Comparison of a Subjective and a Physical Approach for Identification of Priority Areas for Soil and Water Management in a Watershed – A Case Study of Nagwan Watershed in Hazaribagh

Environmental Modeling and Assessment 9: 115–127, 2004. 2004 Kluwer Academic Publishers. Printed in the Netherlands.

Comparison of a subjective and a physical approach for identificationof priority areas for soil and water management in a watershed –

A case study of Nagwan watershed in Hazaribagh Districtof Jharkhand, India

Ravinder Kaur a,∗, Omvir Singh a, R. Srinivasan b, S.N. Das c and Kamal Mishra d

a Division of Environmental Sciences, Indian Agricultural Research Institute, New Delhi-12, IndiaE-mail: [email protected]

b Texas Agricultural Experiment Station, Blackland Research Center, Temple, Texas-76502, USAc All India Soil Survey & Land use Planning, Ministry of Agriculture, Govt. of India

d Department of Soil Conservation, Damodar Valley Corporation, Hazaribagh, Jharkhand, India

The present investigation was an attempt to compare the within-watershed prioritization capabilities of a physical model based SDSSwith the SYI and RPI model based subjective-SDSS, conventionally devised for between-watershed prioritizations, by All India Soil Surveyand Land Use Planning Division of Ministry of Agriculture, Government of India. Application of these two approaches on a test watershedsituated in Damodar-Barakar catchment in India, the second most seriously eroded area in the world, demonstrated that the proposedphysical model based SDSS was capable of realistically and logically mimicking the sub-watershed-scaled water and soil losses therebysuggesting its immense application potential for priority area identification within the test watershed. In contrast to the proposed physicalmethod, the subjective approaches, which assigned totally reverse priorities to about 67–93% of the test-sub-watersheds, were observedto be incapable of realistically assessing the impact of topography and varied land use and soil types in the test watershed on their sub-watershed scaled run-off and soil loss generating potential. Besides, the physical approach could be used for assessing the annual dynamicsof the total water and sediment yields, under prevailing resource management systems in the test watershed with good to moderately goodcorrelation coefficients of 0.83 and 0.65; model efficiency coefficients of 0.54 and 0.70; mean relative errors of −4.28% and −17.97% androot mean square prediction errors of 71.8 mm and 9.63 t/ha, respectively.

Keywords: Damodar-Barakar catchment, SWAT, watershed prioritization, water yield, sediment yield, run-off, soil loss

1. Introduction

In India, the problem of soil erosion was recognized wayback in 1850–1860. This was addressed in several compre-hensive ways by means of legislation. Within one year of itsindependence, India passed the Damodar Valley CorporationAct in 1948 and established an inter-state system to undertake an integrated watershed management in catchment ofDamodar River. However at national level, the concept ofwatershed management gained momentum only during fifthfive-year plan (1974–1975). Since then many ministries anddepartments have initiated studies/programs on soil & waterconservation at watershed scale [1,2].

Inspite of all these efforts it is estimated that presentlyout of a total geographic area of 328 Mha, about 150 Mha isunder serious soil erosion problem and another 69 Mha is atcritical stage of deterioration in the country. Out of this totaleroded area, about 32 Mha is subject to wind erosion therebyleaving about 187 Mha subject to varying degrees of watererosion problems. Sheet erosion exists almost throughoutthe country [3]. It has been estimated that about 5334 Mtof soil is lost annually due to agriculture and associated ac-tivities alone thereby leading to an annual average nationalsoil erosion rate of about 16.75 t/ha/yr (which translates into

∗ Corresponding author.

approximately 1 mm each year) against the permissible levelof 7.5 to 12.5 t/ha/yr for various regions [4]. It is estimatedthat one-third of the arable land is likely to be lost withinthe next 20 years, if the present trend is allowed to continue.Out of this soil lost annually, about 2052 Mt is carried by therivers with nearly 1572 Mt (29%) falling in the sea and about480 Mt (10%) getting deposited in various reservoirs therebyleading to a loss of the live storage capacity of these reser-voirs by 1–2% annually [5]. It has been observed that thetrap efficiency of Indian reservoirs is about 90% [6] therebyleading to an equally alarming situation of serious floods inthe country. Floods hit India almost annually and cause dam-age to an average area of 6.7 Mha, resulting into an estimatedannual loss of rupees 1260 million.

Thus, although many soil and water conservation studieshave been initiated at watershed scale since early seventiesyet it has been observed that these programs were ineffec-tive of bringing down the national soil loss rates close totheir soil loss tolerance values. This was majorly becausethese programs lacked adequate, accurate and scientific in-formation on natural resource at watershed level; sufferedfrom inadequacy of organization and skills needed for col-lection and interpretation of data required for precise wa-tershed planning and relied on intuition, engineering expe-rience, thumb rules and site specific simple subjective tech-

116 R. Kaur et al. / Identification of priority areas for soil and water management

niques for making management decisions that had extremeeconomic and environmental implications [7]. All this ledto an acute shortage of systematic impact studies, withoutwhich no decision-making is possible.

Many water and sediment yield-estimating methods havebeen designed so far for watershed prioritization. These in-clude sediment-rating curves between water and sedimentdischarge rates [8,9]; statistical regression relations betweenone or more basin characteristics and/or climatic factors[10,11]; sediment delivery ratio procedures relating sedi-ment delivery ratios to the basin area and shape parame-ters [12–14]; physical models based on the (physical) equa-tions characterizing the hydrological processes controlling abasin’s response to any hydrologic event or a process [15–18] and subjective (parameteric) models relating the indi-vidual and combined effects of various basin-morphometricparameters to the relative sediment yields from each of thehydrologic units within a basin [19–21]. It is important tonotice here that the sediment-rating curve and statistical re-gression techniques require actual stream flow data whilethe rest of the techniques require GIS/remote sensing tech-nique based information, extracted from toposheets/satelliteimageries, on watershed morphology and/or physical or sub-jective models for watershed prioritization. Hence, in con-trast to the first two approaches, the last three techniques canbe used even for prioritizing ungauged watersheds.

Management of eroded lands requires concerted efforts inplanning and implementation of soil and water conservationmeasures at large scales. This requires continuous monitor-ing of sediments in run-off from large eroded lands. How-ever, continuous monitoring of water and sediment yieldscontributed by these large eroded lands is politically, eco-nomically and socially difficult, complex, costly and timeconsuming and it is not possible to collect silt data from allwatersheds/sub-catchments within the country. This prob-lem can be solved to some extend by gauging only those,critical water and sediment yielding, catchments which needpriority attention. The Sediment Yield Index (SYI) and Run-off Potential Index (RPI) are an excellent examples of thesubjective models used for this purpose [22–24] in India.The above (subjective) spatial decision support system forwatershed prioritization was conceptualized by the All IndiaSoil Survey and Land Use Planning (AISSLUP), Ministry ofAgriculture, Government of India, in 1969 and has been inuse since then for prioritizing sub-catchments/watersheds in31 – River Valley Project (RVP) and 10 – Flood Prone River(FPR’s) catchments [3,4]. Both these models have been re-fined and modified over years [25–27].

It is a well-known fact that different areas within a wa-tershed do not yield water and sediments similarly, as theyare exposed to different agricultural systems. Also the avail-able resources with the government are not sufficient totreat (saturate) the entire representative watershed imme-diately. In this context, it becomes essential to even lo-cate those critical water and sediment yielding source ar-eas within a (representative) watershed, which need prior-ity attention. At present the conventional subjective mod-

els have been used for only prioritizing watersheds withinlarge basins/catchments (i.e. inter-watershed comparisons)but not for prioritizing areas within a watershed (i.e. forintra-watershed comparisons) there by leading to, at times,either an unnecessary saturation of representative water-sheds with various conservation strategies or implantationof conservation strategies at not so critical locations. In fact,once these critical source (problem) areas are identified, avariety of techniques can be used to minimize the impactof various agricultural and other activities on environment.These practices are generally area specific and thus generalguidelines are not feasible. The cost effectiveness of thesepractices also varies significantly from site to site. Evalu-ating alternative management practices through experimentsat a site is generally not feasible. In this context, to identifynon-point source pollution problem areas within a watershedand to calculate the effectiveness of existing resource man-agement systems or hypothetical solutions, more quantita-tive spatial decision support systems, including the physicalmodels and geographic information systems (GIS), are of agreat use.

Currently a battery of (physical) hydrologic models areavailable for simulating water and sediment yields at wa-tershed/basin scales. However, at these scales, the mostcommonly used modeling techniques comprise of SWRRB[17,28]; ANSWERS [15]; AGNPS [16]; LlSEM [29] andSWAT [18]. For developing a SDSS for assessing the impactof existing resource management plans on regional waterand sediment yields it is essential that the selected (regional)hydrologic model is not site specific; has no subjective bias;has ability to assess impact of land use and other resourcemanagement strategies on water quantity, silt production andcrop growth; has ease of use of the methodology and is basedon readily available input-data; has ability to link to GIS andhas sufficient documentation and degree of support avail-able.

Keeping in view the above facts, the proposed study wasthus an attempt to:

1. Develop such a physical (hydrologic) model based quan-titative Spatial Decision Support System (SDSS) whichcould not only estimate (predict) total water and sedi-ment yields from the test-watershed, under current re-source management systems but could even identify thehigh priority water and sediment yielding source areaswithin a test Nagwan watershed in the Damodar-Barakarcatchment (i.e. watershed prioritization) thereby;

(a) Saving on cost and time involved in such large scalemonitoring of erosion processes within a watershedand

(b) Focusing most of the available central/state govern-ment funds in only high priority areas within a repre-sentative watershed; and

2. Compare the within-watershed prioritization capabilitiesof the proposed SDSS with the conventional SYI andRPI model based subjective SDSS of AISSLUP. It is im-portant to mention here that till date the conventional

R. Kaur et al. / Identification of priority areas for soil and water management 117

subjective models have been tested for only their poten-tial to prioritize watersheds within large catchments withvarying agro-climatic conditions [19–21,30,31]. Thusthrough this investigation an attempt to test its within-watershed prioritization capabilities was also made.

2. Materials and methods

2.1. General characteristics of test watershed site

Encompassing a total area of nearly 9576 ha, the test Nag-wan watershed (No. 7 A in figure 1) is located at the up-per part of the Sewani river between 23◦59′–24◦05′N and85◦18′–85◦23′E within the Damodar-Barakar catchment inIndia, the second most seriously eroded area in the world[32]. The test watershed is just 7 km from the soil con-servation department of D.V.C. at Hazaribagh, Jharkhand;is well connected by road/rail network and is having welldocumented long term time-series database on hydrologyand sediment-yield; right from 1959 onwards. Its measur-ing station is located at 85◦24′E and 24◦03′N. Geologically,the area is quite complex, having rocks of varying compo-sition. The soils of the area are mainly of clay loam type.The maximum and the minimum elevations of the area are637 m and 564 m, respectively. About 86% of area is un-der a slope range of 1–6%. The area experiences sub-humidsub-tropical monsoon type of climate, characterized by hotsummers (40◦C) and mild winters (4◦C). The total annualprecipitation of 1206 mm is distributed mainly between Juneto September, with about 15 rainy days per month. The av-erage storm intensity, by considering storms of more than30 min duration, is about 10 cm/hr. Out of the total wa-tershed area of 9576 ha about 55.00% is under agriculture,

17.22% is under forest and about 21.77% is a wasteland.The main agricultural crops grown during Kharif season arepaddy and maize and in Rabi season are wheat, gram andmustard. Paddy–mustard, paddy–wheat, and maize–mustardare the main crop rotations followed in the area but a ma-jority of the watershed is mostly single cropped with paddyas the major crop and maize as the second most commoncrop. The agriculture is mostly rainfed as only 20% irri-gation is available in the area through sources other thanrain and the cropping intensity is also quite low at 98%.The irrigation is received mainly by wells. Prevalence ofconventional cultivation practices, characterized by conven-tional tillage or no tillage; low fertilizer/manure consump-tion and local varieties of the crops is mainly responsible forthe low crop productivity in the area. The area was fullytreated about 30 years back, with about 50% of the area re-quiring re-treatment. All this information on the test areawas obtained through secondary sources such as Directorateof Economics and Statistics, Ministry of Agriculture; Direc-torate of Census (data Center); DVC, Hazaribagh officialsand Sadar block office of Hazaribagh district, and throughprimary PRA exercises conducted in the test area. Theanalysis of above data thus revealed a subsistence level agri-culture, rainfed farming, dominance of marginal and smallfarmers, fragmented land holdings, low productivity, lacu-nae in the resource management practices and obsolete tech-nical know-how on improved crop production as the domi-nant socio-economic characteristics of the test-watershed.

2.2. Proposed physical model based SDSS

2.2.1. StructureThe proposed SDSS comprised of a Soil and Water As-

sessment Tool (SWAT). The SWAT is a river basin or water-

Figure 1. Location map of test Nagwan watershed.


shed scale hydrologic model developed by the USDA Agri-cultural Research Service (ARS). It operates on daily timestep and is developed to predict the impact of land manage-ment practices on water, sediment and agricultural chemicalyields in large complex watersheds with varying soils, landuse and management conditions over long periods of time.It is in fact a combination of the SWRRB, GLEAMS andROTO models and hence it is able to model both the hy-drology and water quality of a watershed. The model is re-ported to be able to operate on both raster and sub-watershed(hydrologic response unit) basis. In addition, the model islinked to GIS packages like GRASS via the SWAT–GRASSinterface [33] and ArcView through the SWAT–ArcView in-terface [34], thus easing the task of data input and outputdisplay. The current SWAT–ArcView interface (AVSWAT)is a tight coupling between model and GIS. The export ofdata from GIS to the SWAT model and the return of re-sults for display is accomplished by Avenue routines that areaddressed directly by the interactive tools of GIS and theexchange of data is fully automatic. The SWAT–ArcViewsystem consists of 3 key components: (1) preprocessor gen-erating sub-basin topographic parameters and model inputparameters; (2) editing input data set and execute simulation;(3) postprocessor viewing graphical and tabular results. Fig-ure 2 illustrates a schematic of the SWAT–ArcView systembased proposed SDSS.

2.2.2. Generated input maps and data filesMap themes

The SDSS basically requires four map themes viz. Digi-tal Elevation Model (DEM), Land use/Land cover, Soil andGroundwater in ArcInfo–ArcView Grid format.

(1) Digital Elevation Model (DEM) grid. A DEM is thesurface interpolation from contour data points of an area.Figure 2 illustrates the DEM generated [35] from a basic 10mt interval contour data (1 : 25,000 scale) for the test Nag-wan sub-watershed in Hazaribagh district of Jharkhand.

(2) Land use grid. The land use/land cover map (figure 2)of Nagwan (1 : 25,000 scale) was obtained from the SoilConservation Department of the Damodar Valley Corpora-tion, Hazaribagh, Jharkhand in analog form and was con-verted into digital form.

(3) Soil grid. Soil map (1 : 25,000) of Nagwan watershed(figure 2) was also obtained from the Soil Conservation De-partment of the Damodar Valley Corporation, Hazaribagh,Jharkhand in analog form. The map comprised of 10 differ-ent themes viz. Soil series, Surface texture, Slope, Depth,Stoniness, Rockiness, Gravelliness and Land-Capability andErosion classes. This map was also converted into digitalform for further processing.

(4) Groundwater grid. From a grid file of groundwater al-pha parameter, the interface extracts the groundwater para-meters needed for SWAT run. In the present study due to

lack of the groundwater data, this file could not be devel-oped and default values for groundwater parameters used bythe SWAT were used.

ArcView Tables

(A) In (.DBF format)

(1) Soil class table. A table that connected the categorycode of the soil map with the corresponding soil file (.Sol)name was made in Excel and was saved as .Dbf IV file type.This file comprised of 6 soil category codes: (1) LMSND(Loamy Sand); (2) SLOAM (Sandy Loam); (3) LOAMY(Loam); (4) SILM (Silty Loam); (5) CLYLM (Clay Loam);and (6) SICLM (Silty Clay Loam).

(2) Land use class table. A table which connected the cat-egory code of the land use/land cover map with the corre-sponding SWAT land use/land cover classes was also madein Excel and was saved as .Dbf IV file type. This file com-prised of 5 land use category codes: (1) (Upland) RICE;(2) (Upland) CORN; (3) (Long Duration Kharif ) RICE;(4) (Forest with good canopy) FRST; (5) (Gullied land withpoor canopy forest) FRST.

(3) Tables to import raingauge, temperature gauge andweather generator gauge locations. For the present studythe nearest meteorological station to Nagwan watershed wasIndia Meteorological Department’s (IMD) Hazaribagh sta-tion. Since both rainfall and temperature were measuredfrom the same station therefore for all the three files samelongitude and latitude (i.e. 85◦24′E and 24◦03′N, respec-tively) were used.

(4) Tables to store daily precipitation and temperature dataTables for storing 11 years daily precipitation and temper-ature data, from 1981 to 1992, corresponding to the testwatershed-gauging station were prepared from the daily me-teorological data procured, for the above 11 years period,from IMD.

(B) In (.TXT format)

(1) Soil files. Since 6-soil category codes were distributedover Nagwan watershed, therefore 6-independent files with*.SOL extension were prepared for each soil type. Table 1gives a detailed account of the soil input parameters requiredby the SWAT hydrologic model, in the proposed SDSS, foreach of these soil types. For making these files, informationon sand, silt, clay, organic carbon percentages and soil hy-drologic group were obtained from the soil conservation de-partment of DVC, Hazaribagh while the other soil parame-ters such as bulk density, available water capacity, saturatedhydraulic conductivity, moist soil Albedo, soil erodibilityfactor (which were not available from the soil conserva-tion department of DVC, Hazaribagh), were estimated fromthe (available) basic soil properties such as sand, silt, clayand organic carbon and soil hydrologic group information


Figu

re2.

Stru

ctur

eof

prop

osed

SDSS

for

Nag

wan

test

wat

ersh

ed.


Table 1Soil class specific data for Nagwan watershed.

Soil properties Soil classSilty loam Loamy sand Sandy loam Loam Clay loam Silty clay loam

Hydrologic group C B B C D DSand (%) 51.30 80.40 62.95 50.77 34.35 39.74Silt (%) 29.40 11.70 20.09 22.52 22.26 24.53Clay (%) 19.30 7.90 16.96 26.71 43.39 35.73Organic carbon (%) 0.56 0.47 0.29 0.26 0.22 0.31Estimated bulk density (g/cc) 1.43 1.62 1.48 1.39 1.28 1.32Estimated available water 0.12 0.09 0.10 0.11 0.12 0.12

capacity (cc/cc)Estimated saturated hydraulic 8.40 40.87 10.33 3.90 1.71 2.28

conductivity (mm/hr)Estimated moist soil albedo 0.26 0.29 0.35 0.37 0.38 0.35Estimated soil erodibility (K) 0.33 0.29 0.34 0.26 0.19 0.26

through a stand alone software developed by USDA/ARSGrassland soil and water research laboratory for determin-ing soil properties required for SWAT implementation. Soilerodibility factors, for each soil layer and soil type, were es-timated through the following equation [36]:

100K = 2.1M1.14(10−4)(12−a)+3.25(b−2)+2.5(c−3),

where, M = (%silt + %very fine sand)(100 − %clay), a =%organic matter (obtained by multiplying available %-soilorganic carbon values with a factor 1.72; Davies, 1974), b =structure code (1 – very fine granular; 2 – fine granular; 3 –medium or coarse granular; and 4 – blocky, platy or massive)and c = permeability class (1 – rapid, 2 – medium to rapid,3 – moderate, 4 – slow to moderate, 5 – slow, 6 – very slow).

(2) Crop file. The Crop file contains the crop input para-meters for SWAT. This file resides in the program file direc-tory of the SWAT–ArcView interface (AVSWATPR) and getscopied when a project is created through the interface. In thepresent study, the original Crop.dat file was altered for paddyand maize by replacing the values for the harvest index (HI),maximum potential LAI (BLAI), maximum rooting depth(RDMX), maximum crop height (CHTMX) and lower limitof harvest index (WSYF) parameters with their actual val-ues, obtained from Damodar Valley Corporation, for thesecrop types in the Hazaribagh district (table 2).

(3) Management file. This file contains input data for plant-ing, harvest, irrigation applications, nutrient application,pesticide applications and tillage operations. These opera-tions may be scheduled by month and day or by heat units.In the present study these operations were scheduled bymonth and day. Table 3 details the crop-specific (local) landand water management scenarios incorporated in the pro-posed SDSS. This information on the test area was obtainedthrough the primary PRA exercises conducted in the testarea. Since a majority of the test watershed was either underno or poor conservation practices therefore for the presentexercise whole watershed was assumed to be under no con-servation practice (i.e. with P factor value = 1).

(4) Weather generator file. The weather variables neces-sary for driving SWAT are precipitation, air temperature,solar radiation, wind speed, and relative humidity. If dailyprecipitation and maximum and minimum air temperaturedata are available, they can be inputted directly to SWAT. Ifnot, the weather component of the model can simulate dailyrainfall and temperature. Solar radiation, wind speed, andrelative humidity are always simulated. The inputs parame-ters, in the weather generator file, required for simulatingthese weather variables were obtained from an 11-year dailyweather data, obtained from IMD, by means of a programWXPARM developed at Blackland research center.

2.2.3. Implementation on test watershedThe delineation of the test Nagwan watershed and the

development of the watershed and sub-watershed databasewere the first set of fundamental steps performed by the pro-posed SDSS (i.e. SWAT ArcView preprocessor). The testNagwan watershed and its sub-watershed boundaries weredelineated from the DEM of the test area [35] by setting athreshold value of 300 ha or 834 cells for starting stream de-lineation. This led to the delineation of 15 sub-watershedswithin the test watershed (figure 2). Once watershed andsub-basin boundaries were determined all the geometric pa-rameters of sub-basins and stream reach were determined bythe raster-grid functions and stored as attributes of derivedvector themes. Next the land use and soil grids and relateddata files were loaded into the view and clipped in the water-shed area followed by their re-classification and re-samplingat the DEM cell size. In order to capture the heterogene-ity in soil and land use in each sub-watershed, they werefurther divided into 44 homogenous units viz. Hydrologi-cal Response Unit (HRU) or virtual basins on the basis ofthe threshold values of (20% of) land use and (10% of) soiltypes within it. While selecting the distribution it was keptin mind that each physical watershed had at least one HRUthat satisfied the set of thresholds. Precipitation, tempera-ture and weather generator data were defined by loading thestation location and data files. All the SDSS required inputfiles were generated sequentially through the interface andthe SDSS was run for the calibration and the validation pe-


Table 2Salient crop parameters incorporated in proposed SDSS.

Crop name Max total Max harvest Min harvest Max LAI Max crop Max rootbiomass (t/ha) index index height (m) depth (m)

Upland maize 16 0.313 0.063 3.13 2.0 0.25Long duration 16 0.400 0.125 4.80 1.0 0.30

paddyUpland paddy 12.5 0.368 0.126 4.42 0.45 0.20

Table 3Land and water management practice scenarios incorporated in proposed SDSS.

Crop name Tillage Irrigation Planting/sowing Harvesting Fertilizer Pesticide

Maize Depth: 25 mm NIL 15th June 15th Sept. FYM: 1.5 t/ha NILWith cultirow Biomass_target: 16 kg/ha Urea:10 kg/haon 14th June Harvest index_target: 0.32 (basal)

Long duration Depth: 50 mm 1-Pre-sown 5th Sept. 5th Jan FYM: 1.5 t/ha NILpaddy With cultiweeder on 4th Sept. Initial LAI: 1.1 Urea: 15 kg/ha

on 1st Sept. of 120 mm Initial biomass: 800 kg/ha (basal)Biomass_target: 12.5 kg/ha DAP: 25 kg/haHarvest index_target: 0.32

Upland paddy Depth: 50 mm 1-Pre-sowing 25th June 25th Sept. FYM: 1.5 t/ha NILWith cultiweeder on 24th June Initial LAI: 1.1 Urea: 15 kg/haon 22nd June of 120 mm Initial biomass: 800 kg/ha (basal)

Biomass_target: 12.5 kg/ha DAP: 25 kg/haHarvest index_target: 0.36

Good canopy NIL NIL Initial LAI: 3.5 NIL NIL NILforest Initial biomass: 5000 kg/ha

Poor canopy NIL NIL Initial LAI: 2.5 NIL NIL NILforest Initial biomass: 2500 kg/ha

riods, after setting the initial soil moisture storage and base-flow factor values at 1 and using the Pristley–Taylor methodfor evapo-transpiration estimations.

2.2.4. Calibration and validationThe two hydrologic years, 1984 and 1992, associated

with the highest and the lowest water and sediment yields,respectively were selected from an 11-year period from 1981to 1992 for the proposed SDSS calibration, at a daily timestep. The input variables used for calibration were soil prop-erties and curve number. The curve number [37] was al-lowed to vary within the categories for good and fair hy-drologic conditions and the available water capacity was setwithin the range of its natural uncertainty for the study re-gion. The remaining input data for the 9-years period (com-prising of 1981–1983; 1985–1989 and 1991 years), out ofa total 11 years period, from 1981 to 1992, was used forthe proposed SDSS-validation. Correlation coefficient (R),Mean Relative Error (MRE), Root Mean Square PredictionDifference (RMSPD) [38] and model efficiency coefficient(R_nash) [39] statistical measures were used to assess thewater and sediment yield predicting potential of the pro-posed SDSS, at the test watershed outlet, for the calibra-tion and the validation data sets (tables 4 and 5). Sincethe test watershed had no actually monitored water and sed-iment yield data for each HRU or a sub-watershed containedin it (besides the water and sediment yield data available at

its outlet) therefore for the present study a moderately goodvalidation of the proposed SDSS at the test watershed-outletwas considered to be one of the reasonable criteria for judg-ing the proposed SDSS’s potential to realistically mimic theindividual HRU and sub-watershed scaled-run-off and soillosses and hence to identify priority areas within the test wa-tershed for soil and water management.

2.3. Conventional SYI and RPI-model based subjectiveSDSS

2.3.1. StructureThe Sediment Yield Index (SYI) model considers sedi-

ment delivery from a hydrologic unit into a reservoir or asink area as a multiplicative function of the potential sed-iment detachment represented by the erosivity factor, thetransportability of the detached material (represented by thedelivery ratio) and the area of the hydrologic entity. Theerosivity of any hydrologic entity is expressed in terms of itssediment yield weightage value which is in turn is a func-tion of several erosivity determinants, such as physiography,slope, soil type, land use/land cover and soil management ofthe hydrologic entity. The delivery ratio on the other hand isadjudged by not only the above soil erosivity determinantsbut also by the proximity of the areas to the mainstream orreservoir(s). Mathematically the model was expressed as:

SYI ={∑

(Ai · Wi · Di)/Aw

}· 100, i = 1, . . . , n, (1)


Table 4aGoodness-of-fit tests for observed vs predicted water

yields during calibration period (1984 & 1992).

Calibration year Month(s) Water yield(s) mmObserved Predicted

84 6 210.00 266.937 72.00 57.388 187.00 115.459 45.00 79.92

10 26.00 33.1792 6 51.00 87.26

7 57.00 63.578 96.00 87.439 28.00 27.06

10 26.00 17.53

Mean 79.80 83.57STD 66.50 71.38R2 0.76

Model efficiency 0.71

Table 4bGoodness-of-fit tests for observed vs predicted sediment

yields during calibration period (1984 & 1992).

Calibration year Month(s) Sediment yield(s) t/haObserved Predicted

84 6 13.43 28.707 9.56 3.418 8.34 3.229 3.96 2.07

10 0.79 0.0392 6 10.28 11.16

7 4.80 3.748 8.96 2.639 0.00 0.40

10 0.00 0.05

Mean 6.01 5.54STD 4.77 8.75R2 0.54

Model efficiency −0.67

where, Ai is area of the ith mapping unit in a sub-watershedwithin a test watershed; Wi is the sediment yield weightedvalue, as obtained from the standard look-up tables preparedby AISSLUP, for each mapping unit with a given physiog-raphy, slope, soil type, land use/land cover and managementcondition; Di is the adjusted delivery ratio assigned, as perthe standard look-up tables of AISSLUP, for each mappingunit within a sub-watershed in a test watershed; Aw is thetotal area of a sub-watershed within a test watershed and n

is the number of mapping units in a sub-watershed within atest watershed.

The Run-off Index Model is also a modification and ex-pansion of SYI model wherein the erosivity values of an as-semblage of soil and land parameters are replaced by therun-off weightage values and the delivery ratio factor is nottaken into account. This was mathematically expressed as:

RPI ={∑

(Ai · Wi)/Aw

}· 100, i = 1, . . . , n, (2)

where, Ai is area of the ith mapping unit in a sub-watershed

within a test watershed; Wi is the sediment yield weightedvalue, as obtained from the standard look-up tables preparedby AISSLUS, for each mapping unit with a given physiog-raphy, slope, soil type, land use/land cover and managementcondition; Aw is the total area of a sub-watershed within atest watershed and n is the number of mapping units in asub-watershed within a test watershed.

2.3.2. Implementation on test watershedDelineation of test watershed and sub-watershed bound-aries. Like proposed SDSS, the delineation of the testNagwan watershed and the development of the watershedand sub-watershed databases were the first set of fundamen-tal steps performed by means of the ArcInfo GIS software.The test Nagwan watershed and its sub-watershed bound-aries were delineated from the DEM of the test area as perthe procedures detailed in Kaur and Dutta [35]. This led tothe delineation of 15-sub watersheds within the test water-shed (figure 2).

Digitization of test watershed land use/land cover andsoil map. A detailed land use/land cover and soil map(1 : 25,000) of Nagwan watershed (figure 2) was obtainedfrom the Soil Conservation Department of the Damodar Val-ley Corporation, Hazaribagh, Jharkhand in analog form. Themap comprised of information on 14 different themes viz.Soil series, Surface texture, Slope, Depth, Stoniness, Rock-iness, Gravelliness, Present Land use, Forest Areas, WasteLands, Gullied Lands, Management, Land-Capability andErosion classes for each mapping unit in the test watershedarea. This map was converted into digital form by means ofArcInfo software for further processing.

Intersection of sub-watershed boundary and land use/landcover-soil maps. The sub-watershed boundary and landuse/land cover and soil maps delineated/digitized abovewere intersected, by means of ArcInfo software, to gener-ate a new map comprising of sub-watershed wise spatiallydistributed information on the area, surface texture, slope,depth, stoniness, rockiness, gravelliness, present land use,forest areas, waste lands, gullied lands, management, land-capability and erosion classes for each mapping unit in thetest watershed area.

Assigning sediment yield weightage and delivery ratio val-ues. The computation of the SYI and RPI for each sub-watershed within the test watershed required the evaluationof the sediment yield weightage and/or delivery ratio valuesfor each mapping unit within a particular sub-watershed ofthe test watershed. These values per mapping unit were eval-uated as a function of their physiography, slope, soil type,land use/land cover and soil management, as per the recom-mendations made by the AISSLUP [22]. The information onthe physiography, slope, soil type, land use/land cover andsoil management for each mapping unit were obtained fromthe ArcInfo database file linked to the digitized land use/landcover and soil map for the test watershed. The values of sed-iment yield weightage and delivery ratios so evaluated for


each mapping unit per sub-watershed in the test watershed,together with the mapping unit and sub-watershed-areas ob-tained by means of GIS software, were then used for calcu-lating SYI and RPI for each sub-watershed within the testwatershed by solving equations (1) and (2) in an excel work-sheet.

Categorization and gradation of sub-watersheds. The SYIand RPI evaluated for each sub-watershed were then groupedinto three classes to categorize areas with low, medium andhigh sediment and water yielding potential within the testwatershed.

3. Results and discussions

3.1. Physical model based SDSS and water and sedimentyield estimations

Tables 4a and 5b depict the results of goodness of fittests performed on the (kharif season) observed vs pre-dicted water and sediment yields, respectively for the SDSScalibration-years (1984 and 1992) at the test watershed out-let. It could be observed from these results that the calibratedavailable soil water capacity and curve number values in theproposed SDSS yielded reasonably good correlation coef-ficient and model efficiency coefficient values of 0.76 and0.71, respectively for the observed vs predicted total wateryields for the calibration years. It could be further observedfrom these tables that the proposed SDSS was even associ-ated with moderately good correlation coefficient and modelefficiency coefficient values of 0.54 and −0.67, respectivelyfor the observed vs. predicted total sediment yields for thesame calibration years.

For the model validation period, the actual annual ob-served data for the validation period indicated average totalwater and sediment yields (tables 5a and 5b) of 390.69 mmand 25.35 t/ha from the Nagwan watershed during the Kharifseasons. In comparison to this, the proposed SDSS predictedaverage (annual) total water yield of 383.37 mm, with corre-lation coefficient of 0.83; model efficiency of 0.54; relativeerror of −4.28% and root mean square prediction error of71.8 mm and average (annual) total sediment yield of 21.28t/ha, with correlation coefficient of 0.65; model efficiencyof 0.70; relative error of −17.97% and root mean squareprediction error of 9.63 t/ha, respectively for the same val-idation period. Figure 3 gives a pictorial depiction of theobserved vs. predicted total water and sediment yields fromNagwan watershed for a 9-year validation period. The aboveresults thus clearly showed that even under Indian condi-tions the proposed SDSS could simulate the temporal dy-namics of the total water and sediment yields, at the testwatershed outlet, reasonably well and quite close to thosereported for model runs with regions in the US and Ger-many [40,41].

Table 5aGoodness-of-fit tests for yearly observed vs predicted water yields during

validation period (1981–1983; 1985–1989 & 1991).

Year Observed (mm) Predicted (mm)

1981 267.00 215.911982 309.00 222.981983 290.00 272.191985 472.10 393.571986 599.00 649.161987 495.00 621.331988 315.60 260.091989 417.50 380.711991 351.00 434.38

Mean (mm) 390.69 383.37STD (mm) 112.33 162.52

R2 0.83Model efficiency 0.54RMSPD (mm) 71.77

RE (%) −4.28

Table 5bGoodness-of-fit tests for yearly-observed vs predicted sediment yields dur-

ing validation period (1981–1983, 1985–1989 & 1991).

Year Observed (t/ha) Predicted (t/ha)

1981 19.09 6.721982 19.40 9.021983 25.93 14.481985 34.32 19.921986 51.96 56.201987 22.47 33.081988 13.76 11.051989 18.24 25.141991 23.00 15.89

Mean (t/ha) 25.35 21.28STD (t/ha) 11.52 15.48

R2 0.65Model efficiency 0.70RMSPD (t/ha) 9.63

RE (%) −17.97

3.2. Watershed prioritization

3.2.1. Comparison between physical and subjectiveapproachesThe average annual run-off map, obtained through the

physical model based SDSS, for Nagwan watershed is de-picted in figure 4a. It could be clearly observed from thisfigure that moderate (300–400 mm) to high (>400 mm) run-offs were produced by the sub-watershed nos. 7, 12, 11 and1, 2, 3, 4, 6, 8, 15 respectively while the lowest (<300 mm)run-offs were produced by the sub-watershed nos. 5, 9, 10,13 and 14. In contrast to this, the RPI-model based water-shed prioritization technique (figure 4b) assigned 5, 9, 10,13 and 14 – sub-watersheds, with lowest run-offs predictedby the proposed physical-model based SDSS, in moderate(600–800) to high (>800) run-off potential classes and thesub-watersheds 7, 12 and 1, 2, 3, 4, 6, 8, 15, with mod-erate to high run-offs estimated by the proposed physical-model based SDSS, in the lowest (<600) run-off potential


Figure 3. Temporal dynamics of observed vs. predicted total water and sediment yields from Nagwan watershed for a 9-year validation period.

class. It is important to notice that only sub-watershed 11,with moderate run-off predicted by the proposed physicalapproach, was classified somewhat same as that by the sub-jective RPI approach. Similarly, the average annual soilloss map (figure 5a) of the test watershed, proposed by thephysical-model based SDSS, revealed moderate to high soilloss from the sub-watershed nos. 1, 2, 3, 4, 8, 12, 13, 14 and6, 11 and 15 respectively and the lowest soil loss from thesub-watershed nos 5, 7, 9 and 10. In contrast to this, the SYI-model based watershed prioritization technique (figure 5b)assigned the sub-watersheds 5, 7, 9 and 10 respectively, with

lowest physical-model based soil loss, in moderate (10000–30000) to high (>30000) classes and the sub-watersheds 1,2,4, 8 and 6, 15, with moderate to high soil loss proposedby the physical model based SDSS, in the lowest (<10000)class. Only the sub-watersheds 3, 12, 13, 14 and 11, withmoderate to high soil loss, were observed to be classifiedsimilarly by the two approaches. Hence it could be observedthat in comparison to the proposed physical-model basedSDSS, the conventional subjective-model(s) based SDSS as-signed totally reverse priority for about 67–93% of the test-sub-watersheds.


Figure 4. (a) Physical model based average annual runoff and (b) conventional subjective model based RPI maps of the test (Nagwan) watershed.

Figure 5. (a) Physical model based average annual soil loss and (b) conventional subjective model based SYI maps of the test (Nagwan) watershed.

3.2.2. Evaluation of physical and subjective approachesWater and soil loss from a watershed are in fact a function

of its topography (such as slope), physiography (such as landuse, soil type, soil depth) and climatic and management con-ditions. Table 6 gives a detailed account of sub-watershedwise percentage areas under different management practice,slope, land use and soil textural classes for Nagwan water-shed. In general the test-Nagwan watershed had very deep(>100 cm) soils. It could be clearly seen from table 6 thata majority of Nagwan watershed was either unmanaged orpoorly managed. It could be further observed from this ta-ble that sub-watersheds 1, 2, 3, 4, 6, 8, 10, 11, 12, 13 and14 were primarily (with >75% area) in 0–3% slope classwhile the sub-watersheds 5, 7, 9 and 15 had within 75%area in 0–3% slope class and about 25–47.6% area in >3%slope class. Out of these sub-watersheds in >3% slope class,

sub-watersheds 5 and 7 were the steepest. On comparingthe table 6 with figures 4a and 4b it could be further no-ticed that although sub-watershed 14, with the less steeperslope (i.e. only 1.35% area in >3% slope class) and lesser(i.e. 23%) forest cover than sub-watershed 13 (with 13.46%area under >3% slope class and 38% forest cover), hadhigher run-off (276.928 mm) than the later sub-watershed(233 mm) yet its soil loss was less (10.46 t/ha) than that forthe later (14.16 t/ha) sub-watershed. The relatively higherrun-off values from sub-watershed 14 as compared to sub-watershed 13 could be majorly attributed to the presenceof larger areas under long duration kharif paddy (9%) andlight textured soils (in 70% of area) in sub-watershed 14 ascompared to sub-watershed 13 which had just 5% area underlong duration kharif paddy and about 19% area under lighttextured soils. The long duration kharif paddy, grown be-


Table 6Detailed account of sub-watershed wise percentage areas under different management practice, slope, land use and soil textural classes for Nagwan

watershed.

Sub- Area (ha) Per cent area in Per cent area in Per cent area in land use class Per cent area inwatershed management slope class soil textural class

classNone Poor 0–3% >3% Upland Long Upland Good Poor Silty Clay Silty Loam Sandy Loamy

paddy duration maize canopy canopy clay loam loam loam sandpaddy forest forest loam

1.00 2008.65 5.18 94.82 94.68 5.32 67.71 29.14 0.46 0.00 2.69 0.21 22.22 42.42 8.19 0.46 26.492.00 980.52 20.21 79.79 93.68 6.33 82.82 5.71 0.00 3.61 7.86 23.50 0.61 27.73 2.41 3.64 42.113.00 326.02 52.54 47.46 77.17 22.83 62.20 24.52 0.00 4.70 8.83 5.95 4.90 79.84 2.21 7.10 0.004.00 271.47 31.32 68.68 99.28 0.72 69.40 30.60 0.00 0.00 0.00 0.00 0.00 100.00 0.00 0.00 0.005.00 490.38 97.52 2.48 53.47 46.53 33.64 0.00 14.01 6.10 46.25 86.82 0.00 8.95 2.48 0.00 1.756.00 352.49 43.62 56.38 87.76 12.24 63.69 27.02 6.95 0.00 2.33 12.24 0.00 80.80 0.00 0.00 6.957.00 25.93 100.00 0.00 52.38 47.62 0.00 28.02 47.62 0.00 24.35 34.92 0.00 17.46 0.00 0.00 47.628.00 1090.57 62.73 37.27 94.62 5.38 45.70 40.97 3.81 0.00 9.52 12.04 0.00 59.71 13.16 15.09 0.009.00 675.51 25.17 74.83 63.69 36.31 35.88 40.12 0.00 0.37 23.63 0.79 0.00 97.22 1.99 0.00 0.00

10.00 128.73 89.78 10.22 100.00 0.00 0.00 58.56 7.32 0.32 33.80 51.27 0.00 10.22 0.00 38.51 0.0011.00 126.66 53.96 46.04 93.22 6.78 0.00 46.04 28.24 1.30 24.43 0.00 0.00 31.83 14.00 54.17 0.0012.00 744.69 79.47 20.53 91.72 8.28 20.10 21.00 26.61 8.74 23.56 3.58 0.00 12.35 12.85 71.22 0.0013.00 1549.87 66.85 33.15 86.54 13.46 43.84 4.71 8.39 4.55 38.52 2.43 0.00 18.28 60.54 12.66 6.0814.00 349.94 88.30 11.70 98.65 1.35 68.75 8.82 0.12 0.00 22.31 0.00 0.00 51.64 2.14 46.21 0.0015.00 112.44 0.23 99.77 74.73 25.27 25.50 74.50 0.00 0.00 0.00 0.00 0.00 92.96 6.81 0.23 0.00

tween September–January, generally leaves the land surfacebare and exposed during intense rainfall periods (i.e. July,August) thereby leading to more run-off from areas undersuch land use types (LUTs) [42]. However this high run-offfrom the sub-watershed 14 was not associated with concomi-tantly higher soil losses due to the presence of much largerarea (68.75%) under non-erosive upland paddy and muchsmaller area (0.12%) under erosive maize [42] as comparedto the sub-watershed 13 which had only 43.84% area un-der non-erosive upland paddy and 8.4% area under erosivemaize crop. Sub-watersheds 10, 11 and 12 also, with almostsame forest cover (25.7–34.1%) and rather less steep slopesthan sub-watersheds 9, 13 and 14, contributed much higherrun-offs and soil losses than the later because of higher per-cent areas under long duration kharif paddy (21–46%) andmaize (26.7–28.2%) crops in the former sub-watersheds ascompared to the later sub-watersheds with only (0–9%) and(0–8.4%) area under long duration kharif paddy and maizecrops, respectively. Besides, lesser run-off and soil lossfrom sub-watershed 12 as compared to the sub-watershed11 could be attributed to the predominant presence of lighttextured soils (71%) in the later sub-watershed as comparedto the former sub-watershed with only 54% area under lighttextured soils. It could further be noticed that although sub-watersheds 5 and 7 fell under almost same slope classes(i.e. 46.53 and 47.62% area in >3% slope class and the restarea in 1–3% slope class, respectively), yet sub-watershed7 contributed far higher run-off (395.5 mm) and soil loss(7.77 t/ha) than sub-watershed 5 (175.77 mm and 0.50 t/ha,respectively) because in contrast to only 0% area under longduration kharif paddy and 14.01% area under erosive maizecrop in sub-watershed 5, sub-watershed 7 had about 28.02%area under long duration kharif paddy and 47.62% area un-der erosive maize crop. Although sub-watersheds 5, 7 and 9

fell under steep slope class (>3%) yet their run-off and soilloss values were substantially lower than sub-watersheds 1,2, 3, 4, 6, 8, 11 and 12, with just 1–3% slope, because of ex-tensive forest cover (23–52%) in the former sub-watershedsas compared to the later sub-watersheds (with 0–32.2% for-est cover). Amongst the higher run-off and soil loss con-tributing sub-watersheds 15, 6 and 11, sub-watershed 11 hadlowest run-off but highest soil loss which could be majorlyattributed to the presence of highest (25.7%) forest coverand highest (28.2%) maize crop cultivation respectively inthe later sub-watershed. Amongst sub-watersheds 15 and 6,sub-watershed 15 had higher run-off but lower soil loss thansub-watershed 6 due to no forest cover, no maize crop cul-tivation and 74.5% area under long duration kharif paddy.In contrast to this the sub-watershed 6 was associated with2.32% forest cover; 6.9% maize crop cultivation and 27 percent area under long duration paddy. Hence the proposedphysical model based SDSS was observed to be mimickingeven HRU and sub-watershed-scaled water and soil lossesquite realistically and logically thereby suggesting its im-mense application potential for priority area identification inthe test watershed. As, in contrast to the proposed physi-cal model based SDSS, the subjective approaches assignedtotally reverse priorities to about 67–93% of the test-sub-watersheds therefore it was observed that the subjective wa-tershed prioritization techniques could not account for theimpact of varied land uses and soil types, found in the sub-watersheds, on their total run-off and soil loss generating po-tential realistically. Thus the proposed physical-model basedapproach proved to be a far superior tool for identification ofpriority areas, for soil and water management, within the testwatershed.


4. Conclusions

The present investigation could thus quite realisticallydemonstrate the far superior potential of the proposed phys-ical model based SDSS, in comparison to the conventionalapproach of AISSLUP, for within-watershed prioritizations.Besides, although the subjective approach is simple and lessdata-intensive yet, unlike the physical approach, the investi-gations revealed that it could not be used for assessing theimpact of climate and land-use changes on the erosivity ofa hydrologic entity. Further, unlike the physical models thatcan actually quantify total water and sediment yields fromthe watersheds/catchments with diverse climate, land useand soils, the conventional subjective approach could onlybe used for assigning relative ranks to these hydrologic units.

References

[1] G. Singh, G. Venkataramanan, G. Sastry and B.P. Joshi, Manual ofSoil and Water Conservation Practices (Oxford and IBH Publ. Co.Pvt. Ltd., New Delhi, 1990).

[2] J.S. Bali, Journal of Soil and Water Conservation 42 (1998) 127–143.[3] Anonymous, Working Group Report (IX Plan), Planning Commis-

sion, Govt. of India (1996).[4] S. Bhan, Journal of Soil and Water Conservation 41 (1997) 21–24.[5] V.V. Dhruva Narayan and R. Babu, Indian Journal of Irrigation and

Drainage Engineering 109 (1983) 419–434.[6] J.C. Samuel, Journal of Soil and Water Conservation 43 (1999) 69–77.[7] Vaidyanathan, Integrated watershed development: some major issues,

Foundation Day Lecture, Society for Promotion of Waste Lands De-velopment, New Delhi (1991).

[8] N.E.M. Asselman, Journal of Hydrology 234 (2000) 228–248.[9] J.L. Loizeau and J. Dominik, Aquatic Sciences 62 (2000) 54–67.

[10] R.M. Vogel, I. Wilson and C. Daly, Journal of Irrigation and DrainageEngineering May/June (1999) 148–157.

[11] J. Schmidt, K. Hennrich and R. Dikau, Hydrological Processes 14(2000) 1963–1979.

[12] D.E. Walling, J. Garnier and J.M. Mouchel, Hydrobiologia 410 (1999)223–240.

[13] V. Ferro and M. Minacapilli, Hydrol. Sci. J. 40 (1995) 703–717.[14] U.C. Kotyahari and S.K. Jain, Hydrol. Sci. J. 42 (1997) 833–843.[15] D.B. Beasley and L.F. Huggins, ANSWERS Users Manual, EPA-

905/9-82-001, U.S. EPA, Region V, Chicago, IL (1981).[16] R.A. Young, C.A. Onstad, D.D. Bosch and W.P. Anderson, Agricul-

ture Non-Point Source Pollution Model: A Watershed Analysis Tool(ARS, USDA, Morris, MN, 1986).

[17] J.G. Arnold, J.R. Williams, R.H. Griggs and N.B. Sammon,SWRRBWQ: A Basin Scale Model for Assessing Management Im-pacts on Water Quality (USDA-ARS, West Lafayette, IN, 1991).

[18] J.G. Arnold, J.R. Williams and D.R. Maidment, ASCE Journal of Hy-draulic Engineering 121 (1995) 171–183.

[19] B. Sujata, S. Sudhakar and V.R. Desai, Journal of Indian Society ofRemote Sensing 27 (1999) 155–166.

[20] M.A. Khan, V.P. Gupta and P.C. Moharana, India Journal of Arid En-vironments 49 (2001) 465–475.

[21] J.C. Sharma, J. Prasad, S.K. Saha and L.M. Pande, Indian Journal SoilConservation 29 (2001) 7–13.

[22] Anonymous, Methodology of priority delineation survey, All Indiasoil and land use survey, Ministry of Agriculture, New Delhi (1991).

[23] R.L. Karale, Y.P. Bali and K.K. Narula, J. Indian Soc. Soil Sci. 25(1977) 331–336.

[24] Y.P. Bali and R.L. Karale, A Sediment Yield Index for Choosing Pri-ority Basins (IAHS-AISH Pub., Paris, 1977).

[25] J.S. Bali, Problems in watershed management in various River ValleyProjects (RVPs), in: Proceedings of the National Symposium on R.S.in Development and Management of Water Resources (SAC, Ahmed-abad, 1983).

[26] S.P. Gawande, Overview of remote sensing application in soil andland resource management in India, in: Proceedings of the NationalSymposium on R.S. for Agriculture Application (New Delhi, 1990).

[27] National Remote Sensing Agency, Integrated mission for sustainabledevelopment, Guidelines for field survey and mapping, NRSA (1995).

[28] J.G. Arnold, J.R. Williams, A.D. Nicks and N.B. Sammon, SWRRB:A Basin Scale Simulation Model For Soil And Water Resources Man-agement (A&M Press, Texas, 1989).

[29] A.P.J. De Roo, C.G. Wesseling, V.G. Jetten and C.J. Ritsema, LISEM:A User Guide (Version 3.0) (Department of Physical Geography,Utrecht University, The Netherlands, 1995).

[30] K.S.S. Prasad, S. Gopi and R.S. Rao, International J. Remote Sensing14 (1993) 3239–3247.

[31] G.S. Sidhu, T.H. Das, R.S. Singh, R.K. Sharma and T. Ravishankar,Indian Journal of Soil Conservation 26 (1998) 71–75.

[32] S.A. EI-swaify, E.W. Dangler and C.L. Armstrong, Soil Erosion byWater in Tropics, Research Extension Series, Vol. 24 (University ofHawaii, College of Agriculture, 1982).

[33] R. Srinivasan and J.G. Arnold, American Water Resources Associa-tion 30 (1994) 453–462.

[34] M. Di Luzio, R. Srinivasan and J.G. Arnold, An integrated user in-terface for SWAT using ArcView and Avenue, in: Proceedings An-nual International Meeting of ASAE in Minneapolis, Paper No 972235(1997).

[35] R. Kaur and D. Dutta, Indian J. Soil Cons. 30 (2002) 1–7.[36] W.H. Wischmeier, C.B. Johnson and B.V. Cross, Journal of Soil &

Water Conservation 26 (1971) 189–193.[37] USDA Soil Conservation Service, National Engineering Handbook

(U.S. Dept. of Agriculture, Washington, DC, 1972) Section 4.[38] R.A. Green and D. Stephenson, Hydrological Sciences J. 31 (1986)

395–411.[39] J.E. Nash and J.V. Sutcliffe, Journal of Hydrology 10 (1970) 282–290.[40] R. Srinivasan, T.S. Ramanarayanan, J.G. Arnold and S.T. Bednartz,

J. Am. Water Resources Assoc. 34 (1998) 91–101.[41] K.W. King and J.G. Arnold, Trans. ASAE, Paper No. 98-2228 (1998).[42] G. Singh, R. Babu and S. Chandra, Soil Loss Prediction Research in

India, Bulletin No. T-12/D-9 (Central Soil and Water ConservationResearch and Training Institute, Dehradun, India, 1981).

Documents

Comparison of a Subjective and a Physical Approach for Identification of Priority Areas for Soil and Water Management in a Watershed – A Case Study of Nagwan Watershed in Hazaribagh