Multiple segmented reaches per subwatershed modeling approach for improving HSPF-Paddy water quality simulation

Abstract

Reach segmentation influences predicted water quality concentrations in water quality modeling. Many lumped or semi-distributed watershed models, including Soil and Water Assessment Tool (SWAT) and Hydrologic Simulation Program-Fortran (HSPF), simulate stream/river water quality with a single segmented reach per subwatershed (SSRS) modeling approach. A multiple-segmented reaches per subwatershed (MSRS) modeling approach was developed, and simulated BOD5 concentrations for this approach were compared with results from the SSRS modeling approach using the HSPF-Paddy model. The SSRS modeling approach has potential systemic errors for predicting BOD5 concentration even when the model is well calibrated. When the point source was loaded at the most upstream location with the same decay rate, the predicted BOD5 concentration using the SSRS modeling approach was higher compared with results for the MSRS modeling approach, and the difference between two methods increases with increasing load and decay rate. When BOD5 was loaded more downstream, BOD5 concentration using the SSRS modeling approach was lower compared with results for MSRS modeling. For a case study, simulated streamflow and BOD5 concentration for the SSRS and MSRS modeling approaches demonstrated good agreement with observed data. However, the estimated decay rate for the SSRS modeling approach was smaller than that for the MSRS modeling approach because BOD decays through total volume in the SSRS modeling approach, although BOD may be loaded anywhere in the reach. The MSRS modeling approach can minimize systematic errors and provide more detailed variation of water quality concentrations along a river length. The MSRS modeling approach does not always need to be applied to all subwatersheds but is recommended for reaches significantly polluted by point source pollution.