FORECAST METHOD OF TYPHOON EXTREME PRECIPITATION BASED ON SVM MODEL OPTIMIZATION

The frequent occurrence of extreme precipitation events has attracted widespread attention from all over the world, and it has a significant impact on the sustainable development of coastal cities. At present. the error in the prediction of heavy rainfall above the torrential rain level is relatively large. In this paper. taking the coastal areas of southeast China as an example. an optimized SVM model is used to quantitatively evaluate extreme precipitation under typhoon conditions. The research results show that the precipitation intensity in the study area is mainly distributed between 80 mm and 340 mm, which basically presents a normal distribution characteristic. The precipitation is scored using the ETS method proposed by the previous. The relationship between precipitation and score is a power exponential relationship. As the precipitation increases from 0 to 500 mm, the score decreases from 0.45 to about 0.05. This method has better forecasts of average precipitation in outer rain belts than in inner rain belts. For rain belts within the range of 0 similar to-80 km from the tropical cyclone center, the forecasted extreme precipitation threshold is 195 mm, which is very close to the actual extreme precipitation threshold of 205 mm. On the whole, this method has better forecasting ability of extreme precipitation in the inner rain belt of tropical cyclones than the outer rain belt. The actual measurement and prediction results of extreme precipitation based on the SVM model are very consistent, and the prediction results are slightly higher than the actual measurement results. This study shows that the extreme rainfall evaluation results based on the SVM model are reliable. This method can provide a reference for accurate prediction of extreme precipitation.