Testing a new hybrid runoff generation module in four typical watersheds across different hydrometeorological zones in China

Runoff generation in semi-humid regions is always characterized by a complex nonlinear process influenced by both saturation excess mechanism and infiltration excess mechanism. A hybrid runoff generation module is proposed in this study to delineate the mixed rainfall-runoff process by integrating an infiltration module, based on a modified Horton equation, with the saturation excess runoff generation module of Xinanjiang model at grid scale. A new distributed hydrological model, termed grid-Xinanjiang-infiltration-excess (GXAJ-IE) model, is subsequently developed in the context of grid-Xinanjiang model. Not only in semi-humid regions, but GXAJ-IE model is also expected to achieve acceptable performance in other hydrometeorological zones due to its superimposed runoff generation structure. Thus GXAJ-IE model is tested in four watersheds across different hydrometeorological zones (humid, semi-humid, semi-arid and arid) of China, and two models with single runoff generation mode, grid-Xinanjiang (GXAJ) model and grid-infiltration-excess (GIE) model, are set as benchmarks for comparison purpose. The results indicate that compared with the two benchmark models, GXAJ-IE model has higher flexibility and robustness in reproducing the flood hydrographs, especially the flood peaks, driven by various rainfall patterns in the semi-humid Dongwan and Maduwang watersheds. Furthermore, GXAJ-IE model could well capture the spatiotemporal characteristic of the saturation and infiltration excess runoff components, and delineate the evolution of their contributing areas within a flood event. Yet rainfall input with low spatiotemporal resolution still remains a limitation to give full play to the advantage of GXAJ-IE model. None of the models performs well in the arid and semi-arid Suide watershed, even though, GXAJ-IE model shows comparable simulation accuracy with GIE model whereas GXAJ model absolutely loses its edge. In the humid Tunxi watershed, GXAJ-IE model produces comparably good performance with GXAJ model while GIE model is slightly inferior. Overall, GXAJ-IE model is fairly adaptable to different hydrometeorological regions in China and shows great potential for universal application, with an especially promising prospect in improving the flood forecasting accuracy for the semi-humid watersheds.