UAV imagery in natural disasters: Real-time damage assessment of flash flooding events
Alejandro Román, Antonio Tovar‐Sánchez, Marcos Larrad, Francisco José Rubiano-Sánchez, José Manuel Zafra, Rafael Piñeiro Rodríguez, Ángel Castillo, Félix A. López, Ana Lucía Vela, Ana Allende, Glòria Sánchez, Alberto Martínez-Alonso, Daniel Samper, Juan Carlos García-Davalillo, Inés Galindo, Gabriel Navarro
Abstract
Unmanned Aerial Vehicles (UAVs) have proven to be essential for real-time emergency management, impact forecasting, and damage assessment of flash flood events. This study integrates the results obtained during the cut-off low of October 29th, 2024, in Valencia (Spain), from deploying a wide range of UAV-mounted sensors to monitor the emergency in real-time, alleviate impacts, and support the early stages of the reconstruction phase. Several UAV operations conducted over six days in November and December 2024, allowed to: i) generate a 3D reconstruction and a point cloud of the entire Poyo's gully and the main affected buildings at neighbouring municipalities; ii) use multispectral technology to cover some areas where sludge of varying composition and thickness had accumulated; iii) employ optical RGB technology to develop an algorithm for detecting vehicles accumulated in open fields or trash items carried by the water flow; iv) sectorize the affected area and select sampling sites for microbiological, chemical, and physicochemical analyses; v) quantify potential atmospheric pollution in populated areas through the application of gas sensors; and vi) flash flood forecasting simulations based on Digital Elevation Models (DEMs) derived from LiDAR technology. This study provided crucial data for emergency services and established UAVs as fast, precise, and safe tools for monitoring flash flood events, with potential applications in other natural disasters. Moreover, the data collected by UAVs in the days following the event is also being used to assess environmental impacts and to identify potential public health risks stemming from this extreme weather event. • Several UAV flights monitored the Valencia flash flood (Cut-off Low 2024) in real-time. • 3D reconstruction of Poyo's gully is provided showing affected buildings in nearby towns. • Multispectral technology was used to assess sludge composition type and accumulation. • Deep learning detected vehicle buildup and water-carried debris in agricultural fields. • Biogeochemical analysis and flood forecasting were performed in worst-hit areas.