Abstract: The distributed rainfall-runoff model has the advantage of being physically-based. However,this model is difficult to apply because of the demands on the input data,much of which is not directly measurable,and computational time. This fact has led to the development of simplified distributed models,such as the Distributed Time Variant Gain Model(DTVGM), by a combination of physical or conceptual processes with a nonlinear system approach. In the Banqiao Basin of Yellow River, considering the spatial variation of precipitation,the DTVGM makes positive response to the distributed input data and obtains higher precision in streamflow simulation than that of the lumped model. In Chaobaihe River basin of North China,by analyzing the variation of the time variant factor G for different land use types,the response trend of runoff to the land use and cover change is preliminarily showed:in case of the same soil moisture condition,the value of G decreases,which means that the surface runoff generation ability decreases with the increase of the forest cover.