Diabatic Digital Filter Initialization for Tropical Cyclone Model Forecasting
Chi-Sann Liou (email@example.com)
Naval Research Laboratory, Monterey, CA
A dynamic initialization method using diabatic digital filtering has been developed and tested for initializing high-resolution dynamic models for the prediction of tropical cyclone intensity and structure. The diabatic digital filtering removes unbalanced high-frequency components from the initial conditions through a weighted inverse Fourier transform. In this method, a dynamic model is first integrated adiabatically backward and then diabatically forward to obtain model states used in the inverse Fourier transform. After applying an efficient window function to the weights of the inverse transform, numerical experiments show that 1-h backward and 2-h forward initialization integrations are sufficient to obtain balanced initial conditions for tropical cyclone forecasts. The short integration length not only reduces the cost of the dynamic initialization, but also allows us to use fixed boundary conditions in the initialization integration. Since the diabatic digital filtering initialization includes model consistent diabatic forcing in providing a balanced state, the resulting initial conditions are better balanced than those obtained by static initialization methods, which usually involve adiabatic dynamics only.
The diabatic digital filtering initialization has been implemented and tested in the Coupled Ocean-Atmosphere Mesoscale Prediction System (COAMPS®1). Case studies show that the diabatic digital filtering initialization effectively removes high frequency oscillations in the initial spin-up of COAMPS tropical cyclone forecasts, providing much smoother tendencies in the forecasts of tropical cyclone structure and intensity. Statistics of high-resolution COAMPS tropical cyclone forecasts show that the dynamic initialization improves tropical cyclone track forecast as well.
The diabatic digital filtering initialization has been implemented in a test version of the hurricane Weather Research and Forecasting (HWRF) model. The implementation integrates the diabatic digital filtering routines into the WRF infrastructure, which uses the Earth System Modeling Framework (ESMF) clock utilities to control time integration and object oriented programming and recursive calls to manage nested grids. The implementation method and code have been reviewed by the WRF infrastructure designer. Testing and evaluation is currently underway.