在干旱区，冰雪融化是水资源形成的主要来源，为此，积雪资源的形成、转化与利用是新疆水资源开发利用研究的重要内容，而水文模型是水资源形成与转化量确定的关键途径。以中国科学院天山积雪站野外试验区为研究基地，以气象数据为自变量，以融雪量为因变量，研究了基于气温变化的融雪模型，并对所建单因素简易模型进行了率定和验证，同时分析了试验区多年融雪变化规律以及融雪对气温的响应过程。结果表明：在一定的低温状态下，冬季仍有融雪发生，在天山山区本项目研究流域积雪消融的日平均气温临界值约为-7 ℃，当日平均气温低于-7 ℃时，融雪基本处于暂停状态，体现了干旱区融雪特征。在模型方面，基于气温构建的单因素简易融雪模型在模拟山区融雪量时呈现出良好的代表性，在率定期（2016—2020年），融雪量观测值与模拟值间的相关性参数偏差、平均绝对误差、均方根误差、纳什效率系数和决定分别为-0.037、0.367、0.482、0.870和0.876；而验证期的值分别为-0.210、0.292、0.577、0.845和0.811。验证期的模拟结果和相关性系数显示，该模型的模拟值与观测值具有良好的一致性和稳定性，其优点是通过易获取的气象数据就可以估算流域融雪量。研究成果为干旱区积雪消融计算提供较为简便的算法，也为水文模型提供简便有效的融雪子模块。本研究对于掌握该地融雪变化规律及后续的融雪径流模拟及预测具有重要的参考价值。

In arid areas，snowmelt is the main source of water resource formation. Therefore, the formation, transformation and utilization of snow resources are important research topics on the development and utilization of water resources in Xinjiang, China. Further, hydrological model is a key to determine the formation and transformation of water resources. This study uses the field experimental area of Tianshan Snow Station, Chinese Academy of Sciences, as the experimental area. The meteorological data and snowmelt amount were adopted as the independent and dependent variables. A snowmelt model based on temperature change is studied, and a single-factor simple model is calibrated and validated. Simultaneously, the variation law of snowmelt in the experimental area for many years and the response process of snowmelt to temperature are analyzed. The results indicate that snowmelt still occurs in winter in certain low-temperature ranges. In the study area of Tianshan Mountains, the critical value of daily average temperature of snowmelt is approximately -7 ℃. When the temperature is lower than -7 ℃, snowmelt is basically suspended, reflecting the characteristics of snowmelt in arid areas. In terms of the model, a single factor simple snowmelt model based on temperature exhibits good performance in simulating snowmelt in mountainous areas. During the calibration period (2016—2020), the correlation parameters of Bias, MAE, RMSE, NSE and R² between the observed and simulated snowmelt values were -0.037, 0.367, 0.482, 0.870, and 0.876 respectively. The values of validation period were -0.210, 0.292, 0.577, 0.845, and 0.811, respectively. The simulation results and correlation coefficients during the validation period indicate that the simulated value of the model has good consistency with the measured value. The advantage is that the amount of snowmelt in the basin can be estimated through the available meteorological data. The results provide a relatively simple algorithm for snowmelt calculation in arid areas and a simple and effective snowmelt submodule for a hydrological model. This study has an important reference value for understanding the change law of snowmelt and the simulation and prediction of snowmelt runoff in subsequent studies.