Improving flood damage assessments in data-scarce areas by retrieval of building characteristics through UAV image segmentation and machine learning – a case study of the 2019 floods in southern Malawi

• DOI: 10.5194/nhess-21-3199-2021
• Publication Date: 2021-10-27T07:31:27Z
• Authors: Lucas Wouters; Anaïs Couasnon; Marleen C. de Ruiter; Marc J. C. van de...; Aklilu Teklesadik; Hans de Moel
• Keywords: Ground sample distance; Vulnerability

Abstract. Reliable information on building stock and its vulnerability is important for understanding societal exposure to floods. Unfortunately, developing countries have less access availability of this information. Therefore, calculations flood damage assessments use the scarce available, often aggregated a national or district level. This study aims improve current by extracting individual characteristics estimate based buildings' vulnerability. We carry out an object-based image analysis (OBIA) high-resolution (11 cm ground sample distance) unmanned aerial vehicle (UAV) imagery outline footprints. then support vector machine learning algorithm classify delineated buildings. combine with local depth–damage curves economic three villages affected 2019 January river floods in southern Shire Basin Malawi compare conventional, pixel-based approach using land denote exposure. The extent obtained from satellite (Sentinel-1) corresponding water depths determined combining elevation data. results show that OBIA footprints much closer OpenStreetMap data, which tends overestimate. Correspondingly, estimated total lower (EUR 10 140) compared 15 782). A sensitivity illustrates uncertainty derived larger than hazard research highlights potential detailed UAV determine risk data-poor regions.