Water Balance Approach for Watershed Discretization and Parameter Simulation

Abstract

The simulation results of the hydrological model are impacted by number, size, and scale of sub-basins. The rationale of this study is to evaluate the impacts of size and number of sub-basins used in the models to simulated river flow using the water balance approach. The study area Bagmati River Basin (BRB), impacted by natural and anthropogenic activities, lies in the central region of Nepal and basin area is 3892.4 km2. We divided the basin into four different level of spatial discretization limiting to the threshold area of 20, 30, 50 and 150 sq. km resulting to 114, 68, 47 and 20 number of sub-basins respectively. Three observed discharge locations were used for analyzing the impact of the spatial discretization to simulated flow. Each location represents the different topography, hilly terrain (slope > 20%), mid plain terrain (5% < slope < 20%) and plain terrain (slope < 5%) from top to bottom respectively. These impacts were analyzed with different temporal scale such as daily, monthly and annual scale too. The results show that average annual river flow increases significantly with an increase in the number of sub-basins. Furthermore, more numbers of discretization seem to be possible to represent basin characteristics more realistic. The water balance for the 16 sub modeling scenario using the cross-validation approach was tested. The result showed that discretization of the watershed has significant (p < 0.05) impact on water balance components. We also observed that impacts were related to different factors, such as geographical and topographical factors, and level of impacts was varied at different temporal scale. The sensitivity analysis of the modeling scenario was made using the different statistical parameters and the result displays that the model with the maximum number of subbasin has better performance (NSE = 0.85, PBIAS = +3.29% and R² = 0.86) than the other.

Keywords: BRB, Distributed Hydrological Model, GIS, spatial discretization, SWAT