The role of groundwater flow in streamflow generation within two small forested watersheds of the Canadian Shield

Abstract

The role of subsurface flow in streamflow generation during storms was investigated in two forested watersheds of the Canadian Shield in south-central Ontario. Storms were intensively monitored in the Harp 4-21 catchment (3.7 ha), where coarse-textured glacial till is up to 15 m thick, and in Harp 3A (21.7 ha), where the till is less than 0.5 m thick. Measured data included stream discharge in up to five subcatchments per watershed, soil moisture content by time domain reflectometry, groundwater levels, and groundwater and stream chemistry.

Subsurface flow is the dominant component of storm runoff. More than 75% of stream discharge in Harp 4-21 and 85% in Harp 3A was pre-event water (soil and till water). Three-component hydrograph separations based on two tracers (18O and dissolved silica) in Harp 4-21 show that groundwater flow through glacial till contributed 29 and 62% of total runoff for two storms. Vertical hydraulic gradients show that increased flow from the tills to the soils cannot account for the large and rapid increase of the till water component in the stream. Pre-event till water that has been discharged to the soils prior to storm onset is probably flushed from the soils into the stream during storm events.

Increased subsurface flow is caused by increased water content and hydraulic conductivity, rather than by increased hydraulic gradient. Since increases in hydraulic gradient were small (<0.04), groundwater ridging was insufficient to produce the observed increases in subsurface flow. Instead, the rising water table saturated permeable soil horizons, thereby increasing hydraulic conductivities and downslope subsurface flow. Flow within the transiently saturated horizons can account for observed subsurface stormflow if hydraulic conductivities are high. In Harp 4-21 and Harp 3A, sufficiently high saturated hydraulic conductivities of approximately 10^-3 m/s were estimated from soil water balances and infiltration experiments. However, macropores must be present to account for the high measured hydraulic conductivities of these soils.

Most of the subsurface storm runoff in the two catchments was generated within the soils because soil structure and macropores led to higher hydraulic conductivities in the soil than in the underlying glacial till or bedrock. Furthermore, subsurface flow and the efficiency of runoff production from a hillslope were greatest when groundwater levels rose into shallow soil horizons. Results were most consistent with the transmissivity feedback model with dual porosity flow in which the large increase in subsurface flow is attributed to saturation of macropores in shallow soil horizons by a rising water table. This conceptual model also provides an explanation for the displacement of pre-event water through macropores to the stream.

The thickness of glacial till influences the area from which runoff is generated and, consequently, the magnitude of storm runoff. In Harp 4-21, where till is thick, upslope areas store infiltrating precipitation and generate little runoff. Groundwater flow through the till maintains high groundwater levels and water contents in soils adjacent to the stream throughout the year. Wet areas near the Harp 4-21 stream respond rapidly to storms and contribute consistently to storm runoff along shallow flowpaths even after dry weather conditions. In Harp 3A, where till is thin, upslope areas cannot sustain groundwater flow during dry periods, areas adjacent to the stream dry out, and summer storms produce minimal runoff. However, when antecedent conditions in Harp 3A are wet, the water table develops in middle and upper hillslope soils and subsurface runoff is generated rapidly. As a result, storm runoff is more variable (effective runoff ratios of 0.0-0.67) in Harp 3A than in Harp 4-21 (0.07-0.38).

Dissolved organic carbon (DOC) concentrations and budgets demonstrate the importance of storms to the DOC export from catchments. DOC concentrations in the stream increase during storms by as much as 100 and 410% in Harp 3A and Harp 4-21 respectively. Storms were responsible for the export of between 57 to 68% of the total DOC in autumn and between 29 to 40% in spring. Riparian groundwater levels and flowpaths influence DOC concentrations and sources to the stream during storms. In Harp 4-21, riparian areas contributed between 73 and 84% of the stream DOC export during an autumn storm because groundwater flowed through shallow organic horizons as demonstrated by near-stream piezometers. In Harp 3A, riparian areas of hillslopes contributed less than 50% of the stream DOC because riparian flowpaths were predominantly through the lower B horizon.

This thesis contributes to the methodology, data collection, interpretation, and conceptual understanding of subsurface flow in streamflow generation studies. General equations for the three-component hydrograph separations were developed and applied to this study. The magnitude and timing of runoff production from hillslope sites were quantified by a soil water balance method. This method relates runoff generation to changes in water storage instead of water fluxes. The concept of a variable contributing area, which defines the area that contributes to surface and subsurface runoff generation, was introduced. Existing conceptual models of streamflow generation were re-interpreted and classified according to the mechanism (increase in hydraulic conductivity, increase in hydraulic gradient), the spatial extent and the zone in which increased subsurface flow occurs (saturated, transiently-saturated or unsaturated).

The results of this research have several implications for the study of subsurface flow during storms. This study has demonstrated the importance of relating physical and hydraulic properties such as slope, sediment texture, till thickness, hydraulic conductivity and characteristic curves to subsurface flow generation. Since increases in water content and hydraulic conductivity are ore important than increases in hydraulic gradient for producing subsurface flow on the catchment scale, a greater focus on measurements of the unsaturated properties of undisturbed sediments in future studies is warranted. Many gaps in our understanding of subsurface flow generation during storms have been identified, and several opportunities for additional research suggested.