A REVIEW OF METHODS FOR ASSESSING
NONPOINT SOURCE CONTAMINATED GROUND-
WATER DISCHARGE TO SURFACE WATER
-------
&EPA
United States
Environmental Protection
Agency
Office of Water
(WH-550G)
EPA 440/6-90-007
September 1990
A Review of Methods for
Assessing Nonpoint Source
Contaminated Ground-Water
Discharge to Surface Water
-------
A Review of Methods for Assessing Nonpoint Source
Contaminated Ground-Water Discharge to Surface Water
August 27, 1990
Office of Ground-Water Protection
U.S. Environmental Protection Agency
-------
Table of Contents
I. Introduction 1
II. Methods for Measuring Nonpoint Source Contaminated Ground-Water
Discharge to Surface Water 4
A. Studies involving use of seepage meters or mini-piezometers
to measure ground-water discharge to surface water 4
B. Ground-water quality sampling and measurements of ground-
water flow to estimate loading to surface water 24
C. Studies involving geophysical techniques to estimate ground-
water discharge to surface water 32
D. Studies involving hydrograph separation, regression
analysis, or mass balance approaches to estimate the
contribution of ground water to stream flow 41
E. Numerical models of surface-water/ground-water interactions . 54
F. Studies involving the application of functions estimating
nonpoint source loading to surface water for various land
use types 68
G. Studies using environmental isotope methods to estimate the
contribution of ground water to stream flow 74
III. The Impact of Nonpoint Source Contaminated Ground-Water Discharge
to Surface Water and the Waste in Water Quality-Limited Water
Bodies: Determining Total Maximum Daily Load and Waste Load
Allocations 86
A. Statutory and Regulatory Mandate for Determining WLAs and
LAs Under the TMDL Process 87
B. Determining the Total Maximum Daily Load 89
C. Assessment of nonpoint source contaminated ground-water
discharge to surface-water analysis methods as components of
Waste Load Allocation 97
D. Summary 100
Appendix A - Waste Load Allocation References
-------
Chapter 1
Introduction
A. Purpose of this Report
This report presents a summary of methods that have been applied to
measure or estimate nonpoint source contaminated ground-water discharge to
surface water. The U.S. Environmental Protection Agency (EPA) Office of
Ground-Water Protection (OGWP) developed this analysis as part of an effort to
broaden the understanding of the manner in which human activities can affect
water quality in all phases of the hydrologic cycle within a watershed. EPA
undertook this project in response to the growing awareness that contaminated
ground-water discharge is a significant source of nonpoint source contaminant
loading to surface waters in many parts of the country. In particular, this
report is intended to stimulate understanding of the methods that may be
applied to better account for nonpoint sources of contaminant loading.
Improved characterization of nonpoint source loads to surface water may, in
turn, lead to more comprehensive approaches for setting total maximum daily
loads for surface waters and waste load allocation.1 This report provides an
overview of these methods, rather than a manual for employing the methods
presented. Readers who intend to apply the methods summarized here should
study the primary references cited.
While ground water and surface water are generally thought of as
separate systems, they are highly interdependent components of the hydrologic
cycle. The hydrologic cycle refers to the circulation of water among soil,
ground water, surface water, and the atmosphere. Within a watershed, water
may enter the basin through precipitation, upstream inflow, and ground-water
discharge. Water leaves the watershed through downstream outflow,
evaporation, and ground-water outflow (see Figure 1-1). Some rainwater never
reaches surface water due to the evaporation of intercepted rainfall from
vegetative surfaces and the soil matrix and transpiration of water by plants,
returning water vapor back into the atmosphere.
Rainfall that reaches surface water may travel to the stream or lake as
subsurface storm runoff, overland flow, or ground water. In most humid
environments, about 80 percent of rainfall will infiltrate into the soil
rather than travel by overland flow. Overland flow is more predominant in
semi-arid rangelands, roadways, and cultivated fields in regions with high
intensity rainfall. Rainfall that percolates into the soil matrix is held by
capillary forces. As the soil moisture increases, older soil water is
displaced and percolates laterally and/or vertically. Lateral percolation may
eventually enter streams as subsurface storm runoff, while vertical
percolation generally enters the saturated ground-water zone. Ground water
moves more slowly than subsurface storm runoff and will eventually discharge
and provide water to streams, wetlands, and lakes.
1 See Chapter 3 of this document for a discussion of total maximum daily
load and the waste load allocation process.
-------
Most streams and lakes are surrounded by bank storage zones that
increase during storm events. During a rainstorm, precipitation entering
areas of high soil moisture content which are closer to the stream displaces
the water held in storage, thereby providing the stream with water from bank
storage. In addition, when the topsoil is underlain by a less permeable
horizon, water accumulates above that horizon and flows downhill through the
soil. This represents a shorter route to the stream2.
Saturated ground water enters surface waters in the form of springs or
recharging zones to bank storage. If this ground water is contaminated by
disperse nonpoint sources, the discharging ground water may affect surface-
water quality over a wide area. This paper is concerned with the movement of
nonpoint source contaminants through the ground water saturated zone to
surface water and the methods that have been developed and applied for
measuring this nonpoint source loading to surface water.
In preparing this report, EPA contacted over 100 individuals who are
actively involved in developing or applying methods for measuring nonpoint
source contaminated ground-water discharge to surface water. EPA also
reviewed over 200 papers addressing this topic from the technical and
professional literature. This report represents a synthesis of the
information collected from these sources.
B. Organization of this Report
This report is organized in three chapters with one appendix. Following
this introduction, Chapter 2 presents a summary of the analytical methods
identified by EPA for measuring or estimating nonpoint source contaminated
ground-water discharge to surface water. Chapter 3 presents an overview of
the total maximum daily load assessment and waste load allocation process and
discusses the applicability of the methods described in Chapter 2 to support
these analyses. Finally, Appendix A presents a brief list of EPA guidance
documents relevant to the waste load allocation process. In addition, an
annotated bibliography of the papers that formed the basis for the analysis
presented in Chapter 2 is provided in a companion volume to this document.3
2 Dunne, Thomas, and Luna B. Leopold: Water in Environmental
Planning. W.H. Freeman and Company, 1978, pp. 255-277.
3See "An Annotated Bibliography to the Literature Addressing Nonpoint
Source Contaminated Gound-Water Discharge to Surface Water," September 1990,
EPA 440/6-90-006.
-------
Figure 1-1
>""** *** VX'"%"'
:.**» i**'*v"/«*"*-
I" '"""V"V
Cloud and v*x>r ttor»g»
rtd VlpOf ItOCBQV ^^^^^^^.
S£;trJ--^*\\vs/ ^
iX-*>-^<*v-.. >:;^W. J
^.^£$:»s&L
$£:'£:?. ftyv?*^.*"^ .'P. yi'.''-''-^'^^^^^'/;'^^^.'^^
^>V^/.ffffifeff*W#.«^^ JQftt.ny^
Dunne, Thomas, and Luna B. Leopold: Water in Environmental Planning.
W.H. Freeman and Company, 1978, p. 5.
-------
Chapter 2
Methods for Measuring or Estimating Nonpoint Source Contaminated Ground-Water
Discharge to Surface Vater
Introduction
Seven groups of methods have been identified in the literature for
measuring or estimating contaminated ground-water discharge to surface water.
Each section of this chapter presents a general description of the method,
assumptions and limitations for the method, summarizes the data inputs and
outputs for the method, describes the environmental settings and contaminant
types that have been evaluated using the method, and presents a general
evaluation of the suitability of the method for other applications. Each
section also concludes with tables summarizing information presented in the
annotated bibliography in Appendix A to this document.
A limitation common to all of the methods discussed in this report is
the high degree of uncertainty inherent in the study of ground water. The
heterogeneity of geologic formations presents a major problem in ground-water
study. For example, hydraulic conductivity values can range from 10"1 cm/s to
less than 10"10 cm/s in different geologic settings. Furthermore, hydraulic
conductivities and other hydrogeologic parameters can vary significantly over
even small distances. Thus, errors inherent in ground-water parameter
estimates can vary by 50 percent or more, whereas an acceptable error for
surface-water work is about 10 percent. As a result, the reader should note
that the methods described in this chapter may inherently encompass broad
ranges of uncertainty in their estimates.
A. Studies involving use of seepage meters or mini-piezometers to measure
ground-water discharge to surface water.
The papers cited in this section are summarized in Section I of "An
Annotated Bibliography of the Literature Addressing Nonpoint Source
Contaminated Ground-Water Discharge to Surface Water," September 1990, EPA
440/6-90-006
i. General description of method
a. Description of method or procedure
Seepage meters and mini-piezometers may be used to measure the quantity
and quality of ground water discharging to surface water. These methods
measure a "point-location" ground-water discharge rate and allow for water-
quality sampling over a very small area at the surface-water/sediment
interface. In order to characterize larger areas, several measuring/sampling
points must be selected. Areas with different sediment types may be mapped
and several seepage meters/mini-piezometers installed in each sediment type.
The total discharge and loading rate to the surface-water body as a result of
ground-water discharge can be estimated by applying average measurements per
-------
sediment type to the entire bottom area. Alternatively, reconnaissance over a
large area may be used to identify areas where the greatest quantity of
contaminants is entering the surface water body. Seepage meters and mini-
piezometers may then be used to monitor and quantify discharge zones.
Seepage meters and mini-piezometers have been used to investigate
ground-water discharge into lakes, streambeds, and marine environments. They
are best suited for use in moderately permeable soils and relatively quiet
waters, but adaptations allow successful use under more adverse conditions.
They may be used in combination with one another or with piston corers to
analyze soil permeability, ground-water quality, and ground-water discharge.
An important function of these methods is to provide field verification for
geophysical techniques that may be used to estimate ground-water discharge to
a surface water body (see Section II.C).
Seepage Meter
In its simplest form, a seepage meter can be a 55-gallon drum with the
bottom cut off and a vent hole placed in the closed end. The open end of the
drum is pushed into the bottom sediments of the surface-water body until only
the closed top of the drum is exposed (Lee, 1977). The vent hole remains
unstoppered and the seepage meter equilibrates with the sediment environment.
After several days, a collection system consisting of a tube and a deflated
bag is attached to the vent hole (see Figure 2-1). One can use seepage meters
to estimate discharge velocity of ground water to surface water. Dividing the
collected volume of seepage by the duration of the collection period and by
the area of the seepage meter produces an estimate of ground-water discharge
velocity. Multiplying by the surface area of the stream or lake bottom
estimates the total ground-water discharge rate through that area.
Provided that consideration is given to chemical alteration, seepage
meters might be used to determine ground-water quality from collected seepage
samples. Multiplying the measured chemical constituent concentration in the
seepage by the calculated ground-water discharge rate to the surface-water
body estimates the constituent's loading to the surface water (Goodman et al.,
1989).
Mini-piezometer
Description and installation
Piezometers are devices consisting of pipes with slotted tips or well
points on the end. They are used to measure hydraulic head in saturated
geologic materials. Piezometers are usually installed in machine-drilled
boreholes. Mini-piezometers are similar to piezometers, but are smaller in
size and installed manually. A mini-piezometer consists of a small-diameter
tube perforated over a short distance at one end. Nylon mesh covering the
perforated tube keeps sediment from clogging the mini-piezometer. To place a
mini-piezometer, a length of thick-walled pipe, with an inside diameter
slightly larger than the tubing is hammered into the sediment. A temporary
-------
Figure 2-1
water surface
Full section view of seepage meter showing proper
placement in the sediment. A. 4 liter, 0.017 mm membrane
plastic Baggies Alligator bag (open end was heat sealed); B,
rubber-band wrap; C. 0.64 cm inside diameter. 6 cm long,
polyethylene tube: 0, 0.79 cm inside diameter. 4.5 cm long,
amber-latex tube; F. 15 cm x 57 cm diameter epoxy-coated
cylinder (end-section of a steel drum).
Lee, David R., and John A. Cherry: "A Field Exercise on Ground-Water
Flow Using Seepage Meters and Mini-Piezometers," Journal of Geologic
Education. 1978, Volume 27: p. 8.
-------
plug attached to the end of the pipe keeps sediment from entering the pipe
during placement. The plug is knocked free before the mini-piezometer is
inserted to the bottom of the pipe, perforated tip first. The mini-piezometer
tube is held in place as the length of pipe is removed. The pipe used in
installation may be pulled back to expose the perforated section but left in
place to provide added protection to the tubing. A collection system similar
to the seepage meter tube and collapsible bag system can be used to collect
seepage samples (see figure 2-2).
Mini-piezometer variations
One difficulty often associated with the installation of mini-
piezometers is their tendency to move before the sediment collapses around the
tube or if the tube is pulled later on. Lee and Welch (1989) have tested a
harpoon piezometer which helps to alleviate this problem (see Figure 2-3).
"Barbs" on the tip of the piezometer grip the sediment and help to keep the
screen at the desired depth as the driving rod, or pipe, is withdrawn or if
the screen in moved during use.
Another variation of the mini-piezometer is the bundle-type mini-
piezometer, consisting of several small tubes placed within the pipe at one
time. The tubes are placed at selected depths to allow detailed vertical
resolution of head and pore-water chemistry at the selected mini-piezometer
location. If bundle-type mini-piezometers are placed at selected points along
a vertical plane, patterns of flow and geochemical processes in the subsurface
are observable.
Mini-piezometer measurements
An alternative to direct measurement of ground-water discharge is to use
hydraulic conductivity and hydraulic head data obtained from mini-piezometers
to calculate the ground-water discharge rate to surface water using Darcy's
Law. Comparing the hydraulic head in the mini-piezometer with the hydraulic
head of the surface-water body determines the hydraulic gradient across bottom
sediments. Hydraulic head differential may be measured using a manometer (see
Figure 2-4) or a continuous water level recorder (see Figure 2-5). Head
differential is divided by the depth of the piezometer screen below the
sediment-water interface to obtain the vertical hydraulic gradient. Hydraulic
conductivity of the bottom sediments may be estimated or be measured using
either a constant head or falling head test. A constant head test has been
developed using sections of sediment cut directly from a thin-walled piston
core barrel (Munch and Killey, 1985). Once the hydraulic gradient and
hydraulic conductivity have been determined, Darcy's Law may be used to
determine the ground-water flux through the sediment (see Section A.i.d).
Multiplying the calculated flux by the surface area of the surface-water
bottom yields the ground-water discharge rate to the surface-water body.
The mini-piezometer yields seepage samples using a syringe or other
sampling device. Multiplying the measured chemical constituent concentration
in the ground water by the calculated ground-water discharge rate yields the
loading rate to surface water.
-------
Figure 2-2
General features and method of installation of a mini-
piezometer. A, casing driven into the sediment; B, plastic tube
with screened tip inserted in the casing; C, plastic tube is a
piezometer and indicates differential head (h) with respect to the
surface water: 0, plastic bag attached to the piezometer collects
sediment-porewater.
Lee, David R. , and John A. Cherry: "A Field Exercise on Ground-Water
Flow Using Seepage Meters and Mini-Piezometers," Journal of Geologic
Education. 1978, Volume 27: p. 7.
-------
Figure 2-3
1/4" O.D.
polyethylene
tubing
1/4" perforations
3/8" 0.0. polyethylene tube
groove to fit
drive rod
polyethylene tip
^ xstainle*B-tteel wire
S to tecure two tubes
Harpoon piezometer tip, screen and tube. Dimensions for the small
type are shown here. The screen is 10 cm long, has 8 1/4" diameter
perforations and is covered with 3 layers of 240 pm mesh tightly
rolled around the 3/8* O.D. polyethylene tube to prevent entry of
sediment. The drive rod, not shown, fits loosely in groove. The
barbs" are folded back before driving in sediment to ensure that
they grip in the sediment.
Lee, D.L. and S.J. Welch: "Methodology for Locating and Measuring
Submerged Discharges: Targeting Tool, Harpoon Piezometer and More," FOCUS
Conference on Eastern Regional Ground Water Issues," Kitchener, Ontario,
Canada, October 17-19, 1989, p. 8.
-------
10
Figure 2-4
rubber b«lb
B
vet Ion'
9mm ID
el«*r plastic
M«t«r stick
to fit
The manometer used to measure differential heads in
minipiezometers. A, principle of operation: B, the field ap-
paratus.
Lee, David R., and John A. Cherry: "A Field Exercise on Ground-Water
Flow Using Seepage Meters and Mini-Piezometers," Journal of Geoloeic
Education. 1978, Volume 27: p. 8.
-------
11
Figure 2-5
75 cm 10 sfsndpipe
Continuous measurement of head differences between
piezometric level and river level using two water-level
recorders.
Lee, David R., and Stephen J. Welch: "A Method for Installing and
Monitoring Piezometers in Beds of Surface Waters," Ground Water. Volume 27(1)
p. 89.
-------
12
In geologic units having permeabilities too low to allow easy withdrawal
of water from a piezometer tip, a piston corer may be used to obtain a
continuous vertical profile of sediment (Lee, 1988). Munch and Killey (1985)
have used a modified piston corer featuring a thin-wall core barrel and
wireline recovery to sample both cohesive and cohesionless sediment from
depths up to 30 m below the water table. The porewater can then be extracted
from the piston core for chemical analysis.
b. Assumptions involved in using seepage meters and mini-
piezometers
Flow rate is uniform through the sampling interval
Seepage meters and mini-piezometer infer that measured ground-water
quantity and quality are representative of actual conditions throughout the
sampling interval. The ground-water discharge rate recorded using a seepage
meter or mini-piezometer represents the average ground-water discharge rate
for the collection period. If a single discharge measurement or a series of
discharge measurements recorded over a short time period are used to determine
the ground-water discharge rate to a surface-water body, the calculated
discharge rate may vary from actual rates.
Sampled interval is representative temporally
A series of discharge measurements taken over a short duration can vary
substantially due to tidal cycles, storm events, and seasonal changes.
Furthermore, seasonal variations within and between years may be substantial.
It is possible that a series of discharge measurements recorded over a long
duration may not be representative of actual conditions if the measurements
were recorded in excessively wet or dry years.
Sampling placement is representative spatially
Seepage meters and mini-piezometers provide point measurements that
determine the ground-water discharge rate and loading rate to a surface-water
body through extrapolation (Goodman et al., 1989). The representativeness of
the sampling locations and the number of locations influence the accuracy of
the results. For example, Belanger and Connor (1980) not only found
decreasing seepage rates with increasing distance from shore, but also that
ground-water recharge occurred toward the center of East Lake Tohopekaliga.
An overestimation of ground-water seepage would result if seepage meters used
in the study were all located near shore. Conversely, if all the seepage
meters were located toward the center of the lake, one would erroneously
conclude that the entire lake was recharging ground water.
Measured samples are representative of ground-water discharge
quality
Measured ground-water quality may not be representative of actual
conditions because of interactions occurring at the sediment/surface-water
interface. Belanger and Mikutel (1985) concluded that direct determination of
water quality using seepage meters overestimated nutrient loading to lakes due
-------
13
to the enclosure of bottom sediments, which results in anaerobic conditions
and increased release rates of ammonium, nitrogen, and phosphate. In
addition, seepage meter or mini-piezometer samples from shallower, shoreline
locations may be influenced by bank storage water.
c. Limitations of the methods
Placement on different surface-water bottoms
Not all bottom areas of a surface-water body are conducive to
installation of seepage meters or mini-piezometers. Because seepage meters
and mini-piezometers require insertion into bottom sediments, ideal
installation locations are areas with relatively soft, fairly thick,
moderately permeable sediments containing few cobbles or stones. However,
German has successfully installed seepage meters in cobbles and rocks using
bentonite placement.4
Deep surface waters require additional expertise and equipment
Seepage meters and mini-piezometers located in deep water require scuba
abilities and equipment for installation, sampling, and maintenance. Depths
that limit divers' safe performance control installation depths (Woessner and
Sullivan, 1983). Additionally, some bottom locations are not suited to
installation of seepage meters and mini-piezometers.
Strong currents and harsh seasons
Without modification, mini-piezometers and seepage meters should not be
used with strong currents. Acceptable installation locations vary with
seasons in areas due to wave and current action. Additionally, in colder
climates, ice covering surface waters may limit seepage meters and mini-
piezometers sampling and maintenance activities.
Sklash has overcome some of these problems.5 In his investigation,
handles placed on seepage meters aided divers in fast currents. Also, once
the seepage meter is placed, bolts are used to clamp it down and ensure its
stability. To protect seepage bags from the elements, Sklash used rapid
disconnects for the bags and placed rigid containers around them.
Maintenance
Seepage meters and mini-piezometers equipped with sample collection
* German, Dave, personal communication, Nonpoint Source Contaminated
Ground-water Discharge to Surface Water Workshop, Chicago, IL, November 30,
1989.
5 Sklash, Mike, personal communication, Nonpoint Source Contaminated
Ground-water Discharge to Surface Water Workshop, Chicago, IL, November 30,
1989.
-------
14
devices require substantial maintenance. Without frequent changes, the
increased pressure associated with a full catchment device reduces the amount
of ground-water flow into the seepage meter. The sample collection tubing for
seepage meters and mini-piezometers requires regular cleaning or replacement
to prevent algae growth. There may be a need to periodically replace seepage
meters and mini-piezometers due to wave and current action.
Anaerobic environment within equipment
A final limitation of seepage meters and mini-piezometers is that
anaerobic conditions develop within the seepage meter, altering the chemistry
of the discharging ground water (Belanger and Mikutel, 1985). As a result,
calculated loading rates to surface water, determined using seepage meters,
may not be representative of actual seepage from the ground water.
d. Representative equations
Darcy's Law is used to calculate the ground-water discharge rate to
surface water when using mini-piezometers. The form of Darcy's Law used is:
Q - K(dh/dl)A
where
Q - Ground-water discharge rate [L3/T]
K - Hydraulic conductivity [L/T]
dh/dl - Hydraulic gradient [L/L]
A - Surface area of the bottom of the surface-water body [L2] .
The loading rate to surface water as a result of ground-water discharge is:
LR - QC
where
LR = Loading rate [M/T]
Q = Ground-water discharge rate [L3/T]
C = Chemical constituent concentration in ground water [M/L3] .
e. Description of field equipment
Equipment and materials often used for installation, sampling, and
maintenance of seepage meters include:
open-ended 55-gallon drum with vent hole,
tubing,
plastic seepage bag,
boat, and
scuba gear.
Typical equipment and materials required for installation, sampling, and
-------
15
maintenance of mini-piezometers include:
metal pipe,
end plugs,
tubing,
nylon mesh,
hammer,
sample bag, and
boat.
For a detailed description of seepage meters installation and sampling, see
Lee (1977). Installation and sampling of seepage meters in deeper, more
turbulent water often requires additional equipment.
f. Expertise needed to apply method
Siting sampling locations requires a sufficient understanding of
regional geology and hydrology. Extrapolating sampling results also requires
expertise in geostatistical methods needed to delineate areas of similar
sediment type for seepage meter or mini-piezometer location. In near shore
areas where wading is possible, seepage meters and mini-piezometers are
relatively easy to install and maintain. Farther from the shore, installation
of mini-piezometers may be made from a surface platform as described by Welch
and Lee (1989). The surface platform sits on top of two 14 foot boats, and
consists of plywood flooring, 2 by 4's, ropes, bolts, and a central reinforced
joist (see Figure 2-6). As water depth increases, seepage meter and mini-
piezometer installation and maintenance may require scuba divers. The need to
frequently change and check sample collection bags renders the method labor
intensive, especially with dive team involvement.
ii. Data inputs for the method
Extrapolation of sample results to areas with similar sediment
characteristics requires knowledge of the spatial distribution of bottom
sediments. If mini-piezometers are used, an estimate of the hydraulic
conductivity of the bottom sediment is needed to determine the ground-water
discharge rate.
ill. Outputs from the method
Seepage meters and mini-piezometers provide a direct measurement of
ground-water discharge to surface water at a given location. In conjunction
with data characterizing the areal distribution of sediment types, these data
-------
16
Figure 2-6
SIDE VIEW
plywood box beam on
center joist only
rope tied to
2x4 joist
PLAN VIEW
platform
eye bolt (3/8 in < 8 m)
joists
aluminum boat
polypropylene
rope (5/6 in did)
20 cm x 20 cm hole
joist (2xt mx13 ft)
(3-4x8 ft 5/8 m thick,
T & G plywood)
14 ft aluminum boat
Side and plan view. Working platform for installing
piezometers and coring sediments from the water surface.
Lee, David R., and Stephen J. Welch: "A Method for Installing and
Monitoring Piezometers in Beds of Surface Waters," Ground Water. Volume 27(1)
p. 87.
-------
17
can provide estimates of the total loading from ground-water seepage.
a. Ground-water quantity discharge to surface water
A wide range of seepage rates can be measured using seepage meters and
mini-piezometers. For example, in Minnesota and Nova Scotia, seepage
velocities have been measured in the range of 0.1 to 2.58 pm sec"1 using the
simple seepage meter described above (Lee, 1977).
b. Ground-water quality discharge to surface water
Seepage meters allow for the collection of samples for water quality
analysis. The quantification limit for the sample is a function of the
detection limits for the constituent of concern. The anaerobic conditions
within these sampling devices may affect sample integrity.
iv. Settings in which the method has been applied and
contaminants that have been measured using this method
Summaries in Table A-l describe some of the locations where seepage
meters and mini-piezometers have been used successfully and the contaminants
measured using the method.
v. Evaluation of the method
Seepage meters and mini-piezometers provide a simple, direct method to
measure the quantity and quality of ground-water discharge to surface water.
As with any ground-water method, the point measurements obtained represent
moments in time and space for estimates of the quantity and quality of ground-
water discharge to surface water. The inherent variability of most earth
materials will require a large number of spatially distributed measurements in
order to characterize discharge and loading rates accurately. Use of this
method over large areas will require a substantial commitment of resources.
However, the fact that sampling and flow measurements are made in or near the
surface-water body can provide an accurate indication of the contaminant
inflows without the necessity of installing monitoring wells on land and then
extrapolating results to the points of discharge.
vi. References to annotated bibliography
References to the annotated bibliography presented in the accompanying
volume to this document are provided in Table A-2.
-------
18
Table A-l. Sumazy of Settings In Nhicb the Method Has Been Applied and the Contaminants Measured.
Location
Aquifer
Contaminant
Author
Ontario
Sites near Leamington,
Ontario and at Cape Cod
Massachusetts
Chalk River, Ontario
Key Largo National
Marine Sancturay, Florida
Key Largo National
Marine Sanctuary, Florida
Osceola County, Florida
Wisconsin
Michigan and Wisconsin
Brevard County, Florida
Upper Great Lakes
Connecting Channels
Orlando, Florida
South Dakota
Colorado
Floridan Aquifer
Shallow glacial,
glacial bedrock inter-
face and bedrock units.
Nitrates
(Ontario)
Pesticides
Heavy Metals
Nitrates, selected
cations, total phosphates
Phosphorous, Nitrogen
Nitrate, Phosphorous,
Ammonia
Chloride
Phosphorous, Chromium,
Lead, Barium, Zinc, Cobalt
Nickel, Phenols
Nitrate
Nitrogen, Phosphorous,
Pesticides
D. R. Lee, S. J. Welch
D. R. Lee
J. H. Munch, R. W. Douglas
6. M. Simmons Sr.
F. G. Love
G. M. Simmons Jr.
J. Netherton
T. V. Belanger
D. F. Mikutel
T. D. Brock, D. R. Lee,
D. Janes, D. Winek
D. A. Cherkauer,
J. M. McBride
J. N. Connor, T.V. Belanger
EPA Non Point Source Work
Group
C. R. Fellows, P. L. Brezonik
J. Goodman et al.
J. W. LaBaugh, T.C. Winter
Eastern Ontario
D. R. Lee, J. A. Cherry,
J. F. Pickens
Eastern Ontario
Tritium
D. R. Lee, J. A. Cherry
-------
19
Table A-l. Summary of Settings in Which the Method Has Been Applied and the Contaminants Measured. (Continued)
Location
Minnesota, Wisconsin
North Carolina, Nova Scotia
Southern Ontario
Barbados, West Indies
Southeastern Virginia
Minnesota
Holbrook, Massachusetts
Mahantago Creek,
Pennsylvania
Chicago, Cook County,
Illinois
Aquifer Contaminant
Phosphates, Nitrates,
Ammonia, Chloride
Nutrients
Barbados Aquifer Nitrogen, Phosphorous
Shirley, Yorktown, Inorganic Nitrogen
and Tabb Formations
Phosphorous, Nitrogen
Volatile Organics and
Inorganics
Manhantago Creek Basin Nitrogen, Phosphorous
Cambrian and Ordovician E.P. Metals
Aquifers
D.
D.
J.
W.
G.
W.
G.
J.
W.
J.
H.
N.
M.
et
R.
R.
B.
G.
H.
G.
M.
K.
R.
S.
B.
J.
S.
al
Author
Lee
Lee, H.B.N. Hynes
Lewis
Maclntyre,
Johnson ,
Reay,
Simmons , Jr .
Neel, R. M. Brice
Norman, D. P. Ostyre,
Bob in
Poinke, N. J. Gburek,
Gburek et. al.
Henebry, M. Demissie,
Virginia's Eastern Shore
Sault Ste. Marie, Ontario
Wast Thorton, New Hampshire
Lake Mead, Neveda
East Coast of Florida
Tertiary-Cretaceous
Gale-Hills Formation
Nitrate, Ammonia, Total
Phosphorous
TDS, calcium-sulfate
Phosphate
G. M. Simmons, Jr.
S. J. Welch, 0. R. Lee
T. C. Winter
W. W. Woessner, K. Sullivan
C. F. Zimmerman,
J. R. Montgomery,
P. R. Carlson
-------
Table A-2. References to Annotated Bibliography
20
Author
Citation
Reference to Annotated
Bibliography
T. V. Belanger, D. F. Mikutel
R. Carr, T. C. Winter
D. A. Cherkauer, J. M. McBride
J. N. Connor, T. V. Belanger
EPA Non Point Source Work Group
C. R. Fellows, P. L. Brezonik
J. Goodman et al.
J. W. LaBaugh, T. C. Winter
"On the Use of Seepage Meters to Estimate pp.2-3
Ground-Water Nutrient Loading to Lakes,"
Water Resources Bulletin. 1985, Volume
21(2): 265-272.
"An Annotated Bibliography of Devices Developed p. 8
for Direct Measurement of Seepage," U.S. Geological
Survey Open File Report 80-344, 1980.
"A Remotely Operated Seepage Meter for Use in Large pp.9-10
Lakes and Rivers," Ground Water. 1988, 26(2):
165-171.
"Ground Water Seepage in Lake Washington and the Upper pp.11-12
St. Johns River Basin, Florida," Water Resources Bulletin,
1981, 17(5): 799-805,
"Upper Great Lakes Connecting Channel Study, Waste Disposal pp.13-17
Disposal Sites and Potential Ground Water Contamination,
St. Clair River," Non Point Source Work Group Report,
April, 1988.
"Fertilizer Flux into Two Florida Lakes Via Seepage," pp.18-19
Journal of Environmental Quality. 1980, Volume 10(2):
174-177.
"Oakwood Lakes - Poinsett: Rural Clean Water Program pp.20-21
Comprehensive Monitoring and Evaluation Technical Report,
Project 20," Rural Clean Water Program Comprehensive
Monitoring and Evaluation Technical Report, Project 20,
May, 1989.
"In Impact of Uncertainties in Hydrologic Measurement on pp.22-23
Phosphorous Budgets and Empirical Models for Two Colorado
Reservoirs," Limnology and Oceanography. 1984, Volume 29(2):
322-339.
-------
21
Table A-2. References to Annotated Bibliography (Continued).
Author
Citation
Reference to Annotated
Bibliography
D. R. Lee
D. R. Lee, S. J. Welch
T. D. Brock, D. R. Lee,
David Janes, David Winek
D. R. Lee, J. A. Cherry, J. F. Pickens
D. R. Lee, J. A. Cherry
D. R. Lee
D. R. Lee, H.B.N. Hynes
J. B. Lewis
"Six In-Situ Methods for Study of Groundwater Discharge,"
Proceedings of the International Symposium on Interaction
Symposium on Interaction Between Groundwater and Surface
Hater, 30 May-3 June, 1988, Ystad, Sweden, edited by
Peter Dahlblom and Gunner Lindh, Department of Water
Resources Engineering, Lund University, Sweden.
"Methodology for Locating and Measuring Submerged
Discharges: Targeting Tool, Harpoon Piezometer and
More," FOCUS Conference on Eastern Regional Ground Water
Issues: October 17-19, 1989, Kitchener, Ontario,
Canada, Co-sponsored by the Association of Ground Water
Scientists and Engineers, Division of NWWA and Waterloo
Center for Groundwater Research, University of Waterloo.
"Ground-Water Seepage as a Nutrient Source to a Drainage
Lake; Lake Mendota, Wisconsin,"
"Ground-Water Transport of a Salt Tracer through a
Limnology and Oceanography. 1980, Volume 25(1):
45-61.
"A Field Exercise on Ground-Water Flow Using
Seepage Meters and Mini-piezometers," Journal of
Geological Education. 1978, Volume 27: 6-10.
"A Device for Measuring Seepage Flux in Lakes
Lakes and Estuaries," Limnology and Oceanography.
1977, Volume 22(1): 140-147.
"Identification of Groundwater Discharge Zones in a
Reach of Hillman Creek in Southern Ontario," Water
Pollution Research Canada. 1978, 13: 121-133.
"Measurements of Ground-Hater Seepage Flux onto a
Coral Reef: Spatial and Temporal Variations,"
Limnology and Oceanography. 1987, 32(5):
1165-1169.
pp.27-29
pp.37-39
pp.6-7
pp.32-34
pp.30-31
pp.24-26
pp.35-36
pp.40-41
-------
22
Table A-2. References to Annotated Bibliography (Continued).
Author
Citation
Reference to Annotated
Bibliography
W. 6. Maclntyre, G. H. Johnson,
H. 6. Reay, G. M. Simmons, Jr.
J. K. Heel, R. M. Brice
H. R. Norman, D. F. Ostrye,
J. S. Hobin
H. B. Fionke, J. R. Hoover,
R. R. Schnabel, W. J. Gburek,
J. B. Urban, A. S. Rogowski
F. E. Ross, M. S. Henebry,
J. B. Risatti, T. J. Murphy,
M. Demissie County, Illinois
G. M. Simmons Jr., F. G. Love
G. M. Simmons Jr., J. Netherton
"Ground-Water Non-Point Sources of Nutrients to the
Southern Chesapeake Bay," Proceedings of Ground
Hater Issues and Solutions in the Potomac River
Basin/Chesapeake Bay Region. Co-sponsored by the
Association of Ground Hater Scientists and Engineers,
pp. 83-10*.
"Watershed and Point Source Enrichment and Lake
State Index," US EPA, April 1979, EPA-600/3-79-046.
"Use of Seepage Meters to Quantify Ground-Water Discharge
and Contaminant Flux into Surface Water at the Baird and
McGuire Site (NFL No. 14)," Proceedings of Third Annual
Eastern Regional Ground Water and Conference. 1986,
p. 472-491.
"Chemical-Hydrologic Interactions in the Near-Stream Zone,"
Zone," Water Resources Research. 1988, Volume 24(7):
1101-1110.
"A Preliminary Environmental Assessment of the Contamination
Associated with Lake Calumet Cook Hazardous Waste Research and
Information Center. Illinois State Water Survey, 1988, HWRIC
RR-019, 88/300.
"Water Quality of Newly Discovered Submarine Ground Water
Discharge into a Deep (Coral Reef Habitat," NOAA Symposium
series for Undersea Research. Volume 2(2): 155-163.
"Groundwater Discharge in a Deep Coral Reef Habitat:
Evidence for a New Biogeochemical Cycle?" Diving for
Science...86, Proceedings of the Sixth Annual Scientific
Diving Symposium (1986), Tallahassee, Florida, Charles T.
Mitchell, editor.
pp.42-43
pp.44-46
pp.47-49
pp.50-52
pp.53-55
pp.60-61
pp.62-64
-------
23
Table A-2. References to Annotated Bibliography (Continued).
Author
Citation
Reference to Annotated
Bibliography
G. M. Simmons, Jr.
6. M. Simnons, Jr.'
S. J. Welch, D. R. Lee
T. C. Winter
W. W. Woessner, K. Sullivan
C. F. Zimmerman, J. R. Montgomery,
P. R. Carlson
"Understanding the Estuary Advances in Chesapeake
Research," Proceedings of a Conference, March 29-31,
198, Baltimore, Maryland, Chesapeake Research Consortium
Publication 129. CBP/TRS 24/88.
"The Chesapeake Bay's Hidden Tributary: Submarine Ground-
water Discharge," Proceedings of Ground Water Issues and
Solutions in the Potomac River Basin/Chesapeake Bay Region.
Co-sponsored by the Association of Ground Water Scientists and
Engineers, pp. 9-29.
"A Method for Installing and Monitoring Piezometers in Beds of
in Beds of Surface Haters," Ground Water. 1989 27(1): 87-90.
"Geohydrologic Setting of Mirror Lake, West Thorton, New Hampshire,
1984, U.S. Geological Survey Water Resources Investigations Report,
84-4266, 61 pp.
"Use of Seepage Meters and Mini-piezometers for Identifica-
tion of Reservoir - Groundwater Interactions in Lake Mead,
Neveda," Desert Research Institute Water Resources Center,
1983, PB 83-226894.
"Variability of Dissolved Reactive Phosphate Flux Rates in
Nearshore Estuarine Sediments," Estuaries. 1985, 8(2B): 228-236.
pp.56-57
pp.58-59
pp.65-66
pp.67-68
pp.69-70
pp.71-72
-------
24
B. Ground-water quality sampling and measurements of ground-water flow to.
estimate loading to surface water
The papers cited in this section are summarized in Section VIII of "An
Annotated Bibliography of the Literature Addressing Nonpoint Source
Contaminated Ground-Water Discharge to Surface Water," September, 1990, EPA
440/6-90-006.
i. General description of method
a. Description of procedures
Water-level elevation measurements from piezometers and ground-water
wells provide an indication of the quantity of ground water discharging to
surface water in a watershed. This method uses water-level measurements
obtained from wells located in the watershed to develop a water table contour
map. Darcy's Law is then applied to calculate the discharge rate of ground
water to surface water by incorporating estimates of hydraulic conductivity
and the cross-sectional area of the aquifer. By assuming the aquifer
underlying the watershed is homogenous and isotropic, the water-level contour
map can be used to determine flow directions and horizontal gradients in the
basin. In a homogeneous, isotropic aquifer, flow lines will be perpendicular
to equipotentials. Hydraulic conductivity can be estimated for a particular
rock or soil type or can be measured in-situ via aquifer tests. Aquifer
geometry is estimated by examining lithologic logs of wells in the watershed.
This method is often used in conjunction with other methods, such as
mini-piezometers, seepage meters, tracer studies, isotopic studies, or water
and mass balances analyses to verify study results. This method has been
practiced in both marine and fresh water environments and has been used on a
large scale, such as in Long Island (Franke and McClymonds) and North Central
Kansas (Spruill) and on a smaller scale such as in South Farmingdale, New York
(Perlmutter and Lieber) and the Stockett-Sand Coulee coal field, Montana
(Osborne, et.al.). This method has been used for glacial and dolomite
aquifers. Contaminants studied include metals, nutrients, and some organic
constituents.
Ground-water samples taken from wells within the basin can be used to
characterize the spatial distribution of ground-water quality as a means of
estimating nonpoint source contaminant load. To properly characterize ground-
water quality in a drainage basin, potential nonpoint source loading areas
should be identified and the underlying ground water sampled. Agricultural
areas located on soils allowing rapid infiltration of precipitation are of
particular concern. Such areas are identified from soil and land use maps
(Hallberg et al., 1983). Evaluation of ground-water quality beneath nonpoint
source loading areas over time will indicate qualitatively whether the loading
rate to surface water, as a result of ground-water discharge, will increase or
decrease in the future.
Ground-water quality in wells adjacent to the surface-water body are
assumed to be representative of ground water discharging to surface water. By
using the constituent concentrations and the calculated ground-water flux, the
-------
25
immediate loading rate to surface water from ground-water discharge can be
calculated.
b. Assumptions involved in these methods
To use Darcy's Law to calculate ground-water discharge to surface water,
it is assumed that the aquifer is homogeneous, isotropic, of constant
thickness, and that flow is horizontal. The assumption that the aquifer is
homogeneous and isotropic is necessary to ensure that flow lines are
perpendicular to equipotentials (Perlmutter and Lieber, 1979). By assuming a
constant aquifer thickness and horizontal flow, the one-dimensional version of
Darcy's Law can be used. Additionally, water quality in sampled wells is
assumed to be representative of the quality of the water discharging to the
stream.
c. Limitations of the method
Aquifer characteristics
The limitations of this method reflect the natural variability of
aquifers and the availability of information on aquifer characteristics. In
nature, considerable heterogeneity exists and few, if any, aquifers are
homogeneous, isotropic, and of constant thickness. In a heterogeneous,
anisotropic aquifer, ground-water flow is not perpendicular to equipotentials,
and the angle between flow direction and equipotentials is not constant.
Because the predicted flow path length differs from the actual flow path
length, the calculated hydraulic gradients will not be representative of the
actual gradient. Additionally, horizontal gradients determined using wells
screened at different depths below the water table or in different geologic
formations may not represent the actual horizontal gradient. Temporal changes
in the hydraulic conductivity due to changes in seepage face from
precipitation events increase the difficulty of estimating an average ground-
water discharge. Hydraulic conductivity also varies spatially and
directionally in a heterogeneous, anisotropic aquifer. It would be difficult,
if not impossible, to determine an accurate equivalent hydraulic conductivity
and aquifer thickness for the basin. Because of the difficulties in
determining horizontal hydraulic gradient, hydraulic conductivity, and aquifer
thickness, precise determination of ground-water discharge to surface water is
problematic (Koszalka, 1983).
Well installation
Installing the number of wells needed to properly characterize ground-
water quality in a watershed is resource intensive. As an alternative to
installing costly monitoring wells, existing production, domestic, and stock
wells may be sampled. In many cases, however, these wells will not be in
optimum locations or open to the geologic formation of interest.
Additionally, water-quality results can be altered by well construction
materials, faulty well construction, and sampling procedures. Consequently,
ground-water quality in the basin may not be accurately characterized due to
the construction and location of the well.
-------
26
d. Representative equations
Darcy's Law is used in determining the quantity of flow entering surface
water. The form of Darcy's Law used is:
Q - K(dh/dl)A
where
Q - Ground-water discharge rate (L3/T)
K - Hydraulic conductivity (L/T)
dh/dL - Hydraulic gradient (L/L)
A - Cross sectional area of the aquifer (L2).
The loading rate to surface water, as a result of ground-water discharge, is
determined by multiplying the ground-water discharge rate by the concentration
of the constituent in the ground water. The loading rate equation is:
LR - QC
where
LR = Loading rate (M/T)
Q - Ground-water discharge rate (L3/T)
C - Concentration in ground-water (M/L3) .
e. Description of field equipment
After the wells have been installed, equipment is required to measure
water levels and obtain ground-water samples. Suggested equipment includes:
steel tape and chalk or electric well sounder,
submersible pump,
centrifugal pump,
bailer,
pH meter,
conductivity meter,
thermometer,
appropriate sample containers, and
coolers.
f. Expertise needed to apply method
This method involves an initial review of well log records for existing
wells in the watershed to evaluate whether the existing well network can
properly characterize the hydrogeology and water quality of the watershed
(Hallberg et al., 1983). If more wells are needed, the number, locations, and
depths are determined. After the well network for the watershed has been
identified, water-level measurements and ground-water samples are obtained.
Obtaining water-level measurements and ground-water samples is relatively
easy, but often labor intensive. The water-level measurements are used to
construct a water-table contour map from which the horizontal hydraulic
-------
27
gradient is determined. The greatest difficulty in applying the method is in
determining aquifer characteristics and geometry. Aquifer tests are labor
intensive and sometimes difficult to interpret. Additionally, aquifer
characteristics determined for one portion of a watershed must be extrapolated
to other portions of the watershed (Koszalka et al., 1985).
ii. Data inputs
To apply Darcy's Law and estimate the ground-water loading rate for a
watershed, level elevations, hydraulic conductivity, aquifer geometry, and
chemical constituent concentrations in ground water are needed. Well
construction information is essential to determine the subsurface zones that
the hydraulic head and water quality measurements represent.
ill. Outputs from the method
a. Quantity of ground water
A wide range of ground-water discharge to surface water rates can be
estimated using this method. The factors controlling the quantity of ground-
water discharge to surface water are hydraulic conductivity and gradient. If
the hydraulic conductivity of the aquifer is low, and the hydraulic gradient
across the aquifer is minimal, the ground-water discharge rate to surface
water will also be low. Conversely, a high ground-water discharge rate to
surface water will occur when the hydraulic conductivity and gradient for an
aquifer are high.
b. Quality of ground water
A broad range of loading rates to surface water as a result of ground-
water discharge can be predicted using this method. The ability to determine
ground-water quality in sampled wells is limited only by the characteristics
of the well materials and the quantitation limits for the individual
constituents.
iv. Settings in which the method has been applied and
contaminants that have been measured using this method
Some of the locations where this method has been used and the
contaminants that have been measured using the method are summarized in Table
B-l.
v. Evaluation of the method
Most watersheds contain observation or water supply wells that can be
used to obtain water level elevations and water quality data, making this
method applicable in many locations. The method can qualitatively determine
the amount of ground-water discharge and the loading rate to surface water
-------
28
within a watershed. Increasing the number of sampling locations will improve
the predictive capabilities of the method.
Because the method is often applied with limited knowledge of aquifer
characteristics, a large number of sampling points will not necessarily result
in accurate quantification of ground-water discharge or loading rates to
surface water. A comparable qualitative indication of the loading rate to
surface water as a result of ground-water discharge can be obtained by
observing water-quality trends in the watershed. If, within a watershed,
chemical constituent concentrations in ground water increase with time, the
future loading rate to surface water as a result of ground-water discharge can
also be expected to increase. In addition to using the method to
quantitatively predict ground-water loading rates to surface water, this
method can be used to qualitatively assess the impacts of various regulatory
scenarios governing fertilizer and pesticide usage on ground-water quality.
vi. References to annotated bibliography
References to the accompanying annotated bibliography are located in
Table B-2.
-------
Table B-l. Summary of Settinsa in Which the Method Has Been Applied and the Contaminants Measured.
Location
Aquifer
Contaminant
Author
Patchogue,
Long Island, New York
Penfield. New York
Upper Great Lakes
Connecting Channels
Lockport dolomite
Shallow glacial,
glacial bedrock
interface, & bed-
rock units
Nitrate Nitrogen
Sodium Chloride
Zinc, Phenols,
Phosphorous
D. Capone, M. Bautiota
L. R. Davis
EPA Non-Point Source
Group
Long Island, New York
Clayton City, Iowa
Perth, Australia
Niagara County,
New York
Stockett and Sand
Coulee, Montana
Nassau County,
New York
Upper Glacial,
Magothy, and Lloyd
Galena
Kootenai Formation,
Morrison Formation
Magothy
IDS, Inorganic
Metals
Nitrates, Herbicide
Pesticides, Bacteria &
turbidity
Nitrate
Inorganic & Organic
Constituents
Heavy Metals
Cadmium, Chromium
0. L. Franke, N. E. McClymonds
G. R. Hallberg, B. E. Hoyer
E. A. Bettis, III, R. D. Libra
R. E. Johannes
E. J. Koszalka, J. E. Paschal
T. S. Miller, P. B. Duran
T. J. Osborne, J. L. Sonneregger,
J. J. Donovan
N. M. Perlmutter,
M. Lieber
North Central Kansas
Almena,
Kansas Bostwick,
Cedar Bluff Units
Sulfate, Sodium
Chloride, Calcium
T. B. Spruill
Butte, Mead and
Lawrence, South Dakota
Arkansas River Basin
Schwatka Lake, Yukon
Territory, Canada
Arsenic,
Selenium
TDS, Salinity,
Chloride
Nitrogen,
Phosphorous
R. L. Stach, R. N. Belgerson,
R. F. Bretz, M. J. Tipton,
D. R. Biessel, J. C. Harksen
J. D. Stoner
F. B. Whitfield, B. McNaughton,
H. G. Hhitley
-------
Table B-2. References to Annotated Bibliography
30
Author
Citation
Reference to Annotated
Bibliography
D. Capons, M. Bautista
L. R. Davis
EPA Non Point Source Group
0. L. Franke, N. E. McClymonds
G. R. Hallberg, B. E. Hoyer,
E.a. Bettis, III, R. D. Libra
R. E. Johannes
E. J. Koszalka, J. E. Paschal,
T. S. Miller, P. B. Duran
I. J. Osborne, J. L. Sonneregger,
J. J. Donovan
N. M. Perlmutter, M. Lieber
"A Ground-Water Source of Nitrate in Nearshore
Marine Sediment," Nature. 1985, Volume 313: 214-216.
"The Effects of Deicing Salt Usage on Surface and Ground
Hater Quality," 1982 International Symposium on Urban
Hydrology. Hydraulics & Sediment Control. University of
Kentucky, Lexington, Kentucky, July 27-29, 1982.
"Upper Great Lakes Connecting Channel Study, Haste Disposal
Sites and Potential Ground Hater Contamination, St. Clair River"
Non Point Source Hork Group Report, April 1988."
"Summary of the Hydrologic Situation on Long Island, New York,
as a Guide to Hater-Management Alternatives," U. S. Geological
Survey Professional Paper 627-F, 59p.
"Bydrogeology, Hater Quality, and Land Management in the Big
Spring Basin, Clayton County, Iowa," Iowa Geological Survey,
Open-File Report 83-3, 1983 Report on contract 82-5500-002.
"The Ecological Significance of the Submarine Discharge of
Groundwater," Marine Ecology Progress Series. 1980,
3: 365-373.
"Preliminary Evaluation of Chemical Migration to Ground Hater
and the Niagara River from Selected Haste - Disposal Sites,"
USEPA, March 1985, EPA 905/4-85-001.
"Interaction between Groundwater and Surface Hater Regimes
and Mine-induced Acid - Mine Drainage in the Stockett-Sand
Coulee Coal Field," Montana Joint Hater Resources Research
Center, 1983, Project No. A-129MONT, Bozeman, Montana.
"Dispersal of Plating Hastes and Sewage Contaminants in Ground
Hater and Surface Hater, South Farmingdale - Massapequa Area,
Nassau County, New York," U.S. Geological Survey Hater Supply
Paper 1879-G.
pp.198-199
pp.200-201
pp.202-206
pp.207-209
pp.210-211
pp.212-213
pp.214-216
pp.217-219
pp.222-223
-------
31
Table B-2. References to Annotated Bibliography (Continued)
Author
Citation
Reference to Annotated
Bibliography
T. B. Spuill
R. L. Stach, R. N. Helgerson.
R. F. Bretz, M. J. Tipton,
D. R. Biessel, J. C. Harksen
J. D. Stoner
P. H. Whitfield, B. McNaughton
"Statistical Evaluation of the Effects of Irrigation on
Chemical Quality of Ground Hater and Base Flow in Three
River Valleys in North Central Kansas," U.S. Geological
Survey Hater Resource Investigation Report 85-4156, 1985.
"Arsenic Levels in the Surface and Ground Waters along
Whitewood Creek, Belle Fourche River, and a portion of
the Cheyenne River, South Dakota," Completion Report,
Project Number A-054-SDAK, Agreement Number 14-34-0101-6043,
July, 1978.
"Dissolved Solids in the Arkansas River Basin," National
National Hater Summary 1964: Hydrologic Events. Selected
Hater Quality Trends, and Ground-Hater Resources. U.S.
Geological Survey Hater Supply Paper 2275.
"Indications of Ground-Hater Influences on Nutrient Transport
Through Schwatka Lake, Yukon Territory," Hater Resources Bulletin,
1982, 18(2): 197 - 203.
pp.224-226
pp.227-228
p. 229
pp.230-232
-------
32
C. Studies involving geophysical techniques to estimate ground-water
discharge to surface water
The papers cited in this section are summarized in Section II of "An
Annotated Bibliography of the Literature Addressing Nonpoint Source
Contaminated Ground-Water Discharge to Surface Water," September 1990, EPA
440/6-90-006.
i. General description of method
a. Description of method or procedure
Ground water discharging to surface water is controlled by the hydraulic
properties of the sediments of the surface-water body and the hydraulic
gradient across those sediments. The sediment hydraulic properties of large
water bodies, such as the Great Lakes or the Chesapeake Bay, are difficult to
measure due to the depth of the sediments in open water. Standard methods of
drilling and sediment sampling become slow and costly endeavors in deep
aquatic environments. Less costly shipboard geophysical systems offer a
method that continuously characterizes bottom sediments along the ship's
track. Combining seismic and electrical geophysical measurements provides
data to estimate sediment type, thickness, and sequence, as well as relative
vertical hydraulic conductivity. Based on this information, one can calculate
the volume of ground water discharging to surface water.
Geophysical methods have primarily been applied to lakebeds. Bradbury
and Taylor (1984) collected geophysical data at an offshore site in Lake
Michigan with sediment thicknesses ranging from 0.3 to 37 m and water depths
from 2.5 to 27 m. Other investigators have used geophysical techniques in
smaller lakes and in channels connecting the Great Lakes (see Bradbury and
Taylor, 1984; Cherkauer and Taylor, 1987; Lee, 1989, and Taylor and Cherkauer,
1984). Zektser and Bergelson (1989) have used continuous measurements of
temperature and electric conductivity and continuous seismoacoustic profiling
to detect temperature and salinity anomalies in Lake Issyk-Kul in the
southeastern USSR. One major difficulty associated with geophysical
techniques is the need for field tests to verify the results. Field
verification can be difficult to obtain in deep water.
Seismic
Seismic exploration involves generating seismic waves and measuring the
time required for the waves to travel to a series of receiving devices called
geophones. In seismic studies of large surface-water bodies, a shipboard
seismic profiling system can generate and receive the seismic waves. Seismic
waves generated on board the ship travel downward through the lake bottom
sediment until they reflect off a hard surface and back up through the
sediment to the ship's geophones. Information on sediment type, thickness,
and sequence can be inferred through interpretation of the travel times of the
seismic waves (Taylor and Cherkauer, 1984).
-------
33
Induced Potential or Electrical Charge
This method involves charging the sediment with a current, shutting off
the power source, and measuring the rate of current decay. An electrical
array towed behind the boat charges bottom sediments and measures the rate of
current decay. The relative clay content of bottom sediments can be
determined using this method. These determinations are then used to estimate
the vertical hydraulic conductivity of the sediment. Taylor and Cherkauer
(1984) describe the equations characterizing the use of electrical
conductivity and seismic readings used to estimate seepage (see Section
C.i.d).
Resistivity
Resistivity methods also employ an artificial source of current which
enters the subsurface through point-electrodes. Receiving electrodes measure
the potentials of the electric flow field, which are influenced by the
composition of the subsurface materials. An electrical array of source and
receiving electrodes towed behind a boat (see Figure 2-7) measures the
resistivity of an induced electrical field in the sediments. Sediment type,
thickness, and sequence affect the configuration of the induced electrical
field. Investigators infer the effective longitudinal conductance of the
bottom sediments through interpretation of the resistivity of the induced
electrical field. The effective longitudinal conductance, combined with
sediment thickness information from seismic techniques and clay content
estimates from electrical/resistivity techniques, provides data used to
determine the effective vertical hydraulic conductivity of the sediment
sequence. The effective vertical hydraulic conductivity, the hydraulic
gradient over the sediment sequence (the change in hydraulic head over
distance, measured at various points over a large area or assumed constant
over the study area), and surface area of the water body bottom provide data
to assess the likelihood and quantity of ground water discharging to surface
water (Cherkauer and Taylor, 1984).
Temperature and Electrical Conductance
Another indirect method for locating ground-water discharge areas
involves measuring temperature and bulk electrical conductance. A sediment
probe with temperature and electrical conductance sensors is towed behind a
boat along the bottom of a surface water body. From the continuous record of
temperature and conductance, anomalies in temperature and bulk electrical
conductance are located. These anomalies indicate the likelihood of ground-
water inflow. Knowledge of the sediment type, water depth, and other geologic
or hydrologic information concerning the nature of the possible discharge area
may be needed for data interpretation. Investigators may correlate measured
temperatures and conductivities with other techniques to better characterize
the nature of sediment anomalies (Lee, 1989).
-------
34
Figure 2-7
Disc Drive
rt
C. P. U.
A To 0 Converter
Comouter
Terminal
Block diagram of electrical instrumentation
used in iieid survey
Taylor, Robert W., and Douglas S. Cherkauer, "The Application of
Combined Seismic and Electrical Measurements to the Determination of the
Hydraulic conductivity of a Lake Bed," Ground-Water Monitoring Review. 1984,
Volume 4(4): p. 80.
-------
35
b. Assumptions involved in using these models
Seismic
Reflective seismic methods assume an essentially horizontal reflective
layer is located some distance beneath the sediment and will reflect the
seismic waves. If the reflective layer is located a great distance beneath
the sediment/surface-water interface, the seismic wave may be sufficiently
damped, making reception and subsequent interpretation difficult. Seismic
profilers have measured sediment thicknesses up to 37 m (Bradbury and Taylor,
1984).
Because the hydraulic gradient across the sediment sequence on the
bottom of a surface-water body is difficult to measure, a uniform hydraulic
gradient is often assumed for the entire body.
Electrical Charge and Resistivity
Electrical charge and resistivity studies require the assumption that a
proportional relationship exists between clay content, hydraulic leakage
(across the sediment/surface-water boundary) and the scaling coefficient. The
scaling coefficient is required for the determination of hydraulic
conductivity from electric and seismic readings (see Section C.i.d). For water
bodies where clay content does not vary much, the scaling constant is assumed
to be constant for the layered bottom sediments (Bradbury and Taylor, 1984).
Temperature and Electrical Conductance
Temperature and conductivity measurements require that temperature and
conductivity differences in the sediment result from ground-water seepage.
The temperature technique relies on the theory of ground-water flow distortion
of thermal gradients, as indicated by ground-water discharge often producing
relatively strong thermal gradients, while recharge gradients remain
relatively weak. Other phenomena may be caused by sediment temperature
anomalies, such as sediment storage of summer heat. The electrical
conductance technique rests on the assumption that the distribution of solutes
at the sediment/surface-water interface is expected to differ in areas where
there is ground-water discharge as compared to areas where there is no ground-
water discharge to the surface-water body (Lee, 1985).
c. Limitations of the methods
The uncertainties associated with geophysical methods limit their
usefulness in determining ground-water discharge rates to surface water.
Because these methods indirectly determine sediment thickness and effective
vertical hydraulic conductivity for a sediment sequence, the uncertainty
associated with the predicted values is greater than the uncertainty
associated with values measured directly. If it is assumed that a single
hydraulic gradient estimate represents the entire water body, further
uncertainties may be introduced. No attempts were made, in the studies
reviewed in Appendix A, to quantify the effects that these assumptions had on
the predicted rates of ground-water discharge to surface water. These
-------
36
uncertainties highlight the necessity of ground- truthing to verify the
analytical results derived from these techniques.
The resistivity methods used to determined ground-water discharge rates
are best suited to surface water environments where sediment overlies bedrock.
The high resistivity of bedrock terminates the summation of the longitudinal
electrical conductance equation and allows longitudinal conductance to be
determined from a single electrical measurement.
d. Representative equations
The equations describing the longitudinal electrical conductance of a
sequence of sediments are described by Taylor and Bradbury (1984) :
where :
S - Longitudinal electrical conductance [1/MLT]
b£ - Thickness of layer i [L]
Pi - Electrical resistivity of layer i [1/ML2T].
The longitudinal electrical conductance, thickness, and clay content of a
sediment sequence provide data for an equation describing the effective
vertical hydraulic conductivity of the sediment sequence. To utilize the
equation, it is necessary to assume clay content of a sedimentary sequence is
accurately represented by an empirical scaling factor. The equation relating
effective vertical hydraulic conductivity with longitudinal electrical
conductance, sediment thickness, and a scaling factor is described by
(Cherkauer and Taylor) :
Kv - (C0bT)/S
where :
Kv = Effective vertical hydraulic conductivity [L/T]
bT Total thickness of sediment sequence [L]
S - Longitudinal electrical conductance [1/MLT]
C0 = Scaling factor [1/MLT2].
After estimating the vertical hydraulic conductivity for a sediment sequence,
one can apply Darcy's law to determine the quantity of ground water discharged
to surface water by assuming a hydraulic gradient across the sediment
sequence, and summing that gradient across the areal distribution of those
sediment sequences with the same effective hydraulic conductivity. Darcy's
Law is as follows :
Q - K^dh/dlA
where :
Q - Ground- water discharge rate [L3/T]
-------
37
KV - Effective vertical hydraulic conductivity [L/T]
dh/dl - Hydraulic gradient across the sediment sequence [L/L]
A - Area on bottom of surface-water body with similar
effective vertical hydraulic conductivity [L2] .
e. Description of field equipment
These methods usually require a boat, and sometimes a sizeable ship, to
contain and deploy the geophysical instruments and support equipment used to
characterize bottom sediments. Shipboard seismic instrumentation, used to
determine sediment thickness, consists of a high resolution sounder and
recorder. Electrical resistivity and chargeability equipment, used to
determine the electrical longitudinal conductance and clay content of a
sediment sequence consists of a long multiconductor cable equipped with source
and receiving electrodes. The cable is towed behind the ship. A Loran
navigation system determines the location of the ship's position for each
measurement. A computer stores the position and measurement data. The same
computer can assist in the interpretation of the data (Taylor and Cherkauer,
1984).
f. Expertise needed to apply the methods
Review of the papers summarized in Appendix A suggests that geophysical
methods require considerable expertise. Prior experience helps one to
properly configure the instrumentation, conduct the tests, and interpret the
results. Also, because a large boat must be used to house the geophysical
system, these methods require navigational and piloting skills. For a more
complete discussion of the expertise required to apply geophysical methods
readers are referred to Taylor and Cherkauer (1984), Bradbury and Taylor
(1989), Cherkauer and Taylor, and Lee (1985, 1989).
ii. Data inputs for the method
The geophysically determined effective hydraulic conductivity of bottom
sediments is estimated for use with Darcy's Law to determine the quantity of
ground water discharging to a surface-water body. Additional input data
include a representative vertical hydraulic gradient for the entire surface-
water body and the area of the bottom of the surface-water body.
When bottom sediment temperatures and conductivities are used to predict
ground-water discharge areas to surface water, one known source of ground
water seepage aids in calibrating the equipment. This technique is limited to
general information about potential ground-water seepage zones.
iii. Outputs from the method
a. Ground-water quantity discharge to surface water
Geophysical methods estimate essentially any amount of ground-water
discharge to surface water. The discharge rate relates to the effective
-------
38
hydraulic conductivity of the sediments and the hydraulic gradient across the
sediments. Ground-water discharge rates to surface water of up to
approximately 2.5 m3/s were recorded using geophysical methods (Cherkauer and
Taylor, 1984).
b. Ground-water quality discharge to surface water
Geophysical techniques do not determine ground-water quality, but
conductivity measurements offer general water quality information.
Conductivity values relate to the amount of dissolved chemical constituent in
the water. A high conductivity value indicates the presence of dissolved
chemical constituents in water, while a low conductivity value indicates the
relative absence of chemical constituents in the water.
iv. Setting in which the method has been applied
Table C-l summarizes some of the locations where this method has been
used.
v. Contaminants that have been measured using the method
Geophysical methods do not directly measure the concentrations of
chemical constituents in surface or ground water. Water quality can be
qualitatively related to conductivity measurements. Generally the greater the
chemical constituent concentration in water, the greater the conductivity
measurement.
vi. General evaluation of the method
Geophysical methods characterize the spatial distribution of sediments
on the bottom of surface-water bodies over large areas in a cost effective
manner, as compared to other direct measurement techniques. As a result,
there is an inherent trade off in the accuracy of the derived data.
Geophysical techniques are of greatest value when applied to generally
characterize the ground-water discharge characteristics of a large area of
bottom sediment. Ground-truthing using conventional hydrogeologic
measurements, such as mini-piezometers, piston corers, and seepage meters
(described in Section A of this chapter), provides more precise estimates of
discharge rates for specific areas. By identifying on-shore recharge zones
for these areas and regulating land use in the recharge zones, surface-water
quality can be better protected.
vii. References to annotated bibliography
References to the accompanying annotated bibliography are provided in
Table C-2.
-------
Table C-l. Summary of Settings in Which the Method Has Been Applied and the Contaminants Measured.
Location
Aquifer
Contaminant
Author
Ontario
Green Bay, Hi
Detroit Metropolitan
Area, MI
Mequon, WI
Great Lakes
Shallow glacial,
glacial bedrock
interface, bedrock
units
organic solvents
D. R. Lee, S. J. Welch
K. R. Bradbury,
R. W. Taylor
I
D. S. Cherkauer,
R. H. Taylor
D. S. Cherkauer,
B. Zvibleman
EPA Non Point
Source Group
Hardwick and New
Braintree, MA
FDover, NJ
Chalk River
Nuclear * Laboratories,
Ontario
Southeastern
U.S.S.R.
W. W. Lapham
D. R. Lee
I. S. Zekster
G. M. Bergelson
-------
Table C-2. References to Annotated Bibliography
Author
Citation
Reference to Annotated
Bibliography
K. R. Bradbury
R. W. Taylor
Determination of Hydrogeologic Properties of
Lakebeds Using Offshore Geophysical Surveys,"
Ground Hater. 1984 Volume 22(6): 690-695.
pp.74-75
D. S. Cherkauer
R. W. Taylor
D. S. Cherkauer
B. Zvibleman
"Geophysically Determined Ground Water Flow into
the Channels Connecting Lakes Huron and Erie,"
Proceedings of the Second National Outdoor Action
Conference on Aquifer Restoration. Ground Water
Monitoring and Geophysical Methods. Volume 2,
Presented by the Association of Ground Water
Scientists and Engineers and EPA/EMSL - Las Vegas,
pp. 779-799.
"Hydraulic Connection between Lake Michigan and a
Shallow Ground-Water Aquifer," Ground Water. 1981,
Volume 19(4): 376-381.
pp.76-78
pp.79-80
EPA Non Point Source
Work Group
"Upper Great Lakes Connecting Channel Study, Waste
Disposal Sites and Potential Ground Water Contamination
St. Clair River," Non Point Source Work Group Report,
April, 1988.
pp.81-85
W. W. Lapham
D. R. Lee
"Use of Temperature Profiles beneath Streams to Determine
Rates of Vertical Ground-Water Flow and Vertical Hydraulic
Conductivity," Draft Water Supply Paper No. 2337.
"Method for Locating Sediment Anomalies in Lakebeds that
that can be caused by Ground-Water Flow," Journal of
Hydrology. 1985, 79: 187-193.
pp.86-87
pp.88-89
R. W. Taylor
D. S. Cherkauer
"The Application of Combined Seismic and Electrical Measurements
to the Determination of the Hydraulic Conductivity of a Lake Bed,
Ground-Water Monitoring Review. 1984, Volume 4(4): 78-85.
pp.92-93
I. S. Zekster
fl M RoT-OQlcfl
"Effect of Ground Water on Lake Water Quality," Water Quality
B.illaHn Tay.ua IQflQ pp ^fl-1^
pp.94-95
-------
Al
D. Studies involving hydrograph separation, regression analysis, or mass
balance approaches to estimate the contribution of ground water to
stream flow
The papers cited in this section are summarized in Section III of "An
Annotated Bibliography of the Literature Addressing Nonpoint Source
Contaminated Ground-Water Discharge to Surface Water," September, 1990, EPA
440/6-90-006.
i. General description of method
a. Description of method or procedures
The methods discussed in this section have been applied by investigators
in areas throughout the U.S., in Ontario, Canada, and in the United Kingdom.
Hydrograph separation has been used in conjunction with graphical techniques
to estimate the distribution of ground-water flux to areas of the Great
Lakes.6 Other investigators have used analysis of conservative tracers along
with hydrograph separation data to estimate ground-water flux and contaminant
loading. The regression analysis and soil moisture balance methods rely on
equations developed for specific regions. Arihood and Glatfelter (1986) have
developed regression equations for northern Indiana, while Bevans's (1986)
work was in eastern Kansas. Wilson and Ligon (1979) applied a water balance
model to the Piedmont and Sandhill Regions of South Carolina.
Hvdrograph Separation
Precipitation entering a watershed travels to a stream by three main
routes: surface runoff, interflow (or subsurface storm flow), and ground-
water flow. The amount of water contributed to the stream by each of the
three processes is reflected in the shape of the stream hydrograph, a graph of
stream discharge at a particular point in the watershed versus time. The
hydrograph for a single, short duration precipitation event, occurring over
the entire watershed, follows a general pattern (see Figure 2-8). The
hydrograph shows a period of increasing stage, or increasing discharge, known
as the rising limb, that culminates in a peak or crest. Following the peak
discharge, the hydrograph shows a period of decreasing discharge, referred to
as the recession limb. Hydrograph separation techniques are applied to the
recession limb to estimate contributions to stream flow from surface runoff,
interflow, and ground-water flow.
When the hydrograph is plotted on semilogarithmic graph paper (discharge
on the semilogarithmic y-axis), the recession limb has three identifiable line
segments of different slopes, (see Figure 2-9). The slope of the line segment
immediately after the peak discharge is the steepest and represents
contribution to stream flow as a result of surface runoff and subsurface
6 Pranckevicius, Pranas, personal communication, Nonpoint Source
Contaminated Ground-water Discharge to Surface Water Workshop, Chicago, IL,
November 30, 1989.
-------
42
Figure 2-8
r 21-
Hydrograph of streaxnflow in response to a rainstorm from a
100-square-kiiometer basin.
Dunne, T. and L. Leopold. (1978) Water in Environmental Planning. San
Francisco: W.H. Freeman and Co.
-------
43
Figure 2-9
Total runoff
Surface runoff plus
subsurface runoff
Surface runoff
>-Groundwater runoff
^Subsurface runoff
28 29 30 31 I 234 5678
August-September, 1951
Semilogarithmic plotting of a hydrograph, showing separation of runoff com-
ponents. (Panther Creek at £1 Paso, Illinois.)
Chow, V. (ed.) (1964) Handbook of Applied Hydrology. New York: McGraw-
Hill.
-------
44
runoff, which includes interflow and ground-water storage depletion. When
surface runoff storage is depleted, the slope of the recession limb flattens.
This portion of the recession limb represents contribution to stream flow as a
result of interflow and ground-water storage depletion. The slope of the
recession limb of the hydrograph changes again when interflow storage is
depleted and contribution to stream flow is a result of ground-water storage
depletion only. The ground-water contribution to stream flow is referred to
as baseflow (see Figure 2-8). Surface runoff and interflow are often combined
and referred to as direct runoff. The slope of the final segment of the
recession limb is the ground-water recession constant, Kr, for the watershed.
The line segment representing baseflow is extended back in time to a point
under the hydrograph peak to determine maximum ground-water discharge to the
stream as a result of the precipitation event. The ground-water recession
constant for a watershed and the maximum ground-water discharge rate are used
in an empirical formula to estimate ground-water discharge to surface water at
any time after a precipitation event.
O'Brien (1980) has developed a "dynamic method" of hydrograph separation
which matches the hydrograph of an index well with the stream hydrograph to
determine the moment of maximum ground-water discharge for two small wetland-
controlled basins in Massachusetts. The advantage of the method is that it is
not rigidly tied to ground-water stage, and it accommodates variations in
ground-water inflow and loss in channel storage in response to temperature,
vegetation, stream stage, and change in seasons, causing shrinking and
swelling of the peat and muck in the wetlands.
Regression Analysis
Equations developed with regression techniques that relate basin
characteristics to baseflow characteristics in gaged streams can be used to
estimate baseflow in ungaged streams. Examples of basin characteristics used
in the regression analysis include drainage area of the watershed and flow
duration ratio. The flow duration for a stream at a given point in the
watershed is the proportion of time that discharge is less than a specific
discharge value. Flow duration is commonly expressed as a curve representing
the percent of time discharge is less than an indicated value versus discharge
per area of the watershed, (see Figure 2-10). The flow duration ratio is the
20-percent flow duration divided by the 90-percent flow duration. The
drainage areas of the watersheds and the flow duration ratios are transformed
to logarithmic units and a regression equation is developed by backward
elimination and maximum R2 improvement procedures. For more information on
regression analysis see Arihood and Glatfelter (1986) and Bevans (1986).
Soil Moisture Balance
The ground-water discharge component to a stream can also be estimated
using a soil moisture mass balance approach, where inflow (precipitation)
equals outflow (baseflow). Soil moisture water balance methods for a
watershed assume that any excess soil moisture below the root zone ultimately
will contribute to baseflow. The soil characteristics of the major soil types
within the watershed are used to estimate the water-holding capacity of the
different soil types. Excess soil moisture content below the root zone is
-------
45
Figure 2-10
50
2"
'5
i 10
I 5
I
3
5
I 1
8. .5
JO
-i
5
.01
i i n i i i i i i i i i i i .
Maaragua R. at /
Maaragua < 138 sq mi),
Uaso Nyiro at
Humes Bridge (733 sq mi)
i i
I I I I I I I
J I
02 1 5 10 20 40 60 80 90 95 99
Percent of time the flow is less than the indicated value
Flow duration curves for the River Maaragua in
humid, central Kenya (mean annual rainfall 60 inches) and
for the Uaso (River) Nyiro in semi-arid, north-central Kenya
(mean annual rainfall 35 inches). The dashed lines indicate
the flow values below which discharge declines for 10 percent
of the time. The curves were constructed from records for the
period 1956-1970.
Dunne, T. and L. Leopold. (1978) Water in Environmental Planning.
Fransico: W.H. Freeman and Co.
San
-------
46
predicted using precipitation data, the evapotranspiratibn rate, and estimates
of surface runoff for the watershed. The watershed is divided into two zones
based on a predetermined depth to ground water from land surface. In the zone
where depth to water from land surface is less than the predetermined depth,
excess soil moisture below the root zone is assumed to discharge immediately
to the stream. In the zone where depth to water is greater than the
predetermined depth, excess soil moisture below the root zone is assumed to
discharge to the stream in uniform increments, based on the time between
precipitation events. For more information on soil moisture balance methods,
see Wilson and Ligon (1979).
b. Assumptions involved in using these methods
Hydrograph Separation
Use of hydrograph separation techniques assumes that precipitation
entering a watershed is evenly distributed and of the same intensity for the
duration of the storm. Additionally, hydrograph separation techniques assume
that the semilogarithmic plot of the recession limb of the stream hydrograph
will have three identifiable segments of different slopes.
Regression Analysis
An important assumption when using regression equations to predict
baseflow in unaged streams is that the basin characteristics used in the
regression analysis are similar to basin characteristics of the unaged stream.
Basin characteristics of concern are (a) the ground-water gradient, (b) the
direction of the ground-water gradient, (c) the topography of the watershed,
(d) the slope of the stream channel, and (e) the length of overland flow.
Also, the geologic material underlying the basin will influence the shape of
the stream hydrograph (Arihood and Glatfelter, 1986, and Bevans, 1986).
Soil Moisture Balance
The soil moisture water balance model assumes that any excess soil
moisture below the root zone ultimately contributes to baseflow. Excess soil
moisture below the root zone in the zone nearest the stream is assumed to
enter the stream immediately following a precipitation event. Excess soil
moisture below the root zone, in the zone farthest from the creek, is assumed
to reach the stream in uniform increments, based on the time between
precipitation events. Additionally, when the water table is below the root
zone, it is assumed that no evapotranspiration occurs. Ground-water
boundaries are assumed to correspond to surface-water boundaries, and there
are no losses of ground water to other watersheds (Wilson and Ligon, 1979).
c. Limitations of the methods
Hydrograph Separation
In theory, it is straight forward to separate the recession limb of a
stream hydrograph into three segments of different slopes from which the
quantity of water contributed to the stream by surface runoff, interflow, and
-------
47
ground-water flow can be determined. In practice, separating the recession
limb of a stream hydrograph into three segments of differing slope is an
arbitrary process. Often no clear-cut change in slope exists. Given that
precipitation events are not often of constant intensity or evenly distributed
and considering the heterogeneity of a typical watershed, this is not
surprising. Additionally, the effects of bank storage will make separating
the recession limb of the storm hydrograph into three segments even more
difficult. Because of the difficulties in determining the slope of the
baseflow component of the recession limb of the stream hydrograph, predictions
of ground-water contributions to stream flow may not be precise. Also, a
continuous record of stream stage or discharge must be available to use this
method.
In addition, several authors have questioned the ability of hydrograph
separation techniques to determine the ground-water component of storm runoff
accurately. Sklash and Farvolden (1979) report that ground water plays a much
more active, responsive, and significant role in the generation of storm and
snow-melt runoff in streams than hydrograph separations may predict. This
increase in ground-water discharge may be caused by a rapid rise in hydraulic
head along the perimeter of transient and perennial discharge areas.
Regression Analysis
Regression analysis equations developed using basin characteristics from
one region should not be used to predict baseflow of streams located outside
that region. Baseflow characteristics are dependent on the geology and
geographic location of the watershed (Arihood and Glatfelter, 1986).
Different geologic units will have different hydrologic properties, and
different geographic locations will be exposed to different weather conditions
and patterns. A given geologic formation may underlie one watershed and be
absent in the neighboring watershed and weather conditions may be markedly
different in adjoining watersheds as a result of orographic effects. Also,
basin characteristics in a region may change gradually with distance from the
study area. Because regression equations must be applied to basins with
characteristics similar to those of the basins used in the regression analysis
and because applicable basin characteristics are often difficult to define,
the potential exists for regression analysis equations to be misapplied,
resulting in inaccurate baseflow predictions.
Soil Moisture Balance
Like regression equations, mass balance soil moisture equations are
based on basin characteristics common to a region. If the selection of
constants in these questions is based on characteristics of basins that lie
outside the region of concern, estimates of soil moisture balance and baseflow
may be inaccurate.
d. Representative equations
If the break between direct runoff and baseflow on the semilogarithmic
graph of the recession limb is difficult to define, an empirical equation can
be used to estimate the number of days after the peak discharge at which
-------
48
direct runoff essentially ends. The empirical equation is as follows:
N - A0-2
where:
N - Number of days after the peak when baseflow begins [dimensionless]
A - Watershed area in square miles [L2] .
The equation to determine the quantity of ground water discharging to surface
water at any time after a precipitation event is:
Q(t) - Q^V"
where:
Q(t) - Ground-water flow at time t after the peak discharge
[L3/T]
Qo - Ground-water flow at time t - 0 [L3/T]
Kr Ground-water recession constant (derived from the
hydrograph)
t - Time [T].
Equations for regression analysis and soil moisture balance methods are not
presented here, as the equations are region specific and not universally
applicable. Readers interested in these techniques are advised to read Wilson
and Ligon (1979), Arihood and Glatfelter (1986), and Bevans (1986).
e. Description of equipment needs
Stream stage data are obtained using a continuous-chart recorder.
Ideally the recorder should be located in a controlled section of the stream
channel so a stream stage/discharge relationship can be developed.
For water balance methods, topographic maps are used to determine the
area of a watershed. A rain gage can be used to measure precipitation; soil
maps are used to determine soil types and to estimate soil properties.
f. Expertise required to apply this method
The greatest difficulty in using this method is in selecting the portion
of the baseflow recession hydrograph to use in determining the ground-water
recession constant (Kr) for the watershed. Estimation of Kr requires
knowledge of basin characteristics and the temporal distribution of
precipitation in the basin as well as considerable professional judgement.
Once calculated, Kr can be used in the empirical ground-water discharge
formula to determine the quantity of ground water discharging to surface water
in a watershed as a result of a precipitation event.
Computer programs, as well as PC-based spreadsheets, are available that
determine equations relating dependent and independent variables through
regression analysis. The difficulty in using computer programs to determine
-------
49
the relationship of stream stage to basin characteristics is in determining
which basin characteristics are appropriate input parameters. Additionally,
it may be difficult to determine the area over which a regression equation is
applicable. As with hydrograph separation, this method is best utilized by a
hydrologist who is familiar with the subject basin.
A significant level of effort may be required to use soil moisture
balance models to predict baseflow in a watershed. The upper and lower
watershed zones must be delineated based on estimates of depth to ground water
below land surface. Additionally, surface runoff and evapotranspiration rates
must be estimated for the watershed. Because of a lack of representative
precipitation measurements, the precipitation entering a watershed often must
be estimated from precipitation measurements taken some distance away. Again,
familiarity with the hydrologic and geologic characteristics and the temporal
and spatial distribution of precipitation in the subject basin as well as
surrounding basins is highly recommended for the successful use of these
models.
ii. Data inputs for the method
Hydrograph separation techniques require a continuous record of stream
stage or discharge to determine ground-water contribution to stream flow.
Continuous stream stage/discharge data are available for many watersheds from
the United States Geological Survey.
The drainage area of the watershed must be known to use regression
equations to determine baseflow in an ungaged stream (Arihood and Glatfelter,
1986).
Soil moisture balance models require the following data: precipitation
records, water-holding capacity of major soil classes in the watershed, area
of the watershed, and drainable porosity measurements (Wilson and Ligon,
1979).
iii. Outputs for the method
a. Ground-water quantity discharge to surface water
Essentially any quantity of ground-water discharge to surface water can
be predicted using hydrograph separation techniques. If the time between
precipitation events is sufficiently long, the predicted ground-water
discharge rate to surface water will decrease over time. The maximum ground-
water discharge rate to surface water will be a function of the length and
intensity of the precipitation event and the amount of ground water currently
stored in the watershed.
As with hydrograph separation, any quantity of baseflow can be predicted
using regression equations and soil moisture water balance models.
-------
50
b-. Ground-water quality discharge to. surface water
Hydrograph separation, regression equations, and soil moisture water
balance methods do not predict the quality of ground water discharging to
surface water. Baseflow determined using these methods can be used in
conjunction with ground-water quality measurements obtained in wells located
adjacent to the surface-water body to estimate the ground-water loading rate
to surface water.
iv. Settings in which the method has been applied
Some of the settings in which this method has been used are presented in
Table D-l.
v. General evaluation of the method
Hydrograph separation techniques are an established method for
estimating the ground-water discharge rate to surface water in a watershed.
The method is well understood, simple to apply, and continuous stream
stage/discharge data are readily available for many watersheds. The key to
approximating actual ground-water discharge rates to surface water using this
method involves correctly determining the slope of the baseflow portion of the
recession limb of the hydrograph. Different slopes will produce markedly
different predictions of ground-water contributions to stream flow.
Baseflow in ungaged streams can be estimated using regression equations.
The equations are developed using regression analysis, relating drainage basin
characteristics to baseflow, at gaging stations located in the same region.
Because baseflow characteristics are dependent on the geology and geographic
location of a drainage basin, a regression equation developed using drainage
basin characteristics from one region should not be used to predict baseflow
of streams located outside that region. Therefore, accurate estimation of
baseflow in ungaged streams is dependent on the ability of the individual
applying the method to identify regions having similar basin characteristics.
Therein lies the problem-basin characteristics that influence baseflow are a
result of a combination of components, some obvious, such as geographic
location, and some not so obvious, such as geology. Additionally, the degree
of interaction between the components affecting baseflow in a drainage basin
is not well understood. Therefore, the potential exists that regression
equations will be misapplied, resulting in inaccurate baseflow predictions.
Soil moisture balance techniques can be used to estimate baseflow in
ungaged watersheds. Because so many of the input parameters for the model
must be estimated, the error associated with baseflow predictions made using
this method may be large.
vi. References to annotated bibliography
References to the accompanying bibliography are summarized in Table D-2.
-------
Table D-l. Stannary of Settings in Which the Method Has Been Applied and the Contaminants Measured.
Location
Aquifer
Contaminant
Author
Lincoln, Massachusetts
Northern and
Central Indiana
Upper Coastal Plain
of South Carolina;
North Carolina;
Georgia
United Kingdom
Eastern Kansas
Elliot Lake, Ontario
Clayton County
Iowa
Clayton County
Iowa
Illinois
Quebec and Ontario
Cedar River Basin,
Iowa-Minnesota
Piedmont and
Sandhill Regions,
South Carolina
Upper Coastal
Plain Aquifer
Galena
Aquifer
Galena
Aquifer
Cedar River
Basin
Sulfate; Coal
mine drainage
Pyrite, Accessory
metals, Radio-
nuclides
Herbicides,
Pesticides, Nitrates
and other agricultural
inorganics
Nitrate nitrogen;
Pesticides
Herbicides
A. L. O'Brien
L. D. Arihood, D. R. Glatfelter
W. R. Aucott, R. S. Meadows,
G. G. Patterson
M. D. Bako, Ayodele Owoade
Hugh E. Bevans
D. W. Blowes, R. W. Gillham
G. R. Hallberg, R. D. Libra,
E. A. Bettis, III, B. E. Hoyer
R. D. Libra, G. R. Hallberg
B. E. Hoyer, L. G. Johnson
Michael O'Hearn, James P. Gibb
M. G. Sklash, R. N. Farvolden
P. J. Squillace, E. M. Thurman
T. V. Wilson, J. T. Ligon
-------
Table D-2. References to Annotated Bibliography
Author
Citation
Reference to Annotated
Bibliography
L. D. Arihood, D. R. Glatfelter
H. R. Aucott, R. S. Meadows,
G. G. Patterson
M.D. Bako, Ayodele Owoade
H. E. Bevans
D. W. Blowes, R. W. Gillham
G. R. Hallberg, R. D. Libra,
E. A. Bettis, III., B. E. Hoyer
R. D. Libra, G. R. Hallberg,
B. E. Hoyer, L. G. Johnson
A. L. O'Brien
M. O'Hearn, J. P. Gibb
W. A. Pettyjon, R. J. Menning
"Method for Estimating Low-Flow Characteristics of pp.97-98
Ungaged Streams in Indiana," U.S. Geological Survey,
Open-File Report 86-323, 1986.
"Regional Ground-Water Discharge to Large Stream in pp.99-101
the Upper Coasted Plain of South Carolina and Georgia,"
USGS Hater Resource Investigations Report 86-4332, 1987.
"Field Application of a Numerical Method for the Deviation pp.102-103
of Baseflow Recession Constant," Hvdrological Process. 1988,
2: 331-336.
"Estimating Stream-Aquifer Interactions In Coal Areas pp.104-105
of Eastern Kansas by using Streamflow Records," USGS
Water Supply Paper 2290 (January, 1986).
"The Generation and Quality of Streamflow on Inactive pp.106-107
Uranium Tailings Near Elliot Lake, Ontario," Journal of
Hydrology. 1988, 97: 1-22.
"Hydrogeologic and Water Quality Investigations in the Big pp.108-110
Spring Basin, Clayton County, Iowa," Iowa Geological Survey,
1984, Open-File Report 84-4.
"Agricultural Impacts on Ground-Water Quality," Proceedings of pp.111-112
the Agricultural Impacts on Ground Water. 1986, National Water
Well Association, Omaha, Nebraska, pp. 253-273.
"The Role of Ground Water in Stream Discharges from Two Small pp.113-114
Wetland Controlled Basins in Eastern Massachusetts," Ground Water.
1980, Volume 18(4):
State Water Survey Report Number 246. 1980 Illinois Institute of pp.115-116
Natural Resources.
"Preliminary Estimate of Regional Effective Ground Water Recharge pp.117-119
Rates, Related Streamflow and Water Quality in Ohio," Water Resources
Center, Preliminary Estimate of Regional Effective Ground Water
Recharge Rates in Ohio. Project Completion Report, 323 pp., 1979.
52
-------
Table D-2. References to Annotated Bibliography (Continued)
Author
Citation
Reference to Annotated
Bibliography
P. J. Squillace, E. M. Thurman
T. V. Wilson, J. T. Ligon
"Surface-Water Quality of the Cedar River Basin,
Iowa-Minnesota, With Emphasis on the Occurrence and
Transport of Herbicides, May 1984 through November
1985," U.S.G.S. Toxic Substances Hydrology Program.
Abstracts of Technical Meeting. Phoenix. Arizona.
September 26-30, 1988.
"Prediction of Baseflow for Piedmont Watersheds,"
Office of Water Research and Technology, Water
Resources Research Institute, Report Number 80,
1979, 47 pp.
pp.120-122
pp. 123-125
53
-------
54
E. Numerical models of surface-water/ground-water interactions
The papers cited in this section are summarized in Sections IV, V, and
VII of "An Annotated Bibliography of the Literature Addressing Nonpoint Source
Contaminated Ground-Water Discharge to Surface Water," September, 1990, EPA
440/6-90-006.
i. General description of methods
a. Description of method or procedure
Mathematical ground-water modelling simulates an aquifer or watershed
system using a series of equations governing flow and/or mass balance
properties. When developing a model, transport properties should be
constructed using a framework of measured variables. Modelling represents a
cumulative process of data gathering and model verification to ensure an
accurate depiction of real world phenomena in the computer simulation. Models
should not be used without field data and ground truthing, and the transient
conditions of the study locations should be understood and incorporated into
the analysis.
Mathematical ground-water models consist of sets of differential
equations that describe or "govern" ground-water flow and/or contaminant
transport. These equations can be solved to develop an analytical solution;
however, field situations may be complex and difficult to solve exactly, and
the assumptions that must be made to obtain the analytic solution are often
unrealistic and are not representative of the flow or transport problem under
consideration. In these situations, numerical methods can be used to solve
the differential equations and obtain an approximate solution that can be used
to simulate relatively complex ground-water flow and contaminant transport.
This process is presented in Figure 2-11. Two popular numerical methods used
to convert differential equations into algebraic equations are the finite
difference method and the finite element method.
To utilize a numerical flow model, a flow system is defined and
discretized into a finite number of rectangular blocks, in the case of finite-
difference models, or triangles or quadrilaterals, in the case of finite-
element models. Figures 2-12 and 2-13 show finite difference and finite
element representations of an aquifer bounded on three sides by an impermeable
boundary (i.e., no flow into or out of the aquifer) and on the fourth side by
a river into which discharge from the aquifer occurs. Each cell in the flow
region is assigned its own hydrologic properties based on measurements or
observations from the flow region being modeled. Boundary conditions are then
incorporated into the numerical model. Typical boundary conditions are
ground-water divides (no flow), surface-water bodies (fixed head), and
specified flow. The numerical model is run on a computer, and typically, the
calculated head-field distribution at nodal points (the intersections of the
lines delineating the region or centers of the blocks) is compared to the
actual head-field distribution (obtained through measurement of water levels
in wells) in the flow region and, if available, the results of an analytical
solution (see Figure 2-11). If the predicted and actual head fields are not
in close agreement, the model is adjusted by manipulating boundary conditions
-------
55
Figure 2-11
Set of differentia]
equations (mathematical
model)
Method of finite
differences or
finite elements
Set of algebraic
equations (discrete
model)
Calculus
techniques
i
Analytical solution
(possible for a limited
number of cases)
Compare if
possible
Iterative methods
or
direct methods
Approximate solution
1
Compare
Field
observations
Compare
.4
Relationships between mathematical model, discrete algebraic model, analytical
solution, approximate solution, and field observations.
Wang, H. and M. Anderson. (1982) Introduction to Ground water Modeling:
Finite Difference and Finite Element Methods. San Francisco: W.H.
Freeman and Co., 237 p.
-------
56
Figure 2-12
(a)
River
,.'.' ... ., -.;...'fry »"'
V OvierWnOB WVU
(b)
Fuhediffi
ind block
jjl ~ »
/I
Finite diffcpcBoc *n4 finite <
at itpiLJuuioom of u> aquifer region.
(a) Map view of aquifer showmg well field, observinon well*, and boundaries.
(b) Finite difference ghd with block-centered node*, where toe is the spaoag in the .T
direction. Ay H the ipaong m the r direction, and b a the aquifer thirir nrti *
Wang, H. and M. Anderson. (1982) Introduction to Ground water Modeling:
Finite Difference and Finite Element Methods. San Francisco: W.H.
Freeman and Co., 237 p.
-------
57
Figure 2-13
(ci
/
f-
Z.
I/
e Smfcc/tmfc nodt
o Soureewnk node
(c) Fmite difference grid with moh-cenieted nodo.
(d) Finite etement mesh with tnanguttr etemenu where * is the aqufer thickness.
(Adapted from Mercer and Pause 1980*.)
Wang, H. and M. Anderson. (1982) Introduction to Ground water Modeling:
Finite Difference and Finite Element Methods. San Francisco: W.H.
Freeman and Co., 237 p.
-------
58
and/or the hydrologic properties of the individual cells. When close
agreement is reached between predicted and actual head-field distribution, the
model is considered calibrated.
After the numerical flow model is calibrated, the velocity field for the
flow region calculated by the flow model is used as input to a solute
transport model, which simulates the movement of dissolved chemical
constituents through the aquifer by advection and dispersion. In many models,
the advection-dispersion equation is solved numerically. In the alternative
"random walk" method, a random process, based on a normal probability density
function, is used to simulate dispersion (Prickett et al, 1981). Dissolved
chemical constituents are represented by a finite number of discrete
particles. Each particle is assigned a mass which represents a fraction of
the total mass of the chemical constituent involved. Nonpoint chemical
constituents are introduced to the subsurface flow system in the recharge
input for the model. The mass of chemical constituent input to the system is
a function of factors such as land use and crop type. During each time step,
the particles are moved advectively with the ground-water velocity field.
Between time steps, the particles' positions are adjusted by moving individual
particles a random distance in any direction. The magnitude and orientation
of the random displacement is predicted by the normal probability density
function.
Individual cell concentration for each time step is calculated by
determining the number of particles located in the cell. Because each
particle has a constant mass and the cell has a known volume, chemical
constituent concentration in the cell can be determined. Predicted
concentrations are compared to known concentrations in the flow region. To
calibrate the model, input concentration, volume, and location are adjusted
until predicted concentrations agree with observed concentrations. When both
the predicted head and concentration fields reasonably agree with known values
in the flow region, the model is considered calibrated and can be used for
predictive purposes.
b. Assumptions involved in using these models
The primary assumption when using numerical models is that the predicted
flow field is a close approximation of the actual flow field. Loading rates
to surface water as a result of ground-water discharge are largely controlled
by the ground-water velocity field. Even after calibration, using the
assumptions that predicted and actual hydraulic head distributions are
similar, the predicted velocity field may not be correct. This phenomenon
results because a given hydraulic head distribution is not unique for a given
combination of aquifer properties. When characterizing a watershed, a
priority should be placed on delineating the spatial distribution of aquifer
properties rather than on obtaining additional hydraulic head measurements.
If input data are scarce, either in time or space, the predictive capabilities
of the numerical model may be compromised.
An assumption of most finite-difference models is that there is only one
set of principal anisotropic directions in the flow region and the principal
directions are aligned with the grid system. Care should be taken when
-------
59
designing the model grid to align grid axes with principal directions of
isotropy.
Chemical inputs to the flow region as a result of agricultural or other
practices are assumed from generalized land use (Gburek et al., 1989). The
input is assumed constant with time for a given land use for all contaminant
sources within a study region. Instantaneous mixing is assumed to occur in
each cell for each time step. Additionally, in the random walk process of
transport simulation, it is assumed that dispersion in porous media can be
considered a random process having a normal distribution.
c. Limitations of the models
The major limitation of numerical models is the large amount of data
required to accurately calibrate them. To accurately calibrate a numerical
model, information on the spatial and temporal distribution of land use,
recharge, chemical input, hydraulic head, ground-water quality, and surface-
water quality is needed. Also of prime importance is the spatial distribution
of aquifer characteristics. Often, ground-water models do not take into
account the variable effects of near shore phenomena. Generally, models will
not simulate ground-water quality changes associated with seasons, or reflect
the hydraulic conductivity changes associated with seepage face growth and
capillary response to precipitation.
Prior to using a model, the scale and the geographic conditions of the
study area must be incorporated into the model. For instance, fracture flow,
macropore flow, karst terrain, and anthropomorphic effects on the study area's
ground water may require that adjustments be made to the model's structure.
Few watersheds have been monitored sufficiently to provide the data needed to
calibrate a numerical model. Knowledge about one watershed in a region will
assist in characterizing another watershed in the same region, but additional
data probably will be required before the model is considered calibrated and
can be used for predictive purposes. Without being able to reproduce flow-
field conditions or chemical concentrations, little confidence can be placed
in a model's predictive capabilities.
An additional limitation of numerical models results from the
uncertainty associated with them. This uncertainty is a result of numerical
models being based on mathematical expressions that are a simplification of
the real world and the measurement error associated with input data. This
uncertainty could result in predicted values that deviate significantly from
the actual flow in the region being modeled. However, proper field data-
collection techniques and the use of well-tested models by experienced
personnel combine to produce reliable predictions in most cases.
d. Representative equations
The partial-differential equation that describes time-variable flow in a
heterogeneous, anisotropic two-dimensional aquifer (one in which hydraulic
conductivity varies both in direction and space throughout the aquifer) is:
*/8x(K3C-8h/»x) + 8/«y(Ky-8h/Sy) - Ss- fib/fit
-------
60
where :
K, - Hydraulic conductivity in x direction [L/T]
Ky - Hydraulic conductivity in y direction [L/T]
Ss - Specific storage [I/1]
h - Hydraulic head [L]
x - x direction [L]
y - y direction [L]
t - Time [T] .
A finite -difference numerical model approximates the above differential
equation using a series of finite -difference equations. The two-dimensional
finite-difference equation for a homogeneous, isotropic medium, where the grid
spacings in the x- and y- directions are the same and hydraulic conductivity
is constant and isotropic throughout the aquifer (K, =» Ky) , is:
where :
h - Hydraulic head [L]
T - Transmissivity [L2/T]
S - Storativity [dimensionless]
t - Time increment [T]
x - Width of the grid spacing, where »x - *y [L]
i " Column number [dimensionless]
j - Row number [dimensionless]
k - Time step or iteration index [dimensionless].
Column and row numbers in this equation correspond to those in the finite
difference grid presented in Figure 2-14. Nodes (intersections of grids) are
spaced horizontally by »x and vertically by *y. For the first iteration, or
solution of the equation, the modeler estimates the value for the hydraulic
head at each node. The head values of the first iteration (k-1) are used to
calculate the head values for the second iteration (k=2). The equation is
solved several times in this manner until the difference between the head
values of the final iteration and the previous iteration is less than a value
specified by the modeler, called a convergence criterion.
The partial-differential equation that describes solute transport in a
two-dimensional, homogeneous aquifer through dispersion and advection is:
DL-82c/Sx2 + DT-S2c/iy2 - V
dispersion advection
-------
61
Figure 2-14
(i.n
(5.1)
,4)
('-!.;)
(1.7-1)
(/.;)
(« + !.;)
A
I
(i.; + i) (5.4)
Ax-J
Finite difference grid showing index numbering convention.
Wang, H. and M. Anderson. (1982) Introduction to Ground water Modeling:
Finite Difference and Finite Element Methods. San Francisco: W.H.
Freeman and Co., 237 p.
-------
62
where:
DL - Longitudinal dispersion coefficient [L2/T]
DT - Transverse dispersion coefficient [L2/T]
C - Concentration [M/L3]
Vx - Average pore velocity in the x direction [L/T]
x = x direction (direction of flow) [L]
y - y direction [L]
t - Time [T].
In the random walk approach, solute transport in a porous medium is
represented by a series of equations. Dissolved chemical constituents are
represented by a finite number of discrete particles each having a mass
representing a fraction of the total mass of the chemical constituent involved
(Pricket et al., 1981). The total distance a particle travels between time .
steps is:
dj - d + d*
where:
dj = Total distance traveled per time step [L]
d - Distance traveled as a result of advection per time step [L]
d* - Distance traveled as a result of dispersion per
time step [L].
The equation representing the distance a particle is transported by advection
is:
d = vt
where:
d - Distance particle travels for each time step [L]
v Ground-water flow velocity [L/T]
t - Time step duration [T].
After the particles have been moved advectively, the position of each
particle is adjusted a random amount in any direction to account for
dispersion. The one-dimensional equation representing the influence of
dispersion on a particle's position is:
d* - N- (2-dL-V-t)*
where:
d* = Distance traveled as a result of dispersion per
time step [L]
dL = Longitudinal dispersivity [L]
V - Ground-water velocity [L/T]
t - Time step duration [T]
-------
63
N - A number between -6 and 6, drawn from a normal distribution
of numbers having a standard deviation of 1 and a mean of
zero [dimensionless] .
The equation determining individual cell concentration for each time step in a
two-dimensional model is:
where :
C - Concentration per unit width of chemical constituent [M/L3]
n - Number of particles in a cell [dimensionless]
mp - Mass per particle [M]
x - Cell length in x direction [L]
y - Cell length in y direction [L]
i - Column number [dimensionless]
j - Row number [dimensionless].
e. Description of computer hardware or software needs
Essentially any of the commercially- available numerical models can be
run on a PC or mainframe computer. The requirements for a PC computer are a
large memory capacity (640K is usually sufficient) and a math coprocessor
chip. Without a math coprocessor chip, numerical models with a large number
of cells can take hours to converge on a solution or even longer for large
time-variable problems. Small memory capacity will limit the number of nodes
used in simulating a flow region. Generally, the greater the number of nodes
used, particularly near boundaries between property types or in areas of steep
hydraulic gradients, the greater the model's accuracy. Limiting the number of
nodes might compromise the accuracy of the simulation. Suggested computer
hardware includes :
a PC computer with math coprocessor chip and graphics card,
a high resolution monitor (for plotting results on the screen),
a printer, and
a plotter.
Numerous software packages, both in the private and public domain, are
available for simulating ground-water flow and contaminant transport. Some of
the more popular models are codes by Konikow and Bredehoeft7, Prickett and
Lonnquist8, Trescott et al.9, and the International Ground-Water Modeling
7 Konikow, L. and J. Bredehoeft. (1978) Computer model of two-dimensional
solute transport and dispersion in ground water. U.S. Geological Survey,
Techniques of Water Resources Investigations Book 7, Chapter C2, 90 p.
8 Prickett, T. and C. Lonnquist. (1971) Selected digital computer
techniques for groundwater resource evaluation. Illinois State Water Survey
Bulletin 55, 62 p.
-------
64
Center at-Butler University in Indiana.
f. Expertise needed to run the models
To adequately simulate flow situations using numerical modeling
techniques, knowledge of hydrology, numerical methods, computer language, and
computers is needed. In addition to these skills, an intuitive sense, derived
from experience, is needed to determine the proper grid spacing, boundary
conditions, and geometry of aquifer properties. Also, once the numerical
model has been calibrated, expertise is needed to determine if predicted
values are as expected and are correct.
ii. Data inputs for the models
Basic input requirements for numerical models are the spatial
distribution of hydraulic conductivity, anisotropy ratios, and boundary
conditions. Examples of common boundary conditions are ground-water divides,
surface-water bodies, and known flux boundaries. The spatial and temporal
distribution of the hydraulic head field for the flow region is needed for
calibration purposes. Also, for chemical transport modeling, the spatial and
temporal distributions of chemical inputs and concentrations in ground water
and surface water are required. For nonpoint studies, chemical loading may be
a function of crop distribution and type; thus, agricultural records are
required.
ill. Outputs from the models
a. Ground-water quantity discharge to surface water
Numerical models can simulate essentially any quantity of ground-water
discharge to surface water. Discharges less than 1 cubic foot per second
(cfs) have been simulated (Eddy and Doesburg, 1985).
b. Ground-water quality discharge to surface water
As with quantity of ground-water discharge to surface water, essentially
any concentration of chemicals in both ground water and surface water can be
simulated.
iv. Settings in which the methods have been applied and contaminant
discharge that has been modeled
With enough information, a numerical model can be calibrated to simulate
9 Trescott, P., G. Finder, and S. Larson. (1976) Finite-difference model
for aquifer simulation in two dimensions with results of numerical
experiments. U.S. Geological Survey Techniques of Water Resources
Investigations, Book 7, Chapter Cl, 116 p.
-------
65
any watershed in any geologic region of the country. Given enough input data,
the transport of any chemical in ground water can be simulated. Contaminant
transport models are capable of simulating the effects of decay, retardation,
and sorption on chemical constituent distribution and concentration. Table E-
1 summarizes the settings in which numerical models have been used and the
contaminants that have been transported.
v. General evaluation of the method
Numerical models can be used to simulate various nonpoint source loading
scenarios for complex aquifer conditions. Before a numerical model can be
used for predictive purposes, however, a large amount of input data is often
required to properly calibrate the model. The amount of data required will
depend on the type of model used, the objectives of the study, and the level
of accuracy required. Acquisition of the needed data can require considerable
time, expertise, and expense. Because of these constraints, the numerical
model may have limited usefulness in cases where data are scarce and funding
is limited.
The numerical model's strength is its usefulness as a screening tool.
Numerical models calibrated to simulate watersheds in different regions of the
country could be used to assess the general effect of various regulatory
scenarios on ground-water quality in those watersheds located in those
regions.
vi. References to annotated bibliography
References to the accompanying annotated bibliography are summarized in
Table E-2.
-------
Table E-l. Sumiazy of Settings in Which the Method Baa Been Applied and the Contaminants Measured.
Location
Aquifer
Contaminant
Author
Wiiconiin
Kent, Washington
Pennsylvania
Juneau, Alaska
Northwestern
Indiana
Central Sand Plain,
Wisconsin
Trimmers Rock
and Catskill
Formations
Mendenhall Basin
Calumet
Aquifer
Organics, chloroform and tri-
chloroethylene; zinc
Nitrates
Agricultural chemicals
D. S. Cherkauer, B. R. Hansel
C. M. Eddy, J. M. Ooesburg
W. J. Gburek, R. R. Schnabel
S. I. Potter
D. I. Siegel
L. R. Watson, J. M. Fenelson
C. Zheng, K. R. Bradbury
M. P. Anderson
-------
Table E-2. References to Annotated Bibliography
Author
Citation
Reference to Annotated
Bibliography
V. K. Harwell, D. R. Lee
0. S. Cherkauer, B. R. Hensel
V. T. Dubinchuk
C. M. Eddy, J. M. Doesburg
W. J. Gburek, R. R. Schnabel
S. T. Potter
T. A. Prickett, T. G. Naymik
C. G. Lonnquist
D. I. Siegel
L. R. Watson, J. M. Fenelson
C. Zheng, K. R. Bradbury
M. P. Anderson
C. Zheng, H. F. Wang
M. P. Anderson, K. R. Bradbury
"Determination of Horizontal-to-Vertical Hydraulic
Conductivity Ratios from Seepage Measurements on Lake
Beds." Water Resources Research. 1981, 17: 565-570.
"Ground-Water Flow into Lake Michigan from Wisconsin,"
Journal of Hydrology. 1986, Volume 84: 261-271.
"Radon and Radium Discharge to Surface Streams,"
Water Resources. 1981, 8(1): 102-116, translated
from Vodnve Resursv.
"Remedial Action Modeling Assessment Western Processing
Site, Kent, Washington," Report prepared for U.S.
Environmental Protection Agency, Region X, Seattle,
Washington 98101, July 1985.
"Modeling the Effect of the Shallow Weathered Fracture
Layer on Nitrate Transport," Unpublished Draft Report.
"A 'Random Walk' Solute Transport Model for Selected Ground
Quality Evaluations," Illinois State Water Survey,
Champaign, IL, 1981, ISWS/BUL-65-81.
"The Recharge-Discharge Function of Wetlands near Juneau,
Alaska: Part I. Hydrogeological Investigations," Ground
Water. 1988, 26(4): 427-434.
"Geohydrology of a Thin Water-Table Aquifer Adjacent to Lake
Michigan, Northwestern Indiana," (in press).
"Role of Interceptor Ditches in Limiting the Spread of
Contaminants in Ground Water," Ground Water. 1988,
Volume 26(6): 734-742.
"Analysis of Interceptor Ditches for Control of Ground Water
Pollution," Journal of Hydrology. 1988, 98: 67-81.
pp.173-174
pp.175-177
pp.178-179
pp.180-181
pp.182-184
pp.185-187
pp.188-189
pp.190-191
pp.192-193
pp.194-196
67
-------
68
F. Studies involving the application of fvinetions estimating nonpoint
source loading to surface water for various land use types
The papers cited in this section are summarized in Section VI of "An
Annotated Bibliography of the Literature Addressing Nonpoint Source
Contaminated Ground-Water Discharge to Surface Water," September, 1990, EPA
440/6-90-006.
i. General description of method
a. Description of method or procedure
Nonpoint source loading models combine surface runoff, sediment yield,
and ground-water discharge with empirical loading rates to obtain estimates of
nitrogen and phosphorous chemical concentrations in surface water. Runoff in
the watershed is calculated from daily weather data using the U.S. Soil
Conservation Service's Curve Number Equation. Sediment yield is calculated
using the Universal Soil Loss Equation in conjunction with the Richardson
daily rainfall erosivity index. Ground-water discharge is calculated from
daily water balances for the unsaturated and saturated zones in a watershed or
by using hydrograph separation techniques. Loading rates for runoff, sediment
yield, and ground-water discharge are assigned based on land use. Land use is
divided into residential, commercial, industrial, and agricultural categories.
Agricultural land is further subdivided based on land use, crop type, and land
management practices. The land use loading rates for runoff, sediment yield,
and ground-water discharge are summed and multiplied by the total area of
similar land use in the watershed to obtain the empirical loading rate as a
result of that land use category. The total nonpoint source loading rate for
the drainage basin rate is obtained by summing the calculated loading rates
for each land use category (Haith and Shoemaker, 1987, and Ritter, 1986).
The estimation of ground-water discharge from functions is best used in
conjunction with verification methods such as mass and water balance
estimates, ground-water monitoring, piezometer sampling, and seepage meter
monitoring. Functions have been used to estimate discharges in inland
watersheds in Pennsylvania (Gburek, et. al.) and Wisconsin (Uttormark, et.
al.), and in Inland Bays in Delaware (Ritter) and the Chesapeake Bay (Schnabel
and Gburek). Most of the studies utilizing this method have examined nutrient
loadings into surface waters.
b. Assumptions involved in using these models
The major assumption of nonpoint loading models is that the empirical
loading rates assigned a land use category for runoff, sediment load, and
ground-water discharge are representative of actual loading conditions. The
assigned runoff and ground-water discharge loading rates for a land use
category are assumed independent of topography, soil type, or tillage methods
(Schnabel and Gburek, 1983).
Another assumption is that the transport process is not scale dependent;
that is, the empirical loading rates for land use categories are equally
applicable for both large and small drainage basins (Schnabel and Gburek,
-------
69
1983).
c. Limitations of the methods
The ultimate purpose of loading models is to predict the impact various
land management schemes will have on surface-water quality in a watershed
through use of empirical loading factors. Ideally, loading factors for
various land types should only be representative of on- field processes, such
as tillage and fertilization practices. In reality, loading factors are a
combination of on- field and off -field processes. Off -field processes such as
non-crop plant nutrient uptake, deposition of sediment in buffer strips near
streams in the watershed, and mixing of interflow and baseflow components of
different chemical composition, are included in loading factors.
Additionally, loading factors make no distinction between flowpaths,
effectively masking the processes which contribute to sediment and chemical
loss from a watershed. As a result, loading factors mask the interaction of
on- and off -field processes and cannot be adjusted to account for individual
changes in either on- and off -field management practices. Thus, as a
predictive tool, loading factors may have limited use.
There was significant uncertainty associated with the input parameters
for the model applications referenced in this review. Precipitation and
temperature data were collected at one or two locations in a watershed and
were assumed to be representative for the entire watershed (Haith and
Shoemaker, 1987). The shallow ground-water storage value and recession index
were assumed to represent the entire watershed even though several aquifers
may discharge ground water to surface water. Because of these uncertainties
associated with input values to the model, predicted loading rates to surface
water may not be representative .
d. Representative equations
Ground-water discharge to a surface water is determined using a lumped
parameter water balance model based on daily water balances from the
unsaturated and shallow saturated zones (Haith and Shoemaker, 1987). The
equation describing ground-water discharge is as follows:
where :
Gt = Ground- water discharge [L3/T]
St - Shallow saturated zone moisture content [L3]
r - Ground-water recession constant [1/T].
The loading rate to surface water as a result of ground-water discharge is :
LR - Gt-C
where :
LR - Loading rate to surface water [M/T]
-------
70
Gt - Ground-water discharge rate to surface water [L3/T]
C - Concentration of chemical constituent in ground water
[M/L3].
e. Description of equipment needs
Equipment needed for the method includes
rain gages,
continuous-chart recorder,
soil maps,
crop distribution maps, and
land-use distribution maps.
A PC computer-based spreadsheet would be very useful to an application
of the method.
f. Expertise needed to use the method
A significant amount of effort is required to use this method. The
greatest level of effort is required to classify land into the various
categories, which involves correlating soil type distribution with land use
and crop distribution in the watershed. Knowledge of relationships between
soil type, land use, and crop distribution within the watershed is useful.
After the watershed has been sectioned into representative land-use categories
and the recession constant has been determined, the method becomes a
bookkeeping exercise. A computer spreadsheet can be utilized to multiply and
add the calculated values to estimate the loading rate to surface water.
ii. Data inputs for the model
The model requires data describing land use and soil type distribution
and daily precipitation. The ground-water recession constant can be estimated
using standard hydrograph separation techniques and stream gage data.
iii. Outputs from the model
The method estimates the loading rate to surface water as a result of
ground-water discharge, sediment load, and surface runoff for various land
uses.
a. Ground-water quantity discharge to surface water
The amount of ground-water discharge to surface water predicted by this
method is a function of the recession constant and the storage capacity of the
shallow saturated zone. If both the recession index and the storage value are
small, the predicted discharge rate will be small; conversely, if both are
large, the predicted discharge rate will be large.
-------
71
b. Ground-water quality discharge to surface water
The chemical constituents commonly modeled using this method include
nitrogen and phosphorous (Haith and Shoemaker, 1987). The concentrations
predicted in surface water using these methods are a function of the loading
rates assigned to the various land types and land use distribution.
iv. Settings in which the models have been applied and contaminants
that have been modeled
The settings and the contaminants that have been modeled using this
method are summarized in Table F-l.
v. General evaluation of the method
The method has obvious appeal for many applications because information
concerning land use, soil type distribution, precipitation and temperature
data, and stream stage are readily available for most watersheds.
Additionally, the method is relatively easy to use. Once a computer spread
sheet containing the required inputs has been established for a given
watershed, by inputting weather data, the loading rate to surface water ca.n be
estimated. However, ultimately any model used for management decisions must
be able to predict future loading rates as a result of changes in management
practices. Because loading models rely on a multicomponent loading factor,
the effect of changing one component of the loading factor on surface-water
quality may be difficult to determine, making the models less suited for
management applications.
vi. References to annotated bibliography
References to the accompanying annotated bibliography are summarized in
Table F-2.
-------
Table F-l. Stannary of Settings in Which the Method Has Been Applied and the Contaminants Measured.
Location
Aquifer
Contaminant
Author
Pennsylvania
Walton, Hew York
Inland Bays, Delaware
Chesapeake Bay, Virginia
Wisconsin
Trimmers Rock
Catskill formations
Nitrate
Nitrogen, phosphorous
Nitrogen, phosphorous, BOD
Nitrogen, phosphorous
Nitrogen, phosphorous
K. M. Green
W. J. Gburek, J. B. Urban
R. R. Schnabel
D. A. Haith, L. L. Shoemaker
W. T. Ritter
R. R. Schnabel, W. J. Gburek
P. D. Uttormark, J. D. Chapin
-------
Table F-2. References to Annotated Bibliography
Author
Citation
Reference to Annotated
Bibliography
W. J. Gburek, J. B. Urban
R. R. Schnabel
D. A. Haith, L. L. Shoemaker
W. T. Ritter
R. R. Schnabel, W. J. Gburek
P. D. Uttormark, J. D. Chapin
"Nitrate Contamination of Ground Water in an Upland
Pennsylvania Watershed," Proceedings of the Agricultural
Impacts on Ground Water, A Conference, Omaha, Nebraska,
August 11-13, 1986, pp. 352-380.
"Generalized Watershed Loading Functions for Stream Flow
Flow Nutrients," Water Resources Bulletin. 1987, 23(3):
471-478.
"Nutrient Budgets for the Inland Bays," Report to Delaware
Department of Natural Resources and Environmental Control,
August, 1986.
"Calibration of NFS Model Loading Factors," Journal of
Environmental Engineerinn. 1983.
"Estimating Nutrient Loading of Lakes from Non Point Sources,
Office of Research and Monitoring, 1974, Environmental
Protection Agency report number PA 660/3-74-020.
pp.160-162
pp.163-165
pp.166-167
pp.168-169
pp.170-171
73
-------
74
G. Studies using environmental isotope methods to estimate the contribution
of ground water to stream flow
The papers cited in this section are summarized in Section XI of "An
Annotated Bibliography of the Literature Addressing Nonpoint Source
Contaminated Ground-Water Discharge to Surface Water," September, 1990, EPA
440/6-90-006.
i. General description of method
a. Description of method or procedure
An assessment of naturally-occurring (or, in some cases,
anthropogenically-produced) environmental isotope distributions can assist in
the differentiating among sources of water supplying streams. Three
environmental isotopes are commonly used in studies of runoff generation in
catchments: oxygen-18 (180) , deuterium (D or 2H) , and tritium (T or 3H) .
These isotopes are almost ideal tracers for runoff generation studies, due to
two principal characteristics: (1) since 180, D, and T are constituent parts
of natural water molecules (e.g., H2180, HD160, and HT160) , they travel at the
same rate through the catchment as "average" water (H2160) ; and (2) 180, D, and
T are chemically conservative at the low temperatures associated with most
small watershed systems (i.e., their concentrations do not change by reactions
with the catchment materials). The isotope concentrations in the flow system
are altered only by physical processes such as: mixing, diffusion,
dispersion, and radioactive decay.
Both 180 and D are stable isotopes which occur naturally, accounting for
about 0.2% of all oxygen atoms and about 0.015% of all hydrogen atoms,
respectively. Their average concentrations in water, expressed as H2180 and
HD160 are about 2000 and 320 ppm, respectively. T is a radiogenic isotope of
hydrogen whose half-life is in the order of 12.4 years. T atoms represent an
extremely small proportion of terrestrial hydrogen (about 10"1A to 10"16%) of
all hydrogen atoms. Concentrations of T are expressed in tritium units (TU)
in which ITU = IT/1018 atoms of hydrogen. Although very few T measurements
were made on precipitation prior to the introduction of anthropogenically-
produced T into the atmosphere, indications are that precipitation contains
from 4-25 TU of naturally-produced T. Since the advent of atmospheric testing
of thermonuclear devices in 1952, T produced as a by-product of this testing
has been the dominant source of T in precipitation.
The term "hydrograph separation," discussed in section D of this report
is normally associated with graphical hydrograph separation techniques which
have been used for decades in predicting runoff volume and residence times.
Another type of hydrograph separation, based on natural chemical or isotopic
tracers in water, has been developed as a more "physically-based" runoff
separation technique than the graphical technique described previously. This
hydrograph separation technique apportions storm and snowmelt hydrographs into
contributing components based on the distinctive chemical or isotopic
signatures carried by each of the contributing components. The distinctive
signature of each component is developed as the water molecules passing
through the catchment take different flow paths and have different residence
-------
75
times.
The tracer-based hydrograph separation technique normally involves a
two-component mixing model for the stream. The model assumes that water in
the stream at any time during a storm or snowmelt runoff event is a mixture of
two components: "new water," which is water from the current rain or snowmelt
event, and "old water," which is the subsurface water which existed in the
catchment prior to the current rain.
During baseflow conditions (the low flow conditions which occur between
storm and snowmelt runoff events) in humid, headwater catchments, all the
water in a stream is discharged ground water. The chemical and isotopic
character of stream water at a given location during baseflow represents an
integration of the upstream "old water" discharges. During storm and snowmelt
runoff events, however, "new water" (rain or snowmelt) is added to the stream.
If the "old" and "new" water components are chemically or isotopically
different, the stream water becomes "diluted" by the addition of the "new"
water. The extent of this dilution is a function of the relative
contributions from the "old" and "new" water components.
The precursor to using natural isotopes as tracers in the simple two-
component mixing model involved the use of various chemical parameter tracers
such as total dissolved solids, chloride, and electrical conductivity. These
mixing models are based on the general observation that "old water" has higher
concentrations of most chemical parameters than "new water".
The major problem associated with separating hydrographs on the basis of
chemical tracers is that most chemical tracers are not conservative. The
chemistry of the "new water" may vary both areally and temporally as the rain
or snowmelt water interacts with the catchment materials on the way to the
stream. The chemical tracer technique could lead to an overestimate of the
"old water" contribution since the "new water" progressively gains more
solutes on its way to the stream and its chemistry becomes more like that of
the "old water". One can, however, derive some valuable information
characterizing runoff processes by using natural chemical tracers. Some
parameters, such as silica are fairly conservative and can give reliable "old"
and "new" water components. Other parameters such as chloride in coastal
areas, can give some indication of the amount of "washoff" of atmospherically-
deposited particulates by overland flow.
During the 1960's, hydrologists began to use the separation equations
with anthropogenically-produced radioisotopes.
b. Assumptions involved in using the method
The environmental isotope hydrograph separation method uses an
environmental isotope as the tracer in the separation equations (equations [2]
to [4] in Section G.i.d). If the isotopic signatures of the "old" and "new"
water are different, the contributions of "old" and "new" water in a storm or
snowmelt hydrograph can be determined.
The isotopic signature of the "old" water in a catchment results from
-------
76
the mixing of rain and snowmelt which falls over a long period of time. For
the stable isotopes, the "old" water del values (obtained from solving
equation [1]) remain essentially constant year after year or they develop an
annual sinusoidal cycle with the most depleted values in the winter and the
least depleted values in the summer. Whether the "old" water del values are
constant or cyclical is controlled by the residence time of the "old" water
and the degree of dispersive flow in the watershed. The "old" water tritium
concentrations in small watersheds generally show a gradual decrease with
time, reflecting the progressively lower tritium concentrations in recent
(post-1963) precipitation and tritium decay.
The isotopic signature of the "new" water component provides the
contrast needed for the isotopic separation. Although the general seasonal
variations in the 180, D, and T concentrations in precipitation and the
general long-term decline in tritium levels are documented, it must be
emphasized that there is no guarantee that the "new" water in an individual
rain or snowmelt event will be isotopically different from the "old" water.
Four assumptions govern the reliability of hydrograph separations using
environmental isotopes:
The "new" and "old" water component can be characterized by a
single isotopic value for each component or variations in each
component's isotopic content can be documented.
The isotopic content of the "old water" component is significantly
different from that of the "new water" component.
Vadose zone water contributions to the stream are negligible
during the event or they must be accounted for (use an additional
tracer if isotopically different from ground water).
Surface water storage (channel storage, ponds, swamps, etc.)
contributions to the stream are negligible during the runoff
event.
c. Limitations of the method
One major assumption in using environmental isotopes is that the
baseflow represents the "old" water component and the source of ground-water
flow to the stream during storm and snowmelt events is the same as the source
during baseflow conditions. However, ephemeral springs remote from the stream
or deeper ground-water flow systems may contribute differently during events
and if their isotopic signatures differ from that of baseflow, the assumed
"old water" isotopic value may be incorrect. Although the occurrence of such
situations could be tested by hydrometric monitoring and isotopic analyses of
these features, qualification could be difficult.
Catchments with significant surface storage cannot be accurately
characterized using isotope hydrograph separation methods. Isotopic
enrichment of surface water in lakes, ponds, and swamps by evaporation may
introduce complications in the simple two component model.
-------
77
During some events, the "old" and "new" water isotopic contents may be
too similar for meaningful hydrograph separations. Considering that
substantial time may be spent waiting for and then sampling an event and
considerable costs may be incurred for isotopic analyses, it is prudent to
monitor an event using chemical tracers as well as isotope analysis. These
other methods may include: (1) testing for other independent isotopes such as
T rather than 180 if 180 is unsuitable and (2) testing for a conservative
chemical constituent, such as silica, or other less conservative parameters
such as electrical conductivity.
Estimating the actual isotopic composition of the rainfall reaching the
ground surface is complicated in forested catchments because of the
interception loss (by evaporation) from the forest canopy during rainfall.
Evaporation from water stored on the forest canopy typically occurs at rates
of 0.1-0.5 mm/hr and can account for the loss of about 20% of the gross
rainfall. Depending on the ambient relative humidity at the canopy level,
evaporation of 20% of the rainfall could substantially enrich the D or 180
composition of throughfall and net rainfall compared with that of the gross
rainfall usually measured and sampled. This is a potentially serious problem
only when the gross rainfall is isotopically lighter (more negative in delta
notation; equation [1] below) than the prestorm stream water.
d. Representative equations
Since D and 180 concentrations in natural waters are much smaller than
their common light isotopes (XH and 160) , D and 180 concentrations are
generally expressed in the conventional delta (ft) notation as per mil (0/00)
differences relative to the international standard, SMOW:
ftD or 8180 = (Rsample - RSMOW) X 1000 [1]
RSMOW
Analytical precision for ftD and 6180 by mass spectrometry is better than 2 and
0.2%, respectively, with a confidence level of 95%.
Between storm events, stream base flow reflects the isotopic composition
of the "old" (stored) water. During storm runoff events, however, the
isotopic character of the stream may be altered by the addition of "new" water
from rainfall. The "old" and "new" water contributions at any specified time
can be calculated by solving the mass balance equations for the water and
isotopic fluxes in the stream. These equations are expressed as:
Qs - Qo + Qn [2]
CsQs - C0Q0 + CnQn [3]
Qo = (Cs - Cn) x Qs [4]
(Co - Cn)
where Q is discharge, C expresses tracer concentration, and the subscripts s,
o, and n refer to the stream, "old water," and "new water," respectively. The
-------
78
utility of the mass balance equations for any particular storm event is
controlled mainly by the magnitude of (C0 - Cn) relative to the analytical
error and the recognition of areal and temporal variations in C0 and Cn. The
equations can also provide estimates of "old" and "new" water percentage
contributions to throughflow and overland flow.
Environmental isotope data can also used in estimating the areal extent
of overland flow contributing areas and in calculating mean residence time of
the "old" water in the catchment. Assuming that overland flow is generated
entirely as saturated overland flow, the overland flow contributing area is
estimated by the following equation:
Y - Vn [5]
Vtn
where Y is the discharge area expressed as a fraction of the total basin area,
Vn is the total volume of "new water" which leaves the catchment and Vtn is
the total volume of "new water" which falls on the catchment. The mean
residence time of the "old water" in a catchment can be estimated by comparing
the seasonal variations in del values for the precipitation and the streamflow
or by analyzing the tritium input (precipitation) and output (streamflow)
functions.
e. Description of equipment needs
Stage or precipitation activated time discrete automatic water samplers are
needed to ensure sampling at the start of an event, especially for night-time
storms or for remote catchments. Snowmelt lysimeters are needed for sampling
snowmelt. To sample soil water, lysimeters are required. Ground-water samples
are obtained by installing piezometers. Measurements of streamflow from
catchments require weirs with stage recorders.
The concentrations of 180 and D in a water sample are normally measured
using a double-collecting mass spectrometer which compares the concentration of
180 or D in the water sample to a standard water. Water samples for T analysis
are measured using a liquid scintillation counter.
f. Expertise required to apply this method
The method requires knowledge of basin characteristics and the temporal
distribution of precipitation and runoff in the basin as well as considerable
professional judgement. Site water sampling requires a sufficient
understanding of regional geology and hydrology.
ii. Data inputs for the method
To determine the "old" water isotopic value, the samples to be obtained
include the following: ground water at various sites (shallow or deep) within
the catchment, soil moisture at several sites (shallow or deep), and baseflow
in a first order stream in the catchment or baseflow in a larger order stream.
Most isotopic studies have used either ground water or baseflow to
-------
79
characterize the isotopic content of the "old" water. The Isotopic value of
stream baseflow is a good approximation of the isotopic value of ground-water
discharging into the stream. Soil moisture is also appropriate for the "old"
water component in certain hydrologic environments.
Many samples are needed when conducting environmental isotope studies of
storm and snowmelt runoff; i.e., one must take as many time discrete samples
of the precipitation (and snowmelt) and the stream as possible. Depending on
the nature of the study, sampling frequency may vary from minutes to day or
longer depending on the size of the catchment, type of sampling equipment,
type of event, and detail desired. Through flow, overland and macropore flow
sources can also be separated using isotopic methods.
iii. Outputs for the method
a. Ground-water quantity discharge to surface water
Essentially any quantity of ground-water discharge to surface water can
be estimated using isotopic tracer-based hydrograph separation techniques.
Isotopes leave signatures of stored water (in the unsaturated and saturated
zones) that can be detected at the discharge points. Environmental isotope
results can be used to test whether integrated water quality models represent
catchment processes appropriately.
b. Ground-water quality discharge to surface water
Observed contaminant concentrations in surface water can be correlated
with the runoff components indicated by the isotopic data. For example, if
stream flow is found to be dominated by "old" water during a precipitation
event and observed contaminant concentrations rise above baseflow
concentrations, the increased contaminant levels may presumably arise from
subsurface discharge.
iv. Settings in which the method has been applied
Some of the settings assessed and the isotopes analyzed in
representative studies are presented in Table G-l. The predominant conclusion
from these studies is that "old" water components normally dominate storm and
snowmelt runoff in humid, headwater catchments. These studies demonstrate how
isotope tracer studies can improve the characterization of runoff processes
beyond those findings based upon hydrometric and/or hydrochemical data.
v. General evaluation of the method
Because isotopic tracers are constituent parts of natural water
molecules, they can be used as excellent tracers of water origin and movement.
The long term and widespread application of these tracers analyses, will allow
researchers to study runoff generation on scales ranging from macropores to
portions of catchment slopes to first and higher order streams (Sklash 1990).
-------
80
Several disadvantages may arise in the use of isotopic tracers, however.
Conditions for their use are not met in every event, and sample analysis is
expensive. In some catchments, the isotopic content of "old" and "new" waters
is not distinguishable in the snowmelt, and variability in the "new water"
isotopic component may decrease the precision of the separation.
vi. references to annotated bibliography
References to the accompanying annotated bibliography are provided in Table
G-2.
-------
81
G-l. Summary of Settings in Which the Method Has Been Applied and the Isotopes Measured
Location
Catchment Size (km2)
Isotope
180 D
Author(s)
France
24
E. Crouzet,
P. Hubert,
P. Olive,
E. Siwertz,
A. Marce (1970)
2 United Kingdom 1.0
D.S. Biggin (1971)
3 Netherlands
650 ha
W.G. Mook,
D.J. Groenveld
A.E. Bouwn,
A.J. Van Ganswyk (1974)
4 Canada
22, 1.8
P. Fritz,
J.A. Cherry,
K.U. Weyer,
M.G. Sklash (1976)
5 Canada
73 to 700
M.G. Sklash,
R.N. Farvolden,
P. Fritz (1976)
6 Canada
1, 1.2, 3.9
M.G. Sklash,
R.N. Farvolden, (1976)
-------
82
G-l. Summary of Settings in Which the Method Has Been Applied and the Isotopes Measured
# Location
Catchment Size (km2)
Isotope
180 D
Author(s)
7 Canada
10.5, 1.24, 1.76
D.J. Bottomley,
D. Craig,
L.M. Johnston (1984)
8 USA
620
V.C. Kennedy,
C. Kendall,
G.W. Zellweger,
T.A. Wyerman,
R.J. Avangion (1986)
9 New Zealand 3 . 8 ha
10 New Zealand 3.8 ha, 1.6 ha,
0.3 ha, 2.8
11 Sweden 33.5
12 Australia 82 ha
13 Czechoslovakia 2.65
* A
M
M
* M
M
A
* G
E
G
* * J
D
R
* * T
B
T
J
E
.J. Pearce,
.K. Stewart,
.G. Sklash (1986)
.G. Sklash,
.K. Stewart,
.J. Pearce (1986)
. Jacks ,
. Olofsson,
. Werne (1986)
.V. Turner,
.K. Macpherson,
.A. Stokes (1987)
. Dincer,
. R . Payne ,
. Florkowski ,
. Martinec,
. Tongiorgi (1974)
-------
83
G-l.
Summary of Settings in Which the Method Has Been Applied and the Isotopes Measured
// Location Catchment Size (km2) Isotope Author(s)
T 180 D
14
15
16
17
18
19
20
21
Switzerland 43.4 * J
H
H
E
Canada 11.4, 111 * H
G
H
Canada 2.8 * P
H
P
West Germany 18.7 * A
W
Canada 368 * * F
Canada 2.8 * * M
R
Sweden 4.0, 6.8 * A
Sweden several * A
. Martinec ,
. Siegenthaler ,
. Oescheger,
. Tongiorgi (1974)
.R. Krouse,
. Holecek,
. Steppuhun (1978)
.M. Wallis,
.B.N. Hynes, ;
. Fritz (1979)
. Herrmann,
. Stichler (1980)
.W. Schwartz (1980)
.G. Sklash,
.N. Farvolden (1980)
. Rodhe (1981)
. Rodhe (1984)
-------
84
G-l. Summary of Settings in Which the Method Has Been Applied and the Isotopes Measured
# Location Catchment Size
22 Canada 10.5
23 Norway 41 ha
24 USA 42.2
25 Canada 60
26 Canada 10.5
27 USA
(km2) Isotope Author(s)
T 180 D
* D.
D.
L.
* N.
S.
A.
* R.
C.
* M.
M.
* A.
D.
L.
* J.
J. Bottomley,
Craig,
M. Johnston (1984)
Christophersen,
Kjaernsrod,
Rodhe (1985)
P. Hooper,
A. Shoemaker (1986)
M. Obradovic,
G. Sklash (1986)
J. Bottomley,
Craig, ;
M. Johnston (1986)
1
R. Lawrence (1987)
-------
85
6-2. References to Annotated Bibliography
Author Citation Reference to Annotated
Bibliography
Sklash "Environmental Isotope Studies of Storm and Snowmelt p. 289
and Runoff Generation," In Surface and Subsurface Processes
in Hydrogeology, M. G. Anderson and T. F. Burt, (ed.),
John Wiley and Sons Ltd., Sussex, England, 73 p., 1990,
In print.
Sklash, "Environmental Isotope Tracer Studies of Catchment Processes: pp.293-295
Moore, Tools for Verifying Integrated Water Quality Models," In:
Burch Proceedings of the USDA, AIRS-81, pp. 459-478. International
Symposium on Water Quality Modeling of Agricultural Non-point
Sources.
Hooper, "A Comparison of Chemical and Isotopic Hydrograph Separation," pp.285-286
Shoemaker Separation," Water Resources Research. 1986, pp. 1444-1454.
.oszewski, "Application of Flow Models in an Alpine Catchment Area Using pp.287-288
tert, Tritium and Deuterium Data," Journal of HvdroloKY. 1983, 66:
chler, W. Herrmann 319-330.
Sklash, "The Role of Groundwater in Storm Runoff," Journal of Hydrology. pp.291-292
Farvolden 1979, 43: 45-65.
-------
Chapter 3
The Impact of Nonpoint Source Contaminated Ground-Water Discharge
to Surface Water in Water Quality-Limited Water Bodies:
Determining Total Maximum Daily Load and Waste Load Allocations
Introduction
a. Overview of the chapter
This chapter provides a general overview of the process for determining
the Total Maximum Daily Load (TMDL) for water quality-limited water bodies and
the allocation of point source waste loads and nonpoint source loads to
achieve the TMDL.
As used in this chapter, TMDLs are defined as the assimilative capacity
of a waterbody, which is the sum of the individual Waste Load Allocations
(WLAs) for point sources and Load Allocations (LAs) for nonpoint sources and
natural background (see 40 CFR 130.2(h)), plus a safety factor. A Waste Load
Allocation is the portion of a receiving water's loading capacity that is
allocated to one of its existing or future point sources of pollution (see 40
CFR 130.2(g)). Similarly, Load Allocations (LAs) are the portions of a
receiving water's loading capacity that is attributed either to one of its
existing or future nonpoint sources of pollution or to natural background
sources (see 40 CFR 130.2(f)). In sum, the TMDL should encompass the
contaminant waste loads from point sources and nonpoint sources. However, the
nonpoint source load allocation may be accounted for simply as a component of
background contaminant concentrations. This chapter provides a preliminary
discussion of the rationale for applying the methods described under Chapter 2
above to better measure or estimate the nonpoint source component of the load
allocation under a TMDL.
Under Section 319 of the Clean Water Act, by August 4, 1988 the States
were required to identify those water bodies that were not expected to attain
or maintain their respective water quality standards due to point or nonpoint
source loads. In addition, the States were directed to develop a program to
alleviate these problems by describing how they will utilize the TMDL process
to control nonpoint source pollution in accordance with Section 319 (b) (2)
(B) of the Clean Water Act. This Section calls for "an identification of
programs to achieve implementation of the best management practices (BMPs) by
the categories, subcategories, and particular nonpoint sources designated
under subparagraph (A)." Subparagraph (A) requires an identification of the
BMPs and measures which will be undertaken to reduce pollutant loadings
resulting from each category, subcategory, or particular nonpoint source
designated under Section 319 (a) (1) (B). Presently,.this requirement is the
only regulatory tool available under the Clean Water Act to promote nonpoint
source controls. To date, the Agency has prepared a variety of guidance
documents and models to assist in determining TMDLs as part of the water
-------
87
quality.-based permitting process (see.-Appendix. A of-. this document.). * Thisv.
discussion specifically addresses the manner in which nonpoint source loads
may be accounted for in this process.
b. Organization of the chapter
The chapter is organized in four sections. Following this introduction,
Section 2 introduces the regulatory concepts and statutory authorities
mandating the TMDL and load allocation processes. Section 2 also provides a
summary of the status of WLA applications and water quality-based permitting.
Section 3 discusses the theory behind the TMDL process and the application of
WLAs with regard to estimating water quality impacts from biochemical oxygen
demand, nutrients, and toxic substances and provides a limited overview of WLA
modeling approaches. Finally, Section 4 reviews the applicability of the
methods described in 2 to supply the data needed to assess nonpoint source
loads as part of a TMDL analysis.
A. Statutory and regulatory mandate for determining WLAs and LAs under the
TMDL process
i. Rationale behind waste load allocation and water-quality based
permitting
Under Section 303 of the Clean Water Act, the States are required to set
water quality standards that protect the public health or welfare, enhance the
quality of water, and serve the purposes of the Act for all waters in the
State.2 These standards are based on water quality criteria developed by
U.S. EPA3 and are to guarantee the achievement of a designated use for the
water body. The State may not set a water body's designated use at less than
fishable/swimmable without performing a use attainability analysis.*
is in the process of preparing a series of nine Waste Load
Allocation guidance documents. Those documents that are currently available
from the Monitoring and Data Support Division/U.S. EPA are listed in Appendix
A to this document.
2 See also 48 FR 51400 for the regulations implementing the water
quality standard process.
3 These water quality criteria are developed under Section 304(a) of the
Clean Water Act. Quality Criteria for Water 1986. published May 1987, is the
most recent EPA summary of water quality criteria.
4 Use attainability analyses involve a determination of the level of
aquatic protection that can be achieved for a water body. The analysis
includes an assessment of (1) what are the aquatic uses(s) currently being
achieved in the water body, (2) what are the potential uses that can be
attained based on the physical, chemical, and biological characteristics of
-------
88
Sections 302 and 304(1) of the Clean Water Act require the States to
identify those waters for which technology-based effluent limitations are not
sufficiently stringent to attain the water quality standards. The technology-
based limits are mandated under Sections 301 and 307 of the Act and are
implemented by the Agency through the promulgation of industry-specific
effluent guidelines.5 The States must rank their water quality-limited
stretches for planning purposes and set total maximum daily loads of
pollutants in the stretches that will achieve the applicable standard.
Finally, the TMDLs are to be converted to wasteload allocations through
modeling and ultimately to water quality-based effluent limitations on
individual point source dischargers in the limited stretch.
. 11. Implementing waste load allocations and load allocation in water
quality-limited water bodies
Water quality-based controls are implemented for any stream segment in
the water body, and (3) what are the causes of any impairment of the uses?
See Technical Support Manual: Water Body Surveys and Assessments for
Conducting Use Attainability Analyses. November 1983. USEPA/OW.
5 Under Sections 301 and 307 of the Clean Water Act, as amended in 1972,
EPA is responsible for promulgating technology-based effluent guidelines, and
applying these guidelines in permits to industrial point source dischargers.
EPA is to review standards annually and to revise them every three years.
Equivalent technology based standards also apply to municipal discharges;
these have been defined by EPA as secondary treatment of municipal wastewater.
Under the 1972 amendments, industrial point sources were required to
apply the best practicable control technology currently available (BPT) to
their processes by July 1, 1977. BPT was interpreted as involving mainly
"end-of-pipe" controls that imposed control costs and economic impacts that
were not "wholly out of proportion" with water quality benefits. In the
second phase of pollution control, the Act mandated that industries were to
adopt best available technology economically achievable (BAT), or, if
feasible, zero discharge, by July 1, 1983. In contrast to BPT, BAT was
thought of primarily as in-plant process changes that had been or were capable
of being achieved. Compliance costs were considered in setting BAT, but no
cost-benefit analysis was necessary as with BPT. Finally, new sources were
expected to immediately comply with strict standards of performance based on
best available demonstrated control technology (BACT), a standard comparable
to BAT for existing sources.
Under the 1977 amendments to the Act, Congress modified the original
technology-forcing approach somewhat to include a new category of control,
best conventional pollutant control technology (BCT), to be achieved by July
1, 1984. However, EPA found the BCT cost tests difficult to apply and, as a
result, for most industry categories the BCT effluent limitations are
virtually equivalent to BPT requirements.
-------
89
which.it is; know that water quality does not meet applicable water quality
standards, and/or is not expected to meet applicable water quality standards,
even after the application of the technology-based effluent limitations
required by Sections 301 (B) and 306 of the Clean Water Act (see 40 CFR
130.2(i)). For these segments, water quality-based effluent limits may be
imposed on point source dischargers under the authority of Sections 302 and
402 of the Clean Water Act. Water quality-based effluent limits are derived
to protect water quality in the receiving water regardless of cost or waste
stream treatability. In addition, however, Section 302 of the Act makes
provision for permitees to apply for variances from the water quality-based
effluent limits based upon the "relationship between the economic and social
costs and the benefits to be obtained from achieving such limitation."
Control programs for nonpoint sources are developed as part of the
planning process described under Section 319 of the Clean Water Act. In many
cases, States do not have certified or established techniques or procedures
for completing load allocations. In those cases, the State should present its
overall schedule for implementing the TMDL process and a schedule with
milestones for establishing the appropriate load allocations. Until the load
allocation is approved by EPA, the State should pursue a technology-based
approach. Technology-based controls are to be based upon water quality
considerations and not just resource protection. For example, while
agricultural management activities are directed at minimizing soil erosion for
productivity purposes, the technology-based approach requires the
landowner/operator to include not only productivity based controls but offsite
measures such a filter strips and sediment and water control structures, as
well. If BMP implementation is not adequate, the State should develop an
action plan to develop additional BMPS, including a schedule to assess the
water quality conditions and determine if standards are being met within an
appropriate timeframe.
B. Determining the Total Maximum Daily Load
This section presents a brief description of the scientific
understanding of the processes underlying estimates of total maximum daily
load. Because the approach for assessing water quality impacts from different
categories of contaminants varies, each of three classes of substances,
biochemical oxygen demand, nutrients, and toxics is discussed separately.
i. Theory behind Waste Load Allocation
Basic Principles
Total Maximum Daily Load assessments provide information to assist in
making effective decisions on levels of treatment required for a source or
sources of pollutant load. WLAs are water quality oriented and are directed
at establishing a qualitative relationship between a particular waste load and
its impact on water quality. These relationships make it possible to compare
incremental changes in concentrations of specific constituents in the
receiving water system. One is then able to identify the maximum waste load
-------
90
that can be discharged-without violating a-water quality standard.6
Because of the array of variable elements (e.g., temperature, stream
flow, load level, reaction rates) that must be considered in WLAs,
computerized mathematical models are generally employed to make the necessary
calculations. Furthermore, the factors and model formats also differ
depending on the water contaminant under investigation. The approaches taken
for each of three major classes of water contaminants assessed in determining
TMDLs are discussed below.
Biochemical Oxygen Demand/Dissolved Oxygen
Biochemical oxygen demand (BOD) is a measure of the amount of oxygen
used in stabilizing a biodegradable material. Both carbonaceous organic
compounds (CBOD) and nitrogenous forms (NBOD), such as ammonia and organic
nitrogen, are subject to bio-oxidation. In most WLA applications, the amount
of oxygen consumed for biodegradation over a five-day period (BOD5) is used as
the standard measurement. However, full oxidation of organic compounds
generally requires in excess of twenty to thirty days for completion.7
When an organic waste is loaded into a water body, it is subjected to
two processes that influence the transport of the waste: (1) advection, which
represents the downstream transport of the waste load in stream flow, and (2)
dispersion, which encompasses the turbulent and eddying processes that tend to
mix the waste load with upstream and downstream waters. Under steady-state
conditions (i.e., constant waste load and stream flow), the advection and
dispersion processes can be assumed to be constant. In many WLA applications,
one may assume steady-state conditions because critical low-flow conditions
are modeled (e.g., the 7-day, 10-year low flow period). However, if
conditions vary, the transport processes will also vary.
The biochemical oxygen demand and the resulting dissolved oxygen levels
in the water body are a function of the ability of naturally occurring
bacteria to decompose the organic waste materials, thereby utilizing the
oxygen resources of the water body. Replenishment of the oxygen resources in
the water body occurs either through transfer of atmospheric oxygen into the
water column or, to a lesser extent, through oxygen production by aquatic
plants.
The interaction of these processes produces the reduction in dissolved
oxygen levels which is the focus of WLA modeling. The critical factor in the
protection of water quality is an understanding of the rate at which
reaeration takes place and the magnitude of this rate in relation to the rate
of oxygen consumption. This relationship is generally expressed in terms of
6 Technical Guidance Manual for Performing Waste Load Allocations.
General Guidance.
7 Technical guidance Manual for Performing Waste Load Allocation. Book
2: Streams and Rivers. Chapter 1: Biochemical Oxygen Demand/Dissolved Oxygen.
PB86-178936. September 1983. p. 2-13.
-------
91
an oxygen deficit, which is defined as the difference in concentration between
the saturation value and the actual dissolved oxygen concentration.8
Nutrients and Eutrophication
The major water body impact associated with nutrient loading is
eutrophication, or enrichment of the biological productivity of the water
body. Waste load allocations for control of eutrophication are generally
designed to reduce nutrient inputs. This strategy presumes that the nutrient
to be controlled limits the rate of growth and subsequent population of
phytoplankton. It further presumes that reducing the population level of
phytoplankton will provide the desired control of the complex process of
eutrophication and eliminate undesirable water qua-ltty situations such as
algal blooms. Therefore, it should be noted that WLAs to control
eutrophication in water bodies focus directly on nutrient reductions and
indirectly on phytoplankton and dissolved oxygen conditions that result from
overstimulation by nutrients.9
Nutrient levels in water bodies are controlled by external and internal
sources. External sources of nutrients include municipal and industrial point
sources, stream inputs, atmospheric deposition, urban runoff, ground-water
discharge, agricultural drainage, and other nonpoint sources. Internal
sources include sediment release, biological recycling, and nitrogen fixation.
Chemicals and Toxic Substances
The procedure for developing realistic mathematical models for chemical
fate is similar to the mass-balance approach used for other measures of water
quality, such as biochemical oxygen demand. The main differences involve the
modeling of processes affecting the chemical constituent. These processes
include chemical partitioning between the soluble phase and adsorption onto
particulate matter, chemical transfers and kinetics involved in the decay or
volatilization of the constituent, and sedimentation processes. In conducting
WLA for toxic substances, all of these processes are accounted for in a mass
balance equation. The result is a prediction of chemical concentrations in
the water column, sediment, and, in some cases, in the biota present in the
water body'.
The fundamental transfer and kinetic characteristics are known for a
wide variety of chemicals based upon laboratory analyses. These
characteristics can be combined with other relationships, such as advection
and dispersion predictions, to account for the manner in which any material is
8 Ibid, p. 2-19.
9 Technical Guidance Manual for Performing Waste Load Allocations. Book
4: Lakes and Impoundments. Chapter 2: Nutrient/Eutrophication Impacts. PB86-
178928. August 1983. p. 1-4.
-------
92
transported in a water body.10
ii. Waste Load Allocation models: Steady-state conditions
General Approach
Conservation of mass is the fundamental principle which is used as the
basis of all mathematical WIA models of real world processes. All material
must be accounted for whether transported, transferred, or transformed. A
rate equation which conforms to the requirements of mass balance is
V-dc/dt - J + *T + *R + *W
where
c - concentration of the chemical
J = transport through the system
T - transfers within the system
R - transformation reactions within the system
W - chemical inputs
V = volume of water body
t - time.
This fundamental model forms the basis for assessing pollutant load to a
water system. Most WLA applications also assume steady-state conditions,
thereby eliminating the need to measure changes in parameters over time. The
simplified steady-state framework for chemical WLA modeling also assumes
complete mixing throughout the water body.
A steady-state model requires single, constant inputs for effluent flow,
effluent concentration, background receiving water concentration, receiving
water flow, and meteorologic conditions. As a result, the effects of
variability in nonpoint source and point source contaminant discharge on
receiving water quality cannot be predicted accurately using these steady-
state techniques. Nonetheless, steady-state models provide a relatively
simple and conservative tool for estimating water quality impacts from
contaminant discharges. The specific analytical approaches for steady-state
WLA modeling for BOD, nutrients, and toxic substances are described below.
Biochemical Oxygen Demand and Dissolved Oxygen Profile
A dissolved oxygen profile for a stream reach is based upon a simple
mass balance which accounts for the mass of BOD entering a stream reach, the
mass leaving the reach, and the biodegradation and reaeration processes that
occur within the reach that result in the oxygen sag. At steady-state, the
10 Technical Guidance Manual for Performing Waste Load Allocations. Book
4: Lakes. Reservoirs and Impoundments. Chapter 3: Toxic Substances Impact.
EPA 440/4-87-002. December 1986. p. 6.
-------
93
following mass balance applies:11
MASS IN - MASS OUT + SOURCES - SINKS - 0
QC - Q-(C + dC/dx-»x) + Ka-(Ca-C)-V - Kj-L-V - 0
if U - Q/A and V - A-»x, then:
Q-dC/dx-*x - (Q-*x)-dC/dx - (U-A-*x) dC/dx - U-V-dC/dx, and
-U-dC/dx + Ka- (C8-C) - Kd-L - 0
if the oxygen concentration (C) is expressed in terms
of oxygen deficit (D) and the saturated oxygen
concentration (Cs), then C - Cs - D, and
-U-d(C8-D)/dx + Ka-D - K.J-L - 0
if Ca is constant over all x, then
U- dD/dx + Ka- D - Kd- L - 0
the rate of change in biochemical oxygen demand
concentration (L) is expressed as:
L - L^
therefore,
U-dD/dx + Ka-D = Kd
integrating and solving the equation for the condition
that D-DO at x-0 yields the following:
D - D0-e<-K--*/u> +
where
Q - river flow rate
C - concentration of dissolved oxygen entering the segment
Cs - saturation concentration of dissolved oxygen
QC - mass of oxygen entering the segment
dC/dx - rate of change of oxygen (C) with distance (x); equivalent
to rate of change with time (t) when converted by velocity
(U)
dC/dx-*x = change in oxygen concentration during time of passage
11 Technical Guidance Manual for Performing Waste Load Allocation. Book
2: Streams and Rivers. Chapter 1: Biochemical Oxygen Demand/Dissolved Oxygen.
September 1983. PB86-178936, p. 2-40.
-------
94
through segment of length *x
Ka - atmospheric reaeration rate coefficient
Kr - BOD removal rate coefficient (-K,j).
This is a steady-state solution for the oxygen deficit in a stream
segment of length *x. The source term for biochemical oxygen demand (L)
represents the concentration of BOD within the mixing zone below a point
source outfall. However, the source may also include nonpoint loading from
ground-water discharge.
Nutrients
In lakes and estuaries, nutrient inputs may promote increased biological
productivity or eutrophication. Such eutrophication processes depend upon
continual input of nutrients, as a net sedimentation of nutrients occurs over
time. This process can be expressed in a general steady-state, mass balance .
equation which assumes a completely mixed water body. The removal rate of the
nutrients is assumed to be proportional to their water concentration, which is
expressed as follows12:
V-dP/dt - EQi-Pi - Ks-p-V - Qp
where
ZQi-Pi - the sum of all the mass rates of total nutrients
discharged to the lake from all sources (point source
and nonpoint source) [M/T] ; QA = flow [L3/T] ; and PA -
the initial nutrient concentration [M/L3]
p - lake nutrient concentration [M/L3]
V » lake volume [L3]
Kg- the net sedimentation rate of the nutrient [1/T]
Q - lake outflow [L3/T]
assuming steady-state (dp/dt = 0), and letting W = EQj/Pi:
p - W/(Q + K.-V)
if V - A-z (where A- lake surface area and z = mean depth), then:
p - W/[A-z((Q/V) + K,)]
This expression provides a simple estimate of the ambient lake water
12 Technical Guidance Manual for Performing Waste Load Allocations.
Book 4: Lakes and Impoundments. Chapter 2: Nutrient/Eutrophication Impacts.
PB86-178928. August 1983. p.3-18.
-------
95
nutrient concentration.given a loading, rate of- W. However, the equation-does
not provide any indication of the water quality impacts resulting from the
nutrient loading. Such an estimate could be made based upon the ambient
nutrient concentration.
Toxic Substances
As mentioned above, the modeling approach for toxic substances is
similar to that used for BOD and dissolved oxygen depletion WLAs. One of the
principal differences between the approaches arises in the modeling of the
various fate processes and reaction rates affecting organic and heavy metal
toxic constituents. Several references list reaction rates for toxic
constituents.13 If a first-order decay rate estimate-for aparticular
constituent is available, the following equation for estimating the downstream
concentration of the contaminant may be used:
C - Co- e-K(*/u>
where
C - downstream concentration
C0= concentration at the mixing zone
x - distance downstream of mixing zone
u = river velocity
K - measured decay rate.
However, the fate of many toxic constituents currently is not well understood
because of the confounding effects of varying temperature, pH, and other
environmental conditions in a water body.
EPA's TMDL guidance for modeling individual toxicants in streams and
rivers recommends the following steady-state models: Simplified Lake/Stream
Analysis (SLSA), Michigan River Model (MICHR1V), Chemical Transport and
Analyses Program (CTAP), Exposure Analysis Modeling System (EXAMS), and Metals
Exposure Analysis Modeling System (MEXAMS). All of these models except MEXAMS
can simulate both organic chemical and heavy metal fate and transport in
rivers. EPA also recommends these steady-state models for modeling individual
toxicants in lakes and reservoirs.
In addition to steady-state models, research has continued on the
development of dynamic or continuous simulation models. These modeling
approaches are discussed below. EPA recommends the following continuous
simulation models for rivers, streams, and lakes: Estuary and Stream Quality
Model (WASTOX), Chemical Transport and Fate Model (TOXIWASP), Channel
Transport Model (CHNTRN), Finite Element Transport Model (FETRA), Sediment
Contaminant Transport Model (SERATRA), Transient One-dimensional Degradation
and Migration Model (TODAM), and Hydrologic Simulation Program-Fortran (HSPF).
All of these continuous models are designed for multiple reach, multiple
13 For example, see Water Related Fate of 129 Priority Pollutants.
Volumes I and II. EPA-440/4-81-014.
-------
96
source analyses of both organics and heavy metals.14
Detailed descriptions of these steady-state and dynamic models are
provided in the EPA guidance documents referenced in Appendix A at the end of
this document.
ill. Dynamic Wasteload Allocation Modeling
At the present time, most States and EPA Regions use steady-state
models, which assume the wastewater is completely mixed with the receiving
water and the contaminant source loads are constant, to calculate WLAs for
pollutants. This assumption may he adequate for conventional pollutants
because the greatest environmental impact in the receiving water, such as
severe oxygen depletion, is found downstream of the pollutant outfall.
However, for toxic pollutants the highest concentration in the receiving water
(i.e., near the pollutant outfall) may serve as the critical level for
determining the waste load for the contaminant. As a result, dynamic modeling
approaches are increasingly being applied to better account for variations in
point source loads and variations in ambient water conditions resulting from
changes in nonpoint source loads and other factors.
Dynamic TMDL models calculate an entire probability distribution for
receiving water concentrations rather than a single worst case based on
critical conditions. The prediction of complete probability distributions
allows the risks inherent in alternative treatment strategies to be
quantified. The dynamic modeling techniques have an additional advantage over
steady-state modeling in that they determine the entire effluent concentration
distribution required to produce the desired frequency of criteria compliance.
Continuous or dynamic simulation models use each day's effluent flow
(Qe) and concentration data (Ce) with each day's receiving water flow (QB) and
background concentration (C8) to calculate downstream receiving water
concentrations. The model predicts these concentrations in chronological
order with the same time sequence as the input variables. The daily receiving
water concentrations can then be ranked from the lowest to the highest without
regard to time sequence. A probability plot can be constructed from these
ranked values, and the occurrence frequency of any one-day concentration of
interest can be obtained.15
Several methods are available to compute the probability that downstream
toxicants (or effluent toxicity) will exceed criteria. These approaches
include an approximate method of moments and numerical integration.16 The
14 Technical Support Document for Water Quality-based Toxics Control.
EPA-440/4-85-032.
15 Technical Support Document for Water Quality-based Toxics Control.
September 1985. EPA-440/4-85-032. p. 40.
16 Ibid, p. 41.
-------
97
method of moments is based on the following equation:
Ct - Ce- [Q,/(Q0-KJ»)] + C8- [l-(Qe/(Qe-K}8))]
where Ct - downstream concentration of the contaminant at time t. Estimates
of the mean and variance of the effluent concentration, effluent flow, and
upstream concentration can be made by regressing the natural log of each of
these variables against a standard lognormal random variable. More specific
information concerning the variation in each of these terms may also be
applied.
An additional dynamic modeling approach involves the use of Monte Carlo
simulation. Monte Carlo combines probabilistic and deterministic analyses
since it uses a fate and transport mathematical model with statistically
described inputs. While Monte Carlo simulations require more input data and
calibration data than steady-state modeling approaches, they can account for
interactions of time-varying water quality, flow, temperature, and point and
nonpoint source loading terms.
The above discussion presents a brief overview of modeling approaches
for WlAs. More detailed descriptions of the approaches outlined above are
available in the documents listed in Appendix A. The following section
discusses how information characterizing nonpoint contaminated ground-water
discharge to surface water may be incorporated in the WLA process.
C. Assessment of Nonpoint Source Contaminated Ground-water ischarge to
Surface-water Analysis Methods as Components of Waste Load Allocation
i. NFS loading in current TMDL models
In steady-state TMDL models, all source terms, including NFS loads, are
assumed to be constant. Therefore, one may conclude that NFS loading is
accounted for as a component of ambient water quality conditions. However,
because the application of steady-state models typically focuses only on point
source loads, the contribution of nonpoint source inputs within the water body
segment of concern may not be adequately assessed. Furthermore, if the WLA is
determined for other than low flow conditions, the NFS load may vary greatly
and a significant contaminant source term may be overlooked.
The need for a more thorough understanding of the change in ambient
water body conditions brought about by contaminated ground-water discharge
increases for dynamic WLA modeling applications. This understanding should
include an assessment of the magnitude of NFS loading throughout the water
body segment of concern and the spatial and temporal variations in that
loading.
The accuracy of the waste load allocation process is highly dependent on
the quality of the data used for simulation modeling. This data includes
information concerning the ambient conditions in the water body, spatial and
temporal variations in the source loading terms, and a detailed understanding
of the kinetics of contaminant fate and transport. WLA models should also be
calibrated and verified prior to allocating waste loads. Sufficient
-------
98
historical data to accomplish these objectives are often lacking, however, or
of the wrong type. Therefore, improved data collection is often needed to
better quantify ambient conditions and anticipated loadings.17
In addition to characterizing ambient conditions, a firm understanding
of the contaminant source terms should be obtained. For systems that are not
in low flow (i.e., near steady-state) conditions, this understanding includes
a quantitative measure of source constituent and concentration levels over
time. For point sources, this information is readily available through
analyses of permit conditions or past operating practices. For nonpoint
sources, however, the amount of information available to characterize the
source terms accurately typically is limited.
A sampling program to support a WLA assessment should, at a minimum,
include the following sampling locations within the stream segment: upstream
boundary, point source, upstream of point source, mouth of any tributaries
entering the segment, upstream of the tributaries, upstream of any nonpoint
sources, downstream of nonpoint sources, and downstream boundary of segment.
In areas where significant nonpoint source loadings are known to exist, both
the flow rate and constituent concentrations should be measured. If this area
is not so large that other water quality changes are likely to occur during
the travel time through the area, it is reasonable to assume that the changes
in concentrations are due to the nonpoint sources and to use these differences
as a basis for estimating the loads.18 However, if the level of nonpoint
source loading is significant, a more thorough characterization of the
nonpoint source term may be needed.
The following section reviews the applicability of the methods described
in 2 above to better characterize nonpoint source loading as part of the TMDL
assessment process.
ii. Analysis of contaminated ground-water discharge to surface water
assessment methods as sources of data for waste load allocation
As described above, there is no single analytical approach to waste load
allocation. The TMDL analysis and the type and amount of data required for an
assessment will differ depending upon the water body characteristics, the
point and nonpoint source contaminant loads, and the level of water quality
impairment. Furthermore, in many situations there may be no need to
characterize the component of the ambient water contaminant concentrations
contributed by nonpoint source loads. Such a circumstance may arise if the
nonpoint source load is minimal and limited controls on the point sources
within the watershed will achieve the applicable water quality standard. On
17 See Stream Sampling for Waste Load Allocation Models. EPA/625/6-
86/013.
18 Stream Sampling for Waste Load Allocation Applications. EPA-625/6-
86-013. p. 2-7.
-------
99
the other hand, if nonpoint sources contribute a large portion of the ambient
contaminant concentrations in a water body and stringent controls on point
source discharges will not achieve the water quality standard, there may be a
strong incentive to characterize the contaminant load provided by ground-water
discharge to support development of a nonpoint source management strategy.
This section discusses the applicability of the various contaminated ground-
water discharge to surface water analysis methods for supporting such nonpoint
source load assessments.
All of the methods described in Chapter 2 will provide an estimate of
the loading of nonpoint source contaminants to a water body. However, the
methods differ significantly in the level of effort required, the degree of
specificity of the collected data, and the ability to accurately assess
temporal and spatial variations in nonpoint source loads. As a result,
different methods may be suitable for different levels of analysis. For
example, in water bodies that are severely water quality-limited and that have
high nonpoint source loads, the ability to accurately predict changes in
contaminated ground-water discharge may be critical to support dynamic load
allocation modeling. In contrast, for water bodies that are not as severely
water quality-limited, a fundamental understanding of the component of the
water contaminant concentration contributed by ground-water discharge at base
flow or steady-state conditions may be sufficient for determining the TMDL.
The following table summarizes four attributes of the contaminated
ground-water discharge to surface-water analysis methods. The
characterizations are very general in nature and are intended to provide only
a "first-cut" assessment of the various attributes of the methods. Because
the categorization of the methods necessarily combines several different
approaches under one general method heading, a more detailed review of each
method application is needed to better assess the relevance of the approach to
a particular situation. Nonetheless, this summary allows one to compare and
contrast the suitability of the methods for specific applications.
The attributes are as follows: (1) resources needed to implement the
method; (2) ability to assess spatial variations in ground-water discharge to
a stream segment or water body; (3) ability to measure changes in ground-water
discharge levels over time; and (4) the level of confidence in the method's
ability to provide data that accurately reflects the "true" level of
contaminated ground-water discharge to a water body. The attributes for each
of the methods are ranked relative to one another. A more detailed analysis
of specific method applications would be needed to provide absolute measures
of the attributes for each method.
-------
100
Seepage Meter/ Hydrograph Total Flux Numerical Loading Geophysical Isotopic
Mini-piezometer Separation Measurement Models Functions Methods Methods
Resources:
Spatial
Variation:
Temporal
Changes:
Data
Specificity:
moderate
possibly, with
multiple sample
points
yes, with
repeated
samples
high
low
no
yes
moderate
moderate high
no
yes
moderate
yes
yes
low
yes
no
moderate to low
high
high
yes
yes
moderate
high
yes
yes
high
This analysis indicates that no one method may be suitable for all
applications. Nonetheless, one or more of the methods can provide
sufficient data to support load allocations for many applications and
environmental settings.
D. Summary
The preceding discussions outlined several methods that have been
applied in a variety of environmental settings to assess nonpoint source
contaminated ground-water discharge to surface water. Each of the methods
is suitable for different applications and settings and the resources
required to implement each of the methods also differ. An enhanced
understanding of nonpoint source contaminated ground-water loadings to
surface water may also improve the total maximum daily load assessment
process in water quality-limited water bodies. These methods can support
point and nonpoint source load allocations by better characterizing the
component of ambient water quality contamination contributed by nonpoint
sources under steady-state conditions and by improving the ability to
characterize and predict changes in contaminated ground-water loading in
dynamic simulation models. The manner in which several of the methods
described above can be applied to better account for nonpoint source
loading to a stream is the focus of a companion volume to this document.
------- |