and salinity measurements from moorings are used to calculate the analogous source, which represents the salinity difference between the North Pacific and North Atlantic, as well as runoff from rivers into the Gulf of Anadyr and Norton sound.
Three preliminary model experiments have been conducted so far. The fresh-water has been run as a "passive" tracer, in which case it plays no role in the equation of state. In these experiments, forced by the 1990-1994 ECMWF reanalysis fields (and by 30-day surface relaxation to the monthly mean Levitus T and S fields), the fresh-water is a simple dye, which tags each of the river sources with a different "color". Two experiments were done with the runoff included in the density calculations. One of the active runs uses the same forcing as the passive tracer experiment, while the second run, still in progress, uses the ECMWF forcings for 1979-1994. So far, the surface relaxation to Levitus has not been weakened in the "active" experiments.
3. RESULTS
The initial model results are quite encouraging. The modeled tracer distributions show the strong topographic guidance for which the Arctic is so well known. Distribution of tracer with depth is in good agreement with the observations, with increases of vertical penetration over the ocean ridges. Deep vertical convection is visible in several of the locations where we believe that density plumes do, in fact, flow down the shelf-front into the deep basins. The primary difference between the active and passive tracer results is that the active tracer appears to stay on the shelf more than the passive tracer, and the former is also seen to advectively fill the surface layers of the Canadian Basin more uniformly, while the former exits the Chukchi Plateau and tends to advect more toward the Alpha Ridge with only minor meanders into the Canadian Basin. These differences seem to reflect a subtle difference in buoyancy of river plumes after entering the Arctic Ocean, with the active tracer results showing less mixing with saltier waters than the passive tracer (that is, the fresh water, i.e. T and S, prescribed in active experiments introduces additional stratification in the water column, thus decreasing vertical mixing).
In the current formulation, which emphasizes advection pathways over eddy diffusion, the model subducts river runoff primarily along the Laptev Sea margin and the coast of Greenland. Focusing on the Laptev Sea subduction, a number of interesting characteristics of the advective-dominated flow paths are apparent. In particular, waters from the Ob and Yenisey flow eastward along the coast-line, passing through the Vil'Kitskiy Strait (separating Severnaya Zemlya and the Taymyr Peninsula) to enter the Laptev Sea. This is in addition to a strongly diffusive influence of the Ob, Yenisey, Pechora and Severnaya Dvina rivers on the Barents Sea branch of the inflowing Atlantic Water (AW) within the Kara and Barents Seas. This is consistent with the long time scales often attributed to the freshening of the AW within this region. A "diffusive" fresh water source suggests that the return of modified AW component to intermediate depths along the continental shelf, where it joins with the Fram Strait branch, via the St. Anna and Voronin Troughs, is dominated by well mixed waters.
Conversely, within the Laptev Sea, the Ob/Yenisey waters converge with those from the Lena and Khatanga (the latter consisting of the Kotuy and Kheta Rivers entering the Khatanga Gulf). The slightly more dense Ob/Yenisey waters sink to intermediate depths along the shelf at the confluence of the Ob/Yenisey with the Lena/Khatanga waters, whereas the Len/Khatanga sink just to the east of the confluence. These two streams then follow the topography, but, instead of returning along the Lomonosov or Mendeleev Ridges, they continue eastward until they encounter the Chukchi Plateau, where they finally follow the topography northward along the shelf until they encounter the open ocean waters at the northern extreme of the Canadian Basin. At this point, the streams take on a meandering course that, in the passive runs, tends more toward the Makarov Basin and Alpha Ridge, with occasional sweeps across the Canadian Basin, whereas the active run tends more toward a complete coverage of the Canadian Basin. The fact that the waters do not turn northward until reaching the Chukchi Cap is consistent with the expansion of the AW across the Makarov Basin. However, the current results are for forcings from 1990-1994, and we can not be assured of the significance of this Makarov appearance unless the results from the 1979-1994 forcings show a northward turn at the Lomonosov Ridge prior to the 1990s.
Once the waters reach the Canadian shelf they continue their cyclonic path by moving either along and between the Queen Elizabeth and Victoria Islands (for the active runs), or along the Alpha Ridge, until reaching the Lincoln Sea whereupon they exit the Arctic through Fram Strait. A small component re-enters the cyclonic circulation of the Eurasian Basin, from the Morris Jesup Rise following the Nansen-Gakkel Ridge. This last feature suggests that there may actually be two cyclonic
