Alberta uses AI and intuition to fight wildfires

The Province started to utilize a tool from AltaML to predict the area of new wildfires

Canada’s 2023 wildfires were the worst on record.

According to the Canadian Interagency Forest Fire Centre, the fires burned 17 million hectares across the country.

Ed Trenchard, a wildfire manager in Alberta, notes first responders enlist traditional methods, such as the Canadian Wildland Fire Information System, when it comes to fire planning. The system uses several environmental factors, including weather and forest conditions, to rate the danger of fires.

Other important methods are intuition and local knowledge of the area.

But Alberta Wildfire, the Province’s forest firefighting agency, switched things up in 2022 to utilize an AI-powered tool to help responding officers strategically prepare for ensuing wildfires and resource use.

Using machine learning, the tool analyzes thousands of data points to forecast, based on region, how likely new fires will be the next day. Given that it’s trained through historical fire data, it can also make predictions through regional weather and forest conditions.

It further incorporates specifics on global carbon emissions and the day of the week to see if there are any correlations with wildfires, climate change and human behaviour.

AltaML, an Edmonton, Alberta, organization focusing on AI solutions, built the tool with Microsoft’s Azure Machine Learning. AltaML can now forecast a new wildfire 80 percent of the time.

“[The tool] typically supports what experienced duty officers are thinking and enables more inexperienced duty officers to gain that knowledge without taking years to get there,” Trenchard said.

The model can help Alberta Wildfire save between $2 million and $5 million in yearly operating costs.

However, AltaML needs to do more to continue to improve the tool. Trenchard said he would like to see the tool be more precise with when and where a fire will start and how it’ll spread over ten days.

GovLab.ai, an innovation hub under AltaML, worked on the project. The Government of Alberta and Mitacs, a non-profit organization that uses research to address societal problems through AI, also contributed.

Image credit: Government of Alberta 

Source: Microsoft