A JHT-Funded Project: Transition of GFDL Hurricane Prediction System to HWRF
Morris A. Bender, Timothy P. Marchok Isaac Ginis, Biju Thomas ** Robert E. Tuleya * Geophysical Fluid Dynamics Laboratory, NOAA Princeton, New Jersey 08542 (Tel. 609-452-6559; FAX 609-987-5063, email: Morris.Bender@noaa.gov)
Graduate School of Oceanography ** University of Rhode Island EMC/SAIC *
Major physics upgrades to the GFDL Hurricane Prediction System that were developed and tested during the past 2 years through JHT funding were made operational in time for the 2006 hurricane season. These physics changes were made in close collaboration with scientists at NCEP (National Centers for Environmental Prediction) and URI (University of Rhode Island). During the past 6 months they have been fully transitioned into the Hurricane WRF (HWRF) model that is being developed to run in parallel with the GFDL model in 2007.
As was anticipated, the new physics upgrades (i.e., NCEP’s Ferrier microphysics and an improved momentum flux parameterization), significantly improved the GFDL model’s skill in intensity prediction both in the Atlantic and eastern Pacific. For the first time the GFDL model exhibited superior performance compared to the statistical models in most forecast time periods, beating even the official forecast in the Atlantic and in the longer forecast time periods in the eastern Pacific. These results will be summarized in this presentation, as well as the overall performance of the GFDL hurricane forecast system during the past 4 years of JHT funding.
Once the 2006 versions of the GFDL forecast system became fully operational in June 2006, the emphasis at GFDL shifted to assist in the transition of the GFDL physics to the new HWRF model under development at NCEP. The strategy adopted by NCEP is to implement HWRF with the same physics packages that are used in the 2006 version of the GFDL model. A set of storms from the 2004 (Frances, Ivan, Jeanne, Lisa), 2005 (Dennis, Emily, Katrina, Philippe, Rita, Wilma) and 2006 hurricane season (Ernesto and Helene) was selected to serve as a suite of test cases to evaluate the new model. Forecasts on these storms have been run with a preliminary version of HWRF with the results compared to forecasts made with the 2006 GFDL model. A preliminary evaluation of the track forecasts of these 2 models will be presented for a 220 case sample size as well as distributions of the individual storm errors at 2, 3 and 5 days.
Finally, NCEP is testing a major upgrade to its Global Forecasting System (GFS) with implementation scheduled for spring 2007. This upgrade will involve major changes to its global data assimilation as well as adoption of a hybrid pressure-sigma vertical coordinate system. To evaluate the effect of this major implementation on the performance of both the GFS and GFDL models, the major part of the 2005 and 2006 hurricane system is currently being rerun with this new hybrid GFS. The impact of the new global modeling system on the GFDL model will be rigorously evaluated on over 500 forecasts in the Atlantic and over 400 forecasts in the eastern Pacific. Comparison of the track forecasts of the current operational GFS and this new hybrid version will also be made and presented, along with the differences in the track and intensity forecasts with the GFDL model using both versions of the GFS. It should be pointed out that these results will serve as the benchmark for which the final version of the HWRF model will be evaluated, later this spring.