Use of Synthetic Profiles to Diagnose Simulated Tropical Cyclones in Regional Hurricane Models

Forecasts from regional hurricane models often suffer significant degradation when the structure of the simulated storm departs markedly from the observed storm. To assess and identify the deficiencies that lead to structural errors, it is necessary to develop alternative verification and diagnostic approaches that go beyond the computation of errors and biases in track and intensity. The plethora of aircraft reconnaissance and research flights taken each year offer an opportunity for direct comparison between the kinematic and thermodynamic quantities in the observed storm and those in the modeled storm. To do so, the observations need to be compared within a framework that is consistent with the model's resolution and simulated storm location.

This abstract outlines the application of synthetic profiles to evaluate the intensity and structure of simulated tropical cyclones (TCs) in operational and retrospective runs of the Hurricane WRF model (HWRF). To accomplish this goal, the various NOAA and Air Force Reserve flight level data for a given storm are first standardized into a common NetCDF data file. Because the simulated cyclone does not typically follow the actual path taken by the real cyclone, it is necessary to translate the observational data into coordinates relative to the moving storm center of the actual storm, and then sample the model space along these transects in a frame moving with the center of the simulated storm. The resulting synthetic radial profiles of the model's simulated flight level and surface data can then be directly compared with the observed 1-Hz flight level once an appropriate spatial smoothing is applied. Likewise, the model's simulated surface wind field can be sampled and compared with observed surface wind data from Stepped Frequency Microwave Radiometer (SFMR).

Results will be presented for several storms in the 2010-2013 period with an aim of determining the efficacy and usefulness of this methodology. If time permits, we will also explore the potential to apply the synthetic profile approach to the curving trajectories of dropsondes. As the resolution of regional hurricane models increases to ever finer scales, this approach may provide a more accurate way to look at the low-level vertical structure in simulated storms. Potential real-time applications may also be explored.