Simulation Evaluation And Error Correction Of CWRF Model In Summer Extreme Precipitation Over China

Daily rainfall data simulated by the Climate-Weather Research and Forecasting model and observed by China Meteorological Administration from June to August during 1980-2015 are used to evaluate the performance of temporal and spatial simulation ability of climate of various parameterization schemes of CWRF model in simulating summer extreme precipitation index by area average Taylor Diagram,Taylor score,regional average time series for M2 index and correlation coefficent,etc in China and three regions of North China,Central China and South China.the probability bias correction model XCDFt is proposed for extreme rainfall by introducing generalized Pareto distribution(GPD)into the correcting process and assessed in the applicability of extreme rainfall correcting.The distribution characteristics of the recurrence level of extreme precipitation before and after revision are discussed.Finally,the revised effects of different methods and parameterized schemes are compared.The main conclusions are as follows:(1)The performance of spatio-temporal simulation ability of each parameterized scheme is relatively consistent.Different parameterization schemes are consistent in the simulation of the whole country and Central China,south China and North China.Each parameterized scheme can simulate the extreme index better but there are big differences between parameterized schemes.The results show that the first four parameterized schemes are C13,C14,C12 and C1,respectively,the last four parameterized schemes with poor comprehensive simulation ability are C6,C4,C3 and C10.The simulation effect of 14 kinds of parametric sets is better than that of optimal selection and inferior selection,and better than that of control scheme.(2)XCDFt can improve the consistency of GPD scale parameters between model simulation and observation in North China,Central China and South China.After model simulation correcting by XCDFt,the 20-year return levels of spatial average,root mean square error and spatial correlation coefficient for extreme rainfall are closer to observation,which shows that the revised data are more reliable.(3)Different error correction methods have good correction effect on the mean of extreme precipitation,but he correction effects of each revision method on the mean value of extreme precipitation in different regions are different.The RQ-lin revision method is better than other revision methods,the error of model simulation extreme index is reduced effectively.Evaluation of the effects of different parameterized ensemble schemes on the correction of extreme precipitation found that the 14 sets of parameterized scheme members were revised and then set was the most close to the observed results.