Vortex structures from an ensemble-based data assimilation and hurricane prediction system during the 2009 Atlantic tropical cyclone season

An important issue regarding hurricane prediction is the improvement of the initial estimate of the vortex structure through the use of data assimilation. Although many operational models make use of data assimilation techniques, most use fixed error covariance statistics when mapping observations to model grid points, which may not be adequate for defining the smaller scale structure of tropical cyclones. As the intensity of tropical cyclones may depend both on the large-scale environmental conditions as well as the smaller-scale storm structure, not accurately initializing across all scales may have significant impacts on intensity forecasts. Here, a 96-member Ensemble Kalman Filter (EnKF), which assimilates observations using flow-dependent covariance statistics, is used with the Data Assimilation Research Testbed (DART) assimilation system to initialize forecasts in real-time for the Advanced Hurricane WRF (AHW) model during the 2009 Atlantic hurricane season.

Results from both the analyses and corresponding AHW forecasts of select cases will be presented and compared with operational models, with an emphasis on the vortex structure to better understand the sensitivity of the initial structure to intensity change. AHW forecasts were initialized from an ensemble member containing the minimum position and intensity error with respect to NHC best track data, and used a 2-way nested grid system with grid spacings of 12, 4, and 1.33 km. Results show that the analyses and forecasts perform comparably well with other operational models, particularly with regard to minimum sea-level pressure and wind within the outer-most vortex radii, and that environmental shear and model physics both contribute to intensity error, especially in the case of Erika. Senstivity experiments show that parameterized convection in Erika leads to reduced intensity skill, and further analysis isolating the impact of the environmental shear will be presented.