Seepage Meter Paper - version 10

From GeoMod

Neilans, D., and Urbano, L.D., Assessing meter-scale heterogeneity of groundwater discharge using streambed thermal mapping, Journal of Hydrology, (in review)

Title: Assessing meter-scale heterogeneity of groundwater discharge using streambed thermal mapping

Abstract

Understanding the spatial distribution of groundwater seepage into streams is a crucial component in understanding the physical, biological and chemical interactions that occur at the sediment-water interface. This work was undertaken to evaluate the ability of stream-bed thermal measurements using a simple temperature probe to rapidly and accurately map groundwater discharge into a gaining, sand-bedded stream. Seepage meters used in tandem with temperature sensors imbedded at various depths in the sediment (20 cm, 10 cm, and bed surface) indicate a good (R² = 0.76) correlation between stream-bed temperatures and seepage rate. Using this method, weekly maps were produced of a 30x20 meter reach of a small, channelized river in Western Tennessee on a 1-meter grid. The thermal maps show that persistent and concentrated discharge of groundwater in the form of sub-aqueous discharge foci or "springs" were the primary source of seepage into the river. These discharge foci persisted throughout the two-month study period even after reorganization of the shallow (< 1 m depth of sediment) stream-bed geomorphology during a storm event. Comparison of thermal and bed elevation maps indicate that the discharge foci contribute significantly to the geomorphic form of the riverbed; discharge foci were able to maintain depressions in the stream bed even with passage of sand bars over their locations. The significant heterogeneity in the groundwater discharge pattern indicates that point-based methods of groundwater seepage measurements need to account for the intermediate scale variability in discharge if their results are to be generalizable to larger scales.

Introduction

Seepage meters designed to intercept discharging groundwater are considered particularly reliable because they directly measure seepage (Becker et al., 2004). Belanger and Montgomery (1992) investigated the accuracy of seepage meters using a large (3 m diameter) laboratory sand tank that was filled with layers of gravel and sand to simulate a uniform riverbed. Seepage meters made from 50 gallon drums were inserted into the model riverbed and water was then forced up through the bed material. While the overall water supply rate to the apparatus was constant and the bed material was installed as uniformly as possible, variations in seepage rate were found throughout the tank. Tank discharge rates were found to be consistent for each seepage meter over the duration of the experiment. However, spatial heterogeneity was found to exist over sub-meter scales, that when averaged gave the expected discharge rate. Thus, in a controlled environment, seepage rates vary significantly across sand and gravel bed material. Accounting for the spatial variability of discharge in the field is difficult. Installing a dense network of thermocouples and seepage tanks quickly becomes prohibitively time consuming and expensive. Furthermore, the extensive use of these intrusive methods is likely to have a significant effect on the hydraulic properties of the streambed sediments.

Using a single or even a few point measurements to infer general seepage rates for a river is inherently flawed because natural river system sediments are not uniform. Becker et al. (2004) conducted a groundwater seepage survey of Ischua Creek in New York using a single-point temperature measurement method to infer groundwater seepage rates. Although the measurement accuracy at individual points was very high, when extrapolated for the entire reach and compared to traditional stream gauging measurements of groundwater discharge, the calculated values were off by an order of magnitude. Meandering or anastamosing streams continually change channel location and their stream beds can become layered with high- and low-conductivity sediment. Sand bars, ripples, and gravel lags constantly change position and scale due to changing flow parameters such as flow velocity, turbidity, and bed friction (Knighten, 2002). Bed-scale features can also result in heterogenous clogging of pore spaces by fine-grained suspended sediments (Packman and MacKay, 2003). These near-surface sedimentological changes can potentially affect the spatial pattern of groundwater discharge sourced from both regional and hyporheic flow (Kasahara and Wondzell, 2003).

Numerical modeling of small reaches also indicates the potential for heterogeneous groundwater discharge rates. Storey et al. (2003) found that the cross-sectional shape of the stream bed has a profound effect on the local groundwater system. In a gravel-bedded, gaining stream, they found that the shape of a riffle imposes a hydraulic boundary that focuses groundwater flow toward the river banks. Their model showed seepage at the boundaries to be twice as high and more variable than mid-river discharge.

Identifying the meter-scale patterns of groundwater discharge into streams is difficult because of the lack of techniques for efficiently measuring groundwater seepage rates at this intermediate scale. The most common means of stream/groundwater exchange measurement is by gauging river discharge at discrete locations and then attributing any gain or loss in water volume to the underlying aquifer. The large scale of this method limits the scope of the observation to that of an entire reach. Alternatively, seepage meters that intercept water discharging into the stream (Paulson et al., 2001) and buried thermocouples that measure groundwater flow induced changes in the geothermal gradient are frequently used to determine point seepage rates. Point measurements, however, may not be characteristic of reach scale groundwater seepage due to the heterogeneity of the groundwater discharge pattern (Belanger and Montgomery, 1992). Currently, the only way to obtain high resolution temperature data of a substantial area is through the use of FLIR (forward looking infrared) Thermal Infrared imaging. The thermal camera is able to record temperature data in very high detail, but is limited to measuring the temperature of the water’s surface (Torgerson et al., 2001). Any mixing that occurs within the river distorts the groundwater discharge signal.

As part of a study into the geomorphic effects of groundwater on small river systems, we tested a method of rapidly determining the spatial pattern of groundwater discharge at the meter scale using portable thermal probes. These probes have the advantage of measuring surface temperature very rapidly while still being very accurate, as well as making possible the estimation of discharge for the reach along with fine spacial detail of discharge. Unlike penetrating probes or seepage meters, they do not effect the underlying hydraulics and avoid most of the error that may be incurred from disturbing the underlying sediment.

Study Area

The Mississippi Embayment offers an ideal location to study the spatial and temporal patterns of groundwater seepage to streams due to the abundance of small meandering rivers and relatively simple near-surface geology. The embayment itself is shaped similar to a southward plunging syncline composed of shallowly dipping alternating layers of sand and clay (Cushman et al. 1964, Hosman and Weiss, 1991). The syncline axis follows the Mississippi River. Regional groundwater flow tends to follow the gradient of the regional topography (Authur and Taylor, 1998), which reflects the embayment syncline.

The rivers in Western Tennessee tend to cross the sand and clay outcrops perpendicular to strike direction, flowing from east to west towards the Mississippi River. Previous studies have indicated that large systematic changes in groundwater-surface water exchange can be found along the Loosahatchie River, although the exact location and degree of these changes is not yet known (Urbano et al., 2006). In 1968, the Loosahatchie River was channelized by the Shelby County Drainage District to reduce flooding and increase drainage potential. Prior to channelization, the Loosahatchie River was highly sinuous with a general east/west orientation. Channelization of the Loosahatchie has resulted in substantial downcutting, creating steep-sided banks.

The Loosahatchie River was chosen for this study due to its gaining nature and bed composition of uniform medium sand. Repeated weekly sediment sampling in various locations of the study area has shown that the bed sediment is extremely uniform and evenly distributed. No underlying shallow geologic or depositional structure could be found to explain the presence of heterogeneous discharge through the homogenous sediment, although it is important to note that the confining layers in the Mississippi embayment are not of uniform thickness, nor are they always continuous. The possibility exists for the confing clay layer to be thin or absent throughout the study area(Renken, 1998), which could result in significant localized changes in groundwater discharge (Urbano et al., 2006). The laboratory model constructed by Belenger and Montogomery (1992) was also composed of an extremely homogenous sand, yet they too found a natural tendency of discharging water to find and/or develope a preferred path resulting in non-uniform distribution of flow.

A 30 meter reach (23 meter width) that we had previously determined to be gaining was chosen for this study. The area contained a uniform bed of medium sand under approximately one-half meter of water at base-flow conditions. Data was collected during low flow conditions in September and October to ensure that the seepage rate would not change during sampling.

Methods

Bed temperatures were obtained by attaching a type-K wire thermocouple to the bottom of a round 16-cm diameter plate at the base of a 2-meter long pole and then by placing the exposed thermocouple directly on the riverbed surface for the two seconds needed for an accurate measurement. The thermocouple lead was then plugged into a data logger with digital readout attached to the top of pole. Measurements were recorded by hand and with the logger to ensure continuity of the measurements. The temperature probe was insulated between itself and the plate to reduce the plates influence on temperature measurements. The purpose of the plate is to isolate the temperature probe from the river water, only permitting interaction with the riverbed surface. The type K thermocouple used has an accuracy of 0.1 °C with a response time of two seconds.

To measure temperatures below the bed surface, a 20-cm length of 2-cm diameter PVC pipe was connected to the pole below the plate at the base of the pole. Type K wire thermocouples were then attached to the PVC pipe with Duct tape at the desired depth, leaving only the tip of the wire exposed. A traditional sheathed thermocouple was initially used, however the significant amount of hematite in the sand interacted with the metalic sheath material causing a significant amount of electrical interferance. The two-millimeters exposed on the wire thermocouple did not offer a large enough surface area to cause any interferance and also added an extremely short response time. Polyvinyl sheathing of the wire prevented any interference from the sediment or water.

We compared stream-bed temperatures to temperatures at greater depth that were not affected by diurnal temperature changes and to seepage rates measured using seepage meters.

Determining the depth of no diurnal influence

To determine endmember temperature values for river water and groundwater, the temperature of the river water was logged for 48 hours as well as the temperature of the water in a nearby well to a depth of 30 feet. It was found that between the hours of 1:00 PM and 4:00 PM the temperature of the river water remained within a half degree Celsius of the peak diurnal temperature, which was 21.0 °C on that day. Since the river water directly contacts the bed sediment, any change in river water temperature has the potencial to affect the riverbed temperature. Bed temperature measurements were recorded during the stable river temperature interval to ensure that changing riverwater temperatures would not be a source of error. The temperature of the river water was also measured at varying depths to confirm that the stream water was well mixed. Water at all depths above the bed were uniform within the error of the thermocouple. In contrast, the temperature of the water in the well remained at 16.4 °C for the entire 48 hour period.

The effects of diurnal heating and cooling are known to transmit to some depth in the bed sediment. Wells and Annear (2002) found that the depth of diurnal influence decreased with decreasing bed sediment size but was, in general, lost within a one-half meter depth beneath the bed surface. At this depth of no influence, an accurate calculation of water transit rate from temperature can be made (Becker et al., 2004). Because we ultimately used stream-bed temperatures to estimate discharge rates it was necessary to prove that the temperature profile at the depth of no diurnal influence was proportional to the bed surface temperature.

The sediment temperature probe was used to record temperatures over several days at various depths below the sediment-water interface. At 20 cm depth beneath the stream bed, the diurnal temperature range was 0.3 degrees C, compared to a 3.1 degree range in the water column. We conclude that 20 cm closely approximates the depth of no diurnal thermal influence. To minimize any residual diurnal effects, subsequent transect measurements were only made between 1:00 PM and 4:00 PM when river water remained at a relatively constant temperature.

Figure 1 – Temperatures recorded on 10/01/2006 at various depths in the river bed sediment to monitor the diurnal change with depth. Free flowing water in the river fluctuated by several degrees Celsius, whereas water 20 cm below the bed surface only fluctuated 0.3 degrees Celsius.

Correlating bed surface temperature to temperature at depth

A temperature probe was then set to record temperature at depths of 20 cm, 10 cm and 0 cm below the riverbed surface along three transects across the river at one meter intervals. Temperatures measured at the riverbed were systematically higher than temperatures at depth (Figure 2). There is however, a convincing correlation between these temperatures; the correlation coefficent (r²) between 20 cm and the bed surface was 0.65. This lends to the confidence that the variability of bed surface temperatures is proportional to the variability at depth.

Figure 2 – Temperatures recorded on June 16th, 2006 at various depths below the bed surface of the Loosahatchie River, Tennessee. Measurements yielded an r2 value of 0.85 between 20 cm and 10 cm, and an r2 value of 0.65 between 20 cm and the bed surface.

Bed Surface Temperature and Seepage Meter Discharge

Temperature differences observed across the stream transects were related to seepage meter measurements to determine if the correlated to differences in groundwater discharge rates. On one date in the Summer and one in the Fall of 2006, seepage meters were installed in pairs at random locations within the 30 by 23 meter reach and allowed to collect groundwater. On September 6, three locations were chosen to place pairs of seepage meters while on October 1, six locations were used. Individual seepage meter rates in each pair of our study were consistent over time, but were different from each other despite being placed less than 10 cm apart (Fig. 3). In accordance with findings of Belenger and Montgomery (1992) the average seepage rate was used for each pair of tanks. The temperature of the riverbed was collected before insertion of meters to eliminate false readings incurred by disturbing the bed sediment.

A linear regression of the temperature data and the seepage rate data for September 6, yielded a regression value of 0.98. Data for October 1, yielded a regression value of 0.82.

Figure 3 – Combined regression plot of data from September 6th and October 1st, 2006. At each location, seepage rates were repeatedly measured: 4 times per location on September 6, and twice per location on October 1. Positions with higher discharge values tend to stay closer to the trend line.

On both sampling occasions, a high degree of correlation was found between the groundwater discharge rate and the temperature of the riverbed’s surface. Data from both sampling days were plotted together to see if a direct quantitative measure of seepage could be inferred using bed temperature measurements (Figure 3). The combined plot, however, yields a regression value (r²)of 0.47, substantially lower than the regression values for individual dates. On September 6th, the river water was 2.5 °C warmer than on October 1st. This difference in temperature gradient between the river and ground waters produced separate regression curves. It should be possible to eliminate this effect if the temperature of the bed is measured below the effects of diurnal heating and cooling. When this was tried in the field, however, it was determined to be far too time-consuming to be practical at a field scale.

Results and Discussion

Mapping the spatial variability of temperature and discharge

Based on the correlation between seepage rates and bed-surface temperature, riverbed temperatures can be used to map the spatial variability of groundwater discharge. The method only works well during the warmer months of the year in temperate climates when a noticable difference in groundwater and surface water temperature exists. Temperatures were mapped along a 25-meter reach of the Loosahatchie River weekly throughout the low-flow period of the summer and early fall of 2006. A one-meter grid was laid out across the study reach by placing paired numbered landscape stakes on a one meter interval for 25 meters along each bank. Masonry line marked off in one meter increments was then attached to paired numbered stakes across the river. The streambed temperature probe was then used at each meter mark on the masonry line to record bed temperature. A meter stick attached to the temperature probe measured vertical distance from the river bed to the masonry line. Using a datum set two meters below the masonry line, relative elevation of the node was determined. This process was continued at each pair of landscape stakes until every point on the grid had been measured for bed temperature and elevation. Each measurement point required two seconds to stabilize, with the entire grid requiring approximately two hours to measure. This enabled us to remain comfortably within the window from 1:00 PM to 4:00 PM when river water temperatures are stable.

Upon analyzing variograms of the data sets, a linear Kriging model was used to contour maps for weekly bed temperature and elevation using temperature as a proxy for seepage rate. The same kriging model was used for all weekly measurements. The kriging algorithm was chosen to give a weighted distance value between points, preventing small, sub meter scale, features from appearing as diffuse as other algorithms may contour them.

Figure 4 – Bed temperature deviation and elevation maps of the study area for July 20th through August 3rd, 2006. Temperature displayed is the difference in stream-bed and -water temperature in degrees Celsius. Bed elevations are displayed in height (meters) above a reference datum set two meters below the grid constructed with landscape stakes and masonry line. Measurement locations are designated with a (.)

Figure 5 – Bed temperature-deviation and elevation maps of the study area for August 10th through August 24th, 2006. Temperature displayed is the difference in stream-bed and -water temperatures in degrees Celsius. Bed elevations are displayed in height (meters) above a reference datum set two meters below the grid constructed with landscape stakes and masonry line. Measurement locations are designated with a (.)

Spatial distribution of groundwater discharge

The contour maps show that significant seepage consistently occurs in well-defined foci or springs. Two persistant spring locations were located 2 to 4 meters from the bank in the northeast and southeast corners of the study grid that continued to discharge throughout the duration of the study (designated as Spring A and Spring B in Figures 4 and 5). The Spring A continued to discharge on most measurement days, whereas discharge from Spring B was more intermittent. Groundwater was discharging at such a rate in some areas that a very noticeable difference in temperature of more than 1.8 degrees Celsius could be felt upon stepping barefooted on the riverbed.

The absolute position of each spring tended to wander within a two-meter area, possibly as a result of the amount of sand cover derived from the migrating sand bars. These sand bars could add as much as one-half meter of additional sand cover as they migrated downstream at a rate of approximately one-meter per day. One such sand bar had migrated over Spring A following a heavy rainstorm August 11th, as shown in Map 5-B (Fig. 5). Increased pore pressure supplied by the spring likely increased the localized entrainment rate (Simon et al., 2000), creating a depression in the bar (Map 5-B; Fig. 5) until the spring's bed elevation returned to pre-bar elevation on August 24th (Map 6-B; Fig. 5). Very loose sediment with near quick sand conditions were observed in these depressions. A similar sand bar migrated over Spring B in Map 2-B (Fig. 4), but lower discharge rates at this position were not as effective at aiding in removal of bar sediment. The spring bed elevation in this location required more than twice the amount of time to return to pre-bar elevation (Map 6-B; Fig. 5).

Unfocused bank discharge rates also seem to be tied to the presence of migrating sand bars. On July 20th (Map 1-A), no significant sand bars had entered the grid and discharge appeared to be relatively equal along both banks. Over the next two weeks, discharge from the south bank steadily decreased as a large sand bar covered the southern half of the grid (Map 2-A,B and 3-A,B). Discharge from the north bank increased as the bar was washed away (Map 5-A,B and 6-A,B). The bank discharge does not appear to be directly linked with the spring discharge patterns. At times, a bank may show relatively high discharg while a nearby spring is not (Map 5-A, Figure 5, Spring B and South Bank). While at other times the spring showed a high discharge while there was very little discharge at the nearby bank (eg. Map 4-A, Figure 5, Spring B and South Bank). Conditions were also observed when both the spring and nearby bank were both exhibiting high discharge (Map 3-A, Figure 4-A, Spring B and South Bank) or both barely discharging at all (Map 6-A, Figure 5, Spring B and South Bank). The disparity between these observations lends evidence that the banks and the springs are separate discharge regimes.

Discharge in mid-river was initially focused along lineations running perpendicular to the stream-flow direction. Although originally suspected to be an artifact of the measurement method, the lineations were also present in subsequent maps at the same locations. Direction of groundwater flow in the mid river lineations tended to alternate independantly of the bank or spring discharge locations. Discharging mid-river areas in Map 1-A become recharging areas in Map 2-A. Discharge position and orientation remained stable from July 27th through August 10th with steadily decreasing rates of groundwater seepage. After a large rainfall event on August 11th caused elevated river stage, mid-river discharge areas inverted in Map 5-A (Fig. 5) but immediately reverted a week later in Map 6-A. The lineations are most likely controlled by underlying geology. Approximately 30 meters downstream, a thin, broad, clay layer was noticed approximately one-half meter below the normal bed level. The clay unit was covered by sand within a week and not encountered again. Linear joints along the clay unit may permit water from the underlying aquifer to exchange with the river.

The discontinuity in relative discharge rate between the bank discharge and mid-river discharge may indicate two separate groundwater flow-regimes, with spring representing a third possible flow regime or simply the focusing of the local of regional regimes. Bank discharge position remains constant with a fairly steady estimated discharge rate, whereas springs do not follow the same discharge trends as either the bank or mid-river discharge processes. This may be due to a piping process in which springs derive their water from a greater distance(Toth 1971; Kolb, 1976). Storey et al., (2003) found that the hydraulic geometry of many gaining rivers systems results in groundwater seepage focused near the edges of the stream. Their numerical models of a single riffle in a gravel-bed stream indicated groundwater seepage rates to be more variable and twice as high near the stream banks compared to near the center of the stream. We, however, observed that discharge locations mid-river fluctuate to a greater degree. As the regional water table falls during the low-flow season, discharging mid-river areas and bank locations steadily decrease in discharge volume until precipitation raises the water table enough to promote additional discharge. Short rainfall events appear to only promote discharge for a very limited amount of time.

While relative discharge rate of bank areas and mid-river lineations are not consistant, highly discharging bank areas do line up with the discharging mid-river locations in all maps. Discharge rates of bank and mid-river areas also fluctuate to a greater degree than springs and are much more dependant on the local flow regime and local precipitation events, as can be seen in Maps 4-A, 5-A, and 6-A (Fig. 5). Bank and mid-river areas are most likely connected in the shallow local flow regime with primary discharge through the banks. As the water table is raised, pressure forces more groundwater beneath the river sediment to discharge mid-river along lineations. More stable spring discharge rates may indicate groundwater derived from a deeper regional source made possible by any number of confining unit windows that may be present in the Mississippi Embayment (Renken, 1998).

Conclusions:

Bed-surface temperature can be used to determine relative rate of groundwater discharge seepage in a gaining stream. A direct quantitative measure of seepage rate can made using bed temperature as long as the correlation of seepage rate and river water temperature has been previously performed. This has the potential to save a substantial amount of time in measuring point discharge rates. Seepage meters can require days per measurement point compared to the two seconds required to make a bed-temperature reading using our method. The cost of performing the bed surface temperature method is also very inexpensive when considering time saved and cost of equipment.

We show that seepage along a reach of a homogenous sandy riverbed is not diffuse, but rather occurs in discharging springs which retain a fairly fixed location, transient linear regions in mid-river, and along banks. The presence of springs also increases localized entrainment of bed material, creating low points in the riverbed. Determining a trend in discharge locations may aid in understanding the geomorphic processes at work in rivers with unconsolidated beds. Application of this method could also be very useful in determining specific discharge locations in karst or fractured rock settings where dye tests are not feasable.

References

• Arthur, J. K. and Taylor, R. E., 1998. Ground-water flow analysis of the Mississippi Embayment aquifer system, South-central United States, USGS Professional Paper 1416-I.

• Becker, M. W., Georgian, T., Ambrose, H., Siniscalchi, J., Fredrick, K., 2004. Estimating flow and flux of ground water discharge using water temperature and velocity, Journal of Hydrology, 296, 221–233.

• Belanger, T. V. and Montgomery, M. T. 1992. Seepage meter errors, Limnology and Oceanography, 37(8), 1787-1785.

• Kasahara, T., and Wondzell, S.M., 2003. Geomorphic controls on hyporheic exchange flow in mountain streams, Water Resources Research, 39 (1).

• Knighton, David, 1998. Fluvial Forms and Processes, Oxford University Press, New York.

• Kolb, C. 1976. Geologic Control of Sand Boils Along Mississippi River Levees. Army Corp of Engineers Report, 1-34.

• Packman A.I., and MacKay J.S., 2003. Interplay of stream-subsurface exchange, clay particle deposition, and streambed evolution, Water Resources Research 39 (4), p. ESG41-ESG49.

• Paulsen, R. J. Smith, C. F. O Rourke, D. Wong, 2001. Development and Evaluation of an Ultrasonic Ground Water Seepage Meter, Ground Water, 39(6), 904-911.

• Renken, R. A. 1998. Groundwater Atlas of the United States, USGS HA 730-F.

• Simon, Andrew, Curini, Andrea, Darby, Stephen E., Langendoen, Eddy J., 2000. Bank and near-bank processes in and incised channel, Geomophology 35, 193-217.

• Storey R.G., Howard K.W.F., and Williams D.D., 2003, Factors controlling riffle-scale hyporheic exchange flows and their seasonal changes in a gaining stream: A three-dimensional groundwater flow model, Water Resources Research 39 (2), p. SBH81-SBH817.

• Toth, Jozsef, 1971. Groundwater Discharge: A common generator of diverse geologic and morphologic phenomenon. International Association of Scientific Hydrology Bulletin 16(1-3):17-24.

• Torgersen, Christian E., Faux, Russel N., McIntosh, Bruce A., Poage, Nathan J., Norton, Douglas J., 2001. Airborne thermal remote sensing for water temperature assessment in rivers and streams, Remote Sensing of Environment, 76, 386–398.

• Urbano, Waldron, Larsen, and Shook, 2005. Groundwater-surfacewater interactions at the transition of an aquifer from unconfined to confined, Journal of Hydrology, 321(1-4), 200-212.

• Wells, S. A. and Annear, R. L. 2002. Temperature Effects of Streambed Heating, Water Resources Research Institute Annual Technical Report 2002, p. 3–10.

Bibliography

• Arthur, J. K. and Taylor, R. E., 1998. Ground-water flow analysis of the Mississippi Embayment aquifer system, South-central United States, USGS Professional Paper 1416-I.
• Becker, M. W., Georgian, T., Ambrose, H., Siniscalchi, J., Fredrick, K., 2004. Estimating flow and flux of ground water discharge using water temperature and velocity, Journal of Hydrology 296, 221 – 233.
• Belanger, T. V. and Montgomery, M. T. 1992. Seepage meter errors, Limnology and Oceanography 37(8), 1787-1785.
• Cleland, Carol E., 2001. Historical science, experimental science, and the scientific method, Geology 29, 987 – 990.
• Einstein, H. A. and Shen, H. W., 1964. A study on meandering in straight alluvial channels, Journal of Geophysical Research 69, 5239.
• Ferguson, R. I. 1979. River meanders: regular or random? In Wrigley, N. (.ed), Statistical Applications in the Spatial Sciences. London: Pion, 229-241.
• Ferguson, R. I. 1984. Kinematic model of meander migration. In Elliot, C. M. (.ed), River Meandering. New Orleans: American Society of Civil Engineers, 942-951.
• Hantush, Hohamed M., Modeling stream-aquifer interactions with linear response functions. Journal of Hydrology, 311, 2005, pgs. 59 – 79.

• Howard, A.D., 1992. Modeling channel migration and floodplain sedimentation in meandering streams. In Carling, P.A. and Petts, G.E. (eds), Lowland floodplain rivers. Cichester: Wiley, 1 -41.
• Hudson, Paul F., Kesel, Richard H., 2000. Channel Migration and meander bend curvature in the lower Mississippi River prior to human modification, Geology 28, 531 – 534.
• Jackson II, Roscoe G., 1975. Velocity-bed-form-texture patterns of meander bends in the lower Wabash River of Illinois and Indiana, Geologic Society of America Bulletin 86, 1511 – 1522.
• Keller, E. A. 1972. Development of alluvial stream channels: a five stage model, Bulletin of the Geological Society of America 83, 1531-1536.
• Knighton, David, 1998. Fluvial Forms and Processes, Oxford University Press, New York.
• LaFleur, Robert G., 1999. Geomorphic aspects of groundwater flow, Hydrology Journal 7, 78 – 93.
• Larkin, Randell G., Sharp, John M., 1992. On the relationship between river-basin geomorphology, aquifer hydraulics, and ground water flow direction in alluvial aquifers, Geologic Society of America Bulletin 104, 1608 – 1620.
• Lewin, John, 1976. Initiation of bed forms and meanders in coarse-grained sediment, Geologic Society of America Bulletin 87, 281 – 285.
• Malin, Micheal C., Carr, Micheal H., 1999. Groundwater formation of martian valleys, Nature 397, 589 – 591.
• Markham, A. J. and Thorne, C. R. 1992. Geomorphology of gravel-bed river bends. In Billi, P., Hey, R. D., Thorne, C. R. and Tacconi, P. (eds), Dynamics of gravel bed rivers. Chichester: Wiley, 433-450.
• Paulsen, R. J. Smith, C. F. O Rourke, D. Wong, 2001. Development and Evaluation of an Ultrasonic Ground Water Seepage Meter, Ground Water, VOL 39; PART 6, 904-911.
• Pederson, Darryll T., 2001. Stream Piracy Revisited: A Ground water Sapping Solution, GSA Today September 2001, 4 – 10.
• Renken, R. A. 1998. Groundwater Atlas of the United States, USGS HA 730-F.
• Simon, Andrew, Curini, Andrea, Darby, Stephen E., Langendoen, Eddy J., 2000. Bank and near-bank processes in and incised channel, Geomophology 35, 193-217.
• Smith, Charles E, 1998. Modeling High sinuosity meanders in a small flume, Geomoprphology 25, 19 - 30.
• Thompson, A. 1986. Secondary flows and the pool-riffle unit: a case study of the processes of meander development, Earth Surface Processes and Landforms 11, 631-641.
• Torgersen, Christian E., Faux, Russel N., McIntosh, Bruce A., Poage, Nathan J., Norton, Douglas J., 2001. Airborne thermal remote sensing for water temperature assessment in rivers and streams, Remote Sensing of Environment 76, 386 – 398.
• Toth, Jozsef, 1999. Groundwater as a geologic agent: An overview of the causes, processes, and manifestations, Hydrogeology Journal 7, 1 – 14.
• Urbano, Waldron, Larsen, and Shook, 2005. Groundwater-surfacewater interactions at the transition of an aquifer from unconfined to confined, Journal of Hydrology, accepted August 5, 2005, 1-13.
• Van Arsdale, Roy, Waldron, Brian, Ramsey, Natasha, Parrish, Shane, Yates, Rhonda, 2003. Impact of River Channelization on Seismic Risk: Shelby County, Tennessee, Natural Hazards Review, February 2003, 2 – 11.
• Yalin, M. S. 1971. Of the formation of dunes and meanders, Precedings of the 14th International Congress of the International Association for Hydrualic Research 3, Paper C13, 1-8.
• Wells, S. A. and Annear, R. L. 2002. Report for 2002OR2B: Temperature Effects of Streambed Heating, Water Resources Research institute Annual Technical Report 2002, pgs 3 – 10.