Descripción del proyecto
Floods are major natural disasters with severe impacts in urban areas, where increasingly high population density and infrastructure together with worsening climate change trends exacerbate the risk. Reliable and timely monitoring is critical for resilience in terms of pre-disaster preparedness, emergency response, and post-disaster relief. Remote Sensing accurately evaluates flood extent, but data can be expensive and inaccessible. Open RS represents a not fully exploited potential limited by the relatively coarse resolution. Moreover, recent advances in computer vision techniques based on deep learning (DL) and novel observational opportunities can provide valuable information on both flood extent and depth in combination with RS.
STURM aims to advance urban flood knowledge by combining open RS Sentinel imagery and crowdsourcing (semantic and visual data) using DL with the ambition of overcoming the constraints of spatial resolution and limited information. STURM leverages free and newly available opportunistic observing systems providing a globally consistent, open-source-based, smart method for improved multi-source observations of hydroclimatic hazardous events in urban areas. The research objectives are to assess and accurately map urban flood extent and depth with enhanced spatial resolution (sub-pixel mapping and measurements from street-level images) and validate the methodology against real disaster events. STURM’s novel data fusion paradigm suits the demand to fill data and knowledge gaps at the urban scale while offering a benchmark solution for hydrological model validation. The DL-based pipeline combines the strengths of globally available data sources while reducing human and economic resource consumption. Global applicability, low cost, and immediate usability are methodological pillars of the STURM project that determine its high impact potential to enhance urban flood resilience and face the global hydrological challenges of the 2030s and beyond.