CMIP6对川渝地区降水特征的模拟偏差分析

Analysis of the Simulation Bias of CMIP6 on Precipitation Characteristics in Sichuan and Chongqing

摘要: 为研究CMIP6模式模拟川渝地区盛夏降水的能力，本文使用国家气象信息中心(NMIC)提供的川渝地区173个气象站点的日降水观测数据和第六次国际耦合模式比较计划(CMIP6)中三个全球气候模式(BCC-CSM2-MR、EC-Earth3和GFDL-ESM4)历史试验期间的模拟数据，从时空分布方面定量评估三个模式对川渝地区盛夏降水特征的模拟能力。结果表明：1) 总体上，在盛夏川渝地区，CMIP6三个模式模拟的平均降水、强降水均偏弱，模拟的弱降水均偏强。2) 空间分布上，CMIP6模式在川渝地区盛夏降水的模拟能力受地形因素影响明显，普遍在地势较高地区模拟降水偏强，在地势较低地区模拟降水偏弱，整体而言，EC-Earth3、GFDL-ESM4模式模拟情况较好，BCC-CSM2-MR模式模拟情况相对较差。3) 在年际变化上，BCC-CSM2-MR模式模拟情况较好，EC-Earth3、GFDL-ESM4模式模拟情况相对较差。

Abstract: In order to study the ability of CMIP6 model to simulate summer precipitation in Sichuan-Chongqing area, In this paper, we use the daily precipitation observation data of 173 meteorological stations in Sichuan and Chongqing region provided by the National Meteorological Information Center (NMIC) and the simulation data of three global climate models (BCCCSM2-MR, EC-Earth3 and GFDL-ESM4) during the historical experiment of the Sixth International Coupled Model Comparison Program (CMIP6). The ability of the three models to simulate the characteristics of summer precipitation in Sichuan-Chongqing region was quantitatively evaluated from the aspect of temporal and spatial distribution. The results show as follows: 1) In general, the average precipitation and heavy precipitation simulated by CMIP6 models are weak, and the simulated weak precipitation is strong in the Chuan-Chongqing area in midsummer. 2) In terms of spatial distribution, the ability of CMIP6 model to simulate summer precipitation in Sichuan and Chongqing region is obviously affected by topographic factors, and the simulated precipitation is generally stronger in higher terrain areas and weaker in lower terrain areas. On the whole, the simulation results of EC-Earth3 and GFDL-ESM4 models are better, the simulation of BCC-CSM2-MR model is relatively poor. 3) In terms of interannual variation, the simulation performance of BCC-CSM2-MR model is better, while the simulation performance of EC-Earth3 and GFDL-ESM4 model is relatively poor.