Assessment of Flash Flood Hazard in A Semiarid Area Through Satellite and Social Media Data Mining



Journal Title

Journal ISSN

Volume Title



Since 2009, flood events have shown an increasing trend in Saudi Arabia. Moreover, most floods occur in cities and may be related to rapid urbanization. Meanwhile, flooding in semiarid areas is usually short-term flash floods within small areas. Therefore, satellite-based flash flood observations are still challenging, while conventional observations are usually sparse in semiarid areas. This study combines machine learning techniques, the statistical analysis of frequency ratio, the logistic regression, and the analytic hierarchy process (AHP) method to identify flood-prone areas in a semiarid area in southern Saudi Arabia. This study integrates thirteen flood-contributing factors such as rainfall, elevation, aspect, slope, flow accumulation, stream power index (SPI), topographic wetness index (TWI), drainage density, distance from the river, distance from roads, soil types, urban area as represented by impervious area, and normalized difference vegetation index (NDVI). Ground observations from social media, such as Twitter and YouTube, validate the prediction results. The objectives of this study include: First, analyze the impacts of the selected thirteen flood-contributing factors. Second, build a decision-tree model between a flash flood and the influencing factors. Third, create a flood susceptibility map in southern Saudi Arabia using the AHP method. The susceptibility map shows the levels of flood risk and their respective percentages in the study area: very low 5%, low 44%, moderate 39%, high 1%, and very high 11%. The results are validated against the ground observations from social media, such as Twitter and YouTube. This research indicates 30.76% commission error and 35.71% omission error from the derived flood susceptibility map with very high and high flood risks, while the overall accuracy can reach 90.37%.



Data Mining, Flash Flood, Flood, GIS, Satellite, Social Media