Embed Size (px)
Citation preview
USING 3-D MODELING TO DESCRIBE THE RELATIONSHIP BETWEEN PEAK
STAGE, STORM DURATION AND BANK STORAGE AND THE IMPLICATIONS
ALONG A MEANDERING STREAM IN CENTRAL ILLINOIS
Lucas P. Chabela
57 Pages
Groundwater and surface water are often studied as different system; however,
one commonly affects the other through various scenarios. Bank storage is an important
hydrogeological process, which involves the temporary storage and release of stream
water in adjacent aquifers. Bank storage can contribute a considerable amount of
discharge to a river and can be an important process for movement of contamination
downstream. Several studies document the effects of increasing stage and increasing
storm duration; however, these controls are often separated. This project examines peak
stage and storm duration to determine which factor is statistically more influential on the
bank-storage process. The study focuses on a small reach of Little Kickapoo Creek, a
third-order, perennial stream located in central Illinois. A 3-D, transient-state numerical
model (MODFLOW) was created of the study site, and 36 simulations were ran at
various peak stages and storm durations.
Results illustrate that peak stage and storm durations, while both influential, affect
different areas of the bank-storage process. However, peak stage is statistically more
influential in controlling the maximum volume of bank storage (~3.6x) that occurs in the
system and of the storage that remains in the system at 100 hours (~1.1x). While
statistically less influential in determining storage remaining at 100 hours, increasing
storm duration appears to be more effective on the retention of bank storage. Parafluvial
exchange is important for evaluating bank storage along a meandering stream, suggesting
that at least 2-D, ideally 3-D, models are used to evaluate bank storage. Lastly, this study
recognizes that bank storage is the mechanism for road salt storage along the stream
during winter months and the subsequent flushing of the chlorides back to the stream
during the summer months.
KEYWORDS: Bank Storage, storm duration, peak stage, groundwater modeling,
MODFLOW, meandering streams, chloride storage
USING 3-D MODELING TO DESCRIBE THE RELATIONSHIP BETWEEN PEAK
STAGE, STORM DURATION AND BANK STORAGE AND THE IMPLICATIONS
ALONG A MEANDERING STREAM IN CENTRAL ILLINOIS
LUCAS P. CHABELA
A Thesis Submitted in Partial Fulfillment of the Requirements
for the Degree of
MASTER OF SCIENCE
Department of Geography-Geology
ILLINOIS STATE UNIVERSITY
2017
© 2017 Lucas P. Chabela
USING 3-D MODELING TO DESCRIBE THE RELATIONSHIP BETWEEN PEAK
STAGE, STORM DURATION AND BANK STORAGE AND THE IMPLICATIONS
ALONG A MEANDERING STREAM IN CENTRAL ILLINOIS
LUCAS P. CHABELA
COMMITTEE MEMBERS: Eric Peterson, Chair Catherine O’Reilly Johnathan Thayn
i
ACKNOWLEDGMENTS
First, I would like to thank my advisor, Dr. Eric W. Peterson. He guided me
through this thesis from the very beginning and every step of the way. He took the time to
show me field techniques, assisted in conceptual and numerical model development and
provided papers and outlets to find needed information. Most important he taught me the
time and effort need to complete a research project.
I would also like to thank my committee members. Dr. Catherine O’Reilly taught
me the foundation of every project begins with a research question. She was a critical part
of this thesis as she helped me understand how begin a research project. Dr. Thayn
provided valuable insight into the statistical aspect of this paper and understanding the
basics of linear modeling.
Finally, I would like to thank my fellow graduate students who helped me along
the way. A substantial portion of this project came from countless hours spent at research
meetings, study sessions, and just bouncing ideas off my classmates. They helped shaped
my proposal through the research seminar class as well. I also appreciate the students,
graduate and undergraduate, who assisted in data collection and sample analysis used in
this study.
L. P. C.
ii
CONTENTS
Page
ACKNOWLEDGMENTS i
CONTENTS ii
TABLES iv
FIGURES v
CHAPTER
I. INTRODUCTION AND OBJECTIVES 1
Introduction 1 Research Objectives 6 Research Questions 6 II. STUDY SITE AND METHODS 7 Study Site: Little Kickapoo Creek 7 Field Data 11 Conceptual Model 13 Numerical Model 15 III. RESULTS 20 Model Calibration 20 Peak stage Results 22 Storm duration Results 28 Multiple Linear Regression Analysis 35 LKC Chloride and Stage Results 37
iii
IV. DISCUSSION 41
Peak stage Influences 41 Storm duration Influences 43 Implications 44
V. CONCLUSION 49 Conclusions 49 Further Considerations 50 REFERENCES 52 APPENDIX A: FULL BANK STORAGE DATA SET 56 APPENDIX B: SUPPLETMENTAL 3-D MODELING FILES 57
iv
TABLES
Table Page 1. Initial and calibrated values used in the numerical model 19
2. Percent of storage remaining at 100 hours 30 3. Maximum bank-storage values used for a multiple linear regression 36 4. Storage remaining at 100 hours used for a multiple linear regression 36 5. Multiple linear regression results for predicting maximum bank storage 37 6. Multiple linear regression results for predicting storage at 100 hours 37 7. Groundwater chloride concentrations at LKC 40
v
FIGURES
Figure Page 1. LKC Study Site: Conceptual Model 10 2. Representative storm-event hydrograph of LKC 12 3. LKC Study Site: Well Locations 13 4. Methods: Bank storage and Input Hydrographs 18 5. Calibration Results 21 6. Peak stage: Bank-Storage Results 24 7. Peak stage: Contouring of Modeled Hydraulic Head 25 8. Storm duration: Bank-Storage Results 26 9. Storm duration: Contouring of Modeled Hydraulic Head 27 10. LKC Chloride Concentration and Stage Changes 31 11. LKC Chloride Load and Stage Changes 32 12. Storm duration: Bank-Storage Results 33 13. Storm duration: Contouring of Modeled Hydraulic Head 34 14. LKC Chloride Concentration and Stage Changes 39 15. LKC Chloride Load and Stage Changes 40
1
CHAPTER I
INTRODUCTION AND BACKGROUND
Introduction
Groundwater and surface water are often studied as separate systems; however,
both systems interact in a variety of physiographic and climatic landscapes (Sophocleous,
2002). Along any stream, there is some type of interaction between the stream and the
surrounding groundwater. Three basic types of interaction are: 1) streams gain water
(gaining streams) from the groundwater, 2) streams lose water (losing streams) to the
groundwater and 3) streams are gaining and losing simultaneously along the entire stream
(Winter et al., 1998). Another type of interaction is parafluvial exchange across a
meander bend, which is due to the increase in hydraulic gradient across the bend (Boano
et al., 2006, Peterson and Sickbert, 2006, Cranswick and Cook, 2015). Parafluvial
exchange can be an important source of interaction were meanders are the dominant river
pattern (Boano et al., 2006). Since waters exchange water horizontally and vertically,
flow dynamics and direction are typically three-dimensional (Boulton et al., 1998;
Sophocleous, 2002). Exchange processes from groundwater and surface water are
controlled by the distribution and magnitude of hydraulic conductivities within the
channel and surrounding sediments, the relation of stream stage to groundwater level
(gaining or losing), and the geometry and position of the stream channel within an
2
alluvial flood plain (Wroblicky et al., 1998; Woessner, 2000; Sophocleous, 2002).
Precipitation events can alter hydraulic head, which controls the direction of flow
(Brunke and Gonser, 1997). Flow exchange paths can range in scale from hundreds of
meters with years of transport time to centimeter-long flow paths, which could transport
in only minutes (Harvey et al., 1996). Understanding of the basic principles of
interactions between groundwater and surface water is needed for effective management
of water resources because contamination of one system can commonly affect the other
(Sophocleous, 2002; Winter et al., 1998).
An interaction that occurs in nearly all rivers is the infiltration of water into the
stream banks due to an increase in stream stage (Winter et al., 1998). An increase in stage
can occur from storm precipitation, rapid snowmelt, or the release of water from an
upstream reservoir (Winter et al., 1998). In a gaining stream, an increase in stream stage
temporarily reverses the normal groundwater hydraulic gradient away from the stream
(Squillace, 1996). Water from the stream then infiltrates into the stream bank and even
the streambed, which displaces pre-existing groundwater or hyporheic water and
recharges the adjacent aquifer (Cranswick, 2002; Squillace, 1996; Brunke and Gonser,
1997). As the stage falls and the hydraulic gradient returns to pre-storm stage, some or all
of the infiltrated water returns to the river (Squillace, 1996; Winter et al., 1998). The
process where water is temporarily stored in the stream banks and later returned is called
bank storage. Bank storage only effects the sediment near the stream instead of filling all
3
the pores at water table elevation of the flood. Complete saturation is generally not
achieved due to the short duration of the elevated flow (Kondolf et al., 1997).
Bank storage has important hydrological and ecological implications since it
directly affects the sediments surrounding the stream (Mencio et al., 2014). Multiple
studies document that bank-stored water can contribute a considerable amount of
discharge to the river (Cooper and Rorabaugh; 1963; Kondolf et al., 1987). With this,
return discharge of the bank-stored water can contaminant the surface water (Squillace,
1996). Modeling solute exchange through the aquifer-stream interface, McCallum et al.
(2010) concludes groundwater discharging into a stream following a flood event has the
potential to affect 1) return of bank storage during the event 2) return of bank-stored
water of past flood events and 3) discharge of the regional groundwater. Squillace et al.
(1993) documented the movement of high concentration of atrazine from the river into
the bank sediments, with the return flow providing a source for atrazine during baseflow
conditions (Squillace et al., 1993).
Bank storage, like many hydrologic processes, can vary spatially and temporally
due to the many controls on bank-storage processes. Hydrologic controls include:
transmissivity of an aquifer (Brunke and Gonser, 1997), storage capacity of an aquifer
(Brunke and Gonser, 1997), the presence or absence of an unsaturated zone (Doble, et al.,
2012), bank slope (Doble, et al., 2012; Siergieiev et al., 2015), hydraulic conductivity of
an aquifer (Siergieiev et al., 2015), heterogeneity/homogeneity of an aquifer (Cardenas et
4
al., 2004), sinuosity of a stream (Cardenas et al., 2004; Boano et al., 2006), stream stage
(discharge) (Sjodin et al., 2001; Ha et al., 2008, Siergieiev et al., 2015), and storm
duration (Ha et al., 2008, Siergieiev et al., 2015).
Ultimately, the volume of bank storage depends on the duration, height, and shape
of the hydrograph (Brunke and Gosner, 1997). Several studies document the effects of
increasing the peak stage and the storm duration of the elevated stage on bank storage.
Typically, floods rising to higher stages and lasting for a longer time, allows more stream
water to be available to infiltrate into the bank sediments (Todd, 1955; Chen and Chen,
2003; Ha et al., 2008; Siergieiev et al., 2015). Chen and Chen (2003) examined the
effects of increasing peak stage and storm duration to changes in the storage zone. The
storage zone is defined as the depth of penetration of the stored water into the stream
banks (Chen and Chen, 2003). As peak stage and storm duration increase, the size of the
storage zone increases as well (Chen and Chen, 2003). Storage zone increases means
water is infiltrating further into the stream bank sediments thus delaying the return time
and increasing the residence time (Siegieiev et al., 2015). Siegieiev et al. (2015) further
documented the differences of peak stage and duration by examining the discharge
between the stream and aquifer interface. As the peak stage increases, the flux across the
boundary increases as well (Siegieiev et al., 2015). While these studies document how
increases in peak stage and storm duration are documented, the effects are commonly
grouped together and do not explain which factor is more influential.
5
Previous research documenting the effects of increasing peak stage and storm
duration on bank storage use both analytical (Cooper and Rorabaugh, 1963; Gill, 1985;
Barlow et al., 2000; Hunt, 2005; Ha et al., 2008) and numerical flow models (Squillace,
1996; Chen and Chen et al., 2003; Ha et al., 2008; Siegieiev et al., 2015). Using
numerical models, studies use 1-D (Fernald et al., 2001) and 2-D numerical models
(Squillace, 1996; Wroblicky et al., 1998; Bates et al., 2000; Li et al., 2008; Brunner et al.,
2009, McCallum et al., 2010; Welch, et al., 2013; Siergieiev et al., 2015). Very few
studies use 3-D models (Chen and Chen et al., 2003; Ha et al., 2008). Since groundwater
and surface water interactions are complex and occur over all dimensions, 3-D modeling
is needed to be fully understand the bank-storage process (Sophocleous, 2002). Arguing
that the longitudinal (y) dimensional component is only important at the beginning and
end of an event, Bates et al. (2000) justifies modeling bank-storage under the two
dimensions, often the horizontal (x) and vertical (z) dimensions. However, along natural
meandering streams, two dimensions might skew the results. As mentioned previously,
meandering channels induce parafluvial exchanges across the meander bends (Cardenas
et al., 2004; Boano et al., 2006; Peterson Sickbert, 2006). This should increase the
amount of bank storage occurring across the meander bends. Cardenas et al. (2004)
studied the effects of channel curvature on groundwater and surface water exchange and
found that the exchange processes are better understood when incorporating stream
curvature.
6
Research Objectives
The purpose of this work is to model bank-storage processes along a meandering
stream to describe the effects of increasing peak stage and storm duration of a storm
event. More specifically, the variables will be examined separately to determine whether
one variable influences the amount of bank storage more than the other. The final
objective is to examine the implications of bank storage along the meandering stream.
These objectives will be met by answering the following research questions:
Research Questions
1) How does peak stage of a hydrograph influence the bank-storage process?
2) How does storm duration of a hydrograph influence the bank-storage process?
3) Which variable, peak stage or storm duration, is statistically more influential in
determining the amount of bank storage?
4) What are the implications of these results along a meandering stream?
7
CHAPTER II
STUDY SITE AND METHODS
Study Site: Little Kickapoo Creek
The study area is along a small reach of Little Kickapoo Creek (LKC), located
south of Bloomington, Illinois (Figure 1). LKC is a third order, low gradient (0.002),
perennial stream with its headwaters beginning in Bloomington, IL and drains into
Kickapoo Creek (Peterson and Sickbert, 2006; Basu, 2007). The climate of central
Illinois is humid continental with cold winters and warm summers (Peterson and
Sickbert, 2006). The average annual precipitation of the area from 2005 to 2015 is 0.95
m, which is evenly distributed throughout the year (Illinois State Geological Survey,
2016). The stage of LKC can change dramatically in response to high precipitation or
snowmelt events. A gage located south of the study site shows the base level stage
fluctuates between 0.9 m and 1.1 m (with a datum of zero) and during high precipitation
or snow melting events, the stage can peak up to 4 m depending on the location along the
stream. The gauge nearby also shows that the stage post storm events is asymmetric.
After a storm, the stage of LKC typically reaches its maximum stage within a few hours
before stage steadily returns to baseflow (Figure 2). The steady stage drop can take
several hours and even up to a several days depending on the amount of precipitation
(Figure 2).
In the study area, the stream is unmodified and meanders through an alluvial
valley (Peterson and Sickbert, 2006; Sickbert and Peterson, 2014) (Figure 1C). Created
8
during the Wisconsinan glaciation episode, the alluvial valley is about 300 m wide
(Beach and Peterson, 2013). The valley was created by meltwaters that eroded through
glacial till (Basu, 2007). The valley was filled with coarse-grained sands and gravels
creating a glacial outwash plain that pinches out near the sides of the valley (Van der
Hoven et al., 2008). Lastly, alluvium was deposited over the valley as LKC meandered
through the valley (Basu, 2007).
The geology of the valley consists of thick muddy alluvial flood plain deposits
(Cahokia Alluvium) on top of sands and gravels (Henry Formation), which mantles a
low-permeable clay-till (Wedron Formation) (Peterson and Sickbert, 2006). The Cahokia
Alluvium averages a thickness of 2 m across the valley and consists of fine-grained sand
and clayey silts with organic material (Ackermann et al., 2015). Roots of plants, animal
burrows, and worm holes have created macropores in the alluvium increasing its
permeability to have a saturated hydraulic conductivity of 3.5 x 10-6 m/s and an average
porosity of 0.25 (Ackermann et al., 2015). The Cahokia Alluvium extends across the
study site and pinches out at the valley walls. The Henry Formation is composed of
poorly-sorted gravels with sands with a thickness up to 10 m near the center of the valley
and pinches out towards the edges of the valley (Ackermann et al., 2015; Peterson and
Sickbert, 2006). Due to the high porosity (0.35) in the glacial outwash, it acts as an
aquifer with a horizontal hydraulic conductivity at 1.0 x 10-4 m/s (Ackermann et al.,
2015; Peterson and Sickbert, 2006). The specific storage and specific yield values of the
Henry Formation are 0.0007 and 0.021 respectively. The Wedron Group is a glacial till
that is least 70 m thick. The till is composed of clay-till with sand and gravel lenses
(Ackermann et al., 2015). The hydraulic conductivity of the glacial till is 1.0 x 10-8 m/s
9
(Hensel and Miller, 1991). Due to the contrast in hydraulic conductivity values between
the Wedron Formation and the Henry Formation, the glacial till acts as a confining unit
restricting flow within the Henry Formation.
LKC flows near the eastern side of the valley and the modern floodplain
approximately extends the full width of the valley (Ackerman et al., 2015). The banks of
the stream range from sharply incised banks that cut into the alluvium to low-relief
depositional point bars (Peterson and Sickbert, 2006). The water table in the glacial
outwash aquifer is near the boundary between itself and the overlying alluvium during
baseflow conditions (Van der Hoven et al., 2008). Throughout the glacial outwash
aquifer, groundwater flows towards the south-southeast (Van der Hoven et al., 2007). The
glacial outwash aquifer has been documented to be well connected to the stream as the
streambed of the LKC is composed of the Henry Formation (Peterson and Sickbert, 2006;
Van der Hoven et al., 2008). Changes in stream stage similarly affect wells close to the
stream, while further away wells are controlled more by precipitation (Ackerman et al.,
2015).
Bank storage has been documented in the study site, but not explored. Peterson
and Sickbert (2006), used thermal transport to track hyporheic exchange through a
meander neck in the study site. Due to the increased gradient across the meander neck,
water flows through the neck and the aquifer provides short-term bank storage (Peterson
and Sickbert, 2006). Van der Hoven, et al. (2007) confirms these findings with a chloride
transport model of the study area. Chloride concentrations in LKC range from ~150 mg/L
to >1000 mg/L, while groundwater concentrations have chloride concentrations of ~11.8
mg/L (Van der Hoven, et al., 2007). Across the meander neck; however, groundwater
10
chloride concentrations (~152.4 mg/L) are an order of magnitude higher than
groundwater up gradient from the stream (Van der Hoven et al., 2007).
LKC is also impacted by non-point source pollution of road salts. Through the
watershed, LKC meanders through a mixed urban and agricultural watershed (27%
urban; 69% agricultural). Chloride concentrations along LKC have been documented to
range from 46.6 mg/L to 372.8 mg/L with the highest concentrations through the deicing
or winter months (Lax, 2007). Chloride/Bromide ratios determined the source of
chlorides to be the application of road salt (Lax, 2007).
Figure 1: A) Study site location in central Illinois B) Aerial view of conceptual model with outline domain in white and boundary conditions in red and C) cross-sectional view of conceptual model from the A-A` line in B.
11
Figure 2: Representative hydrograph of LKC post storm event. Stage increases rapidly to a peak of 2 m, which is approximately 1 m of change from baseflow. Other storm events produced similar or lower peak stage changes.
Field Data
Groundwater depths were measured from 18 wells across the study site on a bi-
weekly basis from 2/2/2016 through 8/02/2016 (Figure 3). Measured groundwater depths
from the wells were compared to steady-state simulations and used to calibrate the
numerical model. Pressure transducers located within LKC on the northern and southern
edges of the study site recorded water elevation in the stream. These data were used to
construct the real-world storm scenarios used in each simulation.
0.75
1
1.25
1.5
1.75
2
2.25
9/13/2015 9/18/2015 9/23/2015 9/28/2015 10/3/2015 10/8/2015
LKC
Sta
ge (m
)
Time (days)
Representative storm-event hydrograph of LKC
12
Stream water sampling and discharge measurements occurred on a bi-weekly
basis from 2/2/2016 through 12/28/2016. An electromagnetic flow meter was used to
measure stream velocity at 0.6 depth, which was incorporated into the velocity-area
method to calculate stream discharge. Stream samples were collected in triple-rinsed
bottles and filtered before being analyzed for anions (specifically chloride) with a Dionex
DX-120 Ion Chromatograph. Two separate sampling events occurred where groundwater
samples in six wells were analyzed for chloride on 6/22/2016 and 9/8/2016. Wells
included in the sampling were 104, 53, 41, 40, 39, and 21 (Figure 3). Stream water
sampling occurred along the reach towards the southern edge of the study area near well
105 (Figure 3). This sampling location is referred to as LKC South.
13
Figure 3: LKC well field study site with well locations used for chloride sampling.
Conceptual Model
This study uses boundary conditions and aquifer properties from previous models
and research of the LKC study site. A conceptual model was constructed to identify
boundary conditions of the study site (Figure 1). The east/west (x) dimension of the
domain was 336 m and the north/south dimension (y) was 429 m. The elevation and
extent of the study site was defined using LiDAR data in ArcMap (Illinois State
Geological Survey, 2014). Borehole data were used to delineate the subsurface domain
14
and extent of the glacial outwash (Basu, 2007; Van der Hoven et al., 2008). The Cahokia
Alluvium and the Henry Formation were the only two hydrogeologic units included. The
thickness of the Cahokia was 2 m across the entirety of the domain, while the thickness
of the Henry Formation was 10 m at its thickest under the stream, and pinches out
towards the eastern and western edges. Due to no observed spatial differences in
hydraulic conductivities and no differences in vertical and horizontal hydraulic
conductivities, both the Henry Formation and the Cahokia units were considered
homogenous isotropic (Cahokia=3.5 x 10-6 m/s; Henry = 1.0 x 10-4 m/s). The porosity
was 0.25 for the Cahokia and 0.35 for the Henry Formation. Specific storage and specific
yield values for the Henry Formation were 0.0007 and 0.021, and the values for the
Cahokia were based on unpublished aquifer tests, 0.001 and 0.01 respectively. The
Wedron Formation acted as a confining unit due to its low hydraulic conductivity.
The boundary conditions along the top of the model were designated as a constant
flux, which represented recharge from precipitation. The recharge value was initially set
to 10% of the average annual rainfall of 3.01 x 10-8 cm/s, which was similar to previous
research along LKC (Basu, 2007; Ackerman et al., 2015; Ludwikowski, 2016). The
bottom boundary is a no flow boundary representing the low permeable Wedron Group
(Basu, 2007; Ackerman et al., 2015). The northern, southwestern, and southeastern
boundary were constant head boundaries. The southwestern boundary was set using
surveyed elevations of the nearby tributary; while the northern boundary were established
using collected observed groundwater elevations (Basu, 2007). The southeastern
boundary has two components with the first being set using observed groundwater
elevations in the area and the second was the ending reach of LKC. The eastern and
15
western boundary were considered no flow boundary coinciding with the location that
Henry Formation and Cahokia Alluvium pinch out along the value edge where the
Wedron Formation is present.
Numerical Model
MODFLOW 2000 (Harbaugh et al., 2000), a 3-D finite difference groundwater
flow simulation program, was used to simulate groundwater and surface water
interactions. Initial parameters and boundary conditions used in the numerical model
have been mentioned previously and are displayed in Table 1. Model construction
occurred in Groundwater Vistas, which is a groundwater modeling environment program.
The Cahokia Alluvium was the top layer (2 m); the Henry Formation had 10 layers (10
m). The contact between the Henry Formation and the confining unit, the Wedron Group,
served as the lower no-flow boundary. The cell size of the top layer was 3x3x2 m and the
10 layers in the Henry Formation were 3x3x1 m. MODFLOW simulations were
conducted under the following assumptions: 3-D, unconfined, heterogeneous, isotropic
aquifer.
The model was calibrated under steady-state conditions from measured
groundwater elevations data collected from the field and calibrated values used in the
model are shown in Table 1. Post-calibration, 36 numerical simulations were conducted
under transient-state conditions to investigate the influence of peak stage and storm
duration on bank storage. Six different peak stages used were 0.15 m, 0.3 m, 0.46 m, 0.61
m, 0.76 m, and 0.91 m. Six different storm durations used were 1, 3, 5, 7, 9, 10 hours. All
16
combinations of peak stage and storm duration were used to total create the 36
simulations. These hydrographs were constructed based off realistic hydrographs along
LKC (Figure 2). A gauging station located along LKC south of the study site, showed
that most of the storm events over a given year generate a peak stage of ~ 1 m and lower.
Thus, increments of 0.15 m were used. Peak stage was defined as the peak stage of the
flood hydrograph and storm duration was defined as the length of time that the peak stage
was sustained throughout each simulation (Figure 4). All hydrograph combinations of
peak stage and storm duration were constructed and inputted into the numerical model.
For example, if the combination, peak stage of 0.61 m, storm duration of 5 hours, were
used, the hydrograph would steadily increase towards a peak of 0.61 m at the ninth hour.
Then, for 5 hours, the peak stage would be sustained until the 14th hour, then begin to
steadily decrease to pre-peak stage. Each simulation ended at 100 hours and bank-storage
volumes were then calculated. Bank storage was defined as 𝑉𝑉 = −∫ 𝑄𝑄𝑄𝑄𝑄𝑄𝑡𝑡0 where V is the
volume of bank storage per unit length of the stream at time t and Q is the flow of water
into the stream. However, this study calculated the volume of bank storage (m3) by using
change in storage from the 1st hour of the simulation across the entirety of the first layer
(Cahokia Alluvium). Storage values were calculated and recorded to determine the
effects of peak stage and storm duration. Properties of storage values observed were
changes in maximum bank storage, flux, amount of storage remaining after 100 hours,
gradient changes, and storage zone changes. Lastly, two multiple linear regression
models were constructed using peak stage and storm duration to predict maximum bank
storage and storage remaining at 100 hours. Using a statistical software (IBM SPSS
Statistics 24), two multiple linear regression models were created. The first model (LMBS)
17
used peak stage and storm duration as independent variables to predict the dependent
variable of maximum bank storage. The second model used peak stage and storm
duration as independent variables to predict the dependent variable of storage remaining
at 100 hours. The software then generates standardized coefficients used to predict the
dependent variable. The standardized coefficients are generated by normalizing both
independent variables and the dependent variable so independent variables can be
compared.
18
Figure 4: A) Example of hydrograph input into the numerical model displaying the variables: peak stage and storm duration. B) Example of bank-storage output values displaying the variables of maximum bank storage and the storage remaining after 100 hours.
19
Table 1: Initial and calibrated values used in the numerical model
Parameter Initial Values Source Calibrated Values Henry Formation K 1.0 x 10-4 m/s Ackerman et al.,
2015 1.1 x 10-4 m/s
Henry Formation Porosity 0.35 Ackerman et al., 2015
0.35
Henry Formation Sy 0.021 Ludwikowski, 2016 0.021
Henry Formation Ss 0.0007 Ludwikowski, 2016 0.0007
Cahokia Alluvium K 3.5 x 10-6 m/s Ackerman et al., 2015
6.07 x 10-7 m/s
Cahokia Alluvium Porosity 0.25 Ackerman et al., 2015
0.25
Cahokia Alluvium Sy 0.01 Unpublished 0.01
Cahokia Alluvium Ss 0.001 Unpublished 0.001
Recharge Rate 3.01 x 10-8 cm/s Illinois State Water Survey
3.17 x 10-7 cm/s
20
CHAPTER III
RESULTS
Calibration
Prior to conducting the simulations under transient-state conditions, the numerical
model was calibrated under steady-state flow conditions. The numerical model was
calibrated using observed water table elevation data collected from field collection
(Figure 4). The residuals between the measured and simulated head values ranged from -
0.38 m to 0.41 m with a mean absolute error of 0.15 m and a root mean square error of
0.19 m (Figure 5). General flow direction of groundwater in the calibrated model was
north to south except for areas within four meters of the stream.
21
Figure 5: Measured Head, groundwater head values observed in the field, versus Simulated Head, modeled head values at each measuring location. The residuals analysis results in a mean absolute error (MAE) of 0.15 m. Black line shows the 1:1 ratio for an ideal model solution.
Following model calibration, 36 model scenarios were simulated for the various
combinations of stages and storm durations. After each simulation, storage over the entire
model was recorded at each hour and subtracted from the first hour to calculate the
change in bank storage. Bank storage was calculated as a volume and reported in m3.
Each simulation ended at 100 hours. The following results examine the influence of peak
stage and storm duration. For clarity and easier comparison, only 12 simulations, with
storage changes over 100 hours, are presented. Six of the simulations examine the
variation of peak stage over a constant storm duration of 5 hours; the second six scenarios
focus on a constant peak stage over the various storm durations. The 12 simulations are
representative of the other simulations; graphs and tabular data for all 36 scenarios are
presented in Appendix A. However, graphs and tables of maximum bank-storage values
and storage remaining at 100 hours of all 36 scenarios are shown below.
221221.2221.4221.6221.8
222222.2222.4222.6222.8
223
221 221.2 221.4 221.6 221.8 222 222.2 222.4 222.6 222.8 223
Sim
ulat
ed H
ead
(m)
Measured Head (m)
MAE: 0.15 m
22
Peak stage
To examine the influence of increasing stage on bank storage, six simulations
with a constant storm duration of 5 hours and varying stages are shown in Figure 6. Each
scenario, with a peak stage between hours 9 and 14, had a maximum storage value
between hour 12 and 14 (Figure 6). Increases in stage increased the maximum bank
storage and increased the amount of bank storage remaining after 100 hours. For
example, the maximum bank storage for a 0.3 m increase is 416.5 m3 and ending value of
215.3 m3, while the 0.91 m increase simulation maxed at 1053 m3 with an ending value
of 348.3 m3 (Figure 6). Therefore, increasing the stage 3-fold increases the maximum
bank storage 2.5-fold and a 1.6-fold increase in storage remaining at 100 hours. While
increases in the stage increase the banking storage remaining after 100 hours these
increases appear to level off. The percent of storage remaining in the aquifer compared to
the maximum bank storage was 61% for the 0.15 m simulation and 33% for the 0.91 m
increase (Figure 6). These relationships are representative and consistent with the other
numerical simulations. Increases in peak stage appear to have a consistent increase in the
maximum bank storage that occurs over each simulation (Figure 7). In addition, the noted
leveling off the storage remaining at 100 hours is consistent with other storm durations
(Figure 78. As peak stage increases the percentage of bank storage remaining at 100
hours decreases, and is consistent across all simulations (Table 2).
23
After each simulation begins, the storage values increase until a maximum bank
storage is reached. During this time, water is leaving the stream and entering the adjacent
aquifer and is referred to as the fill time. After the maximum storage occurs, water then
returns from the aquifer to the stream and is referred as the return time. Changes in slope
of the fill and return time reflect changes in the discharge of water entering or leaving the
aquifer. Increasing peak stage has effects on the discharge of the fill time and return time.
Here, increases in peak stage appear to increase both the fill time and return time as the
slope of the higher peak stages have steeper slopes (Figure 6). Along LKC, infiltration
time for bank storage was much shorter than the return time back to the stream and even
after 100 hours, a large portion (33-64%) of the bank storage had not returned to the
stream (Figure 6; Table 2). Overall, the shape of the output bank-storage curve is
asymmetrical with a sharp increase until the maximum bank storage is reached, then a
sharp decline until the stage of LKC returns to baseflow. Last, the water then slowly
returns to the stream over several days.
To examine how increasing stage influences bank storage, hydraulic gradients and
extent of the storage zone were examined with a constant storm duration of 5 hours and
changing stages (Figure 9). All graphics are shown at the time of maximum bank storage.
At all stages, most of the bank storage occurs within ~4 m. of the stream (Figure 9).
However, on the most northern reach of the meander, the storage zone extends the
furthest with little change occurring along the straight reaches of LKC (Figure 9). The
storage zone increases appears to leveling off as the peak stage increases (Figure 9).The
hydraulic gradient between the stream and ~4 m of the banks changes as the stage
increases (Figure 9). Over all simulations, the hydraulic gradient reverses resulting in
24
storage along the banks. The reversed hydraulic gradient near the stream increases as the
stage increases (Figure 9).
Figure 6: A) Hydrograph input used in numerical model simulation for the results provided in Figure 7B. B) Simulation results for models scenarios with a peak stage peaking at hour nine and held constant for 5 hours at various stages. Storage values are measured as change from beginning of simulation (1st hour).
25
Figure 7: Positive relationship between maximum bank storage and peak stage for all 36 numerical simulations.
0
200
400
600
800
1000
1200
1400
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Max
imum
Ban
k St
orag
e(m
3 )
Peak Stage (m)
Relationship between increasing peak stage and maximum bank storage
1 hr 3 hr 5 hr7 hr 9 hr 10 hr
26
Figure 8: Positive relationship between storage remaining at 100 hours and peak stage for all 36 numerical simulations.
0
50
100
150
200
250
300
350
400
450
500
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Ban
k St
orag
e at
100
hou
rs(m
3 )
Peak Stage(m)
Relationship between increasing peak stage and storage remaining at 100 hours
1 hour 3 hour 5 hour7 hour 9 hour 10 hour
27
Figure 9: Scenario analysis of how increasing storm duration effects bank storage along a meandering reach of LKC. Each model scenario are the same as Figure 2 with a constant storm duration of 5 hours. Green line represents contour line 221.76 and blue line represents contour line 221.66. Yellow numbers represent the increases in storage zone. These values were calculated from change in baseflow.
28
Storm duration
To understand the influence of storm duration on bank storage, six simulations
with a constant stage of 0.61 m and varying storm durations are shown in Figure 10. As
storm duration increases the maximum bank storage increases and the max value occurs
later (Figure 10). For example, the 1-hour duration simulation maxed at hour 9 at a value
of 653.9 m3 while the 10-hour duration had a maximum bank storage of 902.2 m3 at hour
18. Therefore, a 10-fold increase in storm duration increased the maximum bank storage
by 1.4-fold. As opposed to increasing peak stage, there are no observed flux
differentiations with altering the storm duration due to constant slopes of the fill time and
return time as storm duration increases. There is also a proportional increase in maximum
bank storage as there is in bank storage remaining after 100 hours (Figure 10). For the 1-
hour duration, bank storage remaining after 100 hours was 214 m3, while the 10-hour
simulation was 408.8 m3. With a 10-fold increase in storm duration, the bank storage
remaining increased by 1.9-fold. The percent of storage remaining in the aquifer
compared to the maximum bank storage was 33% for the 1-hour storm duration and
increased to 45% percent for the 10-hour duration (Figure 10). These relationships are
representative and consistent with the other numerical simulations. Increases in storm
duration appear to increase the volume of maximum bank storage minimally and is
consistent at all peak stages (Figure 11).Unlike the changes in stage however, the leveling
off effect of the bank storage remaining after 100 hours and increases proportionally with
increases in storm duration (Figure 12). The percentage of bank storage remaining at 100
hours also increases with increasing storm duration at all peak stages (Table 2).
29
The shape of the bank-storage output curve has a similar asymmetrical shape to
that of increasing peak stage with a few differences (Figure 10). Like increasing peak
stage, the fill time increases sharply to the maximum amount then sharply decreases until
the stream returns to baseflow (Figure 10). Once the stream is at baseflow, the remaining
storage then steadily returns to the stream over several days (Figure 10). However, the
slope of the fill time and return time do not change and indicates the increasing storm
duration as little effect on discharge of the water entering and returning to the stream
(Figure 10). The timing of the maximum bank storage occurs later with increasing storm
duration (Figure 10). Same with peak stage, even after 100 hours a large portion (25-
64%) of the storage remains across all simulations (Table 2).
To examine how increasing storm duration influences bank storage, the extent of
the storage zone was examined with a constant stage of 5 hours and changing storm
durations (Figure 13). All graphics are shown at the time of maximum bank storage.
Similar to Figure 9, a majority of the hydraulic changes occurs within ~ 4 m of the
stream. As storm duration increases, the hydraulic gradient along the straight reaches
remains constant. Along the meander bend, however, the storage zone size increases with
increasing storm duration (Figure 13). As opposed to increasing peak stage, the increases
in storage zone with increasing storm duration do not appear to level off and actually
would continue to increase if the storm duration went above 10 hours.
30
Table 2: Percentage of bank storage remaining at 100 hours. Values increases with storm
duration and decrease with peak stage.
Table 2: % Storage Remaining at 100 hours (% = volume at 100/maximum bank storage)
Peak Stage (m) Storm
Duration (hours) 0.15 0.3 0.46 0.61 0.76 0.91
1 33% 42% 37% 33% 28% 25%
3 56% 47% 40% 37% 32% 29%
5 61% 50% 44% 40% 35% 32%
7 63% 52% 47% 43% 38% 34%
9 64% 55% 49% 44% 40% 37%
10 64% 55% 52% 45% 41% 38%
31
Figure 10: A) Hydrograph input used in numerical model simulation for the results provided in Figure 7B. B) Simulation results for models scenarios with a 0.61 m stage peaking at hour nine and held constant for various storm durations. Storage values are measured as change from beginning of simulation (1st hour).
32
Figure 11: Positive relationship between maximum bank storage and storm duration from all 36 numerical model simulations.
0
200
400
600
800
1000
1200
1400
0 2 4 6 8 10 12
Max
imum
Ban
k St
orag
e(m
3 )
Storm Duration(hours)
Relationship between increasing storm duration and maximum bank storage
0.15 m 0.3 m 0.46 m0.61 m 0.76 m 0.91 m
33
Figure 12: Positive relationship between storage remaining at 100 hours and storm duration from all 36 numerical model simulations.
0
50
100
150
200
250
300
350
400
450
500
0 2 4 6 8 10 12
Ban
k St
orag
e at
100
hou
rs(m
3 )
Storm Duration(hours)
Relationship between increasing storm duration and storage remaining at 100 hours
0.15 m 0.3 m 0.46 m0.61 m 0.76 m 0.91 m
34
Figure 13: Scenario analysis of how increasing storm duration effects bank storage along a meandering reach of LKC. Each model scenario are the same as Figure 2 with a constant storm duration of 5 hours. Green line represents contour line 221.76 and blue line represents contour line 221.66. Yellow numbers represent the increases in storage zone. These values were calculated from change in baseflow.
35
Multiple Linear Regression Analysis
To determine and measure the influence of each variable, peak stage and storm
duration, maximum bank-values and volumes of bank storage remaining at 100 hours
were recorded from all 36 numerical simulations (Table 3; Table 4).
LMBS predicted the maximum bank storage with an R2 value of 0.993. The
standardized coefficients and p-values are shown in Table 4. L100 predicted the volume of
storage remaining at 100 hours well and had an R2 value of 0.936. The standardized
coefficients and p-values are shown in Table 5. The flood stage is ~3.6 times more
influential in predicting the peak bank storage that will occur over each simulation (Table
5). The low significance of the flood stage variable comparative to the flood duration
significance in LMBS indicate that the flood stage is more significant in the prediction as
well (Table 5). However, the flood stage is only ~1.1 times more influential in determine
the amount of storage remaining at 100 hours (Table 6). The similar p-value between
flood stage and flood duration in L100 indicate the equal significance between the two
variables (Table 6).
36
Table 3: Maximum bank-storage values used in the first of two multiple linear regression analyses. Peak stage and storm duration are the independent variables used to predict the dependent variable of maximum bank storage.
Table 3: Predicted Maximum Bank Storage (m3)
Peak Stage (m) Storm
Duration (hours) 0.15 0.3 0.46 0.61 0.76 0.91
1 185.51 342.11 489.97 653.86 800.37 946.53 3 216.48 380.03 500.90 694.80 858.78 1008.36 5 227.26 420.80 584.89 753.67 916.38 1068.97 7 253.32 464.38 632.78 803.41 974.67 1130.80 9 285.69 496.63 682.16 872.90 1042.70 1201.81
10 301.77 523.41 712.29 902.21 1091.95 1226.68
Table 4: Maximum bank-storage values used in the first of two multiple linear regression analyses. Peak stage and storm duration are the independent variables used to predict the dependent variable of maximum bank storage.
Table 4: Predicted Storage Remaining at 100 hours (m3)
Peak Stage (m) Storm
Duration (hours) 0.15 0.3 0.46 0.61 0.76 0.91
1 60.88 145.01 182.50 213.95 223.88 239.94 3 120.86 176.95 198.06 255.62 273.26 288.39 5 138.34 208.60 256.27 298.69 320.12 337.66 7 159.05 240.39 296.31 343.02 366.21 389.03 9 181.43 272.73 336.12 387.09 419.94 442.22
10 193.62 287.97 368.20 408.82 443.98 466.84
37
Table 5: Multiple linear regression results for predicting maximum bank storage.
Table 5: Linear Model: LMBS
Independent Variables Standardized Coefficients (Beta)
p-value (Significance)
Peak Stage 0.962 7.14 x 10-37
Storm Duration 0.262 7.47 x 10-19
Table 6: Multiple linear regression results for predicting storage at 100 hours.
Table 6: Linear Model: L100
Independent Variables
Standardized Coefficients (Beta)
p-value (Significance)
Peak Stage 0.750 2.21 x 10-17
Storm Duration 0.664 7.32 x 10-16
Seasonal Stage and Chloride Results
The stage of LKC varied across the year with ~19 precipitation events, which
caused spikes in the stage of the stream (Figure 14-15). A majority of the precipitation
events occurred from March – September. LKC experienced two separate base levels
across the sampled year. During the winter to late spring, the base level was ~1.2 m.
During the summer to fall, the baseflow stage was ~1.0 m. Chloride concentrations
38
sampled at the LKC South varied seasonally (Figure 14). The winter months (Jan 1st-
Feb. 2nd; Nov. 19th – Dec. 31st) had maximum chloride concentrations at 257.5 mg/L,
minimum of 41 mg/L with an average concentration of 143 mg/L. Spring and summer
(March 1st – August 18th) had maximum and minimum concentrations of 182.3 and 49.1
mg/L respectively, with a mean chloride concentration of 83.4 mg/L. The fall months
(Sept. 8th – Nov. 11th) revealed a 67.2 mg/L maximum, 38.1 mg/L minimum and a mean
concentration of 56.3 mg/L. Chloride load varied seasonally at the LKC South site as
well (Figure 15). The winter months (Nov. 29th – Feb. 2nd) had maximum chloride load at
158 g/s, minimum of 49 g/s with an average load of 81 mg/s. Spring and summer (March
1st – August 18th) had maximum and minimum load of 216 and 19 g/s respectively, with a
mean chloride load of 59.9 g/s. The fall months (Sept. 8th – Nov. 11th) revealed an 81 g/s
maximum, 0.6 g/s minimum and a mean concentration of 21.8 g/s.
The majority of the stage fluctuations throughout are through the late spring and
summer months (Figure 14-15). During this time, the mean chloride concentrations and
loads are lower than winter but higher than the fall. There appears to be an increase in
chloride concentration following storm events. For the storm events on 3/30/2016,
7/14/2016, and 7/25/16, stream sampling occurs a few days later where the stage was
slightly elevated due to the effects of the storm. Chloride concentrations were 88.8,
112.62, and 75.19 mg/L respectively. For the storm events on 3/30/2016 and 7/14/2016,
the chloride loads were 54 g/s and 69.4 g/s respectively.
Groundwater concentrations reveal chloride concentrations higher than LKC in
most wells located near the stream and across the meander bend (Table 7). However,
wells located further from the stream, receiving source water from the northern western
39
areas of the aquifer, had similar or lower chloride concentrations than LKC (Table 7).
There is also a decrease in chloride concentration between the two sampling events
(Table 7). Chloride concentrations were higher on 6/22/2016 and lower on 9/8/2016 with
large differences occurring in wells 21 and 41 (Table 7).
Figure 14: Seasonal changes in chloride concentrations (mg/L) and stage (m) of LKC. Stage measured from Jan 1st through Dec. 31st and chloride [Cl-] concentrations measured from Feb. 2nd through Dec. 28th.
0.75
1.25
1.75
2.25
2.75
3.25
3.75
0
50
100
150
200
250
300
Stag
e (m
eter
s)
Chl
orid
e C
once
ntra
tion
(mg/
L)
Annual chloride concentrations and LKC stage changes
LKC South [Cl-]
LKC Stage
40
Figure 15: Seasonal changes in chloride load (g/s) and stage (m) of LKC. Stage was measured from Jan 1st through Dec. 31st and chloride load was measured from Feb. 2nd through Dec. 28th.
Table 7: Groundwater sampling events for groundwater chloride analysis. Sample locations relative to stream are shown in Figure 2. Green boxes indicate wells away further away from the stream, blue boxes indicate wells near the stream and yellow boxes indicate wells across the meander bend.
0.75
1.25
1.75
2.25
2.75
3.25
3.75
0
20
40
60
80
100
120
140
160
180
Stag
e (m
eter
s)
Chl
orid
e Lo
ad (g
/s)
Annual chloride load and LKC stage changes
LKC South Chloride Load
LKC Stage
Sample Date 6/22/2016 9/8/2016Sample Location
[Cl-] (mg/L)
[Cl-] (mg/L)
104 24.8 27.553 33 33.441 102.6 52.940 55.7 38.539 42.8 36.721 87.7 36.7
LKC 30 38
41
CHAPTER IV
DISCUSSION
Peak stage and storm duration are both key factors in controlling the bank-storage
process. Previous research group both peak stage and storm duration together and
document increases in both increase the amount of storage that occurs (Todd, 1955; Chen
and Chen et al., 2003; Ha et al., 2008; Siergieiev et al., 2015). The purpose of the project
was to identify which factor influences the amount of bank storage more. The results here
are consistent with previous research that suggests peak stage and storm duration affect
different areas of the bank-storage process (Chen and Chen et al., 2003; Ha et al., 2008;
Siergieiev et al., 2015). This suggest that neither factor, storm duration or peak stage, are
more important than the other because they affect different properties of bank storage.
However, the linear regression analysis suggests that peak-stage is statistically ~3.6 times
more influential in determining the maximum bank-storage volume that occurs after a
storm event (Table 5). When determining the storage remaining at 100 hours, peak stage
is still statistically more influential in determining that value.
Peak stage Influences
Peak stage is statistically more influential than storm duration in determining the
maximum bank storage that occurs over a given storm event (Table 5). This is consistent
with the relationships between storm duration, peak stage and maximum bank storage.
Increases in peak stage appear to change the maximum bank storage volumes more than
42
increasing storm duration (Figure 7; Figure 11). The increases in maximum bank storage
due to increasing peak stage documented here are consistent with previous work (Todd,
1955; Ha et al., 2008; Siergieiev et al., 2015). Ha et al. (2008) used a 1-D analytical
model to examine the influences of peak stage on bank storage. Using a 1, 2, and 3 m
peak stages, maximum bank storage values were approximately 30, 60, and 90 m3/m
respectfully. Thus, a 3-fold increase in peak stage resulted in a 3-fold increase in
maximum bank storage as well (Ha et al., 2008). In the current study, results were similar
as a 3-fold increase in peak stage resulted in a 2.5 fold increase in maximum bank
storage.
As peak stage increases, the maximum bank storage increases as well, but the
time of each maximum amount occurs at the same time (Figure 6). This suggests that
increasing the peak stage increases the discharge of water in and out of the aquifer
(Figure 6). The increases in discharge entering and leaving the banks are result from the
observed increasing in hydraulic gradient near the stream (Figure 9). Thus, increasing
peak stage allows a large volume of water to enter the bank sediments and at a faster rate
compared to lower peak stages. Since high peak stage storm events allow higher
maximum bank-storage volumes to occur but less actual retention of that storage, these
high peak stage storm events can act as flushing events. These events have the potential
to flush previous bank storage.
While peak stage is statistically more influential in determining the volume of
storage that occurs, it has less effect than storm duration on retention of storage. This is
indicated by two observations: 1) Increases in peak stage decrease the percentage of
storage left in the aquifer at 100 hours (Table 2), and 2) at high peak stages the storage
43
zone increase are minimized (Figure 9). Due to the limitations of Cahokia Alluvium
properties (hydraulic conductivity, porosity and permeability), the stream water does not
infiltrate deep into the bank sediments and allows the water to return to the stream faster
after a stage decrease. This limitation explains the asymmetrical shape of the bank-
storage output curve associated with increasing peak stage. Increases in maximum bank
storage due to increasing peak stage occur due to the exposure of more bank sediment
material. However, the water returns a large partition of that bank storage returns quickly
because the water does not infiltrate deep at higher peak stages.
Storm duration Influences
Although peak stage is statistically more influential than storm duration when
determining the storage volume remaining at 100 hours (Table 6), storm duration appears
to be more efficient at the retention of bank storage than peak stage. This is due to three
items: 1) the percent of bank storage remaining at 100 hours increases with storm
duration and decreases with peak stage (Table 2), 2) at storage remaining at 100 hours
appears to increase continuously for storm duration (Figure 12) and appears to level off
for peak stage (Figure 8) and, 3) increasing storm duration increases storage zone size
while higher peak stages provided little change in storage zone size.
Maximum bank storage increases slightly with increasing storm duration (Table
3). Since the peak stage is constant, the additional bank storage occurs though expanding
the storage zone laterally. As noted before, soil properties of the bank sediments play a
large role in dictating the amount of bank storage than can infiltrate into the bank
44
sediments. Therefore, increasing the peak stage allows a larger maximum volume to
occur but the water does not infiltrate deep enough to be retained for long periods.
Increasing the storm duration, however, minimizes the effects of the soil properties and
bank storage water penetrates deeper into the bank sediments (Chen and Chen, 2003).
Increases in storage zone then delays the return time and allows effective retention of the
bank storage (Chen and Chen, 2003). Increases in storm duration causing an increase in
maximum bank-storage volume and storage zone size are consistent with previous
research (Todd, 1955; Chen and Chen, 2003; Siergieiev et al., 2015). Chen and Chen
(2003) used a 3-D numerical model to examine bank storage changes with increasing
storm duration. Increases in storm duration increase the maximum bank storage and each
maximum occurred later. In addition, increases in storm duration increases storage zone
with the longest storm duration having the largest storage zone size (Chen and Chen,
2003).
Implications along LKC
Bank storage studies have used 1-D, 2-D, and 3-D models to evaluate how
variables control the bank storage process, however, three dimensions are needed to
accurately quantify volume of bank storage that occurs over each simulation. Bank
storage is defined as the change in volume of storage per unit of the stream (m3/m);
however, the results above were in change in volume across the entirety of the first layer
in the model (m3). The length of LKC during each simulation was 1152 m; thus, the
range of maximum bank storage values are from 0.185 m3/m to 1.05 m3/m. During most
45
of the 36 simulations less than 1.0 m3 occurred along each meter of the stream. However,
many of the above results suggests that most of the bank storage occurs across the
meander bends and less across the straight reaches of the stream. Groundwater chloride
concentrations immediately near the stream and across meander bends are similar to
chloride concentrations along LKC (Table 7; Figure 3) (Van der Hoven et al., 2007).
Groundwater chloride concentrations further away from the stream are much lower than
LKC (Table 7; Figure 3) (Van der Hoven et al., 2007). This suggests that the most of the
interaction between the stream and adjacent aquifer occurs immediately near the stream
and across meander bends. Most of the hydraulic gradient change due to the increase in
peak stage occurs within ~4 m of LKC (Figure 9). However, as the peak stage and storm
duration increase, the storage zone size of water penetrating the aquifer, only increase
across the meander bends and little change occurs along the straight reaches of the stream
(Figure 9). This suggests increases of bank storage across the entirety of the model occur
mostly across the meander bends. The shape of the output bank-storage curve indicate
this as well, especially for the higher peak stage simulations. At high peak stages, the
bank storage increases quickly to the maximum volume and then decreases quickly until
LKC returns to baseflow (Figure 6). Then, the storage remaining slowly returns to the
stream over several days. This indicates a large portion of the bank storage is quickly
returned to the stream, primarily because it occurs along more straight reaches and does
not infiltrate deep into the bank sediments. The water that does infiltrate across the
meander bend, infiltrates farther and has a longer return time. Therefore, the shallow
slope return time beginning as soon as LKC returns to baseflow can be attributed to
storage across the meander bends. This explains why the remaining bank storage takes
46
days to return to the stream, because the flow paths across the meander bends are long.
Short-term shortage across the meander bend has be explored, as parafluvial exchange
was tracked using thermal transport thought the meander necks along LKC (Peterson and
Sickbert, 2006).
The influence of the meander bend would not have been captured if the model
were ran in two dimensions. There are two consequences of using a two-dimensional
model along a meander stream. If a two-dimensional model were created along a straight
reach and the bank-storage values were extrapolated along the river, the values would be
under estimated. Likewise, if a two-dimensional model were created along the meander
bend, the values would be overestimated. This finding is consistent with previous 1-D, 2-
D and 3-D models. Ha et al. (2008) used both 1-D analytical solution and a 3-D
numerical solution to examine the differences in peak stage and storm duration and how
they affect bank storage. 1-D analytical solutions used along a reach of the river had a
majority of the maximum bank-storage values above 10 m3/m and up to 90 m3/m. When
using a 3-D numerical solution to examine the field application, the maximum bank
storage found was ~0.05 m3/m. While using a 1-D or 2-D model to estimate the bank
storage affect along a meandering stream can be overestimated or under estimated, using
less dimesons can be useful to gain insight about how changing different factors
(hydraulic conductivity, peak stage, porosity) affect the bank-storage process.
Along LKC, non-point source pollution of road salts has been identified. (Lax,
2007). LKC stream chloride concentrations vary seasonally with the highest
concentrations in the winter and lowest concentrations in the fall. The minimum chloride
47
concentration sample was 38 mg/L. While low compared to winter concentrations, 38
mg/L is still considered elevated compared to the documented background concentration
of ~ 15 mg/L in central Illinois (Panno et al., 2006). These seasonal variations are similar
to the research involving road salts (Van der Hoven et al., 2007; Kelly, 2008; Meriano et
al., 2009; Corsi et al., 2015). Traditional research suggest elevated stream chloride
concentrations are attributed to large masses of chlorides that moves through the
groundwater originating from areas with high urban land use (Bester et al., 2006; Kelly,
2008). The chlorides accumulate in the groundwater and then later discharge into the
streams over the non-deicing months (Corsi et al., 2015). However, this research suggests
that there is not a large chloride mass building in the watershed as chloride concentrations
away from the stream in many areas are low (~ 20 mg/L). Instead, high chloride
concentrations have been found along roadside ditches (Lax and Peterson, 2008;
Ludwikowski, 2016), in LKC (Lax, 2007; Van der Hoven et al., 2007) and along
meander bends and groundwater near LKC (Van der Hoven et al., 2007).
The data suggest chlorides are transported into storage along the bank sediments
and across meander bends of LKC during the winter months. Bank storage has previously
been documented but not explored at LKC (Peterson and Sickbert, 2006; Van der Hoven
et al., 2008). In streams with high sinuosity, meander bends extend the hyporheic zone
due to the increased hydraulic gradient across the meander bends (Cardenas et al., 2004;
Peterson and Sickbert, 2006; Boano et al., 2006). Van der Hoven et al. (2008) used
MODPATH to model chloride movement through the meander bend presented in Figures
9 and 13. It takes a minimum of 60 days and up to 200 days for chloride particles to move
from the stream, through the meander bend and then back to the stream. During the
48
winter months, when chloride concentrations are the highest, stream water moves into the
meander bends and are temporarily stored. Throughout summer, several high peak stage
events occurred due to an increase in high precipitation events (Figure 14-15). These high
peak stage and short storm duration events transport water with lower chloride
concentrations into the bank sediments and across meander bends allowing high
concentration water in storage to be flushed out back into the stream. This is consistent
with the above results as the several high peak stage and low storm duration events occur
in the summer. Even with the increase in discharge through LKC, the mean summer
concentrations are still higher than the fall.
49
CHAPTER V
CONCLUSION
Conclusions
Neither peak stage nor storm duration are more important than the other, in terms
of effecting the bank-storage process. Other research document similar effects due to
increasing peak stage and storm duration. However, this study, using a 3-D numerical
model, documents peak stage being statistically more influential in determining the
maximum bank storage (~ 3.6 x) and the bank storage occurring at 100 hours (~1.1 x).
While it is more influential at determining the storage remaining at 100 hours, storm
duration appears to be more effective at the retention of bank storage by allowing stream
water to infiltrate farther into the bank sediments. In addition, as peak stage increases the
storage remaining at 100 hours begins to level off suggesting that even at the highest
stages, water can only infiltrate so far before quickly returning to the stream.
Meandering bends along a stream appear to induce bank storage. Output bank-
storage curves, distribution of chloride concentrations across the study sight, and storage
zone changes along LKC show most of the stream/aquifer interactions occur along the
meander bends. Also along LKC, the chloride concentrations are highest during the
winter/spring and lowest in the fall. These high concentrations are attributed to road salt
deposition in the winter months. Findings here suggest that stream water with elevated
chloride concentration is stored along the stream and across meander bends through stage
increases during late winter and early spring. During the summer, high peak stage and
50
low storm duration events with lower chloride concentrations flush the bank-stored
chlorides back into the stream. This suggests that road salt storage and release into the
stream through the non-deicing season can be attributed to the bank-storage process.
Chlorides act conservatively; therefore, they do not react with the soils and groundwater
across the meander bends. However, increased storage across meander bends can be
important when examining stream ions that react more with the bank sediments and
aquifer water. While these findings recognize the role of parafluvial exchange in bank
storage and road salt distribution along a stream, the results may vary across different
streams, rivers and study sites.
Further Consideration
As mentioned before, the numerical model used in this study was only calibrated
in steady-state and not under transient-state conditions. Each simulation was modeled as
a single event while no re-wetting simulation took place. Both of these are justified since
this study examined the influences of peak stage and storm duration on bank-storage
processes under hypothetical conditions (construction of individual hydrographs). While
these hydrographs were constructed, the asymmetric shape (sharp increase with a gradual
decrease) of each graph reflected real world storm event induced stage changes. As stated
previously, the bank-storage volumes were measured using the change in model storage
from the first hour of each simulation across only the first layer (Cahokia Alluvium, ~ 2
m). Chen and Chen (2003) documented a significant portion of the bank storage can
occur underneath the stream channel, which was not included each bank-storage
calculation. Squillace (1996) documented comparable results and estimated 70% of the
51
bank-storage water moves underneath the streambed and the remaining 30% infiltrates
into the banks. However, examining the hyporheic interaction beneath the channel was
out of the scope of this study.
52
REFERENCES
Ackermann, J. R., Peterson, E. W., 2015, Quantifying nutrient removal from groundwater seepage out of constructed wetlands receiving treated wastewater effluent, Environmental Earth Science, v. 74, p. 1633-1645.
Barlow, P. M., DeSimone, L. A., Moench, A. F., 2000, Aquifer response to stream-stage and recharge variations. II. Convolution method and applications, Journal of Hydrology, v. 230, p. 211-229.
Basu, A., 2007, Quantifying N cycling between groundwater and surface water using numerical molding and mass flux calculations, Master’s thesis, Department of Geography-Geology Illinois State University.
Bates, P. D., Stewart, M. D., Desitter, A., Anderson, M. G., Renaud, J.-P. 2000, Numerical simulation of floodplain hydrology, Water Resources Research, v. 36, p. 2517-2529.
Beach, V., Peterson, E. W., 2013, Variation of hyporheic temperature profiles in a low gradient third-order agricultural stream – a statistical approach, Open Journal of Modern Hydrology, vol. 3, pg. 55-66.
Bester M. L., Frind, E. O., Molson, J. W., and Rudolph, D. L., 2006, Numerical investigation of road salt impact on an urban wellfield, Groundwater, v. 44, no. 2, p. 165-175.
Boano, F., Camporeale, C., Revelli, R., Ridolfi, L., 2006, Sinuosity-driven hyporheic exchange in meandering rivers, Geophysical Research Letters, v. 33, L18406.
Boulton, A. J., Findlay, S., Marmonier, P., Stanley, E. H., Valett, H. M., 1998, The functional significance of the hyporheic zone in streams and rivers, Annual Review of Ecology and Systematics, v. 29, p. 59-81.
Brunke, M., Gonser, T., 1997, The ecological significance of exchange processes between rivers and groundwater, Freshwater Biology, v. 37, p. 1-33.
Brunner, P., Cook, P. G., Simmons, C. T., 2009, Hydrogeologic control on disconnection between surface water and groundwater, Water Resources Research, v. 47, W1422, p. 1-13.
Cardenas, M. B., Wilson, J. L., Zlotnik, V. A., Impact of heterogeneity, bed forms, and stream curvature on subchannel hyporheic exchange, Water Resources Research, v. 40, W08307.
53
Chen, X. and Chen, X., 2003, Stream water infiltration, bank storage, and storage zone changes due to stream-stage fluctuations, Journal of Hydrology, v. 280, p. 246-264.
Cooper, H. H., Rorabaugh, M. I., 1963, Ground-water movements and bank storage due to peak stages in surface streams, Ground-water hydraulics, Geological Survey Water-Supply Paper 1536-J.
Corsi, S. R., De Cicco, L. A., Lutz, M. A., and Hirsch, R. M., 2015, River chloride trends in snow-affected urban watersheds: increasing concentrations outpace urban growth rate and are common among all seasons, Science of the Total Environment, v. 508, p. 488-497.
Cranswick, R. H., Cook, P. G., 2015, Scales and magnitude of hyporheic, river-aquifer and bank storage exchange fluxes, Hydrological Processes, v. 29, p. 3084-3097.
Doble, R., Brunner, P., McCallum, J., Cook, P. G., 2012, An analysis of river bank slope and unsaturated flow effects on bank storage, Ground Water, v. 50, p. 77-86.
Fernald, A. G., Wigington, P. J.-J., Lander, D. H., 2001, Transient storage and hyporheic flow along the Willamette River, Oregon: Field measurements and model estimates, Water Resources Research, v. 37, p. 1681-1694.
Gill, M. A., 1985, Bank storage characteristics of finite aquifer due to sudden rise and fall of river level, Journal of Hydrology, v. 76, p. 133-142.
Ha, K., Koh, D., Yum, B., Lee, K., 2008, Estimation of river stage effect on groundwater level, discharge, and bank storage and its field application, Geosciences Journal, v. 12, p. 191-204.
Harbaugh, A. W., Banta, E. R., Hill, M. C., McDonald, M. G., 2000, MODFLOW-2000, the U.S. Geological Survey modular ground-water model – user guide to modularization concepts and the ground-water flow process, U.S. Geological Survey, Open-File Report 00-92.
Harvey, J. W., Wagner, B. J., Bencala, K. E., 1996, Evaluating the reliability of the stream tracer approach to characterize stream-subsurface water exchange, Water Resources Research, vol. 32, pg. 2441-2451.
Hensel, B. R., Miller, M. V., 1991, Effects of wetlands creation on groundwater flow, Journal of Hydrology, v. 126, p. 293-314.
Hunt, B., 2005, Bank-storage problem and the Dupuit Approximation, Journal of Hydrologic Engineering, v. 10, p. 118-124.
Illinois State Geological Survey, 2014, Illinois geospatial data clearinghouse, LiDAR data set, McLean County, https://clearinghouse.isgs.illinois.edu/data/elevation/illinois-height-modernization-ilhmp-lidar-data
54
Illinois State Geological Survey, 2016, Illinois State Climatologist Data, Illinois State Water Survey: http://www.isws.illinois.edu/data/climatedb/data.asp (accessed May 2016).
Kelly, W., 2008, Long-term trends in chloride concentrations in shallow aquifers near Chicago: Ground Water, v. 46, p. 772-781.
Kondolf, G. M., Maloney, L. M., Williams, J. G., 1987, Effects of bank storage and well pumping on base flow, Carmel River, Monterey Country, California, Journal of hydrology, v. 91, p. 351-369.
Lax, S., 2007, Estimating stream chloride concentrations as a function of land use change for two small watersheds in central Illinois, [Master thesis]: Normal, Illinois State University, 44 p.
Lax, S., Peterson, E. W., 2009, Characterization of chloride transport in the unsaturated zone near salted road, Environmental Geology, v. 58, p. 1041-1049.
Ludwikowski, J., 2016, The Transport and fate of chloride within the groundwater of a mixed urban and agricultural watershed, [Master thesis]: Normal, Illinois State University, 65 p.
Li, H., Boufadel, M. C., Weaver, J. W., 2008, Quantifying bank storage of variably saturated aquifers, Ground Water, v. 46, p. 841-850.
Mencio, A., Galan, M., Boix, D., Mas-Pla, J., 2014, Analysis of stream-aquifer relationships: A comparison between mass balance and Darcy’s law approaches, Journal of Hydrology, v. 517, p. 157-172.
Meriano, M., Eyles, N., and Howard, K. W. F., 2009, Hydrogeological impacts of road salt from Canada’s busiest highway on Lake Ontario watershed (Frenchman’s Bay) and lagoon, City of Pickering, Journal of Contaminant Hydrology, v. 107, p. 66-81.
McCallum, J. L., Cook, P. G., Brunner, P., Berhane, D., 2010, Solute dynamics during bank storage flows and implications for chemical base flow separation, Water Resources Research, v. 46, W07541, p. 1-11.
Peterson, E. W., Sickbert, T. B., 2006, Stream water bypass through a meander neck, laterally extending the hyporheic zone, Hydrogeology Journal, vol. 14, pg. 1443-1451.
Panno, S., Hackley, K., Hwang, H., Greenberg, S., Krapac, I., Landsberger, S., and O'Kelly, D., 2006, Source identification of sodium and chloride in natural waters: Ground Water v. 44, p. 176-187.
Sickbert, T. B., Peterson, E. W., 2014, The effects of surface water velocity on hyporheic interchange, Journal of Water Resource and Protection, vol. 6, pg. 327-336.
55
Siergieiev, D., Ehlert, L., Reimann, T., Lundberg, A., Liedl, R., 2015, Modelling hyporheic processes for regulated rivers under transient hydrological and hydrogeological conditions, Hydrology Earth System Science, v. 19, p. 329-340.
Sjodin, A., Lewis Jr., W. M., Saunders III, J. F., 2001, Analysis of groundwater exchange for a large plains river in Colorado (USA), Hydrological Processes, v. 15, p. 609-620.
Sophocleous, M., 2002, Interactions between groundwater and surface water: the state of the science, Hydrogeology Journal, v. 10, p. 52-67.
Squillace, P. J., Thurman, E. M., Furlong, E. T., 1993, Ground water as a nonpoint source of atrazine and deethylatrazine in a river during base-flow conditions. Water Resources Research, v. 29, p. 1719-1729.
Squillace, P. J., 1996, Observed and Simulated Movement of Bank-Storage Water, Ground Water, v. 34, p. 121-134.
Todd, D. K., 1955, Ground-water flow in relation to a flooding stream, American Society of Civil Engineers, v. 81, p. 1-20.
Van der Hoven, S. J., Fromm, N. J., Peterson, E. W., 2008, Quantifying nitrogen cycling beneath a meander of a low gradient, N-impacted, agricultural stream using tracers and numerical modeling, Hydrological Processes, v. 22, p. 1206-1215.
Winter, T. C., Harvey, J. W., Franke, O. L., Alley W. M., 1998, Ground water and surface water – a single source, U.S. Geological Survey Circulation 1139.
Woessner, W. W., 2000, Stream and fluvial plain ground water interactions: rescaling hydrogeologic thought, Ground Water, v. 38, p. 423-429.
Wroblicky, G. J., Campana, M. E., Valett, H. M., Dahm, C. N., 1998, Seasonal variation in surface-subsurface water exchange and lateral hyporheic area of two stream-aquifer systems, Water Resources Research, v. 34, p. 317-328.
Welch, C., Cook, P. G., Harrington, G. A., Robinson, N., 2013, Propagation of solutes and pressure into aquifers following river stage rise, Water Resources Research, v. 49, p. 5246-5259.
56
APPENDIX A FULL BANK STORAGE DATA SET
57
APPENDIX B SUPPLEMENTAL 3-D MODELING FILES
Location: Geo(\\casfiles01) -> PetersonGroup -> Chabela_Lucas
