Sensitivity of Tropical Cyclone Simulations to Air-Sea Flux Parameterizations and Implications for Parameter Estimation

Tropical cyclones (TCs) are fueled by large fluxes of sensible and latent heat from the air-sea interface. These fluxes (as well as momentum fluxes) cannot be explicitly resolved by numerical weather prediction (NWP) models and therefore must be parameterized. Unfortunately, there is a great deal of uncertainty as to the behavior of air-sea surface fluxes, especially under strong (hurricane-force) winds. Thus, the numerous NWP parameterizations of surface fluxes introduce model error into TC forecasts, which limits the accuracy of predictions of TC intensity. In fact, recent research has shown that numerical simulations of TCs are highly sensitive to parameterizations of the surface exchange coefficients for drag (Cd) and moist enthalpy (Ck). In the present study, observations are assimilated via the ensemble Kalman filter (EnKF) to provide better estimates of Cd and Ck in an attempt to improve model forecasts of Hurricane Katrina (2005). Preliminary results are promising: the EnKF system is able to update the values of Cd and Ck based on the ingested observations (including airborne Doppler radar data). In particular, strong correlation structures are found between Cd and the wind fields. Simultaneous state and parameter estimation via EnKF is shown to improve the forecast, in agreement with earlier work. Ongoing tests are investigating the feasibility of adopting EnKF parameter estimation into a real-time hurricane forecasting system.