An Alternative Approach to Overcome the Limitation of HRUs in Analyzing Hydrological Processes Based on Land Use/Cover Change

Since the concept of hydrological response units (HRUs) is used widely in hydrological modeling, the land use change scenarios analysis based on HRU may have direct influence on hydrological processes due to its simplified flow routing and HRU spatial distribution. This paper intends to overcome this issue based on a new analysis approach to explain what impacts for the impact of land use/cover change on hydrological processes (LUCCIHP), and compare whether differences exist between the conventional approach and the improved approach. Therefore, we proposed a sub-basin segmentation approach to obtain more reasonable impact assessment of LUCC scenario by re-discretizing the HRUs and prolonging the flow path in which the LUCC occurs. As a scenario study, the SWAT model is used in the Aksu River Basin, China, to simulate the response of hydrological processes to LUCC over ten years. Moreover, the impacts of LUCC on hydrological processes before and after model modification are compared and analyzed at three levels (catchment scale, sub-basin scale and HRU scale). Comparative analysis of Nash-Sutcliffe coefficient (NSE), RSR and Pbias, model simulations before and after model improvement shows that NSE increased by up to 2%, RSR decreased from 0.73 to 0.72, and Pbias decreased from 0.13 to 0.05. The major LUCCs affecting hydrological elements in this basin are related to the degradation of grassland and snow/ice and expansion of farmland and bare land. Model simulations before and after model improvement show that the average variation of flow components in typical sub-basins (surface runoff, lateral flow and groundwater flow) are changed by +11.09%, -4.51%, and -6.58%, and +10.53%, -1.55%, and -8.98% from the base period model scenario, respectively. Moreover, the spatial response of surface runoff at the HRU level reveals clear spatial differences between before and after model improvement. This alternative approach illustrates the potential bias caused by the conventional configuration and offers the possible application.