Using the JPL Tropical Cyclone Information System (TCIS) to evaluate physical parameterizations in WRF simulations of hurricanes: Can we find the right set of parameterizations?

Improving forecasting of hurricane intensity remains a significant challenge for the research and operational communities. Even track forecasts, generally more accurate, still exhibit a level of uncertainty that needs to be decreased. Further forecast improvements require the use of high-resolution cloud-resolving models with accurate representation of the important physical processes. Yet, determining which parameterizations produce realistic forecasts and which do not remains a difficult task even today. The problem is accentuated by the difficulty of obtaining high-resolution 3D in-situ observations within these very powerful storms.

To address the issue, we have developed a comprehensive set of satellite remote-sensing observations that describe the 3D structure of the storm and capture the important environmental characteristics (the JPL Tropical Cyclone Information System –TCIS). Coming soon, we will also incorporate into the database airborne remote-sensing and in-situ observations.

We use the TCIS database, together with instrument simulators (observation operators), to evaluate the impact of different physical parameterizations in WRF model simulations of hurricanes. In particular, we address the model forecast sensitivity to microphysical parameterizations and to the parameterizations of the boundary layer processes.

We produce ensemble simulations, using different model physics. We then use the WRF-produced hydrometeors and thermodynamic variables to forward simulate satellite observables (e.g. radar reflectivity, microwave brightness temperatures and scatteromer-like backscattering cross-sections). Next, we use a variety of techniques (e.g. scatter plots, 2D maps, azimuthal averages, CFADs and other PDFs) to compare model-produced observables to such measured by TRMM-PR, TRMM-TMI and QuikSCAT with the goal to find the set of parameterizations that produce the most realistic storms.

Our preliminary results indicate that multi-parameters satellite observations can, indeed, help discriminate among simulations with different physical parameterizations, pointing to the set of assumptions that produce the closest to observed hurricane structures.