A sub basin scale dynamic runoff-generation identification method was proposed to address the complex runoff mechanism and unclear response relationships of key water cycle elements in the Qinghai Tibet Plateau. Based on MODIS satellite remote sensing products and the classification results of dominant runoff-generation modes in sub-basins, the distributed hydrological model iRainSnowHydro was applied for rainfall-snowmelt runoff simulation and water source division. Quantitative analysis was conducted on the seasonal variation characteristics of runoff composition in the Zhimenda and Shigu sub-basins, as well as the autocorrelation coefficients, partial autocorrelation coefficients, and cross-correlation coefficients of precipitation and runoff at different response times. The results show that the proposed method can effectively identify the seasonal dominant runoff patterns in each sub-basin of the Jinsha River Basin, accurately simulate the runoff process of the basin, and achieve a Nash efficiency coefficient of 0.75. Interflow and groundwater are identified as the dominant runoff components in the Zhimenda and Shigu sub-basins, while the proportion of surface runoff increases significantly during spring and summer. The decay time of surface runoff and base flow is 3-10 and 10-80 d, respectively.