Uncertainty quantification in Lagrangian clustering analysis

Partitioning ocean flows into regions that minimally mix with their surroundings can identify materially coherent vortices and assist in search and rescue planning by reducing the search domain. One method for such partitioning is the Lagrangian clustering analysis, which identifies sets of trajectories that move as a compact set. This method has been applied to deterministic, chaotic systems, revealing underlying transport barriers. For ocean models, however, in addition to the complex dynamics, there are several sources of uncertainty, such as model initialization and parameters, limited knowledge of oceanographic processes, and ocean boundary conditions and forcing. Therefore, the Lagrangian clustering analysis, when applied to ocean forecasts, should incorporate uncertain parameters and the resulting coherent structures should be robust to model uncertainty. Through application to a geostrophic flow, we present an investigation of the sensitivity of the spectral clustering method to uncertain parameters and an approach for applying this method to an ensemble of simulations.