Analysis of a Dual-Wavelength Radar Technique for Estimating Liquid Water Content and Droplet Size

Previous studies have suggested a dual-wavelength radar system comprised of X- and K_a-bands is well-suited for ground-based remote sensing of single and mixed-phase (ice and liquid) clouds. Transmitted radiation at K_a-band is measurably attenuated by liquid water; the range-differentiated difference between the reflectivities is proportional to the amount of liquid. In practice, factors such as non-Rayleigh scattering, the presence of ice crystals, measurement errors and sensitivity of the instruments have confounded evaluations of liquid water content (LWC) retrievals during field trials.

In this paper a numerical analysis of a dual-wavelength radar technique for estimating LWC and particle size is presented. The key component for modeling range- dependent dual-wavelength reflectivities is the droplet spectrum. The study uses actual spectra measured from the NASA Twin Otter research aircraft in the Ohio Valley for modeling dual-wavelength radar reflectivity range profiles. The measured spectra included a wide range of conditions: cloud droplets (maximum diameter less than or equal to 50 m), drizzle (20 - 500 µm) and rain (> 500 µm), with LWC of 0.01 - 0.5 g m^-3. Radar reflectivities were calculated for both radar wavelengths using these spectra. A linear least-squares fitting of reflectivity difference (X minus K-band) along an assumed radar beam through a cloud was used to estimate attenuation.

The analyses confirm that retrieval of LWC and median volume diameter from radar reflectivity alone is not feasible. An earlier study defined a radar estimated size (RES), based on reflectivity and attenuation measurements available from the dual-wavelength system. Analysis of the simulation results suggests in the case of cloud and drizzle conditions, dual-wavelength radar observations are capable of retrieving LWC and RES. Mixed-phase radar measurements were simulated using a one-dimensional Gaussian distribution with a specified mean, standard deviation and correlation length for ice reflectivity. Small ice crystals (<~1 mm diameter) do not affect the LWC retrieval but the RES estimate is biased upward by their contribution to total reflectivity. In the case of non-Rayleigh scattering from larger ice crystals or raindrops, the difference in reflectivity between the two wavelengths is no longer monotonically increasing as is the case for pure Rayleigh scattering; in these cases, the local minimum reflectivity differences were used to estimate attenuation. As a result, spatial resolution of the LWC estimate is compromised in the mixed-phase regions. The effect of radar measurement error on attenuation estimation was also investigated using various range-averaging lengths. Based on these analyses, an optimum design and data processing scheme for a dual-wavelength system is presented.