The University of Utah’s latest endeavor takes its research to the rugged terrains of Alaska, where critical river systems flow largely unchecked. Funded by a hefty ~$700K grant from the National Water Prediction Service, this project isn’t just another academic exercise. The team, led by Professor Ryan Johnson, aims to develop new tools for predicting streamflow and water temperature, essential for various aspects of Alaskan life. The grant will back a hands-on research effort, working directly with communities and agencies in the often-neglected Southcentral and Southeastern regions of Alaska.
Bounding over long-standing challenges associated with Alaska’s vast and inaccessible landscapes, the Johnson team turns to innovative solutions. With traditional data collection methods impractically slow for such an immense and remote area, the researchers are leveraging machine learning (ML) and artificial intelligence (AI) to fill the gaps. By using transfer learning, they hope to adapt knowledge from data-rich environments to Alaska’s unique needs, as Ryan Johnson pointed out to At The U, “It’s a way to expand predictive capability without waiting decades to collect new data.”
The importance of streamflow and temperature might sound niche, but these factors underpin key elements of Alaska’s ecosystems and economy. Streamflow data improve flood risk prediction, vital for emergency responses and wildlife management. Meanwhile, stream temperature is a beacon of ecosystem health, particularly for salmon, a species woven deeply into the cultural and economic tapestry of Alaskan life. Minor deviations in their aquatic environment can ripple through the food chain with dramatic consequences. Early and accurate predictions can, therefore, become powerful tools in maintaining the delicate equilibrium of these river systems…