Earth Observation supports landslide risk assessment in Medellín, Colombia

Abstract

Landslides are the most frequent natural disaster in Colombia, coupled with cities growing unplanned and informal towards high-risk areas due to land scarcity. Earth Observation (EO) has a powerful capability for supporting risk reduction and adaptation. The ongoing German-Colombian research project ‘Inform@Risk’ is developing a low-cost and site-specific Early Warning and Evacuation Systems (EWES) in a low-income and self-built neighborhood (https://www.bmbf-client.de/en/projects/informrisk). This multi-scale project illustrates how the combination of EO with other data sources improves knowledge on landslide risks of many urban dwellers. We use very high resolution, multi-temporal and multi-sensoral remote sensing data (e.g., from drones, Landsat, Sentinel and TanDEM-X data) combined with ancillary data (e.g., census population and hazard maps) to produce up-to-date and better estimates of areas, people an assets at risk. At the city level, we found exposed areas doubled in the last 25 years. Formal and informal developments were located using the morphologic proxy of built-up structures, combined with downscaled population revealed that 1 out of 10 people is simultaneously exposed to landslides hazard and living in informal settlements. At the neighborhood level, we produced unprecedentedly detailed geo-information. We identified an average of 50 houses being built per year in a high risk-prone area of 40 hectares. Estimates showed nearly 4,600 residents in the EWES location, of which 40% are at very high risk. This project exemplifies of how environmental, economic and social risks proxied from space and combined with other data sources provides a comprehensive picture for decision-making.