Development of stream water chemistry during spring melt in a northern hardwood forest

The role of snowmelt and subsurface hydrology in determining the chemistry of a small headwater stream in the Turkey Lakes Watershed (TLW) was evaluated for the spring melt periods 1992 to 1996. Spring runoff is the dominant hydrological event at the TLW each year. Processes occurring within the snowpack during snowmelt were principally responsible for the above-ground changes in chemical fluxes relative to bulk deposition (the effect of winter throughfall was minimal). Large changes in chemical fluxes occurred below ground. Organic matter decomposition, weathering, nitrification, and element cycling are some of the more important below-ground processes that operate during the snow accumulation and ablation season and control the composition of the water ultimately appearing in the stream. Maximum stream discharge was accompanied by elevated concentrations of H super(+), NO sub(3) super(-), K super(+), NH sub(4) super(+), DOC, Al and Mn, but reduced levels of Ca super(2+), Mg super(2+), SO sub(4) super(2-) and SiO sub(2). The concentration-discharge relationships were consistent with water movement through and above the forest floor during peak discharge, a flowpath facilitated by rapid infiltration of meltwater and the existence of a relatively impermeable layer in the mineral soil creating a perched water table. Averaged over the five periods of snow accumulation and ablation, it was estimated that pre-melt stream flow, and water routed through the forest floor and through the upper mineral soil contributed 9, 28 and 63%, respectively, of the discharge measured at the outlet of the catchment. The forest floor contribution would be greater at peak discharge and at higher elevations. An end-member mixing model estimated concentrations of SO sub(4) super(2-), NO sub(3) super(-), Cl super(-), Ca super(2+), Mg super(2+), Na super(+) and Al that were comparable to average values measured in the stream. Other variables (NH sub(4) super(+), H super(+), K super(+) and DOC) were over-estimated implying retention mechanisms operating outside the model assumptions