The authors report on the development of an efficient and accurate tool for
computing Lagrangian trajectories using time-varying velocities. By linear
ly interpolating the velocities both in space and time one can obtain analy
tical expressions for the trajectory inside each model grid box. In combina
tion with a numerical determination of the transit times through each grid
box, these expressions allow for the implementation of expedient algorithms
for offline analyses of large datasets resulting from general circulation
models. The authors demonstrate the efficacy of this approach on a time-var
ying two-dimensional model gyre.