Stochastic Transport and Identification of Moving Passive Tracer Sources with Application to Marine Traffic in the Mediterranean Sea

Abstract

Source Reconstruction Problems are widely used for the aim of determining the sources of contaminants and pollutants in the case of deliberate or accidental release scenarios. This involves the inverse problem and inference of the sources parameters given a set of observed and measured data for direct emergency actions and services while estimating at the same time the nature of the threat in order to avoid and manage escalating consequences.

The aim of this thesis report is to introduce the methodology and results of the inference and source reconstruction problems with an application to contaminants and pollutants transport in the Mediterranean Sea in the presence of a stochastic velocity field. This will allow immediate and accurate actions in emergency cases such as any unexpected release scenario.

The forward Lagrangian model, adopted in transporting the pollutants, is first introduced followed by a detailed discussion of the full implementation of the stochastic transport of moving passive tracers in the Mediterranean Sea in the presence of a stochastic velocity field. This also involves building the probability maps for various release scenarios along a given ship path.

In addition, a new sampling based approach, similar to MCMC algorithms and based on the evaluation of the cost function between the modelled likelihood of the contributing events and the observation patch, is introduced. This algorithm allows the inference of single and multiple sources in the Mediterranean Sea with a quantitative measure of the relative contribution of these sources to the observation patch while quantifying the uncertainty in the solution. The results obtained from this proposed inference algorithm are illustrated and are in accordance with the true solution obtained using a deterministic optimization approach.