|Documentation||Vadose Zone||Conjunctive Simulation||Sleepers River||Watershed Flowpath|
The hydrology and geochemistry of the Sleepers River watershed has been widely studied and that research has identified a number of processes important for controlling stream flow. However, to date the watershed has not been examined with a fully-coupled physically-based numerical groundwater/surface water model. In the work presented here a portion of the W-9B watershed is examined with MODFLOW-WHT to compare the experimentally-identified flow processes with the integrated variably-saturated groundwater/surface water model. In particular, two concepts are tested: 1) that subsurface pre-event water represented most of the storm flow and 2) that transmissivity feedback controls the response of the stream.
The numerical model suggests that groundwater hydraulics plays a controlling role in flow to the stream. Precipitation and snowmelt along the flowpath increase the hydrostatic pressure, which increases baseflow and return flow/saturation overland flow. The aerial extent of overland flow increased during recharge events, but it was relatively insensitive to recharge intensity. The model data support the concept that most of the storm-derived flow to the stream was "pre-event" water that existed in the subsurface prior to the storm event. In general terms, the model data do not support the "transmissivity feedback" concept. Variations in hydraulic conductivity in the model changed the proportions of overland and base flow, but flow to the stream was not sensitive to the vertical distribution of hydraulic conductivities.
Model Grid and Initial and Boundary Conditions
The model grid is composed of 99 columns and 12 layers. The top layer is used to simulate overland flow. The next five layers represent the soil, the next three represent the till, and the bottom three layers represent the bedrock. The soil and peat in the model range in thickness from 0.86m to 1.27 meters and the till thickness ranges from 0.26 to 0.67 m. The saturated hydraulic conductivity of the upper three soil layers was initially set to 0.6m/hr, the next two soil layers were set to 0.3 m/hr and the till was set to 0.1 m/hr. The capillary pressure/saturation data used for both the soil and till are listed in Table 1. The initial water table was estimated using the McGlynn (1999) data from the period prior to significant rainfall and snow melt. The end and bottom boundaries were all "no flow". There were two constant head cells at the downgradient end of the flowpath to represent the stream (head changes in the stream were insignificant in comparison to head changes within the flowpath.)
Results and Discussion
Normalized total flow to the stream from the flowpath is shown in Figure 1, along with normalized stream flow from the W-9 watershed as a whole. The figure indicates that agreement between the model results and the experimental data is good, both in magnitude and in timing. (Both the model results and experimental data are average daily flow values). Results from the variably-saturated groundwater/surface water flow model indicate that the majority of flow to the stream came via pathways that involved the subsurface. For the conditions simulated, about half of the flow during the spring melt period came as base flow and have came as return flow and saturation overland flow (Figure 2). The flowpath area contributing saturation overland flow was on the order of 20%. This means that during recharge conditions on the order of 20% of the recharge will fall on this contributing area and be transported directly to the stream via overland flow. Figure 2 also shows that during low-recharge conditions, saturation overland flow comes to a halt and all of the flow reaching the stream is as base flow.
For the period simulated (which represents the highest recharge rates of the year) the primary factors controlling flow to the stream were the hydraulic conductivity of the stream and the hydraulic head generated by recharge throughout the watershed. In the latter context, while groundwater residence times in the subsurface are much longer than the observed stream flow responses, baseflow still provides an important pathway for rapid stream response because the hydraulic pressure is quickly transmitted down the flow path and results in rapid increases in base flow due to displacement of the pre-event water into the stream. In this context, the results of the numerical model are in agreement with previous work at the site (Hjerdt, 2002; McGlynn, 1999). However, the model results generally do not support the idea that "transmissivity feedback" is an important mechanism. Regardless of the hydraulic conductivity distribution of the soil, during "drier" seasons of the year much of the recharge flows to the stream as baseflow. When the capacity of the aquifer is reached, the soils become saturated and return flow is initiated. (In the context of this and other forested watersheds, return flow and saturation overland flow may actually be occurring within the forest litter layer at ground surface.)
Hjerdt, N, 2002, Deconvoluting the Hydrologic Response of a Small Till Catchment: Spatial Variability of Groundwater Level and Quality in Relation To Streamffow, Ph.D. Dissertation, State University of New York, Syracuse, New York.
McGlynn, B.L., J.J. McDonnell, J.B. Shanley, C. Kendall, 1999, Riparian zone flowpath dynamics during snowmelt in a small headwater catchment, Journal of Hydrology 222 75-92
|Return to: Top | MODFLOW-WHaT | MF-WHaT Documentation ||