Skip to main content

About the project

Predictive Forest Fire Risk & Spread Simulation Platform

Wildfires can escalate rapidly, threatening forests, wildlife, communities, and critical infrastructure. Our project aims to help users move beyond simply detecting fires toward understanding where fire risk may be highest and how a fire could spread over time.

The platform combines satellite imagery, historical fire records, weather conditions, vegetation indices, and terrain characteristics to generate 24-hour forest fire risk predictions across a geographic grid of approximately 500 m × 500 m. Using an XGBoost-based machine learning model, it estimates fire risk for individual grid cells and visualizes risk levels through an interactive GIS dashboard.

Beyond risk prediction, a Cellular Automata–based simulation engine models potential fire spread over the next 12 hours, incorporating factors such as wind, slope, aspect, and land-cover characteristics. Users can explore geographic risk patterns, inspect available active-fire observations, and visualize simulated fire boundaries over a timeline.

Built around geospatial data processing, machine learning, spatial databases, and web technologies, the project is designed as an extensible open-source platform. Its initial focus is regional wildfire risk analysis in Himachal Pradesh, with potential for expansion to other fire-prone regions.

Our goal is to transform environmental data into accessible, map-based intelligence that can support wildfire awareness, preparedness, and planning. The platform is a research and decision-support prototype; its predictions and simulated boundaries require further independent validation before operational emergency use.

Built with

pythontensorflowpytorchnext.jspostgresql

AI tools

Coded withAntigravityGemini 3.8 Flash

Prize Categories Entered

  • DigitalOcean

    Best Open-Source AI Project

    Submitted
  • Event prize

    Hacktoberfest Hack Day Winners

    Submitted