Comparison of Monthly IMERG Precipitation Estimates with PACRAIN Atoll Observations

Abstract

AbstractSatellite-based precipitation estimates provide valuable information where surface observations are not readily available, especially over the large expanses of the ocean where in-situ precipitation observations are very sparse. This study compares monthly precipitation estimates from the Integrated Multi-satellitE Retrievals for GPM (IMERG) with gauge observations from 37 low-lying atolls from the Pacific Rainfall Database for the period June 2000 – August 2020. Over the analysis period, IMERG estimates are slightly higher than the atoll observations by 0.67% with a monthly correlation of 0.68. Seasonally, DJF shows excellent agreement with a near-zero bias, while MAM shows IMERG is low by 4.6%, and JJA is high by 1.2%. SON exhibits the worst performance, with IMERG overestimating by 6.5% compared to the atolls. The seasonal correlations are well-contained in the range 0.67 – 0.72, with the exception of SON at 0.62. Furthermore, SON has the highest RMSE at 4.70 mm/day, making it the worst season for all metrics. Scatterplots of IMERG versus atolls show IMERG, on average, is generally low for light precipitation accumulations and high for intense precipitation accumulations, with best agreement at intermediate rates. Seasonal variations exist at light and intermediate rate accumulations, but IMERG consistently overestimates at intense precipitation rates. The differences between IMERG and atolls varies over time but does not exhibit any discernable trend or dependence on atoll population. The PACRAIN atoll gauges are not wind-loss corrected, so application of an appropriate adjustment would increase the precipitation amounts compared to IMERG. These results provide useful insight to users as well as valuable information for future improvements to IMERG.