Real-time steering of mesoscale forecast models using objective techniques

Real-time steering of mesoscale forecast models using objective techniques allows data assimilation and computational resources to focus on Regions of Interest (ROI) where active weather will likely occur. In developments inspired by a presentation at the 2005 AMS Annual Meeting by Kelvin Droegemeier, mechanisms for using existing real-time data systems and analysis tools to steer a local forecast model to a region where "interesting" weather would occur during the forecast period have been employed which enable the model domain to evolve over successive forecast runs while providing research and education users with products and data based on the forecast domain.

The system implemented within a week following the San Diego AMS meeting uses operational forecast model fields and an objective weighting method to select the region of interest. GEMPAK's Gaussian Weighted Filter using normal distribution of weights is used to create a 24 hour predictive field where the model domain is centered over selected maximas. The method currently uses the 24 hour accumulated precipitation field produced from forecast hours 6 through 30 of NCEP's 12km ETA (aka NAM) model, which is distributed operationally via NOAAPORT, as the predictor for the ROI function. Once the model domain is determined, Workstation ETA and WRF models are run. As a result, the system sucessfully tracked the major ice storm that moved across the Southeast US and effectively shutdown Atlanta and other cities in the area. Other appications of the system will be discussed.