Assessment of satellite precipitation product estimates over Bali Island

Satellite precipitation product estimates (SPPEs) provide rainfall data on regional and global scales and have the potential to be applied in various fields. Several satellite precipitation estimates are available for their various features in retrieval algorithms, used sensor instrument, spatial-temporal resolution, and coverage area. Global Satellite Mapping of Precipitation (GSMaP), Integrated MultisatellitE Retrievals (IMERG), and Climate Hazards Group Infrared Precipitation with Station (CHIRPS) are global coverage precipitation datasets with high spatial resolution (0.1 degrees-0.05 degrees) and high temporal resolution (from 30 min to daily updates). The objective of this study was to assess the performance of GSMaP, IMERG, and CHIRPS over Bali Island from 2015 to 2017 in terms of ground rain gauge data over a high density of rain gauge stations (27 in-situ rain gauges) and at various elevations, rainfall intensities, and temporal scales (i.e., daily, penta-days, monthly, and seasonal). A traditional point-to-pixel-based method along with a new introduced continuous, categorical, and volumetric statistical indices comparison approach were implemented to evaluate satellite products. The assessment results demonstrated that IMERG products achieved the highest performance on daily, penta-day, and seasonal time scales, whereas CHIRPS outperformed the other two products on the monthly time scale. Moreover, IMERG was more efficient in detecting rainfall events at different alludes, but it tended to overestimate rainfall events at high alludes. With respect to their abilities to detect rainfall events, GSMaP, IMERG, and CHIRPS tended to underestimate the frequency of light rainfall events (0-1 mm/day) and heavy rainfall events (> 50 mm/day) but overestimate the frequency of moderate rainfall events (5-10 mm/day). Our result not only highlight IMERG products exhibited better performance in comparison to GSMaP and CHIRPS in Bali Island but also recommend that further improvement on the precipitation estimate algorithm is required by considering complex terrain over small island in the maritime continent area.