Abstract and subjects
In Part I of this study, a new algorithm for retrieving the latent heat field in tropical cyclones from airborne Doppler radar was presented and fields from rapidly intensifying Hurricane Guillermo (1997) were shown. In Part II, the usefulness and relative accuracy of the retrievals is assessed by inserting the heating into realistic numerical simulations at 2-km resolution and comparing the generated wind structure to the radar analyses of Guillermo. Results show that using the latent heat retrievals as forcing produces very low intensity and structure errors (in terms of tangential wind speed errors and explained wind variance) and significantly improves simulations relative to a predictive run that is highly calibrated to the latent heat retrievals by using an ensemble Kalman filter procedure to estimate values of key model parameters.
Releasing all the heating/cooling in the latent heat retrieval results in a simulation with a large positive bias in Guillermo's intensity that motivates the need to determine the saturation state in the hurricane inner-core retrieval through a procedure similar to that described in Part I of this study. The heating retrievals accomplish high-quality structure statistics by forcing asymmetries in the wind field with the generally correct amplitude, placement, and timing. In contrast, the latent heating fields generated in the predictive simulation contain a significant bias toward large values and are concentrated in bands (rather than discrete cells) stretched around the vortex. The Doppler radar based latent heat retrievals presented in this series of papers should prove useful for convection initialization and data assimilation to reduce errors in numerical simulations of tropical cyclones.