Community-oriented Wildfire Planning with Agentic AI: Automated Fire Spread Simulation and Interactive 3D Visualization

Abstract. With the increasing frequency and severity of Wildland–Urban Interface (WUI) wildfires, there is a growing need for community-oriented planning approaches that integrate foresight, preparedness, and participatory decision-making. This paper presents an intelligent and interactive framework that engages communities, urban planners, and emergency responders in exploring firere-silient strategies through agentic AI-driven wildfire simulation and interactive 3D visualization. The proposed system enables users to define scenario-based queries via a conversational interface, which are automatically translated into orchestrated 2D and 3D fire spread simulations. A virtual, interactive environment supports intuitive exploration of alternative wildfire scenarios and community-scale mitigation strategies. The framework aims to enhance public awareness and adaptive behavior while providing science-informed decision support for evaluating fire dynamics under varying terrain, infrastructure, and fuel conditions. Built upon a Large Language Model (LLM)-powered agentic architecture, the system automates simulation workflows and lowers technical barriers, enabling users with diverse backgrounds to participate in wildfire planning processes. Simulation outputs are delivered through web-based 3D visualization and immersive holographic interfaces, supporting iterative and interactive analysis of fire spread in urban environments. Through this approach, communities can assess risk, compare mitigation strategies, and improve preparedness for real-world wildfire events.