Facing the Flames: How We Model Wildfire Risk
By Jorge Veiras & Valeria Beljaeva
It’s official, 2024 is the hottest year ever recorded (NOAA, 2025), a strong reminder of the accelerating impact of climate change. Forest restoration projects, like Afforestation, Reforestation, and Revegetation (ARR) initiatives, play a vital role in fighting climate change. They absorb carbon, restore ecosystems, and generate carbon credits. But they’re not immune to risks, wildfires, for example, can devastate years of progress in a matter of hours.
That’s where Artio comes in. We provide insurance from the earliest stages against carbon credit under-delivery designed specifically for ARR projects, protecting them against climate risks like wildfires. A key part of what we do is our proprietary risk model, which analyses threats such as fire to help project developers and investors stay one step ahead.
Our Approach To Modelling Fire Risk
Using historical satellite data from MODIS, we analyse where fires have occurred in the past, how they spread, and what factors influenced their behaviour. We often see that areas with a history of dry seasons and strong winds are especially vulnerable. By studying these patterns, we can gain insights into which regions are most at risk and why. This knowledge forms the foundation of our fire risk models.
Wildfires don’t just happen, they need the right mix of conditions to ignite and spread. That’s where the Fire Weather Index (FWI) comes in. Think of it as a kind of wildfire forecast that evaluates four key ingredients; temperature, humidity, wind and rainfall.
The FWI combines all these factors where a higher number means a higher risk of fire. We simulate future FWI values using climate data and explore different scenarios based on global emissions. For example, in a low-emission future, there might be fewer days with dangerous fire conditions. In a high-emission world, the risk increases significantly.
Once we know the conditions that make fires more likely, we use the CLIMADA framework (Siguan, 2023) to simulate how wildfires could behave. Our model breaks the process into three stages:
Ignition which predicts where and how fires start
Propagation which predicts how fires spread
Terminate which predicts when and where fires die out or get extinguished
This allows us to create maps showing potential fire paths and risks for both the present and the future.
Once we have a picture of where and when fires could occur, we calculate fire severity (damage). We use climate conditions, satellite images and tree biometric data (diameter, age, height etc) when calculating the loss.
VCS987: A Case Study
Figure 1. Overview of the project area with its boundaries highlighted.
To show how our risk model works, we assessed a live project: the "Reforestation of Degraded Forest Reserves in Ghana" (Verra ID: VCS987). This project aimed to restore 15,000 hectares of degraded forest. However, between 2008 and 2012, wildfires swept through areas around the Asubima Forest Reserve, damaging the project area and releasing stored carbon. This resulted in 42.5 ha being burnt contributing to a material under-delivery in the first issuance.
We analysed the 2011 fire and compared its behaviour to our predictive model. Our results showed a close match: while the model slightly overestimated the fire’s size, it captured how the fire spread and which areas were most vulnerable.
Figure 2. Comparison of the 2011 fire, MODIS vs Artio’s model prediction. The blue outline represents the project area.
While all fires can’t be prevented, mitigation measures can reduce their impact. Alongside insurance, Artio can offer tailored guidance on which preventative measures are most appropriate for the project. For example:
Firebreaks can stop fires from spreading between forest blocks.
Community training ensures local residents know how to detect and fight fires early.
Patrols during dry seasons help catch small fires before they grow.
Firefighting tools like pumps and protective gear make rapid response possible.
Following the fires, the project in Ghana implemented the steps outlined above to safeguard against future wildfires. Between the periods of 2019-2024 the project has seen significantly fewer fires in the project area as shown below.
Figure 3. Historical Burned Area (2019–2024) from 1 km MODIS Data.
Fires Under Future Climate Scenarios
2025 has just begun, but it's already shaping up to be a year with significant climate impacts. Using our models, we’ve projected fire risks for 2025 of the same project under four different climate scenarios. The results reveal a clear pattern: warmer and drier conditions lead to more intense fires.
Figure 4. Artio’s prediction of fire intensity for 2025 under four different climate scenarios (SSP1-26, SSP2-45, SSP3-70, and SSP5-85). Warmer colours indicate hotter, more intense fires, which noticeably become more intense as climate change progresses.
We can observe in Figure 4 that the southeastern part of the project area looks relatively safe, making it a good candidate for new plantings. Meanwhile, the northern and western zones show higher fire risks and will need close monitoring and additional fire prevention measures.
At Artio, we don’t just insure ARR projects, we try to give them a high chance to succeed. By combining innovative risk modelling with practical advice and insurance, we help buyers and investors navigate the challenges of climate change.
If you’d like to learn more about our work supporting ARR projects please book a call here.
1. 2024 was the world’s warmest year on record (2025) National Oceanic and Atmospheric Administration. Available at: https://www.noaa.gov/news/2024-was-worlds-warmest-year-on-record (Accessed: 23 January 2025).
Siguan, G.A. et al. (2023) Climada-project/climada_python: V4.0.1, Zenodo. Available at: https://zenodo.org/records/8383171 (Accessed: 23 January 2025).
Cover Photo by Gerd Altmann from Pixabay
