Вы находитесь на странице: 1из 11

ISSN 00978078, Water Resources, 2011, Vol. 38, No. 6, pp. 827–836. © Pleiades Publishing, Ltd., 2011.

INTERACTION BETWEEN CONTINENTAL WATERS


AND THE ENVIRONMENT

Impact of Climate Change on Stream Discharge and Sediment Yield


in Northern Viet Nam1
D. B. Phana, C. C. Wub, and S. C. Hsiehb
a
Institute of Tropical Agriculture and International Cooperation, National Pingtung University of Science and Technology,
Pingtung 912, Taiwan
bDepartment of Soil and Water Conservation, National Pingtung University of Science and Technology, Pingtung 912, Taiwan

Received June 15, 2011

Abstract—The purpose of this paper is to apply “Soil and Water Assessment Tool (SWAT)” model to assess
the impacts of climate change on stream discharge and sediment yield from Song Cau watershed in Northern
Viet Nam. Three climate change scenarios B1, B2, and A2; representing low, medium, and high levels of
greenhouse gas emission, respectively, were considered in this study. The highest changes in stream discharge
(up to 11.4%) and sediment load (15.3%) can be expected in wet season in 2050s according to the high emis
sion scenario (A2), while for the low emission scenario the corresponding changes equal to 8.8% and 12.6%.
The results show that the stream discharge is likely to increase in the future during the wet season with
increasing threats of sedimentation.

Keywords: SWAT, climate change, sediment yield, stream discharge.


DOI: 10.1134/S0097807811060133

1
INTRODUCTION climate change scenario resulted in a higher mean
stream discharge might induce greater flooding, but
Climate change is one of the most significant chal the frequency of the normal and low stream discharges
lenges that all living matters on the Earth needs to decreased. Gosain et al. (2006) [5] simulated the
face. It will affect living habitats, productions, and
environment worldwide. Temperature and sea level impacts of a 2041–2060 climate change scenario on
rises may cause floods and salt water intrusion, thus stream discharges from 12 major river basins in India,
may damage agriculture as well as pose risks to indus ranging in size from 1668 to 87180 km2. Stream dis
tries and socialeconomic system. Over the past 50 charge was found to generally decrease, and the sever
years average temperature in Viet Nam has increased ity of both floods and droughts increased in response
about 0.5–0.7°C, and the consequences of climate to the climate change projection. Rosenberg et al.
change are considered to be serious and present signif (2003) [22] simulated the effect of downscaled Hadley
icant threats to the achievement of Millennium Devel Centre Coupled Model version 2 (HadCM2) [10] cli
opment Goals and nation’s sustainable development mate projections on the hydrology of 18 major water
[13]. Changes in water availability due to global warm resource regions (MWRRs) with SWAT within the
ing add further pressure on the adaptability of the hydrologic unit model of the United State (HUMUS)
water systems. However, assessing the impacts of cli framework. Water yields were predicted to change
mate change on water resource management is com from –11 to 153% and from 28 to 342% with respect to
plicated by the spatial resolution of climate change baseline conditions across the MWRRs in 2030 and
predictions. Limited efforts have been carried out in 2095, respectively.
Viet Nam regarding the potential impacts of climate
change on water resource management systems [17, Rosenberg et al. (1999) [23] found that driving
21]. SWAT with a different set of 12 climate projections
resulted in Ogallala Aquifer recharge decreases up to
Soil and Water Assessment Tools [19] known as
SWAT has been used since 1993 mainly by hydrolo 77% within the Missouri and ArkansasWhiteRed
gists for watershed hydrology related issues [12, 24]. water resource regions. Nearing et al. (2005) [18]
Muttiah and Wurbs [15] used SWAT to simulate the compared runoff and erosion estimates from SWAT
impacts of historical climate trends versus a 2040– versus six other models in response to six climate
2059 climate change projection for the 7300 km2 San change scenarios for the 150 km2 Lucky Hills water
Jacinto River basin in Texas. They reported that the shed in southeastern Arizona. The responses of all
seven models were similar across six scenarios. They
1 The article is published in the original.
concluded that climate change could potentially result

827
828 PHAN et al.

in significant soil erosion increases if necessary con Erosion and sediment yield are estimated for each
servation efforts are not implemented. subbasin with the Modified Universal Soil Loss
Hence, the objective of this study is to apply SWAT Equation (MUSLE) [29]. Sediment yield estimation
model in Song Cau watershed to assess the impact of in MUSLE is written as
climate change on stream discharge and sediment Sed = 11.8 ( Q surf q peak area hru )
0.56
K USLE C USLE P USLE
yield in northern Viet Nam. (2)
× LS USLE CFRG,
where Sed is the sediment yield on a given day (metric
METHODOLOGY tons), Qsurf is the surface runoff volume (mm/ha), qpeak
is the peak runoff rate (m3/s), areahru is the area of the
Brief Descriptions of SWAT Model hydrologic response unit (HRU) (ha), KUSLE is the
SWAT is a physically based, continuoustime Universal Soil Loss Equation (USLE) soil erodibility
hydrologic model with an ArcViewGIS interface factor (0.013 metric tonm2 hr/m3metric toncm),
developed by the Blackland Research and Extension CUSLE is the USLE cover and management factor,
Center and the United States Department of Agricul PUSLE is the USLE support practice factor, LSUSLE is
ture—Agricultural Research Service (USDAARS) the USLE topographic factor [30], and CFRG is the
[2] to predict the impact of land management prac coarse fragment factor.
tices on water, sediment and agricultural chemical The channel sediment routing equation uses a
yields in large, complex basins with varying soil type, modification of Bagnold’s sediment transport equa
land use and management conditions over long peri tion [3] that estimates the transport capacity of the
ods of time [20]. streamflow as a function of flow velocity. The model
SWAT delineates watersheds into subbasins inter either deposits excess sediment or reentrains sedi
connected by a stream network. Each subbasin is fur ment through channel erosion depending on the sedi
ther divided into hydrologic response units (HRUs) ment load entering the channel [25].
based upon unique soil and land class characteristics.
SWAT sums the flow, sediment and nutrient loading Study Area and Its Schematization
from each subbasin HRU and the resulting loads are
then routed through channels, ponds, and reservoirs The Song Cau watershed is located in northern Viet
to the watershed outlet [1]. A single growth model in Nam, between 21°07′–22°18′ N and 105°28′–
SWAT, based on a simplification of Erosion Produc 106°08′ E (Fig. 1a). The watershed has an area of
tivity Impact Calculator (EPIC) crop model, is used 2941 km2 and agriculture, mainly arable and pasture
for simulating all crops development [26]. Phenologi lands, dominants the land use. The elevation ranges
cal development of a crop is based on daily heat unit from 24 to 1498 m above sea level with the mean eleva
accumulation. SWAT also uses WXGEN weather gen tion of 285 m (Fig. 1b). The length of the main stream
erator model [21] to generate climate data or to fill is 125 km. High rainfall events in combination with
gaps in the measured records. less permeable soil formations caused significant run
off and subsequently high soil losses and sediment
The hydrological processes of SWAT are divided yield.
into two phases [8]: 1) the land phase, which controls
the amount of water, sediment and nutrients received Data from Thai Nguyen Meteorological and
by a water body and 2) the water routing phase, which Gauging station show that the study area is influenced
simulates water movement through the channel net by tropical monsoon climate. The average annual
work. rainfall and temperature is 2000–2500 mm and 22°C,
respectively, and the highest precipitation usually
The hydrologic cycle of SWAT is based on the water occurs from June to August which constitutes more
balance equation than 70% total annual rainfall. The winter season,
t November–March, is mostly dry with the average
SW t = SW 0 + ∑ (R day – Q surf – E a – W seep – Q gw ), (1) temperature under 15°C and the lowest temperature
recorded in recent years is 3°C. The summer, April–
i=1
October, is the hot rainy season with the average tem
where SWt is the final soil water content (mm H2O),
perature of 26°C, the highest temperature recorded so
SW0 is the initial soil water content on day i (mm far is 39°C [6].
H2O), t is the time (days), Rday is the amount of precip The entire Song Cau watershed was divided into
itation on day i (mm), Qsurf is the amount of surface 27 subwatersheds (Fig. 1c) based on the similarity of
runoff on day i (mm), Ea is the amount of evapotrans land use to allow consideration of significant spatial
piration on day i (mm), Wseep is the amount of water detail for climatic, soil, topographic, and land cover
entering the vadoes zone from the soil profile on day i variations within Song Cau watershed. Table 1 pre
(mm), and Qgw is the amount of return flow on day i sents a listing of the respective land cover types, sub
(mm) [19]. watershed areas, and USLE C factors [11] for each

WATER RESOURCES Vol. 38 No. 6 2011


IMPACT OF CLIMATE CHANGE ON STREAM DISCHARGE 829

DEM OF SONG CAU WATERSHED


CHINA
Lao Cai

NANOI
N
Haiphong Song Cau
watershed
LAOS Gulf of
Vinh Hainan
Me Tonkin (CHINA)
ko
ng

Hue
THAILAND Da Nang

Qui Nhon
CAMBODIA
Nha Trang
Cam Ranh 24–169
170–336
M Ho Chi Minh City 3337–503
Dao Phu Long ekong 504–670
South 671–837
Quoc Xuyen Can Tho
China 838–1004
Gulf Con Dao Sea 1005–1171
of 0 100 200 km 1172–1338
Thailand 1339–1498

0 100 200 m

WATERSHED DELINEATION OF SONG CAU

N OuBets
1 Linking stream added Outlet
2 6 Manually added Outlet
4 3 Streams

Subbasins
23
7
8
11 9
13
12 16 14
10 15 18

17
25
20
22
19 Weagages

Raingages
26
27

Fig. 1. a—Location, b—Digital Elevation Model (DEM), and c—Watershed delineated of Song Cau watershed.

land cover type delineated. The land use type and GIS system were obtained from the Soils database
topography of each subwatershed were both extracted contained in SWAT. The sources and types of data col
from the GIS database. lected are shown in Table 2.

Data Climate Change Scenarios

Data collected for this study included Soil, DEM, The climate change is believed to be primarily
and Land use map, forest area, land cover, precipita dependent upon greenhouse gas emission level.
tion, temperature, solar radiation, relative humidity, Greenhouse gas emission has found to be correlated
and wind speed, stream discharge, and sediment yield. with human activities such as industry, agriculture,
Measured daily maximum and minimum tempera transportation, deforestation etc. Hence, grounds for
tures were obtained from Bac Kan, Dinh Hoa and determining greenhouse gas emission scenarios are:
Thai Nguyen stations. Monthly stream discharge (1) economic growth at a global scale, (2) global pop
measured at the watershed outlet (Gia Bay station) for ulation and consumption, (3) living standards and
44 years (1964–2008) was designated for model cali behavior, (4) energy sources and consumption,
bration and validation, while for sediment, 37 year (5) technology transfer, and (6) land use change, etc.
observations (1972–2008) were used. Soil parameters In the Special report on emission scenarios in
for the corresponding soil series extracted from the 2000, Intergovernmental Panel on Climate Change

WATER RESOURCES Vol. 38 No. 6 2011


830 PHAN et al.

Table 1. Areas, Land cover types, and USLE C factor delineated in the Song Cau watershed

SWAT Area USLEC


Land Use Type
Code km 2 % Factor

WATR Water 399.48 13.58 0.00


URMD Urban 163.84 5.57 0.008
FRSD ForestDeci duous 152.14 5.17 0.001
FRSE ForestEvergreen 383.91 13.05 0.001
FRST ForestMixed 173.32 5.89 0.001
RICE Rice 266.90 9.08 0.030
PAST Pasture 1350.69 45.93 0.003
AGRL Agricultural LandGeneric 1.26 0.04 0.200
AGRR Agricultural LandRow Crops 49.28 1.68 0.200
Total 2940.86 100

Table 2. Sources and types of data collected for SWAT simulation


Types of data Sources of data
Precipitation (rainfall) Thai Nguyen, Bac Can, and Dinh Hoa weather station
Temperature solar radiation, relative humidity,
and wind speed.
Runoff discharge, Sediment yield (observed data Gia Bay, Thac Buoi, and Thac Gieng gauging stations
Topography (DEM) and Land use map Department of Information and Communication technology for Natural
Resources and Environment, Ministry of Natural Resources and Environ
ment plus field survey data
Soil map Viet Nam Soil and Fertilizers Research Institute
Forest, Agricultural land (as shown in Table 1) Department of Natural Resources and Environment of Thai Nguyen and
Bac Can provinces, and field survey data

(IPCC) introduced 40 scenarios that reflect relatively mented and slow (high emission scenario, similar to
diversified possibilities of greenhouse gas emission in A1FI).
the 21st century. These emission scenarios are classi B1 family; Rapid economic growth as in A1 family
fied into four families with the main characteristics as but with rapid changes toward a service and informa
follows [9]: tion economy; global population reaches the peak in
A1 family: Rapid economic growth; global popula 2050 and declines thereafter; reduction in material
tion peaks at 9 billion in 2050 and declines thereafter; intensity and introduction of clean and resourceeffi
cient technologies; emphasis on global solutions to
the rapid introduction of new and more efficient tech economic, social and environmental sustainability
nologies; a convergence of world income and way of (low emission scenario, similar to A1T).
life; a convergence among regions and increasing cul
tural and social interaction. The A1 family further B2 family: Continuously increasing population but
develops into 3 groups based on the technological at a rate lower than A2 family; the emphasis is on local
emphasis: (1) A1FI: an emphasis on fossilintensive rather than on global solutions to economic, social
(high emission scenario); (2) A1B: an emphasis on a and environmental sustainability; intermediate levels
balance across all energy sources (medium emission of economic development; less rapid and more diverse
technological change than B1 and A1 families
scenario); (3) A1T: an emphasis on nonfossil energy (medium emission scenario, in the same group of
sources (low emission scenario). A1B).
A2 family: A very heterogeneous world; selfreli Criteria of method selection for climate change
ance and preservation of nations; continuously scenario development in Viet Nam [13] include:
increasing population in the 21st century; regionally (1) plausibility of global climate change, (2) level of
oriented economic development; technological climate change, (3) inheritability, (4) updating,
change and per capita economic growth are more frag (5) local appropriateness, (6) completeness of scenar

WATER RESOURCES Vol. 38 No. 6 2011


IMPACT OF CLIMATE CHANGE ON STREAM DISCHARGE 831

Table 3. Changes in Annual mean temperature, °C, and Annual rainfall, %, relative to those from 1980–1999 for northern
Viet Nam
Scenario B1 Scenario B2 Scenario A2
Items
2020s 2030s 2050s 2020s 2030s 2050s 2020s 2030s 2050s
Rainfall
Mean annual 1.4 2.1 3.6 1.4 2.1 3.8 1.6 2.1 4.0
Mean wet season 3.3 4.8 8.4 3.3 4.8 8.9 3.8 4.9 9.2
Mean dry season –0.5 –0.6 –1.2 –0.5 –0.6 –1.3 –0.6 –0.7 –1.2
Mean temperature
Mean annual 0.5 0.7 1.2 0.5 0.7 1.3 0.5 0.8 1.3
Mean wet season 0.5 0.7 1.2 0.5 0.7 1.2 0.5 0.7 1.2
Mean dry season 0.6 0.8 1.3 0.6 0.8 1.4 0.6 0.9 1.5

ios, and (7) possibility of self updating. Based on the using daily historical weather information. The simu
analysis of the above mentioned criteria, the results lated stream discharge (1964–2008) and sediment
from Stochastic Weather Generator (LARSWG) yield (1972–2008) were compared to the stream dis
software [27] and statistical downscaling method are charge and sediment yield measured in the same
selected for the development of climate change sce period at the same gaging stations (Gia Bay station).
narios. Hence, climate change scenarios for tempera Considering that SWAT is not a ‘parametric model’
ture and rainfall were developed for seven climate with a formal optimization procedure to fit any data
zones in Viet Nam, and the northern zone was selected and that it uses physicallybased inputs, original
for this study. parameters were adjusted to provide a better fit. The
The resultant scenarios B1, B2, and A2 represent curve number (CN2) were adjusted within 4 percent
low, medium, and high level of greenhouse gas emis from the tabulated curve numbers to reflect conserva
sion, respectively. The changes of rainfall and temper tion tillage practices and soil residue cover conditions
ature are given in Table 3. Rainfall changes are highly of the watershed and others parameters as ESCO,
variable with the highest changes in wet season show EPCO, GW_REVAP were adjusted too (Table 4).
ing increases for all scenarios in the 2020s, 2030s, and Also, the Cover or management factor (C FACTOR),
2050s. Increases in rainfall in wet season could result linear factor (SPCON) and exponential factor
in high surface runoff and that could aggravate the (SPEXP) for channel sediment routing and Residue
existing problem of flooding in lower Song Cau. Mean decomposition coefficient (RSDCO) were yield in the
annual temperatures increase about 0.5°C for all sce area. The sequence of adjusting the model parameters
narios in the 2020s while 1.2°C for B1 and 1.3°C for were based on the procedures outlined by Santhi et al.
B2 and A2 scenario in the 2050s. (2006) [25]. Table 4 shows a summary of the optimized
The daily data was obtained and used together with model parameters.
observed data in LARSWG to generate monthly rain This study uses NashSutcliffe efficiency (NSE)
fall and temperature changes. Three stations: Thai [16], and percent bias (PBIAS) to evaluate model per
Nguyen, Bac Kan, and Dinh Hoa were chosen for cli formance.
mate change scenario generation. The NSE value is calculated using the following
Eq. (3)
Model Calibration and Validation n

∑ (Q
i i 2
obs – Q sim )
To account for spatial variability in topographic,
soil types, and land use factors in Song Cau watershed, i=1
NSE = 1 – 
 , (3)
n
parameters governing stream discharge response in

i i 2
SWAT were calibrated in a distributed fashion. Stream ( Q obs – Q obsmean )
discharge records from 1964 to 1984 were used for i=1
model calibration and records from 1985 to 2008 were i
designated for validation. For sediments, 19 year data where n is the number of registered data points, Q obs ,
recorded from 1972 to 1990 and 18 year data from i
and Q sim are the observed and simulated data, respec
1991 to 2008 were used for calibration and validation, tively on the th time step, and Qobsmean is the mean of
respectively. i
Using the SWAT default parameters, the watershed observed data ( Q obs ) across the n evaluation time
conditions were simulated from 1964 through 2008 steps.

WATER RESOURCES Vol. 38 No. 6 2011


832 PHAN et al.

Table 4. The most sensitive parameters in SWAT and values used in this study
Variable name Model processes Description Normal range Value used
CN2 Flow Curve number ±5 ±4
ESCO Flow Soil evaporation compensation factor 0.00–1.00 0.70
EPCO Flow Plant uptake compensation factor 0.00–1.00 0.02
GW_REVAP Flow Groundwater “revap” coefficient 0.02–0.40 0.2
CFACTOR Sediment Cover or management factor 0.001–0.20 0.001–0.20
SPCON Sediment Linear factor for channel sediment routing 0.0001–0.01 0.002
SPEXP Sediment Exponential factor for channel sediment routing 1.0–1.5 1.0
BIOMIX Sediment, organic Biological mixing efficiency 0.2–0.5 0.3
RSDCO Sediment, organic Residue decomposition coefficient 0.01–0.05 0.05

PBIAS is used to determine if the average tendency Δc = 


 c ( k + Δk ) 
– c ( k – Δk) . (7)
of the simulated data is either larger or smaller than its Δk 2Δk
observed counterparts [7]. PBIAS is calculated using
the following equation: The condition number provides a transfer function
n
to propagate the relative error of the parameter into
the relative error of the prediction. The bigger the con
∑ (Q
i i
obs – Q sim ) dition number, the more sensitive the parameter is for
=1  × 100%.
PBIAS = i (4) the specific model prediction. A negative condition
n
number indicates that the parameter has an opposite

i
Q obs effect on prediction. The most sensitive parameters are
i=1 summarized in Table 4.
All parameters in Eq. (4) shares the same defini
tions as that shown in Eq. (3). RESULTS
Simulation of Stream Discharge and Sediment Load
Sensitivity Analysis Comparison of measured versus simulated monthly
Sensitivity analysis is a technique for assessing the hydrographs from Gia Bay gauge station at the main
comparative change in model response resulting from stream of Song Cau showed very good agreement as
a change in model inputs. It helps identify the param that shown in Fig. 2. For the calibration period from
eters that affect the model’s output significantly. Iden 1964 to 1984, monthly NSE was 0.82, while PBIAS
tification of sensitive input parameters may help the was—1.6%, For the validation period (1985–2008)
modeler adjust those parameters during model cali NSE and PBIAS were lower than those for the calibra
bration. tion period with respective values of 0.77 and 5.9%
Spruill et al. (2000) [28] indicated that the most (Table 5).
sensitive parameters of SWAT model for application in In general, SWAT performance in simulating sedi
a central Kentucky watershed were saturated hydraulic ment response was considered good based on monthly
conductivity, alpha base flow factor, drainage area, NSE values for the calibration and validation periods,
channel length, and channel width. which were 0.66 and 0.69 respectively, Moriasi et al.
An expression of sensitivity called “condition (2007) [14] concluded that a model performance of
number” [4] is used in this study. sediment load is very good if PBIAS < ±15%; good if
Consider c as a function of each of the model ±15% ≤ PBIAS < ±30%; satisfactory if ±30% ≤
parameters and forcing functions: that is, c = f(k1, k2, PBIAS < ±55% and unsatisfactory if PBIAS ≥ ±55%.
k3, …). The condition number can be expressed as In this study the PBIAS value was satisfactory
(⎯36.1%) for calibration and good (–26.4%) for vali
∂c ( k,
CN k = k) dation period (Table 5). The negative value of PBIAS
(5)
x ∂k indicated that the sediment load was overestimated.
where CNk is defined as the condition number for the SWAT tends to overestimate sediment loads for all
parameter k. If the derivative form is difficult to months of year, except August of the validation period
obtain, a discrete form is used for the derivative (Fig. 3).
The conceivable reason contributing to this result is
kΔc
CN k =  , (6) the land use map generated based on year 2008 survey
cΔk data which may cause differences in land use installed
where in the previous period. Another reason causing dis

WATER RESOURCES Vol. 38 No. 6 2011


IMPACT OF CLIMATE CHANGE ON STREAM DISCHARGE 833

Monthly Mean Strem Discharge (cms) Monthly Mean Strem Discharge (cms) Calibration Calibration

Avarage Monthy Sediment (tons)


180 200000
160 180000 Observed
Observed 160000
140 Simulated Simulated
140000
120
120000
100
100000
80 80000
60 60000
40 40000
20 20000
0 0
JanFebMarAprMay Jun Jul AugSepOctNov Dec Jan FebMarAprMayJun Jul Aug Sep OctNovDec

Avarage Monthy Sediment (tons)


Validation
200000
Validation 180000
180
Observed 160000 Observed
160 Simulated Simulated
140000
140
120000
120 100000
100 80000
80 60000
60 40000
20000
40
20 0 Jan FebMarAprMayJun Jul Aug Sep OctNovDec

0
JanFebMarAprMay Jun Jul AugSepOctNov Dec Fig. 3. Observed versus Simulated Average monthly Sedi
ment load during the Calibration and Validation periods of
Song Cau watershed.

Fig. 2. Observed versus Simulated Average monthly stream


discharge during the Calibration and Validation periods of a management tool to predict the effects of land use
Song Cau watershed. changes in midsize watersheds.

crepancy between simulated and observed sediment Impact of Climate Change on Stream Discharge
yield may be attributed to channel erosion, especially and Sediment Yield
during high flows. Other factors include SWAT’s inad
From SWAT simulation output (Table 6 and
equate description of channel scouring process and Fig. 4a), climate change scenario B1 increases the
the presence of temporary channel embankment used mean wet season (May–October) runoff by an average
by farmers to retard channel flow velocity. Neverthe of 8.8% in 2050s, but decreases by 3.9% for mean dry
less, the overall adequacy of SWAT to simulate sedi season (November–April). In the 2050s, the percent
ment yields in the watershed indicates its usefulness as ages of stream discharge of mean wet season increase

Table 5. Monthly Stream discharge and Sediment load coefficient of NashSutcliffe efficiency (NSE), and Percent bias
(PBIAS) of Song Cau

Items Simulation period Period of Record Monthly NSE PBIAS, %

Stream Calibration 1964–1984 0.82 –1.6

Discharge Validation 1985–2008 0.77 5.9

Sediment Calibration 1972–1990 0.66 –36.1

Load Validation 1991–2008 0.69 –26.4

WATER RESOURCES Vol. 38 No. 6 2011


834 PHAN et al.

Table 6. Percentages of Stream discharge and Sediment load change with respect to Baseline period for Mean annual, Wet
season and Dry season
Scenario B1 Scenario B2 Scenario A2
Items
2020s 2030s 2050s 2020s 2030s 2050s 2020s 2030s 2050s
Stream discharge change
Mean annual 1.4 1.2 2.4 1.4 1.0 2.2 1.5 1.0 3.0
Mean wet season 3.9 5.3 8.8 4.1 5.3 9.0 4.5 5.7 11.4
Mean dry season –1.1 –2.8 –3.9 –1.3 –3.0 –4.7 –1.5 –3.8 –5.3
Sediment load change
Mean annual 1.2 1.8 4.0 1.7 1.9 4.7 2.0 1.5 3.8
Mean wet season 3.6 6.3 12.6 5.1 7.4 14.6 5.7 8.0 15.3
Mean dry season –1.3 –2.7 –4.6 –1.7 –3.6 –5.3 –1.7 –5.1 –7.7

by 8.8, 9.0 and 11.4% for Scenarios B1, B2 and A2 tion, which was impacted by the increased variability
respectively. For the B2 scenario, the mean annual of monthly precipitation as projected by LARSWG.
stream discharge increases by 2.2% while mean dry Greater variability in daily precipitation distribution
season decreases by 4.7% with respect to baseline led to increased occurrence of large storms and there
period (1980–1999). The increase in stream discharge fore greater stream discharge and soil loss.
was caused by increased variability of daily precipita In general, A2 climate change scenario produces
higher stream discharge as compared to that generated
from B1 and B2 for the 2020s. Table 6 shows that the
(a) percentages of flow discharge increases in the wet sea
14 son from 3.9% (B12020) to 11.4% (A22050) in par
Percentage of Flow change (%)

12 Mean annual
Mean wet season allel with mean wet season rainfall increase from 3.3 to
10 Mean dry 9.2%, respectively. In dry season, when rainfall
8 season decreases from 0.5 (B12020) to 1.2% (A22050) and
6 temperature increases from 0.6 to 1.5°C, streamflow
4 reduces from 1.1 to 5.3%. That means more floods in
2 wet and drought in dry season may occur in the future.
0 In general, change in stream discharge in this study
–2 was lower than in other previous research. For
A2 20

A2 30

0
B2 20

B2 30

A2 50
B1 30

B2 50
B1 20

05
0

0
0

0
0

instance, Nguyen et al. (2011) applied SWAT to model


_2

_2

_2
_2

_2

_2
_2

_2

_2

–4
B1

–6 the impact of climate change on stream discharge in


–8 central Viet Nam reported that when precipitation
Scenarion increased by 4.8 % in 2030 and by 9.2% in 2050,
stream discharge increased by approximately 4.8 and
Percentage of Sediment load change (%)

(b)
16.9%, respectively. Muttiah et al. (2002) summarized
20 his study that the stream discharge increased by 21.5%
Mean annual in 2040 when precipitation increased by 11.3% with
Mean wet season respect to baseline period (1990–2000), etc.
15 Mean dry
season Referring to sediment loads in Song Cau under dif
10 ferent climate change scenarios (Table 6 and Fig. 4b),
it can be seen that increasing in flow discharge will in
5
general increase mean annual sediment load, while
0 decrease in flow discharge will decrease mean sedi
ment loads for all scenarios. For instance, Scenario
A2 20

A2 30

0
B2 20

B2 30

A2 50
B1 20

B1 30

B2 50

05

B1, B2, and A2 produces significant decrease in sedi


0

0
0

0
0

–5
_2

_2

_2
_2

_2

_2
_2

_2

_2

ment loads during mean dry season by 4.6, 5.3, and


B1

–10 7.7%, respectively in the 2050s; whereas, significant


Scenarion increase of sediment loads by 3.6% (B12020) to
15.3% (A22050) was predicted with respect to base
Fig. 4. Percentage of Stream discharge (a) and Sediment
line period for mean wet season. Figures from Table 3
yield change in Mean annual (b), Wet season and Dry sea and Table 6 show a close correlation between changes
son with respect to Baseline scenario. in precipitation variability and changes in monthly soil

WATER RESOURCES Vol. 38 No. 6 2011


IMPACT OF CLIMATE CHANGE ON STREAM DISCHARGE 835

loss. An increase in precipitation variability was often Hydrologic Engineering, 1999, vol. 4, no. 2, pp. 135–
accompanied by an increase in soil loss, and vice versa 143.
[21]. These results indicate that soil loss prediction is 8. Hiroaki, S., and Ikuo, T., Numerical Analyses on Sea
sensitive to changes in precipitation variability. sonal Variations of Nutrient Salts and Load Discharges
in Abashiri River Basin, 2009 International SWAT Conf.
Proc., 2009, pp. 241–248.
CONCLUSIONS 9. Intergovernmental Panel on Climate Change (IPCC)
Working Group I. Climate Change 2001: the Scientific
The SWAT model was parameterized and cali Basis. Contribution of Working Group I To the Third
brated in selected Song Cau watershed in the northern Assessment Report of the IPCC, Cambridge: Cambridge
Viet Nam with an aggregate area of 2941 km2 to simu University Press, 2001.
late the effects of climate change on stream discharge 10. Johns, T.C., Carnell, R.E., Crossley, J.F., et al., The
and sediment yield. Results indicated that SWAT ade second Hadley Centre coupled oceanatmosphere
quately predicted the stream discharge of the water GCM: Mode description, spinup, and validation, Clim.
shed with NashSutcliffe Efficiency ranging from 0.77 Dynam., 1997, vol. 13, no. 2, pp. 103–134.
to 0.82. Likewise, the model adequately predicted the 11. Karaburun, A., Estimation of C Factor for Soil Erosion
sediment yield with NSE ranging from 0.66 to 0.69. In Modeling Using NDVI in Buyukcekmece Watershed,
the 2050s, for scenario A2, the percentage of stream Ozean Journal of Applied Science, 2010, vol. 3, no. 1,
discharge increases by 11.4% in mean wet season and pp. 77–85.
decreases by 5.3% in mean dry season with respect to 12. Keshta, N., Elshorbagy, A., and Carey, S., A Generic
baseline scenario. That means more floods in wet sea System Dynamics Model for Simulating and Evaluat
son and draught in dry season may occur in the future. ing The Hydrological Performance of Reconstructed
Consequently, this will increase sediment yield by Watersheds, J. Hydrol. Earth Syst. Sci., 2009, vol. 13,
15.3% in wet season with respect to baseline scenario. pp. 865–881.
This study showed that SWAT model can be used as a 13. MORE (Ministry of Natural Resources and Environ
management tool for modeling the impact of climate ment). Climate change, Sea Level Rise Scenarios for Viet
change in Viet Nam watersheds. Nam, Ha noi, 2009.
14. Moriasi, D.N., Arnold, J.G., Van Liew, M.W., et al.,
Model Evaluation Guidelines for Systematic Quantifi
ACKNOWLEDGMENT cation of Accuracy in Watershed Simulations, Trans.
ASABE, 2007, vol. 50, pp. 885–900.
The authors’ sincere thanks go to Viet Nam Gov
15. Muttiah, R.S. and Wurbs, R.A., Modeling the Impacts
ernment Agencies for providing data on Song Cau of Climate Change on Water Supply Reliabilities, Intl.
watershed and for preparing the detailed land use map. Water Resources, 2002, vol. 27, no. 3, pp. 407–419.
In particular, the corresponding author is especially 16. Nash, J.E.. and Sutcliffe, J.V., River Flow Forecasting
grateful to Taiwan Scholarship Office for funding his through Conceptual Models. Part IA Discussion of
Ph.D. programme. Principles, J. Hydrology, 1970, vol. 10, no. 3, pp. 282–
290.
REFERENCES 17. Nguyen, K.L., and Nguyen, V.T., Assessing Climate
Change Impacts and Adaptation in Central Viet Nam
1. Arnold, J.G., Allen, P.M., and Morgan, D.S., Hydro Using Watershed and Community Based Approach:
logic Model for Design and Constructed Wetlands, Case Study in Quang Nam Province, 2011 Interna
Wetlands, 2001, vol. 21, no. 2, pp. 167–178. tional SWATSEA Conference Proceedings, 2011,
2. Arnold, J.G., Srinivasan, R., Muttiah, R.S., and Will pp. 125–137.
iams, J.R., Large Area Hydrologic Modeling and 18. Nearing, M.A., Jetten, V., Baffaut, C., et al., Modeling
Assessment. Part 1: Model Development, J. of Amer. Response of Soil Erosion and Runoff to Changes in
Resource Assoc., 1998, vol. 34, no. 1, pp. 73–89. Precipitation and Cover, Catena, 2005, vol. 61, nos. 2–
3. Bagnold, R.A., Bedload Transport in Natural Rivers, 3, pp. 131–154.
Water Resour. Res., 1977, vol. 13, no. 2, pp. 303–312. 19. Neitsch, S.L., Arnold, J.G., Kiniry, J.R., and Williams, J.R.,
Soil and Water Assessment Tool Theoretical Documenta
4. Chapra, S.C., Surface Water–Quality Modeling, New
tion, Version, 2005a, Temple, Tex.: USDAARS Grass
York: McGraw–Hill, 1997.
land, Texas: Soil and Water Research Laboratory, 2005.
5. Gosain, A.K., Rao, S., and Basuray, D., Climate 20. Panagopoulos, I., Mimikou, M., and Kapetanaki, M.,
Change Impact Assessment on Hydrology of Indian Estimation of Nitrogen and Phosphorus Losses to Sur
River Basins, Current Sci., 2006, vol. 90, no. 3, pp. 346– face Water and Groundwater through the Implementa
353. tion of the SWAT Model for Norwegian Soils, J. of Soils
6. GSOV (General Statistics Office of Viet Nam). Statistical & Sediments, 2007, vol. 23, no. 4, pp. 95–105.
Year Book 2008, Ha Noi: National Political Publishing 21. Phan, D.B., Wu, C.C., and Hsieh, S.C., Impact of Cli
House, 2008. mate Change and Deforestation on Stream Discharge
7. Gupta, H.V., Sorooshian, S., and Yapo, P.O., Status of and Sediment Yield in Phu Luong Watershed, Viet
Automatic Calibration for Hydrologic Models: Com Nam, J. Envir. Science and Engineering, 2011, vol. 5,
parison with Multilevel Expert Calibration, Journal of no. 1, pp. 92–101.

WATER RESOURCES Vol. 38 No. 6 2011


836 PHAN et al.

22. Rosenberg, N.J., Brown, R.A., Izaurralde, R.C., and 26. Sharpley, N., and Williams, J.R., EPICErosion Pro
Thomson, A.M., Integrated Assessment of Hadley ductivity Impact Calculator Model Documentation, U.S.
Centre (HadCM 2) Climate Change Projections in Department of Agriculture, Agricultural Research Ser
Agricultural Productivity and Irrigation Water Supply vice, Tech. Bull., 1990.
in the Conteminous United States: I Climate Change
Scenarios and Impacts on Irrigation Water Supply Sim 27. Semenov, M.A., and Barrow, E.M., LARSWG: A Sto
ulated with the HUMUS Model, Agric. For. Meteor., chastic Weather Generator for Use in Climate Impact
2003, vol. 117, no. 1–2, pp. 73–96. Studies, User Manual, 2002.
23. Rosenberg, N.J., Epstein, D.L., Wang, D., Vail, L., 28. Spruill, C.A., Workman, S.R., and Taraba, J.L., Simu
et al., Possible Impacts of Global Warming on the lation of Daily and Monthly Stream Discharge from
Hydrology of the Ogallala Aquifer Region, Clim. Small Watersheds Using the SWAT Model, Trans.
Change, 1999, vol. 42, no. 4, 677–692. ASAE, 2000, vol. 43, no. 6, pp. 1431–1439.
24. Santhi, C., Arnold, J.G., Williams, J.R., et al., Valida
tion of the SWAT Model on A Large River Basin with 29. Williams, J.R., Jones, C.A., and Dyke, P.T., A Model
Point and Nonpoint Sources, J. Am. Water Resour., ing Approach to Determining the Relationship between
2001, vol. 37, no. 5, pp. 1169–1188. Erosion and Soil Productivity, Trans. of the ASABE.,
1984, vol. 21, pp. 129–144.
25. Santhi, C., Srinivasan, R., Arnold, J.G., and
Williams, J.R., A Modeling Approach to Evaluate the 30. Wischmeier, W.H., and Smith, D.D., Predicting Rain
Impacts of Water Quality Management Plans Imple fall Erosion Losses—A Guide to Conservation Plan
mented in a Watershed in Texas, J. Envir Modelling & ning, US Department of Agriculture, Agricultural Hand
Software, 2006, vol. 21, pp. 1141–1157. book, 1978, no. 537.

WATER RESOURCES Vol. 38 No. 6 2011


Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Вам также может понравиться