Application of Satellite Data and Google Earth Engine for Flood Assessment in the HyDEPP Project

Abstract

This study examines the effectiveness of freely available satellite data and the Google Earth Engine (GEE) platform for early flood detection and information sharing, within the constraints of the COVID-19 pandemic, which restricted field surveys. The case study focuses on Typhoon Ulysses in the Philippines in November 2020, covering three basins: the Cagayan, Pampanga, and the Pasig–Marikina–Laguna (PML) basins. Flooded areas were extracted from Sentinel-2 using the modified normalized difference water index (threshold =0.15) and Sentinel-1 synthetic aperture radar (SAR) using a VH (vertical transmit polarization and horizontal receive polarization) backscatter decrease of 5 dB. These were overlaid with WorldPop 100 m population data to estimate the “potentially affected population,” which was then compared with evacuee counts reported by the National Disaster Risk Reduction and Management Council. Satellite observations taken on the day after landfall enabled the detection of wide-area floods. The estimated affected populations differed from the actual evacuees (e.g., ∼10 times in Cagayan and around 70% in PML). The detection rate of evacuation occurrence was generally high (up to 100% with Sentinel-2), suggesting its value as an early warning indicator. Limitations were identified in detecting urban/under-canopy flooding, as well as uncertainty in population distribution estimates. The GEE-based web application integrated rainfall, track, and inundation data, facilitating rapid information sharing among international teams and demonstrating potential for monitoring during the recovery phase. Future improvements in population data accuracy and automated processing are expected to enhance support for administrative decision-making.

Flood map displayed in the GEE app