A Tropical Cyclone Vortex Dynamic-Initialization Methodology Applying In Situ Observations and Experiments Using a Coupled Atmosphere-Ocean Model

The consensus within the tropical cyclone (TC) forecasting community is that numerical weather prediction (NWP) models are unable to forecast TC intensity with consistent skill. The reasons include (but are not limited to) the modeling of air-sea interactions, the use of accurate atmosphere and ocean initial conditions, and grid-length resolutions which are sufficient to simulate scales ranging from eye/eye-wall to the larger meso- and synoptic-scale interactions. A recent effort within the Center for Ocean-Atmosphere Prediction Studies (COAPS), at the Florida State University (FSU), has produced a high-resolution coupled atmosphere-ocean model within which a TC vortex dynamic-initialization algorithm has been implemented. The atmosphere and ocean models which constitute the coupled model are the Advanced Weather Research and Forecasting (WRF-ARW) and the HYbrid Coordinate Ocean Model (HYCOM), respectively. The initial and lateral boundary conditions for the respective models are obtained from the National Center for Environmental Prediction's (NCEP) Global Forecasting System (GFS) and the Naval Oceanographic Office (NAVOCEANO) HYCOM ocean prediction system [Chassignet et al. 2009]. Within WRF-ARW, a TC vortex dynamic-initialization procedure -- similar to that discussed by Kurihara et al. [1993] and Kurihara et al. [1995] and implemented with the Geophysical Fluid Dynamics Laboratory (GFDL) TC prediction model, has been developed. This methodology offers an advancement beyond that of the GFDL scheme. Rather than relying simply on the traditional parameters for maximum wind-speed intensity, the radius of maximum winds, and the environmental flow, in situ TC-relative wind field observations (H*WIND; Powell and Houston [1996]) are used to define the initial TC vortex via the dynamic-initialization of the atmosphere (WRF-ARW) model. Further, grid-length invariant filtering techniques for the 3-dimensional kinematic and thermodynamic attributes of the analysis (first-guess) vortex also provide additional enhancements to the GFDL method. We present NWP simulations for TCs Gustav (2008), Hanna (2008), and Ike (2008), using atmosphere initial conditions defined by both the GFDL [Kurihara et al. 1993; Kurihara et al. 1995] and the aforementioned H*WIND procedure. Simulations are conducting using configurations which employ both the atmosphere (WRF-ARW) model and the coupled atmosphere-ocean (WRF-ARW/HYCOM) model. The simulated TC track and intensity forecasts are skilled against the Hurricane Best-Track Reanalysis [Jarvinen et al. 1984] and suggestions for further improvements to both the TC vortex initialization methodology and the coupled atmosphere-ocean model are provided.