In this report we reanalyse the Great Lakes Drogue Studies data taken in 1964. Original drogue position data are edited into smoothed position data for this purpose. The edited data are then processed using linear regression procedures to calculate not only Lagrangian deformations but also velocity gradient parameters, i.e. divergence, vorticity, deformation rates. The linear regression method also enables evaluation of turbulent characteristics, in particular momentary eddy diffusivities. The results show that i) the drogue area is controlled primarily by the cumulative effect of horizontal divergence, ii) a simple vorticity balance is established, iii) turbulence characteristics are generally consistent with the previous estimates by Okubo and Farlow in 1967, iv) momentary eddy diffusivities are relatively small and average 2 x 10 2 cm2/sec for all experiments. Values of the Lagrangian deformation and momentary eddy diffusivities are fed into the analytical solution of time-dependent advection-diffusion equation and the result shows favorable comparison with observed gross-scale drogue dispersion. Inspection of patterns within gross-scale drogue groups revealed that each group was composed of an average of five clusters. Small-scale variability was examined by computing the velocity gradient parameters for each cluster. Drogues tend to cluster over local convergences while the clusters themselves tend to diverge; that is, the gross-scale expands, small-scale tends to contract. A statistical test shows that 10-15 drogues are required to achieve a standard deviation equal to the 95% confidence limit of the gross-scale drogue group. This provides a guideline for designing future dispersion experiments.

Technical Report

Great Lakes (North America)