Hydrologic modeling response to NEXRAD and raingage spatial variability and strategic watershed management

Watershed models are fundamental to water resources management. Natural spatial and temporal precipitation patterns cause uncertainty in model response. The goal of this study was to improve model utility for supporting watershed management decisions. Specific objectives were: (1) to refine procedures for determining local bias adjustments for rainfall estimates from NEXRAD Stage III data; (2) to evaluate the uncertainty in predicted streamflow response to spatial rainfall input resolution; and (3) to demonstrate an approach for targeting areas to maximize water-quality benefits from BMP implementation. The study area was 6,316 km2 Smoky Hill River or Kanopolis Lake Watershed in central Kansas, the study period was from 1992-2002, and the hydrologic model used was SWAT.; Hourly NEXRAD Stage III rainfall depths were accumulated for 24-hr periods to match the observation time of raingage data. NEXRAD overestimated monthly average rainfall depths (45% to 184%) in warm months (April to September) and underestimated depths (33% to 64%) in cold months (December to February); depths in the remaining months were within +/-20%. Bias adjustment factor threshold limits of 0.15 (lower) and 2.0 (upper) resulted in the best agreement between daily predicted versus observed streamflows.; The bias-adjusted NEXRAD data, with original resolution of 4 km x 4 km, was aggregated in space with incrementally coarser resolutions of 8 km x 8 km, 16 km x 16 km, 32 km x 32 km, 64 km x 64 km, 124 km x 124 km, and 256 km x 256 km. There was no consistent and clear relationship between rainfall resolution and model performance, but generally, coarser resolution data gave better performance. The best performance was obtained by rainfall input resolutions of 64 km and coarser in the upstream (drier) half of the watershed, compared to 32 km and coarser in the downstream (wetter) half.; Strategic watershed management involved ranking subwatershed sediment yields and evaluating the impacts on sediment, N, and P yields of implementing reduced tillage, edge-of-field vegetative buffers, and contour-terraced practice either randomly or using targeting. Targeting resulted in greater reductions, both overland and at watershed outlet. The benefits of targeting were greater for the initial increments of BMP adoption and decreased as the proportion of BMP adoption on targeted land areas increased.