A Method for Assimilation of Lagrangian Data

Abstract

Difficulties in the assimilation of Lagrangian data arise because the state of the prognostic model is generally described in terms of Eulerian variables computed on a fixed grid in space, as a result there is no direct connection between the model variables and Lagrangian observations that carry time-integrated information. A method is presented for assimilating Lagrangian tracer positions, observed at discrete times, directly into the model. The idea is to augment the model with tracer advection equations and to track the correlations between the flow and the tracers via the extended Kalman filter. The augmented model state vector includes tracer coordinates and is updated through the correlations to the observed tracers. The technique is tested for point vortex flows: an NF point vortex system with a Gaussian noise term is modeled by its deterministic counterpart. Positions of ND tracer particles are observed at regular time intervals and assimilated into the model. Numerical experiments demonstrate successful system tracking for (NF, ND) = (2, 1), (4, 2), provided the observations are reasonably frequent and accurate and the system noise level is not too high. The performance of the filter strongly depends on initial tracer positions (drifter launch locations). Analysis of this dependence shows that the good launch locations are separated from the bad ones by Lagrangian flow structures (separatrices or invariant manifolds of the velocity field). The method is compared to an alternative indirect approach, where the flow velocity, estimated from two (or more) consecutive drifter observations, is assimilated directly into the model.