In November, CU Boulder’s Environmental Data Science Innovation & Inclusion Lab (ESIIL) hosted its first-ever virtual hackathon, Environmental MosAIc. The three-day event brought together 45 participants from around the world to work in teams, using environmental data science and AI to create solutions for environmental problems, like an app to track water quality by neighborhood.

“One of the objectives of ESIIL is to build a community of environmental data scientists,” said Virginia Iglesias, ESIIL hackathon coordinator and a CIRES scientist in Earth Lab. “We thought that it would be interesting to bring together the AI and the environmental data science communities and foster a collaborative space where the environmental scientists, with a deep understanding of data and information needs, would pose the research questions, and the AI community would help leverage analytical techniques, leading to informed, innovative approaches to pressing environmental issues.”

Iglesias said the idea came out of last May’s ESIIL Innovation Summit. Environmental scientists and environmental data scientists expressed a desire to use AI, but very few people knew how.

Participants were given access to CyVerse, an NSF-funded cyberinfrastructure also used by data scientists at ESIIL. The platform was new to many, but before the hackathon, all participants were trained on how to use it efficiently. 

For three days, participants worked into the night exploring environmental data and AI solutions. And they did it fast. 

Day 1: participants were divided into seven teams, each including one ESIIL mentor, and got to work analyzing data presented in a data cube designed by the ESIIL data analytics team. 

Day 2: Teams developed a question from the data and framed it as an AI problem.

Day 3: Each team presented their final product in a seven-minute pitch session to a panel of expert judges. 

• Team A: Biodiversity and ecosystem function: Predicting resistance to wildfire from spectral diversity
• Team B: How do natural hazards affect measures of community, cultural, and biological diversity? 
• Team C: The impact of the droughts in the Amazon watershed: An AI approach
• Team D: Community water: Critical water data, monitoring, prediction, and alerts for everyone
• Team E: Using AI to predict the effects of climate on fire and post-fire forest regeneration in Alaska 
• Team F: Predicting the impacts of fire on different measures of biodiversity

Judges represented a diverse group from the public and private sectors, environmental scientists, and experts in AI. “It was amazing to see the creativity and innovative proposals,” said Christine Wiedinmyer, associate director of science for CIRES and one of the judges. “I was so impressed by what the teams accomplished in such a short time.”  

The winning group, Team A, received funding for a week-long in-person meeting in Boulder with the ESIIL team. They’ll receive cyberinfrastructure and technical support to make progress on their project. 

ESIIL has more hackathons and events planned for 2024, including working with a French AI company to predict floods. 

The lab’s goal is to run one hackathon annually moving forward, and Iglesias is already looking forward to next year’s hackathon. “It was impressive to see how participants who had not met before and had diverse skills and backgrounds could work as inclusive teams that thought on their feet, prioritized tasks, and delivered creative presentations in just two and a half days.”