Non-linear Optimization of Multi-Vehicle Ocean Sampling Networks for Cost-effective Ocean Prediction Systems

The problem of optimally deploying a suite of sensors to estimate the oceanographic environment is addressed. The best way to estimate (nowcast) and predict (forecast) the ocean environment is to assimilate measurements from dynamical and uncertain regions into a dynamic ocean model. A Genetic Algorithm (GA) approach to this problem is presented. The scalar cost function is defined as a weighted combination of a sensor suites sampling of the ocean variability, ocean dynamics, transmission loss sensitivity, model uncertainty (and others). An example with 3 Gliders, 2 REMUS powered vehicles, and 3 moorings is presented to illustrate the optimization approach in the complex Mid-Atlantic Bight region off the coast of New Jersey.