Inter-comparison Of Long-term Simulations Of Climate Over China By PRECIS And RegCM3

Climate warming and more and more frequent extreme climate events had severe impacts on living of people, ecological environment and economic development, so climate change became a significant research field worldwide. As an indispensable tool for climate change research, climate numerical model had been widely used in climate simulation and prediction. Typical of fine resolution and detail description of regional information, regional climate model (RCM) had become one of the most important methods for climate change research.PRECIS, developed by Hadley centre of climate prediction and research, and RegCM3, developed by the Abdus Salam International Centre for Theoretical Physics, were widely applied to climate simulation and prediction in China, while the inter-comparison analysis of long-term climate simulations over China by the two models had not been carried out. The long-term simulation capacity of temperature and precipitation over China by PRECIS and RegCM3 under the same configuration of boundary condition, domain and horizontal resolution was analyzed and assessed in this paper. Firstly, the two RCMs were driven by ERA40 for a 40-years (1961-2000) climate simulation over China. And then their performance in simulating climatology and inter-annual variability of temperature and precipitation over China and four typical areas of East China was comprehensively analyzed. Lastly, the simulated extreme climate indicators of temperature and precipitation were compared with the observed ones, so as to assess the simulation capacity of extreme climate events of the two models.In detail, the main conclusions are given as follows:(1) PRECIS and performed well in simulating temperature climatology over China, and the simulation results generally matched well with the observation. However, there were systematic warm bias and cold bias in the simulation by PRECIS and RegCM3, respectively. The two models could well reproduce the inter-annual variability of temperature, but there were still some differences about the location and intensity of centre in the simulated results of the two models. Seen from the analysis of time coefficients, both models could well simulate the mean inter-annual change in temperature averaged over four typical regions of China, and could reproduce some observed typical years; for the annual mean scale, RegCM3 performed better, while for the summer and winter, the two models both had advantages and disadvantages.(2) With regards to precipitation in China, the two models could basically reproduce the spatial distribution and inter-annual variability. Compared with RegCM3, PRECIS were generally much better at simulating the location and intensity of heavy precipitation centre and areas with greater variability. Among four areas of East China, both models performed best at simulating the inter-annual change of precipitation over the lower-middle reaches of Yangtze Rive, and South China, North China and Northeast China followed. The PRECIS simulation of inter-annual change range of precipitation was larger than RegCM3, but had a better correlation with observation than RegCM3.(3) Both models performed well at simulating the spatial distribution of selected extreme temperature indicators, and the simulation were highly positively related with observation. Both models could well reproduce the inter-annual change of extreme temperature indicators over China, especially the change under the background of globally accelerating warming since 1990s. PRECIS simulation of some extreme temperature indicators were larger than observation, but had a higher correlation with observation. RegCM3 simulation of some extreme temperature indicators were less than observation, and less areas passed the significance test in the correlation analysis.(4)In terms of extreme precipitation, both models had certain capacities for simulating the spatial distribution and inter-annual change of extreme precipitation indicators, and could basically reproduce the distribution of high and low values of extreme precipitation indicators. The simulated indicators were positively correlated with the observed ones over most part of China, of which RR1 and CDD were simulated best, whereas SDII and RX5day were poorly simulated. Seen from the comparison of coefficient between the two models, PRECIS simulation was better correlated with observation than RegCM3 as a whole. For the inter-annual change in extreme precipitation indicators averaged over China, both models had good (such as RR1) and poor (such as RX5day and CDD) performance.