CMIP6多模式对欧亚大陆夏季降水的模拟能力评估

Evaluation of the Performance of CMIP6 Multi-Models in Simulating Summer Precipitation over Eurasia

摘要: 基于全球气候降水中心(GPCC)和参与第六次耦合模式比较计划(CMIP6)的六个模式的月平均降水数据，采用模式评估的相关方法：偏差、标准差、相关系数等，评估了六个模式对欧亚大陆1979~2014年夏季降水的模拟能力。结果表明EC-Earth3模式和IPSL-CM6A-LR模式的模拟能力较好，FGOALS-g3的模拟效果较差。各模式都能较好地模拟出欧亚大陆夏季降水的空间分布特征，除西亚和中亚地区偏差较小外，其他地区都存在一定的高估或低估。各模式夏季区域平均差值不大，其区域平均降水主要以年际变化为主，模式可以模拟出降水大致的年际变化特征，但各模式均整体低估了欧亚大陆夏季降水的强度。印度、青藏高原南麓、东南亚地区和中国东南沿海地区是降水变率大值区，各模式均有一定的高估和低估，其他地区各模式模拟的降水变率与GPCC数据的比较一致。各模式在亚洲季风区均方根误差均偏大，中亚和西亚普遍很小。各模式和GPCC数据的相关性整体都较差，只有部分地区相关性为正且通过显著性检验。

Abstract: Based on the monthly mean precipitation data from the Global Precipitation Center for Climate (GPCC) and six models that participated in the Sixth Coupled Model Intercomparison Program (CMIP6), the performance of the six models in simulating the summer precipitation over Eurasia from 1979 to 2014 is evaluated using the relevant methods of model evaluation: bias, standard deviation, and correlation coefficient. The results show that the EC-Earth3 and IPSL-CM6A-LR show a better performance, and the FGOALS-g3 shows a worse performance. All models can simulate the spatial distribution characteristics of summer precipitation over Eurasia better, with some overestimation or underestimation except for West and Central Asia, where the deviation is small. The regional average summer difference of each model is not large, and its regional average precipitation is mainly dominated by interannual variability, and the model can simulate the approximate interannual variability, but each model underestimates the intensity of summer precipitation over Eurasia as a whole. India, the southern foothills of the Tibetan Plateau, the Southeast Asian region, and the southeast coastal region of China are areas with large values of precipitation variability, and each model has some overestimation and underestimation, while the precipitation variability simulated by each model in the other regions is more consistent with the GPCC. The root-mean-square errors of the models are large in the Asian monsoon region and generally small in Central and West Asia. The correlations between the models and the GPCC are generally poor, with only some regions having significant correlations.