NSF2026: EAGER: An Ecologically Inspired Human-Machine Intelligence Approach to Recognizing Similitude in Multi-Scale Watershed Research

点击翻译按钮获取中文摘要

With support from the Division of Earth Sciences (GEO/EAR) and the NSF 2026 Fund Program in the Office of Integrated Activities, Professors Hamshaw, Lee, Pespeni, Rizzo and Underwood at the University of Vermont & State Agricultural College are awarded this EAGER grant. This project addresses the NSF 2026 Idea Machine topic of Universal Similitude Across Scales, focusing on the broad area of Earth Sciences. Freshwater resources face growing pressures from extreme events and land use changes, which result in varying trends in water quantity and quality at multiple scales (from small creeks to large rivers and from storm events to decadal cycles). To build a greater understanding of the impacts to water quality, this interdisciplinary research project will investigate trends in short-term and long-term water quality and streamflow data and possible similarities and associations to watershed attributes (e.g. land use, topography). The project relies on water quality data archived at research stations across the continental U.S. available through databases of streamflow and water quality. To analyze these large quantities of water quality data, a combination of artificial intelligence and traditional scientific methods will be used. The societal importance of this project includes applying advances in machine learning and artificial intelligence to help researchers and environmental managers analyze and leverage limited monitoring resources collected at different sites and to develop best practices for monitoring environmental change. The project will fund a workshop that brings together graduate students from engineering and hydrological sciences with those from biology and microbiology to share data, analytical methods, and pursue interdisciplinary approaches that translate across disciplinary fields.Identifying similarities in watershed characteristics and water quality conditions across the continental U.S. can help identify the causes of environmental change and enables the translation of research from individual studies to larger regions. Because the sheer size and diversity of these long-term monitoring data present challenges for traditional scientific and statistical methods, we will employ artificial intelligence methods along with domain experts in hydrology and ecology in an integrated human-machine learning framework. In so doing, we aim to identify the common environmental variables and parameters that are linked to similarity across scales and investigate trends in short-term and long-term water quality and streamflow data. Organizational frameworks from ecological and biological disciplines will be applied to create a new approach to organizing and summarizing similarities in watershed signals, where watersheds are grouped into guilds based on their similar functional traits. By identifying patterns of similitude in these large data sets and extracting watershed attributes with linkages to these patterns, findings of research on ecosystem processes conducted in individual watershed studies can be more readily translated to larger regions.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.