IRP Operable Units and Site Location, BRAGS, CTO-0140, Figure 1

IRP Operable Units and Site Location, BRAGS, CTO-0140, Figure 1

FINAL

BASEWIDE REMEDIATION ASSESSMENT

GROUNDWATER STUDY (BRAGS)

MARINE CORPS BASE

CAMP LEJEUNE, NORTH CAROLINA

CONTRACT TASK ORDER 0140

APRIL 20, 1998

Prepared for:

DEPARTMENT OF THE NAVY

ATLANTIC DIVISION

NAVAL FACILITIES

ENGINEERING COMMAND

Norfolk, Virginia

Under:

LANTDIV CLEAN Program

Contract N62470-89-D-4814

Prepared by:

BAKER ENVIRONMENTAL, INC.

Coraopolis, Pennsylvania

TABLE OF CONTENTS

Page

EXECUTIVE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ES-1

1.0 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-1

1.1 Modeling Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-2

1.2 Report Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3

1.3 Location and Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3

1.4 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3

1.5 Topography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-4

1.6 Surface Water Hydrology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-4

2.0 PREVIOUS INVESTIGATIONS AND COMPUTER SIMULATIONS . . . . . . . . . . . . . 2-1

3.0 HYDROGEOLOGY OF THE CAMP LEJEUNE AREA . . . . . . . . . . . . . . . . . . . . . . . . 3-1

3.1 Physiography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-1

3.2 Geologic and Hydrogeologic Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-1

3.3 Conceptual Model of Groundwater Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-3

3.4 Hydraulic Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-5

3.4.1 Surficial Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-5

3.4.2 Castle Hayne Confining Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-5

3.4.3 Castle Hayne Aquifer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-5

4.0 BRAGS GROUNDWATER FLOW MODEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-1

4.1 Finite-Difference Layered Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-2

4.2 Model Boundary Conditions .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-3

4.2.1 Specified (Constant) Head Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-3

4.2.2 General Head Boundary Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-4

4.2.3 Well Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-4

4.2.4 River Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-5

4.2.5 Drain Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-6

4.3 Steady-State Modeling Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-6

4.3.1 Calibration Targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-7

4.3.2 Calibration Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-7

4.3.3 Statistical Evaluation of Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . 4-8

4.4 Calibrated Results of Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-8

4.4.1 Layer 1 -- Surficial Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-10

4.4.1.1 Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-10

4.4.1.2 Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-12

4.4.2 Layer 2 - Castle Hayne Confining Unit . . . . . . . . . . . . . . . . . . . . . . . 4-12

4.4.3 Layer 3 - Upper Castle Hayne Aquifer . . . . . . . . . . . . . . . . . . . . . . . . 4-13

4.4.3.1 Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-13

4.4.3.2 Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-13

4.4.4 Layer 4 - Castle Hayne Fractured Limestone Unit . . . . . . . . . . . . . . . 4-14

4.4.5 Layer 5 -- Lower Castle Hayne Aquifer . . . . . . . . . . . . . . . . . . . . . . . 4-15

4.4.5.1 Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-15

4.4.5.2 Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-15

4.4.6 Three-Dimensional Analysis of Groundwater Flow . . . . . . . . . . . . . . 4-15

TABLE OF CONTENTS

(Continued)

Page

4.5 Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-17

4.5.1 Effects of Altering Recharge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-17

4.5.2 Effects of Altering Horizontal Hydraulic Conductivity . . . . . . . . . . . . 4-17

4.5.3 Effects of Altering Leakance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-17

4.5.4 Effects of Altering GHB Cell Conductance . . . . . . . . . . . . . . . . . . . . 4-17

4.5.5 Effects of Altering River Cell Conductance . . . . . . . . . . . . . . . . . . . . 4-18

4.5.6 Effects of Altering Drain Cell Conductance . . . . . . . . . . . . . . . . . . . . 4-18

4.5.7 Recommended Changes to the Model . . . . . . . . . . . . . . . . . . . . . . . . 4-18

4.6 BRAGS Groundwater Flow Model Summary . . . . . . . . . . . . . . . . . . . . . . . . . 4-19

5.0 SITE 82 GROUNDWATER FLOW MODEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-1

5.1 Finite-Difference Layered Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-1

5.2 Model Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-2

5.2.1 General Head Boundary Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-2

5.2.2 Well Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-2

5.2.3 River Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-3

5.2.4 Drain Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-3

5.3 Steady-State Modeling Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-3

5.3.1 Pre-Pumping Calibration Targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-4

5.3.2 Calibration Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-4

5.3.3 Statistical Evaluation of Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . 5-4

5.4 Calibrated Results of Pre-Pumping Simulation . . . . . . . . . . . . . . . . . . . . . . . . . 5-4

5.4.1 Layer 1 -- Surficial Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-5

5.4.1.1 Pre-Pumping Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-5

5.4.1.2 Pre-Pumping Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-7

5.4.2 Layer 2 - Castle Hayne Aquifer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-8

5.4.2.1 Pre-Pumping Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-8

5.4.2.2 Pre-Pumping Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-8

5.5 Results of Remediation Scenario Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . 5-9

5.5.1 Layer 1 -- Surficial Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-10

5.5.1.1 Remediation Scenario Input . . . . . . . . . . . . . . . . . . . . . . . . . 5-10

5.5.1.2 Remediation Scenario Output . . . . . . . . . . . . . . . . . . . . . . . . 5-10

5.5.2 Layer 2 - Castle Hayne Aquifer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-10

5.5.2.1 Remediation Scenario Input . . . . . . . . . . . . . . . . . . . . . . . . . 5-10

5.5.2.2 Remediation Scenario Output . . . . . . . . . . . . . . . . . . . . . . . . 5-11

5.6 Site 82 Groundwater Flow Model Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 5-11

6.0 CONCLUSIONS AND RECOMMENDATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-1

7.0 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-1

LIST OF FIGURES

1-1 Operable Units and IRP Site Location Plan

1-2 UST Site Location Plan

3-1 Isopach Contour Map -- Castle Hayne Confining Unit

3-2 Elevation Contour Map -- Top of Castle Hayne

3-3 Typical Annual Water Budget

3-4 Idealized Hydrogeologic Cross-Section of New River

3-5 Idealized Hydrogeologic Profile of New River

4-1 Finite Difference Grid Location Map -- BRAGS Model

4-2 Schematic of 3-D Five-Layer BRAGS Model

4-3 Computed versus Observed Values in Layer 1

4-4 Error versus Observed Values in Layer 1

4-5 Computed versus Observed Values in Layer 3

4-6 Error versus Observed Values in Layer 3

4-7 Computed versus Observed Values in Layer 5

4-8 Error versus Observed Values in Layer 5

4-9 Spatial Distribution of Error - Layer 1- Northern Areas

4-10 Spatial Distribution of Error - Layer 1- Southern Areas

4-11 Spatial Distribution of Error - Layer 3- Northern Areas

4-12 Spatial Distribution of Error - Layer 3- Southern Areas

4-13 Spatial Distribution of Error - Layer 5- Camp Geiger and Montford Point Areas

4-14 Spatial Distribution of Error - Layer 5- Tarawa Terrace and Paradise Point Areas

4-15 Spatial Distribution of Error - Layer 5- Hadnot Point Area

4-16 Spatial Distribution of Error - Layer 5- Southern Areas

4-17 Elevation of Bottom of Layer 1

4-18 MODFLOW Cells in Layer 1

4-19 Simulated Water Table Elevation Contours in Layer 1

4-20 Values of Leakance (ft/day/ft) in Layer 2

4-21 Elevation of Bottom of Layer 3

4-22 Simulated Groundwater Elevation Contours in Layer 3

4-23 Water Supply Well MODFLOW Cells in Layer 4

4-24 Elevation of Bottom of Layer 5

4-25 Simulated Groundwater Elevation Contours in Layer 5

4-26 Map View of 3-D Flow Vectors in Layer 4 Showing Effects of Pumping Water Supply Wells in the Northern Areas

4-27 West-to-East Cross-Section of Simulated Groundwater Flow Vectors (Row 47)

4-28 Effects of Recharge

4-29 Effects of Horizontal Hydraulic Conductivity

4-30 Effects of Leakance

4-31 Effects of GHB Conductance

4-32 Effects of River Conductance

4-33 Effects of Drain Conductance

4-34 Comparison of Mean Error (ME) Values

4-35 Comparison of Root Mean Square Error (RMSE) Values

4-36 Comparison of Mean Absolute Error (MAE) Values

LIST OF FIGURES

(Continued)

5-1 Finite Difference Grid Location Map

5-2 Finite Difference Grid

5-3 Bottom Elevation of Layer 1 (Surficial Unit)

5-4 General Head Boundary, River, and Drain Cells in Layer 1

5-5 Leakance Factor in Layer 1

5-6 Simulated Water Table Surface Contours (Pre-Remediation) in Layer 1

5-7 Bottom Elevation of Layer 2 (Castle Hayne Aquifer)

5-8 Water Supply Well Cells in Layer 2 (Castle Hayne Aquifer)

5-9 Simulated Castle Hayne Groundwater Contours (Pre-Remediation) in Layer 2

5-10 Volatile Organic Compounds in Shallow Groundwater

5-11 Volatile Organic Compounds in Deep Groundwater

5-12 Well Log for Supply Well HP-651

5-13 Shallow Extraction Wells in the Surficial Unit (Layer 1)

5-14 Simulated Water Table Contours During Remediation Showing Capture Zones in the Surficial

Unit (Layer 1)

5-15 Capture Zone Pathline Map - Simulated Remediation Water Table Surface

5-16 Deep Extraction Wells in the Castle Hayne Aquifer (Layer 2)

5-17 Simulated Castle Hayne Groundwater Contours During Remediation Showing Capture Zones

(Layer 2)

5-18 Capture Zone Pathline Map - Simulated Remediation Castle Hayne Piezometric Surface

LIST OF TABLES

1-1 Land Utilization Within Developed Areas of MCB, Camp Lejeune

3-1 Hydraulic Conductivity Data from Sites 6, 82, and 73 - Surficial Unit

3-2 Hydraulic Conductivity Data from the Castle Hayne Aquifer

4-1 Pumping Rates of Water Supply Wells at MCB, Camp Lejeune

4-2 Statistical Summary of BRAGS Simulation

4-3 Hydrologic Budget Summary for BRAGS Simulation

5-1 Statistical Summary of Site 82 Groundwater Flow Model

LIST OF APPENDICES

A Site 82 Pumping Test Data Evaluation

B BRAGS Model Input and Output Files (CD-ROM)

C Site 82 Model Input and Output Files (CD-ROM)

ft/day-s ft/day/ft

FS

G3CTM

IP

IR

K

K ii

K v gpm

GWQ

HFB

HPIA

AFCEE

ARM b

BC

BRAGS

C drn

C riv

CA

CCLs

CCP

CERCLA cfd cis-1,2-DCE cm/sec

CMS

CTO

DCE

DNAPLs

DoN

ESE

FFA k:\62470\140phase\finaldoc\report\finlmodl.wpd

LIST OF ACRONYMS AND ABBREVIATIONS

Air Force Center for Environmental Excellence

Absolute Residual Mean (see MAE)

Aquitard Thickness

Boundary Conditions (MODFLOW)

Basewide Remediation Assessment Groundwater Study

Drain Cell Conductance (MODFLOW)

River Cell Conductance (MODFLOW)

Corrective Action

Ceiling Concentration Limits

Central Coastal Plain

Comprehensive Environmental Response, Compensation, and Liability

Act (Superfund)

Cubic Feet per Day cis-1,2-dichloroethene

Centimeters per Second

Corrective Measure Study

Contract Task Order cis and trans-1,2-dichloroethene

Dense Non-Aqueous Phase Liquids

Department of the Navy

Environmental Science and Engineering, Inc.

Federal Facilities Agreement

Square Feet/Day (unit of transmissivity)

Feet/Day (unit of velocity or hydraulic conductivity)

Feet/Day/Feet (unit of leakance)

Feasibility Study

G-3 Contaminant Transport Model (for groundwater discharge to surface water -- NC DENR)

Gallons per Minute

Groundwater Quality

Horizontal Flow Barrier (MODFLOW)

Hadnot Point Industrial Area

Implementation Plan

Installation Restoration

Hydraulic Conductivity (ft/day)

Hydraulic Conductivity Tensor in the i direction (MODFLOW)

Vertical Hydraulic Conductivity (ft/day) v

April 20, 1998 version

PCE ppb

RASA

RCRA

RF

RI

RI/FS

RM

RMS

RMSE

RSD

L m

ME

MAE

MCB mi

2

MSL n

NC

NC DENR

NPL

OU

VC

VOC

W

WAR

X

SDE

SMP

TCE trans-1,2-DCE

USEPA

USGS

UST

Length of Cell (MODFLOW)

Thickness of River or Stream Sediments (MODFLOW)

Mean Error

Mean Absolute Error

Marine Corps Base

Square Mile

Mean Sea Level

Sample Size

North Carolina

North Carolina Department of Environment and Natural Resources

National Priorities List

Operable Unit

Tetrachloroethene (perchloroethylene)

Parts per Billion

Regional Aquifer System Analysis

Resource Conservation and Recovery Act

Retardation Factors

Remedial Investigation

Remedial Investigation/Feasibility Study

Residual Mean (see ME)

Root Mean Square (see RMSE)

Root Mean Square Error

Residual Standard Deviation (see SDE)

Standard Deviation of the Errors

Site Management Plan

Trichloroethene

Trans-1,2-dichloroethene

United States Environmental Protection Agency

United States Geological Survey

Underground Storage Tank

Vinyl Chloride

Volatile Organic Compound

Width of Cell (MODFLOW)

Water and Air Research, Inc.

X Axis of Cartesian Coordinates associated with Hydraulic Conductivity

(MODFLOW) k:\62470\140phase\finaldoc\report\finlmodl.wpd

vi

April 20, 1998 version

Y

Z

°F

Y Axis of Cartesian Coordinates associated with Hydraulic Conductivity

(MODFLOW)

Z Axis of Cartesian Coordinates associated with Hydraulic Conductivity

(MODFLOW)

Degrees Fahrenheit k:\62470\140phase\finaldoc\report\finlmodl.wpd

vii

April 20, 1998 version

EXECUTIVE SUMMARY

EXECUTIVE SUMMARY

Marine Corps Base (MCB), Camp Lejeune, bisected by the tidal estuary New River, borders the

Atlantic Ocean and encompasses approximately 236 square miles of the Atlantic Coastal Plain of North

Carolina. Ongoing management of water supply withdrawals and evaluations of potential impacts to groundwater due to remediation are required. Although regional geologic and hydrogeologic studies have been conducted by the United States Geological Survey (USGS) at MCB, Camp Lejeune (Harned, et al, 1989 and Cardinell, et al, 1993), the resulting effects of these remedial groundwater pump and treatment systems on the underlying aquifers have not been evaluated. Existing USGS Regional

Aquifer System Analysis (RASA) and North Carolina Department of Environment, and Natural

Resources (NC DENR) groundwater flow models were evaluated; however, over-large scales and lack of detail in the impacted surficial units precluded their use at MCB, Camp Lejeune.

The Fiscal Year 1998 Site Management Plan (SMP) for MCB, Camp Lejeune, the primary document referenced in the FFA, identifies 42 sites that require Remedial Investigation/Feasibility Study (RI/FS) activities. In addition to the RI/FS sites, 135 underground storage tank (UST) sites have also been identified, 126 of which have undergone environmental investigations. Based on information obtained from the SMP and from Base personnel, 28 groundwater remediation systems (i.e., groundwater pumping and treatment or air sparging) are currently operating, waiting construction, or under consideration.

Consequently, the focus of this study is to develop a Basewide groundwater flow model which can be used to evaluate the effects of various groundwater remediation projects under the auspices of the

Basewide Remediation Assessment Groundwater Study (BRAGS) that are active or planned for MCB,

Camp Lejeune. Two three-dimensional groundwater flow models were developed for use at MCB,

Camp Lejeune: a comprehensive Basewide model (referred to herein as the BRAGS model) and a sitespecific model for Site 82 (Piney Green VOC Area). The BRAGS model was constructed first based on composite groundwater elevation data taken from 30 IRP/UST sites at the Base and from published

USGS data collected from the water supply wells at the Base. The site-specific model was then constructed on the foundation laid by the BRAGS model using data primarily from IRP Sites 82, 6, 9, and 3, UST Site 889-891, and from the nearby water supply wells. Both models were constructed with

MODFLOW (a finite-difference numerical flow model) and calibrated to measured head data collected by Baker from 1992 to 1993. MODPATH was used to generate particle pathlines based on the results of MODFLOW.

k:\62470\140phase\finaldoc\report\finlmodl.wpd

ES-1

April 20, 1998 version

The objectives of this modeling effort were to provide LANTDIV and MCB, Camp Lejeune with one or more working groundwater flow models that can be used to:

!

Describe how groundwater flows beneath the entire Base as well as under individual sites of concern.

!

Demonstrate the effects of groundwater withdrawals (supply and remedial) on the aquifers in question (most notably the surficial unit and the Castle Hayne Aquifer).

!

Predict the relative effectiveness of various remediation schemes at individual sites

(including Site 82).

As "working" models, it is imperative that these groundwater flow models be updated as new information becomes available. Only updated models will be effective decision-making tools to optimize groundwater resource management, protection, and restoration. It is envisioned that personnel at LANTDIV or Camp Lejeune (or their representatives) will update and use these models to determine the relative effectiveness of various remedial scenarios at individual sites around the Base.

The BRAGS model was constructed first so that the conceptual model of the entire Base could be tested. After the BRAGS model was calibrated, the model for Site 82 was constructed. This enabled the use of the previously calibrated inputs to be used, with some adjustments, at the site level of detail.

As new information became available during the course of the study, the BRAGS model was updated and recalibrated. The update incorporated the new data from the Site 82 model (including the results of the pumping test) and from the Site 73 model into the BRAGS model.

The BRAGS groundwater flow model presented herein portrays the three-dimensional pattern of groundwater flow within the surficial unit and the Castle Hayne Aquifer. The BRAGS model predicts the elevation and flow direction of the surficial and Castle Hayne groundwater in many areas around the Base where no data currently exist. The BRAGS model also demonstrates that discharge to the

New River and its tributaries is the controlling factor on flow directions in the Castle Hayne Aquifer in the vicinity of Camp Lejeune. The model output indicated that the relatively high-volume withdrawal rates of the supply wells have a localized effect on the water levels in the Castle Hayne; however, large numbers of actively pumping wells in small areas have the potential to induce saltwater intrusion into the upper Castle Hayne Aquifer. This effect is most pronounced in the Paradise Point k:\62470\140phase\finaldoc\report\finlmodl.wpd

ES-2

April 20, 1998 version

area along Brewster Boulevard. To minimize drawdown and the resulting potential for saltwater intrusion, actively pumping water supply wells should not be grouped together in small areas but should be spread out in a line perpendicular to the ambient flow direction (not parallel to it).

One of the concerns that initiated this modeling effort was that the potential number of pump and treat remedial actions at the Base may negatively impact the supply of available groundwater. The BRAGS model strongly indicated that the low volumes of water withdrawn from the surficial unit and/or the

Castle Hayne Aquifer during such remedial actions will not measurably affect the groundwater supply at the Base. The results of the Site 82 model corroborate this theory.

The Site 82 model describes the three-dimensional pattern of groundwater flow in the surficial unit and

Castle Hayne Aquifer. The Site 82 model demonstrates the effects of proposed remedial groundwater withdrawals on the surficial unit and the Castle Hayne Aquifer. The model also demonstrates that the relatively low-volume withdrawal rates of the extraction wells will have a localized effect on the water levels in the surficial unit and the Castle Hayne Aquifer.

The Site 82 model directly addressed the third objective: it clearly showed the relative effectiveness of various site-specific remediation schemes. The locations of the extraction wells in the surficial and in the Castle Hayne Aquifer were finalized by the successful running of the model. "Success" was indicated by complete hydraulic control or "capture" of the contaminant plume. Also, the model indicated that the low volumes of water withdrawn during remedial actions in the surficial unit or the upper Castle Hayne will not measurably affect the groundwater supply at the Base.

The groundwater flow models described herein will be useful in managing the future RI activities at the Base. The BRAGS model will be especially useful for determining the groundwater flow patterns in areas where no data currently exists and it gives a regional perspective on site-specific modeling.

Future groundwater flow and/or contaminant transport modeling done at the site level should be coordinated with the BRAGS groundwater flow model so that the "big picture" of the groundwater flow is consistent across the Base.

k:\62470\140phase\finaldoc\report\finlmodl.wpd

ES-3

April 20, 1998 version

BASEWIDE REMEDIATION ASSESSMENT GROUNDWATER STUDY (BRAGS)

Marine Corps Base, Camp Lejeune, North Carolina

1.0

INTRODUCTION

Marine Corps Base (MCB), Camp Lejeune was placed on the Comprehensive Environmental Response,

Compensation, and Liability Act (CERCLA) National Priorities List (NPL) on October 4, 1989

(54 Federal Register 41015, October 4, 1989). Subsequent to this listing, the United States

Environmental Protection Agency (USEPA) Region IV, the North Carolina Department of Environment and Natural Resources (NC DENR), and the United States Department of the Navy (DoN) entered into a Federal Facilities Agreement (FFA) to conduct remedial investigations at MCB, Camp Lejeune. The primary purpose of the FFA is to ensure that environmental impacts associated with past and present activities at MCB, Camp Lejeune are thoroughly investigated and appropriate CERCLA response/Resource Conservation and Recovery Act (RCRA) corrective action alternatives are developed and implemented, as necessary, to protect public health, welfare, and the environment (FFA, 1989).

The Fiscal Year 1998 Site Management Plan (SMP) for MCB, Camp Lejeune, the primary document referenced in the FFA, identifies 42 sites that require Remedial Investigation/Feasibility Study (RI/FS) activities. In addition to these sites, 135 underground storage tank (UST) sites have also been identified, of which 126 sites have undergone environmental investigations. Based on information obtained from the SMP and from Base personnel, more than 28 of these IRP/UST sites are currently undergoing or are proposed for groundwater remediation actions (i.e., groundwater pumping and treatment or air sparging). Although regional geologic and hydrogeologic studies have been conducted by the United States Geological Survey (USGS) at MCB, Camp Lejeune (Harned, et al, 1989 and

Cardinell, et al, 1993), the resulting effects of these groundwater pump and treatment systems on the underlying aquifers have not been evaluated. Consequently, the focus of this study is to develop a

Basewide groundwater flow model which can be used to evaluate the effects of various groundwater remediation projects under the auspices of the Basewide Remediation Assessment Groundwater Study

(BRAGS) that are active or planned for MCB, Camp Lejeune.

In the course of this modeling effort, two working groundwater flow models were developed for use at

MCB, Camp Lejeune: a comprehensive Basewide model and a site-specific model for Site 82 (Piney

Green VOC Area). The BRAGS model was constructed first based on groundwater elevation data taken from 30 sites at the Base and from USGS data collected from the water supply wells at the Base.

k:\62470\140phase\finaldoc\report\finlmodl.wpd

1-1

April 20, 1998 version

The site-specific model was constructed on the foundation laid by the BRAGS model using data primarily from IRP Sites 82, 6, 9, and 3, UST Site 889-891, and from the nearby water supply wells.

Both models were calibrated to measured head data collected by Baker from 1992 to 1993.

1.1

Modeling Objectives

The objectives of this modeling effort were to provide the Atlantic Division, Naval Facilities

Engineering Command (LANTDIV) and MCB, Camp Lejeune with one or more working groundwater flow models that can be used to:

1.

Describe how groundwater flows beneath the entire Base as well as under individual sites of concern;

2.

Demonstrate the effects of groundwater withdrawals (supply and remedial) on the aquifers in question (most notably the surficial unit and the Castle Hayne Aquifer); and,

3.

Predict the relative effectiveness of various remediation schemes at individual sites

(including Site 82).

As "working" models, it is imperative that these groundwater flow models be updated as new information becomes available. Only updated models will be effective decision-making tools for optimal groundwater resource management, protection, and restoration. It is envisioned that personnel at LANTDIV or Camp Lejeune will update and use these models to determine the relative effectiveness of various remedial scenarios at individual sites around the Base.

The BRAGS model was constructed first so that the conceptual model of the entire Base could be tested. After the BRAGS model was calibrated, the model for Site 82 was constructed. This enabled the use of the previously calibrated inputs to be used, with some adjustments, at the site level of detail.

As new information became available during the course of the study, the BRAGS model was updated and recalibrated as necessary. The update incorporated the new data from the Site 82 pumping test into the BRAGS model.

1.2

Report Organization

k:\62470\140phase\finaldoc\report\finlmodl.wpd

1-2

April 20, 1998 version

The Final BRAGS report is comprised of one text volume which includes appendices of data from previously conducted pump tests at the Base as well as electronic input and output from the groundwater flow models (provided in Appendices B and C on CD-ROM). The section headings included within this text volume are as follows:

! Previous Investigations and Computer Simulations - Section 2.0

! Hydrogeology of the Camp Lejeune Area - Section 3.0

! BRAGS Groundwater Flow Model - Section 4.0

! Site 82 Groundwater Flow Model - Section 5.0

! Conclusions and Recommendations - Section 6.0

! References - Section 7.0

! Pumping Test Data from Site 82 - Appendix A

! BRAGS Model Data - Appendix B

! Site 82 Model Data - Appendix C

1.3 Location and Setting

MCB, Camp Lejeune is located on the Atlantic Coastal Plain of North Carolina in Onslow County. The

Base is in the Tidewater (i.e., tidally-influenced) Region of the North Carolina Coastal Plain (Stuckey,

1965). The facility encompasses approximately 236 square miles and is bisected by the New River.

The New River flows in a southerly direction through Camp Lejeune and forms a large, meandering estuary before entering the Atlantic Ocean. The southeastern border of Camp Lejeune is the Atlantic

Ocean shoreline. The western and northeastern boundaries of the facility are U.S. Route 17 and State

Route 24, respectively. The City of Jacksonville borders Camp Lejeune to the north (see

Figure 1-1 ).

Figure 1-1 also shows the locations of the IRP Sites within each Operable Unit (OU) and

Figure 1-2

shows the locations of the UST Sites around the Base.

1.4 History

Construction of MCB, Camp Lejeune began in April 1941 at the Hadnot Point Industrial Area (HPIA), where major functions of the Base are located today. The MCB, Camp Lejeune complex, designed to be the "World's Most Complete Amphibious Training Base," consists of 12 general geographical locations under the jurisdiction of the Base Command. These areas include Hadnot Point, Paradise

Point, Berkeley Manor/Watkins, Midway Park, Tarawa Terrace I and II, Knox Trailer, French Creek,

Courthouse Bay, Onslow Beach, Rifle Range, Camp Geiger, and Montford Point. Table 1-1 lists the

acreage in each geographical area of different types of land utilization (e.g., training, operations, storage, administration, etc.).

1.5 Topography

Elevations on the Base vary from sea level to 72 feet above the National Geodetic Vertical Datum of

1929 (hereafter referred to as “mean sea level” or msl); however, most of MCB, Camp Lejeune is between 20 and 40 feet above msl. Drainage at MCB, Camp Lejeune is generally toward the New

River, except in areas near the coast where flow is into the Intracoastal Waterway that lies between the mainland and barrier islands. In developed areas of the facility, natural drainage has been altered by asphalt cover (i.e., roadway and parking areas), storm sewers, and drainage ditches. Approximately

70 percent of MCB, Camp Lejeune is comprised of broad, flat interstream areas with poor drainage

(WAR, 1983).

1.6 Surface Water Hydrology

The dominant surface water feature at MCB, Camp Lejeune is the New River. It receives drainage from a majority of the Base. The New River is short, with a course of approximately 50 miles on the central Coastal Plain of North Carolina. Upstream from Camp Lejeune and over most of its length, the

New River is confined to a relatively narrow channel in Eocene and Oligocene limestones. South of

Jacksonville, the river widens dramatically as it flows across less resistant sands, clays, and marls. At

MCB, Camp Lejeune, the New River flows in a southerly direction into the Atlantic Ocean through the

New River Inlet. Several small coastal creeks drain the area of Camp Lejeune not associated with the

New River and its tributaries. These creeks flow into the Intracoastal Waterway, which is connected to the Atlantic Ocean by Bear Inlet, Brown’s Inlet, and the New River Inlet. The New River, the

Intracoastal Waterway, and the Atlantic Ocean converge at the New River Inlet.

SECTION 1.0 TABLES

SECTION 1.0 FIGURES

2.0

PREVIOUS INVESTIGATIONS AND COMPUTER SIMULATIONS

Over 160 IRP/UST investigations have been conducted regarding the hydrogeological characteristics of the subsurface at and near MCB, Camp Lejeune. The geology and hydrogeology of the region and of the area adjacent to Camp Lejeune has been described by the USGS in several recent reports from

1989 to 1993. Of particular pertinence to this effort were the publications by Cardinell et al (1993),

Geise et al (1991), Winner and Coble (1989), and Harned et al (1989). On-site investigative activities conducted by Baker and other firms have added to the existing data with regard to the near-surface geology and hydrogeology.

At least three groundwater flow models have been constructed of the region encompassing the Base.

At the outset of this effort it was thought that one or more of the existing regional groundwater flow models may be adapted for use on a smaller scale. The Regional Aquifer System Analysis (RASA) program generated two regional groundwater flow models and the North Carolina Geological Survey created a model of the Central Coastal Plain (CCP). These three existing models were examined and it was subsequently determined that the they were too large in scale and not detailed enough to yield meaningful results for use at MCB, Camp Lejeune. A brief description of each (and the reasons for its unsuitability to the task at hand) follows:

!

One of the RASA models (Leahy & Martin, 1993) encompassed the entire area of the

Northern Atlantic Coastal Plain (from Long Island, New York to North Carolina). The scale of this model was much too large to be adapted for use at such a comparatively small area like MCB, Camp Lejeune (which fit into one of the cells of the RASA model’s finite difference grid).

!

Another RASA model (Geise, G.L., J.L. Eimers, & R.W. Coble, 1991) covered only the Atlantic Coastal Plain in North Carolina. Unfortunately, the scale of this model was also too large to be used directly because the area of MCB, Camp Lejeune took up only 4 cells wide by 4 cells long in the model grid. However, the inputs to this model were used extensively as background information for the BRAGS model at

MCB, Camp Lejeune.

!

The CCP model (Eimers, J.L., W.L. Lyke, & A.R. Brockman, 1990) covered a smaller region within North Carolina (MCB, Camp Lejeune was encompassed by k:\62470\140phase\finaldoc\report\finlmodl.wpd

2-1

April 20, 1998 version

approximately 10 cells wide by 10 cells long). However, the CCP modeled only the

Cretaceous Peedee Aquifer and below. The surficial unit and the Castle Hayne

Aquifer were not specifically modeled. It is possible that the current modeling effort could be used to generate input parameters in a version of the CCP model for the surficial unit and the Castle Hayne Aquifer. That, however, is beyond the scope of this modeling effort, but may be of interest to water management officials in the future.

None of these existing groundwater flow models dealt with either the surficial unit or the Castle Hayne

Aquifer in a meaningful manner over the area of interest. Because these two potentially vulnerable hydrologic units were of paramount importance in this study, a new model was deemed necessary.

k:\62470\140phase\finaldoc\report\finlmodl.wpd

2-2

April 20, 1998 version

3.0

HYDROGEOLOGY OF THE CAMP LEJEUNE AREA

3.1

Physiography

MCB, Camp Lejeune lies within the Tidewater region of the Atlantic Coastal Plain physiographic province (Stuckey, 1965). The Atlantic Coastal Plain is an eastward-thickening wedge of sediments lying atop the basement of Precambrian bedrock. This wedge varies from a thickness of zero near the

Fall Line to more than 10,000 feet near and under the Atlantic Ocean (Winner & Coble, 1989). The

Tidewater region is the portion of the Atlantic Coastal Plain that is influenced by diurnal ocean tides and is generally low-lying, swampy terrain with elevations ranging from sea level to about 50 ft.

3.2

Geologic and Hydrogeologic Framework

Beneath Camp Lejeune are seven water-bearing units, each comprised of one or more formations: an unnamed surficial unit of recent and Pleistocene age, the Castle Hayne Aquifer of Oligocene and

Eocene age, the Beaufort aquifer of Paleocene age, and four Upper Cretaceous aquifers (the Peedee,

Black Creek, and the Upper and Lower Cape Fear). For practical purposes, the surficial unit is not considered an "aquifer" since it cannot yield sufficient amounts of water even for domestic use. This limitation of its use is probably due to its small thickness (which limits available drawdown) near

Camp Lejeune. The underlying hydrologic units are much thicker and are capable of yielding adequate supplies of water; therefore, the underlying units can be practically considered "aquifers" and are referred to as such in this report.

Each of the six aquifers mentioned above provide drinking water to many industries, municipalities, and private well owners throughout the eastern Carolinas and have been described in detail by many authors including Cardinell et al (1993), Trapp (1992), and Eimers et al (1990). The surficial unit and the Castle Hayne Aquifer were the only hydrologic units modeled in this effort because: 1) the contaminants beneath MCB, Camp Lejeune are either in the surficial unit or in the Castle Hayne

Aquifer; 2) only the Castle Hayne Aquifer provides the drinking water for the Base; and 3) the underlying aquifers are over 400 feet deep and effectively isolated by the Beaufort confining unit. The other five aquifers were not modeled in this effort and are not discussed further here.

According to the data collected by Baker during site-specific RI studies, the surficial unit consists mainly of a fine sand with silt, although medium-grained sand occurs to a lesser extent. Across the k:\62470\140phase\finaldoc\report\finlmodl.wpd

3-1

April 20, 1998 version

Base, the thickness of the surficial unit ranges from 0 to 73 feet. These deposits are undifferentiated

Pleistocene and recent sediments. Also, sand beds of the Belgrade Formation of Miocene age are considered part of the surficial unit (Cardinell et al, 1993). The bottom of the surficial unit is at or near mean sea level over most of the Base.

The Castle Hayne confining unit underlies the surficial unit and overlies the Castle Hayne Aquifer. It is comprised of clay and/or sandy clay from one or more of the following lithologic units: the lower portion of the Miocene Belgrade Formation, the upper portion of the Oligocene River Bend Formation, or the upper portion of the Eocene Castle Hayne Formation (Cardinell et al, 1993). The thickness of this confining unit averages about nine feet near Camp Lejeune and has been breached by the New

River and some of its larger tributaries. This observation is one of the keys to understanding groundwater flow near the Base: the localized absence of the confining unit near the New River (or a large tributary) allows a strong hydraulic communication between the surficial unit and the Castle

Hayne Aquifer. Cardinell et al (1993) graphically contoured the thickness of the Castle Hayne

confining unit ( Figure 3-1 ).

In contrast to the classification of lithologic units, the classification of hydrologic units or aquifers depends only on hydraulic conductivity. There must be a distinction between the two types of classification: the silty sand bed below the Castle Hayne confining clay may be considered lithologically as part of the Belgrade Formation, but hydrologically it belongs to the same hydrologic unit as the Castle Hayne Aquifer. The BRAGS conceptual model used only distinctions in hydraulic conductivity to define layers.

The Castle Hayne Aquifer lies beneath the Castle Hayne confining unit and consists of the lower portions of the Oligocene River Bend Formation and the Eocene Castle Hayne Limestone. In the vicinity of Camp Lejeune, the Castle Hayne Aquifer consists mainly of fine sand, shell rock and limestone. The upper portions of the aquifer consist of calcareous sand with discontinuous silt and clay beds (most likely the River Bend Formation). The calcareous sand becomes more limy with depth

(Cardinell et al, 1993). At Site 73, two conspicuous layers of indurated (and subsequently fractured) fossiliferous limestone occur at elevations of approximately -30 to -50, and -80 to -100 feet referenced to mean sea level (msl) (see boring logs for Site 73 RI Report, 1997). The limestone layers may indicate the top of the Castle Hayne Limestone and seem to be the most productive subunit of the Castle

Hayne Aquifer as evidenced by the screened intervals of the Courthouse Bay supply wells (see Table

4-1 ). Harned et al (1989) constructed a typical figure of water supply wells at the base from data

collected from water supply wells and USGS test wells; the figure shows a single 30-foot thick layer of limestone at elevations from -75 to -105 msl. While the thickness of the limestone layers may not be consistent, the presence of fossiliferous limestone seems to be laterally continuous across the Base.

In the vicinity of Camp Lejeune, the Castle Hayne confining unit and the upper Castle Hayne Aquifer have been incised by the meandering of the New River in ages past. Cardinell and others (1993) graphically contoured the top of the Castle Hayne Aquifer and a buried channel presumably created by

the New River is evident in the southern half of Figure 3-2 . This buried channel is significant in that

it suggests that the Castle Hayne confining clay is breached near Courthouse Bay. This is, in fact, what was found in borings at Site 73, adjacent to Courthouse Bay (Baker, 1996b). This connection would provide hydraulic communication between the surficial unit and the Castle Hayne Aquifer and possibly allow uninhibited contaminant migration into the Castle Hayne Aquifer.

The bottom of the Castle Hayne dips to the east across the Base at an average gradient 0.004 feet/foot

(ft/ft), (Cardinell et al, 1993).

3.3 Conceptual Model of Groundwater Flow

Wilder and others (1978) calculated an overall hydrologic budget for a typical location in the eastern

Coastal Plain in North Carolina (see Figure 3-3 ): precipitation averages about 50 inches/year; five

inches/year is lost to surface runoff; 34 inches/year is lost due to evaporation and plant transpiration.

Total recharge to the water table is then about 11 inches/year. Of this amount, about 10 inches/year is discharged to surface water bodies as base stream flow. The remaining one inch/year leaks into the lower units (e.g., the Castle Hayne Aquifer and underlying units). Other estimates of regional recharge to the water table range from 12 to 20 inches/year (Geise et al, 1991) and also 15 to 22.5 inches/year

(Leahy & Martin, 1993). However, even in studies where there were higher estimates of recharge to the water table, the estimates of vertical seepage to the underlying aquifers remained at 1 inch (Geise et al, 1991 and Eimers et al, 1994).

Precipitation falling on the upland areas of the eastern Coastal Plain generally moves vertically downward and generally flows horizontally toward the nearest groundwater discharge area: stream,

river, bay, etc. (see Figure 3-4 ). As groundwater approaches the nearest discharge point (e.g. a stream

or river), it may encounter a low hydraulic conductivity units (silt or clay) in which leakage through the layer is predominantly vertical. Near the discharge area, the head in the surficial water-bearing zones

is reduced by the change in the surface relief at the surface water body; however, the pressure in the deeper aquifers remains higher than that in the surface water body. In the immediate vicinity of the discharge area, the particle responds to the vertical gradient in the deeper aquifers and moves vertically upward to the surface water body. The resulting flow path of a “typical” particle of groundwater in three-dimensions would therefore result in a curvilinear path from the recharge area to the discharge area.

Most of the precipitation falling in the middle of the Coastal Plain generally does not flow very far vertically, but flows horizontally to the nearest groundwater discharge area: stream, river, bay, etc. (see

Figure 3-5 ); however, some of the precipitation infiltrating into the upland (recharge) areas (estimated

at about 1"/year) manages to move downward toward the bottom of the unconsolidated sediments in response to the downward vertical head. At some depth, depending on the pressure head, groundwater stops migrating downward and starts to move horizontally toward the east.

As groundwater approaches the ocean, the pressure head in the upper aquifers is reduced by the change in the surface relief (ultimately to sea level). The pressure in the deeper aquifers beneath the coast remains higher than sea level and the vertical gradient in the deeper aquifers becomes vertically upward. The fresh groundwater then flows vertically toward the surface in the vicinity of the freshwater-saltwater interface near the ocean.

This freshwater-saltwater interface is the area where the saltwater from the ocean is at equal pressure with the freshwater from the land. As such it represents a no-flow boundary that is relatively stable in position (unless hydraulic stress such as pumping is introduced on either side of the interface). The fresh water then moves upward toward the coast and parallel to the interface. The pressure head in the aquifers is such that the groundwater is forced to the surface through the confining layers. The travel time for a pathway as described here (from the Fall Line to the shore) would be on the order of centuries or millennia. The rivers, lakes, and streams along the coast are the ultimate discharge points for fresh groundwater in the Atlantic Coastal Plain aquifers.

The natural groundwater discharge areas around Camp Lejeune are the New River and all of its tributaries (including swamps, wetlands, and streams) and the Atlantic Ocean. Most of these are at or very near mean sea level. Anthropogenic (man-made) discharges include a system of over 100 water supply wells in the Castle Hayne Aquifer at MCB, Camp Lejeune. In 1993, 68 of those wells pumped an average of almost 7 million gallons per day, according to information supplied by the Base Water

Department. Some of the wells have been taken off-line and/or decommissioned because of high levels of organic contamination (e.g., HP-651), others due to poor well performance.

3.4 Hydraulic Characteristics

3.4.1 Surficial Unit

The hydraulic conductivity of the surficial unit has been measured by slug and pumping tests conducted during various RI and UST investigations. The average of the pumping and the slug testing in the surficial unit at IRP Sites 73 and 82 was 3.0 feet per day (ft/d). These data are presented in

Table 3-1 .

The procedures and results of the shallow pumping test at Site 82 are discussed in Appendix A.

3.4.2 Castle Hayne Confining Unit

Between the surficial unit and the Castle Hayne Aquifer lies the Castle Hayne confining unit. Leakance of an aquitard (e.g., a clay and/or silt confining unit) is defined as the vertical hydraulic conductivity which is within the stated range of Trapp.

3.4.3 Castle Hayne Aquifer

Several pumping tests were performed in deep wells in various locations around the Base: DRW-1

(Site 82 by Baker/OHM), supply well HP-642 (ES&E Inc.), supply well HP-708 (USGS), and test well

X24s2x (NC DENR). The results of these tests indicated that the average hydraulic conductivity of the

Castle Hayne Aquifer is very similar to that of the surficial unit with values averaging 2.85 ft/d (1x 10

-3

Castle Hayne hydraulic conductivity data from various sites and other hydrogeologic studies are

summarized in Table 3-2 . The previous studies by the USGS, NC DENR and ESE (Cardinell et al,

1993) resulted in values of hydraulic conductivity ranging from 2.3 ft/d to 4.9 ft/d, using values for saturated thickness of 308 to 382 feet.

These hydraulic conductivity values are indicative of fine sand and/or silty sand (Heath, 1983). In contrast, several USGS papers have been published that estimate the regional hydraulic conductivity of the Castle Hayne Aquifer in North Carolina as being one or more orders of magnitude greater than the site-specific values stated above (e.g., an estimated average of 65 ft/d, Winner & Coble, 1989). The highly permeable and relatively thin (10-20 feet thick) layers of indurated and fractured limestone within the Castle Hayne may be the reason for such high conductivity value estimates. When a highly permeable layer is tested via pumping (as the USGS did), the resulting transmissivity value is measured directly, independent of the unit’s thickness. The calculation of the hydraulic conductivity value depends upon the interpretation of the thickness of the unit being tested. This may explain the apparent difference between the two sets of hydraulic conductivity data: a single transmissivity value divided by a large thickness (i.e., the entire thickness of the Castle Hayne Aquifer) would yield a lower hydraulic conductivity than for a thinner (limestone) layer. This modeling effort assumed an average thickness of 350 feet for the entire Castle Hayne Aquifer.

Another possible explanation for the difference between the regional and site-specific data could be the natural variations in hydraulic conductivity that can result from different depositional facies within the same chronostratigraphic unit, or perhaps post-depositional reworking by fluvial and/or tidal action.

The large fraction of fine sand and silt in the upper portion of the Castle Hayne near MCB, Camp

Lejeune indicates a relatively low to medium energy, shallow water environment of deposition.

k:\62470\140phase\finaldoc\report\finlmodl.wpd

3-6

April 20, 1998 version

SECTION 3.0 TABLES

SECTION 3.0 FIGURES

4.0

BRAGS GROUNDWATER FLOW MODEL

The groundwater flow regime beneath MCB, Camp Lejeune was simulated by using the model code

MODFLOW (McDonald & Harbaugh, 1988), a numerical groundwater flow code initially developed by the USGS and modified to run on IBM-compatible computers. This code was chosen because it has been extensively tested and documented in many applications and was appropriate for this complex, three-dimensional groundwater flow system.

The simplified governing (partial differential) equation used by the numerical model (MODFLOW) is:

δ

(K xx

δ h/

δ x)/

δ x +

δ

(K yy

δ h/

δ y)/

δ y +

δ

(K zz

δ h/

δ z)/

δ z - W = S s

δ h/

δ t where:

!

!

!

!

!

!

x, y, and z are Cartesian coordinates aligned with the major axes of hydraulic conductivity h is the potentiometric head or water table elevation

W is a volumetric flux per unit volume of aquifer and represents sources and/or sinks of water t is time

This equation describes the movement of water through a porous medium. For a steady-state model such as this, the right side of the equation becomes zero because the change in head with time is assumed to be zero. Together with the specification of initial and boundary conditions, this equation constitutes a mathematical model of groundwater flow.

MODFLOW can accommodate confined or unconfined conditions and uses input parameters of hydraulic conductivity, aquifer thickness, recharge, evapotranspiration, porosity, storativity, and specific yield to calculate water levels at various locations within the model boundaries. Each of the inputs can be varied spatially across the model grid so that by changing the parameters, a match to actual field conditions can be accomplished. k:\62470\140phase\finaldoc\report\finlmodl.wpd

4-1

April 20, 1998 version

In order to use MODFLOW, it was necessary to discretize the domain into cells, each of which had a

"node" containing the properties (e.g., hydraulic conductivity) and/or boundary conditions (e.g., rivers, wells) that approximated the conditions found at the site. For example, a well was reproduced by a

"well" cell that specifies the flow into (recharge) or out of (discharge) the cell. Similarly, rivers, streams, swamps, and no flow boundaries were reproducible by one of the internal boundary cell types within MODFLOW.

Initially, the BRAGS model was to represent the seven aquifers beneath the Base: the surficial unit, the Castle Hayne Aquifer, the Beaufort aquifer, and four Upper Cretaceous aquifers (the Peedee, Black

Creek, and the Upper and Lower Cape Fear); however, it was determined from the initial model runs that the aquifers below the Castle Hayne were not noticeably affected by changes to the top two layers.

Layers representing the Beaufort aquifer and below were subsequently removed from the modeling process. In addition, little to no water level data from these units were available beneath the Base; therefore, the layers representing these aquifers could not be calibrated. This change improved the performance of the model and reduced the necessary memory requirements for its continual use.

Electronic model input and output for the BRAGS model can be found on CD-ROM in Appendix B.

4.1 Finite-Difference Layered Grid

The finite-difference grid superimposed over the subject area has a uniform spacing: 1,000 feet by 1,000

feet square cells (see Figure 4-1 ). The grid has 101 rows (about 19 miles north to south) and 80

columns (about 15 miles east to west) over an area of approximately 285 square miles. The outer limits of the grid were chosen to be far enough away from the area(s) of pumping at the Base such that the boundaries of the grid would not interfere with any drawdown cones generated by pumping wells. Such interference would artificially increase or decrease the simulated drawdown, depending on the type of boundary being affected.

The fully-3D groundwater flow model consists of five layers (see Figure 4-2 ). From top to bottom they

represent the surficial unit (layer 1), the Castle Hayne confining unit (layer 2), and three separate layers representing the upper (layer 3), middle (layer 4), and lower (layer 5) portions of the Castle Hayne

Aquifer. The Castle Hayne Aquifer was divided into three portions because the “deep” monitoring well data represented only the upper portion of the Castle Hayne Aquifer. The water supply wells were generally screened in the middle of the Castle Hayne Aquifer where the fractured limestone occurs.

Therefore, the “deep” monitoring well target data was put into layer 3 and the supply well cells were

put into the more permeable layer 4 (representing the limestone layer). Water elevation targets for the water supply wells were placed into layer 5 since the wells are also screened below the limestone.

4.2

Model Boundary Conditions

Boundaries in MODFLOW include external and internal boundaries. External boundaries can include specified (constant) head or general head boundary cells. General head boundary cells were used around the perimeter of the model to simulate the regional gradients of ambient groundwater flow.

Internal boundaries include well, river, stream, and drain cells. For the BRAGS model, no stream cells were used. Internal boundaries were well, river and/or drain cells. Some (8%) of the cells in the model were inactive (no-flow) due to their location in the Atlantic Ocean.

Any cell in which the water elevation does not change appreciably over time such as the Atlantic Ocean were assigned specified (constant) head cells. There is no input of bottom elevations or conductances to this type of cell. The cell recharges or discharges to/from as much volume as the aquifer needs to keep the water elevation constant.

Any cell in which the elevation of the water and the bottom surface of the water body was used to simulate the surface water to groundwater interaction (recharge or discharge) which also had a relatively constant elevation such as the New River was simulated using river cells.

The remaining streams and swamps that were presumed only to receive groundwater runoff from the surficial unit were simulated by drain cells which are designed only to remove water from the groundwater system based on the elevation differences between the drain and the surrounding water table. Drain cells do not recharge groundwater.

4.2.1

Specified (Constant) Head Cells

The ocean was simulated by specified heads of zero (sea level) along the shore and no-flow cells further east. This arrangement presumes that all groundwater is discharging to the surface just west of (and along) the shoreline. It also presumes no east-west movement of the saltwater-freshwater interface along the shore.

4.2.2

General Head Boundary Cells

k:\62470\140phase\finaldoc\report\finlmodl.wpd

4-3

April 20, 1998 version

General head boundary cells are head-dependant flow cells that allow flow into or out of the cell depending on two things: 1) the head differential between the assigned value and that in the surrounding aquifer and, 2) an assigned constant of proportionality. In the BRAGS model, the assigned value of head represents a theoretical head value (e.g., of a surface water body) at some distance beyond the model boundary and the proportionality constant represents the hydraulic conductivity of the aquifer between the model boundary and the “theoretical” surface water body.

Four of the five layers had general head boundary cells placed along the outer boundaries to simulate the ambient groundwater gradient. The values of head assigned to each cell were chosen to represent the gradient of regional groundwater flow (0.000025 to the east in layer 1 and 0.0000125 to the east in the underlying layers, estimated from Geise et al, 1991). The proportionality constants were adjusted by trial and error during the calibration process until a reasonable fit was achieved at the boundaries.

4.2.3

Well Cells

Wells cells are specified (constant) flux boundaries which keep a constant flow rate throughout the specified time period. Positive values recharge to groundwater and negative values discharge from groundwater. MODFLOW assumes that each well fully penetrates the layer in which it is placed.

These cells were placed at the locations of the water supply wells and assigned negative (discharge) pumping rates in cubic feet per day. All available well locations were plotted even if they were turned off. This will help in the future if they are turned on again. The NC state planar coordinates (NAD

1983) of the water supply wells were converted from the latitude and longitude as recorded in Cardinell et al (1993).

The average pumping rates of the supply wells were calculated from 1993 total pumping data supplied by the MCB, Camp Lejeune, Base Water Department. Since the water supply wells are turned off at night, it was necessary to estimate the fraction of time the wells were pumped each day. This was done by taking the total gallons pumped from each well per year (P, gallons/year) and dividing it by 365 days/yr. This number was the average gallons pumped per day (p, gpd): p = P/365 k:\62470\140phase\finaldoc\report\finlmodl.wpd

4-4

April 20, 1998 version

Then the maximum measured pumping rate (r, gallons/minute, or gpm) for each well was multiplied by 1,440 minutes per day to get the theoretical maximum daily rate (R, gpd) as if the well had been pumping day and night:

R = r x 1,440

Next, p was divided by R to get the fraction of time the well was actually pumping per day (f, unitless): f = p / R

For example, HP-603 pumped a total of 27,586,860 gallons in 1993. It’s maximum pumping rate was

150 gpm (it has since been removed from service). The average daily rate was 27,586,860 gallons/365days = 75,580 gallons/day. The theoretical maximum that HP-603 could produce in one day was: 150 gpm x 1,440 min/day = 216,000 gallons/day. Assuming that when the well was on (pumping at its maximum rate), the fraction of time that the well was on-line was the ratio of average/maximum

(75,580 gpd/216,000 gpd = 0.349) or 35% of the time. That would have been about 8 hours of pumping

every day. This value varied from well to well. Table 4-1 presents the average daily pumping rates

that were calculated for every well in gallons per minute and cubic feet per day.

4.2.4 River Cells

River cells are head-dependant flow cells in which the elevations of the surface water and river bottom are held constant (at surveyed or mapped elevations) and the thickness and conductance of the sediments control the flow rate of water to or from the cell. If the stream or pond level is higher than the surrounding groundwater, the river cell allows water to recharge the groundwater. Conversely, if the water level in the stream or pond is lower than the groundwater, the groundwater discharges to the surface water body.

where:

K = hydraulic conductivity of the river sediments;

L = length of the river in each cell;

W = width of the river in each cell; and,

M = thickness of the river sediments.

4.2.5

Drain Cells

Drain cells function similarly to river cells except that they cannot recharge the groundwater when the ambient water table drops below the drain elevation. Streams and swamps were represented by drain cells because it was reasonably assumed that they only receive groundwater discharge and were not recharging groundwater. In low-lying swamps and wetlands where the elevation of the water is lower than the surrounding water table, this assumption is reasonable as the wetlands would be receiving discharged water most of the year. However, in cases where there are wetlands atop hills where water is ponding above the water table, drain cells may not be the best representation; river cells may be better to provide a source of ponded water in this case.

4.3

Steady-State Modeling Process

In a steady-state groundwater flow model all values of drawdown are assumed to have reached equilibrium.

That is, enough time is supposed to have passed with the wells pumping at constant rates that no additional drawdown is occurring. While rarely true in reality, this assumption can be considered valid when applied over the long term (years or decades) to understand how groundwater flows within the regime. The most important assumption of this approach is that the diurnal pumping schedule of the water supply wells has been averaged as if pumping were a continuous event.

In general, the extent to which the model assumptions match the actual subsurface conditions dictates the accuracy of any subsequent predictions. In order to get a realistic model, it was prudent to calibrate the model to match actual measured values of head. The "targets" of the calibration should be based on the statistics of the historical water level data where possible.

4.3.1

Calibration Targets

Water elevations measured at 21 IRP sites and one UST site around the Base during 1992 and 1993 provided “target” data for the BRAGS model. At these sites, the well locations were in such close proximity to each other that an average water level was calculated for use as the site target in the BRAGS k:\62470\140phase\finaldoc\report\finlmodl.wpd

4-6

April 20, 1998 version

groundwater flow model. This limited the number of targets to a manageable size (142 head targets in layers 1, 3, and 5).

From the average water levels in the shallow wells, 23 targets were established to which the surficial unit

(layer 1) would be calibrated. In layer 3 (the upper portion of the Castle Hayne) there were 29 targets representing the "deep" and "intermediate" water level data. Layer 5 (the lower portion of the Castle

Hayne Aquifer) contained 90 targets developed from the water level data from the supply wells. Because the water levels in the supply wells had been collected by the USGS over many years and in different ways, there was no way of knowing whether the water level data represented static conditions in these pumping wells at the time of collection. Therefore, more credence was given to the data in layers 1 and 3 than those in layer 5; therefore, the calibrations in layers 1 and 3 were deemed more accurate than that in the bottom layer.

4.3.2

Calibration Methods

The calibration process used the "trial and error" method in which the results of each run were examined statistically to determine the degree of "fit" of the results. No comprehensive groundwater contour maps exist for the Base; the only such maps were those pieced together by Harned et al (1989) and those were spaced rather far apart. Statistics used were mean error (ME), mean absolute error (MAE), standard deviation of the errors (SDE), and the root mean square error (RMSE). After each run, one or more input values (e.g., horizontal and/or vertical hydraulic conductivity) and/or their spatial distributions were changed and the model rerun. Changes were made to various parameters in those areas of the grid where the error between simulated and measured water levels was large. In this "trial and error" method, not all changes were for the better, some had to be changed many times to find a "better" value or distribution.

This process continued until a reasonable fit was achieved.

The definition of “reasonable fit” is relative and depends upon many things including the amount of fluctuation in naturally-occurring water levels and upon the reliability of data collection methods. For the

Marine Corps Air Station (MCAS), Cherry Point groundwater flow model (Eimers et al, 1994), statistics were also used to judge the adequacy of “fit.” As discussed below, this effort had comparable statistics to those used at MCAS, Cherry Point.

4.3.3

Statistical Evaluation of Calibration

k:\62470\140phase\finaldoc\report\finlmodl.wpd

4-7

April 20, 1998 version

The difference between a measured value (or target) and a simulated value is called an error. The average of the all the errors should be close to zero for an accurate model because that indicates that the errors higher than the targets are balanced by the errors below the targets. A highly positive or negative mean error (ME) would indicate an inaccurate model in which the water levels are all too low or too high, respectively. The ME is a good indication of accuracy but not of precision (or data dispersion). As long as the errors were balanced a model can be considered accurate, but the fit is better measured by the more useful indicators of precision. These include the mean absolute error, MAE, the standard deviation of the errors, SDE, and the root mean square error, RMSE. These statistical values are better indicators of fit than the ME alone because smaller values indicate less dispersion from the targets and that the simulated values collectively match the targets more closely (see introductory statistical reference such as Dixon &

Massey, 1983).

4.4 Calibrated Results of Simulation

This section describes the inputs and outputs of the calibrated simulation. Unless otherwise indicated, the values used in the model were taken directly from the values discussed in the previous sections.

Table 4-2

presents the errors of each target in the BRAGS model. The end of the table shows the statistics

for each layer as well as for the entire model. The final ME value for all three layers in the BRAGS model was -0.85 feet, which indicates that the average of all the simulated water levels in all three layers was

0.85 feet lower than the measured water levels. The MAE for all three layers was 4.46 feet. The SDE was

6.10 feet, and the RMSE was 6.14 for the final calibration.

The comparable statistics between Camp Lejeune (three targeted layers) and MCAS, Cherry Point (six targeted layers) are as follows:

MCAS, Cherry Point MCB, Camp Lejeune

Mean Error (ME) =

(Eimers et al, 1994) (Baker, current report)

0.20 feet -0.85 feet

Mean Absolute Error (MAE) = 4.35 feet 4.46 feet

Standard Deviation of Errors (SDE) = 5.70 feet 6.10 feet

Root Mean Square Error (RMSE) = 5.70 feet 6.14 feet

Range of Errors = ± 17 feet ± 23 feet

Range of Observed Elevations (R) = 45 [-8 to 37 feet msl] 43 [-5 to 38 feet msl]

SDE/R = 13% 14%

These statistical values of the two models compare favorably; the MAE, RMS, and SDE values indicate that the simulated values at Camp Lejeune are as close to the observed targets as are those at MCAS,

Cherry Point. The ratio of SDE to R can be used as an indication of adequate calibration. The ratios suggest that the two models are calibrated to an equivalent level.

The range of errors may seem rather large for both models; however, for the Camp Lejeune BRAGS model, the largest errors occurred at locations having less than completely reliable measured data (i.e., in water supply wells with limited and sometimes dubious available data). With one exception, all of the errors in the surficial unit and upper Castle Hayne Aquifer (layers 1 and 3) are within ±10 feet of their targets; most of the error in layers 1 and 3 is associated with one data location (Site 69). The significance of this will be discussed below. In layer 5, all the modeled heads are within ±20 feet of the targets.

Figure 4-3 shows the graph of modeled head values as a function of the observed target head values. With

one exception (Site 69), the heads in the three targeted layers (1, 3, and 5) are within ±10 feet of the established targets.

Figure 4-4 shows this more clearly: the target data point at Site 69 was very close to

the New River but was also elevated more than 25 feet above mean sea level (msl). If the data from Site

69 were excluded as targets, the statistics would show that the current groundwater flow model predicts the rest of the data with much better accuracy than the aforementioned numbers indicate: ME = -0.28;

MAE = 2.79; and RMSE = 3.78.

Figures 4-5 and

4-6 show the spatial distribution of the errors in layer 3. Again, with one exception, the

heads are within ±10 feet of the targets. The anomalous value is from Site 69 without which the statistics are much improved: ME = -0.94; MAE = 1.85; and RMSE = 2.57.

Figure 4-7 and

4-8 show that the modeled heads for the lower Castle Hayne Aquifer are within ±20 feet

of their targets. The error for layer 5 is larger than that in layers 1 and 3 due to several factors: 1) water level data from the water supply wells may have been measured before the wells had fully recovered from pumping, 2) even if the supply wells were allowed to fully recover internally before measurement, the influence of nearby active wells would not have allowed truly non-pumping conditions in many wells, 3) the wellheads of the supply wells had not been surveyed for vertical nor for horizontal control; elevations were estimated from topographic maps.

Figures 4-9 and

4-10 show the spatial distribution of error in layer 1.

Figure 4-11 and

4-12 show the

spatial distribution of error in layer 3. Figures 4-13 through

4-16 show the spatial distribution of error in

layer 5. These figures show error bars for each data station: the bars represent ± one standard deviation from the mean (where the data were available); when data were insufficient to produce a standard deviation, a confidence interval of no more than five feet was chosen, depending upon the type of data point.

4.4.1 Layer 1 -- Surficial Unit

4.4.1.1 Input

A uniform value of recharge of 11 inches per year was used in this model. This value was estimated based on several USGS studies (see section 3.3 for discussion) and was also calibrated to site-specific hydraulic conductivity data. Recharge occurred only in layer 1. Layer 1 was unconfined and bottom elevations in

layer 1 ranged from -70 feet to +10 feet msl (see Figure 4-17 ).

The general head boundary (GHB) cells were set to about 50 feet above mean sea level (msl) in the inland

areas around the Base (see Figure 4-18 ). The conductance of the GHB cells (C ) was set at 2,000 ft /d.

The Atlantic Ocean was represented by a specified (constant) head value of +0 feet msl along the shoreline.

River cells were used to represent the New River and its elevation was assumed to be mean sea level (see

Figure 4-18 ). Using the input value of 5,000 ft /d for river cell conductance (C ) is reasonable assuming

the following parameter values for the New River:

L = 1000 feet (average length of river cell in the New River)

W = 1000 feet (average width of river cell in the New River)

M = 2 feet (estimated thickness of river sediments)

This K value is typical of a silty clay sediment which is reasonable for the bottom of the New River.

The elevations of the drain cells were the approximate elevations of the streams as determined by

topographic mapping of the area (see

Figure 4-18

). Drain cells for streams were assigned a uniform

conductance value (

C drn

=5,000 ft /d). This translates to the following values:

L = 1000 feet (average length of drain cell along streams)

W = 10 feet (estimated average width of streams)

M = 1 feet (estimated thickness of stream sediments)

This K value is typical of silt which would be expected in low energy streams.

Drain cells were also used to simulate three areas of upland swamps by assigning mapped elevations and

L = 1000 feet (average length of drain cell in wetlands)

W = 1000 feet (average width of drain cell in wetlands)

M = 5 feet (estimated thickness of confining clay)

This K value is typical of the vertical hydraulic conductivity of a clay which would underlie a swampy area.

adjusted from the average hydraulic conductivity value of 3 ft/d (from shallow pumping and slug tests at

Sites 73 and 82).

Vertical hydraulic conductivity was assumed to be 0.1 times the value of horizontal hydraulic conductivity average value of vertical hydraulic conductivity (1.7 ft/d) by the Neuman method at Site 82 (see

Appendix

A ). This value may not be indicative of the entire Base as the confining unit is absent at Site 82. Vertical

anisotropy is often unknown and is estimated during calibration. Vertical anisotropy ratios ranging from

1 to 1,000 are common in model application (Anderson & Woessner, 1982).

No well cells were used in the surficial layer of the BRAGS model.

4.4.1.2 Output

Figure 4-19 shows the water table contours across MCB, Camp Lejeune. The map shows that the New

River and its tributaries are the main areas of groundwater discharge from the surficial hydrostratigraphic unit. Between the streams (localized discharge areas) are recharge areas; this was the expected pattern of flow in the surficial unit, based on the conceptual model.

With one exception, the simulated water levels in layer 1 were within 8 feet of their targets. Table 4-2

shows that the ME in layer 1 was -1.08 feet, the MAE was 3.48 feet, the SDE was 5.38 feet, the RMSE was 5.37. As indicated by the value of the SDE for layer 1, about 66% of the simulated heads were within

5.4 feet (one standard deviation) of their targets and 95% of them were within 11 feet (2 standard deviations). These compare favorably to: MAE = 3.87, SDE = 5.0, and RMSE = 5.0 for the surficial unit in the MCAS, Cherry Point groundwater model (Eimers et al, 1994).

Table 4-3 is a summary of the simulated hydrologic budget for the BRAGS model. It shows that, of 11

inches of recharge per year, about 1.7 inches infiltrated into the Castle Hayne Aquifer. This value, while larger that the estimate by Wilder et al (1978) of 1.0 inches per year, is still relatively close to the estimate.

The BRAGS value of deep infiltration may be larger due to the effects of supply well pumping which accounts for 0.6 inches per year on average (5,366 gpm). The cessation of pumping would tend to make the value of deep infiltration about 1.1 inches per year.

4.4.2 Layer 2 - Castle Hayne Confining Unit

Layer 2, representing the Castle Hayne confining unit, had a uniform thickness of 10 feet but varied in depth across the entire Base. The top elevation of layer 2 is the same as the bottom of layer 1. The bottom of the confining unit varied from +0 to -80 feet msl. Layer 2 comprised three values of horizontal hydraulic conductivity: 0.1 ft/d over most of the Base area, 0.00073 ft/d in selected places and 5 ft/d where the clay unit was breached. Vertical hydraulic conductivities was assumed to be 0.1 times the horizontal values.

Figure 4-20 shows the leakance values used in layer 2 (where leakance = K /thickness = K /10 feet).

No GHB, well, river, or drain cells were used in layer 2 of the BRAGS model for MCB, Camp Lejeune because it is assumed that the flow in layer 2 is mostly vertical and no groundwater flows laterally through the boundaries of layer 2.

No river cells were needed in layer 2 to simulate leakage to the New River because the vertical permeabilities in layer 2 were used to simulate the presence or absence of the clay unit. Where no confining unit was present, the vertical hydraulic conductivity of layer 2 was much higher than in areas where a confining unit was indicated. The higher the vertical hydraulic conductivity, the more hydraulic

“communication” between vertically adjacent units. The direction of vertical flow depends on the head differences in the adjacent units. The higher heads in the upper Castle Hayne (layer 3) provided the impetus for the upward leakage to the New River in layer 1.

4.4.3 Layer 3 - Upper Castle Hayne Aquifer

4.4.3.1 Input

Top elevations in layer 3 were identical to the bottom elevations of layer 2. Bottom elevations in layer 3

range from -40 feet to -130 feet msl (see Figure 4-21 ). The bottom of layer 3 is sloping from west to east

across the study area. The GHB cells in layer 3 are set so that a very slight regional gradient (0.0000125, estimated from Geise et al, 1991) is flowing to the east (i.e., 25 feet msl at the western boundary and 24 feet msl at the eastern boundary). The uniform hydraulic conductivity value in layer 3 is 7 ft/d. This is a calibrated adjustment to the average from the pumping tests in the upper Castle Hayne Aquifer (3 ft/d).

Vertical hydraulic conductivity was assumed to be 0.1 times the horizontal value in layer 3 (0.7 ft/d).

No well, river, or drain cells were used in layer 3 of the BRAGS model for MCB, Camp Lejeune.

4.4.3.2 Output

Figure 4-22 shows the piezometric surface contours in the upper Castle Hayne (layer 3) across MCB, Camp

Lejeune. The map shows that the New River and its tributaries are still the main areas of groundwater discharge from the upper Castle Hayne Aquifer. The drawdown from the pumping wells is shown especially near Hadnot Point where the resulting groundwater elevation is near sea level.

All of the simulated water levels in layer 2 (with the exception of Site 69) were within 8 feet of their

targets. Table 4-2 shows that the ME in layer 2 was -1.68 feet, the MAE was 2.57 feet, the SDE was 4.68

feet, and the RMSE was 4.89 feet. As indicated by the value of the SDE for layer 2, about 66% of the simulated heads were within 4.7 feet (one standard deviation) of their targets and 95% of them were within

9.4 feet (2 standard deviations). This match is very similar to that of layer 1. These compare favorably to: MAE = 4.89, SDE = 5.50, and RMSE = 6.30 for the upper Castle Hayne Aquifer in the MCAS, Cherry

Point groundwater model (Eimers et al, 1994).

4.4.4 Layer 4 - Castle Hayne Fractured Limestone Unit

Layer 4, representing the highly conductive Castle Hayne fractured limestone, had a uniform thickness of

10 feet but varied in depth across the entire Base. The top elevations of layer 4 were the same as the bottom elevations of layer 3. The bottom of the confining unit varied from -50 feet to -140 feet msl. Layer

4 had a uniform horizontal hydraulic conductivity of 100 ft/d over the Base area. Vertical hydraulic conductivity was assumed to be 0.1 times the horizontal value (10 ft/d). This value was estimated assuming that a higher hydraulic conductivity unit existed in the lower portions of the Castle Hayne

Aquifer. No direct hydraulic conductivity data exist that can confirm this, but it is consistent with the well logs around the Base and with regional values of hydraulic conductivity in the Castle Hayne Aquifer (Geise et al, 1991; Cardinell et al, 1993; Harned et al, 1989).

The water supply wells were placed into layer 4 of the BRAGS model for two reasons: 1) the high yields of these wells suggest a high conductivity layer and, 2) the well logs for most of the supply wells indicate

that they are screened in one or more fractured limestone layers. Figure 4-23 shows the locations of the

water supply wells around Camp Lejeune. Well cells were installed for all identified well locations around

the Base but only those actually pumping in 1993 were given values of discharge as shown in Table 4-1 .

Discharge rate values for MODFLOW are in negative cubic feet per day.

GHB cells were used in layer 4 as boundaries and were identical to those of layers 3 and 5. No river or drain cells were used in layer 4 of the BRAGS model for MCB, Camp Lejeune.

4.4.5 Layer 5 -- Lower Castle Hayne Aquifer

4.4.5.1 Input

The top elevations of layer 5 are identical to the bottom elevations of layer 4. Bottom elevations in layer

5 range from -180 feet to -480 feet msl ( Figure 4-24 ). The bottom of layer 5 is sloping from west to east

across the study area.

The GHB cell boundaries in layer 5 are set the same as in layers 3 and 4. The regional gradient is flowing to the east. No river or drain cells were used in layer 5.

The uniform hydraulic conductivity value in layer 5 is 10 ft/d. Leakance is inactive in layer 5 as it is the bottom of the model. This has the same effect as a no-flow boundary at the bottom.

4.4.5.2 Output

Figure 4-25 shows the piezometric surface contours in the lower Castle Hayne (layer 5) across MCB, Camp

Lejeune. As in layer 3, the map shows that the New River and its tributaries are the main areas of groundwater discharge from the lower Castle Hayne Aquifer. The drawdown from the pumping wells is shown especially along Brewster Boulevard (near Paradise Point) where the resulting groundwater elevation is near sea level.

Table 4-2 shows that in layer 5, all of the errors were less than 20 feet. Table 4-2 shows that the ME in

layer 5 was -0.52 feet, the MAE was 5.33 feet, the SDE was 6.68 feet, and the RMSE was 6.66 feet. As indicated by the value of the SDE for layer 2, about 66% of the simulated heads were within 6.7 feet (one standard deviation) of their targets and 95% of them were within 13.4 feet (2 standard deviations). These compare favorably to: MAE = 5.45, SDE = 6.50, and RMSE = 6.50 for the lower Castle Hayne Aquifer in the MCAS, Cherry Point groundwater model (Eimers et al, 1994).

4.4.6 Three-Dimensional Analysis of Groundwater Flow

MODFLOW was used in the BRAGS model to generate 3-D flow vectors (directions and relative velocity at discrete points) in a map view and in an west-to-east cross-section, respectively. The length and size

of each arrow represents its relative velocity compared to the other arrows.

Figure 4-26

shows the map

view of the northern portion of the Base in layer 4 (limestone) where the pumping wells are concentrated.

The combined effects of the supply wells can be seen along Brewster Boulevard near Paradise Point where the flow directions have been reversed from the river back toward the wells. In Camp Geiger the effect

is less but is still noticeable as the velocity of the ambient flow (from west to east upgradient of the wells) has been slowed considerably.

As shown on Figure 4-26 , the controlling factors in determining where groundwater flows are the streams:

they provide a differential head that allows deep groundwater to migrate upwards because the heads in the underlying aquifer are much greater than those in the streams. In combination with breaches in the confining clay unit, a vertical “escape route” is provided for deep groundwater. That is why the flow vector arrows are so large in upstream portions of each stream; the head differential is typically much larger there than near the New River. Also noteworthy are the larger flow vectors near the edges and the smaller flow vectors toward the middle of surface water bodies; this is also the result of the changing head differentials beneath the surface water.

Figure 4-27 shows an east-west cross-section between Verona Loop Road and Hadnot Point. It shows that

groundwater flows like that envisioned in the conceptual model: downward in the upland recharge areas

and upwards to discharge into local streams or the New River (see also Figure 3-4 ). The flow in the

surficial unit is mostly vertical and the flow in the lower Castle Hayne (layer 5) is mostly horizontal.

Again, the flow vectors are largest where the head differential is largest, in upstream portions of the tributaries to the New River. The flow vectors discharging to the New River itself are much smaller in magnitude because the head differential is smaller there.

Water supply wells intercept some of this water on its way to the New River but the individual effects are localized around the area near each well. However, where the supply wells are grouped together and their

drawdown “cones” overlap are where the resulting groundwater elevations are lowest. Figures 4-22 and

4-25 show such an area along Brewster Boulevard (near Paradise Point) where the model predicts steady-

state groundwater elevations in the upper and lower Castle Hayne Aquifer at or below sea level. These are the “danger zones” for saltwater intrusion into the Castle Hayne Aquifer. In order to mitigate this situation, pumping wells should be spread out as much as possible to avoid the creation of such zones and to preserve the potable groundwater quality of the Castle Hayne Aquifer.

4.5 Sensitivity Analysis

A sensitivity analysis was performed on the BRAGS model. Selected parameters were changed by ± 20 and 50% and the resulting effects were quantified statistically using the values of ME, MAE, and RMSE.

The following six parameters were used in the sensitivity analysis: recharge, hydraulic conductivity

(horizontal), leakance, general head boundary cell conductance, river cell conductance, and drain cell conductance.

4.5.1 Effects of Altering Recharge

Figure 4-28 shows the effect of changing the value of recharge from the calibrated input value (11

inches/year). The model is very sensitive to recharge, but no significant improvements were found based on changes to recharge input values. While a 15% increase in recharge would make the ME closer to zero, the values of RMSE and MAE would increase slightly.

4.5.2 Effects of Altering Horizontal Hydraulic Conductivity

Figure 4-29 shows the effect of changes to horizontal hydraulic conductivity, K. During the analysis,

values of K were changed in every layer at once. While decreases in the K values reduced the value of

ME, they produced unacceptable increases in RMSE and MAE.

4.5.3 Effects of Altering Leakance

Figure 4-30 shows the effect of changes to leakance, which is vertical hydraulic conductivity, K , divided

by the unit thickness, b. During the analysis, values of leakance were changed in every layer at once.

Decreases in leakance values reduced the value of ME toward zero and slightly improved the values of

RMSE and MAE. It is possible that a reduction in the values of leakance may slightly improve the model.

4.5.4 Effects of Altering GHB Cell Conductance

The values of ME, RMS and MAE were not noticeably affected by changes in the conductance of the GHB

cells (see Figure 4-31 ).

4.5.5 Effects of Altering River Cell Conductance

The values of ME, RMSE and MAE were slightly improved by reductions in the conductance of the river

cells, but the changes were not significant (see Figure 4-32 ).

4.5.6 Effects of Altering Drain Cell Conductance

Figure 4-33 shows that the value of ME was slightly improved by decreasing the conductance of the drain

cells by 50%. The values of RMSE and MAE were also slightly reduced. Therefore, a change in the conductance of drain cells may be recommended, depending upon the effects of other recommended changes.

4.5.7 Recommended Changes to the Model

Figures 4-34 through

4-36

show the relative sensitivity of the model to the six parameters. Figure 4-34

shows the response of ME to the changes: in descending order, the ME is most sensitive to recharge, hydraulic conductivity, leakance, drain conductance, and river conductance. The ME was not sensitive to the changes in GHB cell conductance. The ME is most sensitive to changes in recharge; although improvements were obtained by increasing recharge by about 15%, such a change is not recommended because it did not improve the other statistics (RMSE and MAE). Reductions of hydraulic conductivity in all five layers by 50% produced improvements in the value of ME; again however, such a change is not recommended because it did not improve the other statistics (RMSE and MAE). Reductions in drain cell conductances and leakance (50%) seemed to improve the ME. A 50% reduction of river cell conductance also slightly improved the ME.

Figure 4-35 shows that a 50% decrease in the values of leakance, drain cell conductance, and river cell

conductance would improve the RMSE. On this figure it is shown that a 20% increase in K would slightly improve the RMSE. No improvements were noted by changing the calibrated value of recharge. The

RMSE was not sensitive to changes in GHB cell conductance.

Figure 4-36 shows that no significant improvements to the value of MAE are obtained by changing the six

calibrated input values.

In summary, the model is most sensitive to (in decreasing order):

1. Recharge

2. Hydraulic Conductivity (horizontal)

3. Leakance

4. Drain Cell Conductance

5. River Cell Conductance

The model was not sensitive to changes made in the conductance of the general head boundary (GHB) cells.

Recommended changes are to decrease leakance, drain cell conductance and river cell conductance values by at least 50% to lower RMSE and MAE values and to achieve an ME value close to zero. Changes to recharge, hydraulic conductivity, and GHB cell conductance are not recommended.

4.6

BRAGS Groundwater Flow Model Summary

The BRAGS groundwater flow model presented herein portrays the three-dimensional pattern of groundwater flow within the surficial units and the Castle Hayne Aquifer. It achieves the first two of the three objectives described in Section 1.1:

!

Based on the conceptual model described in Section 2.0, the model describes how groundwater flows in three dimensions beneath the entire Base as well as under individual sites (Objective 1). The model reasonably predicts the elevation and flow direction of the groundwater in many areas around the Base where no data currently exist.

!

The model demonstrates the effects of groundwater supply withdrawals on the surficial unit and the Castle Hayne Aquifer (Objective 2). The model demonstrates that discharge to the New River and its tributaries is the controlling factor on groundwater flow directions in the Castle Hayne Aquifer in the vicinity of Camp Lejeune. The model output indicates that the relatively high-volume withdrawal rates of the supply wells have a localized effect on the water levels in the Castle Hayne; however, large numbers of actively pumping wells in small areas have the potential to induce saltwater intrusion into the upper Castle Hayne Aquifer. This effect is most pronounced in Paradise Point along

Brewster Boulevard. Actively pumping water supply wells should not be grouped together in small areas but should be spread out in a line perpendicular to the ambient flow direction (not parallel to it) to avoid this situation.

!

Although the BRAGS model does not directly address the third objective (i.e., predicting the relative effectiveness of various site-specific remediation schemes), it strongly indicated that the low volumes of water withdrawn from the surficial unit and/or the k:\62470\140phase\finaldoc\report\finlmodl.wpd

4-19

April 20, 1998 version

Castle Hayne Aquifer during such remedial actions will not seriously impact the water supply at the Base.

k:\62470\140phase\finaldoc\report\finlmodl.wpd

4-20

April 20, 1998 version

SECTION 4.0 TABLES

TABLE 4-2 -- STATISTICAL SUMMARY OF BRAGS SIMULATION

BRAGS, CTO-0140

MCB, CAMP LEJUENE, NORTH CAROLINA

Location

Name

Target Computed Error

Value Value

BB-220

Site 1

Site 2

Site 3

Site 6

Site 7

Site 9

Site 16

Site 21

Site 24

Site 28

Site 30

Site 35

Site 36

Site 41

Site 43

Site 44

Site 69

Site 73

Site 74

Site 78

Site 86

UST 21

A-5

BA-164

BA-190

BB-43

BB-44

BB-45

CCC-1

CCC-2

LCH-4006

LCH-4007

RR-45

TC-201

TC-202

TC-604

27.00

8.56

28.50

24.00

16.74

4.30

17.75

3.50

20.00

8.50

2.33

34.26

8.80

3.84

9.86

1.35

4.51

25.39

8.54

14.50

15.33

9.86

3.30

4.41

11.00

7.00

1.20

3.70

3.00

3.00

7.00

12.00

14.00

4.00

17.00

16.00

22.00

21.63

9.26

36.02

31.89

14.88

5.55

11.33

4.12

16.80

14.56

2.35

28.33

5.50

4.14

5.67

1.89

6.27

6.70

6.96

14.00

15.70

8.42

3.83

2.70

10.57

6.44

1.57

2.45

1.65

5.06

8.28

7.35

11.92

10.27

14.15

13.92

14.66

-1.71

-0.43

-0.56

0.37

-1.25

-1.35

2.06

1.28

-4.65

-2.08

6.27

-2.85

-2.08

-7.34

Layer 1

-5.37

0.70

7.52

7.89

-1.86

1.25

-6.42

0.62

-3.20

6.06

0.02

-5.93

-3.30

0.30

-4.19

0.54

1.76

-18.69

-1.58

-0.50

0.37

-1.44

0.53

Layer 3

Error Summary

Mean Error

Mean Absolute Error

Root Mean Square Error

Standard Deviation of Errors

Maximum Error

Minimum Error

Mean Error

Mean Absolute Error

Root Mean Square Error

Standard Deviation of Errors

Maximum Error

Minimum Error

-1.08

3.48

5.37

5.38

7.89

-18.69

-1.68

2.57

4.89

4.68

6.27

-22.58

Page1 of 4

TABLE 4-2 -- STATISTICAL SUMMARY OF BRAGS SIMULATION

BRAGS, CTO-0140

MCB, CAMP LEJUENE, NORTH CAROLINA

HP-629

HP-630

HP-633

HP-634

HP-635

HP-636

HP-637

HP-638

HP-639

HP-640

HP-641

HP-601

HP-602

HP-606

HP-609

HP-610

HP-612

HP-613

HP-614

HP-615

HP-626

HP-628

Location

Name

Target Computed Error

Value Value

Site 1D

Site 2D

Site 3D

Site 6D

Site 9D

Site 28D

Site 35D

Site 36D

Site 41D

Site 43D

Site 44D

Site 69D

Site 73D

Site 78D

Site 86D

7.25

1.50

7.00

12.39

10.00

2.33

8.67

5.00

9.88

2.22

5.98

28.03

4.00

15.25

10.00

8.52

1.45

7.11

9.07

8.91

3.04

7.22

4.10

5.77

2.33

6.57

5.45

3.43

13.81

8.20

1.27

-0.05

0.11

-3.32

-1.09

0.71

-1.45

-0.90

-4.11

0.11

0.59

-22.58

-0.57

-1.44

-1.80

Layer 3 (continued)

Layer 5

20.00

12.00

15.00

20.00

6.00

16.00

13.00

4.00

19.00

27.00

12.00

-0.14

15.00

16.70

23.00

5.00

9.60

10.00

13.20

16.00

13.30

14.00

20.70

14.32

6.41

16.35

13.76

13.45

13.11

8.65

17.24

23.12

13.30

10.57

12.53

17.02

16.08

10.81

3.37

5.66

0.45

0.15

14.51

15.87

0.70

2.32

-8.59

-3.65

7.76

-2.55

0.11

4.65

-1.76

-3.88

1.30

10.71

-2.47

0.32

-6.92

5.81

-6.23

-4.34

-12.75

-15.85

1.21

1.87

Error Summary

Mean Error

Mean Absolute Error

Root Mean Square Error

Standard Deviation of Errors

Maximum Error

Minimum Error

-0.52

5.33

6.66

6.68

12.93

-19.24

Page2 of 4

TABLE 4-2 -- STATISTICAL SUMMARY OF BRAGS SIMULATION

BRAGS, CTO-0140

MCB, CAMP LEJUENE, NORTH CAROLINA

Location

Name

Target Computed Error

Value Value

HP-642

HP-643

HP-644

HP-645

HP-646

HP-647

HP-648

HP-649

HP-650

HP-651

HP-652

HP-653

HP-654

HP-655

HP-661

HP-663

HP-698

HP-699

HP-700

HP-701

HP-703

HP-704

HP-705

HP-706

HP-708

HP-709

HP-710

HP-711

M-161

M-168

M-197

M-267

M-628

M-629

M-630

MCAS-106

MCAS-131

MCAS-203

MCAS-4140

NC-52

25.00

23.00

17.00

17.00

15.00

12.00

15.00

13.00

12.00

4.00

6.00

21.00

5.00

12.00

19.00

12.00

5.00

9.00

3.00

15.00

26.00

21.00

26.00

22.00

31.00

13.00

12.00

27.00

2.00

8.00

9.00

3.00

6.00

5.00

5.00

1.00

1.00

3.00

2.00

3.00

17.64

-0.05

-0.53

0.81

2.01

4.24

25.83

29.37

31.60

10.89

22.81

11.21

9.58

12.79

13.14

27.93

0.06

-0.27

0.95

3.59

1.76

3.43

9.91

15.87

33.06

15.14

16.64

21.65

10.66

10.69

8.23

5.21

9.82

6.62

7.27

8.84

8.76

8.51

10.31

10.55

-14.11

-0.19

-5.79

-7.42

-2.21

1.14

12.93

-12.94

-12.27

-3.05

-2.41

-19.24

-1.57

-2.09

-1.36

-12.05

-5.53

-8.19

-0.99

-10.76

-0.17

8.37

5.60

-6.13

2.06

2.14

4.64

-5.35

8.66

2.69

-0.77

2.21

3.82

1.62

2.27

7.84

7.76

5.51

8.31

7.55

Layer 5 (continued)

Error Summary

Page3 of 4

TABLE 4-2 -- STATISTICAL SUMMARY OF BRAGS SIMULATION

BRAGS, CTO-0140

MCB, CAMP LEJUENE, NORTH CAROLINA

Location

Name

Target Computed Error

Value Value

OW-2

OW-3

OW-4

OW-5

RR-47

RR-97

RR-229

T-9

TC-100

TC-104

TC-191

TC-600

TC-901

TC-1001

TC-1253

TC-1255

TC-1256

TT-23/25

TT-26

TT-31

TT-52

TT-53

TT-54

TT-67

X (1950)

X24c2

X24s2x

Y25q2

8.00

6.00

7.00

2.00

3.00

8.00

3.00

4.00

1.00

13.00

15.80

23.20

5.00

10.00

6.00

14.00

8.00

5.00

34.00

6.00

14.00

7.00

8.00

0.00

20.00

19.00

21.00

14.50

6.13

5.80

13.54

7.99

7.55

12.02

8.48

16.04

11.12

11.32

9.44

8.56

9.01

10.99

8.85

9.16

10.94

10.78

12.83

8.99

10.00

12.53

8.34

10.78

15.23

12.31

4.76

36.80

7.32

8.44

-4.44

-6.79

-12.21

3.85

-0.84

2.94

4.78

5.83

6.99

7.00

4.53

5.34

4.78

1.23

4.31

-0.24

2.80

0.13

-8.20

6.54

-0.01

7.55

-7.98

-10.52

-4.96

-3.38

Layer 5 (continued)

Layers 1, 3, & 5

Error Summary

Mean Error

Mean Absolute Error

Root Mean Square Error

Standard Deviation of Errors

Maximum Error

Minimum Error

-0.85

4.46

6.14

6.10

12.93

-22.58

Page4 of 4

LAYER

1

2

3

4

5

TABLE 4-3 -- HYDROLOGIC BUDGET SUMMARY FOR BRAGS SIMULATION

BRAGS, CTO-0140

MCB, CAMP LEJEUNE, NORTH CAROLINA

Discharge to Discharge to Discharge to Discharge to Flow through Recharge

Atlantic Supply Streams & New River Lateral 11

Ocean

(gpm)

Wells

(gpm)

Creeks

(gpm) (gpm)

Boundaries

(gpm) in/yr

(gpm)

-1,542 0 -45,125 -36,171 1,417 95,931 <==> 18,468,000 cfd

0 0

0

0

0

0

0

0

0

0

-3,056

0 -5,366 0 0 -3,042

0

0

0 0 0 0 -3,042 0

TOTAL IN:

TOTAL OUT:

Difference

% Error

97,348

-97,343

5

0.00

INFILTRATION INTO THE CASTLE HAYNE:

Net flow from Layer 1 into Layer 3 which is subsequently lost to lateral boundaries in Layers 3, 4, & 5

(sum of all Layer 1 rows)

=========>

All values taken from the MODFLOW output file (brags-cbc.out -- see Appendix B)

Negative values indicate water discharging from (exiting) the model.

Positive values indicate water recharging (entering) the model.

14,510 gpm

2,793,404 cfd

1.66 inches / year

PAGE 1 OF 1

SECTION 4.0 FIGURES

SURFICIAL UNIT

CASTLE HAYNE CONFINING UNIT

UPPER CASTLE HAYNE AQUIFER

FRACTURED LIMESTONE

LOWER CASTLE HAYNE AQUIFER

-50% -40%

Figure 4-28 Effects of Recharge

-30%

% Change from Calibrated Value

-20% -10%

8

0% 10% 20% 30%

6

4

2

0

-2

-4

Mean Error Mean Absolute Error Root Mean Square Error

40% 50%

-50%

Figure 4-29 Effects of Horizontal K

-40% -30%

% Change from Calibrated Value

-20% -10%

7

0% 10% 20%

6

30%

3

2

1

0

-1

-2

-3

5

4

Mean Error Mean Absolute Error Root Mean Square Error

40% 50%

-50%

Figure 4-30 Effects of Leakance

-40% -30%

% Change from Calibrated Value

-20% -10%

7

0% 10% 20% 30%

6

40%

5

4

3

2

1

0

-1

-2

Mean Error Mean Absolute Error Root Mean Square Error

50%

-50%

Figure 4-31 Effects of GHB Conductance

-40% -30%

% Change from Calibrated Value

-20% -10%

6

0% 10% 20% 30%

3

2

1

0

-1

5

4

-2

Mean Error Mean Absolute Error Root Mean Square Error

40% 50%

-50%

Figure 4-32 Effects of River Conductance

-40% -30%

% Change from Calibrated Value

-20% -10%

6

0% 10% 20% 30%

2

1

0

-1

5

4

3

-2

Mean Error Mean Absolute Error Root Mean Square Error

40% 50%

-50%

Figure 4-33 Effects of Drain Conductance

-40% -30%

% Change from Calibrated Value

-20% -10%

6

0% 10% 20% 30%

3

2

1

0

-1

5

4

-2

Mean Error Mean Absolute Error Root Mean Square Error

40% 50%

Figure 4-34 Comparison of ME Values

-50% -40% -30%

% Change from Calibrated Value

-20% -10%

3

0% 10% 20% 30%

Recharge (Layer 1)

Cdrn (Layer 1)

2

1

0

-1

-2

-3

-4

Kh (All Layers)

Criv (Layer 1)

40%

Leakance (All Layers)

Cghb (All Layers)

50%

-50%

Figure 4-35 Comparison of RMSE Values

-40% -30%

% Change from Calibrated Value

-20% -10%

7

0% 10% 20% 30% 40% 50%

6.5

Recharge (Layer 1)

Cdrn (Layer 1)

6

5.5

5

Kh (All Layers)

Criv (Layer 1)

Leakance (All Layers)

Cghb (All Layers)

-50%

Figure 4-36 Comparison of MAE Values

-40% -30%

% Change from Calibrated Value

-20% -10%

5

0% 10% 20% 30% 40% 50%

4.5

4

Recharge (Layer 1)

Cdrn (Layer 1)

3.5

3

Kh (All Layers)

Criv (Layer 1)

Leakance (All Layers)

Cghb (All Layers)

5.0 SITE 82 GROUNDWATER FLOW MODEL

Site 82, the Piney Green VOC Area, is a forested area approximately 30 acres in size located on the

southern boundary of Wallace Creek and north of the Hadnot Point Industrial Area (see Figure 5-1 ).

Disposal of chlorinated solvents and petroleum products occurred at the site during the 1950's through the

1970's when the adjacent site (Site 6) was used for an open lot storage area (referred to as Storage Lot 203) for various industrial materials and supplies used by the Base. Studies conducted at the site (NUS, 1991;

Baker, 1992 through 1997) indicated high levels (as high as 97,000 ug/L) of the chlorinated solvent trichloroethene (TCE) in the Castle Hayne Aquifer. Consequently, a groundwater pump and treatment system was designed for the site to address the volatile organics in the underlying aquifers and to mitigate possible migration to nearby water supply wells (notably HP-633).

After the BRAGS model for Camp Lejeune was completed, the model for Site 82 was started. Site 82 is immediately adjacent and due north of Site 6. The first step was to use the BRAGS model to get a firstorder approximation of the simulated groundwater conditions near Site 82. The external boundary conditions used in the Site 82 model were derived from the simulated heads in the BRAGS model.

Flow directions were modeled using MODPATH. MODPATH is a particle-tracking code also developed by the USGS (Pollock, 1989) that uses the results of MODFLOW to generate particle traces (or pathlines) that result from groundwater advection (flow) only. Although no dispersion, reaction, and degradation of particles are possible with this type of software, it is most useful to generate capture zones around individual wells to demonstrate contaminant capture. Electronic model input and output for the Site 82 model can be found on CD-ROM in Appendix C.

5.1 Finite-Difference Layered Grid

The finite-difference grid superimposed over Site 82 had variable spacing: square and rectangular cells

range from 25 to 1,000 feet in length (see Figure 5-2 ). The grid was comprised of 72 rows (about 10,400

feet north to south) and 94 columns (about 13,600 feet east to west) over an area of approximately five square miles.

The Site 82 model consisted of only two layers: the top layer representing the surficial unit and the bottom layer the Castle Hayne Aquifer. The Site 82 model was "quasi-3d" meaning that the confining layers were represented not by actual low hydraulic conductivity layers, but by a "leakance factor" used by MODFLOW

to calculate leakance between layers. Because the average thickness of the Castle Hayne confining unit near Camp Lejeune is about 10 feet, this was the thickness of the pseudo-confining layer. That is, the bottom of layer 1 is 10 feet above the top of layer 2.

5.2

Model Boundary Conditions

Boundaries in MODFLOW include external and internal boundaries. External boundaries can include specified head or general head boundary cells. Internal boundaries include well, river, stream, and drain cells. For the Site 82, no specified head or stream cells were used. External boundaries were general head boundary cells and internal boundaries were well, river and drain cells.

5.2.1

General Head Boundary Cells

General head boundary cells are head-dependant flow cells that allow flow into or out of the cell depending on two things: 1) the head differential between the assigned value and that in the surrounding aquifer and,

2) an assigned constant of proportionality. In the Site 82 model, the assigned value of head represents a head value (e.g., of a surface water body) at some distance beyond the model boundary and the proportionality constant represents the hydraulic conductivity of the aquifer between the model boundary and the surface water body.

Both layers had general head boundary cells placed along the outer boundaries to simulate the ambient groundwater gradient (as determined by the BRAGS model). The values of head assigned to each cell were chosen to represent the gradient of regional groundwater flow. The proportionality constants were adjusted by trial and error during the calibration process until a reasonable fit was achieved at the boundaries.

5.2.2

Well Cells

Wells cells are specified (constant) flux boundaries which keep a constant flow rate throughout the specified time period. Positive values recharge to groundwater and negative values discharge from groundwater. These cells were placed at the locations of the water supply wells and at the locations of the existing and proposed extraction wells. As in the BRAGS model, the wells were assigned average daily pumping rates in cubic feet per day (negative to represent groundwater discharge). All available well locations were plotted even if they were turned off. This will help in the future if they are turned on again.

k:\62470\140phase\finaldoc\report\finlmodl.wpd

5-2

April 20, 1998 version

The state planar coordinates of the water supply wells were converted from the latitude and longitude as recorded in Cardinell et al (1993). The state planar coordinates of the monitoring and extraction wells were taken from Site 82 survey data.

5.2.3

River Cells

River cells are head-dependant flow cells in which the elevations of the surface water and river bottom are held constant (at surveyed or mapped elevations) and the thickness and conductance of the sediments control the flow rate of water to or from the cell. If the stream or pond level is higher than the surrounding groundwater, the river cell allows water to recharge the groundwater. Conversely, if the water level in the stream or pond is lower than the groundwater, the groundwater discharges to the surface water body. The equation for river conductance C was given in the previous chapter.

R

River cells were used to represent Wallace Creek near Site 82 where its elevation is mean sea level. This includes the estuary portions of Wallace Creek (as defined on the site maps).

5.2.4

Drain Cells

Drain cells function similarly to river cells except that they cannot recharge the groundwater when the ambient water table drops below the drain elevation. Streams and swamps were represented by drain cells because it was reasonably assumed that they only receive groundwater discharge and were not recharging groundwater. The elevations of the drain cells were the approximate elevations of the streams as determined by topographic mapping of the area.

5.3

Steady-State Modeling Process

Like the BRAGS model, the Site 82 model was steady-state and all values of drawdown are assumed to have reached equilibrium. This assumption is valid when applied over the long term (years or decades) to understand how groundwater flows within the modeled system. Again, the most important assumption of this approach is that the diurnal pumping schedule of the water supply wells has been averaged as if pumping were a continuous event.

In the Site 82 model, new wells were introduced to the system. Therefore, it was necessary to calibrate the model to pre-pumping conditions (in the Site 82 extraction wells) before they could be "turned on." k:\62470\140phase\finaldoc\report\finlmodl.wpd

5-3

April 20, 1998 version

5.3.1 Pre-Pumping Calibration Targets

Pre-pumping water elevations were measured at the monitoring wells at Sites 82 and Site 6 (Storage Lot

201 -- adjacent to Site 82) during 1992 and 1993. Table 5-1 shows that there were a total of 59 head

targets: 33 shallow wells in layer 1 and 26 deep, intermediate, or supply wells in layer 2. Because of the data quality limitations of the water supply well data (discussed in the previous chapter), more credence was given to the data collected from the Site 82/Site 6 wells.

5.3.2 Calibration Methods

As in the BRAGS model, the calibration process used both "trial and error" and "parameter estimation" methods. The "parameter estimation" calibration was used generally at the beginning and at the end of the calibration process with the "trial and error" process used in the middle.

5.3.3 Statistical Evaluation of Calibration

The same statistics were used to calibrate the pre-pumping Site 82 model as were used in the BRAGS model: RM, ARM, RSD, and RMS. Generally, the degree of fit was deemed acceptable when the errors were within 10 feet of the target averages in both layers 1 and 2.

5.4 Calibrated Results of Pre-Pumping Simulation

This section describes the inputs and outputs of the calibrated pre-pumping simulation. Unless otherwise indicated, the input values used were taken directly from the values discussed in the previous sections.

Table 5-1 presents the errors of each target in the Site 82 groundwater flow model. The end of the table

shows the statistics for each layer as well as for the entire model. All simulated heads were within 14 feet of the target values. The RM for all three layers in the Site 82 model was 1.72 feet, which means that the average of all the simulated water levels in both layers was 1.72 lower than the measured water levels.

The ARM for both layers was 3.23 feet. The RSD was 3.99 feet, and the RMS was 4.35 for the final calibration.

5.4.1 Layer 1 -- Surficial Unit

5.4.1.1 Pre-Pumping Input

As in the BRAGS model, a uniform value of recharge of 11 inches per year was used in this model.

Recharge occurred only in layer 1.

The top elevation of layer 1 was assigned an arbitrary uniform value of +80 feet msl. This value is greater than the water table elevation which ensures that layer 1 remains unconfined. The bottom elevations in

layer 1 ranged from -20 feet to -10 feet msl (see Figure 5-3 ).

General head boundary cells were used to simulate the ambient groundwater at the external boundaries of the Site 82 model. On the upgradient (north and east) sides of the grid, head values of either +20 or +28 feet msl and proportionality constants of 10,000 and 5,000, respectively, were assigned to various areas

of general head boundary cells (see Figure 5-4 ). This method achieved a reasonable match of the

"incoming" water elevations. On the downgradient (west) side of the grid, head values of zero (sea level) and a proportionality constant of 50 or 5,000 were used to approximate the New River beyond the western grid boundary.

River cells were used to represent Wallace Creek near Site 82 where its elevation is mean sea level see

( Figure 5-3 ). Using the input value of 5000 ft /d for C is reasonable assuming the following parameter

values for the Wallace Creek:

L = 100 feet (average length of river cell in Wallace Creek)

W = 100 feet (average width of river cell in Wallace Creek)

M = 1 feet (estimated thickness of river sediments)

This K value is typical of a silty fine sand which is reasonable for the bottom sediment of Wallace Creek near Site 82.

As shown in Figure 5-4 , drain cells in the low-lying wetland areas near Wallace Creek were assigned a

L = 100 feet (average length of drain cell in wetlands along Wallace Creek)

W = 100 feet (average width of drain cell in wetlands along Wallace Creek)

M = 2 feet (estimated thickness of wetland sediments)

This K value is typical of silts which would be expected in wetlands.

Hydraulic conductivity in layer 1 was uniform at 3 ft/d. This was the average value from the shallow

pumping tests at Site 82 (see Appendix A ).

Leakance factors (Vcont) in layer 1 had two values: 2x10 and 3.65x10 ft/day/ft (see Figure 5-5 ).

Assuming an average water table height of 40 feet, and given a 10 foot confining layer and a thickness of

300 feet in layer 2, the vertical hydraulic conductivity of the confining layer in a quasi 3-D model can be computed from a rearrangement of the following [McDonald & Harbaugh, 1988, pp 5-16):

Vcont = (

∆ z /2)/K +

∆ z /K + (

∆ z /2)/K where:

∆ z = thickness of the confining unit (10 feet)

∆ z = thickness of the upper unit (40 feet)

∆ z = thickness of the lower unit (300 feet)

K =

∆ z / [(1/Vcont ) - (

∆ z /2)/K - (

∆ z /2)/K ]

This value is indicative of clay for the confining unit.

Where the clay is breached the value is indicative of silts.

5.4.1.2 Pre-Pumping Output

All of the simulated water levels in layer 1 were within 10 feet of their targets at Site 82.

Table 5-1 shows

that the RM in layer 1 was +1.90 feet, which means that the average simulated head value were 1.9 feet lower than the average of the measured heads in the surficial unit. The ARM was 3.36 feet, the RSD was

3.98 feet, and the RMS was 4.41. As indicated by the value of the RSD for layer 1, about 66% of the simulated heads were within 4 feet (one standard deviation) of their targets and 95% of them were within

8 feet (2 standard deviations).

Figure 5-6 shows the water table contours across Site 82 before any extraction wells were activated. The

map shows that Wallace Creek and its surrounding wetlands and tributaries are the main areas of groundwater discharge from the surficial unit near Site 82. The groundwater flow direction from the site generally follows topography due north to Wallace Creek.

5.4.2 Layer 2 - Castle Hayne Aquifer

5.4.2.1 Pre-Pumping Input

The top elevations of layer 2 range from 0 feet (sea level) to -80 feet msl (see Figure 5-6 ). Bottom

elevations in layer 2 range from -30 feet to -140 feet msl (see Figure 5-7 ). The bottom of layer 2 is sloping

from west to east across the study area.

The specified head boundaries in layer 2 are set to about +35 feet msl. The regional gradient is flowing to the northeast (as observed in the BRAGS model).

The uniform hydraulic conductivity value in layer 2 is 5 ft/d. This was the calibrated value and is close

to the average from the pumping tests in the Castle Hayne Aquifer (3 ft/d, see Table 3-2 ).

The locations of the water supply wells near Site 82 are shown on Figure 5-8 . The wells were placed in layer 2 and their “steady-state” pumping rates were determined in Section 4.0 (see Table 4-1 ). Discharge rate values used in MODFLOW are in negative cubic feet per day. No river or drain cells were used in layer 2.

5.4.2.2 Pre-Pumping Output

Figure 5-9 shows the piezometric surface contours in the Castle Hayne around Site 82. The map shows that Wallace Creek and its tributaries are still the main areas of groundwater discharge from the Castle

Hayne Aquifer. The effects of nearby water supply wells are also clearly shown on this figure. HP-633 is pumping an average of 109 gpm and HP-709 an average of 85 gpm. It is apparent from this figure that some potential exists for contaminants detected in the Castle Hayne Aquifer at Site 82 to migrate toward supply well HP-633. To date, no contamination has been reported in HP-633.

The contaminants found in the deep wells at Site 82 had apparently been drawn down into the Castle

Hayne Aquifer by the former supply well HP-651 which had been taken off-line and subsequently decommissioned due to high concentrations of organic contaminants. The screened interval of HP-651 was from 125 feet to about 200 feet bgs and the contaminants originating at the surface were drawn down into the Castle Hayne Aquifer while HP-651 pumped at a maximum rate of about 270 gpm.

All of the simulated water levels in layer 2 were within 10 feet of their targets with one exception: the simulated water level at water supply well HP-633 was 14 feet below the target. However, the measured target in HP-633 (+15 feet msl) is probably a non-pumping water level. The actual water level near HP-

633 when pumping (as simulated in the Site 82) model would be much lower. The simulated value is probably closer to the actual pumping level than this analysis would indicate.

Table 5-1 shows that the RM in layer 2 was 1.48 feet, the ARM was 3.06 feet, the RSD was 3.99 feet, and the RMS was 4.26 feet. As indicated by the value of the RSD for layer 2, about 66% of the simulated heads were within 4 feet (one standard deviation) of their targets and 95% of them were within 8 feet (2 standard deviations). Again, the match for the layer 2 is very similar to that of layer 1.

5.5 Results of Remediation Scenario Simulation

This section describes the inputs and outputs of the remediation simulation. All of the inputs other than extraction wells are identical to the pre-pumping inputs described in section 5.4.

Chlorinated organic compounds have been identified in the surficial and deep groundwater at Site 82. The estimated extents of the plumes within the surficial and deep groundwater evaluated from the July 1997 groundwater are shown in Figures 5-10 and 5-11 , respectively. Originally, three shallow and three deep extraction wells were proposed at Site 82 to contain the off-site migration. The pumping rates of the shallow extraction wells were based on data from previous pumping tests conducted within the surficial unit. The pumping rates of the deep extraction wells were estimated based on the pumping rate of the nearby water supply well HP-651. This supply well was decommissioned because of very high chlorinated organic compound concentrations believed to have originated from Sites 6 and 82. Supply well HP-651 had a maximum pumping rate of 270 gpm and was screened between depths of 125 to 200 feet bgs with most of the water being produced from 125 to 155 feet bgs. The well log for HP-651 is shown on Figure

5-12 .

With the additional information made available by the Site 82 pumping tests, it became apparent that three shallow wells would not be adequate to contain the off-site migration in the surficial unit. Also, the locations and pumping rates of the three deep extraction wells were revisited. Many intermediate remediation schemes were run with the model where locations and pumping rate of wells were slightly altered before the final remedial scenario was chosen.

5.5.1 Layer 1 -- Surficial Unit

5.5.1.1 Remediation Scenario Input

The soil vapor extraction system (adjacent to the existing extraction well SRW-1) was taken off-line after completion of the soil remediation project. However, groundwater contaminant concentrations in this area remain high. SRW-1 is designed to pump from this "hot spot" area of high VOC concentrations in the groundwater. Five additional shallow (30 feet bgs) extraction wells are proposed to be placed in an eastwest line near the northern edge of Site 82. This line of wells is just south of, and adjacent to, the wetland floodplain of Wallace Creek (see Figure 5-13 ). The easternmost well, SRW-2, is 350 feet due north of

SRW-1.

Each of the shallow extraction wells is anticipated to pump up to 5 gpm. The locations of the five additional wells were optimized to contain the shallow groundwater contamination from the area between wells 6MW-32 and 6MW-40. Simulated remedial scenarios with fewer wells were tried unsuccessfully to contain the existing contaminant plumes.

5.5.1.2 Remediation Scenario Output

Figure 5-14 shows the water table contours in a close-up of the Site 82 area with the shallow extraction wells activated. Figure 5-15 shows the capture zones of each well superimposed on the water table contours and the total VOC plume concentrations. The steady-state simulation shows that the proposed wells are able to capture the shallow contamination and prevent further off-site migration toward Wallace

Creek.

5.5.2 Layer 2 - Castle Hayne Aquifer

5.5.2.1 Remediation Scenario Input

Three deep (90 to 110 feet bgs) extraction wells were originally proposed for Site 82. As happened with the design for the shallow extraction system, new information became available with the pumping tests at the site (see Appendix A ). The main difference between the original and the new designs was that the lower anticipated pumping rates reduced the expected radii of the capture zones. After the pumping test, the pumping rate from SRW-1 was lowered from 150 gpm to about 25 gpm. The other two wells were also projected to have similar rates so the designed screen lengths of SRW-2 and SRW-3 were increased from

20 to 30 feet. This change is expected to increase the expected pumping rate in these two wells to at least

40 gpm.

Also, a modification was made to the original design in which SRW-2 was moved about 400 feet north of its original location (see Figure 5-16 ). This change brought the two downgradient wells (SRW-2 and

SRW-3) into a line perpendicular to groundwater flow which is the typical arrangement for an extraction well system.

5.5.2.2 Remediation Scenario Output

Figure 5-17 shows the steady-state piezometric surface contours in the Castle Hayne around Site 82 with all three deep extraction wells activated. On Figure 5-18 , the capture zones around each well show that the contaminants in the Castle Hayne Aquifer can be contained by the proposed remedial design.

5.6 Site 82 Groundwater Flow Model Summary

The Site 82 model describes the three-dimensional pattern of groundwater flow in the surficial unit and

Castle Hayne Aquifer (based on the data to which it was calibrated). It achieves the three objectives described in Section 1.1:

! Based on the conceptual model described in Section 2.0, the Site 82 model describes how groundwater flows beneath Site 82 (Objective 1).

! The Site 82 model demonstrates the effects of remedial groundwater withdrawals on the surficial unit and the Castle Hayne Aquifer (Objective 2). The model demonstrates that the relatively low-volume withdrawal rates of the extraction wells will have an extremely localized effect on the water levels in the surficial unit and the Castle Hayne Aquifer.

! The Site 82 model directly addressed the third objective: it clearly showed the relative effectiveness of various site-specific remediation schemes. The locations of the extraction wells in the surficial and in the Castle Hayne Aquifer were finalized by the successful running of the model. "Success" was indicated by complete hydraulic control or "capture" of the contaminant plume. Also, the model indicated that the low volumes of water withdrawn during such remedial actions will not seriously impact the water supply at the

Base.

SECTION 5.0 TABLES

SECTION 5.0 FIGURES

GW30

82MW03

GW32

4426

GW28

29

4000

GW33

500

0

GW34

6080

60

00

GW01

82MW02

GW03

GW21

2000

GW16

2711

LEGEND

GW08

0

4426.00

Shallow Monitoring Well Location with

Total VOC Concentration

Estimated Extent of Contaminant Concentrations

NOTES:

1. Concentrations presented in micrograms per liter

500 0

Scale

500 1000 Feet

N

W E

S

Loc ID Parameter Result

Units Date

IR06-GW34

IR06-GW28

IR06-GW16

IR06-GW32

TRICHLOROETHENE

TETRACHLOROETHENE

1,1,2,2-TETRACHLOROETHANE

TRICHLOROETHENE

TETRACHLOROETHENE

CHLOROBENZENE

1,1,2,2-TETRACHLOROETHANE

VINYL CHLORIDE

TETRACHLOROETHENE

1,2-DICHLOROETHENE (TOTAL)

TRICHLOROETHENE

310

170

5600

22

7

2700

11

16

110

1500

2800

UG/L

UG/L

UG/L

UG/L

UG/L

UG/L

UG/L

UG/L

UG/L

UG/L

UG/L

07/24/97

07/24/97

07/24/97

07/25/97

07/25/97

07/27/97

07/27/97

07/27/97

07/27/97

07/27/97

07/27/97

MARINE CORPS BASE, CAMP LEJEUNE

NORTH CAROLINA

Volatile Organic Compounds in

Shallow Groundwater

Operable Unit No. 2 - Site 82

Groundwater Flow Model

CTO-0140

FIGURE 5-10

GW36D

GW35D

GW30DW

GW37D

318

1000

5000

GW27D

8321

GW28DW

1650

10000

50000

GW01D

126275

GW40DW

GW40DWA

GW01DA

100000

GW38D

GW15D

GW02DW

MW03D

300

LEGEND

GW28DW

·

1650

Deep Monitoring Well Location with

Total VOC Concentration

Estimated Extent of Contaminant Concentrations

NOTES:

1. Concentrations presented in micrograms per liter

0 300

Scale

600 900 1200 Feet

N

W E

S

Loc ID

IIR06-GW27DW

IR06-GW37D

IR06-GW28DW

IR06-GW01D

Parameter

1,1-DICHLOROETHENE

VINYL CHLORIDE

TRICHLOROETHENE

1,2-DICHLOROETHENE (TOTAL)

TRICHLOROETHENE

1,2-DICHLOROETHENE (TOTAL)

1,2-DICHLOROETHENE (TOTAL)

TRICHLOROETHENE

METHYLENE CHLORIDE

VINYL CHLORIDE

1,1-DICHLOROETHENE

TETRACHLOROETHENE

1,2-DICHLOROETHENE (TOTAL)

TRICHLOROETHENE

Result

11

110

3400

4800

88

230

550

1100

8

320

57

890

28000

97000

Units

UG/L

UG/L

UG/L

UG/L

UG/L

UG/L

UG/L

UG/L

UG/L

UG/L

UG/L

UG/L

UG/L

UG/L

Date

07/22/97

07/22/97

07/22/97

07/22/97

07/23/97

07/23/97

07/25/97

07/25/97

07/26/97

07/26/97

07/26/97

07/26/97

07/26/97

07/26/97

MARINE CORPS BASE, CAMP LEJEUNE

NORTH CAROLINA

Volatile Organic Compounds in

Deep Groundwater

Operable Unit No. 2 - Site 82

Groundwater Flow Model

CTO-0140

FIGURE 5-11

6.0

CONCLUSIONS AND RECOMMENDATIONS

The modeling effort described herein was successful in achieving the three objectives stated at the outset of this report. The objectives were to:

!

Describe how groundwater flows beneath the entire Base as well as under individual sites of concern.

!

Demonstrate the effects of groundwater withdrawals (supply and remedial) on the aquifers in question (most notably the surficial unit and the Castle Hayne Aquifer).

!

Predict the relative effectiveness of various remediation schemes at individual sites

(including Site 82).

The two groundwater flow models were intended to be "working" models, that is, they were meant to be transferred into the hands of Base personnel (or their representatives) to update and modify to site-level work or as new information becomes available. The updated models will be effective decision-making tools for optimal groundwater resource management, protection, and restoration. The models can be used to determine the relative effectiveness of various remedial scenarios at individual sites around the Base.

The BRAGS groundwater flow model presented herein portrays the three-dimensional pattern of groundwater flow within the surficial units and the Castle Hayne Aquifer (based on the data to which it was calibrated). The model reasonably predicts the elevation and flow direction of the surficial and Castle

Hayne groundwater in many areas around the Base where no data currently exist. The BRAGS model also demonstrates that discharge to the New River is the controlling factor on flow directions in the Castle

Hayne Aquifer in the vicinity of Camp Lejeune. The model output indicates that the relatively high-volume withdrawal rates of the supply wells have a localized effect on the water levels in the Castle Hayne.

One of the concerns that initiated this modeling effort was that the potential number of pump and treat remedial actions at the Base may negatively impact the supply of available groundwater. The BRAGS model strongly indicated that the low volumes of water withdrawn from the surficial unit and/or the Castle

Hayne Aquifer during such remedial actions will not noticeably affect the groundwater supply at the Base; however, large numbers of actively pumping water supply wells in small areas have the potential to induce saltwater intrusion into the upper Castle Hayne Aquifer. This effect is most pronounced in Paradise Point k:\62470\140phase\finaldoc\report\finlmodl.wpd

6-1

April 20, 1998 version

along Brewster Boulevard. Actively pumping water supply wells should not be grouped together in small areas but should be spread out in a line perpendicular to the ambient flow direction (not parallel to it) to avoid this situation.

The Site 82 model describes the three-dimensional pattern of groundwater flow in the surficial unit and

Castle Hayne Aquifer. The Site 82 model demonstrates the effects of proposed remedial groundwater withdrawals on the surficial unit and the Castle Hayne Aquifer. The model also demonstrates that the relatively low-volume withdrawal rates of the extraction wells will have an extremely localized effect on the water levels in the surficial unit and the Castle Hayne Aquifer.

The Site 82 model directly addressed the third objective: it clearly showed the relative effectiveness of various site-specific remediation schemes. The locations of the extraction wells in the surficial and in the

Castle Hayne Aquifer were finalized by the successful running of the model. "Success" was indicated by complete hydraulic control or "capture" of the contaminant plume. Also, the model indicated that the low volumes of water withdrawn during such remedial actions will not noticeably affect the groundwater supply at the Base.

The groundwater flow models described herein will be useful in managing the future RI activities at the

Base. The BRAGS model will be especially useful for determining the groundwater flow patterns in areas where no data currently exists and it gives a regional perspective on site-specific modeling. Future groundwater flow and/or contaminant transport modeling done at the site level should be coordinated with the BRAGS groundwater flow model so that the "big picture" of the groundwater flow is consistent across the Base.

It is strongly recommended that the additional hydrogeologic and chemical data collected from the on-going remediation activities and long-term monitoring at Site 82 be incorporated into the Site 82 groundwater flow model. At that time, the Site 82 groundwater flow model should be converted to fully threedimensional so that it is capable of modeling contaminant transport. From future modeling efforts (which should include actual pumping rates, updated groundwater elevations, and contaminant concentrations) recommendations can be provided to address the question of complete capture and the necessity of additional wells at Site 82.

k:\62470\140phase\finaldoc\report\finlmodl.wpd

6-2

April 20, 1998 version

7.0

REFERENCES

Anderson, Mary P. and William W. Woessner. 1992. Applied Groundwater Modeling -- Simulation of

Flow and Adjective Transport. Academic Press, Inc. Harcourt Brace Jovanovich, San Diego. 381 pp.

Baker Environmental, Inc. 1997a. Draft Feasibility Study for Site 73, Operable Unit No. 9, Amphibious

Vehicle Maintenance Facility at Marine Corps Base Camp Lejeune, North Carolina. Prepared for the

Department of the Navy, Naval Facilities Engineering Command, Atlantic Division, Norfolk, Virginia.

February, 1997.

Baker Environmental, Inc. 1997b. Draft Groundwater Modeling Report for Site 73, Marine Corps Base

Camp Lejeune, North Carolina. Prepared for the Department of the Navy, Naval Facilities Engineering

Command, Atlantic Division, Norfolk, Virginia. March 1997.

Baker Environmental, Inc. 1997c. Site Management Plan for Marine Corps Base, Camp Lejeune,

North Carolina. Prepared for the Department of the Navy, Naval Facilities Engineering Command,

Atlantic Division, Norfolk, Virginia.

Baker Environmental, Inc. 1996a. Draft Basewide Remedial Assessment Groundwater Study (BRAGS),

Marine Corps Base Camp Lejeune, North Carolina. Prepared for the Department of the Navy, Naval

Facilities Engineering Command, Atlantic Division, Norfolk, Virginia. April, 1996.

Baker Environmental, Inc. 1996b. Draft Remedial Investigation Report for Operable Unit No. 9 (Site

73), Marine Corps Base Camp Lejeune, North Carolina. Prepared for the Department of the Navy, Naval

Facilities Engineering Command, Atlantic Division, Norfolk, Virginia. November 1996.

Baker Environmental, Inc. 1993. Final Remedial Investigation/Feasibility Study for Operable Unit No.

2 (Sites 6, 9, and 82), Marine Corps Base Camp Lejeune, North Carolina. Prepared for the Department of the Navy, Naval Facilities Engineering Command, Atlantic Division, Norfolk, Virginia. August 1993.

Cardinell, A.P., S.A. Berg, and O.B. Lloyd, Jr. 1993. Hydrogeologic Framework of U.S. Marine Corps

Base Camp Lejeune, North Carolina. USGS Water Resources Investigations Report 93-4049.

k:\62470\140phase\finaldoc\report\finlmodl.wpd

7-1

April 20, 1998 version

Cardinell, A.P, D.A. Harned, and S.A. Berg. 1990. Continuous Seismic Reflection Profiling of

Hydrogeological Features beneath New River, Camp Lejeune, North Carolina. USGS Water Resources

Investigations Report 89-4195.

Dixon, W.J., and F.J. Massey, Jr. 1983. Introduction to Statistical Analysis. McGraw-Hill, Inc., New

York. 678 pp.

Eimers, J.L., C.C. Daniel, III, and R.W. Coble. 1994. Hydrogeology and Simulation of Groundwater

Flow at U.S. Marine Corps Air Station, Cherry Point, North Carolina, 1987-90. USGS Water Resources

Investigations Report 94-4186.

Eimers, J.L., W.L. Lyke, and A.R. Brockman. 1990. Simulation of Groundwater Flow in Cretaceous

Rocks in the Central Coastal Plain, North Carolina. USGS Water Resources Investigations Report 89-

4153.

ES&E. 1988. Confirmation Study -- Marine Corps Base Camp Lejeune. May 24, 1988.

Federal Facilities Agreement (FFA) between United States Environmental Protection Agency, Region

IV; the North Carolina Department of Environment, Health, and Natural Resources; and the United

States Department of the Navy for Marine Corps, Camp Lejeune and Marine Corps Air Station, New

River, North Carolina. December 6, 1989.

Geise, G.L., J.L. Eimers, and R.W. Coble. 1991. Simulation of Groundwater Flow in the Coastal Plain

Aquifer System of North Carolina. USGS Open File Report 90-372.

Geophex, Ltd. 1991. Wellhead Management Program Engineering Study 91-36. Raleigh, NC, 44 pp.

386

Geraghty & Miller, Inc. 1993. ModelCad Documentation (2 volumes). Reston, VA.

Golden Software, Inc. 1991. SURFER Documentation for Version 4.0. Golden, CO.

k:\62470\140phase\finaldoc\report\finlmodl.wpd

7-2

April 20, 1998 version

Harned, D.A., O.B. Lloyd, Jr., and M.W. Treece, Jr. 1989. Assessment of Hydrologic and

Hydrogeologic Data at Camp Lejeune Marine Corps Base, North Carolina. USGS Water Resources

Investigations Report 89-4096.

Heath, Ralph C. 1983. Basic Ground-Water Hydrology. USGS, Water Supply Paper 2220. 75 pp.

Hill, Mary C. 1992. A Computer Program (MODFLOWP) for Estimating Parameters of a Transient,

Three-Dimensional, Groundwater Flow Model Using Non-Linear Regression. USGS Open File Report

91-484. Denver, Colorado. 102 pp.

Leahy, P.P and M. Martin. 1993. Geohydrology and Simulation of Groundwater Flow in the Northern

Atlantic Coastal Plain Aquifer System. Regional Aquifer System Analysis (RASA) -- Northern Atlantic

Coastal Plain. USGS Professional Paper 1404-K.

Lyke, W.L. and M.D. Winner, Jr. 1990. Hydrogeology of Aquifers in Cretaceous and Younger Rocks

in the Vicinity of Onslow and Southern Jones Counties, North Carolina. USGS Water Resources

Investigations Report 89-4128.

McDonald, Michael G., and Arlen W. Harbaugh. 1988. A Modular Three-dimensional Finite Difference

Groundwater Flow Model [MODFLOW]. Chapter A1 in Techniques of Water Resources Investigations of the United States Geological Survey, Book 6, Modeling Techniques.

NUS Corporation. 1991. Preliminary Draft Site Inspection Report, Site 82 Piney Green Road, Marine

Corps Base Camp Lejeune, North Carolina. Prepared for the Department of the Navy, Naval Facilities

Engineering Command, Atlantic Division, Norfolk, Virginia. November 1991.

Pollock, David W. 1989. Documentation of Computer Programs to Compute and Display Pathlines

Using Results from the U.S. Geological Survey Modular Three-dimensional Finite-difference

Groundwater Flow Model [MODPATH]. USGS Open File Report 89-381. 188 pp.

Stuckey, J.L. 1965. North Carolina -- Its Geology and Mineral Resources. North Carolina Department of Conservation and Development. 550 pp..

k:\62470\140phase\finaldoc\report\finlmodl.wpd

7-3

April 20, 1998 version

Todd, D. K. 1980. Groundwater Hydrology. John Wiley & Sons. New York. 535 pp.

Trapp, Henry, Jr. 1992. Hydrogeologic Framework of the Northern Atlantic Coastal Plain in parts of

North Carolina, Virginia, Maryland, Delaware, New Jersey and New York. Regional Aquifer System

Analysis (RASA) -- Northern Atlantic Coastal Plain. USGS Professional Paper 1404-G.

Water and Air Research, Inc. 1983. Initial Assessment of Marine Corps Base Camp Lejeune, North

Carolina. Prepared for the Department of the Navy, Naval Energy and Environmental Support Activity,

Port Hueneme, California. April 1983.

Wilder, H.B., T. M. Robison, and K. L. Lindskov. 1978. Water Resources of North Carolina. USGS

Water Resources Investigations Report 77-81. 113 pp.

Winner, M.D., Jr. and R.W. Coble. 1989. Hydrogeologic Framework of the North Carolina Coastal

Plain Aquifer System. USGS Open File Report 87-690, 155 pp.

k:\62470\140phase\finaldoc\report\finlmodl.wpd

7-4

April 20, 1998 version

APPENDIX A

SITE 82 PUMPING TEST DATA EVALUATION

Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement