Statistical-Dynamical Seasonal Prediction of Tropical Cyclones Making Landfall Along the South China Coast
Samson K.S. Chiu, City University of Hong Kong, Hong Kong, Hong Kong; and J. C. L. Chan, W. L. Ginn, and S. M. Lee
In a previous study, a statistical-dynamical seasonal prediction scheme was developed for tropical cyclones (TCs) making landfall along the U.S. Atlantic coast and a 17-40% skill improvement over climatology in three groups of TCs is found. This study is an extension of the previous one to develop an operational seasonal prediction of TC landfall along the South China (SC) coast. To develop such a real-time operational prediction, a global circulation model (GCM) of long history with no modification in configuration and physics is needed. The Japan Meteorological Agency is therefore chosen. As in the previous study, the prediction equation is fitted using multivariate Poisson regression to correlate the empirical orthogonal function (EOF) coefficients of the model-predicted fields with the number of TCs making landfall along the SC coast. The equation-predicted number of landfalling TCs can be obtained. A categorical prediction (e.g. 0-1, 1-2 or 2-3) would then be issued as the predicted number of landfalls. During the 22-year (1984-2005) hindcast period, 16 years of the dependent samples are correctly “predicted” during the three-month period (ASO) using the model run at base date of July 10th. In cross-validation, 13 years are correctly predicted in ASO. The physics behind the predictors are found related to the location and strength of the subtropical high, the environmental steering flow and the El Nino/Southern Oscillation. The above results and the performance of the real-time prediction for the year 2009 will also be presented.
Poster Session 1, Posters: TCs and Climate, Monsoons, HFIP, TC Formation, Extratropical Transition, Industry Applications, TC Intensity, African Climate and Weather
Tuesday, 11 May 2010, 3:30 PM-5:15 PM, Arizona Ballroom 7
Previous paper Next paper
Browse or search entire meeting
AMS Home Page