878 A Probabilistic Forecast for Rainfall based on Bayesian estimation methond and Its Preliminary Experiments over china

Thursday, 10 January 2013
Exhibit Hall 3 (Austin Convention Center)
Jing Chen, Chinese Meteorological Administration, Beijing, China; and Y. han and M. Jiao

Abstract: The paper applies BPO(Bayesian Processor of output) method based on Bayesian theory to the probabilistic method of rainfall ensemble product. Using ensemble prediction data and historic observational data, we develop a rainfall probability forecast model, and then revise a set of precipitation predicted value into a set of Bayesian precipitation probability forecast in the form of continuous probability distribution or continuous probability density. Besides, we obtain a group value of IS (Informativeness Score), which can express the prediction ability of each ensemble member. Furthermore, we fuse the probability forecast results of each member into an integration Bayesian precipitation probability forecast on the basis of IS and test the results with CRPS. Experiments results show that the reliability of integration Bayesian precipitation probability forecast is higher than ensemble direct probability forecast. Key words: Bayesian estimation method; precipitation ensemble forecast; probability forecast experiments
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner