随着全球气候变化的加剧，研究极端气候指数变化对于理解和预估极端气候事件的发生频率和强度，以及为社会经济活动和政策制定提供科学依据，具有重要的现实意义和迫切性。本研究利用格点化观测数据CN05.1以及26个高分辨率统计降尺度的NEX-GDDP-CMIP6气候模式数据，分析了1961年至2014年中国主要极端气候指数（包括极端高温、极端低温及极端降水）的变化，评估了同期气候模式对主要气候变量及主要极端气候指数的模拟性能，并预估了不同排放情景下（SSP245和SSP585）主要极端气候指数的未来变化（2031至2090年）。主要研究结果如下：

（1）历史时期中国地区极端冷指数总天数全区域均减少，但极端低温的总天数延长；高温总天数均呈现增加趋势，持续暖期延长，日间和夜间的最高气温极值持续升高。极端降水总日数减少，但极端强降水总日数增多。极端降水总量和极端降水强度持续增强。TN10p可作为极端冷指数的代表，TX90p作为极端热指数的代表，R95p作为极端降水指数的代表。CWD作为一个独特的气候特征与其他的极端温度或降水指数的关联性不高。未来时期高排放情景下的变化速率大于中等排放情景的变化速率，冷夜比例TN10p在全区域持续降低，暖昼比例TX90p比例增加，强降水量R95p在大部分区域持续增加。

（2）基于NEX-GDDP-CMIP6数据集对温度变量的模拟效果优于对降水变量的模拟，最高气温的模拟效果优于最低气温的模拟，表征暖事件变化的极端暖指数的模拟效果整体要优于极端冷指数的模拟，尤其是针对相对阈值定义下的极端热指数以及气温极值的模拟。日最低和最高气温对北方大部分内陆地区模拟效果更好，如东北、华北、西北及青藏高原地区，但代表极端冷指数变化的TN10p指数和代表极端热指数变化的TX90p指数都在南方大部分地区表现出更好的模拟效果，如东南、西南地区。代表极端降水指数变化的R95p指数在中国北方大部分地区表现出更优的模拟效果，如青藏高原、东北及西北地区。

（3）气候模式对中国区域平均年最低气温、年最高气温和年降水量的模拟并不能完全替代对主要极端气候指数的模拟。多模式集合平均可以作为模拟极端气候指数的有效参考，其中排名前五和排名前三的模式集合平均表现出对极端气候指数最佳的模拟能力。极端降水指数可以考虑采用排名前三的模式集合平均来提高模拟准确性。

本研究对NEX-GDDP-CMIP6这一高分辨率统计降尺度数据集进行了性能评估，证明了单一气候模式对于极端气候指数模拟存在的局限性，进一步得到针对中国区域不同气候变量及极端气候指数模拟的最优方案。

As global climate change escalates, the examination of shifts in extreme climate indices holds paramount practical significance and urgency in comprehending and forecasting the frequency and intensity of extreme climate events. This endeavor also furnishes a scientific foundation for socio-economic endeavors and policy formulation. This investigation utilizes the gridded observational dataset CN05.1 in conjunction with 26 high-resolution statistically downscaled NEX-GDDP-CMIP6 climate model datasets to scrutinize alterations in key extreme climate indices across China from 1961 to 2014. The study assesses the simulation performance of climate models for crucial climate variables and primary extreme climate indices during the same timeframe, while projecting future changes (2031 to 2090) of these indices under diverse emission scenarios (SSP245, SSP585). The key research findings are outlined as follows:

(1) During the historical period, the total number of extreme cold index days in China decreased across all regions, but the total number of extremely cold days extended; the total number of hot days increased, the continuous warm period extended, and the maximum daytime and nighttime temperatures continued to rise. The total number of extreme precipitation days decreased, but the total number of extreme heavy precipitation days increased. The total amount and intensity of extreme precipitation continued to increase. TN10p can serve as a representative of the extreme cold index, TX90p as a representative of the extreme heat index, and R95p as a representative of the extreme precipitation index. CWD, as a unique climate feature, is not highly correlated with other extreme temperature or precipitation indices. The rate of change under future high emission scenarios is greater than that under medium emission scenarios, with the proportion of cold nights (TN10p) continuously decreasing across all regions, the proportion of warm days (TX90p) increasing, and the amount of heavy precipitation (R95p) continuously increasing in most areas.

(2) The simulation of temperature variables based on the NEX-GDDP-CMIP6 dataset is better than that of precipitation variables, and the simulation of maximum temperature is better than that of minimum temperature. The simulation of extreme warm indices, which characterize warm event changes, is generally better than that of extreme cold indices, especially for extreme heat indices and temperature extremes defined under relative thresholds. The simulation of daily minimum and maximum temperatures is better for most inland areas in the north, such as the Northeast, North China, Northwest and the Qinghai-Tibet Plateau, but the TN10p index representing changes in extreme cold indices and the TX90p index representing changes in extreme heat indices both show better simulation performance in most areas in the south, such as the Southeast and Southwest. The R95p index representing changes in extreme precipitation indices shows better simulation performance in most areas in the north of China, such as the Qinghai-Tibet Plateau, Northeast and Northwest.

(3) Climate models' simulation of China's regional average annual minimum temperature, annual maximum temperature and annual precipitation cannot fully substitute for the simulation of major extreme climate indices. Multi-model ensemble averages can serve as an effective reference for simulating extreme climate indices, with the top five and top three ranked model ensemble averages demonstrating the best simulation capabilities for extreme climate indices. To improve simulation accuracy, the top three ranked model ensemble averages can be considered for extreme precipitation indices.

This study evaluates the performance of the high-resolution statistically downscaled dataset NEX-GDDP-CMIP6, demonstrating the limitations of a single climate model for simulating extreme climate indices, and further identifies the optimal solutions for simulating different climate variables and extreme climate indices in China.