The Main Characteristics Of GRAPES?GFS Global Precipitation Forecast Deviation

The performances of day1(D1)to day5(D5)precipitation forecasts from the operational global model system(GRAPES?GFS)are evaluated in terms of amount,frequency,and diurnal cycle etc,against observation products of the Global Precipitation Measurement(GPM)data of 4 months(Jan,April,Jul and Oct)of 2017,in which special focus is put on the Warm Pool(WP)in the Western Pacific and the Storm Track in the North Hemisphere(STNH).Results show that:(1)The D1-D5 forecasts reasonably capture the global distributions of monthly mean precipitation amount and precipitation frequency,especially reproduce the typical observed features of global zonal mean precipitation(amount and frequency)maximums between 20oS?20oN and in latitudes of 40o?50o with the correlation coefficients above 0.9,and also resemble well the daily variation and diurnal cycle of precipitation over the WP and the STNH.(2)The characteristics of model precipitation forecast deviation in low-latitudes and the mid-latitudes are different.In low-latitudes,the “double peaks” of positive forecast biases of the daily precipitation amount and of the heavy rainfall day((29)25mm/d)frequency occur in the same locations of observed precipitation maximums,while the forecast deviation of wet-day(?0.1mm/d)frequency is quite small;whereas in mid-latitudes,forecast biases in the daily precipitation amount are small,but positive deviation of wet-day frequency(20%-40%)and negative bias of heavy rainfall day frequency are obvious in latitudes of 40?-60°.There is nearly no change in the bias distribution pattern when deviation values vary a little in different season and increase from D1 to D5,also RMSE is several times of AME,indicating the natures of model with systematic precipitation errors and variable performance in daily forecast.(3)In terms of diurnal cycle,the positive bias of prediction amount in WP is due to overprediction in precipitation intensity,whereas the slight negative deviation of precipitation frequency in WP is because of under-prediction in the rainfall coverage;however,the positive bias of precipitation frequency in STNH results from both factors,the over-prediction of rainfall coverage and the failed forecast of weak precipitation events.(4)The obvious differences on the precipitation(amount and frequency)deviation in low-and mid-latitudes,are related to the incongruity about the proportion of the grid and sub-grid scale precipitation in the model.The clues of model improvement point to the trigger function and process of deep convection precipitation in model cumulus parameterization scheme,and the coordination between the cumulus parameterization scheme and the cloud microphysics scheme.