HYDROLOGICAL PROCESSES Hydrol. Process. (2009) Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/hyp.7476
Spatial delineation of soil erosion vulnerability in the Lake Tana Basin, Ethiopia Shimelis G. Setegn,1,2 * Ragahavan Srinivasan,3 Bijan Dargahi1 and Assefa M. Melesse2 1
Division of Hydraulic Engineering, Department of Land and Water Resources Engineering, The Royal Institute of Technology (KTH), Stockholm, Sweden 2 Department of Earth and Environment, Florida International University, Miami, FL, USA 3 Spatial Science Laboratory, Texas A & M University, College Station, TX, USA
Abstract: The main objective of this study was to identify the most vulnerable areas to soil erosion in the Lake Tana Basin, Blue Nile, Ethiopia using the Soil and Water Assessment Tool (SWAT), a physically based distributed hydrological model, and a Geographic Information System based decision support system that uses multi-criteria evaluation (MCE). The SWAT model was used to estimate the sediment yield within each sub-basin and identify the most sediment contributing areas in the basin. Using the MCE analysis, an attempt was made to combine a set of factors (land use, soil, slope and river layers) to make a decision according to the stated objective. On the basis of simulated SWAT, sediment yields greater than 30 tons/ha for each of the sub-basin area, 18Ð4% of the watershed was determined to be high erosion potential area. The MCE results indicated that 12–30Ð5% of the watershed is high erosion potential area. Both approaches show comparable watershed area with high soil erosion susceptibility. The output of this research can aid policy and decision makers in determining the soil erosion ‘hot spots’ and the relevant soil and water conservation measures. Copyright 2009 John Wiley & Sons, Ltd. KEY WORDS
soil erosion; Lake Tana; SWAT; MCE; GIS; hydrologic modeling
Received 11 November 2008; Accepted 12 August 2009
INTRODUCTION Soil erosion and loss of agricultural soils is a major problem in Blue Nile River Basin, Ethiopia. The high rate of surface erosion in the basin and the rate of sediment transport in the river system contributes to increased sedimentation problems in the Lake and reservoirs as well as the downstream areas. Poor land use practices, improper management systems and lack of appropriate soil conservation measures have played a major role for causing land degradation problems in the country. Because of the rugged terrain, the rates of soil erosion and land degradation in Ethiopia are high. The soil depth of more than 34% of the land area is already less than 35 cm [Zemenfes, 1995; Soil Conservation Research Project (SCRP, 1996)]. Hurni (1989) indicated that Ethiopia loses about 1Ð3 billion metric tons of fertile soil every year and the degradation of land through soil erosion is increasing at a high rate. According to Kr¨uger et al. (1996), 4% of the highlands are now so seriously eroded that they will not be economically productive again in the foreseeable future. The SCRP (1996) has estimated an annual soil loss of about 1Ð5 billion tons from the highland. According to the Ethiopian Highlands Reclamation Study (EHRS, 1984), soil erosion is estimated to cost the country $1Ð9 billion between 1985 and 2010. These call for * Correspondence to: Shimelis G. Setegn, Division of Hydraulic Engineering, Department of Land & Water Resources Engineering, The Royal Institute of Technology (KTH), Teknikringen 76-3tr, 100 44 Stockholm, Sweden. E-mail:
[email protected];
[email protected] Copyright 2009 John Wiley & Sons, Ltd.
immediate measures to save the physical quality of soil and water resources of the country. The Lake Tana Basin is one of the most affected area by soil erosion, sediment transport and land degradation. The land and water resources of the basin and the Lake Tana ecosystem are in danger due to the rapid growth of population, deforestation and overgrazing, soil erosion, sediment deposition, storage capacity reduction, drainage and water logging, flooding, pollutant transport, population pressure and overexploitation of specific fish species. The available land and water resources are not used effectively to improve the livelihood and socioeconomic conditions of the inhabitants. Sediments, organic and inorganic fertilizers from the agricultural fields that enter the lake by runoff may result in eutrophication. So far no effective measures have been taken to combat flooding, soil erosion and sedimentation problems. The lack of decision support tools and limitation of data concerning weather, hydrological, topographic, soil and land use are the factors that significantly hinder research and development in the area. To solve the existing soil erosion problems there is a need to identify the most erosion sensitive areas in the region, so that effective conservation measures can be taken. Appropriate tools are needed for the better assessment of the hydrology and soil erosion processes as well as decision support system for planning and implementations of appropriate measures. The tools involve various hydrological and soil erosion models as well as geographical information system (GIS). Many hydrological and soil erosion
S. G. SETEGN ET AL.
models are developed to describe the hydrology, erosion and sedimentation processes. Hydrological models are tools which describe the physical processes controlling the transformation of precipitation to runoff. Erosion modeling is based on understanding the physical laws of landscape processes that occur in the natural environment. There are different soil erosion models such as Chemicals, Runoff, and Erosion from Agricultural Management Systems (CREAMS) (Knisel, 1980), Erosion Productivity Impact Calculator (EPIC) (Williams et al., 1984), Agricultural Non-point Source Pollution Model (AGNPS) (Young et al., 1987, 1989), European Soil Erosion Model (EUROSEM) (Morgan et al., 1998) and Soil and Water Assessment Tool (SWAT) (Arnold et al., 1998). These models allow evaluation of management practices that influence factors contributing to erosion (Srinivasan and Engel, 1994). SWAT model is one of the appropriate watershed models for long-term impact analysis. It is widely applied in many parts of United States (Bingner, 1996; Arnold et al., 1998; Peterson and Hamlett, 1998; Srinivasan et al., 1998; Benaman et al., 2005; Neitsch et al., 2005) and many other countries (Heuvelmans et al., 2004; Bouraoui et al., 2005; Alamirew, 2006; Setegn et al., 2008). A comprehensive review of SWAT model applications is given by Gassman et al. (2007). Analysis of spatial information is becoming an emerging approach which is capable of acquiring, managing and analyzing complex problems of river basins and lake watersheds. In recent years, GIS has shown to be a good alternative to serve as a better decision support tool in the planning, management and implementation of soil and water resources. GIS is a very useful tool for storing, processing and manipulating and visualization of spatial databases. Consequently, the integration of multi-criteria evaluation (MCE) within a GIS context could help users to improve decision making processes. The main purpose of the MCE technique is to investigate a number of alternatives in the light of multiple criteria and conflicting objectives (Voogd, 1983). To carry out that, it
is necessary to generate compromise alternatives and a ranking of alternatives according to their degree of attractiveness (Janssen and Rietveld, 1990). In the last decade, MCE has received renewed attention in the context of a GIS-based decision making (Pereira and Duckstein, 1993; Heywood et al., 1995; Malczewski, 1996). Different studies have been conducted using MCE technique in the area of the natural resources management (e.g. Tecle and Yitayew, 1990; Ceballos-Silva and Lo’pez-Blanco, 2003; Leskinena and Kangas, 2005; Bello-Pineda et al., 2006; Hajkowicz and Higgins, 2006). In this study, MCE seems to be applicable to GIS-based spatial delineation of erosion vulnerability which helps to carry out the delineation of the most erosion prone area in the Lake Tana Basin. Generally, this article explains decision support system with MCEs and physically based SWAT model in identifying erosion hazard areas in the Lake Tana Basin. SWAT calculates the sediment yield within each hydrological response units (HRUs) and sub-basin. The GIS tool combines the slope, land cover, soil and river layers as the factors which contribute to soil erosion. Hence, the main goal of this study is to delineate the erosion vulnerable areas through physically based SWAT model and the MCE technique within a GIS context.
METHODS Study area The Lake Tana watershed, which is one of the subbasin of Blue Nile (Abbay River Basin Integrated Development Master Plans Project, 1999) River Basin, has a drainage area of 15 096 km2 . It is located in the country’s north–west highlands (latitude 12 00 N, longitude 37° 150 E) (Figure 1). This basin is one of the major basins that significantly contribute to the livelihoods of tens of millions of people in the lower Nile River basin. The mean annual rainfall of the catchment area is about 1280 mm. The mean annual actual evapotranspiration and water yield of the catchment area are estimated
Figure 1. Location map of the study area (Setegn et al., 2008) Copyright 2009 John Wiley & Sons, Ltd.
Hydrol. Process. (2009) DOI: 10.1002/hyp
SPATIAL DELINEATION OF SOIL EROSION PRONE AREAS
to be 773 mm and 392 mm, respectively (Setegn et al., 2008). It is rich in biodiversity with many endemic plant species and cattle breeds; it contains large areas of wetlands; and it is home to many endemic birds and cultural and archaeological sites. This basin is of critical national significance as it has great potentials for irrigation, hydroelectric power, high value crops and livestock production, ecotourism and others. The lake covers 3000–3600 km2 area at an elevation of 1800 m and with a maximum depth of 15 m. It is approximately 84 km long and 66 km wide. It is the largest lake in Ethiopia and the third largest in the Nile Basin. Gilgel Abay, Ribb, Gumera and Megech are the main rivers feeding the lake contributing more than 93% of the inflow. It is the main source of the Blue Nile River that is the only surface outflow for the Lake. The climate of the region is ‘tropical highland monsoon’ with main rainy season between June and September. The air temperature shows large diurnal but small seasonal changes with an annual average of 20 ° C. Major soils in the basin are Chromic Luvisols, Eutric Cambisols, Eutric Fluvisols, Eutric Leptosols, Eutric Regosols, Eutric Vertisols, Haplic Alisols, Haplic Luvisols, Haplic Nitisols and Lithic Leptosols (Figure 2a). The majority of the land area, 51Ð3%, of the Lake Tana Basin is used for agriculture, 29% is agropastoral area, 20% of the basin is covered by the lake water (Figure 2b) (Setegn et al., 2008). Models description In this study, two approaches were used to identify erosion prone areas to improve the decision making in planning and implementing soil and water conservation measures. The approaches involved the integration of the spatially distributed SWAT model, GIS and MCE. Description of SWAT model. SWAT is a river basin scale, continuous time and spatially distributed model developed to predict the impact of land management practices on water, sediment and agricultural chemical yields in large complex watersheds with varying soils, land use and management conditions over long periods of time (Arnold et al., 1998; Neitsch et al., 2005). The detail description of the model can be found in SWAT2005 theoretical document (Neitsch et al., 2005). As a physically based model, SWAT uses HRUs to
describe spatial heterogeneity in terms of land cover, soil type and slope within a watershed. SWAT simulates the hydrological cycle based on the water balance equation. SWt D SW0 C
t
Rday Qsurf Ea Wseep Qqw i
iD1
1 In which SWt is the final soil water content (mm), SW0 is the initial soil water content on day i (mm), t is the time (days), Rday is the amount of precipitation on day i (mm), Qsurf is the amount of surface runoff on day i (mm), Ea is the amount of evapotranspiration on day i (mm), Wseep is the amount of water entering the vadose zone from the soil profile on day i (mm) and Qgw is the amount of return flow on day i (mm). SWAT calculates the surface erosion caused by rainfall and runoff within each HRUs with the Modified Universal Soil Loss Equation (MUSLE) (Equation 2) (Williams, 1975). MUSLE is a modified version of the Universal Soil Loss Equation (USLE) developed by Wischmeier and Smith (1965, 1978). USLE predicts average annual gross erosion as a function of rainfall energy. In MUSLE, the rainfall energy factor is replaced with a runoff factor to simulate erosion and sediment yield. This improves the sediment yield prediction accuracy, eliminates the need for delivery ratios (the sediment yield at any point along the channel divided by the source erosion above that point) and single storm estimates of sediment yields can be calculated. Sediment yield prediction is improved because runoff is a function of antecedent moisture condition and rainfall energy. In MUSLE, the crop management factor is recalculated every day that runoff occurs. It is a function of above ground biomass, residue on the soil surface and the minimum C factor for the plant. sed D 11Ð8Qsurf ð qpeak ð areahru 0Ð56 ð KUSLE ð CUSLE ð PUSLE ð LSUSLE ð CFRG
2
where sed is the sediment yield on a given day (metric tons), Qsurf is the surface runoff volume (mm/ha), qpeak is the peak runoff rate (m3 /s), areahru is the area of the HRU (ha), KUSLE is the soil erodibility factor [0Ð013 metric ton m2 h/(m3 metric ton cm)], CUSLE is the cover and management factor, PUSLE is the support practice factor,
Figure 2. (a) Left—soil types (b) right—land cover maps of Lake Tana Basin Copyright 2009 John Wiley & Sons, Ltd.
Hydrol. Process. (2009) DOI: 10.1002/hyp
S. G. SETEGN ET AL.
LSUSLE is the topographic factor and CFRG is the coarse fragment factor. The details of the USLE factors and the descriptions of the different model components can be found in study of Neitsch et al. (2005). The hydrological model component estimates the runoff volume and peak runoff rate that are in turn used to calculate the runoff erosive energy variable. SWAT calculates the peak runoff rate with a modified rational method. The sediment routing model (Arnold et al., 1995) that simulates the sediment transport in the channel network consists of two components operating simultaneously: deposition and degradation. To determine the deposition and degradation processes the maximum concentration of sediment calculated by Equation 3 in the reach is compared with the concentration of sediment in the reach at the beginning of the time step. The maximum amount of sediment that can be transported from a reach segment is a function of the peak channel velocity and is calculated by Equation 3. spexp
concsed,ch,mx D Csp ð vch,pk
3
where concsed,ch,mx is the maximum concentration of sediment that can be transported by the water (ton/m3 or kg/l), Csp is a coefficient defined by the user, vch,pk is the peak channel velocity (m/s) and spexp is exponent parameter for calculating sediment reentrained in channel sediment routing that is defined by the user. It normally varies between 1Ð0 and 2Ð0. The maximum concentration of sediment calculated in Equation 3 in the reach is compared with the concentration of sediment in the reach at the beginning of the time step, concsed,ch,i . If concsed,ch,i > concsed,ch,mx , deposition is the dominant process in the reach segment and the net amount of sediment deposited is calculated from Equation 4. seddep D concsed,ch,i concsed,ch,mx Ð Vch
4
where seddep is the amount of sediment deposited in the reach segment (metric tons), concsed,ch,i is the initial sediment concentration in the reach (kg/l or ton/m3 ), concsed,ch,mx is the maximum concentration of sediment that can be transported by the water (kg/l or ton/m3 ) and Vch is the volume of water in the reach segment (m3 ).
If concsed,ch,i < concsed,ch,mx , degradation is the dominant process in the reach segment and the net amount of sediment reentrained is calculated as Equation 5. sed0 deg D concsed,ch,mx concsed,ch,i Ð Vch Ð Kch Ð CCH 5 where seddeg is the amount of sediment reentrained in the reach segment (metric tons), concsed,ch,mx is the maximum concentration of sediment that can be transported by the water (kg/l or ton/m3 ), concsed,ch,i is the initial sediment concentration in the reach (kg/l or ton/m3 ), Vch is the volume of water in the reach segment (m3 ), KCH is the channel erodibility factor (cm/h/Pa) and CCH is the channel cover factor. The final amount of sediment in the reach is determined from Equation 6. sedch D sedch,i seddep C seddeg
6
where sedch is the amount of suspended sediment in the reach (metric tons), sedch,i is the amount of suspended sediment in the reach at the beginning of the time period (metric tons) and seddep is the amount of sediment reentrained in the reach segment (metric tons). The amount of sediment transported out of the reach is calculated by Equation 7. sedout D sedch Ð
Vout Vch
7
where sedout is the amount of sediment transported out of the reach (metric tons), sedch is the amount of suspended sediment in the reach (metric tons), Vout is the volume of outflow during the time step (m3 ) and Vch is the volume of water in the reach segment (m3 ). Model input. Digital elevation model: A 90 m by 90 m resolution Digital Elevation Model (DEM) for Blue Nile Basin (Figure 3) was downloaded from Shuttle Radar Topography Mission (SRTM) website (Jarvis et al., 2006). A 2 m by 2 m resolution DEM for Anjeni watershed was also obtained from soil and water conservation programme (SCRP), University of Bern, Switzerland. The DEM was used to delineate the watersheds and to analyze the drainage patterns of the land surface terrain.
Figure 3. Digital Elevation Model (DEM) of the Lake Tana Basin (meter above sea level) Copyright 2009 John Wiley & Sons, Ltd.
Hydrol. Process. (2009) DOI: 10.1002/hyp
SPATIAL DELINEATION OF SOIL EROSION PRONE AREAS
Land use and soil data: Land use is one of the most important factors that affect runoff, evapotranspiration and surface erosion in a watershed. The land use and soil map of the study area was obtained from ministry of water resources Ethiopia and SCRP, University of Bern, Switzerland. Figure 2b shows that more than 50% of the Lake Tana watershed is used for agriculture. SWAT model requires different soil textural and physicochemical properties such as soil texture, available water content, hydraulic conductivity, bulk density and organic carbon content for different layers of each soil type. These data were obtained from different sources (Setegn et al., 2008). Weather data and river discharge: In this study, the weather variables used for driving the hydrological balance are daily precipitation, minimum and maximum air temperature for the period of 1978–2004. These data were obtained from Ethiopian National Meteorological Agency for stations located within and around the watershed. Weather data were also obtained from SCRP project office Addis Ababa, Ethiopia (SCRP, 2000). Daily river discharge values for Ribb, Gumera, Gilgel Abay, Megech rivers and the outflow river Blue Nile (Abbay) were obtained from the Hydrology Department of the Ministry of Water Resources of Ethiopia. These daily river discharges at four tributaries of Lake Tana were used for model calibration (1981–1992) and validation (1993–2004). Ten years precipitation, air temperature, river discharge and sediment measurements on Minchet River were used for the simulation of the stream flow and sediment yield in Anjeni gauged watershed. The period from 1984 to 1988 was used for calibration of the model. Whereas the period from 1989 to 1993 was used for validation of the SWAT model. Description of multi-criteria evaluation GIS-based analysis of spatial data is capable of analyzing complex problem of evaluating and allocating natural resources for targeting potential or sensitive areas. MCE model (under IDRISI GIS environment) is a method for decision support where a number of different criteria are combined to meet one or several objectives (Voogd, 1983; Carver, 1991). An objective is thus a perspective that serves to guide the structuring of decision rules, which is the procedure whereby criteria are selected and combined to arrive at a particular evaluation, and evaluations are compared and acted upon. Many GIS software systems provide the basic tools for evaluating such a model. For this study, GIS software called IDRISI which has an MCE module was used. A detailed description of the method can be found in IDRISI32 Guide to GIS and Image Processing (Eastman, 2001). The GIS tool combines the slope, land use and soil layers as a major factor, which contributes for soil erosion and sediment transport. The GIS-based MCE procedures involve a set of geographically defined alternatives and a set of evaluation criteria represented as map layers. The problem is to Copyright 2009 John Wiley & Sons, Ltd.
combine the criterion maps according to the criterion (attribute) values and decision maker’s preferences using a decision rule (combination rule). The primary issue in MCE is concerned with how to combine the information from several criteria to form a single index of evaluation. In the case of Boolean criteria (constraints), the solution usually lies in the union (logical OR) or intersection (logical AND) of conditions. However, for continuous factors, a weighted linear combination (WLC) (Voogd, 1983) is most commonly used. With a WLC, factors are combined by applying a weight to each followed by a summation of the results to yield a suitability map, i.e. SD wi xi 8 where S stands for suitability, wi for weight of factor i and xi for criterion score of factor i. This procedure is not unfamiliar in GIS and has a form very similar to the nature of a regression equation. In cases where Boolean constraints also apply, the procedure can be modified by multiplying the suitability calculated from the factors by the product of the constraints, i.e. SD wi xi ð cj 9 where cj is the criterion score of constraint j and D product. Because of the different scales upon which criteria are measured, it is necessary that factors be standardized before combination using the formulas above, and that they be transformed, if necessary, such that all factors maps are positively correlated with suitability. Voogd (1983) reviews a variety of procedures for standardization, typically using the minimum and maximum values as scaling points. The simplest is a linear scaling such as: xi D Ri Rmin /Rmax Rmin ð standardized range 10 where R is the row score. However, if the continuous factors are really fuzzy sets, we easily recognize this as just one of many possible set membership functions. In IDRISI, the module named FUZZY is provided for the standardization of factors using a whole range of fuzzy set membership functions. The module provides the option of standardizing factors to either a 0–1 real number scale or a 0–255 byte scale. The latter option is recommended because the MCE module has been optimized for speed using a 0–255 level standardization. Importantly, the higher value of the standardized scale must represent the case of being more likely to belong to the decision set. A critical issue in the standardization of factors is the choice of the end points at which set membership reaches either 0Ð0 or 1Ð0 (or 0 and 255). Breaking the information down into simple pairwise comparisons in which only two criteria need to be considered at a time can greatly facilitate the weighting process and will likely produce a more robust set of criteria weights. A pairwise comparison method has the Hydrol. Process. (2009) DOI: 10.1002/hyp
S. G. SETEGN ET AL.
added advantages of providing an organized structure for group discussions and helping the decision making group hone in an areas of agreement and disagreement in setting criterion weights. The technique implemented in IDRISI is that of pairwise comparisons developed by Saaty (1977) which is known as the Analytical Hierarchy Process (AHP). In the procedure for an MCE using a WLC, it is necessary that the weights sum to one. In Saaty’s technique, weights of this nature can be derived by taking the principal eigenvector of a square reciprocal matrix of pairwise comparisons between the criteria. The comparisons concern the relative importance of the two criteria involved in determining suitability for the stated objective. Ratings are provided on a nine-point continuous scale. Calibration, validation and application of SWAT model The SWAT model was calibrated and validated for flow in Lake Tana Basin on a daily basis and for flow and sediment yield in Anjeni gauged watershed on a monthly basis. Before the calibration exercise was implanted, 26 hydrological parameters were tested for sensitivity analysis for the simulation of the stream flow in the study area. The details of all hydrological parameters are found in the ArcSWAT interface for SWAT user’s manual (Neitsch, et al., 2004; Winchell et al., 2007). Both manual and autocalibration methods were implemented for minimizing the difference between measured and predicted flow and sediment yield. In this study, Sequential Uncertainty Fitting (SUFI-2) (Abbaspour et al., 2004, 2007) calibration and uncertainty analysis method was used for autocalibration of the flow and sediment parameters. In SUFI2, Latin hypercube sampling is used to draw independent parameter sets (Abbaspour et al., 2007). The goodness of fit was quantified by the coefficient of determination (R2 ) and Nash–Sutcliff coefficient (NSE) (Nash and Sutcliffe, 1970) between the observations and the final best simulation. This efficiency is commonly used quantitative measure of hydrograph prediction performance that helps to evaluate between the predicted and observed flow as well as observed and predicted sediment yield. NSE is calculated using Equation 11. n
2
Oi Pi
iD1
NSE D 1 n
2
11
Oi O
iD1
where NSE is the prediction efficiency, Oi is the observed condition at time i, Oi is the mean of the observed values over all times, Pi is the predicted value at time I and P is the mean predicted value over all times. The index i refer to storm number for calculating the prediction efficiencies for sediment yield, and refer to time during the storm for calculating the efficiency of a hydrograph for a particular storm. Copyright 2009 John Wiley & Sons, Ltd.
Root mean square error (RMSE) observations standard deviation ratio (RSR): RMSE is one of the commonly used error index statistics (Chu and Shirmohammadi 2004; Singh et al., 2005; Moriasi et al., 2007). RSR standardizes RMSE using the observations standard deviation, and it combines an error index (Moriasi et al., 2007). RSR is calculated as the ratio of the RMSE and standard deviation of measured data, as shown in Equation 12. n obs Qi Qisim 2 iD1 RMSE 12 RSR D D STDEVobs n Qiobs Qimean 2 iD1
Percent bias (PBIAS): PBIAS measures the average tendency of the simulated data to be larger or smaller than their observed counterparts (Gupta et al., 1999). PBIAS is calculated with Equation 13. n obs sim Q Q ð 100 i i iD1 PIBIAS D 13 n obs Qi iD1
In which PBIAS is the deviation of data being evaluated, expressed as a percentage. After setting up of the model, the default simulations of stream flow, using the default parameter values, were carried out in the Lake Tana Basin for the calibration period (1978–1992). An independent precipitation, temperature and stream flow dataset (1993–2004) was used for validation of the model in the four river basins. Periods 1978–1980 and 1990–1992 were used as ‘warm-up’ periods for calibration and validation purposes, respectively. The warm-up period allows the model to get the hydrologic cycle fully operational. The sediment yield was simulated from 1992 to 2004 in the Lake Tana Basin. SWAT calculates the soil erosion and sediment yield within each HRUs in each sub-basin within the watershed. The model gives the magnitude of sediment yield in each sub-basin so that the rate of soil erosion within each sub-basin can be understood. First, the calculated sediment yield was converted from SWAT project file into VIZSWAT (a visualization and analysis tool developed by Baird & Associates for SWAT model output). VIZSWAT is a customized version of Spatial Data Analyzer, a GIS-based data visualization and analysis tool that animates time series and spatial data over GIS maps with impressive display speed. The VIZSWAT aggregated the simulated sediment yield into annual average. Second, the aggregated annual sediment yield was taken to ArcGIS and added as a separate field to the watershed attributable to produce a map. Finally, the map was reclassified into four erosion categories as low, medium, high and very high erosion potential. This shows the areas in the watershed which produce high and low annual sediment yields. Hydrol. Process. (2009) DOI: 10.1002/hyp
SPATIAL DELINEATION OF SOIL EROSION PRONE AREAS
Application of MCE In this approach, the selection of potential soil erosion prone areas through integration of various of GIS layers, spatial analysis and multi-criteria based evaluation is presented. The decision was made after combination of four criteria (factor maps) using MCE decision wizard. The first factor considered is slope factor. Steeper and longer slopes result in high erosion rates. The second criterion is the land cover which controls the detachability and transport of soil particles and infiltration of water into the soil. Soil types also play significant role in erosion and sediment transport process depending upon their physical properties and sensitivity to erosion. A layer which contains all rivers within the catchments was also considered as a contributing factor in the study. It was assumed that places close to streams or rivers are more easily washed especially during high flow seasons. Development of factor maps. Slope factor map: The DEM was first imported from ArcGIS to IDRISI software. The slope map was generated from the DEM using surface analysis tool. The decision wizard requires the criteria maps in raster form. First, a raster image initialization layer was created by defining the spatial parameters. The vector form slope map was converted to raster map using the initialized image layer and a raster conversion module. The raster map consists of the slope class from 0 to 205Ð57%. This slope range was reclassified to six major slope classes depending on the Food and Agriculture Organization (FAO) slope classification (Table I). Each slope category was given an index for their sensitivity to erosion (Table I). The sensitivity index was given values from 0 to 20. This value is relative value with respect to each subfactor. Zero value means less sensitive and 20 means extremely sensitive. Land cover factor map: The land cover map was stored in shape file format and imported from ArcGIS to IDRISI. It was in vector form and converted to raster form (in the same way as for the slope map). The major land cover types were originally subclassified into 151 classes depending upon the specific type of cover, the type of crops and degree of cultivation. The 151 classes of cover type were reclassified into 12 main land cover types. The 12 classes of land cover types were in turn reclassified as per its sensitivity to erosion. Values were set from 0 to 20 to categorize the factor map into common sensitivity Table I. FAO slope categories and assigned sensitivity index New class 1 2 3 4 5 6
Slope categories (%)
Characteristics
Sensitivity index
0–2 2–5 5–10 10–15 15–30 >30
Flat or almost flat Gently undulating Undulating Rolling Moderately steep Steep
0 4 8 12 16 20
Copyright 2009 John Wiley & Sons, Ltd.
index. For instance the highest value of 20 was assigned to the most sensitive land cover type and zero to less sensitive one (Table II). Soil factor map: The soil layer were imported from shape file to IDRISI format and then converted into raster. Originally the soil layer had 200 subclasses and they were reclassified to 12 major soil types. Each soil type was assigned values from 0 to 20 depending upon their degree of sensitivity to soil erosion. The sensitivity of the soil to erosion was based on the soil physical characteristics (texture and structure). These characteristics are also studied by different organization and their characteristics on erosion vulnerability were listed in Abay master plan study (Table III.). Rivers factor map: The GIS layer which consists of all rivers in the Lake Tana watershed was imported to IDRISI. For this river layer, a distance analysis was performed so that the decision wizard could consider the distance from the river with a certain value depending upon the criteria set. In this case, it is considered that the area nearest to the river may have the chance to be easily washed because of the high flood current especially during high flow seasons. Table II. Land cover type in the Lake Tana catchments Land cover
Afro Alpine Dominantly cultivated Moderately cultivated Forest Grassland Water body Swamp Plantations Shrub land Urban Woodland open
Area (km2 )
Area (%)
Sensitivity index assigned
100Ð04 7732Ð79 3364Ð88 13Ð11 424Ð96 3041Ð42 19Ð82 8Ð93 429Ð40 27Ð36 11Ð18
0Ð7 51Ð0 22Ð2 0Ð1 2Ð8 1 0Ð1 0Ð1 2Ð8 0Ð2 0Ð1
1 20 15 2 1 0 3 5 7 10 5
Table III. Major soil types in the Lake Tana Catchments (source for soil map: Ethiopian Ministry of water resources) No.
Soil type
Area (km2 )
Area (%)
Sensitivity index assigned
1 2 3 4 5 6 7 8 9 10 11 12
Chromic Luvisols Eutric Cambisols Eutric Fluvisols Eutric Leptosols Eutric Regosols Eutric Vertisols Haplic Alisols Haplic Luvisols Haplic Nitisols Lithic Leptosols Urban Water
4240Ð9 2Ð1 1847Ð6 4913Ð0 44Ð9 2064Ð8 1498Ð8 4507Ð3 544Ð4 456Ð6 10Ð5 3042Ð6
18Ð3 0Ð1 8Ð0 21Ð2 0Ð2 8Ð9 6Ð5 19Ð4 2Ð3 2Ð0 0Ð0 13Ð1
6 4 8 20 4 4 12 6 12 2 2 0
Hydrol. Process. (2009) DOI: 10.1002/hyp
S. G. SETEGN ET AL.
Integration of different layers. Once the criteria maps (factors and constraints) have been developed, an evaluation or aggregation stage is undertaken to combine the information from the various factors and constraints. The MCE module offers three logics for the evaluation/aggregation of multiple criteria: Boolean intersection, WLC, and the ordered weighted average. The most simplistic type of aggregation is the Boolean intersection or logical AND. This method is used only when factor maps have been strictly classified into Boolean suitable/unsuitable images with values 1 and 0. The evaluation is simply the multiplication of all the images. However, for continuous factors, a WLC (Voogd, 1983) is a usual technique. The details of the different logic for the MCE are described in IDRISI32 help tutorial manual (Eastman, 2001). Before combining the four factor maps they were first standardized to a scale of 0–255, where 0 is not sensitive and 255 is more sensitive. The wizard facilitates standardization of quantitative factor images using the module FUZZY. The membership function shape and type and set control points from 0 to 20 were determined for each factor maps. Factor weights are assigned to specify the relative importance of each one in determining the aggregate output value using AHP. The comparisons concern the relative importance of the two criteria involved in determining suitability for the stated objective. In AHP analysis, ratings are provided on a nine-point continuous scale (Table IV). The factor weights were derived by taking the principal eigenvector of a square reciprocal matrix of pairwise comparisons between the criteria for final analysis (Table V). In this study, six scenarios were adapted depending upon the priority given to the factors. For instance, in the first scenario slope factor was given the highest priority followed by land cover, soil and river maps. As the complete pairwise comparison matrix contains multiple paths, it is also possible to determine the degree of consistency that has been used in developing the ratings. The consistency ratio (CR) indicates the probability that the matrix ratings are randomly generated. Saaty (1977) suggested that matrices with CR ratings greater than 0Ð10 should be reevaluated. After completing the whole process, the final MCE erosion potential area map was produced which show the degree of erosion sensitivity of each area. The final map was ranked from 0 to 255 scales and it was reclassified into four categories (nil, slight, moderate and severe) depending upon the combined degree of sensitivity to erosion. The reclassified map was exported from IDRISI to ArcGIS for visualization and mapping.
Table V. A pairwise comparison matrix for assessing the comparative importance of factors to identify erosion sensitive areas Scenario 1 Criteria
Slope
Land cover
Soil
Rivers
1 1/3 1/5 1/9
1 1/3 1/5
1 1/3
1
Slope Land cover Soil Rivers
RESULTS AND DISCUSSION SWAT model calibration and application The Lake Tana Basin was divided into 34 sub-basins and 284 HRUs. Surface runoff volume and peak runoff rate (m3 /s) are the flow components which determine the rate of soil erosion and sediment yield. First, the model was calibrated for flow in the Lake Tana Basin. The sediment parameters were generated after calibrating the SWAT model in Anjeni gauged watershed which is located very near to the basin in a similar topography and agroclimatic zone. The parameter sensitivity analysis was carried out using the ArcSWAT interface for the whole Lake Tana Basin and Anjeni watershed. Twenty-six hydrological parameters were tested for sensitivity analysis for the simulation of the stream flow in the study area. The most sensitive parameters considered for calibration in both watersheds were soil evaporation compensation factor, initial SCS Curve Number II value, base flow ˛ factor (days), threshold depth of water in the shallow aquifer for ‘revap’ to occur (mm), (days), available water capacity (mm WATER/mm soil), groundwater ‘revap’ coefficient, channel effective hydraulic conductivity (mm/h) and threshold depth of water in the shallow aquifer for return flow to occur (mm). Table VI shows sensitive parameters for flow and calibrated flow parameters values for both Anjeni gauged watershed and Lake Tana Basin. Whereas the most sensitive parameters for predictions of sediment yield are linear parameter for calculating the maximum amount of sediment that can be entrained during channel sediment routing, channel cover factor, USLE equation support practice factor, exponent parameter for calculating sediment reentrained in channel sediment routing and minimum value of USLE C factor for land cover/plant. These sediment parameters are listed in Table VII with their calibrated values. SWAT model was first calibrated and validated for flow in Anjeni watershed. The model was calibrated for sediment after calibrating the flow parameters in the
Table IV. The continuous rating scale (Eastman, 2001) Rating scale 1/9 Extremely
1/7 1/5 Very strongly Strongly Less important
Copyright 2009 John Wiley & Sons, Ltd.
1/3 Moderately
1 Equally
3 Moderately
5 7 Very strongly Strongly More important
9 Extremely
Hydrol. Process. (2009) DOI: 10.1002/hyp
SPATIAL DELINEATION OF SOIL EROSION PRONE AREAS
Table VI. Sensitive parameters for flow and calibrated flow parameters values for both Anjeni gauged watershed and Lake Tana Basin No.
2 1 3 4 5 6 7 8 9 10 11
Sensitive flow parameters
CN2 ESCO ALPHA BF REVAPMN SOL AWC GW REVAP CH K2 GWQMN Gw Delay Gw Revap Sol Z
Lower and upper bound š25%a 0–1 0–1 0–500 š25%a š0Ð036 0–5 0–5000 0–500 š0Ð036 š25
Final fitted value Gilgel Abay River
Megech River
Ribb River
Gumera River
Anjeni watershed
10 0Ð8 0Ð1 300 20 20 4Ð6 108
9 0Ð8 0Ð1 289 20 0Ð1 3Ð2 17
10 0Ð8 0Ð1 372 10 0 1Ð9 333
8 0Ð8 0 446 20 0Ð1 0Ð7 98
5 0Ð75 0Ð048 50 0Ð717 28 0Ð087 C21%Ł
a The percentage with which the original values changed. ESCO-Soil evaporation compensation factor, Sol Awc-Available water capacity (mm WATER/mm soil), Gw Delay-Groundwater delay (days), Gw Revap-Groundwater ‘revap’ coefficient, Ch K2-Channel effective hydraulic conductivity (mm/h), Sol Z-Soil depth (mm), REVAPMN-Threshold depth of water in the shallow aquifer for ‘revap’ or percolation to the deep aquifer to occur (mm, H2 O), GWQMN-Threshild depth of water in the shallow aquifer required for return flow to occur (mm, H2 O), Sol Z- Soil depth (mm).
Table VII. Sensitive and calibrated sediment parameters and for Anjeni gauged watershed Parameter
Linear parameter for calculating the maximum amount of sediment that can be reentrained during channel sediment routing (Spcon) Channel cover factor (Ch Cov) Channel erodibility factor (Ch Erod) USLE equation support practice factor (USLE P) Exponent parameter for calculating sediment reentrained in channel sediment routing (Spexp) Minimum value of USLE C factor for land cover/plant (USLE C)
Lower and upper bound 0Ð0001–0Ð01
Rank based on relative sensitivity
Calibrated value
5Ð09
0Ð005
Table VIII. Model evaluation statistics for calibration and validation result for flow in the Lake Tana Basin (Setegn et al., 2008) Objective function
Rivers Gilgel Abay
NSE R2
Gumera
Megech
Ribb
Cal
Val
Cal
Val
Cal
Val
Cal
Val
0Ð73 0Ð75
0Ð69 0Ð80
0Ð62 0Ð69
0Ð60 0Ð70
0Ð18 0Ð19
0Ð04 0Ð32
0Ð51 0Ð59
0Ð48 0Ð55
Cal, calibration; Val, validation; NSE, Nash–Sutcliff coefficient; R2 , coefficient of determination.
0–1Ð00
4Ð08
0Ð35
0–1Ð00
3Ð12
0Ð50
0–1Ð00
1Ð44
0Ð8
1–2Ð00
1Ð06
1Ð39
š25
0Ð09
0Ð27
USLE, universal soil loss equation.
watershed. The main reason not able to calibrate sediment parameters in the Lake Tana Basin is that there is no measured sediment data with in Lake Tana Basin. The comparison between the flow parameters calibrated both at Anjeni and Lake Tana Basin indicated a reasonable similarity between the parameters in the two basins. In both areas the most sensitive parameters for flow and Copyright 2009 John Wiley & Sons, Ltd.
sediment are similar which indicate that the hydrological response to flow and sediment as a result of land use, soil and topographic characteristics are related, so that we were able to upscale the sediment parameters to the Lake Tana Basin. The comparison between the observed and simulated flow discharge values for 12 years of simulations indicated that there is a good agreement between the observed and simulated flows using SUFI-2 algorithms, which were verified by higher values of R2 and NSE for Gilgel Abay, Gumera and Ribb inflow rivers. Calibrated and validated model predictive flow performance for all Lake Tana inflow rivers on daily flows is summarized in Table VIII. The monthly calibration and validation of the SWAT model for flow and sediment yield in Anjeni watershed have shown that the model can predict the flow and sediment yield as well as indicated by model performance evaluation measures, the, R2 the Nash–Suttcliffe simulation efficiency, PBIAS and standardized RMSE. The statistical comparison between the measured monthly sediment yield and best simulation result from SUFI2 algorithms showed a good agreement. The result was verified by NSE D 0Ð81, PBIAS D 28%, RSR D Hydrol. Process. (2009) DOI: 10.1002/hyp
S. G. SETEGN ET AL.
0Ð23 and R2 D 0Ð85 for calibration and NSE D 0Ð79, PBIAS D 30%, RSR D 0Ð29 and R2 D 0Ð80 for validation periods. Both the NSE and RSR results show good result both for calibration and validation periods. The PBIAS values are good for both periods. The R2 statistics also show a good correlation between measured and simulated sediment yields. The simulated sediment yield output of the SWAT model has shown that 18Ð4% of the watershed area has high potential for soil erosion (Table X) which produces an average annual sediment yield of 30 to 65 tons/ha. On the basis of the classes assigned to the annual sediment yield, the map was reclassified into four major categories of soil erosion hazards region such as very low, low, moderate and severe erosion conditions (Figure 4). The result of the SWAT output indicated that significant portions of the area which are known to be highly cultivated area are more vulnerable to soil erosion. Moreover, areas at a higher slope condition have shown higher contribution of sediment yield. Some parts of the watershed which have higher erodibility characteristics because of poor soil physical properties contributed for a higher sediment yield than others. Many of the places which are very near to rivers and stream has shown a considerable contribution for higher soil erosion and sediment yield. The Ribb and Gumera inflow rivers shown to contribute large amount of sediment yield production to the Lake Tana with respect to their size.
MCE analysis In the MCE analysis, six scenarios were tested to find the main factors which play a major role for the rate of soil erosion in the watershed. In the first and second scenarios, the main consideration was given for slope factor followed by land cover and soil, respectively. In all cases river factor networks was given the lowest priority in comparison with others (Table IX). It was assumed that the position of rivers in the watershed play less role for the rate of soil erosion than the other factors. On the basis of the foregoing assumptions, the weights were derived by entering the ratings into a pairwise comparison matrix for each scenario. The pairwise comparison matrix for scenario one indicates that the rating of land cover factor relative to slope gradient is 1 : 3. Soil is less important than slope, river factor is less important than slope factor and so on. The computed CR is 0Ð03 which is within the acceptable range ( 0Ð50. The SWAT model showed that 18Ð5% of the watershed is high erosion potential areas. The MCE result for scenarios that gives high priority to land cover and slope factor, indicated 28% and 25% of the land area are vulnerable to soil erosion, respectively. The comparison of the maps produced by the two approaches showed a considerable similarity indicating areas with possible erosion risk. The result indicated that land use factor play a significant factor in the rate of soil erosion and land degradation. The output of this study may support planners and decision makers to take relevant soil and water conservation measures and thereby reduce the alarming soil loss and land degradation problems in the basin.
ACKNOWLEDGEMENTS
The authors would like to thank the Applied Training Project of the Nile Basin Initiative for the financial support of this research. They extend thanks to Meteorological Agency of Ethiopia and Ministry of Water Resources for the data used in the study. They also acknowledge the Associate Editor and the two anonymous reviewers for their constructive comments and suggestions.
REFERENCES Abbaspour KC, Johnson CA, van Genuchten MT. 2004. Estimating uncertain flow and transport parameters using a sequential uncertainty fitting procedure. Vadose Zone Journal 3(4): 1340– 1352. Abbaspour KC, Yang J, Maximov I, Siber R, Bogner K, Mieleitner J, Zobrist J, Srinivasan R. 2007. Modelling hydrology and water quality in the pre-alpine/alpine Thur watershed using SWAT. Journal of Hydrology 333: 413– 430. Abbay River Basin Integrated Development Master Plans Project. 1999. Volume IX—Land Resources Development, Part II—Semi-Detailed Soils Survey, May 1999. Alamirew, CD. 2006. Modeling of hydrology and soil erosion in upper Awash river basin. PhD thesis, University of Bonn, Institut f¨ur St¨adtebau, Bodenordnung und Kulturtechnik, Garmany; 235p. Arnold JG, Srinivason R, Muttiah R, Williams JR. 1998. Large area hydrologic modeling and assessment part I: model development. Journal of the American water Resources Association 34(1): 73–89. Copyright 2009 John Wiley & Sons, Ltd.
Arnold JG, Williams JR, Maidment DR. 1995. Continuous-time water and sediment routing model for large basins. Journal of Hydraulic Engineering 121(2): 171–183. Bello-Pineda J, Ponce-Hern´andez R, Liceaga-Correa MA. 2006. Incorporating GIS and MCE for suitability assessment modelling of coral reef resources. Environmental Monitoring and Assessment 114(1– 3): 225–256. Benaman J, Christine AS, Douglas AH. 2005. Calibration and validation of soil and water assessment tool on an agricultural watershed in upstate New York. Journal of Hydrologic Engineering, ASCE 10(5): 363–374. Bingner RL. 1996. Runoff simulated from Goodwin Creek watershed using SWAT. Transactions of the American Society of Agricultural Engineers 39(1): 85–90. Bouraoui F, Benabdallah S, Jrad A, Bidoglio G. 2005. Application of the SWAT model on the Medjerda river Basin (Tunisia). Physics and Chemistry of the Earth 30: 497– 507. Carver SJ. 1991. Integrating multi-criteria evaluation with geographical information systems. International Journal of Geographical Information Systems 5(3): 321–339. Ceballos-Silva A, Lo’pez-Blanco J. 2003. Delineation of suitable areas for crops using a multi-criteria evaluation approach and land use/cover mapping: a case study in Central Mexico. Agricultural Systems 77: 117–136. Chu TW, Shirmohammadi A. 2004. Evaluation of the SWAT model’s hydrology component in the piedmont physiographic region of Maryland. Transactions of the ASAE 47(4): 1057– 1073. Eastman JR. 2001. Idrisi32—release 2. Guide to GIS and Image Processing, vol. 2. Clark Labs, Clark University: Worcester, MA. Ethiopian Highlands Reclamation Study (EHRS). 1984. Annual Research Report (1983–984). Ministry of Agriculture: Addis Ababa. Gassman PW, Reyes MR, Green CH, Arnold JG. 2007. The Soil and Water Assessment Tool: historical development, applications, and future research directions. Transactions of the ASABE 50(4): 1211– 1250. Gupta HV, Sorooshian S, Yapo PO. 1999. Status of automatic calibration for hydrologic models: comparison with multilevel expert calibration. Journal of Hydrologic Engineering 4(2): 135–143. Hajkowicz S, Higgins A. 2006. A comparisons of multiple criteria analysis techniques for water resource management. European Journal of Operational Research 184(1): 255– 265. Heuvelmans G, Muys B, Feyen J. 2004. Analysis of the spatial variation in the parameters of the SWAT model with application in Flanders, Northern Belgium. Hydrology and Earth Systems Sciences 8(5): 931–939. Heywood I, Oliver J, Tomlinson S. 1995. Building an exploratory multi-criteria modeling environment for spatial decision support. In Innovations of GIS 2 , Fisher P, (ed). Taylor and Francis: Leicester; 127–136. Hurni H. 1989. Soil for the Future. Environmental Research for Development Cooperation, Uni Press 62, University of Berne: Berne; 42–46. Janssen R, Rietveld P. 1990. Multi-criteria Analysis and GIS: an application to agriculture land use in The Netherlands. In Geographical Information Systems for Urban and Regional Planning, Scholten H, Stilwell J. (eds.) Kluwer Academic Press: Dordrecht, The Netherlands; 129–138. Jarvis A, Reuter HI, Nelson A, Guevara E. 2006. Hole-Filled Seamless SRTM data V3 , International Centre for Tropical Agriculture (CIAT), available from http://srtm.csi.cgiar.org. Knisel WG. 1980. CREAM: A Field Scale Model for Chemicals, Runoff, and Erosion from Agricultural Management Systems. USDA, Conservation Research Report No. 26, 643pp. Kr¨uger HJ, Berhanu F, Yohannes GM, Kefane K. 1996. Creating an inventory of indigenous soil and water conservation measures in Ethiopia. In Sustaining the Soil: Indigenous Soil and Water Conservation in Africa, Reij C, Scoones I, Toulmin C (eds). EarthScan Publications: London; 170– 180. Leskinena P, Kangas J. 2005. Multi-criteria natural resource management with preferentially dependent decision criteria. Journal of Environmental Management 77(3): 244–251. Malczewski JA. 1996. GIS-based approach to multiple criteria group decision-making. International Journal of Geographical Information Science 10(8): 321–339. Morgan RPC, Quinton JN, Smith RE, Govers G, Poesen JWA, Auerswald K, Chisci G, Torri D, Styczen ME. 1998. The European Soil Erosion Model (EUROSEM): a dynamic approach for predicting sediment transport from fields and small catchments. Earth Surface Processes and Landforms 23(6): 527– 544. Hydrol. Process. (2009) DOI: 10.1002/hyp
SPATIAL DELINEATION OF SOIL EROSION PRONE AREAS Moriasi DN, Arnold JG, Van Liew MW, Binger RL, Harmel RD, Veith T. 2007. Model evaluation guidelines for systematic quantification of accuracy in watershed simulations. Transactions of the ASABE 50(3): 885– 900. Nash JE, Sutcliffe JV. 1970. River flow forecasting through conceptual models. Part I—a discussion of principles. Journal of Hydrology 10: 282– 290. Neitsch SL, Arnold JG, Kiniry JR, Williams JR, King KW. 2004. Soil and Water Assessment Tool, User’s Manual: Version 2005 . USDA Agricultural Research Service and Texas A&M Blackland Research Center: Temple. Neitsch SL, Arnold JG, Kiniry JR, Srinivasan R, Williams JR. 2005. Soil and Water Assessment Tool, Theoretical Documentation: Version 2005 . USDA Agricultural Research Service and Texas A&M Blackland Research Center: Temple. Pereira JMC, Duckstein L. 1993. A multiple criteria decision-making approach to GIS-based land suitability evaluation. International Journal of Geographical Information Systems 75: 407–424. Peterson JR, Hamlett JM. 1998. Hydrologic calibration of the SWAT model in a watershed containing Fragipan Soils. Journal of the American Water Resources Association 34(3): 531–544. Saaty TL. 1977. A scaling method for priorities in hierarchical structures. Journal of Mathematical Psychology 15: 234– 281. SCRP (Soil Conservation Research Programme). 1996. Soil Conservation Research Project Database Report 1982–1993 . Ministry of Agriculture and University of Berne, Series Report III. Hundelafto Research Unit, Institute of Geography, University of Berne: Switzerland. SCRP (Soil Conservation Research Program). 2000. Soil Erosion and Conservation Database. Area of Anjeni, Gojam, Ethiopia: Long-Term Monitoring of the Agricultural Environment, 1984– 1994. Centre for Development and Environment in association with the Ministry of Agriculture, Ethiopia: Berne, Switzerland. Setegn SG, Srinivasan R, Dargahi B. 2008. Hydrological modeling in the Lake Tana Basin, Ethiopia using SWAT model. The Open Hydrology Journal 2: 49–62. Singh J, Knapp HV, Demissie M. 2005. Hydrologic modeling of the Iroquois River watershed using HSPF and SWAT. Journal of the American Water Resources Association 41(2): 343–360.
Copyright 2009 John Wiley & Sons, Ltd.
Srinivasan R, Engel BA. 1994. A spatial decision support system for assessing agricultural non point source pollution. Journal of American Water Resources Association 30(3): 441–452. Srinivasan R, Ramanarayanan TS, Arnold JG, Bednarz ST. 1998. Large area hydrologic modeling and assessment part II: model application. Journal of the American Water Resources Association 34(1): 91–101. Tecle A, Yitayew M. 1990. Preference ranking of alternative irrigation technologies via a multicriterion decision making procedure. Transactions of ASAE 3(5): 1509– 1517. Voogd H. 1983. Multi-Criteria Evaluation for Urban and Regional Planning. Pion, Ltd.: London. Williams JR. 1975. Sediment-yield prediction with universal equation using runoff energy factor. Present and Prospective Technology for Predicting Sediment Yield and Sources: Proceedings of the Sediment YieldWorkshop 1975 , USDA Sedimentation Lab., Oxford, November 28– 30, 1972. ARS-S-40. 244– 252. Williams JR, Jones CA, Dyke PT. 1984. A modeling approach to determine the relationship between erosion and soil productivity. Transactions of the ASAE 27(1): 129–144. Winchell M, Srinivasan R, Di Luzio M, Arnold J. 2007. ArcSWAT Interface for SWAT User’s Guide. Blackland Research Center, Texas Agricultural Experiment station and USDA Agricultural Research Service. Wischmeier WH, Smith DD. 1965. Predicting rainfall-erosion losses from cropland east of the Rocky Mountains. Agriculture Handbook , vol. 282. USDA-ARS. Wischmeier WH, Smith DD. 1978. Predicting rainfall erosion losses: a guide to conservation planning. Agriculture Handbook , vol. 282. USDA-ARS. Young RA, Onstad CA, Bosch DD, Anderson WP. 1987. AGNPS; Agricultural Non-point Source Pollution Model: A large watershed Analysis tool . Conservati Research. Report35, USDA-ARS: Washington, DC; 77pp. Young RA, Onstad CA, Bosch DD, Anderson WP. 1989. AGNPS: a non point Sourse Pollution Model for Evaluating Agricultural Watersheds. J. Soil and Water Conservation 44(2): 168–173. Zemenfes T. 1995. The political economy of land degradation in Ethiopia. Northeast African Studies 2: 71–98.
Hydrol. Process. (2009) DOI: 10.1002/hyp