Climate Extremes Simulated And Projected By Multi-Model Ensemble

As one of the main the tools in studying anthropogenic impact on climate change, the global climate model has been widely used. Outputs from the latest generation of the models that participate in the fifth phase of the Coupled Model Intercomparison Project (CMIP5) has just been released. These models, including more physical processes and at higher resolution than before, have seldom been studied on the issue of climate extremes. In this thesis, extreme temperatures and precipitations in China simulated by CMIP5multi-model ensemble (MME) are evaluated, and the simulation of global monthly temperature extremes by CMIP3and CMIP5models are compared. In addition, the probabilistic projection of future change in temperature extremes is carried on. The main conclusions are as following:The CMIP5MME has the capability in simulating the extreme temperature and precipitation in China. The linear trend and interanuual variability of the spatially averaged extreme temperature indices in China is well represented in the MME. Both temperature and precipitation extremes spatially correlate well with observations, though the latter is systematically overestimated by the model. Individual models show limited ability than the MME. As projected by the MME under the RCP4.5scenario, extreme cold events will decrease, and warm events and extreme precipitations will increase in China. The change is sharp in the early21st century, while it slows down in the late21st century, which is consistent with the projected change in the mean temperature of China. The20-yr return values of annual maximum and minimum temperatures may increase significantly under RCP4.5, and in parts of China it will increase4degrees Celsius.Compared with observations and reanalyses, the monthly mean temperature extremes simulated by the CMIP3and CMIP5MMEs are evaluated and compared. In general, both CMIP3and CMIP5MMEs can well simulate the cold and warm extremes. Globally and zonally averaged, the temperature extreme is overestimated by the models. As first-order approximation, the bias mainly comes from the overestimation in seasonal cycle. On grid-box scale, CMIP5models show better simulation than CMIP3.The CMIP3and CMIP5MMEs project a similar change pattern in the early21st century for extreme monthly mean temperatures. By the late21st century, the changes in monthly temperature extremes projected under the three CMIP3(B1, A1B, and A2) and two CMIP5(RCP4.5and RCP8.5) scenarios generally follow the changes in climatological annual cycles, which is consistent with previous studies on daily extremes. Enhanced changes in extreme temperatures that exceed the global mean warming are found in regions where the retreat of snow or the soil moisture feedback effect plays an important role. Compared with the CMIP3ensemble, the CMIP5ensemble shows a larger intermodel uncertainty with regard to the change in cold extremes in snow-covered regions.In order to obtain the probabilistic projection of temperature extremes, the Bayesian statistical model is constructed and multi-model outputs and observations are incorporated. Since there is no observation for future, it is hard to distinguish the climate change from the bias change. Sensitivity analysis also confirms the influence of the priors for the bias change.