Comparison of Precipitation Phase Algorithms: A Ground Validation Study for GPM

The Global Precipitation Measurement (GPM) Mission has four products; Dual Frequency Precipitation Radar (DPR), Combined radar radiometer, Goddard Profiling and the Integrated Multi-Satellite Retrievals algorithms that monitor and classify precipitation phase on a global scale. The phase of hydrometeors is directly linked to the event accumulation and intensity. To validate these products, an extensive and continuous network of ground-based observations is necessary.

This study evaluates the performance of the NEXRAD S-band radar Hydrometeor Identification (HID) deterministic algorithm, the Multi-Radar Multi-Sensor (MRMS) deterministic algorithm, and the Sims and Liu 2015 (SL15) and Jaunet and Kirstetter 2018 (JK18) probabilistic algorithms; the last three of which, use surface temperature and wet-bulb temperature. The Automated Surface Observing System (ASOS) present weather sensor (PWS) was considered as the reference. The performance of all the algorithms was evaluated during cold rain, rain to snow transition, and snow events.

The focus of this project was the visualization of the ground observation dataset in a user-friendly package. Three modules were created to construct these time series where PWS was the reference. The event-based diagrams were able to visualize the performance of each algorithm especially during phase transitions. The quantitative assessment was done through percent agreement for deterministic and Brier score for probabilistic algorithms. It is feasible that HID can report frozen while liquid precipitation is observed at the surface. The observed and/or model output sounding is examined to determine the phase change in vertical column.