This work addresses the problem of continuously reducing ocean forecast uncertainty using underwater gliders through the application of a Monte Carlo-based method, and the development of the mechanisms needed to apply that method in the context of ocean forecast models. The solution to the problem is developed in several stages, gradually incorporating the features necessary to apply the solution in the real world. The problem is initially examined in an abstract model of uncertainty, in a single dimension. A method, named the Approximately Optimal Next Action (AONA) method, is developed, analysed, and then evaluated in several scenarios designed to emulate important aspects of the uncertainty structures found in the real ocean. The method is then applied to the Lorenz '96 (L96) model, a simple chaotic system that is often used as a crude representation of physical processes. The token model of uncertainty, whilst useful for understanding the principal challenges in uncertainty reduction, does not describe how to directly quantify uncertainties within other models, and so information entropy is identified as a suitable framework for quantifying these uncertainties, in common with some existing work. The AONA method is reassessed within the L96 model, and compared to some existing approaches, against which it performs favourably. Finally, the mechanisms needed to apply the method to a real ocean model, the UK Met Offi�ce FOAM-NEMO MED12 model, are developed. A model ensemble is described, and an analysis of the temporal and spatial distribution of entropy within that ensemble is provided. A model for glider motion is developed that allows the generation of random glider paths, required for the AONA method, that account for the effects of ocean currents. Additionally, a kernel-based method is implemented to provided a mapping between the discrete grid of the ocean model and the continuous real world in which gliders operate.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:658772 |
| Date | January 2014 |
| Creators | Hughes, Chris D. |
| Contributors | Smeed, D. A. |
| Publisher | University of Southampton |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | https://eprints.soton.ac.uk/378968/ |
Page generated in 0.0015 seconds