Tuesday, 11 February 2003
A new statistical method for estimating tropical cyclone intensity from GOES-IR imagery
This paper introduces a new approach for estimating tropical cyclone (TC) intensity from IR imagery. The method builds on the empirical foundations of the Dvorak technique, but considers more formal statistical relationships between IR derived variables and TC intensity. A multivariate linear model is formed by regressing MSLP measured by aircraft reconnaissance (predictand) onto a set of IR parameters (predictors). The model is then tested and compared with the MSLP estimates from the advanced objective Dvorak technique (AODT).
Preliminary results suggest that the linear model is competitive with the existing AODT while streamlining the process considerably and reducing (or eliminating) the decisions (e.g., scene type identification) that are made within the framework of the AODT.