Detection of Two Major Flooding Events on the Island of Trinidad In 2018 and 2022 Using Google Earth Engine and Sentinel-1 SAR Imagery

This study demonstrates the utility of applying freely available synthetic-aperture radar (SAR) data from the Sentinel-1 satellite within a Google Earth Engine (GEE) HYDRAfloods framework for detecting and mapping flooded areas on a small island. The island of Trinidad, situated in the south eastern region of the Caribbean, experienced its largest flooding event in the past 50 years in October 2018. This was followed by an event of similar magnitude, just 4 years later. GEE was utilized to detect water areas in radar images that were captured during the period of the flood. Flood water areas were separated from permanent water areas by also processing a collection of non-event images. Flood distribution maps were then created and total flooded area was calculated for the study region for both years. A buildings polygon layer derived by Google, was then overlayed on the flood area polygons to visually assess the impacts on infrastructure. The methodology applied in this study was effective for detecting and mapping the flooded areas, however, the temporal resolution (revisit-time) of the Sentinel-1 satellite was still a major limitation, since no images of the peak flooding period of either event were captured.