Comparative Assessment of Snowfall Retrieval From Microwave Humidity Sounders Using Machine Learning Methods

Accurate quantification of snowfall rate from space is important but has remained difficult. Four years (2007–2010) of NOAA-18 Microwave Humidity Sounder (MHS) data are trained and tested with snowfall estimates from coincident CloudSat Cloud Profiling Radar (CPR) observations using several machine learning methods. Among the studied methods, random forest using MHS (RF-MHS) is found to be the best for both detection and estimation of global snowfall. The RF-MHS estimates are tested using independent years of coincident CPR snowfall estimates and compared with snowfall rates from Modern-Era Retrospective analysis for Research and Applications Version 2 (MERRA-2), Atmospheric Infrared Sounder (AIRS), and MHS Goddard Profiling Algorithm (GPROF). It was found that RF-MHS algorithm can detect global snowfall with approximately 90% accuracy and a Heidke skill score of 0.48 compared to independent CloudSat samples. The surface wet bulb temperatures, brightness temperatures at 190 GHz, and 157 GHz channels are found to be the most important features to delineate snowfall areas. The RF-MHS retrieved global snowfall rates are well compared with CPR estimates and show generally better statistics than MERRA-2, AIRS, and GPROF products. A case study over the United States verifies that the RF-MHS estimated snowfall agrees well with the ground-based National Center for Environmental Prediction (NCEP) Stage-IV and MERRA-2 product, whereas a relatively large underestimation is observed with the current GPROF product (V05). MHS snowfall estimated based on RF algorithm, however, shows some underestimation over cold and snow-covered surfaces (e.g., Greenland, Alaska, and northern Russia), where improvements through new sensors or retrieval techniques are needed.