Statistical Comparison of IMERG Precipitation Products with Optical Rain Gauge Observations over Kototabang, Indonesia

Abstract

Satellite-based precipitation estimates play a crucial role in many hydrological and numerical weather models, especially to overcome the scarcity of rain gauge data. Globally gridded rainfall product from Integrated Multi-Satellite Retrievals for Global Precipitation Measurement (GPM) (IMERG) has been used in a wide range of hydrological applications. However, the IMERG is inherently prone to errors and biases. This study evaluated the performance of the IMERG-Final run (IMERG-F) product to estimate rainfall in a mountainous area of Sumatra. Validation was carried out using optical rain gauge (ORG) data for 15 years (2002-2016), at Kototabang, West Sumatra, Indonesia. In general, IMERG-F overestimated rainfall in all time scales. The longer the time scale was, the better the performance of IMERG-F we obtained. This feature was indicated by all quantities of continuous and categorical statistical matrices used. The performance of IMERG-F was lower than in other areas of the Maritime Continent, except for the probability of detection (POD) value. IMERG-F could detect rain very well, including for daily and hourly data, but the false alarm rate (FAR) was also relatively high. Such high FAR value may indicate a significant small-scale spatial rainfall variability in mountainous area of Sumatra.