Evaluation of Precipitation Products in the Mid-Atlantic and Storrs, CT

Precipitation is a driving force for the water cycle and is one of the key sources of weather extremes such as drought and flooding. The changing global climate and consequent increase in extreme weather emphasizes the dependence of daily life on precipitation. Observation- and model-based precipitation products are essential for mapping global precipitation but rely heavily on reliable and accurate precipitation measurements. Unfortunately, precipitation measurements (gauges, weather radars) are not available over vast majority of global land. This is due to topography, land use coverage, cost, and remoteness. With the addition of global ocean coverage (> 70%), precipitation climatology relies on spaceborne precipitation retrievals and model outputs. NASA’s multi-satellite precipitation product, IMERG, operated and funded under the Global Precipitation Measurement (GPM) mission, is widely used by operational agencies in many fields. NOAA’s Multi-Radar Multi-Sensor (MRMS) product, on the other hand, has been widely employed to validate satellite and model precipitation estimates over the United States among many other applications. MRMS itself, however, needs to be validated by independent precipitation measurements.

The GPM ground validation program has been deploying Platforms for In situ Estimation Rainfall Systems (PIERS) at granted institutes across US. This study uses six PIERS+ sites which include a PARSIVEL disdrometer and dual tipping bucket gauges. The sites are in Mid-Atlantic region with an additional site in Connecticut. There were additional instruments including a Pluvio weighing bucket gauge at two sites, two more tipping buckets at one site and three more PARSIVEL disdrometers at another site. The study focused on event rainfall totals for the first five months of 2024. The PARSIVEL disdrometer was the reference for the event definition and phase identification, while one of the tipping bucket gauges was selected as reference for rainfall totals following comparison with the collocated instruments. This study investigates the performance of a) MRMS, b) NOAA’s High Resolution Rapid Refresh (HRRR) model, c) European weather forecast center’s ERA5 reanalysis, and d) NASA Global Modeling and Assimilation Office’s MERRA-2 reanalysis. PIERS+ instrumentation is not part of any of operational algorithm and therefore provides independent evaluation. Among PIERS+ sites, the Chesapeake Bridge-Tunnel, Virginia is unique for its location but may suffer from sea spray, sea salt and windy conditions. The MRMS, gauge adjusted radar product, overestimated event totals at all sites except Greenbelt, MD. The distance of the site to weather radar and a reliable operational gauge plays an important role in MRMS performance. All three model and reanalysis underestimated the event totals for majority of sites. Based on mean absolute error, the performance of the HRRR, ERA5, and MERRA2 had a wide range from one site to another. MERRA2 outperformed HRRR and ERA5 despite its coarser resolution.