| Landfalling TCs are associated with heavy rainfall,strong winds and damaging storm surges,of which strong winds is a non negligible disaster causing factor that triggers storm surges and further aggravates heavy rain disasters.Therefore,accurate and timely gale forecasting of TC is critical.As one of the forecast indicators of TC gales,the TC induced potential maximum gales are able to estimate the intensity and affected area of TC related disasters about strong winds.Chen(2021)developed the Dynamical Statistical Analog Ensemble Forecast model for Landfalling Typhoon Gales(i.e.,TC induced potential maximum gale)(hereafter abbreviated to the DSAEF＿LTG model),based on the DSAEF theory proposed by Ren et al.(2020).At present,the DSAEF＿LTG model is only preliminarily formed and its previous studies were concentrated on the simulation experiments for individual cases or small size of samples.Hence,we are motived to carry out forecasting experiments and improvement research,in order to explore its prediction ability and establish a more robust model,so as to provide an effective tool for TC gale forecasting in China.The research is carried out from following three aspects: at first,carry out the initial forecast application of the DSAEF＿LTG model over South China.Then,three comparative forecasting experiments were conducted by introducing TC translation speed similarity and ensemble scheme improvement to establish a more robust model.Finally,based on the improved model,a national large sample forecast experiment was carried out to explore the applicability of the DSAEF＿LTG model over China.The main conclusions can be summarized as follows:(1)The initial forecast experiment of the DSAEF＿LTG model was conducted in South China.The sum of average TS of the DSAEF＿LTG model at the two key thresholds of Beaufort Scale 7 and 10 ranks first with 0.4273,indicating that the overall forecast performance of the model was better than that of dynamic model(CMA,ECMWF,JMA,NCEP).In addition,the dynamic models were prone to missing alarms,while the DSAEF＿LTG model was prone to false alarms(especially for strong wind above Beaufort Scale 10).From the forecast results of individual TCs,the forecast performances of the DSAEF＿LTG model were more stable than dynamic models.The stronger the maximum gale(the largest value selected from the potential maximum gales of all stations for each TC),the better the forecasting performance of TS7 for each model,but this feature was not obvious in TS10.The analysis of two typical cases further showed that target TC with typical track and widesp read gale(e.g.,Hato)makes it easier for the DSAEF＿LTG model to make accurate prediction both in the wind field pattern and magnitude of central wind speeds.However,for sideswiping TC with small scale gales(e.g.,Doksuri),the DSAEF＿LTG model tends to over predict and fails to achieve satisfactory forecasting results.(2)The improvement experiments of the DSAEF＿LTG model were also aimed at South China.Both the introduction of TC translation speed similarity alone or the improvement of ensemble scheme can improve the overall TS7 and TS10 of the DSAEF＿LTG model,reduce the false alarms ratio and missing alarms ratio,while the latter had a more significant improvement(TS7 and 10 increased by 0.0813 and 0.14respectively;FAR7 and 10 decreased by 0.1214 and 0.2611,respectively).The forecasting skills of single TC further showed that the improvement of TC translation speed similarity was not as stable as that of ensemble scheme.The case analysis of Haima showed that the new ensemble scheme(probability matching mean)can adjust the ensemble wind fields of similar TCs,which contributed to more reasonable and accurate forecasting results.(3)Combining the results of the correlation analysis of 353 TCs in China and the results of the improved experiment in South China,as well as accelerating the calcula tion speed and compressing the calculation memory,the model with improved ensemb le schemes solely was used to carry out and compare the national basic forecast experi ment(DSAEF＿LTG＿B)and regional re-ensemble experiment(DSAEF＿LTG＿R).The ir best forecast schemes made difference in the parameters of seasonal similarity,the n umber of best similar TC and similar region.In general,the average TS of DSAEF＿L TG＿R was 0.6938,which was higher than that of DSAEF＿LTG＿B(0.6187).In additi on,the false alarm ratio and average absolute error of DSAEF＿LTG＿R were lower tha n those of DSAEF＿LTG＿B,but the miss alarm ratio was higher.From the forecast res ults of single cases,both the DSAEF＿LTG＿B and DSAEF＿LTG＿R can provide reliabl e forecast results for TC affecting multiple regions,the former did well in Beaufort Sc ale 6,while the latter had a higher hit ratio of Beaufort Scale 8.For the TC landing alo ng the Pearl River Estuary,the model(especially the location of the extreme center)w as sensitive to the predicted track of target TC.What’s more,moderate number of simi lar TC,more strict TC intensity screening and appropriate collection scheme are benef icial to improve the accuracy of sideswiping TC.In the future,with the continuous improvement and more perfect experiments,th e DSAEF＿LTG model will continue to develop and eventually become robust.Further more,it’s expected to provide useful references for the real-time forecast of TC-induc ed potential maximum gale. |
