Abstract: Using national observation data and automatic station precipitation data,based on objective statistical methods,the prediction ability of the precipitation(12 h,24h cumulative rainfall),the surface elements(2m temperature,10m wind) and the height element(such as wind,temperature,height) in southwest China in 2015 of SWCWARMS model and GRAPES model were evaluated and compared,the main conclusions are as follows:(1) the ETS score of precipitation from SWCWARMS model higher than GRAPES model,meantime the BS score were higher than GRAPES model expert 24 hour light rain,both the two model have empty forecast in moderate rain,heavy rain,rainstorm grade;(2) in the score of 12 hour precipitation,TS score of SWCWARMS model were higher than in GRAPES,but SWCWARMS model forecast of precipitation range is too big,empty forecast more than GRAPES mode with forecast growth of aging; SWCWARMS model moderate rain and heavy rain empty forecast larger than other scale precipitation,GRAPES model failure-to-predict torrential rain and other scale precipitation is empty forecast;(3) the prediction of height field and temperature field of two models is better than that of wind field,forecast in the middle troposphere is better than that of the lower,the prediction of height field and temperature field of SWCWARMS model is superior to the GRAPES mode,RMSE of SWCWARMS model is less than GRAPES mode in the summer half year;(4) the two models have shown the RMSE of 2m temperature increased in autumn and reduced in spring,the RMSE of near ground elements from SWCWARMS model were less than GRAPES.