Theoretical framework for updating cross‐sections in river models using non‐standard information coming from crowdsourced images

Abstract During high rainfall events, one reason for flooding in urban rivers is hydraulic blockages or obstructions in channels. These may cause damming and an immediate increase in upstream water levels, leading to inundation of surroundings. The accuracy of flood models is strongly dependent on the representation of the topography of floodplain and river geometry, so incorporation of such blockages and obstructions in rivers' model geometry needs to be taken into account. The traditional techniques to address this problem are field surveys for cross‐section observations, which are time‐consuming and require skilled personnel. One possible cost‐effective solution is in engaging citizens to take photographs of such reaches in the river. Translating the information from these images to a format that is understandable and usable by the models is a problem in itself. This paper proposes a methodology to incorporate the blockages data from geotagged images into a hydrodynamic model, such that the model can be updated with better representation of the river geometry. The developed methodology includes the calculation of new coordinates of river cross‐section based on flow area reduction, as assessed by an expert based on what is visible in the images. A MATLAB GUI was designed to automatically update the river geometry and to run HEC‐RAS with an updated geometry. The study proposes that the developed procedure can act as a blueprint for the use of crowdsourced data in observing cross‐sectional changes in the urban rivers due to blockages and/or obstructions and quantify their inundating effects on the surroundings. The methodology is currently demonstrated using hypothetical crowdsourced data. As a proof of concept, it is applied to a reach of the Kifissos River in Athens, Greece.