Online Program Home
My Program

Abstract Details

Activity Number: 362 - SPEED: Food, Environment, Biomedical Imaging and Physical System Visualization/Learning, Part 2
Type: Contributed
Date/Time: Tuesday, July 30, 2019 : 11:35 AM to 12:20 PM
Sponsor: International Chinese Statistical Association
Abstract #307778
Title: A Fully Bayesian Approach to Typhoon Precipitation Forecast
Author(s): Yu-Chun Huang* and Chuhsing Kate Hsiao
Companies: National Taiwan University and Institute of Epidemiology and Preventive Medicine, National Taiwan University, Taiwan
Keywords: Fully Bayesian approach; Typhoon precipitation; Ensemble forecast

Ensemble forecast has been widely used in meteorology to consider the uncertainty of atmosphere condition. Unfortunately, current ensemble forecast approach tends to be under-dispersed; namely, the predictive value may not cover extreme events. For typhoon precipitation, accurate prediction for extremely large rainfall values is crucial in precaution of disasters. The purpose of this research is to develop a predictive distribution which can be used for real-time forecast as well as providing a more precise probabilistic forecast for large rainfall events. We propose a fully Bayesian approach which combines raw ensemble forecasts with important information from historic record of typhoon precipitation. By using the added information to update raw ensemble forecasts, our predictive distribution not only can conduct a real-time forecast but also provide a more detailed guideline for probabilistic prediction. Leave-One-Out cross validation and real data application will be demonstrated and compared the predictive performance with other existing methods.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program