Improved Statistical Intensity Forecast Models: A Joint Hurricane Testbed Project Year 2 Progress Report
Mark DeMaria and John A. Knaff Mark.DeMaria@noaa.gov and John.Knaff@noaa.gov NOAA/NESDIS/ORA/RAMMB Fort Collins,CO
John Kaplan John.Kaplan@noaa.gov NOAA/OAR/AOML/HRD Miami, FL
Three improvements to existing operational statistical intensity forecast models were proposed in this project as follows. 1) The decay formulation for storm tracks over land in the Statistical Hurricane Intensity Prediction Scheme (SHIPS) has a low bias (too much decay) for storms moving over islands and narrow landmasses such as Florida. A new formulation of the basic decay equation was developed where the decay rate is proportional to the fraction of the storm circulation over land. The impact of the new decay formulation on the SHIPS forecasts will be evaluated. 2) The method for calculating the vertical shear in SHIPS averages the wind over a fairly large annulus (200 to 800 km from the storm center). This large area is needed to avoid the NCEP global model representation of the storm. A method was proposed to remove the storm circulation from the model forecast fields so smaller areas could be tested. 3) As a companion to the SHIPS forecast, a Rapid Intensity Index (RII) was developed to estimate the probability that a storm will rapidly intensify in the next 24 hours. The current RII is based on a simple scaling of the input variables, which have equal weight in the probability calculation. A more sophisticated discriminant analysis method is proposed, which weights the inputs to provide the best probability estimate.
The new decay model was tested in real time during 2005 and 2006, and provided considerable forecast improvement. A method was developed to remove the symmetric part of the tropical cyclone circulation from the GFS model so that smaller areas for calculating the shear could be tested. It was found that the time tendency of the forecasts vortex strength was also a good predictor of intensity. A parallel version of SHIPS with the smaller shear area (0 to 500 km radius) and GFS vortex predictor was run in real time during the second half of the 2006 season. The forecasts for the first half of the season were also re-run to provide a full season for evaluation. Results show that the version of SHIPS with the modified shear and GFS vortex predictor significantly improved the Atlantic SHIPS forecasts, and marginally improved those in the east Pacific. The discriminant analysis version of the RII was run in parallel in real time for the entire 2006 hurricane season. A comparison of the operational and experimental RII forecast verifications will be presented.
The views, opinions, and findings in this report are those of the authors and should not be construed as an official NOAA and or U.S. Government position, policy, or decision.