Estimation of the Raindrop Size Distribution from Polarimetric Radar Data: a Double-Moment Normalization Approach

The raindrop size distribution (DSD) statistically describes the microstructure of liquid precipitation. It is highly variable in space and time. In weather prediction models and radar remote sensing applications it is useful and often required to estimate the DSD at areal scale, or in remote locations in which no direct measurements are available. For this reason, retrieval of the DSD from weather radar data has been a long-standing goal. Polarimetric radar variables provide information on not only the concentration but also the shape of raindrops, and are thus useful for the retrieval of DSD parameters. We present a new technique for estimation of the DSD using such polarimetric variables, which uses an established double-moment normalization of the DSD. Two DSD moments are estimated from polarimetric radar variables, and the DSD is reconstructed, taking advantage of the relative invariance of the generic shape function of the double-moment normalized DSD. As part of the suggested technique, a method to treat noise in measured radar data and therefore to improve DSD-retrieval performance is proposed. We present results from tests of the technique using networks of disdrometers together with X-band radars in France, Switzerland, and the United States. In the French domain, radar-retrieved DSDs were also compared to micro rain radar estimations of the DSD aloft. In each domain the technique was tested against a modern DSD retrieval technique, and was found to perform similarly to and often better than the existing technique.