Recently we have investigated mercury dynamics during snowmelt at Sleepers River (Shanley and others, 2000). Mercury inputs to the watershed are primarily atmospheric, and most of this mercury (Hg) is retained in the watershed. However, Hg export in streamflow is highly episodic. Mercury export is dominated by the particulate fraction, as high concentrations of particulate Hg associated with POC coincide with high flows from rapid snowmelt or rain-on-snow events. During the 2000 snowmelt, we sampled 10 streams of various size and land cover. Hg mobility was strongly linked to organic carbon, with dissolved Hg strongly correlated with DOC and particulate Hg strongly correlated with POC. These relations held when data from all sites were pooled, for example, data were not sensitive to land cover or basin size.

Plots of dissolved mercury in relation to dissolved and particulate organic carbon during snowmelt in 2000.

In Sleepers River streamwater, mercury (Hg) is strongly associated with organic carbon.  In these plots from the 2000 snowmelt, data are pooled from 10 different streams representing a wide range of size and land cover, yet linear relations emerge.  Dissolved Hg is positively correlated with dissolved organic carbon (DOC) and particulate Hg strongly correlated with particulate organic carbon (POC).

Isotopic evidence in our headwater catchment supports the common finding that pre-event water dominates the event hydrograph (Shanley  and others, 1993).  A large reservoir of pre-event water is present  in the glacial till, which has low hydraulic conductivity (Thomas, 1992; K. Kendall, 1997; McGlynn, 1997), and correspondingly high water retention capacity (Shanley, 1995).  In the low-permeability till, water flows preferentially through strata of high conductivity (McGlynn, 1997).  In particular, transmissivity increases toward land surface, giving rise to flow generation via transmissivity feedback during the snowmelt period (K. Kendall and others, written communication 2001; McDonnell and others, 1998), when the water table approaches land surface even in upper hillslope positions.  The relation of water table level vs streamflow has a marked hysteresis (K. Kendall and others, written communication 2001).   In the riparian zone, the water table is higher at a given flow on the rising limb of snowmelt than at the same flow on the  falling limb.  On the hillslope, the hysteresis is reversed.  Thus  after peak snowmelt, the hydrometric and isotopic data suggest that drainage  from the hillslope dominates the stream hydrograph, despite contradictory  chemical data that suggest hillslope water makes only a minor contribution  to streamflow.  In particular, hillslope ground water has excess silica relative to streamwater, because of the abundance of labile silica in the soil (Shanley and others, 1995a).  This intriguing paradox is the object of continuing investigation.

Other findings from the forested headwater catchment are that (1) surface-saturated area appears not to exceed 5% of the basin, even at peak snowmelt (Titus and others, 1995); (2) Sr and Pb isotope ratios can successfully distinguish shallow soil water (atmospheric or silicate weathering signal) from deep ground water (calcite weathering signal) (Bullen and Kendall, in press); (3) analysis of the N and O isotopes of nitrate suggest that the source of nitrate during spring snowmelt is nitrification in the soil, despite the large amount of nitrate in the melting snowpack (Kendall and others, 1995a,b); (4) dissolved organic nitrogen (DON) comprises a significant portion of streamwater N flux (Campbell and others, written communication); (5) Sleepers River is an end member among northeastern U.S. watersheds for its high buffering capacity and Ca concentrations (Hornbeck and others, 1997); (6) therefore, potential for aluminum toxicity to tree roots (as measured by [exchangeable Al / exchangeable Ca] in soil) is low compared to other northeastern sites (Lawrence and others, 1995).  (7) soil CO2 efflux persists year-round, even when a snowpack is present (Winston and others, 1992; Sundquist and others, 1992); and (8) ice layers within the snowpack impede the downward movement of meltwater and solutes (Shanley and others, 1995).

Plot showing nitrate (NO3) and dissolved organic carbon (DOC) dynamics during the 1993 snowmelt at W-9.  Note that each snowmelt peak causes a peak in concentration, but that for nitrate the peaks concentrtions diminish as snowmelt progresses.

Aside from the headwater catchment, we use the gaged watersheds  of differing size and land use and the rich historic dataset from Sleepers  River.  In a set of nested basins, we analyzed oxygen-18 data for a series of annual snowmelt events and a summer storm.  In some years, there is a tendency for new water contributions to increase with increasing  basin size, which could be explained by the net effect of  the topography  in the larger basins that tends to produce more extensive areas of surface saturation  (Wolock, 1995).  However, the new water increases also corresponded to increasing percentage of open land, and in other years the scale effect was not present, suggesting that presence / absence of ground frost might be the cause of the patterns (Shanley and others, in press).  We analyzed 16 years of ground frost data and flow data, which indicate that ground frost does tend to enhance runoff. (Shanley and Chalmers, 1999).  Smith (1997) found that stream chemistry in the three nested basins and a small agricultural basin could be explained as a mixture of three end members representing groundwater,  soil water, and snowmelt (or rainfall).  The sum of the snowmelt and  soil water components agreed very closely with the "new water" component as determined by isotopic hydrograph separation.

Linked Behavior of Mercury and Organic Carbon Transport in an Upland Landscape During Snowmelt 2001 (Abstract from GSA Northeastern Section - 36th Annual Meeting 2001)

Back to Sleepers River Homepage