Projection And Evaluation Of The Temperature Extremes Indices Over China By CMIP5Models

Based on the same trem observation of daily maximum temperature and daily minimum temperature data during1961-2005, the ability of30models provided by Coupled Model Intercomparison Project (CMIP5) have been evaluated in both spatial distribution and time characteristic with5extreme temperature indices including mean maximum temperature(TXAV), mean minimum temperature(TNAV), heat wave duration index (HWDI), frost days (FD) and warm nights (TNF90).Regional relative error, Taylor Figure, space techniques rating (SS) and variability statistic (M2) are used to assess the space-time-comprehensive simulation ability of the30CMIP5models. According to the simulation ability of models, Top5models ensemble projection and probabilistic projection are used to project the change of extreme temperature in21st century. And the main results are as follows:(1) Assessment of the space-time-comprehensive simulation ability of the30CMIP5models shows that CMIP5models can reflect the spatial distrobution of most extreme temperature index, but can’t refected the trend well. Best space-simulation capability index is TNAV and FD, which model average correlation coefficient reached0.95and0.94respectively. Best time-simulation capability index is HWDI, the second is the average maximum temperature,and the worst time-simulation capability index is FD. The correlation coefficient between the spatial and temporal patterns of simulation capability ranked reachs0.496, which means the ability of simulating the spatial and temporal patterns is of high consistency. Based on the assessment results, the space-time simulation ability rank is got and the top5models:MPI-ESM-P, CMCC-CMS, CMCC-CM, EC-EARTH, CESM1-CAM5.(2)By contrast, the ability to simulate extreme temperature of CMIP5model than CMIP3model, especially for HWDI which average error is reduced by68.31%. Besides, Error dispersion among a new generation of global climate models is also reduced, model simulations are more centralized. However, some problems still exist in CMIP5models, such as the ability for continuous extreme temperature simulation is still weak(3)Comparison between the bias of Top5models ensemble to that of multi model ensemble, optimization model can reduce bot the error area and the deviation value. But both Top5models ensemble and multi model ensemble simulation capability in West China are relatively weak.(4)In RCP45scenario, according to the simulation ability of models,projection shows that both in the mid-and late21st century, extreme cold events will be reduced, and extreme heat events will increase in China.In the mid-21st century,there are significant increases in extreme temperature. Both TXAV and TNAV shows an increase of1.5℃with the probability of more than0.9. HWDI and TNF90will increase slightly, while the probability of reducing15days of FD is more than0.7. The late21st century (2081-2100), extreme temperature gets warming by more than preliminary, in China. Both TXAV and TNAV shows an increase of5℃with the probability of more than0.8, which amplitude is greater than mid-21st, The number of heat wave days in China will be a general increase of2days with the probability of0.7in the late21st century. FD will reduce by30days Tibet Plateau and China central area,with the probability of more than0.5. TNF90will also increase, and increase the probability is greater than the North than in South.