A test assessment of numerical forecast products for precipitation in the southwestern Tibetan plateau
Based on the precipitation data of seven conventional meteorological stations in the southwestern part of the Tibetan Plateau from 8 to 8 o'clock and 20 to 20 o'clock from January 1,2019 to February 20,2020,as well as the 24-hour cumulative precipitation reported from 08 and 20 o'clock in the ECMWF_HR,GRAPES_GFS and GRAPES_MESO modes aduring the period,the precipitation forcasting ability of the above three modes for the southwestern Tibet Plateau was tested and evaluated by the traditional dichotomous test method.The results show that there are significant seasonal differences in the forecasting ability of the three numerical models for general pre-cipitation in the southwestern part of the Tibetan Plateau that the TS score of general precipitation in summer is sig-nificantly higher than that in other seasons,while there is no obvious predictability in fall and winter;there are some regional differences in the precipitation forecast in the southern part of the Tibetan Plateau;the prediction ability of each numerical model for small precipitation(<5 mm)is significantly better than that of other magnitudes,and the TS score of the ECMWF model is greater that that of other numerical models in most cases in the prediction ability of precipitation greater than 10mm,especially in the 50mm precipitation range,which is much higher than that of other numerical models;therefore,the ECMWF model has a high reference basis for the prediction of processes with high intensity and long duration.The ability of the three numerical prediction models to predict general precipitation in summer is significantly better than that of other seasons,which is mainly reflected in the low airborne rate and the underreporting rate.The three numerical prediction models also have some regional differences in the prediction of winter precipitation in the southern part of the Tibetan Plateau,and the ECMWF and GRAPES_MESO models are stable and have certain reference value.
NETL
NSTL
万方数据
