EVALUATION OF BUILDING DATA UNDER NATURAL HAZARD IMPACTS USING MULTIPLE GEOSPATIAL PLATFORMS

The application of remote sensing tools is indispensable for building data evaluation, particularly during rapid assessment phases over extensive areas. Established codes like EMS 98 and FEMA offer guidance for the vulnerability and damage assessment of buildings through ground-based surveys. However, currently, there are no established standards for building data evaluation on geospatial platforms. This study aims to develop a criteria-based framework for damage assessment under both mono- and multi-natural hazard conditions, utilizing visual inspections of Very High Resolution (VHR) imagery. The framework is tailored for compatibility, aligning with EMS 98 scales for earthquake damage and adopting a similar format for flood and tsunami assessments. The validity and applicability of the framework were confirmed using Ground Truth Data (GTD) and Google Street View datasets from multiple target areas. A variety of tools were employed to achieve the study's objectives, including QGIS, ERDAS IMAGINE, MATLAB toolbox, and SNAP for image processing. VHR satellite images of selected target areas were acquired, specifically from Palu, Sulawesi Island, Indonesia (2018); L’Aquila, Italy (2009); and Antakya, Turkey (2023); through the Earthquake Damage Analysis Center (EDAC) at Bauhaus-Universität Weimar. Additional remote-sensed imagery was sourced from open-access online platforms such as NASA, Copernicus, and Geofly, mainly for assessing earthquake and flood/tsunami hazards. Obtaining ground survey datasets for the considered hazards and areas was a challenge, leading to the translation of local reports to understand the local building damage classifications. Notably, over 500 pages from the crisis management office in Indonesia were translated to extract relevant ground truth information. To ensure a comprehensive analysis, the study compared change detection results from automated image processing of satellite imagery with those from ground-based surveys and visual inspections of VHR images. The criteria-based visual inspection method not only demonstrated high accuracy but, in some instances, was more reliable than traditional ground survey datasets. This research not only fills a crucial gap in geospatial assessment standards but also underscores the essential role of remote sensing in performing precise, timely, and wide-ranging risk evaluations for various natural hazards. These findings pave the way for future research to integrate this framework with real-time monitoring systems, enhancing proactive disaster management in regions prone to such events.