The Influence of a Digit Filer Initialization on HWRF Hurricane Forecasting

The digital filter initialization (DFI) is a dynamic initialization scheme that can filter out high-frequency noise in the analysis using a discrete Fourier Transform.  In this study, a DFI scheme is applied for the NCEP Hurricane Weather Research and Forecasting (HWRF) system.  Specifically, the DFI is used in each HWRF analysis-forecast cycle (6 hourly) after the vortex initialization and data assimilation.  A series of numerical experiments have been conducted for Hurricanes Earl (2010) and Gonzalo (2014). Results show that the DFI can significantly mitigate the initial vortex spin-down or spin-up problems. Better track and intensity forecasts can be made within a 1.5-3h cut-off window and beyond. Applying the DFI on both the dynamical (i.e. horizontal wind, surface pressure, and geopotential height) and the thermodynamic fields (i.e. temperature, specific humidity) lead to better intensity forecast, compared with the results when the DFI was only applied on the thermodynamic fields.  

        Additional experiments are being performed using an alternative DFI method based on the Empirical Orthogonal Decomposition (EOF) theory. The advantages and disadvantages of these two DFI methods and their impacts on the HWRF hurricane initialization and forecast are investigated. Detailed results will be reported.