Mapping the Future

Zelalem Demissie, Department of Geology, Wichita State University, is one of four assistant professors who received First Awards from Kansas NSF EPSCoR in Fall 2023. These seed grants are designed to help early-career faculty initiate novel research related to the ARISE project. Each award of up to $50,000 supports a year of research and education efforts.

In this interview, Dr. Demissie answers questions about his radar technology and artificial intelligence (AI) techniques to forecast early-stage drought in Kansas.

Q: How has EPSCoR funding supported your research?

A: Early-career scientists or researchers can get support from Kansas NSF EPSCoR. I had submitted my research to NASA for a few years and decided to switch my focus to Kansas NSF EPSCoR. That turned out really good for me. I have been implementing my research on a detailed schedule for almost a year now.

Q: What is your research focused on?

A: We use a type of radar imaging that’s mounted on the European Space Agency’s Sentinel satellites. These satellites revisit specific areas every six to twelve days, capturing data. I use AI and other sophisticated software to process this radar data to detect early-stage drought here in Kansas.

Q: What breakthroughs have you achieved so far?

A: We’re bringing a new concept to the scientific community called the terrestrial index. In this era of climate change, hydrological and meteorological datasets might fluctuate and generate errors when you’re trying to forecast global climate. Here we’re bringing the earth itself into play.

Water supports the earth, like a structural column of a building. When you extract water, the earth’s surface sinks. When precipitation replenishes the aquifer, the surface rises. The terrestrial index is the surface fluctuation of the earth.

With satellite data collected since 2014, we can measure surface behavior with millimeter precision. By training AI models on this data, we can forecast early-stage drought in Kansas.

We’ve also developed an app that maps injection wells in real-time. This app helps determine if injection activities might be linked to local seismic events by monitoring how much material is injected into wells.

Q: How does your research incorporate AI and machine learning?

A: We use these tools to process and interpret complex radar data. We using long-term machine learning algorithms, like LSTM, used to predict things like the New York Stock Exchange. By training the AI on our time-series data, we can predict how the Earth’s surface might respond in the future, helping us forecast drought conditions before they become critical.

Q: Are you involving the community in your research efforts?

A: Yes. We’re involving a wide range of stakeholders, including high school students, undergrads, and graduate students, many from underserved or underrepresented backgrounds. We train them on coding, AI algorithms, and the scientific aspects of the project. Last summer, we held a camp with professors and even the provost to encourage students to pursue geoscience. It’s about building interest and skills in the next generation.

Q: What have you realized because of your research?

A: The situation is grave. We must use water very responsibly because of our excessive over-extraction of groundwater. If you don’t have money in your bank account, you can’t withdraw it, right? That’s what’s happening right now with the aquifers.

–Story by Melinda R. Cordell, Freelance Consultant