Bayesian Inference and Markov Chain Monte Carlo Sampling for Lagrangian Particle Tracking in the Ocean

Estimating Lagrangian particle tracking (LPT) model parameters is challenging due to the irreversible nature of processes driving the particles. In the ocean, several methods have been proposed to tackle this problem, but each with its own limitations. In this work, we present a generic method that employs Bayesian inference and Markov chain Monte Carlo (MCMC) sampling to estimate the probability distributions of parameters of interest. The forward model used within the MCMC machinery enables the use of existing application-specific modules built on top of the LPT model. Furthermore, various types of information can be exploited depending on available data, making it useful for different applications of LPT in the ocean. The method has been tested using a simple LPT advection-diusion model in a double-gyre synthetic flow field. Different data types were tested, ranging from the simple location of particles, to measurements of concentration and corresponding contours. Inference of the location of the source, time and duration of release is finaly presented as a probability distribution that quantifies uncertainties and correlations between the parameters.