Headwater Streams in Porous Landscapes – What’s the contributing area?

Abstract:

Building on a long legacy of hydrogeological investigations at the Savannah River Site in the Sandhills of the Upper Coastal Plain in South Carolina, we began in 2005 a headwater-scale investigation of hillslope flow pathways, streamflow sources, and water quality responses to intensive woody biomass production. The landscape is characterized by blackwater streams flowing slowly through wide flat stream valleys, deep unconsolidated layers of sands and clays, a regional clay layer beneath Fourmile Creek that defines the lower boundary of the surficial aquifer, rolling topography with steeper slopes on the valley margins and gentle slopes elsewhere, and a sandy clay loam argillic layer within 0.2 to 1.5m from the surface. Most water leaves headwater basins by groundwater flow, appearing as streamflow far downstream. Only at scales larger than 50 km² does average streamflow match expectations from water balances. This raises the question, what constitutes the contributing area for headwater streams in porous landscapes? Perching and interflow generation over the argillic horizon is common, but leakage through clay is rapid relative to interflow travel times, so interflow serves to shift the point of percolation downslope from the point of infiltration. Only interflow from the valley-adjacent slopes can contribute to stormflow responses. Our interflow interception trenches and maximum rise piezometer networks reveal high heterogeneity in subsurface flow paths at multiple spatial scales. Streamwater has isotopic and chemical characteristics similar to deep groundwater, but we cannot easily determine the source area for groundwater reaching the first order streams. Our observations suggest that one’s view of hillslope and catchment flow processes depends on the scale, number, and frequency of observations of state variables and outputs. In some cases, less frequent or less numerous observations of fewer tracers would have yielded different inferences. The data also suggest that each hillslope encompasses an ensemble of thresholds and flow paths that vary with moisture content over space and time.