21世纪前期长江中下游流域极端降水预估及不确定性分析

在全球变暖背景下，极端降水的频率、强度以及持续时间均在显著增加，尤其是对于气候变化敏感的长江中下游流域。由于模式本身、温室气体排放情景以及自然变率存在较大的不确定性，因此未来预估变化的不确定性一直备受关注。为了能够得到对于未来极端降水更为准确的预估结果，使用NEX-GDDP（NASA Earth Exchange Global Daily Downscaled Projections）提供的19个CMIP5降尺度高分辨率数据（0.25°×0.25°），给出21世纪前期（2016—2035年）长江中下游流域极端降水的可能变化。根据长江中下游流域178个气象站1981—2005年的逐日降水量数据，计算了能够代表极端降水不同特征的指数，在评估模拟能力的基础上给出了21世纪前期RCP4.5情景下极端降水的变化。结果表明，降尺度结果对长江中下游流域极端降水有很好的模拟能力，除R90N外，所有模式模拟其余指数的空间结构与观测的相关系数均超过了0.6。其中所有模式模拟PRCPTOT和R10的相关系数均超过0.95。21世纪前期，长江中下游地区降水趋于极端化，尤其是在流域的西部地区。极端降水日数的变化在减少，表明对于极端降水的贡献主要来自于极端降水日的较大日降水量，而非极端降水日数。未来预估不确定性的大值区主要位于流域的南部地区，流域的西部地区不确定性较低，西部地区极端降水的增加应该受到更多的重视。 

Projection and uncertainties of extreme precipitation over the Yangtze River valley in the early 21st century

Intensity, frequency and duration of extreme precipitation would increase in the future under a warming climate. This phenomenon will be specifically significant in the Yangtze River valley, where precipitation is sensitive to climate change. Due to uncertainties in models results, the accuracy of extreme precipitation projection is still an open issue. In order to better understand possible changes of extreme precipitation over the middle and lower reaches of Yangtze River, this study evaluates the performance of 19 CMIP5 GCMs from the NASA Earth Exchange Global Daily Downscaled Projections (NEX-GDDP) in simulating extreme precipitation and compare the results with observations at 178 stations in Yangtze River valley during 1981-2005. The projection under the RCP4.5 scenario for the early 21st century is then given. The results show that the downscaling models have excellent performance in simulating extreme precipitation in this region; the spatial correlation coefficients between all models and observations are larger than 0.6 except the R90N, while the correlation coefficients of PRCPTOT and R10 with observations are even higher than 0.95. Extreme precipitation tends to increase during the early 21st century, especially over the western part of the Yangtze River valley. The days of extreme precipitation (R90N) would decrease and the R95T and PRCPTOT would increase, which indicates that the amount of extreme precipitation is mainly attributed to contributions of the R95T instead of more days of extreme precipitation. The largest uncertainties occur in southern part of the region while the smallest uncertainties are found in the western part of the region. More attention should be paid to the increase of extreme precipitation in the western part.