Advances and Promises in Tropical Cyclone Prediction (Invited Presentation)

In this overview talk, I will present some of our recent progresses in tropical cyclone prediction and predictability research that have the potential to improve future hurricane prediction. Highlights will be given to the direct assimilation of high-resolution Doppler radar data and all-sky satellite radiances into convection-allowing regional models hat can lead to better initialization of the detailed vortex structure, and subsequent improved intensity forecasts. For the track forecasting, we have demonstrated the benefit of using an observation-based ensemble setting technique (OBSET) for tropical cyclone track forecasting by selecting about 40% of the short-term best-performing members. The technique has now been successfully used in the operational tropical cyclone prediction of CMA. We have also explored the use of a Bayes Model Averaging (BMA) method that can better optimize the weights of individual deterministic models in the consensus operational track forecasts by giving higher weights to the better performing models in the most recent predictions. I will further speculate on the future promises and challenges in tropical cyclone prediction, and the needs/benefits for close collaborations between researchers and operational forecasters.