No abstract

Alpine snowmelt is an important generation mode for runoff in the source region of the Tarim River basin, which covers four subbasins characterized by large area, sparse gauge stations, mixed runoff supplied by snowmelt and rainfall, and remarkably spatially heterogeneous precipitation. Taking the Kaidu River basin as a research area, this study analyzes the influence of these characteristics on the variables and parameters of the Snow Runoff Model and discusses the corresponding determination strategy to improve the accuracy of snowmelt simulation and forecast. The results show that: (i) The temperature controls the overall tendency of simulated runoff and is dominant to simulation accuracy, as the measured daily mean temperature cannot represent the average level of the same elevation in the basin and that directly inputting it to model leads to inaccurate simulations. Based on the analysis of remote sensing snow maps and simulation results, it is reasonable to approximate the mean temperature with 0.5 time daily maximum temperature. (ii) For the conflict between the limited gauge station and remarkably spatial heterogeneity of rainfall, it is not realistic to compute rainfall for each elevation zone. After the measured rainfall is multiplied by a proper coefficient and adjusted with runoff coefficient for rainfall, the measured rainfall data can satisfy the model demands. (iii) Adjusting time lag according to the variation of snowmelt and rainfall position can improve the simulation precision of the flood peak process. (iv) Along with temperature, the rainfall increases but cannot be completely monitored by limited gauge stations, which results in precision deterioration.snowmelt runoff model (SRM), Kaidu River basin, runoff simulation, runoff coefficient for rainfall, time lagIn the arid area of northwest China, both the snow accumulation and snowmelt play an important role in regional water resource and environment. Alpine snowmelt is the major source of many rivers and significantly contributes to the local populace and social-economic development. But when snowmelt and storms combine, the area is also easily flooded. Exploring the snowmelt rule, simulating runoff processes and forecasting runoff changes in future climates at the basin scale are significant for generating a water utility plan, the sustainable development of animal husbandry and the economy, and flood prevention.The Snowmelt-Runoff Model (SRM) is designed to simulate and forecast daily snowmelt flow in mountain basins. After it was initially developed by Martinec and Rango in the early 1980s, the SRM has been successfully applied to over 80 basins in 25 countries in northern America, western Europe and southern Asia [1] . Though the climate conditions of these regions are much different from those of the arid areas of northwest China, the model performance in arid areas is also satisfying [2][3][4] .The SRM is a relatively simple degree-day model,