SURFACE LATENT HEAT FLUX AND INNER-CORE RAINFALL ASSOCIATED WITH RAPIDLY INTENSIFYING TROPICAL CYCLONES AND ITS APPLICATION IN TROPICAL CYCLONE INTENSITY PREDICTION OVER WESTERN NORTH PACIFIC

摘要

Regression models for 24-h typhoon intensity forecast based on the Statistical Typhoon Intensity Prediction System (STIPS) are developed by adding two new predictors: surface latent heat flux taken from the third release of the Objectively Analyzed Air-sea Fluxes and inner-core rain rate obtained from Tropical Rainfall Measuring Mission (TRMM) Multi-satellite Precipitation Analysis. The microwave optimally interpolated SST produced by Remote Sensing Systems is used to estimate maximum potential intensity. The models predictors include environmental information from National Center for Environmental Prediction Global Forecasting System final analysis and best track from Japan Meteorological Agency Regional Specialized Meteorological Center Tokyo. Results from a stepwise linear regression model show that the inclusion of these two satellite-based predictors provides 1percent approx 3percent reduction in overall intensity changes. To take into account the non-linearity of the forecast, a neural network (NN) model is developed. Results from the NN show a 5percent approx 11percent reduction of the mean absolute forecast error.