Using 2D-hydraulic modelling together with SfM and YouTube to estimate peak discharge

Abstract

Despite the major damages caused by El Niño South Oscillation (ENSO) rainfall events, there remains a paucity of discharge records for flash floods in Manta, one of Ecuador’s largest cities. To address this data gap, this research presents a new workflow pipeline to make use of crowd-sourced data; specifically, we couple conventional two dimensional (2D) hydraulic modelling with Structure-from-Motion (SfM) and YouTube videos to obtain reliable post-flood, indirect estimates of peak discharge for an ungauged ephemeral river. A sub-metre-resolution Digital Elevation Model (DEM) of the river reach was obtained using SfM; the estimated “observed” high water levels from YouTube videos were compared against several 2D hydraulic simulations to find the best fit of 0.43 m. Water velocity observations confirmed these discharge estimates. The research herein indicates that novel workflows, including open-source data freely available on social media platforms, can provide indirect peak discharge estimates and, looking further ahead, can become a tool for flood risk management.