An Improved Optimization Scheme for Representing Hillslopes and Depressions in Karst Hydrology

Abstract Understanding hydrological processes is essential for management of water resources and for promoting catchment sustainability. In karst regions, landscapes are spatially heterogeneous and include discontinuous soil distribution, and complex networks of matrices and conduits in hillslopes and depressions, which result in variation in hydrological processes. However, most previous studies have mainly focused on the effects of the distribution of soil depth and the fast‐slow flow in the matrices and conduits on hydrological processes and ignored the different hydrological processes related to hillslopes and depressions. This study improved the VarKarst model by adding randomly distributed soil and epikarst depths (RSE), fast‐slow flow (FS), and hillslopes and depressions (HD). The improved model was calibrated and validated in six large catchments (1,213–5,454 km 2 ) and one small catchment (1.25 km 2 ). Results showed that the schemes by combining FS and HD (Scenario FS + HD) and combining RSE, FS, and HD (Scenario RSE + FS + HD) improved model performance (calibrated and validated KGE ranged from 0.54 to 0.89 and AIC ranged from −336.49 to 669.77) compared models that included other schemes (original VarKarst, Scenario RSE, Scenario FS, Scenario HD, and Scenario RSE + FS), especially when reproducing discharge of peaks and recessions. These results suggest that there is a need to separate the hillslopes and depressions when modeling karstic hydrological processes.