Validating the Integrated Multisatellite Retrievals for Global Precipitation Measurement in Terms of Diurnal Variability With Hourly Gauge Observations Collected at 50,000 Stations in China

Abstract

AbstractThe diurnal cycle of precipitation is a key feature in the Earth system and related to multiscale physical processes. The Integrated Multi‐satellitE Retrievals for Global Precipitation Measurement (IMERG) data are expected to improve its presentation of diurnal cycle, but how it really performs remains poorly known. This study compares the diurnal characteristics of the half‐hourly IMERG V05 Final Run product with the hourly rain gauge data collected at approximately 50,000 automatic weather stations in China during a 3‐year period (2014–2016). Our results show that IMERG performs well in terms of the diurnal cycle of precipitation amount but less well in terms of frequency and even worse in terms of intensity. Ground observations show that the frequency and intensity are rather stable for the diurnal cycle; however, IMERG shows strong fluctuations. Significant inverse correlation is found between the diurnal cycles of IMERG precipitation frequency and intensity; thus, the biases of these two variables offset and result in a better estimate of amount. A low probability of detection (POD≈50%) and a high false alarm ratio (FAR≈50%) are the major reasons for the inaccurate diurnal cycles of precipitation intensity and frequency. In particular, IMERG presents a false peak of frequency at approximately 23:00–01:00 (Beijing Time) because a large proportion of the data sources at this time contain the indirect estimates from thermal infrared observations. The significant underestimation of the frequency and overestimation of the intensity at approximately 09:00–11:00 and 20:00–22:00 is because the estimates primarily come from less accurate cross‐track microwave sensors.