Towards using remote sensing and social media data for flood mapping

Ghana's capital, the Greater Accra Metropolitan Area, is vulnerable to flooding. This paper proposes a fusion of satellite imagery and social media data to derive informed flood extent maps and to understand affected population reports during a flood disaster. We use a change detection technique and present an automatic thresholding approach for flood extent mapping using Sentinel-1 images. We explore four different speckle filters and compare them using the VV, VH and VV/VH polarizations to determine the best polarization(s) for delineating flood extents. The VV and VH bands together on Perona-Malik filtered images achieved the highest accuracy with an F1-score of 81.6%. Moreover, we add social media data layers representing tweet text and images posted from the region to highlight different ways these heterogeneous data sources can be used. The obtained signals complement each other and help find flooded roads and areas which are not identified by the satellite imagery analysis.