Groundwater Modeling Policy-20070531

Groundwater Modeling Policy
May 31, 2007
NCDENR Division of Water Quality – Aquifer Protection Section
Table of Contents
page
Introduction ..................................................................................................................................... 2
(1) Purpose of policy.................................................................................................................. 2
(2) Basis of technical approach .................................................................................................. 2
(3) Particular groundwater flow or transport model chosen by investigator ............................. 2
(4) Types of predictive calculations or modeling ...................................................................... 3
(a) Groundwater mounding calculations .......................................................................... 3
(b) Groundwater contaminant transport calculations applied to investigating and/or
maintaining groundwater standards ........................................................................................ 3
(c) Groundwater contaminant transport calculations applied to evaluating potential
impact to surface waters.......................................................................................................... 3
Section I: Groundwater Modeling Process .................................................................................... 5
(1) Define study objectives ........................................................................................................ 5
(2) Develop an initial conceptual model and data collection plan ............................................. 6
(3) Collect required data ............................................................................................................ 6
(4) Refinement of conceptual model.......................................................................................... 7
(5) Select a computer model ...................................................................................................... 7
(6) Construct a groundwater model ........................................................................................... 7
(a) Flow sources and sinks ............................................................................................... 7
(b) Boundary conditions and extent of model .................................................................. 8
(c) Regional groundwater gradient ................................................................................... 9
(d) Hydraulic properties.................................................................................................... 9
(e) Parameters used in transport models......................................................................... 10
(f) Groundwater to surface water models ...................................................................... 10
(g) Contaminant source concentration used in transport models ................................... 10
(h) Data deficiencies and uncertainty ............................................................................. 11
(7) Calibrate groundwater flow and transport model............................................................... 11
(8) Use models for predictive simulations ............................................................................... 12
(a) Groundwater flow and mounding predictive simulations......................................... 12
(b) Groundwater transport simulations ........................................................................... 12
(9) Conduct sensitivity analysis of calibrated models and predictive simulations .................. 13
(10) Perform mass balance calculations .................................................................................. 13
(11) Performance Monitoring and Model Refinement ............................................................ 14
Section II: Reporting Modeling Results ...................................................................................... 15
References ..................................................................................................................................... 16
Appendix A ................................................................................................................................... 18
Dispersion, Chemical Retardation Factor, and Chemical Biodegradation Decay Rate
Parameters used in Transport Models ....................................................................................... 18
Appendix B ................................................................................................................................... 20
Example Calculation of Nitrate Source Concentration to be used in a Groundwater Transport
Model for Land Application Systems Utilizing a Cover Crop ................................................. 20
Introduction
The phrase “predictive calculations or modeling”, or variations of this phrase, appears several
times in the North Carolina Administrative Code Section 15A NCAC 02L and 15A NCAC 02T
regulations. This Division of Water Quality (Division) policy provides guidance regarding this
phrase as it pertains to these regulations, and is not intended as a stand alone step-by-step manual
for conducting groundwater modeling.
(1) Purpose of policy
The dual purpose of this policy is to: (a) provide guidance to investigators in selecting and using
appropriate groundwater models for both permitted sites and incident investigations; and (b)
provide guidance for regulators to use in evaluating the adequacy of groundwater modeling
results submitted by investigators.
(2) Basis of technical approach
This policy is based on the following guides published by the American Society for Testing and
Materials (ASTM, available at www.astm.org):
D 5447-93 Application of a Ground-Water Flow Model to a Site-Specific Problem
D 5609-94 Defining Boundary Conditions in Ground-Water Flow Modeling
D 5610-94 Defining Initial Conditions in Ground-Water Flow Modeling
D 5611-94 Conducting a Sensitivity Analysis for a Ground-Water Flow Model Application
D 5490-93 Comparing Ground-Water Flow Model Simulations to Site-Specific Information
E 978-92 Evaluating Mathematical Models for the Environmental Fate of Chemicals
and also on the following other sources:
“Groundwater Modeling Guidance”, by Richard J. Mandle, Groundwater Modeling Program,
Michigan Department of Environmental Quality, 2002.
“Groundwater Modeling for Hydrogeologic Characterization”, by the California Environmental
Protection Agency, July, 1995.
Ground-Water Modeling, by James W. Mercer and Charles R. Faust, National Water Well
Association, 1981.
A Review of Field-Scale Physical Solute Transport Processes in Saturated and Unsaturated
Porous Media, by L.W. Gelher, A. Mantoglou, C. Welty, and K.R. Rehfeldt, Electric Power
Research Institute (EPRI) EA-4190, Palo Alto, CA., 1985.
Contaminant Hydrogeology, by C.W. Fetter, Macmillan Publishing Co., New York, 1993.
Groundwater Transport: Handbook of Mathematical Models, by I. Javandel, C. Doughty, and
C.F. Tsang, American Geophysical Union Water Resources Monograph 10, Washington D.C.,
1984.
Other references were also used in specific areas of the guidelines. See the complete list of
references at the end of this document.
(3) Particular groundwater flow or transport model chosen by investigator
By its nature groundwater modeling must be site specific, and site characterization (as
documented by the report documenting the hydrogeologic evaluation) must precede selection of
an appropriate groundwater model. The Division requires that any model used on a Division-
DWQ GW Modeling Policy 070531.doc 2 OF 21
regulated project: (1) be thoroughly documented in readily accessible published format; (2) be
peer-reviewed in the scientific literature [includes appropriate government publications and
reviews published by or in cooperation with the International Ground Water Modeling Center];
and (3) be appropriate to the site under investigation, as determined by these guidelines. If there
is uncertainty whether the use of a particular model will be accepted, contact the Aquifer
Protection Section (Central Office) at [phone].
(4) Types of predictive calculations or modeling
There are three types of predictive calculations or modeling described by the Division in these
guidelines: (a) groundwater mounding calculations, (b) groundwater contaminant transport
calculations where the groundwater standards are to be protected, and (c) groundwater
contaminant transport calculations to surface water bodies where the surface water quality
standards are to be protected. These three types of predictive calculations or modeling are
performed for the following regulatory purposes:
(a) Groundwater mounding calculations
The permitted disposal and utilization systems in Subchapter 02T have a design criteria
of maintaining a one-foot minimum vertical separation between the seasonal high water
table and the ground surface. The intention of this regulation is to maintain a minimum
of one foot of vertical separation between the applied waste (ground surface if applied on
the ground) and the altered or mounded seasonal high groundwater table in order to
facilitate soil remediation effects of the applied waste. The “altered or mounded” term is
meant to signify the additive or compounded effects of the disposal activity onto the
ambient seasonal high groundwater table. This mounding effect onto the seasonal high
groundwater table is usually obtained via predictive calculations or modeling methods
(often called groundwater mounding analysis). In addition, this analysis may be used to
determine the effects of groundwater lowering or mounding on surface water bodies (e.g.,
wetlands, steams, etc.).
(b) Groundwater contaminant transport calculations applied to investigating and/or
maintaining groundwater standards
It is the intention of Subchapter 02L and 02T regulations to maintain and protect the
groundwater quality of the state. With this goal, the purpose of the predictive
calculations or modeling is to document that the activity in question will not result in
contravention of groundwater standards at a specified receptor or location, or at the
assigned compliance boundary for a permitted facility. In the particular case regarding
treatment and disposal of soil containing petroleum products, the purpose of the
predictive calculations or modeling is to document that the disposal activity will not
result in the contravention of groundwater standards, in addition to other environmental
standards (e.g., surface water).
(c) Groundwater contaminant transport calculations applied to evaluating potential
impact to surface waters
Because one of the intentions of the Subchapter 02T regulation is to not allow any
violations of surface water standards, the Division may require that an evaluation be
made to determine the potential impact of the waste disposal activity or release onto the
surface waters. Predictive calculations or modeling methods may be required for the
following facilities that have or propose a non-discharge disposal activity and there is
DWQ GW Modeling Policy 070531.doc 3 OF 21
reasonable concern that surface waters may be adversely impacted by the subject non-
discharge waste disposal activity:
(i) any facility treating industrial waste,
(ii) any facility with a design flow of over 25,000 gpd, or
(iii) any facility utilizing a high-rate disposal system.
This evaluation would be conducted using a standard hydrogeologic investigation in
combination with predictive calculations or modeling to determine the potential impact to
surface waters. The evaluation would be mainly concerned with the potential impact of
waste nutrients (nitrogen and phosphorus) onto the surface water body, but also could
consider other surface water quality standards as described in 15A NCAC 02B .0200 at
the direction of the Division. The evaluation would predict the resultant impact in terms
of total pounds/day of contaminant to potentially discharge into the surface water body of
concern. The Division may require this “groundwater to surface water” potential nutrient
impact to be evaluated whenever surface waters or groundwater lowering ditches or
drains are located inside the facility’s compliance boundary, or otherwise reasonably
deemed to be “at risk” by the Division.
DWQ GW Modeling Policy 070531.doc 4 OF 21
Section I: Groundwater Modeling Process
The groundwater modeling process involves the following steps:
(1) define study objectives
(2) develop an initial conceptual model and data collection plan
(3) collect required data
(4) refinement of conceptual model
(5) select a computer model
(6) construct a groundwater model
(7) calibrate groundwater flow and transport model
(8) use models for predictive simulations
(9) conduct sensitivity analysis of calibrated models and predictive simulations
(10) perform mass balance calculations
(11) performance monitoring and model refinement
In general, the groundwater modeling process is a direct outgrowth of the hydrogeologic
investigation. Most of the data required by the groundwater modeling process should be
acquired in the hydrogeologic investigation and documented. The overall purpose of the
hydrogeologic investigation is to support a demonstration as to whether or not the groundwater
standards can be met. These predictions are accomplished via predictive calculations or
modeling.
(1) Define study objectives
In this critical first step, complete and detailed objectives of the modeling effort are specified.
These objectives will dictate the level of detail and accuracy required in model simulation.
These objectives should:
• Adequately address any regulatory requirements. These requirements will
typically be:
¾ ensuring that the groundwater standards will be maintained at the facility’s
compliance boundary or specific property location,
¾ ensuring that the contaminant plume will not adversely affect a known or
potential receptor,
¾ estimating the flow and loading to surface water discharge areas;
¾ estimating the zone of influence around an infiltration gallery to ensure a
closed loop infiltration and recovery groundwater system (for infiltration
galleries), or
¾ ensuring that the one-foot water table separation rule can be maintained.
• Identify constituents and processes to be modeled and acceptable model
assumptions to be made.
• Provide acceptable tolerances for model calibration.
The study objectives as defined above should be documented in writing. And a description
should be provided with regards to how the model/predictive calculations will address the study
objectives.
DWQ GW Modeling Policy 070531.doc 5 OF 21
(2) Develop an initial conceptual model and data collection plan
A conceptual model of groundwater flow is an interpretation or working description of the
characteristics and dynamics of the physical hydrogeologic system. This is also a critical step in
the modeling process, for if the investigator incorrectly conceptualizes the hydrogeologic
environment, then groundwater model results likewise will be incorrect and will produce invalid
predictions. The purpose of the conceptual model is to document regional and site-specific
hydrogeologic data into a set of assumptions and concepts that can be evaluated quantitatively
with the numeric or analytic models used for analysis and prediction.. Consult the Division’s
Hydrogeologic Investigation and Reporting Policy for further guidance on developing a
conceptual model and on performing a hydrogeologic investigation.
An initial conceptual model should be developed from available regional and local studies and
information, and initial site visits before significant site-specific data collection efforts are
undertaken. This step is necessary to assure that adequate types and quantities of data are
collected to adequately define the conceptual model and to constrain the numerical model or
calculation basis. The initial conceptual model generally undergoes refinement or modification
as a result of the data collection process, and may be further modified as a result of sensitivity
analyses with the quantitative model.
The conceptual model and quantitative models derived from it should adhere to the principle of
parsimony. That is, the simplest model that adequately describes the operation of the
hydrogeologic system for the expected analysis conditions is always preferred over more
complex models.
For the Mountain and Piedmont regions of North Carolina, the conceptual model of the
occurrence and movement of groundwater described by LeGrand is a good starting point. For
coastal plain applications, the model described by Giese, et al provides an initial model
framework.
The conceptual model should include a written description of the following:
(1) Topography and drainage
(2) Hydrostratigraphic units:
a. Lateral and vertical boundaries
b. Relationship to other units
c. Hydraulic and transport properties within each unit
(3) Boundary and Initial Conditions for Flow and Transport
(4) Time Domain to be used for analyses (steady or transient)
(5) Sources and sinks for water to enter or leave the modeled system
a. Recharge and evapotranspiration
b. Wells and springs
c. Connected surface water bodies
d. Topographic and manmade drains
(3) Collect required data
Data should be collected in accordance with the procedures outlined in the Division’s
Hydrogeologic Investigation and Reporting Policy document.
An important component of the data collection process is the documentation of data variability,
uncertainty, and deficiencies, and a compilation of the uncertainties recorded for each of the
other components.
DWQ GW Modeling Policy 070531.doc 6 OF 21
Groundwater models should not be used as a substitute for site-specific measurements of field
data. Rather, the site-specific measurements should be used to constrain the modeling by
providing data for model calibration, measurements of hydrostratigraphic unit geometries and
properties, as well as sources and sinks to be modeled.
(4) Refinement of conceptual model
Analysis of collected field data may support the initially developed conceptual model. Or, the
analysis can result in a refinement of the initially developed conceptual model.
At this stage the investigator should address the adequacy of the data collection effort. The
collection of additional appropriate site data may be required in order to further refine/confirm
the conceptual model.
(5) Select a computer model
A computer model is a set of one or more mathematical algorithms that simulate the
characteristics of a physical hydrogeologic system. The computer model selected should be
appropriate for the conceptual model developed. Modeling objectives should provide guidance
on the complexity of model required. In general, the simplest model should be used that
adequately matches the conceptual model. If the problem can be conceptualized in two
dimensions, then a three-dimensional model is unnecessary. When selecting an appropriate
groundwater model for a particular application, it is important to consider the amount and quality
of data available. Do not use a complex, multi-dimensional groundwater model if there is not
sufficient on-site data in addition to adequate knowledge of outer hydraulic boundaries, sources
and sinks. However, there may be situations in which a fully developed three-dimensional
numerical model is required, such as multi-aquifer groundwater flow/transport problems or
multi-layer models incorporating multiple soil horizons (where such complex models result from
complex conceptual models, and necessitate greater detailed hydrogeologic data collection).
When selecting a groundwater computer model, the user should consider the track record of the
model. The Division’s Aquifer Protection Section requires that any model used on a Division-
regulated project: (1) be thoroughly documented in readily accessible published format; (2) be
peer-reviewed in the scientific literature [includes appropriate government publications and
reviews published by or in cooperation with the International Ground Water Modeling Center];
and (3) be appropriate to the site under investigation, as determined by this policy.
(6) Construct a groundwater model
Model construction is the process of transforming the conceptual model into mathematical form.
For numerical models, this process usually involves translating the conceptual model into a
discretized flow domain, identifying discrete periods of time for analysis or annually-averaged
conditions, and compiling input parameters for the groundwater computer model, including
initial and boundary conditions and hydraulic properties. For semi-analytical models, the
process is similar, except no spatial discretizing is required. For analytical models, again no
spatial discretizing is required, but care must be taken to ensure that the pre-set boundary
conditions for a particular analytical solution adequately match the site in question, and that the
assumed groundwater flow field is adequate for the site.
(a) Flow sources and sinks
Sources and sinks influence groundwater flow patterns, and their effects should be
documented for inclusion in the selected model. Common sources and sinks that should
DWQ GW Modeling Policy 070531.doc 7 OF 21
be identified include: pumping or injection wells, precipitation and evapotranspiration
(or net groundwater recharge described below), drains, leakage across confining layers,
and flow to or from surface water bodies. Descriptions of sources and sinks should
include rates and temporal (seasonal and otherwise) variability. Development of a water
budget is usually helpful to quantify the contributions of sources and sinks.
Net groundwater recharge (or simply recharge) refers to the portion of precipitation that
infiltrates the soil and enters into the surficial groundwater aquifer, and is a key
parameter in all groundwater flow models. Recharge can be quantified in two general
ways: by either performing stream hydrograph baseflow separation on a regional
(basinwide or sub-basinwide) scale, or evaluating detailed site-specific soil
infiltration/evapotranspiration and surface runoff estimates and performing a site-specific
water balance to estimate recharge. Obtaining accurate estimates of recharge is difficult
without extensive regional and/or site-specific evaluation, and usually published recharge
estimates are used. With any groundwater flow model, there is always a direct
correlation between net groundwater recharge and the aquifer bulk transmissivity, which
is usually evident during the model calibration process (see Section I (7) Calibrate
groundwater flow and transport model below) and easily seen by running sensitivity
analyses. Therefore, selecting the appropriate net recharge is usually balanced with
selecting the appropriate aquifer bulk transmissivity.
At times it can be advantageous to model total precipitation (P) into the groundwater flow
model, and model evapotranspiration (Et) out of the model, with the net groundwater
recharge (R) being estimated as R = P – Et. A benefit of this methodology is that
seasonal changes in recharge (R) can be easily modeled using long-term averaged
precipitation (P) and standard evapotranspiration (Et) models. When modeling recharge
via the P – Et methodology, special care should be taken to check the model water
balance output to ensure that the model calculated recharge (R = P – Et) is within an
acceptable or reasonable range.
(b) Boundary conditions and extent of model
The physical size or extent of the model (length, width and depth of model) often has a
large bearing on the flow sources and sinks that need to be included into the model in
addition to the types of boundary conditions included in the model. In general, the
groundwater flow and transport model should have as many physical boundaries (such as
rivers, lakes, ocean) as possible in order to adequately simulate the regional groundwater
flow conditions at the particular site of interest. These types of physical boundaries can
generally be considered specified head, specified flux, or head-dependent flux
boundaries. Other good physical boundaries to model would be ridge lines or hilltops,
which can usually be considered no-flow boundaries for the surficial aquifer. However,
these types of features may not be no-flow boundaries for deeper confined aquifers.
In general, model boundaries should be located far enough away to minimize their direct
influence on the study area. To accomplish this and help lead the investigator toward
using real surrounding physical boundaries in the model, the physical size (i.e., each
horizontal dimension) of the model should be at least four (4) times larger than the largest
dimension of the facility’s land application system or other source of contamination.
This general rule can also be applied to the area impacted by small point source
contamination sources such as might be encountered in incident investigations. For
DWQ GW Modeling Policy 070531.doc 8 OF 21
example, if a proposed spray irrigation facility has a spray field that is 1000 feet by 500
feet, then in general the model should be at least 4000 feet in length and breadth, or may
need to be larger if appropriate in order bring in a physical boundary, such as a
neighboring river, into the model. Exceptions to this rule may be if a constant head
boundary (river, stream, etc.) or other boundaries are close to the site being modeled.
When in doubt as to how large to make the extent of the model, it should be made larger
in order to take into account neighboring physical boundaries.
Caution should be exercised when modeling groundwater-lowering ditches, which should
not be modeled as constant head boundaries. In general these features should be modeled
as a head-dependent drain boundary, where the drain elevation is the elevation of the
lowest topographical elevation in the ditch or drain pipe. In many situations, these drain
elevations provide important controls on the configuration of the water table and the
depth to water beneath land application units. Additionally, groundwater-lowering
ditches may necessitate an investigation into the potential impact to surface waters.
(c) Regional groundwater gradient
It is important that the predictive groundwater model accurately reflect the regional
groundwater gradient as measured in the field. Failure to do so will generally result in
incorrect groundwater mounding calculations and incorrect groundwater contaminant
transport calculations. It is important to realize that if the model boundary conditions are
correctly established, and if the sources and sinks (which include groundwater recharge
and leakage across confining layers) are modeled correctly, then the model-predicted
regional groundwater gradients will reflect field-measured groundwater gradients. If the
model-predicted groundwater gradients are too high or too low or in the wrong direction,
then this generally indicates that the model boundary conditions are incorrect and/or the
model sources and sinks are incorrect.
It is often the case that groundwater gradients vary seasonally, varying in magnitude and
direction. This again is generally a result of seasonal changes in physical boundary
conditions, such as changing river water level; and seasonal changes in sources and sinks,
such as changing groundwater recharge and evapotranspiration rates from winter to
summer. Whether or not the groundwater model needs to take into account these
seasonal groundwater gradient changes depends on the problem being solved and the
time-scale of the problem. For groundwater mounding problems, the worst-case scenario
will be in the winter and early spring when the seasonal groundwater table is at its
maximum elevation. For groundwater contaminant transport problems, generally the
time-scale of interest is measured in years because of generally slow-moving
groundwater. For time scales measured in years or decades of years, seasonal
fluctuations of groundwater gradient tend to average out, and modeling yearly averaged
gradients is appropriate.
(d) Hydraulic properties
Hydraulic properties include the transmissive and storage characteristics of the aquifer
system, such as transmissivity, hydraulic conductivity, storativity, and specific yield.
They also include the leakage coefficients of stream, lake and riverbeds. Field and
laboratory measurements of these properties should be documented, compared to
accepted ranges for the medium under investigation, and uncertainty associated with the
property measurements estimated. An assessment of heterogeneity and anisotropy over
DWQ GW Modeling Policy 070531.doc 9 OF 21
the aquifer domain for each property should be made, particularly in the Piedmont and
Mountain regions of the State. See the Division document entitled Performance and
Analysis of Aquifer Slug Tests and Pumping Tests Policy for details related to conducting
aquifer tests.
(e) Parameters used in transport models
Groundwater transport models require certain additional hydraulic and chemical
parameters, these being effective porosity, longitudinal and transverse dispersion
coefficients (or dispersivity), chemical retardation factor, and chemical biodegradation
decay rate. See Appendix A for details on how to estimate these parameters.
(f) Groundwater to surface water models
If the conceptual model involves a groundwater discharge to a surface water body, then
consideration needs to be given to how the surface water body will be modeled.
Generally, the surface water body can either be modeled as a constant head boundary
where the yearly-average surface water elevation is used as the constant head, or as a
head-dependent boundary where the surface water elevation is allowed to vary dependent
on groundwater baseflow and upstream conditions. The evaluation should predict the
resultant contaminant impact in terms of total pounds/day of contaminant to potentially
discharge into the surface water body of concern, once the groundwater contaminant
plume has reached steady-state conditions. See Section I (7) Calibrate groundwater flow
and transport model (below) for more details on transport analyses.
(g) Contaminant source concentration used in transport models
Groundwater transport models require a source concentration or mass flux to be
designated for the source of the contaminant plume to be modeled. In some situations,
uncertainty in the timing, magnitude, and mass of chemical sources may contribute the
largest uncertainty in predictions using transport models. In the case of a groundwater
remediation system being modeled, the measured groundwater contaminant concentration
in the source area actually measured in the field may be used. These sources should be
considered constant and continual unless it can be documented that virtually all of the
source mass has been removed.
In the case of a land application system (spray irrigation of treated wastewater, for
example), the correct source groundwater contaminant concentration to be used in a
transport model may be difficult to determine because of the uncertainty of chemical
removal/uptake in the cover crop and shallow soil horizons. There are two cases to
consider:
• If the cover crop and shallow soils are deemed to have no removal
capacity for a particular contaminant chemical of concern, then the treated
wastewater effluent chemical concentration should be used as the source
concentration of contaminant flux into the groundwater system.
• If the cover crop and shallow soils are deemed to have a certain removal
capacity for the particular contaminant, then the Division will allow 50%
removal of the Realistic Yield Expectation (R.Y.E., as documented by
NRCS, NCSU, etc.) to be used in calculating the resultant contaminant
DWQ GW Modeling Policy 070531.doc 10 OF 21
concentration assumed to leach into the groundwater system (see
Appendix A for a detailed discussion).
(h) Data deficiencies and uncertainty
A final component of the conceptual model is the documentation of data deficiencies, a
compilation of the uncertainties recorded for each of the other components, and an
acknowledgment of any alternative conceptual models that could be developed from the
available data. This last component of the conceptual model is an important step, for it
forces the investigator to quantitatively address the adequacy of the data collection effort.
If high uncertainty is associated with the conceptual model, then an elaborate and costly
modeling effort may not be justified.
(7) Calibrate groundwater flow and transport model
A reliable groundwater flow model must be able to simulate the observed movement of
groundwater and/or concentrations of contaminants. Typically, a groundwater flow model is
calibrated by comparing model output, such as a water level or head and discharge to surface
water, with actual measured values. When groundwater flow calibration is involved, the
modeling results should include (1) an evaluation of the calibration process, and (2) the resultant
calibrated groundwater/potentiometric surface(s) with posted head residuals at individual
observation wells. Residual statistics should be evaluated and reported.
Model calibrations are normally conducted with the flow model in “steady-state” mode, where
all the model parameters are fixed and do not vary with time. Typically, annual averaged
groundwater levels are used or approximated. However, in certain situations it may be necessary
for the investigator to also calibrate a transient (or dynamic) flow model. In this situation, the
model output for various time steps is compared to the observed values, such as water levels that
vary monthly, seasonally, or during the course of a pumping test.
In addition to model calibration using individual observation wells, the gradient (magnitude and
direction) of the model-predicted groundwater table/potentiometric surfaces(s) should reflect the
field-measured gradient across the modeled site. It is possible, for example, that part of the
modeled potentiometric surface appears accurate, but another part of the potentiometric surface
is obviously wrong, either in magnitude or direction. Using such a model to predict contaminant
transport may lead to serious errors.
It is important to realize that even though good groundwater flow calibration may be achieved,
this does not imply that the model is “correct” in its representation of the actual hydrogeologic
processes of the modeled site. Often times the groundwater flow model calibration process leads
to the investigator realizing more site/subsurface information is necessary to improve either (a)
the overall model calibration, and/or (b) the overall model water budget. Once the investigator is
satisfied with a particular model calibration, the overall model water budget should be checked
to ensure that a reasonable groundwater recharge value is being used for the particular site being
modeled. For example, a groundwater flow model being used for a particular site in the interior
coastal plain may appear to calibrate well, but if the resultant water budget shows that the net
groundwater recharge is 30 inches/year when a net groundwater recharge of about 10 – 12
inches/year is more generally accepted, then the overall model should be re-evaluated to
determine the source of error. In this particular case, it may likely be determined that the overall
aquifer transmissivity was set too high, which led the investigator to adjust the net groundwater
DWQ GW Modeling Policy 070531.doc 11 OF 21
recharge too high in order to maintain adequate calibration. It may also imply that the overall
hydrogeologic framework is incorrect or not complete.
Numerical groundwater contaminant transport models require that the groundwater flow field
first be evaluated. Therefore, a numerical transport model calibration is really a two-step
process. In Step #1 the groundwater flow model is calibrated, and then the flow field calculated
by the flow model is used in the contaminant transport model. Step #2 involves calibration of
the groundwater transport model to historic data on contaminant concentrations and degradation
rates. Groundwater transport model calibration will require a minimum of two discreet sampling
events from an appropriate time interval from the site. However, calibrating a groundwater
transport model using to too few sampling events, or between time intervals that are relatively
short, can lead to serious errors in predictive calculations.
(8) Use models for predictive simulations
The main purpose of a modeling effort is to generate a representative groundwater flow and/or
transport model that will make accurate predictions based on an altered environment. Predictive
simulations may either be run when using a model in “steady-state” mode or in “transient” mode.
In steady-state mode, all the model parameters are fixed and do not vary with time, whereas in
transient mode certain parameters such as rainfall, evapotranspiration, pumping rates, etc., are
varied seasonally (typically) to generate a seasonal groundwater table variation. Predictive
simulations will generally be of two forms: (a) groundwater flow and mounding simulations,
and/or (b) groundwater transport simulations.
(a) Groundwater flow and mounding predictive simulations
Typically, predictive groundwater flow models are run in steady-state mode, when
dynamic equilibrium is achieved. Transient groundwater flow models are run when
multiple time periods are simulated. If the flow model is being run to predict a
groundwater mound height generated by some type of land application system (such as
an infiltration gallery or spray irrigation system), then the model is typically run for 200 -
360 days, or whenever the groundwater mound height appears to stabilize (or 720 days,
whichever comes first). However, the simulation period should be a year or less, as the
seasonal groundwater table/mound fluctuation is typically cyclical.
(b) Groundwater transport simulations
Groundwater transport models are typically run with the flow model in steady-state mode
using average annual conditions. Because the time span of groundwater contaminant
travel is usually measured in years, over the span of multiple years the seasonal
groundwater flow variations are generally averaged out, and thus performing transport
models with a transient groundwater flow model is generally not required.
A transport model should be run until the contaminant plume has reached steady-state (or
near steady-state) conditions. Assuming the source of the contaminant flux remains
constant (or near constant), at some point in time the shape of the plume will reach a
maximum size and the shape of the plume will remain relatively fixed for future times.
For larger discharging land application systems, steady-state conditions may not be
reached for decades, especially if deeper semi-confined aquifers are involved in the
groundwater flow and transport process.
DWQ GW Modeling Policy 070531.doc 12 OF 21
(9) Conduct sensitivity analysis of calibrated models and predictive simulations
Sensitivity analysis involves varying values for a property, such as hydraulic conductivity, and
observing the effect on model calculated head or concentration values. Usually there is a certain
amount of uncertainty with regard to the actual aquifer hydraulic conductivity or transmissivity
values to be used in the model. Thus, sensitivity analysis is particularly helpful in quantifying
the uncertainty associated with model-predicted future or altered site conditions. By varying the
hydraulic conductivity or transmissivity, or other potentially sensitive parameters, over the range
of potentially expected values, the range of resultant groundwater elevations or concentrations
will be generated. The investigator can then determine expected head or concentration results
with a range of uncertainty associated with it.
The sensitivity analysis should identify a range of values for each sensitive input parameter.
Data collected from on-site testing will help constrain the range of values for sensitive
parameters. On-site data should be used whenever possible in the model domain. For poorly
constrained parameters, use the most conservative input value(s) that can reasonably be expected
to occur for the particular model application. For example, if the groundwater model is used to
predict mounding conditions in response to irrigation, and the sensitivity analysis indicates that
the model is sensitive to changes in transmissivity values, the transmissivity value(s) used in the
model must be those which, within a reasonable range for the given hydrogeologic conditions,
would result in the highest mounding of water levels. If a groundwater contaminant transport
model is being used to predict the maximum distance that a contaminant may be expected to
travel, sensitive input values used must be those which, within a reasonable range for given
hydrogeologic conditions and chemical properties, would result in the furthest distance traveled
for the modeled constituent.
The results of the sensitivity analyses should include a table showing the sensitive parameters
and their ranges, and figures showing the resulting variations in modeled parameters using the
two value endpoints (highest and lowest value) for each sensitive parameter. Additional field
characterization may be required to obtain data for model input parameters that are determined to
be relatively sensitive.
(10) Perform mass balance calculations
Water and mass balance model outputs (or calculations) should be shown describing all flow and
transport fluxes. All source and sink terms should be shown and the net results should balance
within reasonable margins for error (less than 0.5%).
For model calibration, if net groundwater recharge (R) is being modeled as total precipitation (P)
into the model minus evapotranspiration (Et) out of the model (R = P - Et), then the water
balance output should show the total P into the model, and the total Et out of the model, and an
evaluation of the net groundwater recharge (R) into the model should be made. For groundwater
mounding model simulations, the water balance output should show the total additional flux of
water added into the model above the net groundwater recharge (R).
For groundwater transport models where a particular contaminant is reaching a receptor such as a
stream or a well, the mass balance should show the total mass of contaminant (as a function of
time) reaching the receptor, the total mass removed from the model domain at the receptor, and
the total contaminant mass introduced into the model domain at the source.
DWQ GW Modeling Policy 070531.doc 13 OF 21
(11) Performance Monitoring and Model Refinement
Groundwater models can be useful tools to simulate hydrogeologic conditions and contaminant
concentrations over time. Models are most useful when used as working “tools” that are refined
and improved when more information on site hydrogeologic conditions becomes available. As
more site data becomes available, the groundwater model should be checked against this data
and the model may need to be refined in order to more accurately predict future conditions.
Additional wells or monitoring points may be required during the performance-monitoring
period if the performance monitoring data indicates an inadequate monitoring network.
When required, a Performance Monitoring Report should be submitted to the Division’s Aquifer
Protection Section on an annual basis, or at a time interval agreed upon by the Section and the
Responsible Party, which will contain the predicted model outputs compared with data obtained
during the performance monitoring period. If there is a significant discrepancy between the
predicted model output and the performance monitoring data, the groundwater model should be
refined in order to more closely match actual field conditions.
DWQ GW Modeling Policy 070531.doc 14 OF 21
Section II: Reporting Modeling Results
Results from groundwater modeling efforts must be adequately documented. Such
documentation must provide regulators sufficient information to determine the adequacy of the
model and supporting data, and validity of the modeling results. The major reporting elements
shown below must be included in the model report submission. This format is a modified format
taken from ASTM D 5447. The value of this format is that it standardizes criteria that should be
considered in any modeling effort. The detail provided within the format should reflect the
investment that has gone into the modeling effort.
The Division may request that groundwater computer model data inputs and outputs be provided
in electronic form in order to allow staff to evaluate the model using the actual model.

1.0 Introduction
1.1 General Setting
1.2 Study Objectives
2.0 Conceptual Model
2.1 Aquifer System Framework
2.2 Groundwater Flow System
2.3 Hydrologic Boundaries
2.4 Hydraulic Boundaries
2.5 Sources and Sinks
2.6 Water Budget
3.0 Computer Model
3.1 Model Selection
3.2 Model Description
4.0 Groundwater Flow/Transport Model
Construction
4.1 Model Grid
4.2 Hydraulic Parameters
4.3 Boundary Conditions
4.4 Selection of Calibration Targets
5.0 Calibration
5.1 Residual Analysis
5.2 Sensitivity Analysis
6.0 Predictive Simulations
6.1 Flow simulations
6.2 Transport simulations
7.0 Summary and Conclusions
7.1 Model Assumptions/Limitations
7.2 Model Predictions
7.3 Performance Monitoring and Model
Refinement
7.4 Recommendations
8.0 References
9.0 Appendices
9.1 Model Input Files
9.2 Model Output Files

DWQ GW Modeling Policy 070531.doc 15 OF 21
References
ASTM Guide D 5447-93. Standard Guide for Application of a Ground-Water Flow Model to a
Site-Specific Problem.
ASTM Guide D 5609-94. Standard Guide for Defining Boundary Conditions in Ground-Water
Flow Modeling.
ASTM Guide D 5610-94. Standard Guide for Defining Initial Conditions in Ground-Water
Flow Modeling.
ASTM Guide D 5611-94. Standard Guide for Conducting a Sensitivity Analysis for a Ground-
Water Flow Model Application.
ASTM Guide D 5490-93. Standard Guide for Comparing Ground-Water Flow Model
Simulations to Site-Specific Information.
ASTM Guide E 978-92. Standard Guide for Evaluating Mathematical Models for the
Environmental Fate of Chemicals.
Daniel, C.C., III, Smith, D.G., and Eimers, J.L., Hydrogeology and Simulation of Ground-Water
Flow in the Thick Regolith-Fractured Crystalline Rock Aquifer System of Indian Creek Basin,
North Carolina, in Ground-water resources of the Piedmont-Blue Ridge Provinces of North
Carolina: U.S. Geological Survey Water-Supply Paper 2341–C, 1997.
Driscoll, Fletcher G., Groundwater and Wells, 2nd Edition, Johnson Division, St. Paul, MN,
1986.
Fetter, C. W., Applied Hydrogeology, 3rd Edition, Charles E. Merrill Publishing Co., Columbus,
Ohio, 1995.
Fetter, C.W., Contaminant Hydrogeology, Macmillan Publishing Co., New York, 1993.
Freeze, R. Allen and Cherry, John A., Groundwater, Prentice-Hall, Inc., Englewood Cliffs, New
Jersey, 1979.
Gelher, L.W., Mantoglou, A., Welty, C. and Rehfeldt, K. R., A Review of Field-Scale Physical
Solute Transport Processes in Saturated and Unsaturated Porous Media, Electric Power
Research Institute (EPRI) EA-4190, Palo Alto, CA., 1985.
Harned, D.A., and Daniel, C.C., III, The Transition Zone between Bedrock and Regolith:
Conduit for Contamination?, in Daniel, C.C., III, White, R.K., and Stone, P. A., eds., Ground
water in the Piedmont, Proceedings of a Conference on Ground Water in the Piedmont of the
Eastern United States, Charlotte, N.C., Oct. 16–18, 1989: Clemson, S.C., Clemson University, p.
336–348, 1992.
Heath, Ralph, C., Basic Ground-Water Hydrology, U.S. Geological Survey Water-Supply Paper
2220, U.S. Government Printing Office, 1991.
Javandel, I., Doughty, C., and Tsang, C. F., Groundwater Transport: Handbook of Mathematical
Models, American Geophysical Union Water Resources Monograph 10, Washington D.C., 1984.
Kruseman, G. P. and de Ridder, N. A., Analysis and Evaluation of Pumping Test Data, 2nd
Edition, International Institute for Land Reclamation and Improvement, Wageningen, The
Netherlands, 1990.
DWQ GW Modeling Policy 070531.doc 16 OF 21
LeGrand, H.,E. Sr., 2004, A Master Conceptual Model for Hydrogeologic Site Characterization
in the Piedmont and Mountain Region of North Carolina, NCDENR, Division of Water Quality,
Groundwater Section.
Maag, M., Malinovsky, M., and Nielsen, S. M., Kinetics and Temperature Dependence of
Potential Denitrification in Riparian Soils, J. of Environmental Quality, Vol. 26, p. 215 - 223,
1997.
Natural Resources Conservation Service (NRCS), Conservation Practice Standard, Nutrient
Management, Code 590, June 1998.
Nelson, W. M., Gold, A. J., and Groffman, P. M., Spatial and Temporal Variations in
Groundwater Nitrate Removal in a Riparian Forest, J. of Environmental Quality, Vol. 24, p. 691
- 699, 1995.
North Carolina Cooperative Extension Service, Soil Facts: Nitrogen Management and Water
Quality, Publication AG-439-2, August, 1990.
NC State University (NCSU), Department of Soil Science, Realistic Yield Expectations for Soils
of North Carolina, Final Review Draft, April 2000.

DWQ GW Modeling Policy 070531.doc 17 OF 21
Appendix A
Dispersion, Chemical Retardation Factor, and Chemical Biodegradation Decay Rate
Parameters used in Transport Models
Groundwater transport models typically require certain additional hydraulic properties and
chemical properties, these being (a) longitudinal and transverse dispersion coefficients (or
dispersivity), (b) chemical retardation factor, and (c) chemical biodegradation decay rate.
(a) Dispersion measures the natural spreading of a contaminant plume during migration. Site
specific dispersion parameter values are difficult to measure without extensive field
investigations, but fortunately researchers have developed methods of estimating them using
simple formulas. The dispersion coefficient depends not only on the variability of the local
hydraulic conductivity at the site, but also on the scale of the problem (i.e., the distance from the
original plume site to the groundwater receptor or endpoint of travel). See Fetter (1993) or
Gelhar (1985) for more discussions regarding these issues.
Longitudinal dispersion is a measure of the contaminant plume spreading in the direction
(parallel) of groundwater flow. Transverse dispersion is a measure of the contaminant plume
spreading perpendicular to the direction of groundwater flow. The longitudinal dispersion
coefficient can be estimated as follows:
DL = 0.1 L v
where: DL = longitudinal dispersion coefficient in feet2/day
L = distance in feet from the original plume site to the groundwater
receptor of interest or endpoint of travel
v = average groundwater velocity in feet/day
The transverse dispersion coefficient (DT) can be estimated from the relation DL/DT = 6 to 20,
depending on site conditions (Fetter, 1993), but a ratio of DL/DT of 10 is probably good for
typical cases.
Often groundwater transport models will use the longitudinal and transverse dispersivity
parameter instead of the longitudinal and transverse dispersion coefficient. The relationship
between the two parameters is as follows:
DL = αL v + D*
DT = αT v + D*
where αL is longitudinal dispersivity in feet, αT is transverse dispersivity in feet, v is average
groundwater velocity in feet/day, and D* is effective diffusion coefficient, which is related to the
diffusion due to concentration gradients. For typical groundwater transport problems, αL v and
αT v are numerically much larger than D*, and thus D* can often be ignored (See Fetter, 1993.
However, in some transport problems v can be numerically very low, such as leakage through a
liner problem, and D* term will dominate, implying that the main dispersion mechanism is
diffusion via concentration gradients.) In this case:
DL / DT = αL / αT = 10 (typically)
(b) Chemical retardation factor (unitless number) is the measure of the relative migration
velocity of the chemical (contamination) compared to water. For inorganic constituents (such as
cations, anions, including NO3, Cl) and fecal coliform, the retardation factor is normally set to 1.
For organic chemicals, the retardation factor (R) should be based on the following formula:
DWQ GW Modeling Policy 070531.doc 18 OF 21
R = 1 + ρ Koc foc / n
where: ρ = aquifer bulk density in g/cm3, default = 1.8 g/cm3
Koc = organic carbon-water partition coefficient in L/kg
foc = aquifer organic carbon fraction (unitless), default = 0.001
n = aquifer effective porosity (unitless)
The organic carbon-water partition coefficient (Koc) is readily available for most organic
chemicals from various sources. Values other than the default values for aquifer bulk density (ρ)
and organic carbon fraction (foc) should be documented.
(c) The chemical biodegradation decay rate measures the rate at which a contaminant is
attenuated due to biological activity in the subsurface. Setting the decay rate to zero implies no
biodegradation. Many groundwater transport models that allow biodegradation assume a first-
order decay rate. Typically, these models will either require the decay rate in units of either
1/days or 1/years. Some models, however, may require the decay rate to be entered in terms of a
half-life (or lifetime). The relationship between half-life (τ) and first-order decay (k) rate is:
τ = 0.693 / k
where if τ is in days, then k is in 1/days, or if τ is in years, then k is in 1/years, etc. If a non-
zero biodegradation rate is used in a transport model, evidence needs to be presented to justify its
use.
With regards to Nitrate (NO3) transport and decay in groundwater, there is evidence that Nitrate
may decay (NO3 denitrifying in a riparian buffer zone, for example) via a zero-order decay rate
(see Nelson et. al., 1995), or via Michaelis-Menton kinetics that leads to a first-order decay for
smaller concentrations and a shifting to a zero-order decay for larger concentrations (see Maag
et. al, 1997). In this special case of NO3 denitrifying (and other contaminants that may be
similar), special care should be taken when modeling NO3 removal with solely a first-order
decay model, as over-prediction of the NO3 removal rate could potentially occur.
DWQ GW Modeling Policy 070531.doc 19 OF 21
Appendix B
Example Calculation of Nitrate Source Concentration to be used in a Groundwater
Transport Model for Land Application Systems Utilizing a Cover Crop
Groundwater transport models require a source concentration or mass flux to be designated for
the source of the contaminant plume to be modeled.
If the cover crop and shallow soils are deemed to have a certain removal capacity for the
particular contaminant, then the Division will allow 50% removal of the Realistic Yield
Expectation (R.Y.E., as documented by NRCS, NCSU (see
http://www.soil.ncsu.edu/nmp/ncnmwg/yields/), site specific yield records, etc.) to be used in
calculating the resultant contaminant concentration assumed to leach into the groundwater
system (see North Carolina Cooperative Extension Service, 1990).
Note: This 50% R.Y.E. limit with regards to the cover crop is only for the purposes of
calculating a potential “conservative” resultant contaminant concentration assumed to
leach into the groundwater system, and do not imply that the cover crop will not remove
the full R.Y.E. However, studies have shown that certain chemicals of interest (nitrogen,
for example) typically do not accumulate in the soil, and are readily leached downward
through the cover crop root zone into the surficial groundwater aquifer, especially when
the crop is not in its growing season. Therefore, this 50% rule is meant to be
conservative in order to guard against potential contaminant impact to groundwaters
and surface waters of the State.
This calculation should be done according to the following example.
A certain municipal wastewater treatment plant uses spray irrigation to land apply its
treated wastewater. The WWTP has a design flow of 50,000 GPD (0.05 MGD), and the
investigator is concerned about meeting the NO3 (nitrate) groundwater standard of 10
mg/l N at the compliance boundary. The WWTP sprays onto a 15 acre dedicated field
where the cover crop is fescue grass on Goldsboro soils. The WWTP achieves the
following average effluent limits with regard to nitrogen species:
[Ammonia-N] = 8 mg/l
[NO3-N + NO2-N] = 10 mg/l
[TKN] = 15 mg/l
Total nitrogen in the wastewater effluent is thus [TKN] + [NO3-N + NO2-N] = 25 mg/l
(ppm) N. The total pounds/year of N applied to the spray fields is:
Total pounds N/year applied = (25 ppm N) x (0.05 MGD) x (8.34 lbs/gallon) x (365 days/year)
= 3,805 lbs N/year
Calculate the Cover crop R.Y.E. Uptake:
According to the NC State University Realistic Yield Expectations (R.Y.E.) for Soils in
North Carolina (see NCSU, 2000), fescue planted on Goldsboro soil series will yield 4.0
dry tons of hay/acre/year.
According to the NRCS Conservation Practice , Standard Nutrient Management Code
590 document (see NRCS, 1998), Nitrogen Fertilization Rate for fescue is 40 - 50 lbs
N/ton hay (use 50 lbs N/ton hay).
DWQ GW Modeling Policy 070531.doc 20 OF 21
Combining the above two figures, 15 acres of fescue will consume:
R.Y.E. = (4.0 dry tons/acre/year) x (50 lbs N/ton hay) x (15 acres)
= (200 lbs N/acre/year) x (15 acres)
= 3,000 lbs N/year
The Division will allow 50% uptake of the R.Y.E. for the purposes of calculating
contaminant concentrations leaching to the underlying groundwater system:
50% of R.Y.E = 0.5 x (3,000 lbs N/year) = 1,500 lbs N/year
Resultant pounds N/year
assumed to leach into
groundwater system = 3,805 lbs N/year - 50% of R.Y.E.
= 3,805 lbs N/year - 1,500 lbs N/year
= 2,305 lbs N/year
Resultant chemical conc.
of flux leaching into
the groundwater system = 2,305 lbs/year / (0.05 MGD x 8.34 x 365)
= 15.1 mg/l N
which is assumed to all convert (oxidize) to NO3-N by the time the contaminant is in the
groundwater system.
Thus, for the purposes of building the groundwater flow and transport model, the
investigator would apply 50,000 GPD onto the 15 acres at a concentration of 15.1 mg/l
NO3-N and assume that all the contaminant flux recharges into the groundwater system.
From the above analysis, it is clear that if the total nutrients in the effluent is less than or equal to
50% of R.Y.E., then all the effluent nutrient is assumed to be taken up by the cover crop, and
there is no need to perform any groundwater contaminant transport analysis for the nutrients
involved.
DWQ GW Modeling Policy 070531.doc 21 OF 21