Skip to content

State of predicting wildfires in Canada 

Author: Weseem Ahmed 

We know the facts: climate change is getting worse and causing billions in damage every year through adverse flood, fire, and storm disasters. Let’s consider wildfires: outside of historical fire data – which can be difficult to find – we know very little about the actual risk of these events happening. 

To be fair, there have been efforts to estimate the probability of fires breaking out, but this work is nowhere near to the scale needed for Canada-wide coverage, and the techniques used are difficult to implement on this broader scale. Academic papers over the years have looked at specific regions of Canada to estimate how likely it is that wildfires ignite and spread. And while the papers are very advanced, they altogether lack a standardized methodology and geographic scale. 

But this is understandable once you consider how much needs to go into accurately determining wildfire risk. Some of the maps and datasets that are needed in wildfire prediction include topographic maps called digital elevation models, maps of fuel type that describe the propensity of different vegetation types, forest health and age to burn as well as comprehensive weather station data. On top of all this data the model needs to integrate fire behavior patterns. Then all these datasets and mathematical information are combined into Monte Carlo simulations, with hundreds or thousands of iterations. And even with modern advances in computer technology, data quality and acquisition remain a challenge. 

COLDSTREAM, CANADA – Jul 10, 2021: A view of the Clerke Road wildfire along Hwy 97 in Coldstream BC, Canada

A model is only as good as its inputs. Weather stations provide most of the data in these models but it’s easy to put too much faith in them. They may have faulty sensors, or need upgrades, or maintenance and calibration. Even if they are running perfectly with zero margin of error, their distribution across Canada favours large population centres and discounts more remote areas.  

climate85 eliminates geographic bias by using  data from climate models, evenly gridded and consistent for all of Canada. 

Canada needs to know where wildfires are more or less likely to occur, it is a critical first step towards more resilient communities, infrastructure and businesses.  

climate85 is developing Canada’s first purpose-built platform that quantifies climate hazards, including fire risk. We use state-of-the-art methodologies and advances in computer processing to address the hurdles that have historically hindered wildfire studies. Users will be able to search for an address and instantly know what the odds of a fire breaking out over the next 12 months or over the duration of their mortgages. This information will be a game changer for real estate investors, mortgage brokers, and insurance agents.  


Climate Data Analyst

Weseem combines his data analytics skills with knowledge of climate science to help establish Minerva as a leader in climate risk technologies.