Fresh Perspectives: An Oceanographer’s View of Machine Learning in Great Lakes Science
Date:
Why oceanographers should care about the Great Lakes
Abstract:
The Great Lakes - Ontario, Erie, Michigan-Huron, and Superior - are the largest group of freshwater lakes in the world by total area, holding approximately 21% of Earth’s surface freshwater. These interconnected waters are so large that if they were drained at the volume transport rate of the Gulf Stream, it would take over a week to empty them. In addition to their considerable size, the Great Lakes serve as a critical resource for over 34 million people, providing drinking water, supporting commercial and recreational fishing, facilitating transportation and commerce, and offering opportunities for tourism, recreation, and connecting with nature. This dynamic system, managed cooperatively by the United States and Canada, represents a case study for international water resource management and offers a unique natural laboratory for Earth system science research.
Subject to many of the same atmospheric forcings as coastal oceans, the Great Lakes could arguably be considered inland freshwater seas. On the largest scales, the Rossby number of Great Lakes circulation is comparable to that found in basin-scale ocean gyres, highlighting the importance of planetary-rotational effects on their flow. The lakes feature wind-driven gyres, coastal upwelling, mesoscale phenomena, and distinct nearshore and open water dynamical regimes. They exhibit tilted seasonal thermoclines with associated geostrophic flows, as well as prominent seasonal cycles in lake ice and convective mixing. However, there are some differences as well - the baroclinic Rossby radius of deformation is about an order of magnitude smaller than that of the midlatitude ocean, largely due to weaker stratification from the lack of salinity gradients. Additionally, there is no appreciable tidal mixing, and there are no open boundary conditions to consider.
The Great Lakes have served, and will continue to serve, as a test bed for understanding aquatic, terrestrial, and climatological systems around the world. They span a wide variety of physical, ecological, and regulatory conditions, from a deep, sea-like, relatively unregulated lake in a colder climate (i.e., Lake Superior) to a shallow, rapidly changing, heavily regulated lake with maritime influences (i.e., Lake Ontario). Their extensive coastlines (relevant to freshwater and marine coastal systems globally), relatively high density of data per square kilometer in some regions (such as western Lake Erie), closed boundaries, and regulatory influences offer conditions suitable for integrating machine learning (ML) and artificial intelligence (AI) into marine science and management. This includes areas such as data-driven modeling, decision-relevant forecasting, hypothesis generation and testing, and observing network design, as well as physics-based and interpretable/explainable ML/AI approaches.
This presentation will map out the current state and possible future trajectories of Great Lakes science by exploring initial efforts and research opportunities at the intersection of ML/AI and marine science research in the region. By showcasing current Great Lakes applications and future possibilities, this talk argues that these freshwater systems represent a growth area for marine research and technological innovation. Collaborative efforts between freshwater and marine science communities promise mutual advancements, bridging critical gaps, and leveraging strengths from both fields. In addition, the integration of ML/AI in binational management strategies could influence future decision support tools, promoting sustainable and cooperative management of these vital resources and potentially informing wider international management of shared resources. Finally, I highlight the establishment of the Great Lakes AI Lab, which aims to be a nucleus of a collaborative research community in ML/AI for Great Lakes science, management, and restoration.
Slides: Available here
Related Presentations:
- NOAA Pacific Marine Environmental Laboratory, 2024-09-26, Seattle, Washington