Li, Chih-Hsin and Jing-Shan Hong, Central Weather
Bureau, Taiwan
This research evaluates
the performance and the uncertainty of typhoon track forecast in a mesoscale
ensemble prediction system based on the WRF model. This system was designed to
generate the model uncertainty, including the initial perturbation from random
error seeding in the WRF VAR background error covariance and the model physics
perturbation. MORAKOT and FANAPI typhoon forecast experiments were conducted
for all the members run and applied several verification schemes to evaluate
the spread and accuracy of the track forecast.
The evaluation results
show that the ensemble typhoon track forecast of MORAKOT case performs very
well in terms of the ensemble mean and the spread, but the track of typhoon
FANAPI case doesnŐt. The forecasting track of FANAPI case has southern bias,
and it makes the bad spread.
The bogus scheme might
bring about the same initial typhoon structure, and that would cause the bias
of the forecasting tack of FANAPI case. To improve the track forecast, the
typhoon initial vortex perturbation in WRF 3DVAR was applied. Using different
setting of variance and length scale in WRF 3DVAR to conduct new forecast
experiment. The track forecast of new experiment performs well in terms of the
ensemble mean and the spread.