8 Bias-Correction of the 14-Day NCEP GEFS Tropical Cyclone Forecasts in the Western North Pacific by Using the Reforecast Dataset

Tuesday, 17 April 2018
Champions DEFGH (Sawgrass Marriott)
Hsiao-Chung Tsai, Tamkang Univ., New Taipei City, Taiwan; and T. T. Lo, M. S. Chen, and R. L. Elsberry

The objective of this study is to improve the real-time forecast skill of the 14-day tropical cyclone (TC) activity and track forecasts from the 21-member NCEP Global Ensemble Forecast System (GEFS). The reforecast model and current real-time forecast model are identical, except the ensemble size is reduced to 5 members in the reforecasts. An objective TC tracking method developed by Tsai et al. (2011) is used to obtain the western North Pacific TC tracks in the 20-year GEFSv11 reforecasts and also in the real-time GEFS forecasts during the 2017 season. The probabilistic forecast verifications show that the Area Under Curve (AUC) of the Receiver Operating Characteristic (ROC) curve are 0.76 and 0.68 in Week-1 and Week-2 forecasts, respectively. Previous studies have shown that false alarms (predicted TC-like circulations that do not verify) are an important issue in the GEFS and other ensemble systems. Preliminary verifications are also indicating that the false alarms may be detectable in the GEFS reforecasts, especially when the forecast probabilities are higher. The relationships between the large-scale environmental factors (e.g., ENSO, MJO, etc.) and the reforecast skill are also being investigated. Better skill is found for these week-1 and week-2 forecasts in the western North Pacific that are initialized during MJO Phases 5-7. The predictabilities before and after TC formations will also be presented. Finally, a statistical bias correction scheme obtained from the reforecasts will be applied to the real-time forecasts during the 2017 season to investigate whether the accuracy of the raw forecasts can be improved if the model biases are considered.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner