SEARCH

SEARCH BY CITATION

Keywords:

  • groundwater–surface water interaction;
  • ecohydrology;
  • remote sensing

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References

[1] There is no existing method to quantitatively image groundwater processes along a seepage face. Thus, it is often difficult to quantify the magnitude and spatial variability of groundwater flux. The objective of this work is to assess the use of ground-based thermal remote sensing for fine-scale mapping of groundwater discharge and for locating the water table position along a stream bank seepage face. Seepage faces are poorly understood and often neglected in regional hydrologic studies though they likely exert significant influence on hydrologic and ecologic processes in riparian zones. Although the importance of riparian areas is broadly recognized, our ability to quantify hydrologic, ecologic and biogeochemical processes and ecosystem services is hampered by our inability to characterize spatially variable processes such as groundwater discharge. This work employs a new, transferable, non-invasive method that uses heat as a natural tracer to image spatially-variable groundwater flow processes and distinguish between focused and diffuse groundwater discharge to the surface. We report, for the first time, that thermal remote sensing of groundwater at the seepage face provides indirect imaging of both the saturated zone-unsaturated zone transition and groundwater flux at the centimeter scale, offering insight into flow heterogeneity.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References

[2] A seepage face is an external boundary of the saturated zone at which pressure head equals zero and water exits the subsurface [Boufadel et al., 1999]. Seepage surfaces typically occur along the walls of wells, on hill slopes, along stream banks and in drainage canals where a sharp pressure gradient exists between the internal saturated zone and external seepage area. Although they are common features of shallow unconfined groundwater flow systems, seepage faces are often neglected in groundwater modeling studies [Simpson et al., 2003]. At the aquifer scale (>10 km2), seepage faces comprise a small percentage of a modeling domain and are omitted when effects are considered negligible or when explicit representation is impractical due to computational inefficiency [Romanoa et al., 1999]. Limited understanding exists regarding seepage face groundwater flux and improved methods for measuring this flux are needed.

[3] Renewed interest in heat as a natural tracer of groundwater discharge has resulted in a range of techniques [Anderson, 2005]. Traditionally, both shallow and deep temperature measurements in the subsurface were the focus of heat tracer studies that used deviations from conductive temperature profiles and/or histories to quantify groundwater flow [Anderson, 2005]. More recently, thermal remote sensing [Cardenas et al., 2008; Loheide and Gorelick, 2006; Torgerson et al., 2001] and distributed temperature sensor methods [Lowry et al., 2007; Moffett et al., 2008; Selker et al., 2006] have been used to obtain spatially extensive temperature data for the study of stream-aquifer interactions and hydrologic thermal heterogeneity. In addition, thermal imagery has been proposed as a safety tool for detecting dam seepage (e.g., National Dam Safety Program, Research Needs Workshop: Seepage Through Embankment Dams, Federal Emergency Management Agency, Washington, D. C., 2002).

[4] Here, we use thermal remote sensing of a stream bank seepage face to map groundwater discharge at the interface and locate the water table position at the boundary. The work is based on the concept that groundwater has an identifiable thermal signature because groundwater temperature (TGW) is relatively constant, approximately equal to the average annual air temperature, whereas surface temperature varies on diurnal and seasonal cycles. Therefore, in the Upper Midwest (location of the study) TGW is cool relative to the surface in summer and warm relative to the surface in winter. Thermal images provide insight into centimeter scale hydrogeologic heterogeneity increasing our ability to characterize spatially variable groundwater discharge processes in riparian zones through indirect imaging of groundwater flux.

2. Methods

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References

[5] Thermal imagery and digital photography were collected of both sides of an incised drainage ditch in June 2008 and February 2009. Data was collected using a FLIR Systems (North Billerica, MA) A320 thermal infrared camera, which measures surface temperature using a 320 × 240 focal plane array with a spectral range of 7.5–13 μm. A staph with 5 cm calibrations was used to measure distances in the thermal images. Three 8-inch diameter soil cores extracted from regions of high, moderate and low groundwater seepage intensity were analyzed in the lab for hydraulic conductivity using a falling-head permeameter.

[6] Monitoring wells, equipped with continuously recording pressure transducers, were installed on both sides of the drainage ditch and were corrected with an on-site barometric logger. Air temperature (Tair) and relative humidity were recorded continuously on-site using a HOBO ProV2 (Onset Computer Corporation, MA) logger.

[7] Twenty-four hours of thermal imagery was collected in July 2008 of a seepage face in the drainage ditch. Images of this region were recorded at a rate of 5/minute. Soil moisture along the seepage face (from the stream surface to the top of the stream bank) was measured using a 5.7 cm long theta probe (Delta T Devices, United Kingdom) at 5 cm increments immediately after the 24-hour data collection.

[8] Thermal data correction accounted for emissivity, Tair, relative humidity and the distance between the sensor and the seepage face. In this study, we assumed a uniform emissivity of 0.96 based on emissivity values of water (0.98–0.99), green vegetation (0.96–0.99), wet soil (0.95–0.98) and dry soil (0.92–0.94).

[9] Data analysis included the creation of temperature profiles and estimation of the height of the water table above the stream. A program was developed to average the temperature in each row of each 320 × 240 thermal image and determine the vertical range of pixels that clearly were located within the saturated zone as well as a range clearly within the unsaturated zone. The intersection of the two best-fit lines represents the transition between the saturated and unsaturated zones and is also taken as the position of the water table, because a soil characteristic curve adjacent to the channel indicates that the capillary fringe is small (<10 cm) (Figure 1). Height in centimeters was approximated using the wooden dowel scale bar present in each thermal image.

image

Figure 1. Method for using thermal imagery of the seepage face to identify the water table. (a) Summer thermal image of a stream bank seepage face (groundwater appears cold (dark blue)). (b) Average vertical temperature profile of the summer thermal image (black) with best-fit lines of saturated zone and unsaturated zone temperature (red). The intersection of the two lines marks the location of the water table (gray). (c) Winter thermal image of a stream bank seepage face (groundwater appears warm (red)). (d) Average vertical temperature profile of the winter thermal image (black) with best-fit lines of saturated zone and unsaturated zone temperature (red).

Download figure to PowerPoint

3. Results and Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References

[10] Time-lapse thermal imagery, collected over a 24-hour period of a stream bank clearly indicates dissimilar unsaturated zone and saturated zone thermal inertia (Figure 2). Recorded temperatures in two regions of interest (UNSAT and SAT), selected in areas free of stream bank vegetation, highlight the contrast between saturated and unsaturated zone diurnal temperature variability of the stream bank. Saturated zone and unsaturated zone temperatures follow the same trend as Tair; however, the saturated zone temperature maximum, and to a lesser extent minimum, are muted relative to the unsaturated zone temperature. That is to say that the unsaturated zone has a lower thermal inertia as indicated by the higher peak temperature and lower minimum temperature relative to the saturated zone. The lower thermal inertia of the saturated zone is a result of decreases in both the thermal conductivity and heat capacity of the soil with decreasing soil moisture. In addition, the temperature fluctuation in the saturated zone is buffered by the cool groundwater discharge which is much lower in the unsaturated zone. The saturated zone to unsaturated zone temperature contrast is most dominant at the time of peak Tair (3:30–6:00 PM) and does not decline until sundown. The standard deviation of the continuous thermal data (Figure 2b) illustrates that the saturated zone has lower variability that likely results from the higher heat capacity, which is a consequence of greater water content.

image

Figure 2. Twenty-four hours of continuous thermal data collected at one stream bank. (a) Thermal image showing two regions of interest (SAT – saturated zone, UNSAT – unsaturated zone). (b) Standard deviation of the time lapse data, which exhibits the lower thermal inertia of the unsaturated zone. (c) Twelve thermal images at two-hour intervals. (d) Twenty-four hour average temperature history of two regions of interest (SAT and UNSAT). (e) Vertical soil moisture.

Download figure to PowerPoint

[11] Diurnal temperature changes confirm that peak Tair is the optimal time to collect thermal imagery of groundwater. The time-lapse thermal imagery, demonstrates that from 3:30–5:00 PM, the unsaturated zone – saturated zone temperature difference is greatest. In the 2:00 PM and 4:00 PM snapshots in Figure 2, the saturated zone appears as the cooler zone (<18.5°C), which is separated from the warmer unsaturated zone above by the water table and the warmer stream below by the stream surface. The saturated zone appears as a slightly warmer band relative to the overlying unsaturated zone and underlying stream in the 4:00 AM and 6:00 AM snapshots. It is recommended that thermal imaging of groundwater be performed at seasonal Tair highs or lows and maximum or minimum daily temperatures, respectively. At that time, the difference between Tair and seepage face temperature is greatest and the resulting thermal signature of groundwater will be most clearly evident. As a result, it is possible to image the transition between the saturated and unsaturated zone at the seepage face during those times. Ground-based thermal imagery allows for identification of the water table at the seepage face scale and this method can be used to obtain spatially extensive water table elevation data.

[12] A vertical profile of soil moisture demonstrates that the location of the saturated zone corresponds to the region of groundwater seepage identified in the time-lapse data set. Using soil moisture data, collected at 5 cm increments from the stream to the top of incised bank, the inferred water table depth is 0.54 m above the stream (Figure 2e). Both the soil moisture profile and inferred water table height correspond with the temperature profile method employed to estimate the location of the water table in each thermal image. This finding indicates that the method provides an accurate means to obtain the water table height using thermal images.

[13] Thermal imagery collected of both sides of the drainage ditch provides information about groundwater flux at the seepage face scale. Figure 3 highlights a north-south contrast in height of the water table above the stream detected with both thermal imagery and well measurements; the south-side water table is 0.4 m above the north-side water table at the time of data collection. Based on a series of images (not shown) collected along both sides of the channel, seepage face height is generally higher on the south side of drainage ditch and is spatially variable. On average, seepage face height is 0.76 m above the stream on the south side and 0.59 m above the level of the stream on the north side of the drainage ditch. Thermal imagery, using a simple algorithm, can supply high-resolution spatially-extensive data to assess seepage face water table dynamics along stream reaches. The twenty-four hour thermal imagery data set and the data comparing the south to north side of the seepage face provide a visual display of the differences between diffuse and focused groundwater discharge. This time lapse data set (Figure 2) displays a spatially uniform region in the saturated zone which we interpret qualitatively as diffuse groundwater seepage. On the other hand, the data in Figure 3, which was collected approximately 10 meters upstream, highlights discrete, focused areas of groundwater seepage. In tandem, the imagery provides a clear conceptual depiction of the differences between diffuse and focused groundwater flow and the coexisting roles that both play in generating baseflow in this channel.

image

Figure 3. Imagery and analysis showing the heterogeneous nature of groundwater flux and differences on the south and north sides of the drainage ditch. (top) Digital photographs of the south and north side of the drainage ditch with the thermal image location outlined (white rectangle). (middle) Thermal images of both sides of the drainage ditch with water table and stream locations identified. (bottom) Topographic profile of the drainage ditch with well water level (from pressure transducer), seepage face elevation and stream level identified.

Download figure to PowerPoint

[14] Hydraulic conductivity measurements of three soil cores collected in February 2009 in one thermal image frame (Figure 4) indicate that areas of high groundwater seepage correlate with high hydraulic conductivity soils whereas areas of lower groundwater seepage intensity correlate with lower values of hydraulic conductivity. In Figure 4, the high intensity, warm seepage region (A) has a hydraulic conductivity nearly 200 times greater than the hydraulic conductivity of the low intensity seepage region (C). This indicates that the thermal heterogeneity present along the seepage face is a result of variability in hydraulic conductivity. The method allows us to readily distinguish between regions of low, moderate and high groundwater discharge. However, the method is insensitive in differentiating between very low and no flow because of the very slight effect on the heat budget and is insensitive at very high discharge rates when the groundwater temperature is measured at the seepage face. In images such as Figure 4, it is likely that the warm regions below locations A and B are the result of vertical flow down the seepage face rather than groundwater discharge in these regions. Since thermal imagery only provides surface temperature measurements, hand thermometers may be used to assess temporal patterns at depth and differentiate between groundwater discharge and downward flow on the seepage face.

image

Figure 4. Winter thermal imagery and location of soil cores with hydraulic conductivity (K) measurements. (left) Winter visible image of seepage face with soil core locations annotated. (right) Inter thermal imagery of seepage face with soil core locations and K values [A = high intensity seepage, B = moderate intensity seepage, C = low intensity seepage]. Note: snow cover prevents interpretation of the thermal image above the seepage zone.

Download figure to PowerPoint

[15] Imaging of groundwater flux variability has significant implications for hydrogeologic understanding of heterogeneity and has applications to research on the scale dependencies of hydraulic conductivity, contaminant transport, groundwater flow modeling and ecohydrology. New thermal imaging methods employed in this study provide a technique to display heterogeneity at the centimeter to seepage face scale. Although limited to outcrops and seepage faces, the methods and implications of the study are important for consideration in hydrogeologic models.

[16] Understanding geological heterogeneity is critical for conceptualizing subsurface hydrology and for the accurate modeling of groundwater flow [Anderson, 1989; Eaton, 2006]. This new imagery demonstrates that groundwater flow is more discrete than commonly modeled and displays flow heterogeneity that may be important for accurate simulation of transport. The centimeter scale heterogeneity shown in thermal images, such as Figure 3, provides a visual display of why models often simulate contaminant transport at a slower rate than actually occurs. Thermal imagery shows that groundwater flow may occur in preferential flow paths resulting in faster actual rates of contaminant transport than predicted in large-scale models. The slower zones contribute to the long, low concentration tails often observed in concentration histories [Haggerty and Gorelick, 1998]. Models can be inaccurate in predicting the rates of contaminant transport due to an inability to both quantify and model heterogeneity at the decimeter scale [Wang and Bright, 2004].

[17] The importance of understanding seepage face processes may also be critical in the field of ecohydrology, which studies the intricate link between physical hydrology and the biological processes of ecology. Small differences in the water table depth from the land surface in riparian areas result in differing vegetation communities in groundwater dependent ecosystems [Loheide and Gorelick, 2007; Hammersmark et al., 2009; Henszey et al., 2004]. Although neglected in existing coupled hydrologic-ecologic models [Loheide and Gorelick, 2007], seepage face processes should be modeled in order to account for groundwater-controlled vegetation patterning, because the seepage face elevation may be significantly above the level of the stream. Neglecting this process would result in an under-prediction of groundwater depth of ∼0.76 m on the south side of the stream studied here, leading to an expectation of a drier vegetation community than what is actually present.

4. Conclusion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References

[18] Thermal imaging of groundwater at the seepage face provides a new method to monitor this groundwater discharge process. The method is noninvasive, easily transferable, uses heat as a natural tracer and has a variety of research applications. This paper provides a method to 1) observe the water table using thermal imagery, 2) differentiate between diffuse and focused discharge, 3) qualitatively distinguish low, moderate and high discharge rates and 4) enhance our understanding of the heterogeneous nature of groundwater flux. These findings may assist groundwater modelers, water resources managers, ecohydrologists and agricultural engineers in visualizing and monitoring groundwater flow at discharge interfaces.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References

[19] Financial support for this work was provided by the Wisconsin Groundwater Coordinating Council, the Wisconsin Alumni Research Foundation and the Wal-Mart Stormwater Compliance Team. The authors acknowledge E. Booth, N. Pathak and X. Wang for assistance with lab/field work and C. Lowry, K. Singha, S. Tyler and two anonymous reviewers for constructive comments on this manuscript.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References