A Data-Driven Deployment Approach for Persistent Monitoring in Aquatic Environments

Abstract

Processes of scientific interest in the aquatic environment occur across multiple spatio-temporal time scales. To properly assess and understand these processes, we must observe aquatic ecosystems over long time periods. This requires examination of the problem of deploying multiple, inexpensive, and minimally-actuated drifting vehicles. We aim to utilize these persistent assets to explore all locations on the water surface, and examine the entirety an underwater environment through the visibility of downward-facing cameras. In this work, we propose a data-driven approach for the deployment of drifters that creates a stochastic model, finds the generalized flow pattern of the water, and studies the long-term behavior of an aquatic environment from a flow point-of-view. Given the long-term behavior of the environment, our approach finds attractors and their transient groups as the domains of attractions. We then determine a minimum number of deployment locations for the drifters using these attractors and their transient groups. Our simulation results based on actual ocean model prediction data demonstrate the applicability of our approach.