LAUSR

Abstract Estimating flood impacts on daily activities is vital to disaster risk reduction and urban resilience. Human mobility on complex road networks has caused citizens, especially automobile commuters, to suffer from remarkable impacts on rainy days. However, quantitative studies of these harms have been limited. In this paper, we develop a new approach to estimating flood-affected populations, including their quantity, structure, and spatial distribution, by integrating commuter mobility patterns and road inundation records using multi-source data. Mobility patterns were built by the Gravity model with point-of-interest (POI) data and then downscaled to individual routes by Monte Carlo simulation. Inundation locations and impassable roads were retrieved by combining flood hazard maps with social media data. This approach was applied to Wuhan, a large city in Central China, to estimate the affected automobile commuters under a typical flood event (July 6, 2016) and three flood scenarios (10-year, 20-year, and 50-year). Our approach produces an estimate of the affected population that is more accurate and detailed than existing approaches. It also provides high-resolution impact information for urban studies and can be applied easily to other cities and disasters that include road closures. It will help governments to estimate affected populations rapidly and assess indirect economic losses caused by labor disruptions and work delays.