Estimation of drop-size distributions from deconvolved doppler radar data

A 915 MHz Doppler radar was able under certain conditions to provide Doppler spectra containing separate and distinct modes due to scattering by rain and by the clear air. The rain spectra are broadened and shifted on the velocity axis due to turbulent air motions. Information in the clear air mode is used to correct for these effects, leaving in the rain spectra the reflectivity-weighted distribution of drop fall speed, from which the drop-size distribution can be computed.The fast Fourier transform and two iterative methods of deconvolution were applied to simulated Doppler spectra. We concluded that the iterative methods of deconvolution are more appropriate than the fast Fourier transform. Of the two iterative methods, Jansson's method of deconvolution, including prefiltering of the data, was found to give better results than Van Cittert's method.Jansson's method was applied to biomodal spectra obtained from the HARP project of 1990 in Hawaii and to spectra obtained from a similar radar operated in Montreal. Evident is the different characteristics of rainfall in both regions and the appropriateness of the same deconvolution method to different classes of spectra