This article, by Rodrigo Cuellar, is part of a series highlighting members of the Office of Sustainability’s Experts Database. In a collaboration with instructor Hannah Monroe’s course, LSC 561: Writing Science for the Public, students interviewed campus sustainability experts and produced short feature stories.
AI, or artificial intelligence, is an increasingly common subject, but many do not know that it could play a role in the fight against climate change, allowing farmers to more efficiently grow crops, enabling first responders to have faster responses to natural disasters, and contributing to a more sustainable future.
This is what Qunying Huang, a researcher in the University of Wisconsin–Madison’s Department of Geography, hopes to enable by using AI technology. Huang sees linked possibilities: using AI to predict, and also to reduce, the effects of climate change.
Weather predictions, which are based on decades (sometimes centuries) of historical data, are incredibly important resources for farmers and first responders. However, as weather records are seemingly set every day, these resources are becoming outdated due to climate change.
“Natural hazards are becoming more frequent due to climate change, and we need to respond smartly,” Huang said.
Without accurate predictions, we cannot respond appropriately to different situations. For instance, farmers might find that crops are not adequately irrigated, or a first responder might encounter more severe flooding than expected based on historical data. As it is no longer possible to rely on predictions based on historical data, Huang aims to use new methods such as machine learning to tackle this gap.
Machine learning, a specific application of AI, uses computer programs or AI models to spot patterns in data and make predictions. For example, if you feed a model thousands of images of cats and dogs, you could train the model to identify patterns in the images that can be used to determine if a picture that it has not seen before is a cat or a dog. Now, instead of training a model to differentiate cats from dogs, someone can train a model to predict the depth of a body of water in a flooded area. While this might seem like an inconsequential piece of knowledge, it can be invaluable to a first responder assessing the impacts of a natural disaster and deciding how to respond.
“I started looking at flooding events,” Huang said. “How can we quickly map the actual flooding, and how can we do real-time mapping of the impacted area so we can do a real-time response and recovery effort?”
Leveraging machine learning could allow first responders to act more effectively when previous predictions may be less reliable. As for farmers, using the new technology could allow them to monitor their fields and quickly flag any potential problems by analyzing satellite images through AI models to determine changes. This leads to more efficient use of resources, such as water, while maximizing crop yield.
“The stakeholder of my research would be the community that is impacted by natural hazards, climate change,” Huang said. Others would include “first responders, the decision makers, and the policymakers. And, on another level, the local farmers.”
While AI technology can seem abstract, Huang’s research focuses on providing information and using novel ways to analyze information. She hopes that her work allows decision makers to generate more informed, effective responses to natural disasters and irregular weather patterns. She also hopes to contribute to a more sustainable future by using these tools to reduce the effects of climate change, to “utilize all those technologies — special data science, big data, AI and machine learning — to do research for the social good, making sure that we are able to develop a sustainable earth system,” and “achieve social equality and environmental justice.