Comparative Analysis Of Two Reanalysis Datasets Of ERA-40 And ERA-interim In China

Reanalysis data is a source of global atmospheric data.It plays an important role in many research fields,such as climate detection and seasonal forecast,global and regional water cycle,and model assessment.Two reanalysis datasets of the 2 meter temperature(M2T),mean sea level pressure(MSLP)and 700 hpa specific humidity(SPH)was correlation analysised and root mean square error(RMSE)analysised on yearly scales from 1979 to 2001.And the correlation and trend analysis of the five regions of China were selected on daily,seasonal and annual scales.Standard deviation of each Reanalysis data's M2 T,MSLP and SPH was Calculated.The standard deviation's quotients(STD)were used for data discretization correction.The results are as follows:(1)From the spatial distribution characteristics,on yearly scales,the two reanalysis datasets of the M2 T and the MSLP had strong correlation in the plain and basin area and the correlation coefficients are between 0.9~1;But the correlation of the two variables are not good at a plateau and a mountain range above 3000 meters for the correlation coefficients being between-0.5~0.5.The SPH'S correlation coefficients are between 0.88~1 belongs to the strong correlation in some regions of the Tibetan Plateau.(2)In the Chinese five regions,the average of two reanalysis datasets' s corresponding M2 T,MSLP and SPH are almost the same.And the corresponding variables of the two reanalysis datasets have the same variation trend in the same area on daily,seasonal and annual scales.(3)In the China's five regions,on seasonal and yearly scales,the significance and Sen's slope of the two reanalysis datasets of Corresponding M2 T and SPH are almost the same.(4)The difference of mean values of the two reanalysis datasets' s corresponding M2 T,700hPa SPH and MSLP in the central region are lager.(5)On daily scales,compared with the other four areas,the standard deviation's quotients(STD)of M2 T and MSLP in the central regions have bigger fluctuations.However,in the central regions,fluctuations of SPH's theoretical error coefficients are smaller than the other four areas.The greater the fluctuation of the theoretical error coefficient,the greater the discrepancy between the two reanalysis datasets.