Mixed-phase cloud phase partitioning using millimeter wavelength cloud radar Doppler velocity spectra

Abstract

Retrieving and quantifying cloud liquid drop contributions to radar returns from mixed-phase clouds remains a challenge because the radar signal is frequently dominated by the returns from the ice particles within the radar sample volume. We present a technique that extracts the weak cloud liquid drop contributions from the total radar returns in profiling cloud radar Doppler velocity spectra. Individual spectra are first decomposed using a continuous wavelet transform, the resulting coefficients of which are used to identify the region in the spectra where cloud liquid drops contribute. By assuming that the liquid contribution to each Doppler spectrum is Gaussian shaped and centered on an appropriate peak in the wavelet coefficients, the cloud liquid drop contribution may be estimated by fitting a Gaussian distribution centered on the velocity of this peak to the original Doppler spectrum. The cloud liquid drop contribution to reflectivity, the volume mean vertical air motion, subvolume vertical velocity variance, and ice particle mean fall speed can be estimated based on the separation of the liquid contribution to the radar Doppler spectrum. The algorithm is evaluated using synthetic spectra produced from output of a state-of-the-art large eddy simulation model study of an Arctic mixed-phase cloud. The retrievals of cloud liquid drop mode reflectivities were generally consistent with the original model values with errors less than a factor of 2. The retrieved volume mean vertical air velocities reproduced the updraft and downdraft structures, but with an overall bias of approximately −0.06 m s−1. Retrievals based on Ka-band Atmospheric Radiation Measurement Program Zenith Radar observations from Barrow, Alaska, during October 2011 are also presented.