To better understand fire emissions under worst-case climate scenarios, scientists at Pacific Northwest National Laboratory first taught a machine to predict the past. That artificial intelligence is helping scientists better understand wildfire emissions in the Northwest.
Scientists have fed climate change factors into explainable artificial intelligence models. The model was then trained to learn the pattern.
When AI models can predict what happened in previous fire seasons, scientists know they can be trusted to predict the future, said lab colleague Ruby Leung.
“Advancements in machine learning now allow us to focus on many factors, and they are complex and non-linear relationships,” said Leung.
With AI models, scientists have discovered that climate change will increase temperatures, reduce soil moisture and change land use. These key factors are likely to contribute to more unhealthy fire emissions in the future, she said.
Moreover, machine learning models have shown that unhealthy fire emissions can increase dramatically. The study found wildfire emissions in the West could increase by up to 186% over the next few decades.
Generally, people think of machine learning as a black box, said Sally Wang, a data scientist at PNNL in Richland, Wash. Technology can make good predictions, but scientists don’t know what’s going on. Explainable artificial intelligence is not, she said. Scientists train machine learning models to make human-like decisions.
“Explainable AI allows us to understand: ‘Oh, (when) the reason machine learning projects predict future fire emissions will increase is because they see temperatures rising,’ Wang said as an example.
AI can help scientists better understand the areas they need to focus on to reduce wildfire emissions, Wang said.
An increase in wildfire emissions means a lot, Leung said.
One big implication could be in the realm of public health, Leung said. Particulate matter from wildfire emissions can cause a variety of health problems, especially for people with lung and heart conditions.
“There are many sectors in society that can benefit from this type of information,” she said.
That’s why combining information on the cascading impacts of climate change, wildfires and smoke is important, she said.
In the future, Wang said he would like to use AI models to gain a deeper understanding of how human factors contribute to fire activity. [Copyright 2023 Northwest News Network]