Drop-Size Distribution Studies in N. Alabama as Applied to Simulated and Observed C-Band Dual-polarimetric Radar Measurements

This study examines 2D Video Disdrometer (2DVD) drop size distribution (DSD) and coincident ARMOR C-band dual-polarimetric radar data for a variety of storm types and regimes including inland tropical cyclone rain bands, mesoscale convective systems, isolated thunderstorms, and cold season stratiform precipitation events. The methodology we use exploits previous raindrop axis ratio-diameter relationships, combined with diameter-binned 2DVD data and a T-Matrix/Mueller scattering model to develop customized equations relating C-band polarimetric radar parameters to various properties of observed rainfall (e.g., moments of the DSD including median volume diameter).

In the course of our study we experimented with various techniques to fit the DSD data Gamma size distributions for input to the radar scattering model. We applied non-linear least squares curve fitting techniques, conventional method of moments (MOM), and truncated MOM approaches. We then input the various DSD fits to a T-matrix/Mueller simulation code and non-linearly regressed equations for the observed rainfall characteristics with the simulated radar data for the entire range of rainfall regimes sampled. Interestingly, for the MOM approach when performing combined “internal consistency” regressions between Z, KDP and ZDR for ARMOR data in N. Alabama we arrived at a coefficient and exponent nearly identical to those diagnosed by Bringi et al. (2006) for tropical monsoon rainfall in Baiu Front precipitation observed near Okinawa. Similar regressions using the non-linear least squares regression provided far different relationships with much larger standard error and much poorer comparison to moments computed using the original disdrometers data.

To investigate regime variability, we have further partitioned the 2DVD-observed DSDs into both tropical cyclone remnant and non-tropical events. Having the partitioned the data we have been exploring the sensitivity of radar-rain property equations as a function of their impact on the radar retrievals for the various meteorological regimes. Importantly, the regressed equations exhibit a marked sensitivity to resonance effects at C-band in association with the maximum drop size (Dmax) assumed when simulating the dual-polarimetric variables. We find that this sensitivity is regime specific because of the relative lack (presence) of large (D>4 mm) drops in tropical (non-tropical) DSDs over Huntsville.