Fu, J. S. (2023, August 9). AI can help forecast air quality, but freak events like 2023’s summer of wildfire smoke require traditional methods too. The Conversation. https://theconversation.com/ai-can-help-forecast-air-quality-but-freak-events-like-2023s-summer-of-wildfire-smoke-require-traditional-methods-too-210170  

Scientists can predict air quality a few days ahead using two main methods – a chemical transport model or a machine-learning model. Chemical transport models can predict and simulate chemical reactions and air pollutants flow by using “data from emissions inventories reported by local agencies that list pollutants from known sources.” On the other hand, machine-learning models learn air quality patterns and, when comparing those patterns to the present, can predict future air quality. But with random events, like the wildfires in Canada, they are unable to predict because they had not learned those patterns yet.

Understanding more about what models are used to predict air quality (and how AI is struggling with it) is interesting because depending on the future of machine-learning models, we could predict air quality very well. Especially with the rise of AI and the increase of new data, I believe that machine-learning can work around “random events” and lead to a very accurate prediction. I wonder if, in the future, chemical transport models wouldn’t be needed anymore to predict the air quality. Additionally, a better understanding of air pollution prediction can be better for human health as people can better plan ahead to stay inside on days when the air quality is at its worst.