Hydrologic Modeling of the Blue River Basin Using NEXRAD Precipitation Data with a Semidistributed and a Fully Distributed Model

摘要

A semidistributed (DPHMRS) and a fully distributed (MISBA) physically based hydrologic models were applied to the Blue River Basin (BRB) of Oklahoma in a multiyear mode using next-generation radar (NEXRAD) precipitation data. DPHMRS generally over-simulated low flows of BRB during the validation stage partly because rainfall events based on radar precipitation data are likely too high. However, by adjusting the NEXRAD data based on Mesonet gauge measurements, the Nash-Sutcliffe efficiency (Ef) in the validation run is significantly improved from -0.16 to 0.46. The model performance of DPHMRS also depends on the subbasin resolution, and it was found that DPHMRS can optimally model BRB with seven subbasins. Goodness-of-fit statistics of streamflow simulated by both models demonstrate better performance of MISBA compared to DPHMRS in calibration (E_f improved from 0.53 to 0.82) and validation (E_f improved from 0.46 to 0.83) stages. This difference is partly due to spatially distributed information of soil, land use, and precipitation data for BRB that are partially averaged out in the subbasin framework of DPHMRS, while such information are better retained in the fully distributed framework of MISBA. It seems a fully distributed hydrologic model can more fully take advantage of the spatially distributed information of input data for BRB than a semidistributed model if such detailed data are available. However, this may not be necessarily true for river basins where either data are limited, or where river basins have fairly uniform land use and nonmountainous terrain characteristics.