Applying Adaptive Neuro-Fuzzy Inference System to Improve Typhoon Intensity Forecast in the Northwest Pacific

Typhoon intensity forecast is an important issue. The objective of this study to construct a 5-day 12-hourly typhoon model based on the adaptive neuro-fuzzy inference systems (ANFIS) improve in Northwest Pacific. It analyzed improvement ANFIS by comparing it with MLR when only atmospheric factor or both and oceanic factors are considered. This collected SHIPS (Statistical Hurricane Intensity Prediction Scheme) developmental data typhoons Pacific before landing from 2000 2012. input were simplified stepwise regression procedure (SRP). Subtractive clustering (SC) was used determine number rules reduce complexity. Model Index (MI) taken as standard SC network architecture model. simulated results show that MI could effectively radius influence SC. significantly improved after environmental added. RMSE highest at 84 h; underestimated ratio primarily positive. Songda case shows maximum bias greatly improved, 60 h lead time, percentage (39%). Overall, forecast.