Ms. Ko presented a seminar on “An Investigation of the Cyclogenesis Forecast Ability of the HWRF Model by Invoking the GFDL Vortex Tracker”.
The purpose of this project is to validate the capability of the HWRF basin-scale model to predict cyclogenesis events using the GFDL vortex tracker. To accomplish this, the data generated by the basin-scale HWRF for the 2015 hurricane season were used. Using the GFDL Vortex Tracker, the cyclogenesis forecast cases were detected and written in ATCF-format outputs. Then, another post-processing filter was applied on those forecast events. Through comparing the HWRF cyclogenesis forecast outputs to the NHC Best-Track data, the cyclogenesis location errors were computed. The location bias helps to group the genesis results into three categories – “Hit”, “Miss”, and “False Alarm”. “Hit” includes the cyclogenesis cases predicted correctly – within 300km location-bias tolerant. “Miss” indicates those cyclogenesis events that happened without being predicted. “False Alarm” describes predicted cases of genesis but without cyclogenesis occurring near the estimated location. The HWRF cyclogenesis forecasts showed a significant increase in accuracy between 4 and 3 days validation time. Moreover, the function of the post-processing filter is discussed. The advantage of the filter is to increase the cyclogenesis output quality, although some correct predictions were removed. Overall, these analyses may one day allow further improvement upon current hurricane genesis forecasting abilities.
A recording of the presentation is available on the anonymous ftp site: