A hydrometric and geochemical approach to test the transmissivity feedbackhypothesis during snowmelt

To test the transmissivity feedback hypothesis of runoff generation, surfac e and subsurface waters were monitored and sampled during the 1996 snowmelt at various topographic positions in a 41 ha forested headwater catchment a t Sleepers River, Vermont. Two conditions that promote transmissivity feedb ack existed in the catchment during the melt period. First, saturated hydra ulic conductivity increased toward land surface, from a geometric mean of 3 .6 mm h(-1) in glacial till to 25.6 mm h(-1) in deep soil to 54.0 mm h(-1) in shallow soil. Second, groundwater levels rose to within 0.3 m of land su rface at all riparian sites and most hillslope sites at peak melt. The impo rtance of transmissivity feedback to streamflow generation was tested at th e catchment scale by examination of physical and chemical patterns of groun dwater in near-stream (discharge) and hillslope (recharge/lateral flow) zon es, and within a geomorphic hollow (convergent flow). The presence of transmissivity feedback was supported by the abrupt increas e in streamflow as the water table rose into the surficial, transmissive zo ne; a flattening of the groundwater level vs, streamflow curve occurred at most sites. This relation had a clockwise hysteresis (higher groundwater le vel for given discharge on rising limb than at same discharge on falling li mb) at riparian sites, suggesting that the riparian zone was the dominant s ource area during the rising limb of the melt hydrograph. Hysteresis was co unterclockwise at hillslope sites, suggesting that hillslope drainage contr olled the snowmelt recession. End member mixing analysis using Ca, Mg, Na, dissolved organic carbon (DOC) , and Si showed that stream chemistry could be explained as a two-component mixture of groundwater high in base cations and an O-horizon/overland how water high in DOG. The dominance of shallow How paths during events was ind icated by the high positive correlation of DOC with streamflow (r(2) = 0.82 ). Despite the occurrence of transmissivity feedback. hillslope till and so il water were ruled out as end members primarily because their distinctive high-Si composition hail little or no effect on streamwater composition, Ti ll water from the geomorphic hollow had a chemistry very close to streamwat er base flow, and may represent the base now end member better than the mor e concentrated riparian groundwater. During snowmelt, streamwater compositi on shifted as this base flow was diluted - not by shallow groundwater from the hillslope, but rather bq a more surficial O-horizon/overland flow water , (C) 1999 Elsevier Science B.V. All rights reserved.