A Dual-Frequency Radar Retrieval of Two Parameters of the Snowfall Particle Size Distribution Using a Neural Network

Abstract With the launch of Global Precipitation Measurement Dual-Frequency Radar (GPM-DPR) in 2014, renewed interest retrievals snowfall atmospheric column has occurred. The current operational GPM-DPR retrieval largely underestimates surface accumulation. Here, a neural network (NN) trained on data that are synthetically derived from state-of-the-art ice particle scattering models and measured situ size distributions (PSDs) is used to retrieve two parameters PSD: liquid equivalent mass-weighted mean diameter normalized intercept parameter . Evaluations against test dataset showed statistically significantly improved water content (IWC) relative standard power-law approach an estimate algorithm. Furthermore, estimated median percent errors (MPE) were −0.7%, +2.6%, +1% for , IWC, respectively. An evaluation three case studies with collocated radar observations microphysical shows NN MPE −13%, +120%, +10% applied directly provides default algorithm, removing algorithm’s ray-to-ray instabilities recreating high-resolution results within 15% MPE. Future work should aim improve by including PSD collected more diverse conditions rimed particles. different desired outputs such as shape rate could be included future iterations.