Simulation And Projection Of Climate Change In Asian Large River Basin Based On CMIP5 Modes

Climate models are the one of the major tools to better cognize the phenomenon of the climate system and to project the future climate change, which has been widely used in present researches on hydrology, ecology, environment, agriculture, etc. However, due to the complexity of the climate system, the models’ability to simulate the climate change is far from accurate, and large uncertainties exist in the simulation results. With the development of climate models, the latest models which participated in the fifth phase of Coupled Model Inter-comparison Project (CMIP5) have just been released, and the simulation outputs from these models will be widely used in different study fields. Thus it is necessary to evaluate the simulation performance of these models comprehensively and systematically, as to provide reference and suggestions not only for model improvement, but also for the researchers from different fields who will use the results from climate models.The enhancement of model resolution can improve the ability of simulating the temperature change with the terrain, and improve the performance in precipitation level. Based on statistical downscaling BCSD CMIP5 models on a global scale applicability has been verified, but the application does not apply in the regional scale. This research used five statistical downscaling CMIP5 models and multi-model in simulating precipitation and temperature changes of Asian river basin, and on the basis to project the variety of precipitation and temperature in 2015-2050, conclude the following conclusions:(1) The precipitation is abundant in Asian large river basin from June to September during 1961-2005, and it is less in the rest of month. CMIP5 climate models and multi-model can well simulate the distribution characteristics of the precipitation change in monthly and yearly, and it can better simulate the maximum, minimum and average precipitation, but the simulation values were slightly lower than the observed values, and simulation ability of month with less precipitation is superior to the month with abundant precipitation, multi-model simulation ability is superior to the single mode.(2) In 1961-2005, the change of annual precipitation, summer and autumn precipitation are bigger than the precipitation in spring and winter in Asian large rivers; multi-model can better simulate the annual precipitation and seasonal precipitation distribution of maximum, minimum, quartile precipitation and within 15 mm deviation, and its ability on simulating the seasonal and annual precipitation distribution is optimal, the second is ccsm4 model, simulation ability of bnu-esm model is the worst.(3) CMIP5 models and multi-model can well simulate the space distribution of annual precipitation which precipitation in east was more than west in space during 1961-2005, and it can also simulate the seasonal variation and precipitation characteristics on spatial distribution, but there are still overvalued or undervalued areas and systematic errors. The model simulated precipitation graded distribution range is slightly different; the deviation between the CRU data and model outputs which simulated by multi-model is small, and its simulation ability is relatively better than the single mode.(4) In 1961-2005, Asian large river basins in high-temperature concentrated in May to August, the highest temperature in June, and temperature is relatively low in the rest of month. CMIP 5 models and multi-model can well simulate the months that the distribution of average temperature, and can better simulate the maximum, minimum and average temperature. There is widespread systemic error, but it does not affect the overall distribution characteristics, simulation results in high temperature month is more accurate than the month in low tempersture.(5) Spring and winter average temperature change is higher than the annual average temperature, summer and autumn average temperature change cover 1961-2005; multi-model can better simulate the distribution of annual and seasonal average temperature, and maximum, minimum, quartiles and median characteristic value; Each model and multi-model can simulate the rising trend of average annual temperature in Asian river basin cover 1961-2005, with correlation coefficient, standard deviation, and root mean square error comparison results show that the performance of multi-model is optimal and ccsm4 is the second.(6) The single model and multi-model can well simulate the spatial distribution characteristics of the annual average temperature of Asian large river basin. It can simulate the spatial distribution of seasonal average temperature difference, the model simulated precipitation graded distribution range is slightly different; the deviation between the CRU data and model outputs which simulated by multi-model is small, the simulation ability is superior to the single model.(7) The projection of future changes in annual average precipitation, summer and autumn precipitation is bigger than winter and spring; with the rise of emissions scenario, the annual precipitation and seasonal precipitation are slightly elevated. In RCPs scenarios, relative to baseline period, annual and seasonal precipitation distribution was obvious different; the increase in annual, summer and autumn precipitation was bigger than winter and spring.(8) The projection of future changes in mean temperature over Asian river basin will increase under three scenarios during 2015-2050, and average temperature under the RCP 8.5 will rise fastest; In all of three scenarios, annual average temperature is about 19 ℃, and the annual average temperature and seasonal average temperature changed little; Spatially, the temperature relative to the whole basin base period presents the consistent trend, and the larger rise in northern, southern rise less; And with the increase of concentration of emissions, the greater the amplitude of the average temperature rise.