Kris Sankaran investigates how AI can help predict specific flood risks

Assistant Professor Kris Sankaran contributed to a chapter in the new Microsoft book AI for Good: Applications in Sustainability, Humanitarian Action, and Health, published in April 2024.

Assistant Professor Kris Sankaran has contributed his expertise to the new book, AI for Good: Applications in Sustainability, Humanitarian Action, and Health, published by Microsoft this month. In the book, an esteemed group of artificial intelligence researchers investigates how AI can help tackle major problems related to climate change, humanitarian aid, and healthcare.

One chapter, “Mapping Glacial Lakes”, on which Sankaran is a contributing author, discusses how computer vision algorithms can help accurately monitor glacial lake boundaries in satellite images. As the planet continues to warm, Sankaran said, “Monitoring these boundaries is important because growing lakes can increase the risk for glacial lake outburst floods.” And as glaciers keep melting, the risk for future flooding from these lakes will keep increasing, creating a need for more precise methods of monitoring and predicting future floods. That’s where Sankaran’s work in AI for Good comes in.

According to Sankaran, the work shows “the potential to use AI to turn raw data into more meaningful summaries.” For instance, he referenced turning a collection of satellite images into data showing trends in lake growth, using computer vision algorithms. Additionally, Sankaran said, there are many other issues—tracking methane leaks, for example—where “manually processing all the data to guide decision-making would be very difficult.” In these situations, AI can be a game-changer.

In the case of glacial lake flooding, Sankaran is using his statistics and data science expertise to help advance the monitoring and prediction of these disasters in the face of a changing climate.

“Compared to traditional work on the problem, which analyzes the images only at the times where predictions are needed, we show how to leverage historical [data]. This allows us to combine community monitoring efforts with algorithmic learning,” Sankaran explained. The chapter was co-authored by Sankaran, Anthony Ortiz of Microsoft AI for Good Research Lab, Finu Shresta  and Tenzing Choygal Sherpa of the International Centre for Integrated Mountain Development, and Mir Matin of the United Nations University Institute for Water, Environment and Health.

Sankaran’s work, and the AI for Good book overall, highlights the potential for AI to augment researchers’ capabilities across a range of fields and help address pressing challenges. The chapter also shows how faculty members in the Department of Statistics and the School of Computer, Data & Information Sciences are on the forefront of AI and machine learning research, applying technology in innovative ways for research that serves the public good.

For more information about the new book AI for Good: Applications in Sustainability, Humanitarian Action, and Health, visit its website.

To learn more about our faculty’s research areas, visit our Research Interests webpage.