A New Neural Network Retrieval of Liquid and Snow Water Path Using Radiometer and Radar Observations

Cold-air outbreaks over the high latitudes are known to be mixed-phase, but accurate information on the phase partitioning, which has important implications for both weather and climate, remains difficult to acquire. Optimal estimation methods provide robust error estimates, but also require a-priori knowledge of the vertical atmospheric profile and/or are computationally expensive to run. Here we present a neural network approach to retrieve not only liquid water path (LWP), but also snow water path (SWP) using passive microwave measurements combined with radar reflectivities. The approach is an extension of Cadeddu et al 2009, with the addition of radar reflectivity. The neural network is trained using the Passive and Active Microwave radiative TRAnsfer (PAMTRA; Mech et al., 2020) tool applied to output from a large-eddy simulation of a cold-air outbreak sampled during the Cold-Air Outbreaks in the Marine Boundary Layer Experiment (COMBLE) campaign along with model output from other cold-air outbreaks. Brightness temperatures corresponding to the four sidebands of an upward-looking G-band Vapor Radiometer, along with the vertically-integrated radar reflectivity from a zenith-pointing 95 GHz Wyoming Cloud Radar are simulated from the perspective of a near-surface aircraft track. The radar reflectivity helps discriminate the snow contribution to the brightness temperatures. The neural network regression is thereafter tested on simulations of different cold-air outbreaks, and using measurements from the ARM North Slope of Alaska observatory. This simple neural network approach is being developed towards providing robust, computationally-efficient, near-real-time measurements of LWP and SWP during the Cold Air Outbreak Experiment in the Sub-Arctic Region (CAESAR) campaign in February-April 2024.