The ratio T/TL estimates the statistical degrees of feedom for a record length T and a Lagrangian auto-correlation scale TL. Following Pauline and Niiler (1989), I define TL as the integral to the first zero-crossing of the normalized versions of the lagged auto-correlation function which gives TL 〜 3 days. A single drifter thus gives a statistically independent velocity estimate about once every 3 days. The degrees of freedom of the record mean flows shown in Figure 5 thus vary between 7 and 40; each Eulerian ensemble contains data fom at least 6 different drifters. The time scale TL and velocity variance give a spatial decorrelation (or eddy) scale of about 40 km. This scale exceeds the internal deformation radius by a factor of 3 inshore and a factor of 1.5 offshore near the ice edge. The observed eddy scale thus more closely corresponds to processes influenced by the ice edge than those influence by the estuarine discharge of the Kolyma River.
Assuming the surface flow field to be both homogeneous and stationary, I can estimate a single particle diffusivity from the lagged auto-correlation functions (not shown). Figure 6 shows these eddy diffusivities as a function of lag time τ for the along-shore (Kxx) and across-shore (Kyy) direction. The initial dispersion varies linearily with lag for τ<< TL while it varies little after 7 days when it approaches the "random walk" regime (Taylor, 1921). If the assumptions are satisfied, then one can infer zonal and meridional eddy diffusivities of about 3.5×107 and 2.5×107 cm2/s, respectively.
These values are rather large and resemle mid-latitude open ocean values more closely than they represent values often found in the coastal ocean.
6. CONCLUSIONS
Arctic shelf circulation differs in many respect from mid-latitude circulation even though lateral buoyancy and wind forcing as well as gently sloping shallow topography is present in both. The strong vertical stratification during the short summer month of huge estuarine discharge and the absence of tidal mixing may explain some of the differences. Nevertheless, strong upwelling occurs and this process can break down the vertical stratification on the inner shelf. If this happens the salt content of the surface mixed layer increases and provides greater potential for brine production during the freezing process. Melling and Moore (1995) argue along these lines in order to explain interannual variability in the brine production of the Mackenzie shelf, however, the dynamics of inner shelves and especially those of the across-shelf exchanges are poorly understood on shelves anywhere.
As a community we need direct velocity observations as well as process oriented numerical model studies to properly assess and predict the role of Arctic shelf areas on the basin scale thermohaline circulation.
Acknowledgements-I thank Jacqueline Grebmeier of the University of Tennessee and Tom Weingartner of the University of Alaska for the deployment of the Siberian drifters. The study was funded by the Office of Naval Research through the Arctic Nuclear Waste Assessment Program (N00014-95-1-1157) and the High Latitude Program (N00014-95-0915).
(akm 2/98)
