Simulated Satellite Imagery as a Diagnostic Tool for Regional Hurricane Models

Because in-situ observations of tropical cyclones are typically sparse, the analysis of these systems is sometimes limited to satellite-based data. For this reason, it is desirable to use simulated satellite images, derived from model forecasts, as a basis for forecast verification and evaluation. The Environmental Modeling Center (EMC) produces operationally eight different simulated satellite imagery products based on forecasts produced by the operational and experimental versions of the Hurricane Weather Research and Forecasting (HWRF) model. This study utilizes several cases from the 2011 season to show how the suite of simulated satellite images may be used to better understand model behavior. Results show that the simulated imagery can facilitate near real-time diagnostics that have not yet been possible.

The simulated satellite products are produced using the Community Radiative Transfer Model (CRTM) and include four simulated Geostationary Operations Environmental Satellite (GOES) products, Channels 2,3,4 and 6. Four additional microwave products, intended to mimic the AMSRE instrument are also produced. Microwave frequencies include horizontal and vertically polarized 36.5 GHz and 89GHz. While it would be unreasonable to expect the model to reproduce exact convective features, it is possible to subjectively verify the general structure of the forecasted storm through the comparison of real and simulated of microwave and IR images. Subjectively, results suggest that the general structure of the modeled storms can differ significantly from the observed storm, within the first few hours of the simulation. Much like real satellite images, the simulated satellite images are heavily dependent on the presence of hydrometeors within the modeled storm. This suggests that it may be possible to use the simulated satellite imagery as a verification tool for model microphysics. For instance, a strong ice signal in the simulated 89 GHz imagery, in a location where other products suggest that convection is not occurring, might indicate a problem with how the model handles ice particles. A more objective method, which can be applied to both real and simulated satellite images, is the Dvorak technique. Because the Dvorak technique is heavily dependent on the overall cloud pattern, objective T-numbers can be used to compare modeled storm structure with the observed storm. Initial results from the Dvorak analysis indicate that the operational HWRF tends to develop features synonymous with strong hurricanes, such as a well-defined eye, too quickly in the lifecycle of modeled storms, although the reasons behind this behavior remains unclear.