and water flow. The question is to what extent enhanced delivery of Ba associated with higher suspended sediment loads is traded off against dilution by higher volumes of run-off.If the former predominates, then our estimates for Ba input to the Arctic by rivers would be too low.
Bering Strait waters (50-120 nM range, 70 nM average) are enriched in Ba by the Yukon and other coastal rivers as well as upwelling in the Gulf of Anadyr (Falkner et al. , 1994). The fate of the Bering input is discussed below.
3. UPPER HALOCLINE
In summer, as Bering input waters transit the Chukchi Sea, they tend to be stripped of their Ba contents at the surface in association with biological activities. In winter, Bering input waters tend to be more saline and enter the Arctic interior below the surface layer. The Bering Ba input appears most significant to the upper halocline layer whose core occurs at around salinity 33.1 and is associated with a dissolved siliceous acid maximum (Guay and Falkner, 1998a). The Si maximum had been noted to extend to the vicinity of the Lomonosov Ridge before 1989 however, in 1993-1995, the upper halocline as defined by Ba extended only to the Mendeleev Ridge. This is consistent with changes of the relative positions of Pacific and Atlantic water mass assemblies noted by others (McLaughiin et al. , 1996).
Exactly how the upper halocline acquires its characteristics has yet to be directly observed. It is unclear to what extent regeneration from bottom sediments and brine enrichment contribute to the upper halocline Si and Ba maxima. There is a range of evidence to suggest that the Harold and Barrow Canyons may focus flow into this and other depth ranges and so these are important regions of future year-round observations.
4. SURFACE DISTRIBUTIONS
Within the Arctic, the most elevated Ba levels occur persistently in surface waters of the Beaufort gyre region (80-90 nM) (Guay and Falkner, 1998a). The most likely source of this enrichment is the Mackenzie River. Tracer distributions and mass balance considerations suggest that waters circulate in the Beaufort Gyre for on the order of a decade. Ten years of Mackenzie Ba input to the region would explain approximately 70% of the Ba inventory. The remainder is likely due to a combination of upward mixing and occasional direct input of Bering water.
Barium distributions do not support the circulation of Mackenzie influenced waters over the Chukchi shelf that has been indicated by high resolution model river &dyeΔexperiments (Maslowski et al., 1998). Our observations suggest that the influence of the Bering input on circulation in this region is inadequately captured in the models.
A transitional front between Mackenzie influenced BeaufoRt Gyre waters and a different surface regime in the vicinity of the Arlis Plateau/Mendeleev Ridge is indicated in our interpolated data from 1993 and 1994. We have speculated that the predominant pathway of Siberian river waters in 1993-94 was along-shelf toward the East Siberian Sea with entrance into the interior in the vicinity of the Arlis Plateau (Guay and Falkner, 1998a). We have evidence to suggest that in 1995 and 1996, circulation shifted so that Eurasian river influenced waters tended closer to their shelves region of origin with cross shelf transport occurring along the Lomonosov Ridge. Barium inventories in the Laptev Sea more than doubled from 1993 to 1995. River water inventories as indicated by oxygen isotopes displayed a similar change over the same time period (Frank, 1996).
This change we have noted is similar to a shift in ice advection patterns and the average position of the trans polar ice drift stream as monitored by the Arctic Buoy data program (S. Pfirman, personal communication). Our 1993-94 observations correspond to the buoy observations 1989-93 and our 1995-96 observations are more like 1983-87 ice patterns. The observed changes in the transpolar drift stream and ice advection have been reproduced in high resolution coupled ocean-ice-atmosphere modeling and appear to be driven by large scale wind forcing (Zhang and Hunke, 1997). A separate high resolution modeling exercise focused on river dye experiments over the 1979-1993 time frame shows result similar to the ice findings. It remains to be seen whether observed winds will produce the regime shift in the buoy and model data that we have noted in the Ba observations for 1995-96.
This circulation regime represented by the 1989-94 observations and modeling may be connected to a salinification of upper waters in
