Street-scale hydrodynamic estimation from social media videos: A systematic approach to urban floods data collection

Abstract

An increasing number of evidence highlights how climate change and urbanization are contributing to exacerbate floods impacts. In this framework, flood modelling assumes a key role in supporting the analysis of floods dynamics, especially in urban areas. Recent advances in this topic enabled detailed street-level studies, offering significant potential for flood reconstruction, nowcasting, and forecasting. However, calibration and validation of hydrodynamic models face challenges due to limited availability of flood data. Recent studies highlighted the potential of social media as a valuable resource for urban flood analysis, yet significant challenges persist, particularly in leveraging videos data to retrieve floodwater characteristics, leading to the loss of a relevant amount of potentially useful information. In this paper, to tackle this issue, a five-step workflow for the systematic research and extraction of key hydrodynamic variables from flood-related videos uploaded on social media is proposed. The aim of this procedure is the retrieval of diffused and quality-controlled estimates of floodwater characteristics to support hydrodynamic modelling and dampening the gap due to the lack of field measurements. The workflow was tested to the flood occurred in 2020 in the city of Crotone (southern Italy). The results underscore the potential of the proposed procedure to provide detailed data for flood impact assessment, paving the way for improved street-level hydrodynamic studies and model validation. This approach not only could enhance the quality control of the dataset but also allows for the limitation of information loss, which is critical for supporting a broader distributed validation.