Assessment Of The Precipitation Over China Simulated By CMIP5Multi-models

As one of the major tools to better understand the phenomenon of the climate system andto project the future climate change, climate models have been widely used in presentresearches on hydrology, ecology, environment, agriculture, etc. However, due to thecomplexity of the climate system, the models’ ability to simulate the climate system is far fromperfect, and large uncertainties exist in the simulation of precipitation. The latest models thatparticipated in the fifth phase of Coupled Model Intercomparison Project (CMIP5) have justbeen released, and the outputs from these models will be widely used in different disciplines.Thus it is necessary to evaluate the performance of these models comprehensively andsystematically, so as to provide reference and suggestions not only for model improvement, butalso for the researchers from different fields who will use the results from climate models.In this study, the ability of46CMIP5models in simulating the precipitation during thesecond half of20th century is evaluated. Based on multiple observations and different metrics,the ability in simulating both the mean and extreme precipitation in China from CMIP5modelsis quantified and compared on different temporal and spatial scales. Moreover, both the meanstate and the interannual variability of the summer monsoon circulation system are also studied,so as to better understand the bias of CMIP5models in simulating the summer precipitation inthe eastern part of China. Based on the different metrics, the models with rather betterperformances are chosen to project the future change of mean and extreme precipitation inChina, and the results are compared with the results projected by all the models. The mainconclusions of this paper are summarized as follows:The CMIP5models can well simulate both the temporal variation and the spatialdistribution of precipitation in China. Most models can simulate the similar magnitude ofinterannual variability of mean precipitation averaged over China and its eight sub-regions withrespect to observation. The strength of periodic feature and the phase of decadal variation canbe captured by nearly half of the models. As seen from multi-model ensemble mean (MME),the distribution of simulated seasonal and annual mean precipitation shows similar pattern withthat of the observation, but the overestimate in the Qinghai-Tibet Plateau and the underestimatein South China that have already been found in CMIP3models still exist in the latest models.Most models can well simulate the EOF lead mode of seasonal and annual mean precipitationover China, and the lead mode of winter mean precipitation can be better captured by CMIP5models than that of summer couterpart.Judged from a set of indices that have been widely used in describing precipitationextreme events, the CMIP5models have limited ability in simulating the temporal trend ofextreme precipitation over China. Except for consecutive dry days and extremely wet days, the observed significant trend of changes in extreme precipitation cannot be represented in mostmodels. The MME of the climatology of extreme precipitations simulated by CMIP5modelsshows similar spatial distribution with that of the observation, in spite of the systematic bias ofthe overestimate in Northwest China and Qinghai-Tibet Plateau and the underestimate in SouthChina.The underestimate of summer mean precipitation in Jianghuai region may be attributed tothe simulation of a weaker East Asia summer monsoon circulation including less moisturetransport from South China Sea and Tropical Pacific to the South China, weaker subtropicalhigh, and weaker high-altitude jet stream simulated in CMIP5models. Such bias is moreprominent during strong monsoon years than during weak monsoon years, so that the modelstend to underestimate the magnitude of interannual variability of East Asia summer monsooncirculation and the summer mean precipitation over Jianghuai region.The projection of future changes in mean and extreme precipitation over China is madefrom the results of both the MME of all the models (ALL) studied in this paper and the meanof the best five models (BEST5) that show better performance in simulating the presentprecipitation judged from the metrics that can quantify the models’ performance. The ALL andBEST5project similar changes in the annual mean precipitation averaged spatially over Chinaduring the21st century, and the trend in each RCP scenario nearly follows the correspondingradiative forcing. ALL project that the precipitation will increase most significantly in the eastof Xinjiang Province and Northeast China, while BEST5project a similar pattern of futureprecipitation change with more regional detail. According the projection from both ALL andBEST5, more frequent and intense extreme precipitation events will happen in China duringthe21st century.