The present study revisits the statistical modeling of typhoon transition. The objective of the study is to provide insights on plausible statistical typhoon transition models based on extensive statistical analysis. First, the correlation structures of the typhoon transition are estimated in terms of autocorrelation function (ACF) and partial autocorrelation function (PACF). This facilitates to specify a set of plausible models for further investigation. Then, the corrected Akaike Information Criterion (cAIC) is applied to investigate the relative goodness of fit of these models. The spatial inhomogeneity and the seasonality are taken into account by developing the models for different spatial grids and seasons separately. An appropriate size of spatial grids is investigated. The statistical characteristics of the random residual terms in the models are also examined. Finally, Monte Carlo simulations are performed to investigate the overall performance of the proposed model.
Typhoon transition; Statistical analysis; Spatial inhomogeneity; Seasonality; Non-normality; Model selection
Main Research Area:
The Seventh International Colloquium on Bluff Body Aerodynamics and Applications, 2012