为了评估高分辨率模式间比较计划（HighResMIP）中的模式对青藏高原东坡暖季降水的模拟能力，本文选取了其中16个具有不同分辨率的模式，并结合CN05. 1数据集，对青藏高原东坡暖季降水进行了全面评估。通过对比分析多个模式的输出结果与实际观测数据之间的差异，揭示了各模式在捕捉降水时空变化特征、降水强度以及与地形相关的降水机制等方面的优势和不足。研究发现，高分辨率气候模式在模拟青藏高原东坡全年及暖季降水空间分布方面表现出较高的准确性，但不同模式之间的模拟结果存在显著差异。部分模式（CMCC-CM2-HR4、CMCC-CM2-VHR4、FGOAL-f3-H）的年平均降水呈增长趋势，与观测结果一致，而其余模式则表现出稳定或减少趋势。此外，模式在降水频率和强度的模拟上仍存在明显不足。对比高分辨率组和中低分辨率组，所有模式均系统性低估弱降水事件（<1 mm∙d-1），同时高估强降水事件（>4 mm∙d-1）发生频率。中低分辨率组模拟的青藏高原东坡降相位模拟相较观测结果存在约30天的系统性偏差。相比之下，高分辨率组在降水频率的模拟表现优于中低分辨率组。综合降水的时间分布、频率及模式评分，ECWMF模式的模拟效果较好，而FGOAL-f3-H模式存在显著的负偏差。

This study evaluates the simulation capabilities of 16 models with varying resolutions from the HighResolution Model Intercomparison Project（HighResMIP）in reproducing warm-season precipitation over the eastern slope of the Qinghai-Xizang Plateau，using the CN05. 1 dataset as observational reference. Through comparative analysis of outputs from multiple models against observational data，this study elucidates model strengths and limitations in capturing spatiotemporal variability，precipitation intensity，and terrain-related mechanisms. Results indicate that high-resolution climate models demonstrate reasonable accuracy in simulating annual and warm-season precipitation spatial patterns，though notable inter-model discrepancies persist. Three models （CMCC-CM2-HR4，CMCC-CM2-VHR4，FGOAL-f3-H）successfully replicate the observed increasing trend in annual precipitation，while others exhibit stable or decreasing trends. Persistent model deficiencies emerge in simulating precipitation frequency and intensity：all models systematically underestimate light precipitation events（<1 mm∙d-1）while overestimating heavy precipitation frequency（>4 mm∙d-1）. Medium-to-low-resolution models show a systematic phase lag of approximately 30 days in diurnal precipitation cycles compared to observations. In contrast，the high-resolution models performed better in simulating precipitation frequency than the medium low resolution group. Based on comprehensive evaluation of temporal distribution，frequency characteristics，and model skill scores，the ECWMF model demonstrates superior performance，whereas the FGOAL-f3-H model exhibits significant negative biases.