青藏高原地区气候特殊、地形复杂，气象观测站点稀少，对其区域气候和水循环过程的观测和模拟存在很大的困难。本文基于RegCM模式和WRF模式，探究不同水平分辨率10 km、25 km、50 km下区域气候模式对该地区1989—2008年气候时空分布规律的模拟能力。研究结果表明：在10 km水平分辨率下，RegCM模式模拟多年平均气温绝对误差为0.33 ℃，WRF模式模拟绝对误差为1.77 ℃，比25 km和50 km水平分辨率下绝对误差减少1.60~2.12 ℃，且四季气温模拟值与实测值的相关性有所提高；随着水平分辨率的提高，WRF模式对青藏高原东南部和南部的降水量高估有所改善，RegCM模式模拟值逐渐接近实测值（模拟年降水量相对误差由169%降至75%）且对高原北部降水量的模拟有所改善，但整体上模式对降雨的高估依然存在；两个模式随水平分辨率的提高对地形起伏最大的雅鲁藏布江源区降水量的误差减少最为明显。本研究可为揭示气候变化下青藏高原水文响应机理奠定基础。

The Qinghai-Tibet Plateau has a unique climate, complex topography, and few meteorological observation stations, which makes it difficult to observe and simulate its regional climate and water cycle processes. Using the regional climate models RegCM and WRF, the spatial and temporal distribution of the climate in this region from 1989 to 2008 was systematically analyzed, and the simulation capability of the RegCM and WRF models was investigated at 10, 25, and 50 km horizontal resolutions in the Qinghai-Tibet Plateau. Results show that the trend of annual average temperature simulated by both models at 10 km horizontal resolution is 1.60-2.12 ℃ lower than the multiyear average temperature simulation at 25 and 50 km horizontal resolution. With increasing horizontal resolution, the simulation biases of annual and seasonal temperatures simulated by the WRF model decrease, and the cold bias of temperature in the central and western parts of the Qinghai-Tibet Plateau improves. The simulated temperature in the RegCM model at a 10 km horizontal resolution has the lowest error, and it is significantly better for simulating the spatial distribution of temperature in the Qinghai-Tibet Plateau. The correlation between the simulated temperature of both models in different seasons and the observation data has been improved. In the precipitation simulation, the WRF model at a horizontal resolution of 25 km has the best correlation with the observed data but has the largest error. With the increase of horizontal resolution, the overestimation of precipitation in the southeastern and southern Qinghai-Tibet Plateau by the WRF model has been significantly improved, and the annual precipitation simulated by the RegCM model gradually approaches the measured values (the overestimation decreases from about 2.73 times to 1.77 times). However, the overall overestimation of precipitation by both models still exists. In the simulation of the five major river sources on the Qinghai-Tibet Plateau, with increasing horizontal spatial resolution, the WRF model reduces the biases of the air temperature in the source region of the Mekong river and Salween River, whereas the RegCM model reduces the biases of the air temperature in the source region of the Brahmaputra River and Mekong river. The largest reduction in precipitation bias was achieved in the Brahmaputra River source region at 10 km horizontal resolution by the WRF and RegCM models. This study can lay the foundation for understanding the impact of climate change on the water cycle process in the Qighai-Tibet Plateau.