Global Precipitation Measurement (GPM) Satellite Rainfall Data vs Interpolated Ground Measurements in Metro Manila: Error Analysis, Patterns, and Trends from Dense Network of Ground Stations

Abstract

The Global Precipitation Measurement (GPM) mission is a network of satellites that provides improved global observations of precipitation. Satellite-based observations can serve as an alternative source of information if ground-based measurements are not readily available. The data provided by satellites must be validated to ensure reliability and effectiveness of use. In this study, the Integrated Multisatellite Retrievals for GPM (IMERG) Final Run half-hourly product is aggregated to 1-day resolution and compared to interpolated ground measurements derived from sixteen ground stations in Metro Manila for the duration of July 1 – July 8 2014. The space considered in this study is the area within 14.4° – 14.8° latitude and 120.9° – 121.2° longitude, subdivided into twelve 0.1° x 0.1° boxes. The aggregated satellite product is also directly compared to the readings provided by the individual ground stations. The comparisons are done by taking a normalized root-mean-square error between two datasets. The results show that IMERG performed relatively well over the whole Metro Manila area from July 1 to 4 with normalized RMSE values of 0.44 to 0.89 and performed badly from July 5 to 8 with error values of 1.03 up to 4.72. July 3 and July 4 have the highest average amounts of rainfall of 19.475 mm and 19.320 mm and lowest amounts of normalized RMSE error of 0.72 and 0.44, respectively. July 5 had the lowest average amount of rainfall of 1.625 mm and the highest amount of normalized RMSE error of 4.72. The time series of normalized RMSE for each box is also analyzed. The minimum and maximum RMSE values in this analysis are 0.01 and 1.79, respectively. However, onset of typhoon (July 5) can produce normalized RMSE value of 31.77. This can be attributed to the fact that satellite data may observe high water content in the atmosphere but this is not necessarily reflected from ground measurements due to strong winds. Results from the interpolated ground measurements also show that IMERG generally underestimates the amount of rainfall especially in the southern region of Metro Manila which also has the lowest amounts of average error of 0.48 to 0.77. These results indicate that the performance of the satellite product may be dependent on topography (based on the location of the grid boxes) and the amount of rainfall that is estimated.