Abstract
This paper discusses some of the uncertainties that influence kinematic trajectory calculations. The interpolation errors due to different interpolation schemes are examined by degrading high-resolution wind fields from a numerical weather prediction model with respect to space and time. Under typical circumstances, the greatest errors are due to temporal interpolation, followed by horizontal and vertical interpolation. Relative errors in the vertical wind are higher than those in the horizontal wind components. These errors are quite substantial and severely affect the accuracy of trajectories. For instance, a decrease of the temporal resolution from 3 to 6 h leads to average relative interpolation errors of 16% in the horizontal wind components and 40% in the vertical wind component. These errors cause mean transport deviations of 280 km for two-dimensional model-level trajectories and 600 km for three-dimensional trajectories after 96-h travel time. The substantial deviations for three-dimensional trajectories are due to the large interpolation errors of the vertical velocity component. Although the three-dimensional trajectories are more sensitive to interpolation errors, for sufficiently (though not ideally) resolved wind fields they seem to be superior to model-level trajectories. An intercomparison of three-dimensional, model-level, isentropic, and boundary layer trajectories is presented.
