1

PROJECT COMPLETION REPORT INVESTIGATING THE ROLE OF SURFACE-GROUNDWATER INTERACTIONS ON WATER QUALITY Principal Investigators T.P. Clement^, Associate Professor J.G. Melville, Professor R.R. Goswami, PhD Candidate M. Hogan, M.S. Candidate Che-An Kuo, M.Eng. Candidate M.C.M. Kazezyılmaz-Alhan* Department of Civil Engineering 238 Harbert Engineering Center, Auburn University, Auburn, AL 36849 *Currently at Bogazici University, Istanbul, Turkey. ^Telephone: 334-844 6268; Fax: 334-844 6290; E-mail: clement@auburn.edu Sponsor: USGS-Alabama Water Resources Research Institute June 9, 2007

2

1. PROBLEM STATEMENT AND OBJECTIVES In the state of Alabama, most of the population obtain their water supply from a combination of surface water and groundwater sources. Therefore, it is important to understand and solve the problems related to groundwater-surfacewater interactions. In recent years, there has been a growing interest in understanding the mechanisms involved in surfacewater/groundwater interactions since researchers have established that these interactions play a crucial role in controlling the behavior of contaminant transport in both systems. Surface/groundwater interactions can influence the downstream water quality significantly since the concentration distribution in both systems (stream or groundwater) can change due to the biogeochemical reactions that occur between the minerals/nutrients in the subsurface and the stream. In this project, we completed both theoretical and experimental investigations to quantify the contaminant exchange patterns between surfacewater and groundwater. The theoretical analysis was designed to measure the lateral path of a contaminant plume as it interacts between the surfacewater and groundwater along the flow path. Laboratory-scale experiments were completed to demonstrate the existence of a vertical transport pathway for a contaminant plume in the stream as it interacts with the subsurface water beneath it. Recent natural disasters such as Katrina and the Asian-tsunami have created a need for understanding saline surfacewater transport into unconfined groundwater systems after a large inundation event. Therefore, in addition to studying the traditional stream-groundwater interaction pathways, we also explored the interactions between coastal surfacewater and groundwater during a hurricane-induced flooding event. Hurricane surge waves and tsunamis have the potential to inundate large areas of coastal aquifer region with saline water. Estimating the impacts of these flood waters on groundwater quality is a part of any disaster management program. Our second objective was to conduct multiple experiments to quantify the contaminant transport processes and associated times scales within a coastal unconfined aquifer after an inundation event. Controlled experiments were completed by discharging a surge wave over an unconfined aquifer. Observations were made to study the fate of the dense surfacewater as it migrated across the water table boundary. 2. RESEARCH METHODOLOGY AND RESULTS 2.1. Experimental setup to study groundwater stream interactions To observe the vertical flow between surface water and hyporheic zone which lies under the streambed, we created a straight stream by using the small scale plexi-glass flume. We filled the plexi-glass flume with uniform silica sand to represent the porous media and develop a straight channel within the material with a constant depth and cross-sectional area to simulate a stream segment, as shown in the schematic (Figure 1). In this setup, we first maintained a constant flow rate at upstream and then injected a dye at the inlet of the channel. We observed the vertical path of the dye between the stream and the streambed in addition to the longitudinal paths in the stream and in the porous media.. In addition, we will inject a tracer plume into the porous media by selecting a dry streambed as the initial condition and then maintain a constant flow rate at the upstream end. We

3

observed the path of the tracer both vertically between stream and porous media and longitudinally along stream and porous media. All of these experiments were completed to simply to demonstrate the vertical exchange processes which were hypothesized in our theoretical model. 2.2. General experimental setup to study groundwater and coastal water interactions Saltwater flooding experiments were conducted in a rectangular flow tank. The tank was constructed using 6 mm thick Plexiglass™. The internal dimensions of the porous media region are: 53 cm (length) × 2.7 cm (width) × 30.5 cm (height). A conceptual diagram of the experimental setup is shown in Figure 2. The flow tank was divided into three distinct chambers: a central flow chamber containing the porous medium, and two constant head chambers containing salt and fresh waters. The constant head chambers are 5 cm long and are separated from the porous media chamber by two fine screens of mesh size US # 16. Uniform silica beads of average diameter 1.1 mm, obtained from Potter Industries Inc., were used as the porous medium. Self-adhesive measurement tapes were pasted on the sides and at the bottom of the tank to allow direct measurements. The experiments were recorded using Nikon Coolpix digital camera in a high resolution mode. The digital data allowed us to zoom and observe small-scale variations occurring at the millimeter scale.

2.3. Experimental setup to study saltwater discharged from surface-water storage ponds A conceptual diagram of the system studied in this effort is show in Figure 2. This system was used to study the fate and transport of salty plume evolving from a surfacewater pond. The flow tank and the porous medium used in this study were similar to the ones used in the above experiment. As shown in the figure the saltwater was supplied as a continuous point source to the surface of the porous media through a continuous volume displacement peristaltic pump at a constant rate (QL=8.9 mL/min). The flow rate was sufficiently low to prevent any overflow effects. The system was used to simulate a leaking saltwater pond. Three experiments were performed to study the effects of different ambient groundwater flow rates. The rates were controlled by varying

4

the hydraulic gradient in the flow direction by adjusting the water levels in the end reservoirs. Digital data was collected with a 7.1 mega-pixel camera. A small amount of red food coloring was added to the saltwater to differentiate the salty surfacewater from the fresh ground water.

Figure 2 – Schematic of the experimental setup used to study saltwater contaminant infiltrating from a surfacewater storage pond 2.4. Experimental setup to study saltwater discharged from beach-face direct infiltration A conceptual diagram of the system studied in this effort is show in Figure 3. This experimental setup was used to study the fate and transport of a large volume of saltwater dumped over a freshwater aquifer after a large-scale inundation event occurred due to a hurricane or a tsunami wave. The flow container described above was also used for the beach face infiltration experiment. However, for this experiment, the left freshwater boundary was replaced with a saltwater boundary, to represent the coastal interface, by injecting saltwater at a constant rate from a saltwater supply reservoir as illustrated in Figure 3. The saltwater from the boundary was allowed to intrude into the aquifer from the left to form a stable saltwater wedge. The steady initial condition was attained when the saltwater wedge reached a steady position and the lateral movement of the wedge ceased to penetrate further into the porous medium. Furthermore, this initial steady condition, representing the pre-innundation scenario was attained by setting up a regional hydraulic gradient across the length of the model that would force the uncolored freshwater to flow from right to left. The inundation caused by the tsunami wave was simulated by instantaneously discharging approximately 500 mL of saltwater at the top surface of the model. The freshwater flow was continuously supplied to the model from the constant head boundary at the right side. To avoid any variation in density or chemical composition, the same batch of saltwater was used in a particular set of experiments to supply both the saltwater boundary in the left reservoir and the tsunami water from above. The experiments were repeated for different ambient groundwater flow rates.

5

Figure 3 – Schematic of the experimental setup used to study saltwater contamination after large-scale inundation event in a coastal aquifer 3.0 RESULTS AND DISCUSSIONS 3.1 Analysis of stream/GW interactions These are analysis notes describing an open channel – groundwater interaction mechanism. They are preliminary ideas which may help describe, in terms of classical groundwater hydraulics and open channel hydraulics, experimentally observed interaction flows between the groundwater and surface water. For small vertical flow velocities it can be assumed that the variation of y(x) in the surface water defines the head variation in the ground water. Thus, Darcy’s law for the groundwater flow and the gradually varied flow equation for open channel flow can be expressed in terms of y(x)

dh dyq K Kdx dx

= − = − , Darcy’s law.

0 f

2dy S Sdx 1 F

−=

−, the gradually varied flow equation for open channel flow.

vFgy

= = the Froude number.

6

Figure 4. Surface and Ground Water Interaction: y(x) = surface water depth, b = subsurface aquifer thickness, v(x) = surface water velocity, q(x) = subsurface Darcy veloclity, vi(x) = surface transfer velocity, K = aquifer hydraulic conductivity, n = Mannings roughness coefficient for the open channel. For open channel, steady state flow, the depth variation, y = y(x) is nonlinear. Darcy’s law for the subsurface flow indicates that q = constant if the y(x) is linear. Thus, q is not constant which implies there must be some interface velocity between the surface water and the groundwater. Preliminary analysis indicates that this interface velocity is small and it can be either positive or negative describing flow from or into the ground water formation. Preliminary laboratory experiments have shown vertical flow velocity in porous media below the free surface flows. Enhanced physical-chemical modeling of contaminant transport between the surface and ground water depends on further quantification of laboratory and analytical descriptions of the scale of this interface transport velocity.

y(x) v(x)

q(x) b

x x+Δx

vi(x)

K

n(roughness)

7

Figure 5 Laboratory data indicating vertical velocities induced by surfacewater flow For subcritical flow F < 1, and the further constraint that v << (gy)0.5 , then 1-F2 →1. This limiting case then can be analyzed to obtain an approximate method to explain a mechanism of surface water (open channel flow) and ground water flow under the channel bed. 1st case: For horizontal channel: (S0 = 0)

fdy Sdx

≈ −

From Mannings equation for energy dissipation ,

2 / 3 1/ 2f

1.49v R Sn

= where AR yP

= ≈ for a wide channel.

The energy dissipation in open channel flow can be approximated as

10 / 3fS y−= α

2qn1.49 α =

10 / 3dy y

dx−= −α

which can be solved with the boundary condition, @ x = L y = a

8

(1) ( )3/13

13/ 313y L x a3

= α − +

Considering conservation of volume and interface flow from the channel to the ground water, ( ) ( )iq x b v x q x x b+ ∆ = + ∆

or, 2

i 2dq d yv b bKdx dx

= = − (2)

From (1) and (2):

( )

2

i 23/1313/ 3

10 bK3v

13 L x a3

α =

α − +

2

i 23/ 3

10bK3v

y

α =

which describes an approximation of flow from the channel to the groundwater. 2nd case:

Assume: 0 fdy S Sdx

≈ − .

Then, 10 / 30

dy S ydx

−= − α

( )2

13/ 3 23/ 302

d y 10 S y ydx 3

− −= α − α

Substitution into equation (2) then gives an approximation for the interface velocity,

( )13/ 3 10 / 3i 0

10v bK y S y3

− −= − α − α .

9

This result shows that interface velocity may be positive or negative, that is into the aquifer or from the aquifer. It is true that for typical values of the physical parameters these calculated interface velocities are very small. Acting over large areas and over long time periods, the interface velocity could contribute to surface and ground water quality. Transient effects on the interface velocity are more significant. Water depths in the channels can change due to changes in hydrologic conditions in the watersheds and can change due to variable downstream depth controls. Downstream control depths change rapidly due to hurricane surge waves and tsunamis. These types of depth changes are also accompanied with the influence of density differences between the fresh water and sea water which can augment the interface flow. 3.2 Surface water storage pond discharge experiments This section provides a summary of the data collected using the experimental conditions shown in Figure-2. It is assumed that the contaminant source, i.e. flooded lagoon or inundated region, is located at a distance inland such that the effects of the seawater boundary can be ignored. A summary of the hydraulic conditions used in these experiments are given in Table 1.

Table 1: Summary of Sinking Plume Experiment

Experiment ID Loading Rate Hf Hs Head Gradient Rin Rout Lout Flux ErrormL/min cm cm % mL/sec mL/sec mL/sec %

A-1 8.9 28.9 27.0 3.58 4.22 0.56 3.52 3.32A-2 8.9 28.9 27.2 3.21 4.11 0.97 3.15 0.24A-3 8.9 29.0 28.0 1.89 4.11 2.11 1.95 1.22A-4 8.9 29.0 28.4 1.13 4.11 2.62 1.43 1.46A-5 8.9 29.0 28.3 1.32 4.11 2.75 1.25 2.68A-6 8.9 29.0 28.8 0.38 4.01 3.50 0.52 0.25

Sinking Plume Experiment

The experiments are presented in the order of decreasing plume stability. The A-1 experiment had the largest head gradient and therefore was the most stable. The only experiment in which the head gradient did not correlate with plume stability was experiment A-5 in which the experiment was observed to have had a higher head gradient than A-4 but was found to be less stable. This could be attributed to human error of reading the left and right head levels. Figure 5.1 shows the result of experiment A-1 which had a very high head gradient. As the figure shows, the plume behavior appears to be stable since the body of the plume remains concentrated in the upper most region of the model. This means that the forced convection due to the flow is high compared to the free downward convection due to density differences. The contaminant moves relatively quickly through the model compared to the experiments with a lower head gradient.

Experimental Results

10

Stable A-1

T = 1 minute

T = 3 min

T = 5 min

T = 7 min

T = 9 min

T = 11 min

T = 13 min

T = 15 min

11

T = 17 min

T = 19 min

T = 26 min

T = 31 min

Figure 5.1: A-1 Stable plume experiment

Experiment A-2 had only a slightly smaller head gradient than experiment A-1 therefore the plume profiles are similar. As expected, the A-2 plume penetrates deeper into the model than the A-1 plume. The plume still exhibits a stable behavior because fingering and instabilities were not observed along the interface.

T = 1 min

T = 3 min

12

T = 5 min

T = 7 min

T = 9 min

T = 12 min

T = 14 min

T = 16 min

T = 18 min

T = 20 min

13

T = 25 min

T = 30 min

Figure 5.2: A-2 Stable Plume Experiment

The gradient was reduced considerably in experiment A-3. Instabilities and fingering are on the verge of forming which is why this problem can be considered an intermediate case. In fact, this experiment was useful for determining the critical π1 number for our system. This was because the experiment represented a border case in which any decrease in head gradient would have caused noticeable fingering to develop. Another observation, was the length of time it took the plume to reach the left exit boundary was considerably less than the two prior experiments A-1 and A-2.

Stable A-3

T = 1 min

T = 3 min

T = 5 min

T = 7 min

14

T = 9 min

T = 11 min

T = 13 min

T = 15 min

T = 17 min

T = 19 min

T = 24 min

T = 29 min

Figure 5.3: A-3 Stable Plume Experiment

15

Experiment A-4 was the first of three unstable plume problems. Clearly, instability and fingering are generated along the interface. The fingers appear to have a similar shape but are randomly spaced. It appears that the size of the fingers tend to grow as the distance from the source is increased.

Unstable A-4

T = 1 min

T = 3 min

T = 5 min

T = 7 min

T = 9 min

T = 12 min

16

T = 14 min

T = 16 min

T = 18 min

T = 20 min

T = 25 min

T = 30 min

Figure 5.4: A-4 Unstable plume experiment

The head gradient was reduced further for experiment A-5. There was more fingering evident in Experiment A-5 due to the fact that the fingers were more closely spaced in Experiment A-5 than in A-4. The free convective force due to density now appears to be a major factor in the plume’s behavior.

Unstable A-5

17

T = 1 min

T = 3 min

T = 5 min

T = 7 min

T = 9 min

T = 12 min

T = 14 min

T = 16 min

18

T = 18 min

T = 20 min

T = 25 min

T = 30 min

Figure 5.5: A-5 Unstable plume experiment

The head gradient was greatly reduced for the final experiment, A-6. This system is a free convection dominated problem. The behavior of the plume is highly unstable which is evident in the photographs shown in Figure 5.8. The plume sinks vertically downward in the shape of one big finger and accumulates on the bottom of the tank. This degree of instability represents the least desirable to encounter in the field environment as this plume has the potential to contaminate larger regions of an aquifer than the previous experiments, A-1 through A-5.

A-6 (Highly Unstable)

T = 1 min

19

T = 3 min

T = 5 min

T = 7 min

T = 9 min

T = 12 min

T = 14 min

T = 16 min

T = 18 min

20

T = 20 min

T = 24 min

T = 28 min

Figure 5.6: A-6 Unstable plume

T = 35 min

T = 40 min

21

3.3 Beach Face Infiltration Experiments The experimental data for the beach face infiltration experiment discussed below represent an Katrina type surfacewater inundation event, shown in Figure 3. As discussed in the methods section, the surfacewater source was created by instantaneously distributing a fixed volume of saltwater directly on the top boundary of the model. Experiments were performed in a high and low flow system. The dataset shown in Figure 6 shows the transport of this dense saltwater slug over a time period of 11 minutes. The saltwater wedge intruded considerably less in the high flow experiment than in the low flow experiment. Under high flow conditions the saltwater slug was transported quickly through the model and formed a rather shallow contamination zone. On the other hand, when the ambient flow was low, the slug of saltwater migrated downward and contaminated larger volume of aquifer. There was also more intense fingering in the low flow case.

An important observation made in these experiments was that inundated sea water (red-colored salt water) always remained above the regional seawater interface. This should be expected because the density of the tsunami water sources decreased due to dilution as they migrated through the aquifer and mixed with freshwater. Therefore, the tsunami water would never be able to penetrate the regional seawater wedge. Instead, it will remain floating above the wedge and will eventually be advected by the regional freshwater flow along the wedge.

Figure 6: Beach face infiltration in a high GW flow (left photographs) and low GW flow conditions (right photographs)

4.0 SUMMARY The behavior of contaminant transport in streams is greatly influenced by the processes that exchange fluid between surface and ground waters. We derived an analytical approach to

22

model the vertical exchange between a surfacewater body and the stream bed. The exchange process was also demonstrated through a conceptual experimental study. In the second phase of this work, we studied the interactions between groundwater and coastal waters. The recent natural disasters such as Katrina and the Asian-tsunami have created the need for understanding the saline water transport processes within an unconfined groundwater system, after a large-scale, surfacewater inundation event. We provide several detailed data sets to illustrate the interactions between a dense surfacewater body (e.g., ocean) with a local unconfined aquifer. The first type of experiment involves the study of saline water migration from surfacewater storage ponds located along coastal dunes. The second type of experiment involves simulation of saline water migration patterns into a local unconfined aquifer after a large-scale flooding event. This grant partially supported the research efforts of three graduate students. 5. PUBLICATIONS 1) Goswami, R.R. and T.P Clement, Laboratory-scale investigation of saltwater intrusion dynamics, Water Resources Research, Vol. 43, W04418, doi:10.1029/2006WR005151, 2007. 2) Hogan, M.B., R. R. Goswami, K. G. Villholth, T. H. Illangasekare, and T. P. Clement, Understanding the Flow and Mixing Dynamics of Saline Water Discharged into Coastal Freshwater Aquifers, Proceeding of the SWIM/SWICA joint meeting held in Sardinia, Italy, September 25th-29th, 2006. 3) Clement, T.P., R.R. Goswami, M. Hogan, Understanding the dynamics of freshwater and saltwater mixing processes in unconfined aquifers – laboratory scale model results, Proceedings of the international conference on MODFLOW and More 2006 Managing Ground Water Systems, Golden Colorado, May 2006, p. 16-17. 4) Hogan, M.B., Understanding the flow and mixing dynamics of saline water discharged to coastal aquifers, MS Civil Engineering Thesis, submitted to Auburn University. 6. LIST OF STUDENTS Rohit Goswami, PhD candidate Matthew Hogan, MS Civil Engineering (Graduated in Spring 2007) Che-An Kuo, MEng. Civil Engineering (Graduated in Summer 2007) Michal Murdock, BS Civil Engineering (Graduated in Spring 2007) 7. ACKNOWLEDGEMENTS Dr. M.C.M. Kazezyılmaz-Alhan was the original PI of this proposal. Dr Alhan resigned her position last Fall and transferred this project to Drs Clement and Melville. This project benefited from the discussions, about Katrina-related events, we have had with the USGS-Alabama office. We acknowledge the support provided by Ms. A. Clark, Director of USGS-Alabama.