Groundwater Modeling Guidance

Transcription

1 Groundwater Modeling Guidance Richard J. Mandle Groundwater Modeling Program Michigan Department of Environmental Quality Draft /16/02

2 Executive Summary The use of groundwater models is prevalent in the field of environmental hydrogeology. Groundwaterflow and fate and transport models have been applied to investigate a wide variety of hydrogeologic conditions. Groundwater flow models are used to calculate the rate and direction of movement of groundwater through aquifers and confining units in the subsurface. Fate and transport models estimate the concentration of a chemical in groundwater beginning at its point of introduction to the environment to locations downgradient of the source. Fate and transport models require the development of a calibrated groundwater flow model or, at a minimum, an accurate determination of the velocity and direction of groundwater flow that has been based on field data. Models are conceptual descriptions or approximations that describe physical systems using mathematical equations they are not exact descriptions of physical systems or processes. The applicability or usefulness of a model depends on how closely the mathematical equations approximate the physical system being modeled. In order to evaluate the applicability or usefulness of a model, it is necessary to have a thorough understanding of the physical system and the assumptions embedded in the derivation of the mathematical equations. These equations are based on certain simplifying assumptions which typically involve: the direction of flow, geometry of the aquifer, the heterogeneity or anisotropy of sediments or bedrock within the aquifer, the contaminant transport mechanisms, and chemical reactions. It is because of these assumptions, and the many uncertainties in the values of data required by the model, that a model must be viewed as an approximation and not an exact duplication of field conditions. The equations that describe the groundwater flow and fate and transport processes may be solved using different types of models. Some models (e.g. analytical models), may be exact solutions to equations which describe very simple flow or transport conditions and others (such as numerical models) may be approximations of equations which describe very complex conditions. Because of the simplifications inherent with analytical models, it is not possible to account for field conditions that change with time or space. This includes variations in groundwater flow rate or direction, variations in hydraulic or chemical reaction properties, changing hydraulic stresses, or complex hydrogeologic or chemical boundary conditions. Analytical models are best used for: Initial assessments where a high degree of accuracy is not needed, Prior to beginning field activities to aid in designing data collection, To check results of numerical model simulations, or Where field conditions support the simplifying assumptions embedded in the analytical models.

3 Numerical models are capable of solving the more complex equations that describe groundwater flow and solute transport. These equations generally describe multi-dimensional groundwater flow, solute transport, and chemical reactions. Each model, whether it is a simple analytical model or a complex numerical model, has applicability and usefulness in hydrogeological and remedial investigations, in spite of the simplifications inherent in the model equations. However, the selection and proper use of a model must be based on a thorough understanding of the importance of relevant flow or solute transport processes at a site, this requires proper site characterization. Proper site characterization involves the collection of site-specific data that accurately describe the movement of groundwater and disposition of solutes at the site. Without proper site characterization, it is not possible to determine whether the model equations are appropriate or to develop a reliably calibrated model. Groundwater and fate and transport models are used to predict the migration pathway and concentrations of contaminants in groundwater. The accuracy of model predictions depends upon the degree of successful calibration and verification of the model simulations. Errors in the model used for predictive simulations, even though small, can result in gross errors in solutions projected forward in time. Monitoring of hydraulic heads and groundwater chemistry (performance monitoring) will be required to assess the accuracy of predictive simulations. The collection of site-specific data during the remedial action is referred to as "performance monitoring." Performance monitoring is required as a means of physically measuring the actual behavior of the hydrogeologic system and demonstrating compliance with environmental statutes. The model simulations used to predict hydraulic containment or contaminant removal are estimates which cannot be substituted in place of measured field data. The purpose of this document is to assist DEQ staff obtain a better understanding of the model development and evaluation process. It is not the purpose of this document to fully educate DEQ staff so that technical assistance is not required. It is necessary that model results be given a thorough technical review by qualified DEQ staff. This process of verification and review of groundwater flow and solute transport models is performed by the Groundwater Modeling Program (GMP), in the Superfund Section of DEQ Remediation and Redevelopment Division. DEQ project managers requiring model review assistance should contact the GMP manager to obtain assistance from the GMP. 10/16/02 2

4 1.0 Introduction The use of groundwater flow models 1 is prevalent in the field of environmental hydrogeology. Models have been applied to investigate a wide variety of hydrogeologic conditions. More recently, groundwater models have been applied to predict the fate and transport of contaminants for risk evaluation purposes. This document is intended to assist staff in the evaluation of work plans that propose to use groundwater models, and to assist staff in the evaluation of models that have been developed for remedial design, feasibility studies, development of performance monitoring networks, or for risk assessment. The following groundwater modeling guidance document was developed by the Groundwater Modeling Program (GMP) as part of an interdivisional committee of the Department of Environmental Quality (DEQ). The purpose of this Guidance Document is to provide DEQ staff with a basic understanding of the application and development of groundwater flow and solute transport models. The scope of this guidance document is to describe, in general terms: Application of groundwater flow and transport models for saturated flow conditions, The required level of hydrogeological characterization needed to develop a model for a site, Different types of models, Groundwater modeling procedures, The appropriate level of model documentation, The model review submittal procedures, and The need for verification sampling. It is not the intent of this document to provide DEQ staff with a detailed discussion of groundwater modeling concepts or procedures, or of particular groundwater models. A list of selected references, which provide a more thorough discussion of the concepts presented in this document, is presented in Appendix 1. A number of technical terms are used throughout this document when describing various aspects of groundwater modeling. A glossary of these and other commonly used modeling terms and their definitions are contained in Appendix 2. Throughout the text of this guidance document, the first occurrences of technical terms that are contained in the glossary appear in bold type. The reader is referred to both of these appendices, either to locate a source for more information concerning groundwater modeling or for definitions of groundwater modeling terms. 1 All bold terms are defined in the Glossary found in Appendix 2. 10/16/02 3

5 2.0 Groundwater Models In general, models are conceptual descriptions or approximations that describe physical systems using mathematical equations they are not exact descriptions of physical systems or processes. The applicability or usefulness of a model depends on how closely the mathematical equations approximate the physical system being modeled. In order to evaluate the applicability or usefulness of a model, it is necessary to have a thorough understanding of the physical system and of the assumptions embedded in the derivation of the mathematical equations. A detailed discussion of the assumptions and derivations of the equations that are the basis of different groundwater models is beyond the scope of this document. The reader is referred to one of the references included in Appendix 1 for this information. Groundwater models describe groundwater flow and fate and transport processes using mathematical equations that are based on certain simplifying assumptions. These assumptions typically involve the direction of flow, geometry of the aquifer, the heterogeneity or anisotropy of sediments or bedrock within the aquifer, the contaminant transport mechanisms and chemical reactions. Because of the simplifying assumptions embedded in the mathematical equations and the many uncertainties in the values of data required by the model, a model must be viewed as an approximation and not an exact duplication of field conditions. Groundwater models, however, even as approximations are a useful investigation tool that groundwater hydrologists may use for a number of applications. Among these are: Prediction of the possible fate and migration of contaminants for risk evaluation. Tracking the possible migration pathway of groundwater contamination. Evaluation of design of hydraulic containment and pump-and-treat systems. Design of groundwater monitoring networks. Wellhead protection area delineation. Evaluation of regional groundwater resources. Prediction of the effect of future groundwater withdrawals on groundwater levels. It is important to understand general aspects of both groundwater flow and fate and transport models to ensure that application or evaluation of these models may be performed correctly. 10/16/02 4

6 2.1 General Concepts Groundwater Flow Models Groundwater flow models are used to calculate the rate and direction of movement of groundwater through aquifers and confining units in the subsurface. These calculations are referred to as simulations. The simulation of groundwater flow requires a thorough understanding of the hydrogeologic characteristics of the site. The hydrogeologic investigation should include a complete characterization of the following: Subsurface extent and thickness of aquifers and confining units (hydrogeologic framework). Hydrologic boundaries (also referred to as boundary conditions), which control the rate and direction of movement of groundwater. Hydraulic properties of the aquifers and confining units. A description of the horizontal and vertical distribution of hydraulic head throughout the modeled area for beginning (initial conditions), equilibrium (steady-state conditions) and transitional conditions when hydraulic head may vary with time (transient conditions). Distribution and magnitude of groundwater recharge, pumping or injection of groundwater, leakage to or from surface-water bodies, etc. (sources or sinks, also referred to as stresses). These stresses may be constant (unvarying with time) or may change with time (transient). The outputs from the model simulations are the hydraulic heads and groundwater flow rates which are in equilibrium with the hydrogeologic conditions (hydrogeologic framework, hydrologic boundaries, initial and transient conditions, hydraulic properties, and sources or sinks) defined for the modeled area. Figure 1 shows the simulated flow field for a hypothetical site at which pumping from a well creates changes in the groundwater flow field. 10/16/02 5

7 Figure 1. Simulated groundwater flow vectors and hydraulic head. Through the process of model calibration and verification, which is discussed in later sections of this document, the values of the different hydrogeologic conditions are varied to reduce any disparity between the model simulations and field data, and to improve the accuracy of the model. The model can also be used to simulate possible future changes to hydraulic head or groundwater flow rates as a result of future changes in stresses on the aquifer system (Predictive simulations are discussed in later sections of this document) Fate and Transport Models Fate and transport models simulate the movement and chemical alteration of contaminants as they move with groundwater through the subsurface. Fate and transport models require the development of a calibrated groundwater flow model or, at a minimum, an accurate determination of the velocity and direction of groundwater flow, which has been based on field data. The model simulates the following processes: Movement of contaminants by advection and diffusion, Spread and dilution of contaminants by dispersion, Removal or release of contaminants by sorption or desorption of contaminants onto or from subsurface sediment or rock, or Chemical alteration of the contaminant by chemical reactions which may be controlled by biological processes or physical chemical reactions. In addition to a thorough hydrogeological investigation, the simulation of fate and transport processes requires a complete characterization of the following: Horizontal and vertical distribution of average linear groundwater velocity (direction and magnitude) determined by a calibrated groundwater flow model or through accurate determination of direction and rate of groundwater flow from field data. Boundary conditions for the solute. Initial distribution of solute (initial conditions). Location, history and mass loading rate of chemical sources or sinks. Effective porosity. 10/16/02 6

8 Soil bulk density. Fraction of organic carbon in soils. Octanol-water partition coefficient for chemical of concern. Density of fluid. Viscosity of fluid. Longitudinal and transverse dispersivity. Diffusion coefficient. Chemical decay rate or degradation constant. Equations describing chemical transformation processes, if applicable. Initial distribution of electron acceptors, if applicable. The outputs from the model simulations are the contaminant concentrations, which are in equilibrium with the groundwater flow system, and the geochemical conditions (described above) defined for the modeled area. Figure 2 shows the simulated migration of a contaminant at a hypothetical site. Figure 2. Simulated contaminant plume migration. As with groundwater flow models, fate and transport models should be calibrated and verified by adjusting values of the different hydrogeologic or geochemical conditions to reduce any disparity between the model simulations and field data. This process may result in a re-evaluation of the model used for simulating groundwater flow if the adjustment of values of geochemical data does not result in an acceptable model simulation. Predictive simulations may be made with a fate and transport model to predict the expected concentrations of contaminants in groundwater as a result of implementation of a remedial action. The simulated containment of a contaminant plume is shown in Figure 2. Monitoring of hydraulic heads and groundwater chemistry will be required to support predictive simulations. 10/16/02 7

9 2.2 Types of Models The equations that describe the groundwater flow and fate and transport processes may be solved using different types of models. Some models may be exact solutions to equations that describe very simple flow or transport conditions (analytical model) and others may be approximations of equations that describe very complex conditions (numerical model). Each model may also simulate one or more of the processes that govern groundwater flow or contaminant migration rather than all of the flow and transport processes. As an example, particle-tracking models, such as MODPATH, simulate the advective transport of contaminants but do not account for other fate and transport processes. In selecting a model for use at a site, it is necessary to determine whether the model equations account for the key processes occurring at the site. Each model, whether it is a simple analytical model or a complex numerical model, may have applicability and usefulness in hydrogeological and remedial investigations Analytical Models Analytical models are an exact solution of a specific, often greatly simplified, groundwater flow or transport equation. The equation is a simplification of more complex three-dimensional groundwater flow or solute transport equations. Prior to the development and widespread use of computers, there was a need to simplify the three-dimensional equations because it was not possible to easily solve these equations. Specifically, these simplifications resulted in reducing the groundwater flow to one dimension and the solute transport equation to one or two dimensions. This resulted in changes to the model equations that include one-dimensional uniform groundwater flow, simple uniform aquifer geometry, homogeneous and isotropic aquifers, uniform hydraulic and chemical reaction properties, and simple flow or chemical reaction boundaries. Analytical models are typically steady-state and one-dimensional, although selected groundwater flow models are two dimensional (e.g. analytical element models), and some contaminant transport models assume one-dimensional groundwater flow conditions and one-, two- or threedimensional transport conditions. An example of output from a one-dimensional fate and transport analytical model (Domenico and Robbins, 1985) is shown in Figure 3. 10/16/02 8

10 Figure 3. Results from one-dimensional fate and transport model. Because of the simplifications inherent with analytical models, it is not possible to account for field conditions that change with time or space. This includes variations in groundwater flow rate or direction, variations in hydraulic or chemical reaction properties, changing hydraulic stresses, or complex hydrogeologic or chemical boundary conditions. Analytical models are best used for: Initial site assessments where a high degree of accuracy is not needed, Designing data collection plans prior to beginning field activities, An independent check of numerical model simulation results, or Sites where field conditions support the simplifying assumptions embedded in the analytical models. 10/16/02 9

11 2.2.2 Numerical Models Numerical models are capable of solving the more complex equations that describe groundwater flow and solute transport. These equations generally describe multi-dimensional groundwater flow, solute transport and chemical reactions, although there are one-dimensional numerical models. Numerical models use approximations (e.g. finite differences, or finite elements) to solve the differential equations describing groundwater flow or solute transport. The approximations require that the model domain and time be discretized. In this discretization process, the model domain is represented by a network of grid cells or elements, and the time of the simulation is represented by time steps. A simple example of discretization is presented in the following figure (Figure 4). The curve represents the continuous variation of a parameter across the model space or time domain. The bars represent a discrete step-wise approximation of the curve. The accuracy of numerical models depends upon the accuracy of the model input data, the size of the space and time discretization (the greater the size of the discretization steps, the greater the possible error), and the numerical method used to solve the model equations. Figure 4. Example of discretization process. 10/16/02 10

12 Unlike analytical models, numerical models have the capability of representing a complex multi-layered hydrogeologic framework. This is accomplished by dividing the framework into discrete cells or elements. An example of representing a multi-layered aquifer system in a numerical model is shown in Figure 5. Figure 5. Example of discretization of complex hydrogeological conditions by a numerical model. In addition to complex three-dimensional groundwater flow and solute transport problems, numerical models may be used to simulate very simple flow and transport conditions, which may just as easily be simulated using an analytical model. However, numerical models are generally used to simulate problems which cannot be accurately described using analytical models. 10/16/02 11

13 3.0 Groundwater Model Development Process 3.1 Hydrogeologic Characterization Proper characterization of the hydrogeological conditions at a site is necessary in order to understand the importance of relevant flow or solute-transport processes. With the increase in the attempted application of natural attenuation as a remedial action, it is imperative that a thorough site characterization be completed. This level of characterization requires more site-specific fieldwork than just an initial assessment, including more monitoring wells, groundwater samples, and an increase in the number of laboratory analytes and field parameters. Without proper site characterization, it is not possible to select an appropriate model or develop a reliably calibrated model. At a minimum, the following hydrogeological and geochemical information must be available for this characterization: Regional geologic data depicting subsurface geology. Topographic data (including surface-water elevations) Presence of surface-water bodies and measured stream-discharge (base flow) data Geologic cross sections drawn from soil borings and well logs. Well construction diagrams and soil boring logs. Measured hydraulic-head data. Estimates of hydraulic conductivity derived from aquifer and/or slug test data. Location and estimated flow rate of groundwater sources and sinks. Identification of chemicals of concern in contaminant plume. * Vertical and horizontal extent of contaminant plume. * Location, history, and mass loading or removal rate for contaminant sources or sinks. * Direction and rate of contaminant migration. * Identification of downgradient receptors. * Organic carbon content of sediments. * Appropriate geochemical field parameters (e.g. dissolved oxygen, Eh, ph) * Appropriate geochemical indicator parameters (e.g. electron acceptors, and degradation byproducts) * * Required only by fate and transport models. 10/16/02 12

14 These data should be presented in map, table, or graph format in a report documenting model development. 3.2 Model Conceptualization Model conceptualization is the process in which data describing field conditions are assembled in a systematic way to describe groundwater flow and contaminant transport processes at a site. The model conceptualization aids in determining the modeling approach and which model software to use. Questions to ask in developing a conceptual model include, but are not limited to: Are there adequate data to describe the hydrogeological conditions at the site? In how many directions is groundwater moving? Can the groundwater flow or contaminant transport be characterized as one-, two- or threedimensional? Is the aquifer system composed of more than one aquifer, and is vertical flow between aquifers important? Is there recharge to the aquifer by precipitation or leakage from a river, drain, lake, or infiltration pond? Is groundwater leaving the aquifer by seepage to a river or lake, flow to a drain, or extraction by a well? Does it appear that the aquifer's hydrogeological characteristics remain relatively uniform, or do geologic data show considerable variation over the site? Have the boundary conditions been defined around the perimeter of the model domain, and do they have a hydrogeological or geochemical basis? Do groundwater-flow or contaminant source conditions remain constant, or do they change with time? Are there receptors located downgradient of the contaminant plume? Are geochemical reactions taking place in onsite groundwater, and are the processes understood? Other questions related to site-specific conditions may be asked. This conceptualization step must be completed and described in the model documentation report. 10/16/02 13

15 3.3 Model Software Selection After hydrogeological characterization of the site has been completed, and the conceptual model developed, computer model software is selected. The selected model should be capable of simulating conditions encountered at the site. The following general guidelines should be used in assessing the appropriateness of a model: Analytical models should be used where: Field data show that groundwater flow or transport processes are relatively simple. An initial assessment of hydrogeological conditions or screening of remedial alternatives is needed. Numerical models should be used where: Field data show that groundwater flow or transport processes are relatively complex. Groundwater flow directions, hydrogeological or geochemical conditions, and hydraulic or chemical sources and sinks vary with space and time. A one-dimensional groundwater flow or transport model should be used primarily for: Initial assessments where the degree of aquifer heterogeneity or anisotropy is not known. Sites where a potential receptor is immediately downgradient of a contaminant source. Two-dimensional models should be used for: Problems which include one or more groundwater sources/sinks (e.g. pumping or injection wells, drains, rivers, etc.), Sites where the direction of groundwater flow is obviously in two dimensions(e.g. radial flow to a well, or single aquifer with relatively small vertical hydraulic head or contaminant concentration gradients), Sites at which the aquifer has distinct variations in hydraulic properties, Contaminant migration problems where the impacts of transverse dispersion are important and the lateral, or vertical, spread of the contaminant plume must be approximated. 10/16/02 14

16 Three-dimensional flow and transport models should generally be used where: The hydrogeologic conditions are well known, Multiple aquifers are present, The vertical movement of groundwater or contaminants is important. The rationale for selection of the appropriate model software should be discussed in the model documentation report. The choice of model software program for use at a site is the responsibility of the modeler. Any appropriate groundwater flow or fate and transport model software may be used provided that the model code has been tested, verified and documented. However, it is recommended that the model developer contact the GMP at the beginning of the investigation to discuss the selection of appropriate model software. In the event that the software is not currently used by the GMP, a copy of the software and the program documentation must be submitted to the GMP along with the model documentation report. 3.4 Model Calibration Model calibration consists of changing values of model input parameters in an attempt to match field conditions within some acceptable criteria. This requires that field conditions at a site be properly characterized. Lack of proper site characterization may result in a model that is calibrated to a set of conditions which are not representative of actual field conditions. The calibration process typically involves calibrating to steady-state and transient conditions. With steady-state simulations, there are no observed changes in hydraulic head or contaminant concentration with time for the field conditions being modeled. Transient simulations involve the change in hydraulic head or contaminant concentration with time (e.g. aquifer test, an aquifer stressed by a well-field, or a migrating contaminant plume). These simulations are needed to narrow the range of variability in model input data since there are numerous choices of model input data values which may result in similar steady-state simulations. Models may be calibrated without simulating steady-state flow conditions, but not without some difficulty. At a minimum, model calibration should include comparisons between model-simulated conditions and field conditions for the following data: Hydraulic head data, 10/16/02 15

17 Groundwater-flow direction, Hydraulic-head gradient, Water mass balance, Contaminant concentrations (if appropriate), Contaminant migration rates (if appropriate), Migration directions (if appropriate), and Degradation rates (if appropriate). Figure 6. Comparison between measured and computed hydraulic heads. 10/16/02 16

18 These comparisons should be presented in maps, tables, or graphs. A simple graphical comparison between measured and computed heads is shown in Figure 6. In this example, the closer the heads fall on the straight line, the better the goodness-of-fit. Each modeler and model reviewer will need to use their professional judgment in evaluating the calibration results. There are no universally accepted goodness-of-fit criteria that apply in all cases. However, it is important that the modeler make every attempt to minimize the difference between model simulations and measured field conditions. Typically, the difference between simulated and actual field conditions (residual) should be less than 10 percent of the variability in the field data across the model domain. A plot showing residuals at monitoring wells (calibration targets) is shown in Figure 7. A plot in this format is useful to show the goodness-of-fit at individual wells. Figure 7. Residuals from comparison of measured and computed heads at calibration targets. 10/16/02 17

19 The modeler should also avoid the temptation of adjusting model input data on a scale which is smaller than the distribution of field data. This process, referred to as "over calibration", results in a model that appears to be calibrated but has been based on a dataset that is not supported by field data. For initial assessments, it is possible to obtain useful results from models that are not calibrated. The application of uncalibrated models can be very useful in guiding data collection activities for hydrogeological investigations or as a screening tool in evaluating the relative effectiveness of remedial action alternatives. 3.5 History Matching A calibrated model uses selected values of hydrogeologic parameters, sources and sinks and boundary conditions to match field conditions for selected calibration time periods (either steady-state or transient). However, the choice of the parameter values and boundary conditions used in the calibrated model is not unique, and other combinations of parameter values and boundary conditions may give very similar model results. History matching uses the calibrated model to reproduce a set of historic field conditions. This process has been referred to by others as model verification. Figure 8. Comparison between computed and observed heads with time. 10/16/02 18

20 The most common history matching scenario consists of reproducing an observed change in the hydraulic head or solute concentrations over a different time period, typically one that follows the calibration time period (see Figure 8). The best scenarios for model verification are ones that use the calibrated model to simulate the aquifer under stressed conditions. The process of model verification may result in the need for further calibration refinement of the model. After the model has successfully reproduced measured changes in field conditions for both the calibration and history matching time periods, it is ready for predictive simulations. 3.6 Sensitivity Analysis A sensitivity analysis is the process of varying model input parameters over a reasonable range (range of uncertainty in values of model parameters) and observing the relative change in model response (see Figure 9). Typically, the observed changes in hydraulic head, flow rate or contaminant transport are noted. The purpose of the sensitivity analysis is to demonstrate the sensitivity of the model simulations to uncertainty in values of model input data. The sensitivity of one model parameter relative to other parameters is also demonstrated. Sensitivity analyses are also beneficial in determining the direction of future data collection activities. Data for which the model is relatively sensitive would require future characterization, as opposed to data for which the model is relatively insensitive. Model-insensitive data would not require further field characterization. 10/16/02 19

21 Figure 9. Simulated change in hydraulic head resulting from change in parameter value. 3.7 Predictive Simulations or Evaluation of Remediation Alternatives A model may be used to predict some future groundwater flow or contaminant transport condition. The model may also be used to evaluate different remediation alternatives, such as hydraulic containment, pump-and-treat or natural attenuation, and to assist with risk evaluation. In order to perform these tasks, the model, whether it is a groundwater flow or solute transport model, must be reasonably accurate, as demonstrated during the model calibration process. However, because even a well-calibrated model is based on insufficient data or oversimplifications, there are errors and uncertainties in a groundwater-flow analysis or solute transport analysis that make any model prediction no better than an approximation. For this reason, all model predictions should be expressed as a range of possible outcomes which reflect the uncertainty in model parameter values. The range of uncertainty should be similar to that used for the sensitivity analysis. The following examples demonstrate, in a limited manner, how model predictions may be presented to illustrate the range of possible outcomes resulting in model input data uncertainty. 10/16/02 20

22 Figure 10. Simulated uncertainty in hydraulic heads at calibration targets. Figure 10 shows the range in computed head at calibration targets at a particular point in time resulting from varying a model parameter over its range of uncertainty. If the purpose of the predictive simulations is to determine the future hydraulic head distribution in an aquifer, the upper and lower estimate of head should be presented so that appropriate decisions may be made. In the following example (Figure 11), hydraulic heads are predicted for a future time interval in response to changing stresses on the aquifer system. The predicted results using the calibrated flow model are presented as the black line. The red and blue lines show the hydraulic heads predicted using slightly different model input parameters which result in values of hydraulic head which are five percent higher and lower than the heads predicted using the calibrated model. Again, the range in predicted heads should be presented so that appropriate, or conservative, decisions may be made regarding the groundwater resource. 10/16/02 21

23 Figure 11. Predicted range in hydraulic heads. Figure 12 shows an example of delineating a wellhead protection area (WHPA) for a public water-supply. These WHPAs were simulated using the range of hydraulic conductivity values reported from an aquifer test at a municipal well. The WHPA delineated using the low hydraulic conductivity value (shown in brown) is wider than and not as long as the WHPA delineated using the high value of hydraulic conductivity (shown in gold). In order to be protective of the public water supply, it's important that the delineated WHPA is conservatively estimated. In this example, the final recommended WHPA is a composite of the two previously simulated WHPAs. 10/16/02 22

24 Figure 12. Simulated wellhead protection areas using range of hydraulic conductivities. This final example (Figure 13) shows the simulated contaminant concentrations downgradient from a source area. The purpose of the simulation was to estimate the contaminant concentration at the point of compliance, approximately 1000 feet downgradient from the source area. The difference between the two simulations shown in the figure is the value of organic carbon fraction used in each simulation. Increasing the organic carbon concentration increases the degree of attenuation of the contaminant plume. Since the organic carbon content of sediments is very seldom known with a high degree of certainty, and a single value is typically used in fate and transport models, it is best to present the simulation results as a range of concentrations which might result. This can be said for any of the model input parameters used in fate and transport models. 10/16/02 23

25 Figure 13. Simulated contaminant concentrations. 4.0 Groundwater Models and Performance Monitoring Groundwater models are used to predict the migration pathway and concentrations of contaminants in groundwater. The accuracy of model predictions depends upon successful calibration and verification of the model in determining groundwater flow directions, transport of contaminants and chemical reactions, and the applicability of the groundwater flow and solute transport equations to the problem being simulated. Errors in the predictive model, even though small, can result in gross errors in solutions projected forward in time. Among other things, performance monitoring is required to compare future field conditions with model predictions to assess model error. A model may appropriately be considered as part of a compliance issue if specified to be prepared as part of a RAP or negotiated settlement. However, a model cannot provide verification of hydraulic containment of a contaminant plume or the chemical concentration at the point of human or environmental exposure. At best, a model can only provide an estimate of the effectiveness of a remediation system, verification of actual performance must be demonstrated by the measurement of appropriate field parameters. 10/16/02 24

26 4.1 Predictive Simulations Predictive simulations may be used to estimate the hydraulic response of an aquifer, the possible migration pathway of a contaminant, the contaminant mass removal rate from an aquifer, or the concentration of a contaminant at a point of compliance at some future point in time. The predictive simulations must be viewed as estimates, not certainties, to aid the decision-making process. As an example, the design of a groundwater remediation system may be based on predictive model simulations. A model may be used to predict the pumping rate needed to capture a contaminant plume and to estimate the contaminant concentration of the extracted groundwater. Monitoring of hydraulic heads and contamination concentrations must be used to verify hydraulic containment and remediation of the contaminant plume. Predictive simulations are based on the conceptual model developed for the site, the values of hydrogeological or geochemical parameters used in the model, and on the equations solved by the model software. Errors in values of model parameters, or differences between field conditions and the conceptual model or model equations will result in errors in predictive simulations. Models are calibrated by adjusting values of model parameters until the model response closely reproduces field conditions within some acceptable criteria in an attempt to minimize model error. However, the time period over which a model is calibrated is typically very small, especially when compared to the length of time used for predictive simulations. Relatively small errors observed during the time period over which the model calibration or history matching was performed may be greatly magnified during predictive simulations because of the greater time period length typically used in predictive simulations. The growth in errors resulting from projecting model simulation into the future need to be evaluated by monitoring field conditions over the time period of the predictive simulation or until appropriate cleanup criteria have been achieved. There is always some degree of uncertainty in predictive models. Predictive models should be conservative. That is, given the uncertainty in model input parameters and the corresponding uncertainty in predictive model simulations, model input values should be selected which result in a worst-case simulation. Site-specific data may be used to support a more reasonable worst-case scenario. Or stated another way, site-specific data should be collected to limit the range of uncertainty in predictive models. 10/16/02 25

27 Figure 14. Example of growth of model error in predictive simulation. Further re-calibration of the model to minimize the difference between model simulations and performance monitoring data may not be warranted. 4.2 Performance Monitoring As previously stated, groundwater model simulations are an approximation of the actual system behavior and monitoring of field conditions are needed to evaluate error in model predictions. ASTM guidelines state that Predictive modeling is not used in the RBCA process as a substitute for site-specific Verification data (ASTM Emergency Standard E , Appendix X3.4.3). Performance monitoring is required as a means of physically measuring the actual behavior of the hydrogeologic system and demonstrating compliance with environmental statutes. Groundwater model simulations are estimates and may not be substituted in place of measurement of field data. Examples of applications of contaminant transport and groundwater modeling requiring performance monitoring would include, but not be limited to the following: Hydraulic containment systems for which certain geochemical and hydraulic head criteria, which measure the success of the remedial action, have been specified in a remedial action plan. An example of a monitoring network for a hydraulic containment system is shown in Figure /16/02 26

28 Groundwater/surface-water interface (GSI) mixing zones for which the chemistry of the contaminant plume discharging to a surface-water body must meet appropriate criteria. An example of a monitoring system for a GSI is shown in Figure 12. Natural attenuation remedies in which a contaminant transport model predicts that site specific contaminants will reach regulatory limits prior to reaching a specified point. The degree of performance monitoring required at a site depends on the conditions or actions which have been simulated and the associated level of risk to the downgradient receptors, if applicable. As an example, hydraulic containment of contaminant plumes by pump-and-treat systems would require extensive monitoring of hydraulic heads and possibly less frequent monitoring of groundwater quality. Hydraulic-head monitoring points should be at well distributed locations to demonstrate the desired hydraulic response. Monitoring of groundwater chemistry would be needed at locations downgradient of, or lateral to, the hydraulic containment system, with less frequent chemical monitoring at locations within the capture zone of the system. Sites at which contaminant biodegradation has been simulated would require extensive monitoring of appropriate chemical parameters and less frequent monitoring of hydraulic heads. Chemical monitoring would be required at a sufficient number of locations to evaluate the migration or mass removal of contaminants and groundwater quality at the designated points of compliance. The performance monitoring will be based ultimately on actual laboratory analytical data and the chemical criteria established for the site. 5.0 Documentation of Groundwater Flow and Fate and Transport Models A groundwater model developed for a site, whether an analytical or numerical model, should be described in sufficient detail so that the model reviewer may determine the appropriateness of the model for the site or problem that is simulated. Submittal of a model documentation report and model datasets (in digital format) is required. A suggested format for this report is contained in the following sections and in Appendix Report 10/16/02 27

29 Groundwater modeling documentation must detail the process by which the model was selected, developed, calibrated, verified and utilized. The model documentation report must include the following information: A description of the purpose and scope of the model application. Presentation of the hydrogeologic data used to characterize the site. Documentation of the source of all data used in the model, whether derived from published sources or measured or calculated from field or laboratory tests. Description of the model conceptualization. Identify the model selected to perform the task, its applicability and limitations. A discussion of the modeling approach. Documentation of all calculations. Summary of all model calibration, history matching and sensitivity analysis results. Present all model predictive simulation results as a range of probable results given the range of uncertainty in values of model parameters. The organization of the report should include the following sections: Title Page Table of Contents List of Figures List of Tables Introduction Objectives Hydrogeologic Characterization Model Conceptualization Modeling Software Selection Model Calibration History matching Sensitivity Analysis Predictive Simulations or Use of Model for Evaluation of Remediation Alternatives Recommendations and Conclusions References Tables 10/16/02 28

30 Figures Appendices Tables The following is a list of tables that should appear within the body of the model documentation report or in attached appendices: Well and boring log data including: Name of all wells or borings, Top of casing elevation, Well coordinate data, Well screen interval, Hydraulic head data, Elevation of bottom of model, Hydraulic conductivity or transmissivity, and Groundwater quality chemical analyses (if appropriate). Aquifer test or slug test data. Model calibration and verification result showing a comparison of measured and simulated calibration targets and residuals. Results of sensitivity analysis showing the range of adjustment of model parameters and resulting change in hydraulic heads or groundwater flow rates. 10/16/02 29

31 Other data, not listed above, may lend itself to presentation in tabular format. Where appropriate, the aquifer for which the data apply should be clearly identified in each table Figures The following is a list of the types of figures (maps or cross sections) which should be included in the model documentation report: Regional location map with topography. Site map showing soil boring and well locations, and site topography. Geologic cross sections. Map showing the measured hydraulic-head distribution. Maps of top and/or bottom elevations of aquifers and confining units. Areal distribution of hydraulic conductivity or transmissivity. Map of areal recharge (if appropriate). Model grid with location of different boundary conditions used in the model. Simulated hydraulic-head maps. Contaminant distribution map(s) and/or cross sections showing vertical distribution of contaminants (if appropriate). Map showing simulated contaminant plume distribution (if appropriate). Other types of information, not listed above, may be presented in graphic format. Figures that are used to illustrate derived or interpreted surfaces such as layer bottom elevations and hydraulic-head maps should have the data used for the interpolation also posted upon the figure. As an example, measured hydraulichead maps should identify the observation points and the measured hydraulic-head elevation. Similarly, the simulated hydraulic-head maps should locate the calibration target points and the residual between the measured and modeled data. All figures should provide the following information: North Arrow Date of figure preparation and data collection Title Bar Scale Bar 10/16/02 30

32 Legend All maps or cross sections should be drawn to scale with an accurate scale clearly displayed on each figure. When feasible, all figures should be the same scale. Figures that apply to specific aquifers should be clearly labeled Additional Data Additional data may be required to be presented in the model documentation report. Examples of additional data are as follows: Additional studies work plans providing for the collection of additional data where model simulations show data deficiencies, and Groundwater monitoring plans/proposals/recommendations to collect data needed to verify model predictions. Other data may be required, depending on the conditions at the site. These additional subjects should be addressed within the body of the report. This may include additional figures and tables, or report sections. 5.2 Model Review Submittal Procedures It is necessary that model results are verified rather than accepted. This process of verification and review of groundwater flow and solute transport models is performed by the Groundwater Modeling Program (GMP). DEQ project managers requiring model review assistance should contact the GMP manager to obtain assistance from the GMP. A checklist showing items needed to conduct a model review is contained in Appendix 3. The suggested format for a model documentation report is contained in Appendix 4. Finally, a form to be used for requesting a model review is contained in Appendix 5. This form should be submitted along with the items listed in the checklist found in Appendix 3. 10/16/02 31

33 A copy of the model dataset in digital format must be provided as part of model documentation by the party responsible for developing the model. The datasets for the different simulations (model calibration, history matching and predictive simulations) must be provided and clearly labeled. If a model is used that is proprietary and not currently supported by the GMP, it may be necessary for the modeler to provide a copy of the model software for model review purposes only. The copy of this model will be returned after model review has been completed. 10/16/02 32