Validation of Simulated Hurricane Raindrop Size Distributions using Dual-Polarimetric Radar

Due to its ability to remotely sense extreme environments and its sensitivity to precipitation, weather radar has been an indispensable tool for observing hurricanes. The United States' Doppler radar network upgrade to dual-polarization from 2011-2013 provides new observations for estimating the microphysical properties of precipitation of tropical cyclones (TCs). Dual-polarimetric radar observations can provide a targeted validation of the drop size distribution (DSD) by estimating the phase, size, and number concentration of precipitation at high resolution over a broad area. The radar upgrade allows for new assessments of the structure of TCs that approach the U.S. coastline to both improve our understanding of microphysical processes and validate the performance of numerical models.

The work to be presented investigates the use of these new observations using the cases of Hurricanes Arthur and Ana (both 2014) by comparing observed drop size distributions (DSDs). Hurricane Arthur formed in the Atlantic basin and impacted the southeast U.S. coast, while Hurricane Ana formed in the Central Pacific and impacted the Hawaiian Islands. We also derive estimates of the DSDs created by various microphysical parameters in high-resolution WRF simulations of these hurricanes by using T-matrix scattering calculations to create synthetic differential reflectivity observations, a novel approach in hurricane simulations. We find in both cases that the overall track, structure, and intensity of the storms is reasonably well simulated. However, the single-moment microphysics parameterizations do not simulate the range of DSDs well, resulting in frequency distributions that are too narrow. On the other hand, double-moment schemes vary in their accuracy, with some schemes producing reflectivity and differential reflectivity that are too large compared to the observations. The spectral bin model most accurately reproduces the observed DSDs and the Best Track intensity for both storms. However, the bin model requires approximately 60 times the computational power compared to the bulk schemes, making it a valuable research tool but currently unfeasible for operational TC prediction.