|Other Abstract||Using monthly sea surface temperature(SST), horizontal wind field, specific humidity and sea-level pressure reanalysis data from the ERA-20C and ERA-Interim reanalysis datasets of European Centre for Medium-Range Weather Forecasts(ECMWF), and monthly precipitation data from the Climatic Research Unit(CRU) global land grid analysis datasets(CRU TS v4.0) as the reference state, investigating and correcting the biases of above variables simulation of historical experiments and RCP4.5 emission scenario produced by the 24 global climate models participating in the Coupled Model Inter-comparison Project Phase 5(CMIP5). From the aspects of time and space distribution, seasonal change, the biases of global SST, Asian precipitation and the middle and lower troposphere water vapor transport in 2006-2015 are analyzed. Furthermore, based on these biases a variety of bias correction methods including simple removal of climate drift, simple linear regression(with log-transformed), simple regression with year-to-year increment, the Empirical Orthogonal Function(EOF) method and Bayesian Model Average method(BMA) are used to correct the biases of global SST, Asian precipitation and the middle and lower troposphere water vapor transport in 2006-2015 under RCP4.5 emission scenario. Finally, using anomaly correlation coefficient(ACC), root mean square errors(RMSE) and anomaly rate(AR) as evaluation indicators to evaluate corrected results, and screening out the relatively preferable methods. The global SST, Asian precipitation and middle and lower troposphere water vapor transport are projected by the chosen correction methods in the next 30 years(2016-2045). The main conclusions are as follows:The multi-model ensemble has a capacity of projecting the global SST of spatial distribution and the seasonal variation in 2006-2015. The multi-model ensemble SST is overestimated about 2K over the equatorial and low latitude sea areas of the Middle East Pacific and middle latitude sea areas of the southern hemisphere, while underestimated 1-2K over North Pacific, the western equatorial Pacific, North Atlantic Ocean and the Indian Ocean. In addition, the multi-model ensemble SST is warmer in winter than in summer. The test of multi-model ensemble SST projection in 2006-2015 shows that the correction of can effectively reduce bias, and after the correction the deviation of most sea areas is within 1K. After the correction the overestimated sea areas are mainly distributed in the equatorial and low-latitude, especially the equatorial Pacific, while the underestimated sea areas are distributed in the north of North Pacific and North Atlantic Ocean, and most of the sea areas in the mid-latitude of the southern hemisphere. Statistics of single sea area shows that the RMSE and AR in the Pacific and Atlantic Oceans are larger than those in the Indian Ocean. Compared with the average of 1976-2005, the global SST will generally increase in the next 30 years. In the north of North Pacific and North Atlantic Ocean, SST will increase by about 1K, 0.5-1K in the equatorial and low-latitude sea areas, and about 0.5K in the mid-latitude of the southern hemisphere.Under the RCP4.5 emission scenario, the multi-model ensemble precipitation is overestimated in most parts of northern Asia, and it is overestimated over 60%. The annual and warm season precipitation is underestimated 50-60% in the Indian Peninsula and the southeast of China, and the cold season precipitation is generally overestimated in Asia. The annual precipitation deviation is reduced to about ±20% after the correction by linear regression with log-transformed, and the precipitation deviation in cold season was slightly larger than that in annual and warm season. Corresponding to different regression methods, the anomaly and anomaly same-sign distribution of annual, and test of evaluation indicators show that the simple linear regression with log-transformed is superior to the simple linear regression with year-to-year increment in northern Asia, while it is opposite in southern Asia. It indicates that the precipitation has different statistical properties in the middle and low latitudes of Asia. The regression correction method for model precipitation is regional, it is necessary to chose the optimal correction method in different areas. Using the regional combination regression method, which is using single regression with year-to-year increment in southern Asia(the south of 31 ºN and the east of 60 ºE ), meanwhile using single regression with log-transformed in the rest region of Asia, the precipitation ACC and AR in 2006-2015 is increased to 0.218 and 59.8％respectively, larger than any linear regression correction result. The projected precipitation for 2016-2045 is corrected by using the region combination regression method. The result shows that, compared with the average of 1976-2005, the annual and warm season precipitation will increase by 10-40％ in Indian Peninsula, Indo-China Peninsula and the central northwest and southeast of China in the next 30 years, while in the southwest of the Tibet Plateau, the southwest region and the northern of northwest in China, the middle of the region south of the Yangtze River and the northeast of China will decrease by 10-40％.In the cold season, the precipitation will decrease by 20-40％ in most of Indo-China Peninsula and Indian Peninsula, most of southwest, southeast and northeast China.Considering the influence of topography, the monsoon can be approximately expressed as the middle and lower troposphere water vapor transport flux. Although the CMIP5 multi-model ensemble can well reproduce the seasonal variation of the middle and lower troposphere water vapor transport of Asia, there is still an obvious bias between simulation and observation. The multi-model ensemble water vapor transport of Asia in the middle and lower troposphere is weaker in summer but stronger in winter in 2006-2015. The water vapor transport is underestimated about 120kg/(m·s) in the Arabian Sea and the Indian Peninsula in summer, but it is overestimated 100-120kg/(m·s) near the Tibet Plateau. In winter, water vapor transport is overestimated 100-150kg/(m·s) in the Indian Ocean, the East China Sea and its nearby North Pacific. Using correction methods including simple removal of climate drift, linear regression and regression with year-to-year increment, the deviation of water vapor transport is reduced, but the large deviation is still occur in some areas, such as India Peninsula, the Indo-China Peninsula, the Bay of Bengal and coastal area of China. The corrected results of the above three correction methods are superior to the EOF method. In the next 30 years, water vapor transport in the middle and lower troposphere will increase by about 40kg/(m·s) in the south of Asia, in other areas it will increase within 20kg/(m·s). The increase of water vapor transport in summer is mainly distributed from the equator to the Arabian Sea, the Indian Peninsula, Indo-China Peninsula, and extending to the East China Sea and the South China Sea. In winter, it is mainly distributed in the Arabian Sea, the equatorial western Pacific and the Sea of Japan and the North Pacific.|
In a word, under the global warming background, the correction and projection analysis of the precipitation and monsoon in Asia indicate that the Asian monsoon will show a significantly change in the next 30 years. In the western of Asia, the north of South Asian monsoon region, the north of Indo-China Peninsula, the south and the northwest of China, the precipitation tends to decrease, future monsoon change may result in arid and rainless in the above places, while the precipitation presents an increasing trend in the middle of the Indian Peninsula, the nouth of the Indo-China Peninsula, the region of resource of Three Rivers and the Huaihe River basin in China, where are apt to flood disaster. It is noteworthy that there is still uncertain in the above climate prediction, which is only for users to reference.