Buffalo River Watershed SWAT Modeling

Document
Description
Watershed models were developed for the Buffalo River Watershed (BRW) for the purposes of simulating and evaluating hydrology and water quality (sediment, nutrients, and bacteria). The Soil and Water Assessment Tool (SWAT) 2009 model was used to develop separate models for the South Branch and Upper Mainstem of the Buffalo River (Figure 1). The models were developed to simulate conditions from 1995 through 2009; models were calibrated to data from 2001-2006 and validated with data from 1996-2000. Key inputs to the SWAT models include: weather (i.e., precipitation, solar radiation, temperature, relative humidity, and wind speed) data, topography, soils data, and land cover, as well as land management information representative of typical practices in the region. In addition to the SWAT models, an HSPF (Hydrological Simulation Program – FORTRAN) model is also being developed for the BRW. A key goal in developing these models (both SWAT and HSPF) was to ensure consistency in model setup and data sources used, to allow for eventual comparison of the two models for use in water quality planning and management in the BRW and Red River Valley, in general. Local information sources, such as watershed delineations based on LiDAR (light detection and ranging) data and design details of impoundments, were used whenever possible. The models performed well for hydrology for both simulated watersheds. For the Upper Mainstem and South Branch Models, respectively, mean daily flow values were within -3.7 and -31.5% of the observed mean values for the calibration period and within -6.0 and 3.4% of the observed mean values for the validation period. Additional model validation was performed for a downstream gauging station (near the watershed outlet) and predicted mean daily flow values were within -12.9% of observed data for the simulation period (1996-2006). Because observed water quality data did not coincide with concomitant flow records, modeled flow values were used in conjunction with water quality monitoring data to generate pseudo-observed loads of sediment and phosphorus against which to compare model predictions. The SWAT models did a good to excellent job predicting sediment and total phosphorus loads at most monitoring points. A notable exception is model underprediction of sediment at a monitoring point located near the watershed outlet. In contrast to generally good agreement at upland monitoring locations, under-prediction of sediment near the watershed outlet suggests that other sources of suspended solids such as re-worked channel sediments or in-stream primary productivity may be important in downstream reaches of the Buffalo River. Fecal coliform concentrations simulated with the SWAT models did not compare well against observed data and attempts to simulate bacteria in the BRW are considered unsuccessful.
Date Issued
2013
Number of Pages
65
Decade
Corporate Author
Associated Organization
Rights Holder
Minnesota Water Research Digital Library
Rights Management
Public Domain