Blended Satellite and Ground Precipitation Measurements Over the Philippines

The Philippines is an archipelago that receives huge amount of precipitation that varies significantly with the season and location. However, due to sparse synoptic stations in the country, there is inadequate representation of precipitation in some areas. This study analyzed the inter-relationship between available ground and satellite-based measurements and demonstrate a method to blend these data to better represent precipitation in the country.

The study utilized synoptic station measurements from PAGASA and satellite precipitation products from TRMM and GPM. Comparison between station and satellite measurements show that these data agree well at the monthly timescale. Satellite products are prone to ground measurement mismatch at the daily scale. These results may be attributed to the synoptic scale representation of the ground data. Precipitation data from a dense rain gauge network were then used to complement synoptic measurements and capture mesoscale variations. Results showed that as the rain gauge is situated farther from the synoptic station, the agreement between their measurements decreases. Because of this relationship, inverse distance weighting (IDW) was done to combine rain gauge measurements representative of precipitation in a 0.1° cell. IDW values were found to correlate well (r = 0.58 to 0.97) with the overlapping synoptic station measurements inside the cell.

Regression kriging was done to combine ground (IDW values) and satellite (GPM) precipitation measurements. The blended product provides a more detailed monthly precipitation over the country showing least difference to GPM for the months February to May. Compared to GPM, higher precipitation values were observed from the fused product in places with dense rain gauges. In contrast, lower precipitation values were observed from the fused product in places of higher altitude with few rain gauges. The fused product at daily timescales showed a general increase in precipitation with respect to GPM, improving on the countrywide satellite product underestimation. This study successfully demonstrated that by combining satellite with ground data, a better representation of precipitation is achieved.