Venezuela: Earthquake Building Debris Assessment
Published by UNEP/OCHA Joint Environment Unit
About this dataset
This initial quantification of the June 2026 earthquake-induced debris across the most-affected Venezuelan states was derived through a multi-sensor synthetic aperture radar (SAR) change-detection analysis, integrated with rapid building damage assessments and supplemented by visual verification against very high-resolution optical imagery where available. Building footprints, including a modeled height attribute, were sourced from the Global Building Atlas [1] as the baseline layer, with locality-specific damage assessments and an independent SAR-derived damage product incorporated for areas where they were available, as detailed below. Damage to buildings was assessed through multiple sources and methods. A pixel-wise temporal t-test (PWTT) [2], adapted to a z-score formulation to accommodate the limited number of post-event Sentinel-1 acquisitions available at the time of analysis, was applied to Sentinel-1 Ground Range Detected (GRD) imagery, with buildings scoring above 0.4 retained for further assessment. This was complemented by damage assessment products from Copernicus EMS activation EMSR884 [3], as available on 27 June 2026, and by visual verification using PlanetScope imagery released through Planet's Crisis Response Program [4]. Building damage assessments produced by Microsoft's AI for Good Lab [5], generated by applying a classification model to post-event optical imagery acquired on 25 June 2026, were incorporated in their entirety for Catia La Mar, La Guaira, and Caraballeda. For locations with neither a locality-specific damage assessment nor high-resolution optical coverage, a Sentinel-1 SAR-based damage proxy map distributed via NASA's Disasters Program [6] was incorporated. This assessment estimates 96,046 buildings to have been damaged or destroyed., corresponding to nearly 3% of all buildings in the assessed states. For modeling purposes, minimum building height and average story height were considered to be 3m. Each built sq. meter is considered to have generated 1 tonne of debris. For visualization and modelling purposes, results were aggregated into a 7.8 sq. km hexagon grid. A substantial proportion of the debris remains within structurally compromised buildings and will only become accessible following demolition. [1] Zhu, X.X., Chen, S., Zhang, F., Shi, Y., & Wang, Y. (2025). GlobalBuildingAtlas: An Open Global and Complete Dataset of Building Polygons, Heights and LoD1 3D Models [Data set]. Technical University of Munich. [2] Ballinger, O. (2025). Open access battle damage detection via Pixel-Wise T-Test on Sentinel-1 imagery. Remote Sensing of Environment, 331, 115025. [3]Copernicus Emergency Management Service (© 2026 European Union), [EMSR884] Earthquake in Venezuela: Damage Assessment Products. Accessed 27 June 2026. [4] Planet Labs PBC. (2026). Planet Crisis Response: Venezuela Earthquake, June 2026 [Data set]. Source Cooperative. [5] Microsoft AI for Good Lab. (2026). Venezuela Earthquakes: Building Damage Assessment in Catia La Mar, La Guaira, and Caraballeda [Data sets]. Humanitarian Data Exchange (HDX). [6] Scher, C., & Van Den Hoek, J. (2026). Damage analysis of Copernicus Sentinel-1 data, Oregon State University [Data set]. NASA Disasters Program, Earthdata GIS.