The North American Great Lakes are vital for the health, safety, and economy of the Great Lakes Region, but monitoring them effectively with limited resources is a challenge. We're using a machine learning tool called DeepSensor to help determine the best locations for new monitoring stations. Instead of putting sensors everywhere, which is too expensive, we can use DeepSensor to predict temperatures in unmeasured areas based on existing data. With DeepSensor, we can test different strategies for placing sensors. For example, we might decide to put new sensors in areas where we're most uncertain about our predictions, or in places where new data would give us the biggest improvement in our overall understanding.

We're also considering real-world factors like how difficult it is to maintain a sensor at a particular location or how often a spot is covered in ice. By combining this advanced technology with practical considerations, we aim to create a more effective and resource-conscious way to monitor the Great Lakes. This approach can help coastal planners, international governing bodies, and communities make better decisions about managing this crucial resource and could be used for other similar environmental systems too.