A Study On Station Forecast Of Surface Air Temperature And Precipitation In China Using Topographic Correction

This study is mainly based on the effect of terrain correction on the model station prediction of temperature and 24h cumulative precipitation,and the improvement of the model prediction effect by the multi-model ensemble scheme.The model data using Global Forecast System(GFS)1-5 day prediction of the surface 2m temperature,the US National Environmental Forecast Centers(NCEP),the Japanese Meteorological Office(JMA),the China Meteorological Bureau(T639)model 1-6 day of 24h cumulative precipitation.The surface air temperature and precipitation data of China Meteorological Observatory are taken as observational data.The 2m temperature model forecast is corrected in vertical direction,and then the linear regression equation is established,which is compared with the linear regression correction method.The prediction ability of the model itself will affect the affect two types of calibration schemes.If the model topographic altitude bias(TAB)is too large,it may severely affect the forecast skill,and lead to too large forecast error.As the forecast lead time becomes longer,the root-mean-square error(RMSE)of the 2m air temperature forecast increases slightly as well.By comparing the effect of the TAB and the increasing forecast lead time on the model forecast performance,the effect of the TAB on the model forecast skill is more significant.Both two types of calibration schemes,namely the linear regression without vertical correction and with vertical correction can reduce the model forecast errors,and the latter has better calibration performance.The linear regression correction and binary linear regression correction which is added to the terrain factor is added to the station 24h cumulative precipitation of model prediction.The two calibration schemes of linear regression and the correction scheme considering terrain can both improve the prediction effect of the model,and the correction scheme with topographic factors is the best to improve the model.Considering the terrain correction scheme,the effect of heavy rain is better in improving the model station forecast.The NCEP model has the best prediction performance after the terrain correction.In different lead time,different numerical forecasting models,the terrain correction scheme and linear regression correction have different influence on the model.The results of multimodel ensemble precipitation forecasts using three models show that multi-model ensemble forecast is superior to that of each single model.The improvement of the multi-model ensemble forecast over the single model forecast is better than the ensemble scheme to the single pattern after the terrain calibration.The effect of the running training period bias-removed ensemble mean is better than ensemble mean.