Prediction of Consensus TC Track Forecast Error and Correctors to Improve Consensus TC Track Forecasts
James S. Goerss (firstname.lastname@example.org)
Marine Meteorology Division, Naval Research Laboratory, Monterey, CA
Funded by a previous JHT project, a graphical predicted consensus error product (GPCE) was developed and installed on the ATCF at both NHC and JTWC in 2004. Using GPCE’s pool of predictors from the 2001-2005 seasons, revised regression models to be used for the 2006 season were derived and installed on the ATCF at both centers. These regression models are used to determine the radii of circular areas drawn around the consensus model forecast positions within which the verifying TC position is expected to be contained approximately 75% of the time. These circular areas are then graphically displayed on the ATCF for use by the forecasters at NHC and JTWC. For the 2006 Atlantic season, the circular areas displayed by GPCE drawn around the CONU forecast positions contained the verifying TC position 81%, 79%, 74%, 79%, and 77% of the time at 24h, 48h, 72h, 96h, and 120h, respectively. For the 2006 eastern North Pacific season, the GPCE circular areas contained the verifying TC position 61%, 67%, 63%, 68%, and 67% of the time at 24h, 48h, 72h, 96h, and 120h, respectively. For the Atlantic, the performance of GPCE was close to what was expected. For the eastern North Pacific, the circular areas displayed by GPCE contained the verifying TC position less often than expected. Because the GPCE areas contained the verifying TC position much more often than expected for the eastern North Pacific in 2005, an adjustment was made to the radius calculations for 2006 decreasing their size. If that adjustment had not been made, the GPCE circular areas would have contained the verifying TC position 65%, 72%, 70%, 72%, and 71% of the time at 24h, 48h, 72h, 96h, and 120h, respectively, much closer to what was expected.
The techniques used to predict consensus error were then applied to predict the east-west and north- south forecast error of the consensus models. Regression models to predict CONU and GUNA east-west and north-south forecast error for all forecast lengths in the Atlantic were derived using the pool of predictors for the 2001-2005 seasons. These predicted errors were used as correctors to be applied to the consensus models for the 2006 season. The means of the CONU and GUNA east-west and north-south forecast errors for all forecast lengths in the Atlantic were also found for the 2001-2005 seasons to be used as correctors for the consensus models for the 2006 season. From previous work it was found that for both CONU and GUNA the bias correctors were more effective than the statistical correctors derived using the regression models for forecast lengths less than or equal to 72h. For CONU, the application of only the statistical corrector for the north-south error was most effective at 96h and 120h. For GUNA, the application of the statistical corrector for the north-south error and the bias corrector for the east-west error was most effective at 96h and 120h. Using these strategies, corrected consensus forecasts (CCON and CGUN) were produced for the 2006 season and CCON and CGUN were installed on the ATCF as experimental guidance. For the 2006 season, the CCON errors were 50 nm, 93 nm, 142 nm, 184 nm, and 240 nm at 24h, 48h, 72h, 96h, and 120h, respectively. The respective errors for CONU were 50 nm, 96 nm, 144 nm, 187 nm, and 242 nm. The CCON improvements were small with only the 48h improvement significant at the 88% level. For the 2006 season, the CGUN errors were 46 nm, 84 nm, 136 nm, 162 nm, and 227 nm at 24h, 48h, 72h, 96h, and 120h, respectively. The respective errors for GUNA were 48 nm, 86 nm, 139 nm, 168 nm, and 216 nm. The CGUN improvements were significant at the 80% level at 24h, 48h, and 72h.