This study aims to assess the snow cover spatial and temporal variation in the Source Region of the Yellow River, to understand the effect of snowmelt on the hydrology and to evaluate the potential impact of increasing temperature on the spring snowmelt runoff. The Snowmelt Runoff Model (SRM) developed by Rango and Martinec was used and improved by adding diurnal temperature computation and snow storage computation to model the runoff for 2013, 2014, 2016 and 2017. The snow cover assessment was done using Tera-MODIS satellite images between 2001 and 2017. Results showed that spring snowmelt produces 38% of direct runoff before July and 15% of annual runoff in average. The snow cover is very sensitive to temperature and to winter precipitation. There is a significant decrease of snow cover in May since 2001 in parallel to a significant increase of temperature. The snow cover duration has decreased by more than 15 days (20%) in the permafrost area in 15 years. Simulation suggests that a temperature increase of 2°C or 4°C will reduce the spring snowmelt by at least 20% and 30% respectively, and spring peak runoff will occur up to 15 days and one month earlier. Future investigations should focus on the interaction between snow, frozen ground and ground water recharge in order to extend the knowledge on snowmelt impact to soil related processes which play a key role on the ecosystem degradation in the region.