Retrievals of vertical profiles of ice cloud microphysics from radar and IR measurements using tuned regressions between reflectivity and cloud parameters

A new approach is suggested for retrieving vertical profiles of radiatively important microphysical parameters of ice clouds from measurements of radar reflectivity and estimates of cloud optical thickness. This approach is applicable to ground-based, vertical-mode measurements of pure ice phase clouds observed without blockage from low-liquid water clouds. It involves "tuning" the coefficient of the power law reflectivity-ice water content (Z(e)-IWC) relations for each individual profile of radar measurements. This tuning is done based on the independent estimates of cloud ice water path obtained from layer-mean values of radar reflectivity and cloud optical thickness. After a vertical profile of IWC is retrieved using this approach, a corresponding vertical profile of cloud particle median sizes, D-m, can be easily calculated for an assumed type of the particle size distribution. By assuming that the exponent in the tuned Z(e)-IWC relations changes linearly within a small dynamic range, better results are obtained compared with the case when it is assumed to be constant. The accuracy of the tuned regression approach is assessed by comparing the retrieval results for several high-quality observational cases with data from the "reference" method, which uses additional measurements of Doppler velocities. This method was verified by earlier comparisons with simultaneous in situ data. The specifics of using Doppler information, however, limit the applicability of this method and prevent its real-time applications. It is shown that the relative standard deviations of IWC and D-m retrievals using the tuned regression approach compared with the reference method are about 35% and 20%, respectively. These accuracies are much better than those that can be attained by any a priori chosen fixed regression between reflectivity and cloud microphysical parameters. At the same time, these accuracies are comparable to uncertainties of estimates of cloud parameters from direct in situ sampling. The suggested tuned regression approach is easy to implement and can be used for routine processing of ground-based, and, potentially, satellite-aircraft-based radar measurements of ice clouds.