使用有效的偏差订正方法以及将非平稳数据平稳化，能够提升对气温分析的科学准确性，以深入揭示其时空分布特征及演变规律。本研究使用 1970-2014年 ERA5_Land近地表（2 m）月平均气温观测数据集，首先利用泰勒图、泰勒指数、年际变率评估指数、秩打分法对国际耦合模式比较计划第六阶段（CMIP6）的 6 种气候模式和多模式集合（MME）平均模式进行评估及优选，然后用 Delta 偏差订正法和Normal 分布匹配法对较优模式进行订正，最后分析 SSP1-2. 6、SSP2-4. 5 和 SSP5-8. 5 情景下青藏高原2015-2100年气温时空变化特点。结果表明：（1）本文选用的6种CMIP6模式及MME平均模式中，ECEarth3模式模拟气温效果最优。（2）将EC-Earth3模式进行Delta偏差订正后的结果与观测结果对比，其确定性系数和纳什效率系数的区域平均值分别为0. 992和0. 983，而用Normal分布匹配法订正后，其确定性系数和纳什效率系数的区域平均值分别为0. 990和0. 978，相比之下，Delta偏差订正对模式月气温的订正效果更优。（3）通过 EOF-EEMD 结合发现，三种情景下第一典型场年气温呈现全区一致变化，且SSP1-2. 6和 SSP2-4. 5情景下存在共同气温变化敏感区，即藏北高原中部地区；第二典型场气温呈现以扎曲河上游区域逐渐向四周反相变化，其中 SSP1-2. 6情景下高原整体呈东部降温西部升温，SSP2-4. 5和 SSP5-8. 5情景下高原先东部增温西部降温，之后则东部降温、西部增温。本研究可为气候模式数据在青藏高原地区的准确应用提供偏差订正方法的参考，并为深入评估青藏高原气温变化对水资源、生态系统和环境的影响提供了关键的基础信息支持。

Using effective bias correction methods and transforming non-stationary data to stationary can enhance the scientific accuracy of temperature analysis，allowing for a deeper understanding of its temporal and spatial distribution characteristics and evolution patterns. This study utilizes the ERA5_Land near-surface（2 m） monthly mean temperature observation dataset covering the period from 1970 to 2014. Initially，it employs the Taylor diagram，Taylor index，interannual variability skill score，and rank scoring method to evaluate and select among six climate models from the International Coupled Model Intercomparison Project Phase 6（CMIP6）and the multi-model ensemble（MME）average models. Subsequently，the superior models are refined using the Del‐ ta bias correction method and the Normal distribution matching method. Finally，the study analyzes the temporal and spatial temperature variation characteristics of the Qinghai-Xizang Plateau from 2015 to 2100 under the SSP1-2. 6，SSP2-4. 5，and SSP5-8. 5 scenarios. The results indicate that：（1）Among the six CMIP6 models and the multi-model ensemble（MME）average models analyzed in this study，the EC-Earth3 model demonstrates the most effective performance in simulating temperature.（2）When comparing the Delta bias correction results of the EC-Earth3 model with observational data，the regional averages of the coefficient of determination（R²） and the Nash-Sutcliffe efficiency coefficient（NSE）are 0. 992 and 0. 983，respectively. After applying the Nor‐ mal distribution matching method for correction，the regional average values of R² and NSE are 0. 990 and 0. 978，respectively. This comparison reveals that the Delta bias correction method exhibits superior correction efficacy for the model's monthly temperature.（3）According to the combination of EOF-EEMD，the annual temperature of the first typical field of the three scenarios changes uniformly in the whole region，and there is a common sensitive area of temperature change under SSP1-2. 6 and SSP2-4. 5 scenarios，that is，the central region of the Qiangtang Plateau. The temperature dynamics in the second typical field reveal a gradual reverse-phase change from the upper reaches of the Zhaqu River to surrounding areas. Under the SSP1-2. 6 scenario，the plateau experiences overall cooling in the east and warming in the west. Conversely，under the SSP2-4. 5 and SSP5- 8. 5 scenarios，the plateau initially warms in the east and cools in the west，followed by a subsequent cooling in the east and warming in the west. This study provides a reference for bias correction methods that enhance the ac‐ curate application of climate model data in the Qinghai-Xizang Plateau region and offers essential foundational in‐ formation for a comprehensive assessment of the impacts of temperature changes on water resources，ecosystems，and the environment in this area.