Application of SWAT with and without Variable Source Area Hydrology to a Large Watershed

In this study, two different versions of the Soil and Water Assessment Tool (SWAT) model were used to simulate the hydrology and biogeochemical response of the Cannonsville Reservoir watershed, in New York. The first version distributes overland flow in ways that are consistent with variable source area (VSA) hydrology driven by saturation excess runoff, whereas the second version is the standard version of SWAT. These two models were each calibrated for streamflow (Flow), particulate phosphorus (PP), total dissolved phosphorus (TDP), and sediment (Sed) against measured data from the 1,200 km² Cannonsville watershed. The standard version of the model yielded an r² between the measured and simulated data of 0.85, 0.73, 0.70, and 0.72 for Flow, Sed, TDP, and PP, respectively. The VSA version yielded an r² of 0.84, 0.69, 0.72, and 0.53 for Flow, Sed, TDP, and PP, respectively. The two models were then used to determine the maximum upper bound on the reduction in phosphorus loading by removing all of the corn in the watershed. The average reductions between the two models were 65 and 37% for PP and TDP, respectively. The VSA version was also used to estimate the effect of moving corn land in the watershed from the wettest, most runoff prone areas to the driest, least runoff prone areas, which cannot be done directly with the standard SWAT model.     

THE IMPACT OF RUNOFF GENERATION MECHANISMS ON THE LOCATION OF CRITICAL SOURCE AREAS1

ABSTRACT: Identifying phosphorus (P) source areas and transport pathways is a key step in decreasing P loading to natural water systems. This study compared the effects of two modeled runoff generation processes ‐ saturation excess and infiltration excess ‐ on total phosphorus (TP) and soluble reactive phosphorus (SRP) concentrations in 10 catchment streams of a Catskill mountain watershed in southeastern New York. The spatial distribution of runoff from forested land and agricultural land was generated for both runoff processes; results of both distributions were consistent with Soil Conservation Service‐Curve Number (SCS‐CN) theory. These spatial runoff distributions were then used to simulate stream concentrations of TP and SRP through a simple equation derived from an observed relation between P concentration and land use; empirical results indicate that TP and SRP concentrations increased with increasing percentage of agricultural land. Simulated TP and SRP stream concentrations predicted for the 10 catchments were strongly affected by the assumed runoff mechanism. The modeled TP and SRP concentrations produced by saturation excess distribution averaged 31 percent higher and 42 percent higher, respectively, than those produced by the infiltration excess distribution. Misrepresenting the primary runoff mechanism could not only produce erroneous concentrations, it could fail to correctly locate critical source areas for implementation of best management practices. Thus, identification of the primary runoff mechanism is critical in selection of appropriate models in the mitigation of nonpoint source pollution. Correct representation of runoff processes is also critical in the future development of biogeochemical transport models, especially those that address nutrient fluxes.     

HYDROLOGIC PARAMETERIZATION OF WATERSHEDS FOR RUNOFF PREDICTION USING SWAT1

ABSTRACT: The use of continuous time, distributed parameter hydrologic models like SWAT (Soil and Water Assessment Tool) has opened several opportunities to improve watershed modeling accuracy. However, it has also placed a heavy burden on users with respect to the amount of work involved in parameterizing the watershed in general and in adequately representing the spatial variability of the watershed in particular. Recent developments in Geographical Information Systems (GIS) have alleviated some of the difficulties associated with managing spatial data. However, the user must still choose among various parameterization approaches that are available within the model. This paper describes the important parameterization issues involved when modeling watershed hydrology for runoff prediction using SWAT with emphasis on how to improve model performance without resorting to tedious and arbitrary parameter by parameter calibration. Synthetic and actual watersheds in Indiana and Mississippi were used to illustrate the sensitivity of runoff prediction to spatial variability, watershed decomposition, and spatial and temporal adjustment of curve numbers and return flow contribution. SWAT was also used to predict stream runoff from actual watersheds in Indiana that have extensive subsurface drainage. The results of this study provide useful information for improving SWAT performance in terms of stream runoff prediction in a manner that is particularly useful for modeling ungaged watersheds wherein observed data for calibration is not available.     

Evaluation of a Watershed Model for Estimating Daily Flow Using Limited Flow Measurements1

Abstract: The Soil and Water Assessment Tool (SWAT) model was evaluated for estimation of continuous daily flow based on limited flow measurements in the Upper Oyster Creek (UOC) watershed. SWAT was calibrated against limited measured flow data and then validated. The Nash‐Sutcliffe model Efficiency (NSE) and mean relative error values of daily flow estimations were 0.66 and 15% for calibration, and 0.56 and 4% for validation, respectively. Also, further evaluation of the model’s estimation of flow at multiple locations was conducted with parametric paired t‐test and nonparametric sign test at a 95% confidence level. Among the five main stem stations, four stations were statistically shown to have good agreement between predicted and measured flows. SWAT underestimated the flow of the fifth main stem station possibly because of the existence of complex flood control measures near to the station. SWAT estimated the daily flow at one tributary station well, but with relatively large errors for the other two tributaries. The spatial pattern of predicted flows matched the measured ones well. Overall, it was concluded from the graphical comparisons and statistical analyses of the model results that SWAT was capable of reproducing continuous daily flows based on limited flow data as is the case in the UOC watershed.     

ROLE OF WATERSHED SUBDIVISION ON MODELING THE EFFECTIVENESS OF BEST MANAGEMENT PRACTICES WITH SWAT1

Distributed parameter watershed models are often used for evaluating the effectiveness of various best management practices (BMPs). Streamflow, sediment, and nutrient yield predictions of a watershed model can be affected by spatial resolution as dictated by watershed subdivision. The objectives of this paper are to show that evaluation of BMPs using a model is strongly linked to the level of watershed subdivision; to suggest a methodology for identifying an appropriate subdivision level; and to examine the efficacy of different BMPs at field and watershed scales. In this study, the Soil and Water Assessment Tool (SWAT) model was calibrated and validated for streamflow, sediment, and nutrient yields at the outlet of the Dreisbach (623 ha) and Smith Fry (730 ha) watersheds in Maumee River Basin, Indiana. Grassed waterways, grade stabilization structures, field borders, and parallel terraces are the BMPs that were installed in the study area in the 1970s. Sediment and nutrient outputs from the calibrated model were compared at various watershed subdivision levels, both with and without implementation of these BMPs. Results for the study watersheds indicated that evaluation of the impacts of these BMPs on sediment and nutrient yields was very sensitive to the level of subdivision that was implemented in SWAT. An optimal watershed subdivision level for representation of the BMPs was identified through numerical simulations. For the study watersheds, it would appear that the average subwatershed area corresponding to approximately 4 percent of total watershed area is needed to represent the influence of these BMPs when using the SWAT model.     

EFFECTS OF WATERSHED REPRESENTATION ON RUNOFF AND SEDIMENT YIELD MODELING1

ABSTRACT: This paper evaluates the effects of watershed geometric representation (i.e., plane and channel representation) on runoff and sediment yield simulations in a semiarid rangeland watershed. A process based, spatially distributed runoff erosion model (KINEROS2) was used to explore four spatial representations of a 4.4 ha experimental watershed. The most complex representation included all 96 channel elements identifiable in the field. The least complex representation contained only five channel elements. It was concluded that oversimplified watershed representations greatly influence runoff and sediment yield simulations by inducing excessive infiltration on hillslopes and distorting runoff patterns and sediment fluxes. Runoff and sediment yield decrease systematically with decreasing complexity in watershed representation. However, less complex representations had less impact on runoff and sediment‐yield simulations for small rainfall events. This study concludes that the selection of the appropriate level of watershed representation can have important theoretical and practical implications on runoff and sediment yield modeling in semiarid environments.     

HYDROLOGIC CALIBRATION OF THE SWAT MODEL IN A WATERSHED CONTAINING FRAGIPAN SOILS1

ABSTRACT: The Soil and Water Assessment Tool (SWAT) model, designed for use on rural ungaged basins and incorporating a GRASS GIS interface, was used to model the hydrologic response of the Ariel Creek watershed of northeastern Pennsylvania. Model evaluation of daily flow prior to calibration revealed a deviation of runoff volumes (D_v) of 68.3 percent and a Nash-Sutcliffe coefficient of-0.03. Model performance was affected by unusually large observed snowmelt events and the inability of the model to accurately simulate baseflow, which was influenced by the presence of fragipans. Seventy-five percent of the soils in the watershed contain fragipans. Model calibration yielded a D_v of 39.9 percent and a Nash-Sutcliffe coefficient of 0.04, when compared on a daily basis. Monthly comparisons yielded a Nash-Sutcliffe coefficient of 0.14. Snowmelt events in the springs of 1993 and 1994, which were unusually severe, were not adequately simulated. Neglecting these severe events, which produced the largest and third largest measured flows for the period of record, a D_v of 4.1 percent and Nash-Sutcliffe coefficient of 0.20 were calculated on a daily comparison, while on a monthly basis the Nash-Sutciffe coefficient was 0.55. These results suggest that the SWAT model is better suited to longer period simulations of hydrologic yields. Baseflow volumes were accurately simulated after calibration (D_v= -0.2 percent). Refinements made to the algorithms controlling subsurface hydrology and snowmelt, to better represent the presence of fragipans and snowmelt events, would likely improve model performance.     

Representing the Connectivity of Upland Areas to Floodplains and Streams in SWAT+

In recent years, watershed modelers have put increasing emphasis on capturing the interaction of landscape hydrologic processes instead of focusing on streamflow at the watershed outlet alone. Understanding the hydrologic connectivity between landscape elements is important to explain the hydrologic response of a watershed to rainfall events. The Soil and Water Assessment Tool+ (SWAT+) is a new version of SWAT with improved runoff routing capabilities. Subbasins may be divided into landscape units (LSUs), e.g., upland areas and floodplains, and flow can be routed between these LSUs. We ran three scenarios representing different extents of connectivity between uplands, floodplains, and streams. In the first and second scenarios, the ratio of channelized flow from the upland to the stream and sheet flow from the upland to the floodplain was 70/30 and 30/70, respectively, for all upland/floodplain pairs. In the third scenario, the ratio was calculated for each upland/floodplain pair based on the upland/floodplain area ratio. Results indicate differences in streamflow were small, but the relative importance of flow components and upland areas and floodplains as sources of surface runoff changed. Also, the soil moisture in the floodplains was impacted. The third scenario was found to provide more realistic results than the other two. A realistic representation of connectivity in watershed models has important implications for the identification of pollution sources and sinks.     

CHANGING RAINFALL-RUNOFF RELATIONSHIPS IN THE URBANIZING PEACHTREE CREEK WATERSHED,ATLANTA, GEORGIA1

ABSTRACT: Peachtree Creek is a gaged watershed that has experienced a substantial increase in urbanization. The relationships of runoff to rainfall were studied for total annual flows, low flows, and peak flows. For each type of flow the relationship in the later, more urbanized period was compared to that in the earlier, less urbanized period. An increase in total runoff in wet years was observed as urbanization increased, but a decrease occurred during dry years. For low flows a similar decrease of runoff in dry years was found. An increase in peak runoff was observed over most of the range of precipitation. Increasing peak flows and declining low flows can be adequately explained by urban hydrologic theoryshed. which focuses on the effects of urban impervious surfaces upon direct runoff and infiltration. However, a decline of total runoff in dry years can be explained only by taking into account evapotranspiration as well. The concept of advectively assisted urban evapotranspiration, previously discovered by climatologists, is needed to explain such a loss of total runoff. Urban hydrologic theory must take into account vegetation and evapotranspiration, as well as impervious surfaces and their direct runoff, to explain the magnitude of total annual flows and low flows. Urban stormwater management should address the restoration of low flows, as well as the control of floods.     

Modeling Flow,Nutrient, and Sediment Delivery from a Large International Watershed Using a Field‐Scale SWAT Model

A large international watershed, the St. Clair‐Detroit River System, containing both extensive urban and agricultural areas, was modeled using the Soil and Water Assessment Tool (SWAT) model. The watershed, located in southeastern Michigan, United States, and southwestern Ontario, Canada, encompasses the St. Clair, Clinton, Detroit (DT), Sydenham (SY), Upper, and Lower Thames subwatersheds. The SWAT input data and model resolution (i.e., hydrologic response units, HRUs), were established to mimic farm boundaries, the first time this has been done for a watershed of this size. The model was calibrated (2007–2015) and validated (2001–2006) with a mix of manual and automatic methods at six locations for flow and water quality at various time scales. The model was evaluated using Nash–Sutcliffe efficiency and percent bias and was used to explore major water quality issues. We showed the importance of allowing key parameters to vary among subwatersheds to improve goodness of fit, and the resulting parameters were consistent with subwatershed characteristics. Agricultural sources in the Thames and SY subwatersheds and point sources from DT subwatershed were major contributors of phosphorus. Spatial distribution of phosphorus yields at HRU and subbasin levels identified locations for potential management targeting for both point and nonpoint sources and revealed that in some subwatersheds nonpoint sources are dominated by urban sources.     

Riparian vegetation and channel morphology impact on spatial patterns of water quality in agricultural watersheds

A model based on theKLS factors of the Universal Soil Loss Equation (USLE) accurately predicted temporal dynamics and relative peak levels of suspended solids, turbidity, and phosphorus in an agricultural watershed with well-protected streambanks and cultivation to the stream edge. Fine suspended solids derived from surface runoff appeared to be a major component of the suspended solids in this stream. The model did not predict the same parameters in a watershed with unstable channel substrates, exposed streambanks, and heterogeneity in riparian vegetation and channel morphology. The rate of increase in concentration of the water quality parameters was higher than predicted in areas without riparian vegetation and with unstable substrates. Peak levels were higher than predicted where unstable channel substrates occurred, and potential energy of the stream was high because of stream alterations (removal of near-stream vegetation and creation of a uniform, straight channel). Timing of the peak levels of suspended solids, turbidity, and phosphorus in these areas seemed related to major flushes of discharge due to delayed inputs from the surface or subsurface or both or to rapid urban drainage. Higher suspended solids concentration in this stream seemed to involve larger quantities of large particles. Thus, the USLE may not adequately predict relative water quality conditions within a watershed when variation in channel morphology and riparian vegetation exists. We make the following recommendations:

1. Models to predict water quality effects of management programs should combine a terrestrial phase (which details hydrologic and erosion processes associated with surface runoff) with an aquatic phase (which details hydrologic processes of scour and sediment transport in channels). The impact of near-channel areas on these hydrologic processes should receive special attention.
2. Sampling schemes should be designed to account for the impact on water quality of both watershed land surface and inand near-channel processes. In order to help distinguish sources of suspended solids, researchers should emphasize analysis of size distribution of particles transported.
3. Best management systems for improving the broadest range of water resources in agricultural watersheds need to be based on an expanded “critical area” approach, which includes identification of critical erosive and depositional areas in both terrestrial and aquatic environments.

     

Integration of SWAT and HSPF for Simulation of Sediment Sources in Legacy Sediment‐Impacted Agricultural Watersheds

A total maximum daily load for the Chesapeake Bay requires reduction in pollutant load from sources within the Bay watersheds. The Conestoga River watershed has been identified as a major source of sediment load to the Bay. Upland loads of sediment from agriculture are a concern; however, a large proportion of the sediment load in the Conestoga River has been linked to scour of legacy sediment associated with historic millpond sites. Clarifying this distinction and identifying specific segments associated with upland vs. channel sources has important implications for future management. In order to address this important question, we combined the strengths of two widely accepted watershed management models — Soil and Water Assessment Tool (SWAT) for upland agricultural processes, and Hydrologic Simulation Program FORTRAN (HSPF) for instream fate and transport — to create a novel linked modeling system to predict sediment loading from critical sources in the watershed including upland and channel sources, and to aid in targeted implementation of management practices. The model indicates approximately 66% of the total sediment load is derived from instream sources, in agreement with other studies in the region and can be used to support identification of these channel source segments vs. upland source segments, further improving targeted management. The innovated linked SWAT‐HSPF model implemented in this study is useful for other watersheds where both upland agriculture and instream processes are important sources of sediment load.     

HYDROLOGIC SIMULATION OF THE LITTLE WASHITA RIVER EXPERIMENTAL WATERSHED USING SWAT1

ABSTRACT: Precipitation and streamflow data from three nested subwatersheds within the Little Washita River Experimental Watershed (LWREW) in southwestern Oklahoma were used to evaluate the capabilities of the Soil and Water Assessment Tool (SWAT) to predict streamflow under varying climatic conditions. Eight years of precipitation and streamflow data were used to calibrate parameters in the model, and 15 years of data were used for model validation. SWAT was calibrated on the smallest and largest sub‐watersheds for a wetter than average period of record. The model was then validated on a third subwatershed for a range in climatic conditions that included dry, average, and wet periods. Calibration of the model involved a multistep approach. A preliminary calibration was conducted to estimate model parameters so that measured versus simulated yearly and monthly runoff were in agreement for the respective calibration periods. Model parameters were then fine tuned based on a visual inspection of daily hydrographs and flow frequency curves. Calibration on a daily basis resulted in higher baseflows and lower peak runoff rates than were obtained in the preliminary calibration. Test results show that once the model was calibrated for wet climatic conditions, it did a good job in predicting streamflow responses over wet, average, and dry climatic conditions selected for model validation. Monthly coefficients of efficiencies were 0.65, 0.86, and 0.45 for the dry, average, and wet validation periods, respectively. Results of this investigation indicate that once calibrated, SWAT is capable of providing adequate simulations for hydrologic investigations related to the impact of climate variations on water resources of the LWREW.     

COMPARISON OF PROCESS‐BASED AND ARTIFICIAL NEURAL NETWORK APPROACHES FOR STREAMFLOW MODELING IN AN AGRICULTURAL WATERSHED1

ABSTRACT: The performance of the Soil and Water Assessment Tool (SWAT) and artificial neural network (ANN) models in simulating hydrologic response was assessed in an agricultural watershed in southeastern Pennsylvania. All of the performance evaluation measures including Nash‐Sutcliffe coefficient of efficiency (E) and coefficient of determination (R²) suggest that the ANN monthly predictions were closer to the observed flows than the monthly predictions from the SWAT model. More specifically, monthly streamflow E and R² were 0.54 and 0.57, respectively, for the SWAT model calibration period, and 0.71 and 0.75, respectively, for the ANN model training period. For the validation period, these values were ?0.17 and 0.34 for the SWAT and 0.43 and 0.45 for the ANN model. SWAT model performance was affected by snowmelt events during winter months and by the model's inability to adequately simulate base flows. Even though this and other studies using ANN models suggest that these models provide a viable alternative approach for hydrologic and water quality modeling, ANN models in their current form are not spatially distributed watershed modeling systems. However, considering the promising performance of the simple ANN model, this study suggests that the ANN approach warrants further development to explicitly address the spatial distribution of hydrologic/water quality processes within watersheds.     

EVALUATION OF RUNOFF,EROSION, AND PHOSPHORUS MODELING SYSTEM—SIMPLE1

ABSTRACT: The purpose of this study was to evaluate the performance of Spatially Integrated Models for Phosphorus Loading and Erosion (SIMPLE) in predicting runoff volume, sediment loss, and phosphorus loading from two watersheds. The modeling system was applied to the 334 ha QOD subwatershed, part of the Owl Run watershed, located in Fauquier County, Virginia, and to the 2240 ha watershed, Battle Branch, located in Delaware County, Oklahoma. Simulation runs were conducted at cell and field scales, and simulation results were compared with observed data. Runoff volume and dissolved phosphorus loading were measured at the Battle Branch watershed. Runoff volume, sediment yield, and total phosphorus loading were measured at the QOD site. SIMPLE tended to underestimate runoff volumes during the dormant period, from November to March. The comparison between observed and predicted dissolved phosphorus showed better correlation than for observed and predicted total phosphorus loading. Cell level simulations provided similar estimates of runoff volume and phosphorus loading when compared to field level simulations for both watersheds. However, observed sediment yields better compared with the values predicted from the cell level simulation when compared to field level simulation. Finally, results of model evaluation indicated that SIMPLE's predictive ability is acceptable for screening applications but not for site-specific quantitative predictions.     

WATERSHED CONFIGURATION AND GEOGRAPHIC INFORMATION SYSTEM PARAMETERIZATION FOR SPUR MODEL HYDROLOGIC SIMULATIONS1

ABSTRACT: A grid cell geographic information system (GIS) is used to parameterize SPUR, a quasi-physically based surface runoff model in which a watershed is configured as a set of stream segments and contributing areas. GIS analysis techniques produce various watershed configurations by progressive simplification of a stream network delineated from digital elevation models (DEM). We used three watershed configurations: ≥ 2nd, ≥ 4th, and ≥ 13th Shreve order networks, where the watershed contains 28, 15, and 1 channel segments with 66, 37, and 3 contributing areas, respectively. Watershed configuration controls simulated daily and monthly sums of runoff volumes. For the climatic and topographic setting in southeastern Arizona the ≥ 4th order configuration of the stream network and contributing areas produces results that are typically as good as the ≥ 2nd order network. However both are consistently better than the ≥ 13th order configuration. Due to the degree of parameterization in SPUR, model simulations cannot be significantly improved by increasing watershed configuration beyond the ≥ 4th order network. However, a range of Soil Conservation Service curve numbers derived from rainfall/runoff data can affect model simulations. Higher curve numbers yield better results for the ≥ 2nd order network while lower curve numbers yield better results for the ≥ 4th order network.     

Use of SWAT to Estimate Spatial Scaling of Phosphorus Export Coefficients and Load Reductions Due to Agricultural BMPS

Phosphorus export coefficients (kg/ha/yr) from selected land covers, also called phosphorus yields, tend to get smaller as contributing areas get larger because some of the phosphorus mobilized on local fields gets trapped during transport to regional watershed outlets. Phosphorus traps include floodplains, wetlands, and lakes, which can then become impaired by eutrophication. The Sunrise River watershed in east central Minnesota, United States, has numerous lakes impaired by excess phosphorus. The Sunrise is tributary to the St. Croix River, whose much larger watershed is terminated by Lake St. Croix, also impaired by excess phosphorus. To support management of these impairments at both local and regional scales, a Soil and Water Assessment Tool (SWAT) model of the Sunrise watershed was constructed to estimate load reductions due to selected best management practices (BMPs) and to determine how phosphorus export coefficients scaled with contributing area. In this study, agricultural BMPs, including vegetated filter strips, grassed waterways, and reduction of soil‐phosphorus concentrations reduced phosphorus loads by 4‐20%, with similar percentage reductions at field and watershed spatial scales. Phosphorus export coefficients from cropland in rotation with corn, soybeans, and alfalfa decreased as a negative power function of contributing area, from an average of 2.12 kg/ha/yr at the upland field scale (~0.6 km²) to 0.63 kg/ha/yr at the major river basin scale (20,000 km²). Editor's note : This paper is part of the featured series on SWAT Applications for Emerging Hydrologic and Water Quality Challenges. See the February 2017 issue for the introduction and background to the series.     

GIS‐Based Predictive Models of Hillslope Runoff Generation Processes1

Abstract: Successful nonpoint source pollution control using best management practice placement is a complex process that requires in‐depth knowledge of the locations of runoff source areas in a watershed. Currently, very few simulation tools are capable of identifying critical runoff source areas on hillslopes and those available are not directly applicable under all runoff conditions. In this paper, a comparison of two geographic information system (GIS)‐based approaches: a topographic index model and a likelihood indicator model is presented, in predicting likely locations of saturation excess and infiltration excess runoff source areas in a hillslope of the Savoy Experimental Watershed located in northwest Arkansas. Based on intensive data collected from a two‐year field study, the spatial distributions of hydrologic variables were processed using GIS software to develop the models. The likelihood indicator model was used to produce probability surfaces that indicated the likelihood of location of both saturation and infiltration excess runoff mechanisms on the hillslope. Overall accuracies of the likelihood indicator model predictions varied between 81 and 87% for the infiltration excess and saturation excess runoff locations respectively. On the basis of accuracy of prediction, the likelihood indicator models were found to be superior (accuracy 81‐87%) to the predications made by the topographic index model (accuracy 69.5%). By combining statistics with GIS, runoff source areas on a hillslope can be identified by incorporating easily determined hydrologic measurements (such as bulk density, porosity, slope, depth to bed rock, depth to water table) and could serve as a watershed management tool for identifying critical runoff source areas in locations where the topographic index or other similar methods do not provide reliable results.     

INTEGRATION OF GEOGRAPHIC INFORMATION SYSTEMS AND A COMPUTER MODEL TO EVALUATE IMPACTS OF AGRICULTURAL RUNOFF ON WATER QUALITY1

ABSTRACT: This study integrates an Agricultural Non-Point Source Pollution Model (AGNPS), the Geographic Resource Analysis Support System (GRASS) (U.S. Army Corps of Engineers, 1987), and GRASS WATERWORKS (a hydrologic modeling tool box being developed at the Michigan State University Center for Remote Sensing) to evaluate the impact of agricultural runoff on water quality in the Cass River, a subwatershed of Saginaw Bay. AGNPS is used to estimate the amounts, origin, and distribution of sediment, nitrogen (N), and phosphorus (P) in the watershed. GRASS and GRASS WATERWORKS are used to generate parameters needed for AGNPS from digital maps, which include soil association, land use, watershed boundaries, water features, and digital elevation. Outputs of the model include spatially distributed estimates of volume and peak runoff, overland and channel erosion, sediment yields, and concentrations of nitrogen and phosphorus. Management scenarios are explored in the AGNPS model to minimize sedimentation and nutrient loading. Scenarios evaluated include variations in crop cover, tillage methods, and other agricultural management practices. In addition, areas vulnerable to erosion are identified for best management practices.