2016): Differentiating between rain, snow, and glacier contributions to river discharge in the western Himalaya using remote-sensing data and distributed hydrological modeling. change, the relative contributions of rainfall, snow and glacier melt to discharge are not well 26 understood, due to the scarcity of ground-based data in this complex terrain. Here, we 27 quantify discharge sources in the Sutlej Valley, western Himalaya, from 2000 to 2012 with a 28 distributed hydrological model that is based on daily, ground-calibrated remote-sensing 29 observation. Based on the consistently good model performance, we analyzed the 30 spatiotemporal distribution of hydrologic components and quantified their contribution to 31 river discharge. Our results indicate that the Sutlej River's annual discharge at the mountain 32 front is sourced to 55% by effective rainfall (rainfall reduced by evapotranspiration), 35% by 33 snow melt and 10% by glacier melt. In the high-elevation orogenic interior glacial runoff 34 contributes ~30% to annual river discharge. These glacier melt contributions are especially 35 important during years with substantially reduced rainfall and snowmelt runoff, as during 36 2004, to compensate for low river discharge and ensure sustained water supply and 37 hydropower generation. In 2004, discharge of the Sutlej River totaled only half the maximum 38 annual discharge; with 17.3% being sourced by glacier melt. Our findings underscore the 39 importance of calibrating remote-sensing data with ground-based data to constrain 40 hydrological models with reasonable accuracy. For instance, we found that TRMM (Tropical 41 Rainfall Measuring Mission) product 3B42 V7 systematically overestimates rainfall in arid 42 regions of our study area by a factor of up to 5. By quantifying the spatiotemporal distribution 43 of water resources we provide an important assessment of the potential impact of global 44 warming on river discharge in the western Himalaya. Given the near-global coverage of the 45 utilized remote-sensing datasets this hydrological modeling approach can be readily 46 transferred to other data-sparse regions. 47 3