Publication Date : 2005-01-10
Author : Chan, J. C. L.Li, K. K.
Disaster Management Theme :
Disaster Type : Tropical Cyclone
Document Type : Research Paper
Languange : en
Link : http://www.springerlink.com/index/QUTY2Q6G7VERAAKV.pdf
This is the third in a series of papers to investigate the applicability of the ensemble forecasting (EF) technique in the prediction of tropical cyclone (TC) motion. In the previous two papers, either the environment or the vortex was perturbed and the other unperturbed component was then merged onto the perturbed component at the initial time. In the present study, the separately-perturbed environment and vortex fields are combined at this time. The objective is to determine the extent to, and the synoptic pattern under which, such a combination can improve the TC motion forecast compared with perturbing only one component.
The study makes use of the same barotropic model as the previous studies and the same dataset – 66 cases from the Tropical Cyclone Motion Experiment TCM90. Perturbations of the environment and those of the vortex are first generated separately using the breeding of growing modes (BGM) method, and then combined at the initial forecast time. The performance of this combined scheme, labeled as BGMC, is then compared with that of the scheme with only the environment or the vortex perturbations (termed BGME and BGMV, respectively).
The BGMC distribution of ensemble forecast tracks are found to be basically similar to those in BGME but the spread is reduced. Some poor forecast members in BGME also become close to the best track in BGMC. The relative skill scores of the BGMC forecasts relative to the best track are almost all positive but those under the perfect model assumption are negative because the control forecast is better. While both BGMC and BGME schemes can improve TC forecast track under transition synoptic conditions, BGMC also achieve a higher success rate under complicated vortex and environment interactions. In general, the BGMC scheme is superior to the BGMV scheme.