 1984/1986 and 2000/2002. The percentage changed
indicates the percentage area of a particular cover which changed
to other covers while the percentage unchanged represents the
percentage area of the original area of a particular cover which remained unchanged for a given period.
From Table 5, the forest cover changed to other forms by 66.3%,
while bushland changed by 69.1%, grassland by 75.7%, settlement
and cultivation by 56.2%. The woodland and wetland vegetation
changed by 35.1% and 79.7% respectively between the periods under consideration. Nevertheless, some cover areas remained unchanged between 1984/1986 and 2000/2002 (Table 5).
3.4. Variations on detected changes, interpretations and limitations
Variations on results from change detection analysis are inevitable and these could impair the interpretability for the detected
changes (Kashaigili, 2006). In this study some variations on the detected changes have been noted. By scrutinizing the change detection matrix (Table 3), one identies that some of the changes were
unrealistic (e.g. a change from forest to wetland vegetation cover
and change from water to woodland and forest cover). It is highly
acknowledged that ecosystem dynamics response is not linear and
depends on many factors but most arguably the variation in rain-

J.J. Kashaigili, A.M. Majaliwa / Physics and Chemistry of the Earth 35 (2010) 730741

737

Fig. 5. Land cover change for woodland between 1984/1986 and 2000/2002.

fall patter and distribution (Kashaigili et al., 2006). Trend analysis


of rainfall in the Malagarasi Catchment revealed that there was
no signicant increase in rainfall amount between 1984/1986
and 2000/2002 (Majaliwa, 2009). It is possible that the variations
are due to plant phenological effects and spectral resolutions.
The different plant phenological effects are related to which season
an image is acquired on the ground. Studies have shown that dry
season is the most desirable period for image change analysis. As
noted by Burns and Joyce (1981) selecting the driest period of
the year for change analysis will enhance spectral similarity due
to excessive wetness prevailing during other periods of the year.
The wet season spectral separability, which is responsible for class
assignment, becomes somewhat difcult and may result in misclassication. This is likely to be the source of variation in detected
changes as images used for this study were obtained in both wet

and dry season (Table 1). The ground truthing eldwork was done
at the peak of the dry season while some of the available and used
images were acquired in the wet season. In any circumstance, such
seasonal differences could affect their use during class verication.
3.5. Perceptions on the causes for the changes, and problems resulting
from changes in land use and land cover in Malagarasi river catchment
Peoples perception on environmental change has been very
important in setting a clear view of what the stakeholders perceive
on utilization of natural resources (Rohr, 2002). Table 6 presents
the local peoples perception on various identied causes for the
land use and land cover changes, while Table 7 presents the perceptions on problems as a result of land use and cover changes
in the Malagarasi river catchment. Majority of respondents indi-

738

J.J. Kashaigili, A.M. Majaliwa / Physics and Chemistry of the Earth 35 (2010) 730741

Fig. 6. Land cover change for forest between 1984/1986 and 2000/2002.

cated re (99%) and deforestation (96%) to be the main causes for


the land cover changes in the area. The main factors mentioned as
contributing to re were beekeeping and hunting activities, while
for deforestation are commercial logging, charcoals production,
population growth, expansion of commercial farming and food
crops production. Other identied causes for the changes include
farming along rivers and lake shores, overgrazing, poor law
enforcement, insufcient knowledge on environmental issues and
increasing poverty. The majority local people are poor with very
low income. The immediate source of livelihood available in the
catchment is based on direct exploitation of the catchment resources in forests, sheries or most importantly land resources
for agricultural production. Majaliwa (2009) revealed a very low
rate of literacy in the study area, with majority not having attended
school. According to Mahinya (2005) education promotes better
management of household resources and reduces pressure on easily accessible natural resources like water, grazing land, forest cover in the catchment, and is an ingredient for sustainable natural
resources management. It is among the factors that inuence an
individual perception for decision making. Therefore, increased
resource extraction is impacting on the resources and increasing
deforestation in the area.
In many areas of the catchment, particularly along the Malagarasi river banks and the lake shores, riparian areas were reported to
be highly degraded as a result of excessive livestock numbers.

Although there was no reliable data on carrying capacity and numbers of livestock in the area, the majority of respondents were in
agreement that the livestock numbers have increased, and overgrazing may become a serious problem in the future. A study by
Nkotagu and Simon (2004) indicated livestock grazing to be uncontrolled and among the overwhelming problem creating conict
(land and water) between farmers and pastoralist; and in addition
threatening the catchment ecosystem. The increase in number of
livestock in the study area was found to be attributable to increased droughts in neighbouring areas such Shinyanga, Upa
and Urambo during 2005/2006 that caused the agro-pastoralists
to migrate into the area. Studies by Yanda and Shishira (1999)
and DANIDA (1999), observed that the lower catchment of
Malagarasi has been under severe overgrazing pressure over time.
Increased grazing pressure along riverbanks and lake shores prevents natural regeneration of cover. Livestock affect both the soil
structure and the land cover of herbaceous plants. Removal of land
cover exposes soil to erosion (wind and water), which when combined with soil disturbances by livestock, speed the erosive processes leading to reduction in inltration and increased runoff
(Lal, 2001).
Considering the increased pressure on the catchment resources,
numerous problems are now evident (Table 7). Some of the environmental problems identied include drying of streams and rivers, loss of soil fertility, soil erosion and reduced crop yield. As

J.J. Kashaigili, A.M. Majaliwa / Physics and Chemistry of the Earth 35 (2010) 730741

739

Fig. 7. Land cover change for wetland from 1984/1986 to 2000/2002.

regard to rainfall, respondents pointed out that rainy seasons in the


pre-1970s were consistent and dependable, whereas today, it is no
longer easy to tell when the rainy season will start. Past studies in
the catchment has shown that rainfall regime in the area has an
impact on both the size and depth of the lakes/river (Nkotagu
and Simon, 2004). For water ows, it was stressed that the river
ows has been much reduced in recent years when compared to
the past 20 year. About 63% of the respondents reported reduced
ow as an increasing problem and requested for strong measures

and immediate actions by the government to save the river and


the lakes (Majaliwa, 2009). Evidence from elderly respondents
memories revealed that in the past, water in the rivers and lakes
used to ood and inundate some parts of the villages. In contrast,
today the river do not ood, except during the wettest month of
the year, mostly in April. Previous studies over the area by DANIDA
(1999) revealed that competing and unsustainable uses of landbased resources in the upper catchment areas have affected downstream water ow into the Malagarasi river catchment.

740

J.J. Kashaigili, A.M. Majaliwa / Physics and Chemistry of the Earth 35 (2010) 730741

Table 3
Changes detection matrix in different land use/cover coverage (ha) between 1984/1986 and 2000/2002.
Cover in
1984/1986 (ha)

Cover in 2000/2002 (ha)


FO

WD

BS

GR

WET

SET + C

BR

CL

CLS

Total

FO
WD
W
BS
GR
WET
SET + C
BR
CL
CLS
Total

(17695.4)
52579.2
76.5
1907.2
3203.6
5084.4
89.0
1162.3
83.1
1291.2
83171.8

19625.8
(2630167.7)
882.2
248280.4
671781.3
81557.5
85047.7
222482.7
3428.1
25861.1
3989114.2

186.7
7905.0
(34428.9)
6267.0
10820.8
18583.6
103.2
4677.2
26.7
473.5
83472.3

4788.8
272514.5
1346.6
(263214.3)
227721.1
107697.8
38344.9
195174.5
5576.0
28014.8
1144393.3

5926.6
844317.1
721.3
242252.3
(340913.1)
48081.6
114769.1
272263.3
3600.8
16709.0
1889554.2

2656.9
47427.9
16252.2
23902.3
17678.7
(73726.0)
1004.9
14391.9
128.3
1932.3
199101.3

904.1
147483.1
109.3
17177.7
101161.8
5073.4
(187852.2)
44250.4
908.2
5235.4
510155.5

588.0
36154.7
264.9
33153.2
24051.6
10144.0
1086.5
(29642.7)
333.9
2106.9
137526.4

54.8
10949.1
165.0
11388.2
3502.2
9172.6
521.1
4031.6
(8.1)
34.3
39827.0

108.8
5263.5
301.2
3460.0
1462.5
4384.4
301.3
1438.4
6.5
(258.8)
16985.3

52535.7
4054761.7
54548.0
851002.5
1402296.7
363505.1
429120.0
789515.0
14099.8
81917.1
8093301.0

Note: FO = forest; WD = woodland; W = water; BS = bushland; GR = grassland; WET = wetland vegetation/marsh; SET + C = settlement/cultivation; BR = bareland/burnscars;
CL = clouds; CLS = cloud shadow; numbers in brackets indicate cover areas that remained unchanged between the two periods.

Table 4
Detected changes in percentage for some selected cover for the period 1984/1986
2000/2002.
Change (fromto)

1984/19862000/2002
area (ha)

% of the cover

FO ? WET
FO ? SET + C
FO ? BR
WD ? FO
WD ? BR + GR
WD ? WET
WD ? SET + C
WD ? BR
WET ? BS + GR
WET ? SET + C

2656.9
904.1
588.0
52579.2
1116831.6
47427.9
147483.1
36154.7
155779.4
5073.4

5.1
1.7
1.1
1.3
27.5
1.2
3.6
0.7
42.9
1.4

Table 5
Percentage changes of individual cover between 1984/1986 and 2000/2002.
Cover type

Unchanged

Changed

% Cover
unchanged

% Cover
changed

Forest
Woodland
Water
Bushland
Grassland
Wetland vegetation/
marsh
Settlement/cultivation
Bare land/burnscar

17695.4
2630167.7
34428.9
263214.3
340913.1
73726.0

34840.5
1424594.1
20119.2
587788.3
1061383.6
289779.3

33.7
64.9
63.1
30.9
24.3
20.3

66.3
35.1
36.9
69.1
75.7
79.7

187852.2
29642.7

241267.7
759872.3

43.8
3.8

56.2
96.2

Percentage response (%) (n = 120)


Very
serious

Serious

Moderate

Not
serious

24
24

45
26

23
25

8
25

61
91
8
91

20
8
29
8

15
1
18
1

4
0
45
0

(URT, 2003; Nkotagu and Simon, 2004). Furthermore; the presence


of refugee camps hosting about 280,000 refugees at various locations around the boundary of the site has led to intensied cover
changes surrounding the catchment (Majaliwa, 2009). The majority of refugees have recently repatriated and some naturalised.
Although the exact number of repatriated refugees is not known,
the residual effect on the environment as a result of tree cutting
for rewood and housing is still noticeable as revealed on the eld.
A study by (Mungongo, 1995) revealed that a household size is an
important variable in determining sustainability of natural resources in an ecosystem. Similarly, ndings from baseline studies
conducted in MalagarasiMuyovozi Ramsar site by IRA (2002)
pointed out that there is a strong relationship between household
size and environmental degradation in the area. This is due to the
fact that large household tend to over-exploit their resources in order to meet their needs while so doing undermine their source of
livelihood.
4. Conclusion

Table 6
Local peoples perceptions on the causes for the land use and cover.

Deforestation
Cultivation along river/lakes
Overgrazing
Fire
Poverty
Poor law enforcement
Insufcient knowledge
on environmental issues

Type of problems

Availability of rewood
Drying of streams and
rivers
Changes in rainfall
Loss of soil fertility
Soil erosion
Reduced crop yield

Note: FO = forest; WD = woodland; BS = bushland; GR = grassland; WET = wetland


vegetation/marsh; SET + C = settlement/cultivation; BR = bareland/burnscars.

Causes for the change

Table 7
Local peoples perceptions on problems as a result of land use and cover changes.

Percentage response (%) (n = 120)


Agree

Undecided

Disagree

96
30
72
99
87
85
83

3
31
23
1
8
10
9

1
39
5
0
5
5
8

The population in Malagarasi Catchment has increased from


about 24,800 in 1981 to about 57,698 respectively in 2002, with
an average household 5.6 against the national average of 4.9

This study investigated the land use and cover changes in the
Malagarasi river catchment and the perceptions of people about
these changes. This was an integrated assessment combining various methodological approaches in understanding the land resources dynamics and the resulting environmental problems as
revealed by respondents. The ndings have revealed that the study
area has undergone notable changes in terms of land use and land
cover for the period 1984/19862000/2002. The woodland areas
were found to be highly impacted, notably by the increased
anthropogenic activities likewise for the wetland vegetation. The
settlement and cultivated land was found to have consistently increased between the two periods under investigation as well as the
bare land area. Local knowledge revealed various factors associated
to land use and cover change that include re, cultivation along

J.J. Kashaigili, A.M. Majaliwa / Physics and Chemistry of the Earth 35 (2010) 730741

rivers and lake shores, overgrazing, poor law enforcement, insufcient knowledge on environmental issues, increasing poverty and
deforestation. The main factors mentioned as contributing to re
were beekeeping and hunting activities, while for deforestation include commercial logging, charcoals production, population
growth, expansion of commercial farming and food crops production. The most perceived environmental problems include drying
of streams and rivers, assumed change in rainfall, loss of soil fertility, soil erosion and reduced crop yield. The study concludes that,
there has been signicant changes in land use and cover in the
catchment. The study highlights the importance of integrating remote sensing and local knowledge in understanding the catchment
resources dynamics and generating information that could be used
to overcome the catchment problems for the sustainability of the
catchment resources.
Acknowledgements
The authors highly acknowledge the nancial support from Belgium Technical Cooperation and the all people consulted.
References
Alphan, H., 2003. Land use change and urbanization in Adana, Turkey. Land
Degradation and Development 14 (6), 575586.
Burns, G., Joyce, A., 1981. Evaluation of Land Cover Change Detection Techniques
Using Landsat MSS Data. In: Proceeding of 7th Pecora Symposium, Sioux Falls,
South Dakota, pp. 11271134.
Cihlar, J., Jansen, L.J.M., 2001. From land cover to land-use: a methodology for
efcient land-use mapping over large areas. The Professional Geographer 53 (2),
275289.
Coppin, P., Jonckheere, I., Nackaerts, K., Muys, B., Lambin, E., 2004. Digital change
detection methods in ecosystem monitoring: a review. International Journal of
Remote Sensing 25 (9), 15651596.
DANIDA, 1999. Environment, Peace and Stability Facility Environmental Support
Programme. Project Document, Sustainable and Integrated Management of the
MalagarasiMuyovozi Ramsar Site, Ministry of Foreign Affairs, Tanzania, 73p.
Dewan, M.A., Yamaguchi, Y., 2009. Using remote sensing and GIS to detect and
monitor land use and land cover change in Dhaka Metropolitan on Bangladesh
during 19602005. Environmental Monitoring and Assessment 150, 237249.
doi:10.1007/s10661-008-0226-5.
Di Gregorio, A., Jansen, L.J.M., 2000. Land Cover Classication System (LCCS):
Classication Concepts and User Manual. FAO, Rome (Incl. CD-ROM with
application software). <https://www.aginternetwork.net/_base> (site visited on
30.06.07).
ERDAS Field Guide, 1999. Earth Resources Data Analysis System. ERDAS Inc.,
Atlanta, Georgia, 628p.
Foody, G.M., 2002. Status of land cover classication accuracy assessment. Remote
Sensing of Environment 80, 185201.
Fraser, R.H., Abuelgasim, A., Latifovic, R., 2005. A method for detecting large-scale
forest cover change using coarse spatial resolution imagery. Remote Sensing of
Environment 95, 414427.
Fraser, R.H., Olthof, I., Pouliot, D., 2009. Monitoring land cover change and ecological
integrity in Canadas national parks. Remote Sensing of Environment 113,
13971409.
Institute of Resource Assessment (IRA), 2002. Baseline Study of Lake/Sagara and
Nyamagoma Wetlands and the Surrounding Environment in the Malagarasi/
Muyovozi Ramsar Site. A Consultancy Report Submitted to SIMMORS Project in
the Ministry of Natural Resources and Tourism, University of Dar es Salaam,
Tanzania, 99p.
Jensen, J.R., 1996. Introductory Digital Image Processing: A Remote Sensing
Perspective, second ed. Prentice Hall. p. 316.
Jones, C., Hill. J., 1994. The Malagarasi/Moyowosi/Kigosi/Ugalla Riverline Wetland
Ecosystem. Report Submitted to the Director of Wildlife on Current Status with
Recommendation for More Effective Management and Further Research,
Frankfurt Zoology Society, 75p.
Kashaigili, J.J., 2006. Landcover Dynamics and Hydrological Functioning of Wetlands
in the Usangu Plains in Tanzania. PhD Thesis, Sokoine University of Agriculture,
266p.
Kashaigili, J.J., Mbilinyi, B.P., Mccartney, M., Mwanuzi, F.L., 2006. Dyanamics of
Usangu plains wetlands: use of remote sensing and GIS as management
decision tools. Physics and Chemistry of the Earth 31 (2006), 969975.
Kennedy, R.E., Townsend, P.A., Gross, J.E., Cohen, W.B., Bolstad, P., Wang, Y.Q.,
Adams, P., 2009. Remote sensing change detection tools for natural resource
managers: understanding concepts and tradeoffs in the design of landscape
monitoring projects. Remote Sensing of Environment 113, 13821396.
Kidegesho, J.R., 2001. The loss of wildlife habitat in Tanzania: what is the way
forward? Kakakuona 26, 69.

741

Lal, R., 2001. Soil degradation by erosion. Land Degradation and Development 12
(6), 520539.
Lambin, E.F., 1997. Modeling and monitoring land-cover change processes in
tropical regions. Progress in Physical Geography 21 (3), 375393.
Lambin, E.F., Geist, H., Lepers, E., 2003. Dynamics of land use and land cover change
in tropical regions. Environment and Resources 28, 205241.
Li, Y., Liao, Q.F., Li, X., Liao, S.D., Chi, G.B., Peng, S.L., 2003. Towards an operational
system for regional-scale rice yield estimation using a time-series of Radarsat
ScanSAR images. International Journal of Remote Sensing 24, 42074220.
Lillesand, T.M., Kiefer, R.W., 1987. Remote Sensing and Image Interpretation. John
Wiley & Sons, New York. p. 250.
Lopez, E., Bocco, G., Mendoza, M., Duhau, E., 2001. Predicting land cover and land
use change in the urban fringe a case in Morelia City, Mexico. Landscape and
Urban Planning 55 (4), 271285.
Lyaluu, E.A., 1996. Draft Management Plan for Malagarasi/Muyovozi Complex
Riverine wetland System, Institute for Inland Water Management and Waste
Water Treatment RIZA, The Netherlands, 39p.
MacLeod, R.D., Congalton, R.G., 1998. A quantitative comparison of changedetection algorithms for monitoring eelgrass from remotely sensed data.
Photogrammetric Engineering and Remote Sensing 64 (3), 207216.
Mahinya, S.G., 2005. Assessment of Ecological Integrity of Land Use System Using
Birds as Bio-indicators in the MalagarasiMuyovozi Ramsar Site, Northwest
Tanzania-East Africa. Dissertation for Award of MSc Degree at Sokoine
University of Agriculture, Morogoro, Tanzania, 106p.
Majaliwa, A.M., 2009. Impact of Land Use and Land Cover Changes on Water Flows:
A Case of Malagarasi River Catchment in Tanzania. Draft MSc Thesis, Sokoine
University of Agriculture, 108p.
Monela, G., Solberg, 1998. Deforestation Rate and Land Use/Land Cover Changes in
Rainforest of the Nguru Mountains, Tanzania. Faculty of Forest Record No. 68.
SUA, 14p.
Mundia, C.N., Aniya, M., 2006. Dynamics of land use/cover changes and degradation
of Nairobi city, Kenya. Land Degradation and Development 17 (1), 97108.
Mungongo, C.G., 1995. Social Processes and Ecology in the Kondoa Irangi Hills,
Central Tanzania. Meddelanden Series B 93, Department of Human Geography,
Stockholm University.
NEMC/IUCN, 1994. Draft Project Proposal Tanzania Wetland, Conservation and
Management Support project, ICUN Gland, Switzerland, 88p.
Nkotagu, H.H., Simon, G.M.N. (Eds.), 2004. The Malagarasi Wetland Ecosystem: An
integrated study. University of Dar es Salaam Press, p. 230.
Ogungo, P.O., Njuguna, J.W., 2004. Institutions, Incentives and Conicts in Forest
Management. In: Shemweta, D.T.K. et al. (Eds.), A Perspective Proceedings of the
IFRI East African Regional Conference Nairobi, Kenya. pp. 924.
Richards, J.A., 1993. Remote Sensing Digitization: An Introduction to Image
Analysis, second ed. Springer-Verlag, New York, London and Tokyo. p. 340.
Rohr, P.C., 2002. A Hydrological Study Concerning the Southern Slopes of Mount
Kilimanjaro. Unpublished Doctorial Dissertation Submitted to the Norwegian
University of Science and Technology, Trondheim, Norwey, 166p.
Schuck, A., Paivinen, R., Hame, T., Van Brusselen, J., Kennedy, P., Folving, S., 2003.
Compilation of a European forest map from Portugal to the Ural Mountains
based on earth observation data and forest statistics. Forest Policy and
Economics 5, 187202.
Skole, D., Tucker, C., 1993. Tropical deforestation and habitat fragmentation in the
amazon. Satellite data from 1978 to 1988. Journal of Science 260, 19051910.
Slayback, D., 2003. Land cover change in the Takamanda Forest Reserve Cameroon
19862000. Takamanda: Biodiversity of African Rainforest 8, 173179.
Stehman, S.V., Wickham, J.D., Fattorini, L., Wade, T.D., Baffetta, F., Smith, J.H., 2009.
Estimating accuracy of land cover composition from two-stage cluster
sampling. Remote Sensing of Environment 113, 12361249.
URT, 2003. Population Census 2002, Preliminary Report. National Bureau of
Statistics: Presidents Ofce, Planning Commission, Dare es salaam, Tanzania.
WCED, 1987. Our Common Future. The World Commission on Environment and
Development. Oxford University Press, Oxford, New York, 400p.
Wickware, G.M., Howarth, P.J., 1981. Change Detection in the peaceathabasca delta
using digital landsat data. Remote Sensing of Environment 11, 925.
Yanda, P.Z., Shishira, E.K., 1999. Forestry conservation and resource utilization on
southern slopes of Mt. Kilimanjaro: trends, conicts and resolutions. In: Ngana
(Ed.), Water Management in Pangani River Basin, Challenges and Opportunities.
Dar es Salaam University Press, pp. 104110.
Yeh, A., Gar, A., Xia, L., 1996. Urban growth management in Pearl River delta: an
integrated remote sensing and GIS approach. ITC Journal 1, 7778.
Yuan, C., Elvidge, C., 1998. NALC land cover change detection pilot study:
Washington DC area experiments. Remote Sensing of Environment 66, 166
178.
Yuan, F., Sawaya, K.E., Loeffelholz, B.C., Bauer, M.E., 2005. Land cover classication
and change analysis of the twin cities (Minnesota) Metropolitan Area by
multitemporal landsat remote sensing. Remote Sensing of Environment 98
(2&3), 317328.
Zhang, J., Foody, G.M., 2009. Preface. Spatial accuracy in remote sensing.
International Journal of Remote Sensing 30, 52395242.
Zhang, Q., Wang, J., Peng, X., Shi, P., 2002. Urban built-up change detection with
road density and spectral information from multi-temporal Landsat TM data.
International Journal of Remote Sensing 23 (15), 30573078.
Zhou, Q., Li, B., Sun, B., 2008. Modelling spatio-temporal pattern of landuse change
using multitemporal remotely sensed imagery. The International Archives of
the Photogrammetry, Remote Sensing and Spatial Information Sciences XXXVII
(Part. B7), 729734.

Вам также может понравиться