A technique for optimizing the placement of oceanographic sensors with example case studies for the New York Harbor region

The work of this thesis presents a new technique for optimal placement of oceanographic sensors in a small scale, highly variable salinity field. The developed method addresses two fundamental objectives: (i) the accurate and efficient estimation of a synoptic salinity field from a limited number of irregularly spaced measurements, and (ii) the design of an optimization strategy to optimally place a limited number of sensors to produce an accurate estimate of the salinity field.;Objective analysis techniques were analyzed to determine an efficient and accurate method to estimate a complex salinity field in the New York Harbor Estuary from a limited number of sensor measurements. To design an efficient optimization strategy, the performance of various nonlinear optimization techniques to select an optimal set of sensor locations was evaluated. Optimization was carried out by minimizing the cost function defined as the RMS error between estimated salinity fields and ground truth salinity fields derived from a hydrodynamic model. A final optimization method was designed based on the results from initial simulation experiments. Accurate estimates of salinity fields at 1 and 12 hour periods at different depths were constructed, and a heuristic approach for extended time sensor deployments was defined. The final optimization method was selected by comparing it with a series of ground truth simulations. It was subsequently tested by trials in a real-ocean environment (non- simulated), in which a small number of sensor locations were used to develop accurate salinity maps for a complex region of the lower Hudson River. The capability of the optimization method was extended by development of a mobile version to optimally place a set of sensors for accurate estimates of a data field. The mobile version was developed for quick implementation in an unknown environment, with no hydrodynamic model data. A second field trial in a low variability environment in Key West, Florida was performed to test the mobile optimization technique. Results, including tests of accuracy, show that the new technique can select optimal sensor locations for a limited number of sensors and produce highly accurate salinity fields in different environments.