BEGIN:VCALENDAR VERSION:2.0 PRODID:-//pretalx//talks.osgeo.org//foss4g-2026//talk//8DMFAL BEGIN:VTIMEZONE TZID:JST BEGIN:STANDARD DTSTART:20000101T000000 RRULE:FREQ=YEARLY;BYMONTH=1 TZNAME:JST TZOFFSETFROM:+0900 TZOFFSETTO:+0900 END:STANDARD END:VTIMEZONE BEGIN:VEVENT UID:pretalx-foss4g-2026-8DMFAL@talks.osgeo.org DTSTART;TZID=JST:20260902T113000 DTEND;TZID=JST:20260902T120000 DESCRIPTION:Natural and technological disasters pose significant threats to communities\, infrastructure\, and the environment worldwide. Effective d isaster risk management requires robust analytical frameworks capable of s ystematically assessing hazard\, vulnerability\, and exposure components a cross multiple disaster types\, recognizing the complex and cascading natu re of modern risks (Pescaroli & Alexander\, 2018). This paper presents a c omprehensive disaster risk analysis model that integrates Geographic Infor mation Systems (GIS)\, machine learning algorithms\, and Artificial Intell igence (AI) driven interpretation tools (GeoAI) to support multi-hazard ri sk assessment\, urban resilience planning\, and citizen-oriented risk comm unication.\n\nThe proposed model is designed to be generalizable across a wide range of disaster types\, including floods\, wildfires\, earthquakes\ , landslides\, droughts\, and urban heat island events. Rather than develo ping isolated models for each hazard category\, the framework adopts a uni fied\, modular architecture in which the core analytical pipeline is consi stently applied regardless of the disaster type under investigation. This design philosophy ensures scalability and reproducibility\, addressing the critical need for understanding social vulnerability to environmental haz ards (Cutter\, Boruff & Shirley\, 2006). This framework integrates geospat ial analysis with supervised machine-learning classification models. Hazar d and vulnerability layers are constructed using multi-source spatial data sets\, including topographic\, land-use and land-cover maps\, hydrological networks\, soil-type and lithology data\, historical disaster records\, p opulation-density grids\, and socioeconomic indicators. These spatial laye rs are processed and analyzed in QGIS\, building on established approaches that employ geoprocessing to evaluate regional risks of major urban hazar ds and improve public safety (Zhao & Liu\, 2017\; Contini et al.\, 2000).\ n\nOnce the geospatial feature matrices are assembled\, machine learning c lassification algorithms are employed to derive hazard and vulnerability m aps. The study evaluates and compares several state-of-the-art ensemble le arning algorithms\, with particular emphasis on Random Forest (RF) and Ext reme Gradient Boosting (XGBoost). The integration of hybrid models and biv ariate statistics has been shown to significantly enhance the accuracy of hazard mapping in GIS-based comparative assessments (Ali et al.\, 2020). P ython-based machine learning libraries are used to implement\, train\, val idate\, and tune these models rigorously. Cross-validation strategies and hyperparameter optimization techniques are applied to ensure robust and ge neralizable model performance. Accuracy metrics such as the Area Under the Receiver Operating Characteristic Curve (AUC-ROC)\, F1-score\, precision\ , recall\, and Cohen's Kappa coefficient are used to evaluate model output s.\n\nThe exposure component of the risk model quantifies the number and n ature of elements at risk within identified hazard zones\, including popul ation counts\, building footprints\, and critical infrastructure. To accur ately reflect structural vulnerabilities\, the model framework allows for the integration of fragility curves\, a method widely used for generating damage probability estimates for structural typologies like reinforced con crete buildings under seismic loads (Akkar & Yakut\, 2005\; Ramamoorthy\, 2006). Spatial overlay operations between hazard maps and exposure layers yield exposure matrices that inform prioritization in disaster preparednes s.\n\nA particularly novel contribution of this work is the integration of AI agent architectures based on Large Language Models (LLMs) and the Mode l Context Protocol (MCP) into the risk assessment pipeline. The MCP serves as a standardized communication interface enabling LLM-based AI agents to dynamically query\, retrieve\, and reason over geospatial risk data store d in structured databases and spatial APIs. These AI agents are capable of performing automated\, context-aware interpretation of multi-hazard risk maps and statistical outputs\, generating region-specific risk narratives that translate complex quantitative results into clear\, actionable insigh ts. By leveraging the natural language generation capabilities of LLMs\, t he system can produce detailed risk assessment reports tailored to specifi c administrative units\, districts\, or neighborhoods\, significantly redu cing the analytical burden on planners and decision-makers. The MCP-based architecture also enables seamless integration with external data sources and third-party geospatial services\, enhancing the model's capacity for r eal-time or near-real-time risk monitoring.\n\nAll data inputs to the mode l are sourced from open and freely accessible repositories\, including Ope nStreetMap for infrastructure and land use data\, Copernicus Land Monitori ng Service and USGS Earth Explorer for remote sensing products\, NASA EART HDATA for climatological and hydrological datasets\, and national/regional open government portals for administrative and socioeconomic statistics. This open data philosophy not only ensures cost efficiency but also enable s transparency\, reproducibility\, and independent verification of results by the broader scientific community and policy stakeholders.\n\nThe outpu ts of the model are made accessible through a web-based geospatial platfor m that allows real-time visualization of hazard susceptibility maps\, vuln erability indices\, exposure layers\, and composite risk scores across mul tiple administrative levels. The platform is designed with usability in mi nd\, targeting both technical users such as urban planners\, civil protect ion agencies\, and researchers\, and non-technical users including local g overnment officials and ordinary citizens. Interactive map layers\, filter ing tools\, and downloadable reports are provided to support diverse user needs. A key feature of the platform is an integrated AI-powered chatbot\, developed using LLM technology and connected to the underlying risk datab ase via the MCP interface. Citizens and decision-makers can query the chat bot using natural language to obtain plain-language explanations of the ri sk status for their area\, receive guidance on risk-reducing behaviors\, a nd access information about emergency preparedness resources. This convers ational interface significantly lowers the barrier to understanding comple x risk information and fosters greater public awareness and engagement.\n\ nThe model was developed and piloted in the Marmara Region of Türkiye\, e ncompassing 11 provinces: Istanbul\, Bursa\, Kocaeli\, Sakarya\, Tekirdağ \, Edirne\, Kırklareli\, Balıkesir\, Çanakkale\, Yalova\, and Bilecik. The Marmara Region has outstanding strategic importance due to its high po pulation density\, industrial concentration\, seismic activity\, and expos ure to multiple natural hazards. This geographically and socioeconomically diverse study area provided an ideal testbed for evaluating the model acr oss varying physical and demographic conditions\, allowing for a comprehen sive evaluation of its cross-hazard applicability and cross-regional trans ferability. Pilot results demonstrated strong classification accuracy for hazard susceptibility mapping across all tested disaster types\, with XGBo ost consistently achieving higher AUC scores compared to baseline models. The LLM-based risk interpretation layer produced coherent\, factually grou nded\, and contextually relevant narratives that were positively evaluated by domain expert reviewers.\n\nIn conclusion\, the GeoAI-based disaster r isk analysis model presented in this paper offers a scalable\, open\, and interoperable framework that bridges advanced spatial analytics\, machine learning\, and generative AI to support evidence-based disaster risk reduc tion. By making risk information accessible through web platforms and conv ersational AI tools\, the model contributes directly to the development of sustainable\, smart\, and resilient cities aligned with the Sendai Framew ork for Disaster Risk Reduction 2015–2030 and the United Nations Sustain able Development Goals (UNDRR\, 2015). Future research directions include the incorporation of real-time sensor data and Internet of Things (IoT) te chnologies to support global initiatives for dynamic multi-hazard early wa rning systems (UNDRR\, 2024). DTSTAMP:20260717T234909Z LOCATION:Cosmos2 SUMMARY:A Comprehensive Disaster Risk Analysis Model with GeoAI - Muhammed Oguzhan Mete URL:https://talks.osgeo.org/foss4g-2026/talk/8DMFAL/ END:VEVENT END:VCALENDAR