Adaptive, real-time hypoxia measurements using an autonomous boat

We present an autonomous system to measure hypoxia at high spatial resolutions. The approach combines a robotic boat, cloud hosted data services, and a suite of adaptive sampling algorithms to minimize the number of samples required to delineate hypoxic extents. The boat lowers sensors into the water column to provide depth profiles of temperature and oxygen concentrations. An adaptive path-planning algorithm continuously analyzes the in-situ observations and directs the boat to its next measurement location. This significantly reduces number of samples compared to a gridded sampling approach, while simultaneously improving the certainty with which the hypoxic regions are delineated. The method has been evaluated on small lakes throughout Michigan and shows significant promise to scale to the Great Lakes, where hypoxia is common occurrence that adversely affects various stakeholder and ecosystems.