Study On Prediction Of Daily Precipitation Associated With Landfalling Tropical Cyclones Based On The DSAEF_LTP Model

Based on the DSAEF?LTP(Dynamical-Statistical-Analog Ensemble Forecast for Landfalling Tropical cyclones(LTC)Precipitation)model,a simulated prediction(i.e.,all of the so called "prediction" based on the DSAEF?LTP model in this paper refer to "simulated prediction for specific samples of LTC")for accumulated precipitation associated with 10 LTCs occuring over China in 2018 was conducted.Then the DSAEF?LTP model for Daily Precipitation Forecast(DSAEF?LTP?D model)was preliminarily developed,and a simulated prediction to predict daily precipitation(i.e.,the day making landfall and the day before it)associated with 5 LTCs landing East China in 2018 was carried out to evaluate performance of the DSAEF?LTP?D model.Finally,a test to compare the impact of introduction of three variables into the generalized initial value(GIV)of the DSAEF?LTP?D model,including TC speed,TC intensity and TC time(i.e.,the date when daily precipitation starts),was also conducted.The main conclusions are as follows:1)Simulated prediction of accumulated rainfall associated with the 10 LTCs:(a)The DSAEF?LTP model is comparable to three global dynamic models(i.e.,ECMWF,GFS,and GRAPES)in predicting LTC accumulated rainfall.For accumulated precipitation of ?250 mm,the TS(threat score)produced by the DSAEF?LTP model is 0.042,which is better than the highest score from the result of the three dynamic models(0.0375);for accumulated precipitation of ?100 mm,TS produced by the DSAEF?LTP model ranks second among all prediction models.(b)The DSAEF?LTP model can still provide valuable forecasts when the dynamic models do not have the ability to predict heavy rainfall accumulations of certain individual TCs;it can capture more localized distribution of accumulated rainfall of ?250 mm,or the ?100 mm rainfall in southern coast of China,but it tends to predict too wide range of the heavy rainfall region as compared to the observations.2)Technical exploration of the DSAEF?LTP?D model,and its simulated prediction of daily rainfall associated with the 5 LTCs:(a)the DSAEF?LTP?D model has been preliminarily developed,which includes two key technics of "construction of Similarity Region on daily scales" and "Identification of Nearest Point and Screening of Shortest Distance".(b)The DSAEF?LTP?D model is comparable to Shanghai's regional dynamic model and slightly superior to the three globle models in predicting 24-hours accumulated rainfall.Rather,the DSAEF?LTP?D model performs best for the rainfall forecast of the day before landing;as for prediction of precipitation on the day of landfall,the performance of the model is at the medium level among all of the forecast model;LTCs with poor performance of the model mostly appears in the northerly regions.(c)Compared with dynamic models,the DSAEF?LTP?D model can better capture the center of heavy rainfall,and the pattern of its predicted heavy precipitation is similar to that of observations,but there are also different degrees of false alarms and misses in the rainfall distributions predicted by the model.3)Comparision test of introduction of TC speed,TC intensity and TC time:(a)After introduction of the three variables,the DSAEF?LTP?D model has significantly improved the forecast performance of daily precipitation associated with the 5 sample LTCs,where its performance has been increased by 32.7% on the day before landfall,and by 13.3% on the day of landfall.(b)The introduction of TC speed and TC time can significantly improve the capability of the DSAEF?LTP model to predict LTC precipitation,nevertheless the incorporation of TC intensity has little effect on improvement for the model.