Simulation And Projection And Uncertainty Of Climate Change In South Asian Large River Basin Based On Different Models

As the main tool for climate projection and analyzing, climate models has been widely used in various fields of ecological research, hydrology, environment, atmosphere and so on. Based onthecollection of dailytemperature and precipitation data of 21 global climate models of CMIP5 ensenmbled data, dynamic downscling regional climate model COSMO-CLM, statistical downscaling global models GFDL-ESM2M, assessing the simulationcapabilityof the spatial and temporal characteristic changes of temperature and precipitation in South Asian River basin. On the basis doing the projection to temperature and precipitation during the 21st early-term(2016-2035),mid-term(2046-2065) and long-term(2081-2100), estimating and discussingthe uncertaintybetween model projections in the result, conclusions are as follows:(1)According to CRU observational data sets from 1961-2005in South Asian River basin, high-temperatures concentrated in May to August, temperature of June is the highest, the rests wererelatively low, CMIP5 multi-model, CCLM and GDFL-ESM2M data collection could well simulate the maximum, minimum and average monthly temperature and other indicators. They alsohad a good simulation to the temperature distribution of the monthly temperature and the correlation coefficient all through the 95% significance test, the systematic bias were small, simulation results of CMIP5 multi-model data collection in the slightly higher-temperaturemonthwere better than the others.(2)CMIP5 multi-model ensemble, CCLM and GFDL-ESM2M have a good simulation effect onaverage annual temperatures in South Asian river basinfrom 1961-2005,and the characterstic thattemperature in southern basinis higher thanthe North.They can also well simulate the spatial distribution differencesamong each seasonal mean temperature. The deviation between the simulation of annual average andseason temperature of Multi-model ensemble simulations and CRU data was the smallest, and the capabilities were superior to the single model.(3)According to the comparison between observational data sets and APHRODITE indicators, precipitation was concentrated in 6-9 month from 1961-2005 in South Asian River basin, precipitation of the rest months was less.CMIP5 multi-model, CCLM and GDFL-ESM2M data collection could well simulate the maximum, minimum and average monthly precipitation and other indicators. They alsohad a good simulation to the spatial distribution of the monthly precipitation and had ahighcorrelation coefficient with observed data. Meanwhile each model had a good simulation of the maximum minimum and average monthly precipitation, simulation values are slightly higher than the actual observations, and modeling capabilityof lower precipitation months is superior to abundant months, multi-model ensemble average analog value is more similar to the distribution of the observed data.Numerical simulation of multi-model ensemble is more similar to the distribution of the observed data.(4)CMIP5 multi-model ensemble, CCLM and GFDL-ESM2M,each of them can well simulate the spatial distribution pattern ofannual precipitationin South Asian River basin from 1961-2005:west less than the east, north less than the south.It can also simulate the seasonal precipitation differences and spatial distribution. Overvalued and undervalued area exists on the pattern space, each model simulations of precipitation slightly higher than the observation quantitatively; deviation between simulations of multi-model ensemble and data APHRODITE is the smallest, its effect is relatively better than a single analog model.(5)Based onthe projectionto the annual mean temperatureby three climate models, under RCP2.6,4.5,8.5scenario, during the 21st century early-term (2016-2035), the mid-term (2046-2065) and long-term(2081-2100),annual average temperature showing a rising trend, the trendgrowfastest under RCP8.5 scenario, the annual average temperature changed little under RCP2.6 scenario. GFDL-ESM2M estimates the maximum result, multi-model data collection do the lowest. From the point of view of space, thewhole basin shows aconsistent upward trend related to the baseline(1986-2005), and the northern part has a larger rise than southern; average temperature increases with the RCP scenarios the greater the increase.(6) Based onthe projectionto the annual mean temperature by three climate models, under RCP2.6,4.5,8.5scenario, during the 21st century early-term (2016-2035), the mid-term (2046-2065) and long-term(2081-2100), annual mean precipitation mainly between 1110-1200mm,projection of CMIP5 multi-model ensemble is much higher than the other two models;the spatialdistribution of each model’s projectionvaries under RCP2.6,4.5 and 8.5 scenarios relative to the baseline(1986-2005). CMIP5 multi-model ensemble projection shows a rising trend, precipitation reduce regional distribution in Indus River and southern North Mekong River.However, projection of single model showsthat itwill decrease overall the basin in the future.(7) According to the projection of three models in annual average temperature, analysis found that:at the early-term of the 21st century, under the RCP2.6,4.5, and 8.5 scenario, the uncertainty of projection decreases with emissions scenario increasing.In the mid-term and long-term of 21st century, simulation uncertainty is on the decline under the RCP2.6scenario, and increasing over time except the range under the RCP8.5 scenari.;the uncertainty of precipitation projectionpresents a trend of "increase-increase-decrease"under the RCP2.6 and 4.5 scenario during the 21st centuryand increase under the RCP8.5 scenarioover time.