Evaluation And Ensemble Forecast Of The Surface Air Temperature And Precipitation In China Based On CMIP5 Multi-models

Based on CMIP5 muti-models,the capabilities of climate models and ensemble forecast on simulating surface air temperature and precipitation in China have been evaluated.Ensemble forecast includes deterministic forecast and probabilistic forecast.In order to optimize the traditional methods,we presented the optimal weighted model(Op-SE)for deterministic ensemble forecast,then it was compared with equally-weighted ensemble(EE)and superensemble(SE),and the capabilities of probabilistic forecast have also been assessed.What's more,based on ensemble forecast,the temperature and precipitation in 21st century are analyzed so as to understand future climate change.The results are as follows:The climate models can simulate temperature well both in temporal variation and the spatial distribution,but the simulation of precipitation is not good enough in temporal variation,and the model-ensemble precipitation is overestimated in northern,northeast,northwest China,while underestimated in southern,eastern China.The standard deviation between models in temperature is highest in winter and lowest in summer,whereas consistency of precipitation among models is best in winter but much lower than temperature.In terms of correlation coefficient,the simulation of temperature perform best in summer,worst in winter,the correlation coefficient of precipitation is best in autumn but do not pass the significance test at 0.1 level in summer and winter.Besides,according to the relative root mean square error(RMSE')and anomaly correlation coefficient(ACC),the capabilities of models is inconsistent in different regions,but the multi-model ensemble mean is more consistent with observations than results of most individual models.In order to assess the capability of Op-SE,it was compared with equally-weighted ensemble(EE)and superensemble(SE),results show that F statistics improve obviously after optimizing.And ACC of temperature between Op-SE and observation is optimal in most of China,especially in eastern China,while Op-SE do not improve in northwest China and Tibet Plateau.However,ACC of precipitation is much worse than temperature,but Op-SE is also best in most regions,especially in midwest of Xinjiang,east of northwest China,north China and east China.As to RMSE,EE is relatively weak,And Op-SE is better than SE in most China,especially in eastern.Moreover,Op-SE also performs optimally on simulate extreme weather,EE is the worst either.The spatial distribution of probabilistic forecast skill and ACC is consistent.The forecast skill of temperature is relatively better in the African continent,the Pacific,west of the Atlantic,the northern hemisphere.And the skill of northern hemisphere is better than southern hemisphere,but the Eurasia do not performs well.For precipitation,the forecast skill of tropics is better than other regions obviously.Besides,the temperature and precipitation are both overforecasted in high probability,while underforecasted in low probability.Comparing three events,it shows that the capabilities of forecasting temperature in the below-normal category and the above-normal category are better than near-normal category,however the skill of forecasting precipitation in near-normal category is better than other categories.Moreover,the forecast probability is concentrated at 0.3-0.4,it means the consistency of models is not good enough.According to the projection of climate change in China based on the Op-SE,it show the projection are consistent between the different emission scenarios in the early stage of the 21st century(2016-2035),however the projection becomes more sensitive to the scenarios since mid of the 21st century(2046-2100),and the temperature and precipitation in northeast and northwest china are the first to increase,this change is more pronounced in the late 21st century.Besides,the spatial distribution of probabilistic forecast is consistent with deterministic forecast.As time increases,the probability in below-normal category(BN)and near-normal category(NN)is decreasing,but the probability in above-normal category(AN)is increasing,especially under RCP8.5 scenarios,the probability of temperature in AN close to 1,and precipitation is 0.45,which is much higher than other two categories.