Data Science Meets Climate Research in New Summer Program
Duke students use data science methods to help address environmental challenges during the first round of Climate+ projects.
By Amanda Solliday
Even for the data-savvy, interpreting the vast amounts of information collected for climate research questions can be daunting.
To help address this need, one of Duke’s most well-known summer education programs tried something new: pairing students interested in applied data science with interdisciplinary projects focused on climate change.
Climate+ is a fresh direction for the existing Data+ program, a full-time, ten-week summer research experience for Duke students of all class years and majors. The Rhodes Information Initiative at Duke and the merged Nicholas Institute for Environmental Policy Solutions and Duke University Energy Initiative sponsored the first interdisciplinary Climate+ project teams.
“Our project offered me the chance to better understand how data science operates in the real world,” said Austin Brown a rising sophomore majoring in computer science and part of the first Climate+ cohort. “It's more complicated than the clean data sets given to us in our classes.”
Brown was part of a project team building a machine learning model to predict what carbon emissions from wetlands may look like in the future and which factors best predict those emissions levels. The existing data was gathered by the AmeriFlux network from 70 sites across the United States.
Yujia Shen, a rising sophomore majoring in computer science and statistics, joined the same project because she wanted to work at the intersection of data and climate.
“When I looked at Climate+, I applied because I saw a bunch of projects that lined up with these two subjects that I’m interested in,” Shen said.
For each team, two to four undergraduates partner with one graduate student, who serves as a project manager, on climate-related issues ranging from electricity consumption, carbon emissions, real-time tracking of climate impacts and changes in river and ocean ecosystems.
“I’m always looking for places where I can apply my learned skills in data science to real-life problems, and climate is one such problem,” said Biniam Garomsa, a rising senior majoring in data science. Over the summer, he conducted research on winter ice coverage of small river systems.
The project analyzed images of ice collected by seven different cameras on the Hubbard Brook watershed in New Hampshire. With those images, the team began to build a model that would demonstrate how snow and ice coverage has changed over the years with a goal of predicting coverage in future years. Right now, it’s in a proof-of-concept phase, but the team hopes to keep developing the model and apply it to even more river systems.
Climate+ will be offered again next summer, with plans to create additional opportunities for students to learn more about the science behind climate change as well as mitigation and adaptation strategies being considered around the world.
“We’ve now established a thematic area for the program explicitly focused on a key societal challenge of the decades to come,” said Kyle Bradbury, director of the Energy Initiative’s Energy Data Analytics Lab and a leader of the Climate+ program. “This enables students to see how data science tools can contribute to developing greater understanding of and prospective solutions for climate change and related societal challenges.”
This article originally appeared in Duke Today on Aug. 3, 2022.