Retrieving microphysical properties of concurrent pristine ice and snow using polarimetric radar observations

Abstract

Abstract. Ice and mixed-phase clouds play a key role in our climate\nsystem because of their strong controls on global precipitation and\nradiation budget. Their microphysical properties have been characterized\ncommonly by polarimetric radar measurements. However, there remains a lack\nof robust estimates of microphysical properties of concurrent pristine ice\nand aggregates because larger snow aggregates often dominate the radar\nsignal and mask contributions of smaller pristine ice crystals. This paper\npresents a new method that separates the scattering signals of pristine ice\nembedded in snow aggregates in scanning polarimetric radar observations and\nretrieves their respective abundances and sizes for the first time. This\nmethod, dubbed ENCORE-ice, is built on an iterative stochastic ensemble\nretrieval framework. It provides the number concentration, ice water content,\nand effective mean diameter of pristine ice and snow aggregates with\nuncertainty estimates. Evaluations against synthetic observations show that\nthe overall retrieval biases in the combined total microphysical properties\nare within 5\u2009% and that the errors with respect to the truth are well\nwithin the retrieval uncertainty. The partitioning between pristine ice and\nsnow aggregates also agrees well with the truth. Additional evaluations\nagainst in situ cloud probe measurements from a recent campaign for a\nstratiform cloud system are promising. Our median retrievals have a bias of\n98\u2009% in the total ice number concentration and 44\u2009% in the total ice water\ncontent. This performance is generally better than the retrieval from\nempirical relationships. The ability to separate signals of different ice\nspecies and to provide their quantitative microphysical properties will open up\nmany research opportunities, such as secondary ice production studies and\nmodel evaluations for ice microphysical processes.\n