Case Studies

From national-scale analysis to real-time forecasting, our work turns wildfire data into actionable insight.

Australia: Sub-Hourly Wildfire Forecasting Pilot

Partner: Covey Associates (QLD) | University of Canterbury | 2023–2025

In partnership with Covey and researchers in New Zealand and Australia, we deployed our real-time fire potential model to forecast ignition risk at sub-hourly resolution. The system ingests weather station data, updating every 10 minutes. Forecasts were validated across the 20 years over three regions in Australia summer season.

New Zealand: National Fire-Environment Typology

Partner: New Zealand Forest Research Institute Ltd (Scion) | 2025-

We worked with Scion to define New Zealand’s first national fire-environment typology using unsupervised machine learning. By clustering weather, fuel, and topographic data, we identified distinct fire-prone classes across the country. These classes will support suppression planning, land management, and national fire strategy.

New Zealand: Modeling Fire Growth and Shutdown Conditions

Partner: New Zealand Forest Research Institute Ltd (Scion) | 2025-

We trained machine learning models on New Zealand wildfire data to quantify the conditions that drive fire growth and those that lead to suppression. These models will help fire managers understand the roles of temperature, wind, humidity, and rainfall in fire behaviour, improving decisions on pre-positioning and resource deployment.