Towards Improved Prediction of Ocean Processes Using Statistical Machine Learning

We discuss the problem of predicting ocean currents based on historical data and ocean models. This problem is relevant to navigation of autonomous underwater vehicles (AUVs) and has significant scientific importance for analysis of biological processes and weather patterns. Available predictive models provide accurate prediction of these currents, but they typically do not provide confidence estimates of these predictions. We propose prediction methods based on Gaussian Process (GP) regression, which provide improved estimates of ocean currents and also have the potential to provide confidence bounds on these estimates. We demonstrate experimental validation on ocean current estimates from the Southern California Bight provided by the Regional Ocean Modeling System (ROMS) dataset, and we integrate these predictions into probabilistic planners to improve the safety of operation of a simulated AUV. While our techniques achieve modest performance improvements thus far (the work reported here is at an early stage), they show potential for significantly improving the prediction of ocean processes for environmental monitoring applications.