A Diagnosis of Oceanic Precipitation in IMERG-GMI

Abstract Diagnosing errors in spaceborne oceanic precipitation estimates is difficult due to complicated multisatellite algorithms and limited surface-based measurements. The Global Precipitation Measurement (GPM) mission helps to alleviate these challenges with NASA’s Integrated Multi-satellitE Retrievals for GPM (IMERG) product, which is transparently designed to encourage community validation activities, and the GPM Validation Network, which collects observations across global precipitation regimes from over 100 ground-based weather radars to serve as reference datasets for the GPM precipitation products. This study uses the GPM Validation Network’s oceanic precipitation observations from 32 island and coastal radars to diagnose the performance of IMERG V06B and V07B Final Run products during GPM Microwave Imager (GMI) overpasses (i.e., IMERG-GMI) in the period June 2014–September 2021. Errors are traced from the input level 2 (satellite footprint) Goddard profiling algorithm climate (GPROF-CLIM) GMI product through the successive gridding, calibration, and precipitation distribution restoration steps of IMERG’s level 3 (gridded) algorithm. Results highlight that IMERG-GMI V07B outperforms V06B in detecting and quantifying oceanic precipitation, with a significant improvement over high-latitude ocean (V06B: +143%; V07B: +50%). Furthermore, there is a clear oceanic latitudinal trend in the mean relative bias of IMERG-GMI V07B (high latitude: +50%; midlatitude: +10%; tropical: −41%), which largely traces back to GPROF-CLIM V07 (high latitude: +22%; midlatitude: −8%; tropical: −44%), with bias differences driven by IMERG’s passive microwave calibration scheme. This error tracing approach supports future IMERG algorithm developments by disentangling how algorithm steps enhance or mitigate errors. Significance Statement Most precipitation occurs over the ocean, yet precipitation products are rarely evaluated there due to a lack of surface-based measurements. This study utilizes the GPM Validation Network’s datasets from coastal and island radars to conduct the first oceanic evaluation of the NASA IMERG V07B multisatellite precipitation product. The performance of IMERG at different successive algorithm steps is assessed to understand which processing steps reduce or increase errors and to support future algorithm improvements. The latest IMERG version provides more accurate and reliable precipitation estimates than the preceding version. However, IMERG-GMI is biased high by +50% and low by −41% over the high-latitude and tropical oceans, respectively.