Allison Craig

Allison worked on the interaction between ground water and surface water at two streams in Nebraska and Indiana. She conducted field work using seepage meters and did theoretical analyses using analytical solutions and the numerical model vs2dh.

Allison modified some of our earlier designs for seepage meters developed by Susan Kelly. She improved the hydrodynamics, which reduced the scour, and she improved the design for the bag. Allison also evaluated the performance of a seepage meter based on a diluted tracer. That work was required just to get ready for the field. Allison then hauled her new seepage meters out to Nebraska and Indiana as part of the USGS ACT project, where she made hundreds of measurements of seepage and stream characteristics. She developed the most detailed data set that I know of showing how ground water discharge to streams varies spatially. She also evaluated the performance of three different seepage meters and evaluated the use of stream bed temperature as a surrogate for seepage.

For more details, you should read Allison's Abstract, Discussion and Conclusion below, or her thesis. Here is a powerpoint presentation she made. Photos of her field equipment and study area are also available.

_______________________________

ABSTRACT

Groundwater and surface-water interaction plays an important role in the transport and cycling of contaminants and nutrients, which subsequently affects ecosystems throughout subsurface and stream environments. This interaction varies both spatially and temporally and is consequently difficult to characterize. The purpose of this study is to (1) compare several techniques for measuring groundwater discharge to streams, (2) characterize the spatial and temporal variability of groundwater discharge within a study reach and (3) compare patterns of groundwater discharge between two streams in different geologic settings.

Three seepage meter techniques were conducted from the same pan: the pan and bag; piezo-seep; and dilution meter. These techniques were compared at several locations. Pan and bag seepage meters were then deployed several meters apart throughout streams in central Nebraska and Indiana to characterize the spatial variability of groundwater discharge. Sixty-seven locations were sampled with approximately five repetitions at Maple Creek in central Nebraska and 24 locations were sampled with 10 repetitions each at Leary-Weber Ditch in central Indiana. Pan and bag measurements were then repeated over time scales that ranged from hours to weeks to characterize the temporal variability at Maple Creek. Streambed temperatures were also recorded at seepage meter locations. Vertical streambed temperature profiles recorded from buried sensors at one location were used to predict rates of groundwater discharge.

Groundwater discharge varied spatially by one to three orders of magnitude within the reaches, but the arithmetic means of groundwater discharge at the two streams were both 1.0 x 10 -3 cm/sec. The three seepage meter techniques were statistically equivalent; however, the techniques varied in their ability to resolve small-scale temporal variations in flux. The piezo-seep could resolve groundwater discharge with an accuracy of approximately 1 cm/sec whereas the pan and bag seepage meter exhibited measurement errors of approximately 10 cm/sec in some trials. Groundwater discharge was greatest directly downstream of piles of woody debris, along streambed riffles, and in areas of coarse-grained sediments. Fluxes were generally greatest in the deepest parts of the channel and the lowest fluxes were recorded in areas of fine-grained sediments along the edges of the channel. No systematic trend in the temporal pattern of groundwater discharge was observed and the temporal variations are small in comparison to spatial variability. Streambed temperature ranged from approximately 10 to 20°C at both study reaches.

Streambed temperature was negatively correlated with groundwater flux, as expected; however, the details of the correlation were more complicated than anticipated apparently due to variations in groundwater temperature that result from variations in heat transfer surrounding the stream. Results from a theoretical analysis indicated that variations in the height of the stream banks and the thickness of the aquifer may affect the temperature of groundwater discharging to the stream. Subsequently, groundwater discharge rates cannot be predicted from a correlation of discharge and streambed temperature at Maple Creek.

An analytical solution was used to predict groundwater discharge rates from profiles of streambed temperature recorded at one location in Maple Creek, NE. The analysis indicated an average discharge rate of 1.1 x 10 -4 cm/sec for the location. An additional estimate of groundwater discharge rate of 1.3 x 10 -4 cm/sec was obtained by extrapolating from pan and bag seepage meter results recorded during the same time period.

Stream-dwelling organisms depend on groundwater to stabilize stream temperature as well as to supply essential nutrients to the stream. The three order-of-magnitude range in groundwater discharge rate and the 10 °C range in stream bed temperature observed during the study suggest that a variety of microhabitats are available for microorganisms within the two study reaches. Groundwater discharge and streambed temperature are affected by physical stream characteristics such as topography, substrate and debris and as a result, species distributions should adjust in response to these features.

Discussion

Spatial variation

Groundwater discharge varied by an order of magnitude over several meters at both Maple Creek and Leary-Weber Ditch. Spatial variation of groundwater discharge appears to be related to stream bed topography, sediment grain size, and the location of in-stream debris. However, despite large differences in geologic material in the stream bed, the overall average of groundwater discharge at each study reach was similar.

The highest flux value of 5.0 x 10 -3 cm/sec observed at Leary-Weber Ditch was recorded along the downstream end of a riffle (Figure 32). As streambed profile changes with alternating pool-riffle sequences, the elevation of the water surface also changes. Habel (2004) used an analytical model to show that a change in water surface elevation due to wave motion induces upward vertical flow of more than one-half the wavelength and downward flow over the other half. The ratio of the wavelength to the depth of the channel was also found to affect the flux across the stream and streambed interface. Fluctuations in the water surface elevation may induce hyporheic flow paths through the streambed if the downward flux exceeds the rate of groundwater discharge. However, if the ambient groundwater discharge is greater, the upward flux may only be diminished by the downward flux induced by the change in elevation. Subsequently, when the water surface elevation decreases along a riffle, groundwater discharge increases as hyporheic flow is discharged to the stream.

The greatest groundwater discharges recorded at Maple Creek were associated with increased sediment grain size and/or located downstream of woody debris (Figures 26 and 31). The highest flux of 4.0 x 10 -3 cm/sec was recorded directly downstream of the largest woody debris pile in the study reach. The second, third and fourth highest fluxes of 3.9 x 10 -3 cm/sec, 3.4 x 10 -3 cm/sec and 2.8 x 10 -3 cm/sec were adjacent to one another and located in areas with coarse-grained beds and downstream of woody debris piles. The fifth largest flux of 2.2 x 10 -3 cm/sec was also located in an area with a coarse-grained bed. The lowest fluxes of 0.1 x 10 -3 cm/sec were recorded in areas where the bed was fine-grained along the margins of the channel. According to the Hazen method of determining hydraulic conductivity, K is equal to the product of the square of the effective grain size and sorting coefficient. The Hazen method was used to calculate the hydraulic conductivity for three grab samples collected from Maple Creek. Hydraulic conductivity of the surface sediments was found to range from 1.28 x 10 -1 to 2.3 x 10 -2 cm/sec.

No consistent pattern of groundwater discharge across all transects was observed. Fine-grained bed sediments were generally located along the margins of the stream whereas coarse-grained sediments were more prevalent through the thalweg of the study reach, which may have had some effect on the distribution of groundwater discharge. Discharge was greatest along the left side of the channel for the eight most upstream transects (ie. transects 1 through 8). The discharge was greatest on the right side of transects 9 and 10, and in the middle of transects 11, 12, and 13. In each case the areas with the highest groundwater discharge generally corresponded to the deepest section of the channel, similar to the results found by Kelly (2003) but unlike results from Shaw and Prepas (1990a) which suggest that groundwater discharge into lakes is greatest near the margins of the water body.

Large obstructions or debris in stream beds alter the elevation of the water table by increasing the elevation of the water surface on the upstream end of the debris pile, and decreasing on the downstream end. Where the water elevation is higher upstream of debris, groundwater discharge will decrease and may even become negative if the ambient groundwater discharge is low. Groundwater discharge will increase at the downstream end of the obstruction where the water surface elevation falls. If negative discharge is induced at the upstream end of the obstruction, it will discharge at the downstream end causing a shallow hyporheic flow path circulating stream flow through the streambed. This effect was not observed with pan and bag seepage meters during the study because it may be localized to an area several centimeters upstream of the debris. Safety issues due to the high stream flows during the study period prevented sampling from the area directly upstream of debris.

According to a 2000 EPA National Inventory of Water Quality, agricultural contaminants, including pesticides and nutrients, are the number one cause of surface water quality degradation. These contaminants are mainly transported from agricultural areas by surface runoff into streams. As contaminated surface water is circulated through the upper portion of the streambed, microorganisms break down and metabolize the contaminants. In addition, some contaminants are thought to sorb onto streambed sediments where they are stored until after a storm pulse. Obstructions and debris as well as pool-riffle sequences should increase the rate of this cycling by inducing hyporheic flow which circulates surface water through the streambed. Though negative discharge was never observed during the study, increased discharge observed directly downstream of obstructions and debris may be the result of this cycling. Chemical analyses of bag samples would be necessary to provide evidence of this effect.

Temporal variation

Temporal variations were investigated on scales ranging from one day to three weeks. No systematic pattern of temporal variations in groundwater discharge was observed at Maple Creek. In most trials the means varied slightly but were within the error of the method.

Diurnal variations on groundwater discharge result from daily fluctuations in evapotranspiration and barometric pressure. In areas with dense riparian vegetation, evapotranspiration increases with temperature and sunlight during afternoon hours and then decreases during the evening and morning hours. As evapotranspiration increases, groundwater discharge should decrease as the water table is drawn down; however, in areas lacking dense riparian vegetation, including large trees with deep roots that penetrate the groundwater, diurnal fluctuations due to evapotranspiration may be negligible. Maple Creek is characterized by grass and small shrubs along the stream banks but large woody trees are absent from the study reach. The lack of dense riparian vegetation may be why diurnal variations are small at Maple Creek, but larger at other sites where large trees are present (Kelly 2001).

Temporal variations in groundwater discharge that result from fluctuations in stream stage were also investigated throughout the Maple Creek study reach. Seepage meter measurements made along one transect before and after a small storm and at a small difference in stream stage were significantly different at only one location. The spatial pattern of groundwater discharge along the transect was maintained between the two sampling times with the highest fluxes at the left side of the channel and the lowest fluxes to the right of the channel.

Other locations were sampled before and after several large storms and at a greater difference in stream stage. The means at each location were only statistically different at 36% of the locations and no systematic trend in the differences was observed. Because the means were not consistently greater at the lower stage, the differences may be a result of changes in the distribution of streambed sediments that result from large increases in stream flow.

No systematic trend in temporal variations was observed on any scale throughout Maple Creek. The storm pulse and change in stream stage produced different fluxes in some locations, but were significantly equivalent in most trials. The error inherent in pan and bag seepage meters appears to obscure the small temporal variations at Maple Creek. Overall, temporal variations appear to be small in comparison to the spatial variations in groundwater discharge at Maple Creek. Large changes in stream stage likely result in large changes in discharge; however, it was not feasible to use seepage meters in the stream at high stages.

Comparison of techniques

Though pan and bag seepage meters are relatively simple and inexpensive devices, they have several limitations. Errors are associated with bag type and size as well as bag conductance and velocity head (Murdoch and Kelly 2000). These effects have been reduced by using a compliant bag and isolating the bag from stream flow, but errors have not been eliminated. These errors can be investigated by comparing the pan and bag results with other types of seepage meters that do not utilize a collection bag.

Three seepage meter types, the piezo-seep meter, dilution meter, and pan and bag seepage meter, were compared at a subset of locations to examine whether the pan and bag seepage meter results were reliable. In each trial the three tests were conducted from the same pan without moving it to eliminate spatial variability. A one-way ANOVA performed on the data set concluded that the results were statistically the same and the means could not be differentiated from one another. Consequently, the three techniques are all considered to be equally viable options for measuring groundwater discharge. The techniques do, however, vary in their abilities to resolve small fluctuations in groundwater discharge. The pan and bag results exhibited the greatest standard deviation of 2.97 x 10 -4 cm/sec, while the piezo-seep exhibited the lowest standard deviation of 1.91 x 10 -4 cm/sec. The dilution meter results for the seven trials in which all three techniques were compared had a standard deviation of 5.15 x 10 -4 cm/sec.

Differential stream gauging provided an additional estimate of groundwater discharge at Leary-Weber Ditch. The average groundwater discharge was determined by measuring the stream discharge at the upstream and downstream ends of the study reach and dividing the difference by the estimated surface area of the streambed. The average flux was 1.5 x 10 -3 cm/sec which was approximately fifty percent greater than the average groundwater discharge of 1.09 x 10 -3 cm/sec measured with pan and bag seepage meters. Differential stream gauging depends on the accuracy and difference in two volumetric discharge measurements. If the stream flow in a reach is low, the differences are small and therefore the error of the technique becomes a large percentage of the estimated stream flow. Leary-Weber Ditch is a small stream with a typical stream flow of 1 to 2 ft 3/s, so the error inherent in stream gauging at the reach may be a large percentage of the estimated discharge.

Temperature-discharge correlation

Shallow streambed temperatures were correlated with groundwater discharge measurements made from pan and bag seepage meters to determine whether temperature could be used as a simple surrogate for detailed seepage meter measurements. Repeat seepage meter measurements throughout a study reach are time-consuming, but multiple temperature measurements can be made easily and rapidly. If a correlation between seepage meter discharge measurements and temperature measurements can be developed, temperature measurements can be used to infer groundwater discharge at any location within the study reach.

Theoretically, groundwater discharge is negatively correlated with temperature in areas of deep groundwater discharge but may not be in areas with significant lateral or hyporheic flow. The poor correlation between groundwater discharge and temperature at both reaches examined in this study suggests that lateral flow may be occurring in addition to groundwater discharge. Some high discharge areas at both study reaches exhibit correspondingly low temperatures associated with deep groundwater discharge, whereas others do not, possibly as the result of hyporheic flow induced by streambed topography and other features which act to circulate surface water through the streambed. As a result, temperature cannot be used as a quantitative surrogate for seepage meter measurements at the two study reaches, but may identify the source of flow at measurement locations.

In addition, the topography of the area surrounding the stream may affect the temperature of the streambed. Several transects at the downstream end of the Maple Creek study reach exhibited temperatures two to six degrees warmer than temperatures observed throughout the rest of the reach. The area is located adjacent to and downstream from a shallow and unvegetated sandy point bar. As shallow groundwater flows through the point bar it becomes heated and then discharges to the stream in the downstream end of the reach. As a result, areas characterized by high groundwater discharge may also exhibit warm stream bed temperatures, further complicating the correlation of flux and temperature.

The correlation of temperature and flux at Maple Creek was used to delineate nine classes of streambed areas with characteristic ranges of temperature and flux. The classes were created by dividing the ranges of temperature and flux into groups of high, medium, and low temperature and flux. Two classes of low temperature/ low flux, and low temperature/medium flux were dominant throughout the upper three-quarters of the reach, whereas a class of high temperature/high flux occupied the downstream quarter of the streambed. The six remaining classes exhibited a patchy distribution and were each only present at a few of the sampled locations.

A relatively high temperature of 19° C was recorded along transect 10 where the highest flux was exhibited; however, because transect 10 was located along a streambed riffle, the high temperature is considered to be indicative of shallow hyporheic cycling of surface water through the streambed.

Conclusion

The averages of groundwater flux from two streams in different hydrogeologic settings were strikingly similar, but the streams exhibited different ranges of groundwater discharge. Spatial variations in groundwater discharge appear to depend primarily on physical streambed features including pool-riffle sequences, sediment grain size, and the location of in-stream debris. The greatest fluxes generally occurred in the deepest sections of the Maple Creek study reach as observed by Kelly (2001), but the pattern was variable within the Leary-Weber study reach. Low flux areas occur where the bed grain size is relatively fine along the margins of the stream. High flux areas were located along streambed riffles, downstream of debris and in areas of large grain size. Changes in stream surface elevation caused by debris piles and streambed topography may induce hyporheic flow paths which increase the spatial variability of groundwater discharge within the reaches (Devito 1996), which subsequently increase the cycling of nutrients through the streambed and may in turn affect the distribution of stream bed organisms. These effects may be further resolved by chemical analysis of pore water samples and biological sampling of hyporheic water.

Overall, temporal variability appears to be small in comparison to spatial variability within Maple Creek. Temporal variations in groundwater discharge over scales from one day to several weeks could not be resolved by the pan and bag seepage meter. Though the flux means varied between trials at some locations, the variations were small and within the error of the device at most locations. The piezo-seep yielded more consistent hourly results during each sampling day, however no systematic diurnal trends were observed. The piezo-seep meter may, however, be able to resolve diurnal trends in an area with more dense riparian vegetation and greater evapotranspiration, as shown by Kelly (2001).

Design improvements made to the pan and bag seepage meter were effective in eliminating scouring around the pans. The 0.6-meter-long design was the most functional design for connecting the bag container to the pan, whereas the 1-meter-design made the connection of the bag container difficult (especially in a strong current) because the longer carapace made it difficult to access the fitting on the downstream side of the pan. The design of the bag container successfully protected the bag from stream flow and prevented it from folding. By placing the bag and bag attachment on the side of the pan, the seepage meters could to be used in more shallow water than in previous experiments (approximately 3 centimeters), allowing the margins of the stream channel to be sampled.

Fluxes measured using three seepage meter types at the same locations were statistically equivalent. The pan and bag results exhibited the greatest standard deviation whereas the piezo-seep meter had the lowest standard deviation of the three techniques. An additional estimate of groundwater discharge obtained by differential stream gauging at Leary-Weber Ditch was approximately fifty percent greater than the estimate obtained with seepage meters, however because of the low stream flow within the reach, a large percentage of the stream gauging estimate may be attributed to error.

Shallow streambed temperatures were observed to vary by more than 10 degrees at both study reaches, apparently due to variations in groundwater temperature caused by the topography of the area surrounding the stream. Shallow streambed temperature and groundwater discharge exhibited a weak negative correlation. As a result, streambed temperatures may be useful in determining the source and direction of groundwater flow as well as delineating micro-habitats and fish spawning areas, but groundwater discharge rates in the two study reaches cannot be effectively predicted through the correlation temperature-discharge correlation similar to that presented by Conant (2004).

Contour maps of temperature display cool temperatures coinciding with some high discharge areas and warm temperatures in some low discharge areas, but some high discharge areas exhibit surprisingly warm temperatures. High discharge areas displaying warm temperatures may be the result of physical streambed features which induce shallow hyporheic flow paths that circulate surface water through the stream bed (Silliman et al. 1995, Alexander and Cassie 2003). These effects may be restricted to an area within a few centimeters of the stream bed features, and were not directly observed during the study due to the limitations in positioning seepage meter pans directly adjacent to debris in a high-flow current. However, surface water has been observed to circulate through the streambed below pool-riffle sequences (Devito 1996, Alexander and Cassie 2003), beaver dams and boulders (White et al. 1987) using chemical analysis of stream bed water samples.

The height of the banks surrounding the stream may also affect the temperature of groundwater discharging to the stream. Groundwater discharging from below a higher bank will be cooler than groundwater discharging from below a lower bank. As shallow groundwater flows less than a meter below the ground surface, it is heated during warm summer months (especially in regions lacking significant vegetation). Conversely, groundwater flowing several meters below the ground surface is less affected by the atmospheric temperature. As a result, stream bed temperatures will vary due to the height of the adjacent banks.

The high variability in the spatial distribution of temperature and flux throughout the streambeds suggest that a variety of microhabitats for streambed-dwelling organisms may be present within the reaches. Groundwater both stabilizes stream bed temperature and supplies nutrients to the stream (Hayashi and Rosenberry 2002) and as a result, the pattern of species distribution within a streambed is directly related to temperature and flux (Fraser and Williams 1998). Because stream bed temperature and groundwater discharge rates are affected by physical stream characteristics such as topography, sediment grain size, and the location of debris, species distributions should adjust in response to these features.

Last Updated: August 28, 2007 -- Questions or comments, contact Larry Murdoch.
School of Environment, Department of Geological Sciences
340 Brackett Hall, Clemson, SC 29634  USA -- Telephone: (864) 656-3438, Fax: (864) 656-1041
© 2004-2005 Clemson Univeristy, Clemson, SC, USA. All rights reserved.