Optimal estimation of Eulerian velocity field given Lagrangian observations

The problem of estimating an Eulerian velocity field given particle trajectories is formulated as an optimal filtering problem. Under the idealistic assumption that the Eulerian velocity field is delta-correlated in time (Kraichnan model) the exact solution of the non-linear filtering problem is found. In a more realistic Markov model with finite correlation time an approximate solution is suggested and examined by Monte Carlo means.