542R04001a
ฎEPA
    United States
    Environmental Protection
    Agenoy
Demonstration of
Two Long-Term Groundwater Monitoring
Optimization Approaches

Report
           Compliance boundary
         Groundwater flow direction
                                     *  +
                                        The nearest
                                        downgradient
                                        receptor

-------

-------
                                     NOTICE
This report was prepared by Mitretek Systems (Mitretek) for the U.S. Environmental Protection
Agency (U.S. EPA) under U.S. EPA Requisition #B4T024, QT-DC-04-000504, and summarizes
the results of demonstration projects completed by The Parsons Corporation (Parsons) under Air
Force Center for Environmental Excellence (AFCEE) contract (Contract No. F41624-00-D-8024,
Task Order No. 0024), and by Groundwater Services, Inc. (GSI), also under an AFCEE contract
(Contract No. F41624-98-C-8024). Reference to trade names, commercial products, process, or
service does not constitute or imply endorsement, recommendation for use, or favoring by the
United States  Government or any  agency thereof.  The views and opinions of the authors
expressed herein do not necessarily state or reflect those of the United States Government or any
agency thereof.

This document, with its appendices (542-R-04-001b) or without its appendices (542-R-04-001a),
may be downloaded from U.S. EPA's Clean Up Information (CLUIN) System at http://www.clu-
jn.org. A limited number of hard copies of each version also are available free of charge from the
National Service Center for Environmental Publications (NSCEP) at the following address:

U.S. EPA National Service Center for Environmental Publications
P.O. Box 42419
Cincinnati, OH 45242-2419
Phone: (800)490-9198 or (513)489-8190
Fax: (513)489-8695

-------
                                    PREFACE
This report summarizes the results of a demonstration in which optimization techniques were
used to improve the design of long-term groundwater monitoring programs.  Two different
approaches to optimizing groundwater monitoring programs were used in the demonstration:

   •   The  Monitoring  and  Remediation  Optimization  System  (MAROS)  software tool,
       developed by GSI for AFCEE (2000 and 2002), and

   •   A three-tiered approach applied by Parsons.

The report discusses the results  of application of the two approaches to  the  evaluation  and
optimization of groundwater monitoring programs at three sites (the Fort Lewis Logistics Center,
Washington, the Long Prairie Groundwater  Contamination Superfund Site in Minnesota,  and
Operable Unit D,  McClellan Air Force Base, California), and examines  the overall results
obtained using the two monitoring program optimization approaches. The primary goals of this
demonstration  were to highlight  current  strategies for  applying optimization techniques to
existing long-term monitoring programs,  and  to assist  site managers in understanding  the
potential  benefits  associated with monitoring program optimization.  The  demonstration was
conducted as part of an assessment of long-term monitoring optimization approaches, initiated by
the U.S. Environmental Protection Agency's  Office of Superfund Remediation and Technology
Innovation (USEPA/OSRTI) and AFCEE.
                                            n

-------
                           ACKNOWLEDGEMENTS
This report summarizes the results of a demonstration of two different approaches to optimizing
long-term groundwater monitoring programs.  The  demonstration projects summarized herein
were  completed  by The  Parsons Corporation  (Parsons), Dr.  Carolyn  Nobel  as  principal
investigator, and Groundwater Services, Inc. (GSI), Ms. Julia Aziz as principal investigator; the
two teams are commended for the  quality  of their work,  and the principal investigators are
thanked for their helpful cooperation through the course of this project.

This project would not have been  possible without  the cooperation  of the facilities whose
monitoring programs were the subjects of the demonstration:

Fort Lewis, Washington - Richard W. Smith, U.S.  Army Corps  of Engineers, Seattle District,
Point of Contact

Long Prairie Superfund Site, Minnesota - Mark Elliott, Minnesota Pollution Control Agency, and
Eric Gabrielson, Barr Engineering, Points of Contact

Former McClellan Air Force Base, California - Brenda Callan, URS Corporation, and Diane H.
Kiyota, Air Force Real Property Agency (AFRPA), Points of Contact

The authors also wish to acknowledge the reviewers who have improved  this document with their
productive comments.  Their advice and assistance during the project are  greatly appreciated.

The following agencies or individuals can be contacted for additional information:

U.S. Environmental Protection Agency,
  Office of Superfund Remediation and Technology Innovation (U.S. EPA/OSRTI)
MS5102G
1200 Pennsylvania Avenue NW
Washington, D.C. 20460
(703)603-9910
John W. Anthony
Lead Hydrologist
Mitretek Systems
7720 E. Belleview Avenue, Suite BG6
Greenwood Village, Colorado  80111
john.anthony@mitretek.org

Carolyn Nobel, Ph.D.
Senior Scientist
The Parsons Corporation
1700 Broadway, Suite 900
Denver, Colorado 80290
carolyn.nobel@parsons.com
E. Kinzie Gordon
Lead Scientist/Regulatory Specialist
Mitretek Systems
7720 E. Belleview Avenue, Suite BG6
Greenwood Village, Colorado  80111
kinzie.gordon@mitretek.org

Julia J. Aziz
Senior Scientist
Groundwater Services, Inc.
2211 Norfolk Street, Suite 1000
Houston, Texas 77098
i aziz@gsi-net.com
                                            111

-------

-------
                             EXECUTIVE SUMMARY
This report summarizes the results of a demonstration in which optimization techniques were used to
improve  the design  of several  long-term  groundwater monitoring programs.   Two  different
approaches to optimizing groundwater monitoring programs were applied in the demonstration:

   •   The Monitoring and Remediation Optimization System (MAROS) software tool, developed
       by Groundwater Services, Inc. (GSI) for AFCEE (2000 and 2002), and

   •   A three-tiered approach applied by The Parsons Corporation (Parsons).

The report  discusses the results of  application  of the two approaches to  the evaluation and
optimization of groundwater monitoring programs at three sites (the Fort Lewis Logistics Center,
Washington, the  Long Prairie Groundwater Contamination  Superfund Site  in Minnesota, and
Operable  Unit D, former McClellan Air Force Base, California), and examines the overall results
obtained using the two long-term monitoring optimization (LTMO) approaches.  The primary goals
of this demonstration were to highlight current strategies for applying optimization techniques to
existing long-term monitoring (LTM) programs, and to  assist site managers in  understanding the
potential  benefits  associated with  monitoring  program optimization.   The  demonstration was
conducted as part of an assessment of LTMO approaches, initiated by  the U.S. Environmental
Protection Agency's Office of Superfund Remediation and Technology Innovation (USEPA/OSRTI)
and the Air Force Center for Environmental Excellence (AFCEE).

The MAROS tool is a public-domain software package that operates in conjunction with an electronic
database environment (Microsoft Accessฎ 2000)  and performs certain mathematical and/or statistical
functions appropriate to completing qualitative, temporal,  and  spatial-statistical evaluations of a
groundwater monitoring program,  using data that have been loaded into the database (AFCEE, 2000
and 2002).  MAROS utilizes  parametric  temporal  analyses (using linear regression) and non-
parametric  trend  analyses (using  the Mann-Kendall  test  for  trends)  to  assess  the statistical
significance of temporal trends in concentrations of contaminants of concern (COCs). MAROS then
uses the  results of the temporal-trend analyses to develop recommendations  regarding optimal
sampling frequency at each sampling point in a monitoring program by applying a modified Cost-
Effective Sampling algorithm, to assess  the feasibility of reducing the frequency of sampling at
individual sampling points. Although the MAROS tool primarily is used to evaluate temporal data, it
also incorporates a spatial statistical algorithm, based on a ranking  system that utilizes a weighted
"area-of-influence" approach (implemented using Delaunay triangulation) to assess the relative value
of data generated  during  monitoring, and to identify the optimal locations of monitoring points.
Formal decision  logic  and  methods  of  incorporating user-defined secondary  lines  of evidence
(empirical or modeling results) also are provided, and can be used to further evaluate monitoring data
and make recommendations for adjustments to  sampling frequency, monitoring locations, and the
density of the monitoring network.

In  the three-tiered LTMO approach, the  monitoring-program evaluation is conducted in stages to
address each of the objectives  and considerations of monitoring:  a qualitative evaluation first is
completed,  followed in succession by temporal  and spatial evaluations.  At the conclusion of each
stage (or "tier") in the evaluation,  recommendations are generated regarding potential changes in the
temporal frequency of monitoring, and/or  whether  to  retain or remove each  monitoring point
                                             IV

-------
 considered in the evaluation.  After all three stages have been completed, the results of all of the
 analyses are combined and interpreted, using a decision algorithm, to generate final recommendations
 for an effective and efficient LTM program.

 Application of the two approaches to the optimization of LTM programs at each of the three case-
 study example sites generated recommendations for reductions in sampling frequency and changes in
 the numbers and locations of monitoring points that are sampled.  Implementation of the optimization
 recommendations could lead to reductions ranging from only a few percent to more than 50 percent
 in the numbers of samples  collected and analyzed annually at particular sites (Table ES.l).  The
 median recommended reduction in the  annual  number of samples collected, generated during the
 optimization demonstration, was 39 percent. Although available information regarding monitoring-
 program costs at each of the three case-study example sites  is not directly comparable, it is projected
 that depending upon the scale of the particular LTM  program, and the  nature of the optimization
 recommendations, adoption of optimized monitoring programs at each of the case-study sites could
 lead  to  annual cost  savings ranging  from a  few  hundred dollars (using the recommendations
 generated by MAROS for the monitoring program at Operable Unit D [OU D], former McClellan Air
 Force Base [AFB])  to approximately  $36,500  (using the  results  generated  by the  three-tiered
 approach for the monitoring program at the Fort Lewis Logistics Center Area). The results of the
 evaluations also  demonstrate that each of the optimized monitoring programs remains  adequate to
 address the primary objectives of monitoring at the sites. Although the general characteristics of each
 of the three case-study example sites are similar (chlorinated solvent contaminants in groundwater,
 occurring at relatively shallow depth in unconsolidated sediments), the assumptions underlying the
 two approaches, and the procedures that are followed in conducting the evaluations are applicable to
 a much broader  range of conditions (e.g., dissolved metals  in groundwater, or contaminants in  a
 fractured bedrock system).

                  Table ES.l:  Summary of Results of LTMO Demonstrations
Feature of Monitoring Program
Total number of samples (per year) in
current program
Rangeb/ of total number of samples
(per year) in refined program
Percent reduction in number of
samples collected per year
Projected range of cost savings0' (per
year)
Example Site*
Fort Lewis
180
107-113
37-40
$33,500 - $36,500
Long Prairie
51
22-36
29-51
$4,200 -$8,1 00
McClellan AFB OU D
34
17-32
6-50
$300 - $2,550
" Information regarding site characteristics and the site-specific monitoring programs of the three example sites is presented
  in Section 3 (Fort Lewis), Section 4 (Long Prairie) and Section 5 (McClellan AFB OU D), and in Appendices C and D.
bl Ranges of total numbers of samples collected annually in refined programs, percentage reductions in numbers of samples
  collected, and associated potential annual cost savings, reflect the results of the evaluations conducted using MAROS and
  the three-tiered approach.
d Estimates of potential annual cost savings were based on information regarding monitoring program costs provided by
  facility personnel. Costs associated with monitoring include cost of sample collection, sample analyses, data compilation
  and reporting, and management of investigation-derived waste (e.g., purge water).
Prior to initiating an LTMO evaluation, it is of critical importance that the monitoring objectives of
the program to be optimized be  clearly articulated, with all stakeholders agreeing to the stated
objectives, so that the program can be optimized in terms of recognized (and agreed-upon) objectives,

-------
using decision rules and procedures that are acceptable to all stakeholders. The decisions regarding
whether to conduct an LTMO evaluation, which approach to use, and the degree of regulatory-agency
involvement in the LTMO evaluation and implementation of optimization recommendations, must be
made on a site-specific basis.  Factors to be considered  in deciding whether to proceed with an
LTMO evaluation include:

   •   The projected level of effort necessary to conduct the evaluation;

   •   The resources available for the evaluation (e.g., quality and quantity of data, staff having the
       appropriate technical capabilities);

   •   The anticipated degree of difficulty in implementing optimization recommendations; and

   .   The potential benefits (e.g., cost savings) that could result from an  optimized monitoring
       program.


Optimization of a monitoring program should be considered for most sites having LTM programs that
are based on sampling of characterization monitoring points, or for sites where more than about 50
samples are collected and analyzed on an annual basis. Because it is likely that monitoring programs
can  benefit  from  periodic evaluation  as environmental programs evolve,  monitoring program
optimization  also should be undertaken periodically, rather than being regarded as a one-time event.
Overall site conditions should be  relatively stable, with no large changes  in remediation approaches
occurring or  anticipated. Furthermore, successful application of either  approach to the site-specific
evaluation of a monitoring program is directly  dependent upon the  amount and quality  of the
available data - results from a minimum of four to six separate sampling events are necessary to
support a temporal analysis, and results  collected  at  a  minimum  of about six (for a MAROS
evaluation) to 15 (for a three-tiered evaluation) separate monitoring points are necessary to support a
spatial analysis.  It also is necessary to develop an adequate conceptual site model (CSM)  describing
site-specific conditions prior to applying either approach.  In particular, the extent of contaminants in
the subsurface at  the site  must be adequately delineated before  the  monitoring program can be
optimized.

Although the MAROS tool is capable of being applied by an individual with little formal statistical
training, interpretation of the results generated by either approach requires a relatively sophisticated
understanding of hydrogeology, statistics, and the processes governing the movement and fate of
contaminants in the  environment.   Although  many of the  basic assumptions and  techniques
underlying both optimization approaches are  similar,  and  both optimization approaches  utilize
qualitative, temporal, and spatial analyses, there are several differences between the two approaches,
which can cause one optimization approach (e.g., the three-tiered approach) to generate results that
are not completely consistent with the results obtained using the other approach (e.g.,  MAROS).
Nevertheless, each approach is capable of generating sound and defensible recommendations for
optimizing LTM programs.

The most  significant advantage conferred by both  optimization approaches is the  fact that  both
approaches apply consistent, well-documented procedures, which incorporate formal decision logic,
to the process of evaluating and optimizing groundwater monitoring programs.  However, the process
of data preparation, screening, processing, and evaluation can be extremely time-consuming for either
approach.   Both approaches could benefit  from further development  efforts to address current
limitations; and continued development of both approaches is contemplated or in progress.
                                              VI

-------
Typically, a program manager should anticipate incurring costs on the order of $6,000 to $10,000 to
complete an LTMO  evaluation at the  level of detail of the case-study examples described in this
demonstration.  Consequently, an LTMO evaluation may be cost-prohibitive for smaller monitoring
programs.  However, an LTMO evaluation that can be used to reduce the total number of samples
collected at a site by about 5 to 10 samples per annum should be cost-effective.
                                            vn

-------
                        TABLE OF CONTENTS
EXECUTIVE SUMMARY	iv
LIST OF ACRONYMS AND ABBREVIATIONS	xii

1.0 INTRODUCTION	1
      1.1     PROJECT DESIGN	1
      1.2     CASE-STUDY EXAMPLES	2
      1.3     PURPOSES OF GROUNDWATER MONITORING	2
      1.4     LONG-TERM GROUNDWATER MONITORING PROGRAM OPTIMIZATION	4
      1.5     REPORT ORGANIZATION	6

2.0 EVALUATION AND OPTIMIZATION OF LONG-TERM
   MONITORING PROGRAMS	7
      2. l     CONCEPTS IN GROUNDWATER MONITORING	7
      2.2     METHODS FOR DESIGNING, EVALUATING, AND OPTIMIZING MONITORING
            PROGRAMS	9
      2.3     DESCRIPTION OF MAROS SOFTWARE TOOL	10
      2.4     DESCRIPTION OF THREE-TIERED APPROACH	14
      2.5     CASE-STUDY EXAMPLES	16

3.0 SUMMARY OF DEMONSTRATIONS AT LOGISTICS CENTER AREA,
   FORT LEWIS, WASHINGTON	17
      3.1     FEATURES OF FORT LEWIS LOGISTICS CENTER	17
      3.2     RESULTS OF LTMO EVALUATION COMPLETED USING MAROS TOOL	19
      3.3     RESULTS OF LTMO EVALUATION COMPLETED USING
            THREE-TIERED APPROACH	20

4.0 SUMMARY OF DEMONSTRATIONS AT LONG PRAIRIE GROUNDWATER
   CONTAMINATION SUPERFUND SITE, MINNESOTA	 23
      4.1     FEATURES OF LONG PRAIRIE SITE	23
      4.2     RESULTS OF LTMO EVALUATION COMPLETED USING MAROS TOOL	25
      4.3     RESULTS OF LTMO EVALUATION COMPLETED USING
            THREE-TIERED APPROACH	26
5.0 SUMMARY OF DEMONSTRATIONS AT McCLELLAN AFB OU D,
   CALIFORNIA	28
      5.1     FEATURES OF MCCLELLAN AFB OU D	28
      5.2     RESULTS OF LTMO EVALUATION COMPLETED USING MAROS TOOL	30
      5.3     SUMMARY OF LTMO EVALUATION COMPLETED USING
            THREE-TIERED APPROACH	31

6.0 CONCLUSIONS AND RECOMMENDATIONS	33
      6.1     SUMMARY OF RESULTS OF MAROS EVALUATIONS AND
            THREE-TIERED APPROACH	33
      6.2     OTHER ISSUES	41
      6.3     CONCLUSIONS	41

7.0 REFERENCES	44
                                   Vlll

-------

-------
                               LIST OF FIGURES
No.                                     Title                                Page

3.1     Features of Fort Lewis Logistics Center Area	18
4.1     Features of Long Prairie Groundwater Contamination Superfund Site	24
5.1     Features of McClellan AFB OU D	29
                                          IX

-------
                                LIST OF TABLES
No-                                      Title                                  Page
1.1    Characteristics of Monitoring Programs at Three Example Sites Used in
       Long-Term Monitoring Program Optimization Demonstrations	2
2.1    Primary Features of MAROS	12
2.2    Primary Features of Three-Tiered LTMO Approach	15
3.1    Results of Optimization Demonstrations at Logistics Center Area,
       Fort Lewis, Washington	  21
4.1    Results of Optimization Demonstrations at Long Prairie
       Groundwater Contamination Superfund Site, Minnesota	26
5.1    Results of Optimization Demonstrations at McClellan AFB OU D, California	31
6.1    Summary of Optimization of Monitoring Program at
       Fort Lewis Logistics Center Area	33
6.2    Summary of Optimization of Monitoring Program at
       Long Prairie Groundwater Contamination Superfund Site	36
6.3    Summary of Optimization of Monitoring Program at McClellan AFB OU D	38

-------
            LIST OF APPENDICES (included in EPA 540-R-04-001b)
Appendix A - Concepts and Practices in Monitoring Optimization
Appendix B - Description of MAROS Tool and Three-Tiered Optimization Approach
Appendix C - Synopses of Case-Study Examples
Appendix D - Original Monitoring Program Optimization Reports by Groundwater
            Services, Inc. and Parsons
                                         XI

-------

-------
               LIST OF ACRONYMS AND ABBREVIATIONS
a
AFB
AFCEE/ERT

AFRPA
AR
ASCE
P
bgs
BRAC
CAM
CERCLA
CES
CFR
COC
COV
CR
CSM
CT
DCA
DCE
DNAPL
DQO
EGDY
ERPIMS

ESR1
ETD
EW
FS
ft/day
ft/yr
GC
GeoEAS
GIS
GWMP
gpm
GSI
GTS
GWOU
ID
IDW
IROD
LOGRAM
LTM
statistical confidence level
Air Force Base
Air Force  Center  for  Environmental Excellence/Technology  Transfer
Division
Air Force Real Property Agency
area ratio (calculated by MAROS)
American Society of Civil Engineering
statistical power
below ground surface
Base Realignment and Closure Act
chlorinated aliphatic hydrocarbon compound
Comprehensive Environmental Response, Compensation, and Liability Act
cost-effective sampling
Code of Federal Regulations
contaminant of concern
coefficient of variation
concentration ratio (calculated by MAROS)
conceptual site model
concentration trend (calculated by MAROS)
dichloroethane
dichloroethene
dense, non-aqueous-phase liquid
data-quality objective
East Gate Disposal Yard
(US Air Force) Environmental Restoration Program Information
Management System
Environmental Systems Research Institute, Inc.
extraction, treatment, and discharge
extraction well
feasibility study
feet per day
feet per year
gas chromatograph
Geostatistical Environmental Exposure Software
geographic information system
Groundwater Monitoring Plan
gallon(s) per minute
Groundwater Services, Inc.
Geostatistical Temporal/Spatial optimization algorithm
Groundwater Operable Unit
identifier
investigation-derived waste
Interim Record of Decision
revised Logistics Center  monitoring program
long-term monitoring
                                          XH

-------
                LIST OF ACRONYMS AND ABBREVIATIONS (continued)
 LTMO
 LTMP
MAROS
MCL
Mitretek
MMR
MPCA
MS
NPL
NRC
O&M
ORP
OU
Parsons
PCE
POL
QA
QC
RAO
RCRA
RI
ROC
ROD
S
SF
SVE
TCA
TCE
OSRTI
US
USAGE
U.S. EPA
VOC
 long-term monitoring optimization
 long-term monitoring program
 microgram(s) per liter
 Monitoring and Remediation Optimization System
 maximum contaminant level
 Mitretek Systems
 Massachusetts Military Reservation
 Minnesota Pollution Control Agency
 mass spectrometer
 National Priorities List
 National Research Council
 operations and maintenance
 oxidation-reduction potential
 operable unit
 The Parsons Corporation
 tetrachloroethene
 petroleum, oils, and lubricants
 quality assurance
 quality control
 remedial action objective
 Resource Conservation and Recovery Act
remedial investigation
rate-of-change parameter (calculated by MAROS)
record of decision
Mann-Kendall test statistic
slope factor (calculated by MAROS)
soil-vapor extraction
trichloroethane
trichloroethene
U.S. EPA's Office of Superfund Remediation and Technology Innovation
United  States
U.S. Army Corps of Engineers
U.S. Environmental Protection Agency
volatile organic compound
                                          Xlll

-------
                              1.0    INTRODUCTION
This report describes the results of a demonstration in which optimization techniques were used to
improve the design of long-term  groundwater monitoring programs.  The  primary objectives of
optimizing  the particular monitoring programs addressed in this study were to assess the optimal
frequency of monitoring implemented in each program, and to evaluate the spatial distribution of the
components of each monitoring network.  Two different long-term monitoring optimization (LTMO)
approaches were used in the demonstration:

   1.  The Monitoring and Remediation Optimization System (MAROS) software tool,  developed
       by Groundwater Services, Inc. (GSI) for the Air Force Center for Environmental Excellence
       (AFCEE) (2000 and 2002); and

   2.  A three-tiered approach applied by The Parsons Corporation (Parsons).

The primary goals of this demonstration were to highlight current strategies for applying optimization
techniques  to  existing long-term  monitoring (LTM) programs, and  to assist site managers in
understanding the potential benefits associated with monitoring program optimization.  The report
also presents the basic concepts underlying environmental monitoring and monitoring optimization,
so that the  discussion of particular procedures can be understood in terms of an overall monitoring
approach.  The work presented in this document was commissioned  by the  U.S. Environmental
Protection Agency's (U.S. EPA's) Office of Superfund Remediation  and Technology Innovation
(OSRTI).

1.1      PROJECT DESIGN

This project was conducted to demonstrate and assess two different LTMO approaches that can be
used to identify opportunities for streamlining groundwater monitoring programs. The project  was
designed as follows:

   •   Three  sites having  existing long-term  groundwater monitoring programs were selected as
       case-study examples for this demonstration project. The sites were required to meet minimum
       screening criteria to ensure that the available monitoring data were sufficient for the LTMO
       evaluations (refer to Sections 3, 4, and 5, and Appendix C of this report for detailed  site
       information).

   •   GSI and Parsons evaluated groundwater monitoring data from each of the three sites using
       their respective approaches, to assess whether the monitoring programs could be streamlined
       without significant loss of information.  GSI and Parsons then prepared  reports summarizing
       the results of their evaluations.

   •   The summary reports  then were provided to Mitretek Systems (Mitretek) for review. Using
       those summary reports,  Mitretek prepared this document, which summarizes the  LTMO
       evaluations and examines the results.

-------
1.2
CASE-STUDY EXAMPLES
The current LTM programs at the Fort Lewis Logistics Center, Washington (Fort Lewis), the Long
Prairie Groundwater Contamination Superfund Site in Minnesota (Long Prairie), and Operable Unit
(OU) D, McClellan Air Force Base (AFB), California (McClellan AFB OU D), were selected as case-
study example programs, because the numbers and  spatial coverage of wells,  and length of the
monitoring history  at each  site, were  judged to be adequate to  generate meaningful  results.  The
primary  characteristics of the monitoring programs at each of the three sites are presented in Table
1.1.

         Table 1.1: Characteristics of Monitoring Programs at Three Example Sites
           Used in Long-Term Monitoring Program Optimization Demonstrations
Monitoring-Program
Characteristic
Number of distinct water-
bearing units or monitoring
zones addressed by the
monitoring program
Principal contaminants'1'
Total number of wells
included in program
Total number of samples
collected (per year)
Total costc/ of monitoring
(per year)
Example Site"
Fort Lewis
2 (Upper Vashon and
Lower Vashon)
cซ-l,2-DCE,PCE,
1,1,1-TCA,TCE, VC
21 extraction wells
40 upper Vashon
monitoring wells
1 1 lower Vashon
monitoring wells
180
$90,000
Long Prairie
3 (water table [Zone A], base
of upper glacial outwash
[Zone B], lower glacial
outwash [Zone C])
cw-l,2-DCE,PCE,TCE
2 municipal supply wells
6 extraction wells
12 Zone A monitoring wells
1 5 Zone B monitoring wells
8 Zone C monitoring wells
51
$14,280
McClellan AFB OU D
2 (Zones A and B)
1,2-DCA, cw-l,2-DCE,
PCE, TCE
6 extraction wells
32 Zone A monitoring wells
1 3 Zone B monitoring wells
34
Information not provided
^ Information regarding site characteristics and the site-specific monitoring programs of the three example sites is
  presented in Section 3  (Fort Lewis), Section 4 (Long Prairie) and Section 5 (McClellan  AFB OU D), and in
  Appendices C and D.
b/ DCA = dichloroethane; DCE  = dichloroethene; PCE = tetrachloroethene;
  TCA = trichloroethane; TCE  = trichloroethene; VC = vinyl chloride.
c/ Information regarding annual monitoring program costs was provided by facility personnel.  Costs associated with
  monitoring include cost of sample collection, sample analyses, data compilation and reporting, and management of
  investigation-derived waste (e.g., purge water).
1.3
PURPOSES OF GROUNDWATER MONITORING
The U.S. EPA (2004) defines monitoring to be

      "... the collection and analysis of data (chemical, physical, and/or biological) over a sufficient
      period  of time and frequency  to determine  the status  and/or  trend in one or  more
      environmental parameters or characteristics.  Monitoring should not produce a  'snapshot in
      time' measurement, but rather should involve repeated sampling over time in order to define
      the trends in the parameters  of interest relative to clearly-defined management objectives.
      Monitoring may collect abiotic and/or  biotic data using well-defined methods and/or
      endpoints.  These data, methods,  and endpoints should be directly related to the management
      objectives for the site in question."

-------
Monitoring of groundwater systems  has  been practiced for decades.   Monitoring activities have
expanded significantly  in  recent  years,  to  assess  and address  the  problems  associated with
groundwater contamination and its environmental consequences, because the processes active within
a groundwater system, and the interactions of a groundwater system with the rest of the environment,
can be assessed only through monitoring (Zhou, 1996).

There are statutory requirements establishing the necessity for monitoring, and governing the types of
monitoring that  must be  conducted under particular  circumstances.   Passage  of the Resource
Conservation and Recovery  Act  (RCRA) in  1976, and  subsequent  promulgation of  the  first
regulations authorized under RCRA  in  1980, resulted in significant expansion  of the role  of
groundwater monitoring. RCRA and subsequent amendments include provisions for establishing
groundwater monitoring programs at all  of the  hazardous-waste  treatment,  storage, and disposal
facilities, at all of the solid-waste landfills, and at many underground storage tank facilities in the
United States.  In December 1980, the Comprehensive Environmental Response, Compensation, and
Liability Act (CERCLA) was passed, in  part to  address potential threats posed by "uncontrolled"
hazardous waste  sites.  CERCLA statutory  authority  regarding monitoring gives U.S.  EPA  the
authority to undertake monitoring to  identify  threats (42 USC ง9604[b]),  and defines removal and
remedial actions  as inclusive of any monitoring reasonably required  to  ensure  that such actions
protect the public health, welfare, and the environment (42 USC ง9601 [23] and 42 USC ง9601 [24],
respectively).  Therefore, response  actions  at  such  sites  require that monitoring  programs  be
developed and implemented to investigate the  extent of environmental contamination and to monitor
the progress of cleanup activities (Makeig, 1991).

Four inherently different types of groundwater monitoring programs can be distinguished (U.S. EPA,
2004):

   •   Characterization monitoring;

   •   Detection monitoring;

   •   Compliance monitoring; and

   •   Long-term monitoring.

Characterization monitoring is  initiated in an area where contaminants are known or suspected to be
present in environmental media (soil, air,  surface water, groundwater) as a consequence of a release
of hazardous substances. Site characterization involves delineating the nature, extent, and fate of
potential contaminants in the environment, identifying human populations or other biota ("receptors")
that could be adversely affected by exposure to those contaminants, and assessing the possibility that
the contaminants could migrate to a location where a potential receptor could come into contact with
the contaminant(s)  ("exposure  point").   Groundwater sampling is   a  critical  element of site
characterization,  as  it is necessary to establish whether site-related contaminants are migrating in
groundwater to potential exposure points.

Detection monitoring and  compliance monitoring  generally are  required for facilities that  are
regulated under  RCRA.   A groundwater-quality monitoring  program  designed  for  detection
monitoring consists of a network of monitoring points  (wells) in an uncontaminated water-bearing
unit that is  at risk of contamination from an overlying waste  facility.  If the results of periodic
sampling conducted during detection monitoring indicate that a release may have occurred, the owner

-------
 or operator of the facility must implement the next phase of groundwater monitoring - compliance
 monitoring.   During compliance  monitoring, groundwater  samples  are  collected from  locations
 designated as compliance points, and are analyzed for constituents that are known or  suspected to
 have been released.  After it has been established that a release of the type and magnitude suspected
 has occurred, a corrective-action program must be implemented (Makeig, 1991).

 During a corrective action, the owner or operator of a facility must remove, control, and/or treat the
 wastes that have caused the release, so that groundwater quality can be brought into compliance with
 established groundwater  protection   criteria.   (Additional  characterization  monitoring  may  be
 necessary during the selection of a corrective action, so that the actual extent and fate of contaminants
 in the subsurface can be assessed to the extent necessary to support remedy decisions.) Groundwater
 cleanup  criteria usually are established by the individual states, or on a site-specific basis within a
 state.  In  all cases,  the cleanup criteria must be as  stringent as, or more  stringent than, various
 standards established by the federal  government, unless such requirements are waived.   After a
 remedy has been selected and put in  place, groundwater monitoring also  is used in evaluating  the
 degree to which the remedial measure achieves its  objectives  (e.g.,  abatement of groundwater
 contaminants, restoration of groundwater quality, etc.). This type of monitoring - known as LTM -
 typically  is initiated only after a remedy has been selected and implemented, in conjunction with
 some type of corrective-action program. It usually is assumed that after a site enters the LTM phase
 of remediation, site characterization is essentially complete, and the existing monitoring network can
 be adapted, as  necessary,  to  achieve the objectives  of the LTM program (Reed et al, 2000).
 Optimization  techniques  have been  applied to the design  of  monitoring networks  for  site
 characterization, detection  monitoring, and  compliance  monitoring (Loaiciga et al.,  1992).  In
 practice, however, optimization techniques usually are applied  only to LTM programs,  as these
 programs typically provide  well-defined spatial coverage  of the area  monitored,  and have  been
 implemented for a period  of time sufficient to generate  a relatively  comprehensive monitoring
 history.

 1.4      LONG-TERM GROUNDWATER MONITORING PROGRAM OPTIMIZATION

 As of 1993, the National Research  Council  (NRC,  1993) estimated that groundwater had  been
 contaminated at  between 300,000 and 400,000 sites in the United States.  As a consequence of the
 identification of certain technology  limitations and recognition of the potentially significant  costs for
 remediating all  of these sites (approximately  $500 billion to  $1 trillion), the paradigm for
 groundwater remediation recently has shifted to some degree, from resource restoration to long-term
 risk management. This strategy change is expected to result in more contaminants being left in place
 for longer periods  of time, thereby requiring  long-term monitoring (NRC, 1999).   At  many sites,
 LTM can require decades of expensive sampling of monitoring networks, ranging in size from tens to
 hundreds of sampling locations, and resulting in costs of hundreds of thousands to millions of dollars
 per year  for sampling and data management  (Reed et al., 2000).   Development of cost-effective
 monitoring programs, or optimization of existing programs, can produce significant cost savings over
 the life of particular remediation projects.  As a consequence of the resources required to maintain a
 monitoring program for a long period of time, most monitoring optimization efforts, including the
 monitoring optimization evaluations described in this report, have focused on LTM.

 It is critical that the objectives of monitoring be developed and clearly articulated prior to initiating a
monitoring program (Bartram and Balance, 1996), or during the process of evaluating and optimizing
an existing program.  Monitoring program  objectives are dependent upon  the types of information
that will be generated, and the intended uses of that information.  The exact information needs  of

-------
particular monitoring programs usually must be established by considering the program objectives
during the planning stages or during periodic LTM program reviews.  Clearly articulated program
objectives will establish the  end-uses of monitoring data, which in turn will clarify those data that
must be collected.  The connection between the data collected by monitoring and the uses to which
those  data  are  applied is an  important element  in the success of any water-quality  monitoring
program.  Without carefully connecting the acquisition of data  with the production  and use  of
information contained within the data, there is a high probability that data collection will become an
end in itself (Ward et al., 1990).  Because site conditions, particularly in saturated media, can  be
expected to change through time, the objectives of any LTM program should be revisited and refined
as necessary during the course of the program.

Monitoring objectives fall into four general categories (U.S. EPA, 1994b and 2004; Gibbons, 1994):

   •   Identify changes in ambient conditions;

   •   Detect the movement and monitor the physico-chemical fate of environmental constituents of
       interest (COCs, dissolved oxygen, etc.) from one location to another;

   •   Demonstrate compliance with regulatory requirements; and

   •   Demonstrate the effectiveness of a particular response activity or action.

As is clear from the discussion in Section  1.3, the two primary objectives of long-term groundwater
monitoring programs are  a subset of these general objectives, and can be expressed as follow:

   •   Evaluate the long-term temporal state of contaminant concentrations at one or more points
       within or outside  of the remediation zone, as a means of monitoring the performance of the
       remedial measure (temporal objective); and

   •   Evaluate the extent to  which contaminant migration is occurring, particularly if a potential
       exposure point for a susceptible receptor exists (spatial objective).

Ultimately,  the  relative  success of any  remediation system and its  components (including  the
monitoring  program) must  be judged based on  the degree  to  which they achieve  their stated
objectives.  The most important components of a groundwater monitoring program are  the network
density (the number of monitoring wells and their relative locations) and the sampling frequency (the
number of observations or samples per unit time) (Zhou, 1996).  Designing an  effective groundwater
monitoring program involves locating monitoring points and developing a site-specific strategy for
groundwater sampling  and  analysis in  order to maximize  the amount  of relevant  information
(information required to effectively address the temporal and spatial objectives of monitoring) that
can be obtained, while minimizing incremental costs.   The efficiency of a monitoring program is
considered to be optimal if it is effectively achieving its objectives at the lowest total  cost, and/or
with the fewest possible number of monitoring locations (Reed et al., 2000).

While  several  different  LTMO methods  have been  developed and applied in recent years, this
evaluation examines the results obtained  by investigators applying two approaches in  current use.
The MAROS software tool,  developed and applied by GSI, uses parametric and non-parametric trend
analyses  to  assess  temporal  chemical concentration  trends and recommend optimal sampling
frequency, and also uses  spatial statistical techniques to identify monitoring points that potentially are

-------
generating redundant information.  The MAROS software then combines the results of the temporal
trend  analysis  and spatial  statistical  analysis,  and  uses  the  combined  results  to generate
recommendations regarding the frequency of monitoring and spatial distribution of the components of
the monitoring network.  Parsons has applied a three-tiered  approach consisting of a  qualitative
evaluation, a statistical evaluation of temporal trends in contaminant concentrations, and a spatial-
statistical analysis, to assess the degree to which the monitoring program addresses each  of the two
primary objectives of monitoring, and also to address other potentially-important considerations.  The
results of the three evaluations then  are combined and used to assess the optimal frequency of
monitoring and the spatial distribution of the various components of the monitoring network.

1.5      REPORT ORGANIZATION

The main body of this report is organized into seven sections, including this introduction:

   •   Concepts in groundwater monitoring and techniques for evaluating monitoring programs are
       discussed in Section 2; ways in which some of these techniques are  implemented in the
       MAROS software tool and in the three-tiered approach also are described briefly.

   •   Background information relevant to the current groundwater monitoring programs  at the Fort
       Lewis Logistics Center, the Long Prairie Groundwater Contamination Superfund Site, and OU
       D, McClellan AFB is reviewed in Sections 3, 4, and 5, respectively; and the summary results
       of the MAROS and three-tiered evaluations  of each monitoring program are presented in
       those Sections.

   •   Section  6  examines the results of the MAROS  and three-tiered evaluations  of the three
       monitoring   programs,  and   presents  recommendations  for   implementing   program
       improvements.

   •   References cited in this document are listed in Section 7.

Readers interested  in a summary description of the demonstration project, and its results, will  find
this information in the main body of this report (EPA 542-R-04-001a).  Readers interested in more
detailed discussions can find supporting information contained in four appendices:

       •   Concepts and  practices in  groundwater monitoring, and in monitoring  optimization, are
          discussed in detail in Appendix A.

       •   Features of the  MAROS  tool and the three-tiered LTMO approach  are described in
          Appendix B.

       •   Synopses of the MAROS  and three-tiered LTMO  evaluations of the  three monitoring
          programs are included in Appendix C.

       •   The  detailed  results of the MAROS and three-tiered LTMO evaluations of the three
          monitoring programs, as described in reports originally generated by GSI and Parsons, are
          presented in Appendix D.

The main body of the report, together with the appendices, comprise EPA 542-R-04-001b.

-------
               2.0    EVALUATION AND OPTIMIZATION OF
                  LONG-TERM MONITORING PROGRAMS
2.1       CONCEPTS IN GROUNDWATER MONITORING

Designing an effective groundwater-quality monitoring program involves selecting a set of sampling
sites,  suite of analytes, and a sampling schedule based upon one  or more monitoring-program
objectives (Hudak et al., 1993). An effective monitoring program will provide information regarding
contaminant migration and changes in chemical suites and concentrations through time at appropriate
locations, thereby enabling decision-makers to verify that contaminants are not endangering potential
receptors', and that remediation is occurring at rates sufficient to achieve remedial action objectives
(RAOs) in a reasonable timeframe.  The design of the monitoring program therefore should address
existing receptor exposure pathways, as well as exposure pathways arising from potential future use
of the groundwater.

The U.S. EPA (2004) defines six steps that should be followed in  developing and implementing a
groundwater monitoring program:

      1.   Identify monitoring program objectives.

      2.   Develop monitoring plan hypotheses (a conceptual site model, or CSM).

      3.   Formulate monitoring decision rules.

      4.   Design the monitoring plan.

      5.   Conduct monitoring, and evaluate and characterize the results.

      6.   Establish the management decision.

In this paradigm, a monitoring program is founded on the current understanding of site conditions as
documented in the CSM, and monitoring is conducted to validate (or refute) the hypotheses regarding
site conditions that are contained in the CSM. Thus, monitoring results are used to refine the CSM by
tracking changes in site conditions through time. All monitoring-program activities are undertaken to
support a management decision, established as an integral part of the monitoring program (e.g., assess
whether a selected response action is/is not achieving its objectives).

Most past efforts in developing or evaluating monitoring programs have addressed only the design of
the monitoring  plan (Step 4 in the six-step process outlined  above).  The process of designing a
groundwater monitoring plan involves four principal tasks (Franke, 1997):

       1.  Identify the volume and characteristics of the earth material targeted for sampling.

      2.  Select  the  target parameters and  analytes, including  field parameters/analytes and
          laboratory analytes.

-------
       3.  Define the spatial and temporal sampling strategy, including the number of wells necessary
          to be sampled to meet program objectives, and the schedule  for repetitive sampling of
          selected wells.

       4.  Select the wells to be sampled.

 However, this procedure  considers only the physical and chemical data that the monitoring plan is
 intended to generate, and does not completely take into account the objectives that the monitoring
 data are intended to address (Step 1, above), the decision(s) that the monitoring program is(are)
 intended to support (Step 6), or the means by which a decision will be selected (Step 3).  All of the
 six steps outlined by the U.S. EPA (2004) should be considered during the  development or evaluation
 of a  monitoring program, if that  program is  to be effective and  efficient, and also  should be
 considered during optimization of existing programs.

 Most monitoring programs have  been designed and evaluated based on qualitative insight into the
 characteristics of the hydrologic system, and using professional judgment (Zhou, 1996).  However,
 groundwater systems by nature are highly variable in space and through  time, and it is difficult or
 impossible  to account for much of  the existing variability  using  qualitative techniques.  More
 recently, other, more quantitative approaches have been developed, arising from the recognition that
 the results obtained from a monitoring program are used to make inferences about conditions in the
 subsurface on the basis of samples, and on the need to account for natural variability. The process of
 making inferences on the basis of samples, while simultaneously evaluating the associated variability,
 is the province of statistics; and to a large degree, the temporal and spatial variability of water-quality
 data currently are addressed through the application of statistical methods of evaluation, which enable
 large  quantities of data to be managed and interpreted effectively, while  the variability of the data
 also is quantified and managed (Ward et al., 1990).

 All approaches  to the design,  evaluation,  and optimization of effective groundwater monitoring
 programs must acknowledge and account for the dynamic nature of groundwater systems, as affected
 by natural phenomena and anthropogenic changes (Everett, 1980). This means that in order to assess
 the degree  to which a  particular  program is  achieving the temporal  and  spatial objectives of
 monitoring  (Section  1.4), a monitoring-program evaluation must address the temporal and spatial
 characteristics of groundwater-quality  data.  Temporal and spatial data generally are evaluated using
 temporal  and  spatial-statistical  techniques, respectively.    In  addition,  there may  be  other
 considerations that best are addressed through qualitative evaluation.

 In a qualitative  evaluation, the relative performance  of the monitoring program is assessed from
 calculations and judgments made without the use of quantitative mathematical methods (Hudak et al.,
 1993).   Multiple  factors may be considered  qualitatively  in  developing recommendations for
 continuation or cessation of monitoring at each  monitoring point.   Qualitative approaches to the
 evaluation of a monitoring program range from  relatively simple to  complex, but often are highly
 subjective.  Furthermore, the degree  to  which the program satisfies LTM objectives  may not be
 readily evaluated by qualitative methods.

 Temporal data (chemical  concentrations measured at  different points in  time) provide a means of
quantitatively assessing conditions in  a groundwater system  (Wiedemeier and Haas,  1999),  and
evaluating the performance  of a  groundwater remedy and its associated monitoring  program.  If
attenuation or removal of contaminant mass is  occurring  in  the subsurface  as a consequence of
natural processes or operation of an engineered remediation system, attenuation or mass removal will
be apparent as a decrease in contaminant  concentrations through  time at a particular sampling

                                              8

-------
location, as a decrease in contaminant concentrations with increasing distance from chemical source
areas, and/or as a change in the suite of chemicals through time or with increasing migration distance.
Conversely, if a  persistent  source  is  contributing to  groundwater contaminant  plumes  or if
contaminant  migration  is occurring,  this  may  be apparent  as  an  increase  in  contaminant
concentrations through time  at a particular sampling location, or as an increase in  contaminant
concentrations through time with increasing distance from contaminant source areas.

The temporal objective of long-term monitoring (evaluate contaminant concentrations in  groundwater
through time;  Section 1.4) can be addressed by defining trends in contaminant concentrations, by
identifying periodic fluctuations in concentrations, and by estimating long-term average ("mean")
values of concentrations (Zhou, 1996). The frequency of sampling necessary to achieve  the temporal
objective then can be based on trend detection,  accuracy of estimation of periodic fluctuations, and
accuracy of estimation of long-term mean  concentrations.   Concentration trends, periodicity, and
long-term mean concentrations typically are evaluated using statistical methods - in particular, tests
for trends, including the Student's  t-test (Zhou,  1996), regression analyses, Sen's  (1968)  non-
parametric estimator of trend slope,  and the Mann-Kendall test, are  widely applied  (Hirsch et al.,
1991).

Spatial techniques that can be applied to the design and evaluation of monitoring programs fall into
two  general categories  - simulation approaches  and ranking approaches (Hudak  et al., 1993).
Simulation approaches utilize computer models to simulate  the evolution of contaminant plumes.
The  results then are  incorporated into an optimization model which derives an optimal monitoring
network configuration  (Reed et al., 2000).  Ranking approaches utilize  weighting schemes that
express  the relative value to  the  monitoring program  of  candidate sampling  sites distributed
throughout a sampling domain (Hudak et al., 1993). The relative value of a potential monitoring site
can be ranked by assessing its spatial position relative to areas such as contaminant sources, receptor
locations, or  probable  zones  of contaminant migration.    Ranking approaches commonly use
geostatistical  methods to assist in the design, evaluation, or optimization of a monitoring network
(American Society  of  Civil  Engineering  [ASCE],  1990a  and  1990b).   General  concepts in
groundwater monitoring, and techniques used in the design/optimization of monitoring programs, are
discussed further in Appendix A.

2.2      METHODS FOR DESIGNING, EVALUATING, AND OPTIMIZING MONITORING PROGRAMS

Although monitoring network design has been studied extensively in the past, most previous studies
have addressed one of two problems (Reed et al., 2000):

      1.  Application of numerical simulation and formal mathematical  optimization techniques to
          screen monitoring plans for detection monitoring at landfills and hazardous-waste sites; or

      2.  Application of ranking methods, including  geostatistics, to augment or design monitoring
          networks for site-characterization purposes.

A number of studies  (Appendix A)  have addressed detection  monitoring  by applying global
approaches to the design of new monitoring networks.  In contrast, few investigators have formally
addressed the evaluation and optimization of LTM programs at sites having extensive monitoring
networks that were installed during site characterization.  The primary goal  of optimization efforts at
such sites is to reduce  sampling costs by eliminating data redundancy to the extent  possible.  This
type of optimization  usually is not intended to identify locations for new monitoring  wells, and it is
assumed during optimization that the existing monitoring  network sufficiently characterizes the

-------
 concentrations and distribution of contaminants being monitored.  It also is not intended for use in
 optimizing detection monitoring. Two approaches to evaluating monitoring networks - the MAROS
 tool and the three-tiered evaluation approach - were developed specifically for use in optimizing
 existing monitoring programs.  (Although formal mathematical optimization techniques have been
 applied to the problem of optimizing monitoring  programs [Appendix A], neither the MAROS tool
 nor the three-tiered approach incorporates mathematical optimization in the strict sense.  Rather, in
 subsequent   discussion,  "optimization"  refers  to  the  application  of rule-based procedures,
 incorporating statistical analysis and professional judgment, to identify possible improvements to a
 monitoring program that will continue to be effective at meeting  the two objectives of monitoring
 while addressing qualitative constraints and minimizing the necessary incremental resources.) The
 principal features of these two approaches are discussed in the following  sections, and are described
 in detail in Appendix B.

 2.3      DESCRIPTION OF MAROS SOFTWARE TOOL

 The MAROS software originally was developed  primarily for use as a tool to assist non-technical
 personnel (e.g., facility environmental managers) in evaluating and optimizing long-term monitoring
 programs (AFCEE, 2000). As an added benefit, the MAROS tool provides a convenient platform for
 the organization, preliminary evaluation, and presentation of monitoring data in graphical or tabular
 formats. In the years since its development, the performance of the MAROS software tool has been
 assessed critically ("beta tested") by applying the tool to the evaluation  and optimization of actual
 monitoring programs at a number of U.S. Air Force facilities (e.g.,  Parsons, 2000 and 2003a).  In
 response to recommendations for modifications to the MAROS software, generated as a consequence
 of the beta testing, GSI  developed MAROS Version 2, which was issued by AFCEE (2002) for
 additional testing  in 2002.   The public-domain software and accompanying  documentation are
 available free of charge for download on the AFCEE website at http://www.afcee.brooks.af.mil/er/
 rpo.htm .  All case-study  example  monitoring programs examined in  the current demonstration
 project were  evaluated and optimized using MAROS Version 2 (Sections 3.2, 4.2, and 5.2 of this
 report).

 The MAROS tool consists of a software package that  operates in conjunction with an electronic
 database environment (Microsoft Accessฎ 2000) and performs certain mathematical and/or statistical
 functions appropriate to  completing qualitative,  temporal, and spatial-statistical evaluations  of a
 monitoring program, using data that have been  loaded into the database (AFCEE, 2002).  MAROS
 utilizes  parametric temporal analyses (using linear  regression) and  non-parametric trend analyses
 (using the Mann-Kendall  test for trends) to assess the statistical significance of temporal trends in
 concentrations of contaminants of concern (COCs) (Appendix B).  MAROS then uses the results of
 the temporal-trend  analyses to develop  recommendations regarding sampling  frequency  at  each
 sampling point in  a monitoring program by applying a modified Cost-Effective Sampling (CES)
 algorithm, based on the CES method developed at Lawrence Livermore National Laboratory (Ridley
 et al, 1995).  The modified CES method  uses recent and historical  COC measurements to determine
 optimal sampling frequency.

Although the  MAROS tool primarily is used to evaluate  temporal data, it also incorporates a spatial
 statistical algorithm, based on a ranking system that utilizes a weighted "area-of-influence" approach
(implemented using Delaunay triangulation) to assess the relative value of data generated during
monitoring, and to identify the optimal locations  of monitoring points.   Formal decision logic and
methods of incorporating  user-defined secondary  lines of evidence (empirical  or modeling results)
also are provided, and can be used to further evaluate monitoring data  and  generate recommendations


                                            10

-------
for adjustments  to  sampling frequency, monitoring locations, and the density of the monitoring
network.  Additional features (moment analyses)  allow the  user to evaluate conditions and  the
adequacy  of the monitoring network  across a contaminated  site  (rather than just  at individual
monitoring locations.)

MAROS is intended to assist users in establishing practical  and cost-effective LTM goals for a
specific site, by

   •   Identifying the COCs at the site;

   •   Determining whether temporal trends in groundwater COC concentration data are statistically
       significant;

   •   Using identified temporal trends to evaluate and optimize the frequency of sample collection;

   •   Assessing the extent  to which contaminant migration is occurring, using temporal-trend and
       moment analyses;

   •   Evaluating the relative importance of each well in a monitoring network, for the purpose of
       identifying potentially-redundant monitoring points;

   •   Identifying those wells that are statistically most relevant to the current sampling program;

   •   Evaluating whether additional monitoring points are needed to achieve monitoring objectives;

   •   Providing indications of the overall performance of the site remediation approach; and

   •   Assessing whether the  monitoring  program is sufficient to  achieve program objectives on
       local or site-wide scales.

As with any approach to LTM program optimization, successful application of the MAROS  tool to
the site-specific  evaluation of a monitoring program is completely dependent upon the amount and
quality of the available  data (e.g., data requirements for a temporal trend analysis include a suggested
minimum  of six separate sampling events at an individual sampling point,  and a spatial analysis
requires sampling results from a minimum of six different sampling locations). It also is necessary to
develop an adequate CSM (Section 2.1), describing site-specific conditions (e.g., direction and rate of
groundwater movement, locations of contaminant sources and potential receptor  exposure points)
prior to applying the MAROS tool.  In particular, the nature and  extent of contaminants in  the
subsurface at the site must be adequately characterized and delineated before the monitoring program
can be optimized.

MAROS  is designed  to  accept  data  in  any  of  three  formats:   text  files  in  U.S.  Air Force
Environmental Restoration Program Information Management System (ERPIMS) format, Microsoft
Accessฎ files, or Microsoft  EXCELฎ files.  Prior to conducting a monitoring-program  evaluation,
spatial and temporal data are loaded into a database, to include well identifiers (IDs), the sampling
date(s) for each  well, COCs, COC concentrations detected at  each well sampled on each sampling
date, laboratory detection limits for each COC,  and any quality assurance/quality control (QA/QC)
qualifiers  associated with sample collection or analyses.  The spatial analysis also requires that
geographic coordinates (northings and eastings, referenced to some common datum) be supplied for
each well.

                                             11

-------
 Because MAROS can be used to evaluate the spatial and temporal characteristics of a maximum of
 five COCs in a single simulation, one or more COCs must be removed from data sets containing
 more than five COCs, or the data set must be split, so that only five COCs are included in a single
 simulation.   MAROS  is  capable of evaluating  a maximum of 200 monitoring points in each
 simulation.  Prior to applying MAROS to the evaluation of a monitoring network comprising more
 than 200 monitoring points, those monitoring locations providing relatively little information (or
 information that  is not  compatible with the other points in the  network)  can  be identified using
 qualitative methods and eliminated from the evaluation.  As an  alternative, a monitoring network
 comprising more than 200 monitoring points could be  divided into  subsets,  each subset of the
 network could be evaluated using MAROS, and the results of the evaluations then could be combined
 to generate recommendations for the entire network.

 After COCs  have been identified, and the monitoring points in the  network to be used in the
 evaluation have been selected, the MAROS evaluation and optimization of a monitoring program is
 completed in two stages:

   •   A preliminary evaluation  of plume stability is completed for the monitoring network, and
       general recommendations for improving the monitoring program are produced; and

   •   More-detailed temporal  and spatial evaluations then are completed for individual monitoring
       wells, and for the complete monitoring network.

 In general, the MAROS tool is  intended for use  in evaluating  single-layer groundwater systems
 having relatively simple hydrogeologic characteristics (GSI, 2003a).  However,  for a multi-layer
 groundwater system, the user could analyze those components of the monitoring network completed
 in individual layers, during separate evaluations.

 The primary features of MAROS, and the ways in  which it addresses the qualitative, temporal, and
 spatial aspects of environmental monitoring data, are summarized in Table 2.1.  Additional details
 regarding the MAROS software tool,  its functionality, capabilities, and methods of application, are
 presented in Appendix  B.  Details regarding specific examples of its application are presented in
 Appendix D.

                          Table 2.1: Primary Features of MAROS

                                        Infrastructure

 The MAROS  tool is a public-domain software package that operates in conjunction with an electronic
 database environment (Microsoft™  Accessฎ 2000) and performs certain mathematical  and/or
 statistical functions appropriate to completing qualitative, temporal, and spatial-statistical evaluations
 of a monitoring program, using data that have been loaded into the database.	
 The MAROS  software, and accompanying documentation, are available  for download free of charge
 from the AFCEE website.	
Although relatively sophisticated applications of the MAROS tool  are possible, many of the steps in
the evaluation are straightforward, and can be completed by a user unfamiliar with statistical concepts
and practice. In such instances, the recommendations generated by  application of the software should
be reviewed by a more experienced individual.	
                                            12

-------
                          Table 2.1: Primary Features of MAROS
                                  Qualitative Evaluation
Qualitative information is used to make  preliminary recommendations  for the entire monitoring
program rather than for individual wells.  Qualitative considerations also may be applied to develop
recommendations regarding sampling  frequency  at various  stages  throughout the evaluation,
depending upon whether the available data are sufficient to be used reliably by the MAROS statistical
tools.	
                                   Temporal Evaluation

MAROS includes a linear-regression analysis and a Mann-Kendall test to determine whether COC
concentrations at a particular well display a statistically-significant temporal trend.  MAROS also
calculates  the coefficient of variation (COV)  for each statistical test, for use in evaluating whether
COC concentrations displaying no trend at a particular well have a large degree of "scatter" or can be
considered "Stable."             	
MAROS requires the results of a minimum of six sampling events to complete a temporal analysis at
an individual well.	
MAROS  uses the  results  of the temporal-trend analyses to  develop recommendations  regarding
optimal sampling frequency at each sampling location, by applying a modified CES algorithm.	
MAROS uses the results of moment analyses to assess the overall stability of a plume, and can
perform a data-sufficiency analysis, to  assess whether  RAOs have  been/are being  achieved  at
individual  wells and at designated compliance points.	
MAROS assigns the value of the reporting limit (or some fraction thereof) to samples having a
constituent concentration below the reporting limit.     	

                                    Spatial Evaluation

MAROS uses an inverse-distance weighting  algorithm to estimate the concentrations of COCs  at
individual  monitoring locations.        	
MAROS uses a "slope factor", calculated based on the standardized difference between the measured
and estimated concentrations at a particular location, together with the average concentration ratio
and area ratio, to determine the relative value of information obtained at individual monitoring points.
MAROS requires sampling results from a minimum of six different sampling locations to complete a
spatial analysis.	
The  spatial-evaluation algorithm implemented  in  MAROS  can  be used to assess the  spatial
distribution of multiple COCs simultaneously.  	

                                          Overall

MAROS  uses  the  results  of the temporal  evaluation  to generate  recommendations  regarding
monitoring frequency, and uses the results of the spatial evaluation to identify potentially redundant
monitoring points.  Qualitative information is considered only during the preliminary evaluation  of
the monitoring program.   A MAROS evaluation  can  be conducted using a maximum  of five
constituents.	
A monitoring program evaluation completed using MAROS  may cost  in the range of  $6,000  to
$10,000. depending upon the size  of the monitoring program.	
                                            13

-------
 2.4      DESCRIPTION OF THREE-TIERED APPROACH

 As described by Parsons (2003b, 2003c, and 2003d), a three-tiered LTMO evaluation is conducted in
 stages to address each of the objectives and considerations of monitoring:  a qualitative evaluation
 first is completed, followed in succession by temporal and spatial evaluations. At the conclusion of
 each stage (or "tier") in the evaluation, recommendations are generated regarding potential changes in
 the temporal frequency of monitoring, and/or whether to retain or remove each monitoring  point
 considered in the evaluation.  After all three stages of evaluation have been completed, the results of
 all of the analyses are combined and  interpreted, using a decision algorithm, to generate  final
 recommendations for an effective and efficient LTM program.

 In the qualitative evaluation, the primary elements of the monitoring program (numbers and locations
 of wells, frequency of sample  collection, analytes specified in  the  program) are examined, in  the
 context of site-specific conditions, to ensure that the program is capable of generating appropriate and
 sufficient information  regarding plume migration and changes in chemical concentrations through
 time. Criteria used in the qualitative evaluation are discussed in detail in Appendix B, and examples
 of application of these criteria are presented in the detailed case-history examples (Appendices D-l,
 D-2, and D-3). In the temporal  evaluation, the historical monitoring data for every sampling point in
 the monitoring program are examined for temporal trends in COC concentrations, using the Mann-
 Kendall test (Appendices A and B).

 After the Mann-Kendall test for trends has been completed for all COCs at all monitoring points,  the
 spatial distribution of temporal trends in COC concentrations is used to evaluate the relative value of
 information obtained from periodic monitoring at each monitoring well by considering the location of
 the well within (or outside of) the horizontal extent of the contaminant plume, the location of the well
 with respect to potential receptor exposure points, and the presence or absence of temporal trends in
 contaminant concentrations in samples collected from the  well. In the third stage of the three-tiered
 evaluation, spatial statistical techniques are used to assess the relative value of information (in  the
 spatial sense) generated by sampling at each monitoring point in the network.  COC concentration
 data collected during  a single sampling event  are used to identify those areas having the greatest
 uncertainty associated with the estimated extent and concentrations of COCs in groundwater.  At  the
 conclusion of the spatial-statistical evaluations, each  well is ranked, from those providing the least
 information to those providing the most information, based on the amount of information the well
 contributed toward describing the spatial distribution of the COC being examined. Wells providing
 the least amount of information represent possible  candidates for removal from the monitoring
 program, while wells  providing the greatest amount  of information represent sampling points that
 probably should be retained in any refined version of the monitoring program.

 At each stage in the three-tiered evaluation, monitoring points that provide relatively greater amounts
 of information regarding the occurrence and distribution of COCs in groundwater  are identified, and
 are distinguished from those monitoring points that provided relatively lesser amounts of information.
 After all  three stages have been completed, the results of the three stages are combined to generate a
refined monitoring program that potentially can provide information sufficient to address the primary
 objectives of monitoring at the site, at reduced cost.

The qualitative evaluation can be completed by a competent hydrogeologist.  The temporal evaluation
can be completed using commercially-available statistical software packages having the capability of
using non-parametric methods (e.g., the Mann-Kendall test) to examine time-series data for trends.
The spatial-statistical evaluation can be completed by a user familiar with geostatistical concepts, and
having access to a standard geostatistical software package (e.g., the Geostatistical Environmental

                                             14

-------
Exposure Software [GeoEAS;  Englund and Sparks, 1992],  GSLIB [Deutsch and Joumel, 1998] or
similar  package).  In  practice, data manipulation, temporal and spatial  analyses, and  graphical
presentation of results  are simplified, and the quality of the results is enhanced, if a commercially
available  geographic  information  system  (GIS)  software  package   (e.g.,  ArcViewฎ  GIS)
(Environmental Systems  Research Institute, Inc. [ESRI], 2001) with spatial-statistical capabilities
(e.g., Geostatistical Analyst™,  an extension to the ArcViewฎ GIS software package) is utilized in the
LTMO evaluation.

As with the MAROS tool, the site-specific evaluation of a monitoring program using the three-tiered
approach  is directly  dependent upon the amount and quality of the available data.   The primary
features of the three-tiered approach, and the ways in which it addresses the qualitative, temporal, and
spatial aspects of environmental monitoring data, are summarized in Table 2.2.  Additional details
regarding the three-tiered approach, its  functionality, capabilities, and methods of application, are
presented in Appendix B.  Details regarding specific examples of its application are presented in
Appendix D.

               Table 2.2: Primary Features of Three-Tiered LTMO Approach

                                       Infrastructure

A three-.tiered  LTMO evaluation  is conducted in stages  to address each  of the  objectives and
considerations of monitoring:  a qualitative evaluation first  is completed, followed in succession by
temporal  and  spatial evaluations.  At the conclusion of each stage (or "tier") in the evaluation,
recommendations are  generated  to retain or remove each monitoring point considered in the
evaluation. After all three stages have been completed, the results of all of the analyses are combined
and interpreted, using a decision algorithm, to generate final recommendations for an effective and
efficient LTM program.	
No software is required for the qualitative evaluation.  The temporal evaluation can be completed
using commercially-available  statistical  software  packages having  the capability  of using non-
parametric methods to  examine time-series data for trends.  The spatial-statistical evaluation can be
completed using a standard geostatistical software package.  Data manipulation, temporal and spatial
analyses,  and graphical  presentation of results  are  simplified, and the quality  of the  results is
enhanced, if a commercially-available GIS software package with spatial-statistical  capabilities is
used.	
Completion of the qualitative evaluation  requires a  competent hydrogeologist and an adequate CSM.
The temporal and spatial-statistical evaluations require a user familiar with non-parametric statistical
and geostatistical concepts, having access to appropriate software.	

                                   Qualitative Evaluation

Qualitative information is evaluated to determine optimal sampling frequency and removal/inclusion
of each well in the monitoring program based on all historical monitoring results.	

                                   Temporal Evaluation

The three-tiered temporal statistical analysis includes classifications for wells at which a particular
COC has never been detected at a  concentration greater than the reporting limit ("Not Detected") and
for wells  at which a particular COC consistently has been  detected at concentrations less than the
practical quantitation limit ("< PQL").	
The three-tiered approach requires the results of a minimum of four sampling events (if seasonal
effects are not present)  to complete a temporal analysis at an  individual well.	


                                             15

-------
                Table 2.2: Primary Features of Three-Tiered LTMO Approach

                              Temporal Evaluation (continued)

 The three-tiered approach uses the results of the temporal evaluation to develop recommendations
 regarding sampling frequency, and to identify wells to be retained in or removed from the program.
 The approach uses a formal decision framework to develop these recommendations.	
 The three-tiered  approach uses the results of the  temporal  evaluation to assess trends  only at
 individual monitoring points.	
 The three-tiered approach assumes that monitoring points having historical results with "No Trend"
 are  of limited value, while  MAROS  treats a monitoring point  having  "No Trend" in  COC
 concentrations similar to a monitoring point having an "Increasing Trend" in concentrations.	

                                     Spatial Evaluation

 The three-tiered  approach applies geostatistics to  estimate  the  spatial  distribution of COCs.
 Application of this procedure depends upon the development of an appropriate semi-variogram.	
 The three-tiered  approach uses changes in the median kriging error generated  during different
 realizations to rank the relative value of information obtained at individual monitoring  points.  The
 relative ranking (from "Provides Most Information" to "Provides  Least Information") is used to
 develop  recommendations regarding  which  wells should be  retained  in or  removed from  the
 monitoring program.	^^
 The three-tiered  approach requires sampling results from a  minimum  of 15 different sampling
 locations to complete a spatial analysis.	
 Currently, only a single "indicator COC" (typically, the COC that has been detected at the greatest
 number of separate monitoring locations) is used in the three-tiered spatial evaluation.	

                                          Overall

 The three-tiered approach combines the results of the qualitative, temporal, and spatial evaluations to
 generate overall recommendations regarding optimal  sampling frequency and number of monitoring
 points  in a monitoring program.  Although  the spatial evaluation  stage is restricted to a single
 constituent, the qualitative and temporal stages of the  evaluation  can be applied to an unlimited
 number of constituents.	
 A monitoring program evaluation completed using the three-tiered approach may cost in  the range of
 $6,000 to $10,000, depending upon the size of the monitoring program.	

 2.5      CASE-STUDY EXAMPLES

 The MAROS tool and the three-tiered approach each were applied to the evaluation and optimization
 of existing  groundwater monitoring programs at three different sites - the Logistics Center at Fort
 Lewis, Washington, the Long  Prairie Groundwater Contamination Superfund Site in Minnesota, and
 OU D at the former McClellan AFB, California.  Pertinent features of the groundwater monitoring
programs for each site, and the results of the MAROS evaluation and the three-tiered evaluation of
the monitoring program at each site, are summarized in the following sections.
                                             16

-------
 3.0   SUMMARY OF DEMONSTRATIONS AT LOGISTICS CENTER
                    AREA, FORT LEWIS, WASHINGTON
An overview of features pertinent to the groundwater monitoring program at the Logistics Center
area, Fort Lewis, Washington is provided in this section, together with a summary of the results of the
LTMO demonstrations.  The features of the site, and of the monitoring-program evaluations that were
completed using the MAROS tool and the three-tiered approach, are summarized in Appendix C, and
are described in detail in Appendix D-l.

3.1      FEATURES OF FORT LEWIS LOGISTICS CENTER

The  Fort Lewis Military Reservation  is  located near the southern end of Puget Sound in Pierce
County, Washington, approximately 11 miles south of Tacoma and 17 miles  northeast of Olympia.
The  Logistics  Center occupies  approximately 650 acres of the Fort Lewis  Military Reservation.
Process wastes were disposed of at several on- and off-installation locations, including the East Gate
Disposal  Yard  (EGDY), located southeast of the Logistics Center.  Between  1946 and 1960, waste
solvents  (primarily trichloroethene  [TCE])  and petroleum, oils, and lubricants (POL)  generated
during cleaning, degreasing, and maintenance operations were disposed of in trenches at the EGDY,
resulting  in the introduction of contaminants to soils and groundwater at and downgradient from this
former landfill.  The dissolved chlorinated solvent  plume that originates at the EDGY extends
downgradient across the entire width of the Logistics Center,  and beyond the northwestern facility
boundary to the southeastern shore of American Lake (Figure 3.1).  The program that was developed
to monitor the concentrations and extent of contaminants in groundwater in the vicinity of, and
downgradient from the EDGY, and to assess the performance of remedial systems installed to address
contaminants  in  groundwater,  was the  subject  of the  MAROS  and three-tiered  evaluations
(Appendices C and D).

TCE has been identified as the primary COC in groundwater beneath the Logistics Center, based on
its widespread detection in wells across the site.  Other COCs  in groundwater include cis-1,2-
dichloroethene (DCE),  tetrachloroethene  (PCE),  1,1,1-trichloroethane  (TCA), and vinyl chloride
(VC).  TCE, DCE, and TCA have been detected consistently in many wells, while PCE and VC have
been detected only sporadically, in a few wells. The former waste-disposal trenches at the EGDY are
the apparent source of these chlorinated aliphatic hydrocarbon compounds (CAHs) in groundwater
beneath and downgradient from the Logistics Center.

Beginning in December 1995, groundwater monitoring was conducted at the Logistics Center on a
quarterly basis. Under the monitoring program, 38 monitoring wells and 21 groundwater extraction
wells were sampled, resulting in 236 primary samples per year (59 wells each sampled four times per
year) (Appendices C and D). The primary objectives of the monitoring program, as expressed in the
monitoring plan, are to confirm that the groundwater extraction systems are preventing the continued
migration of contaminants in groundwater to downgradient locations, to evaluate potential reductions
in contaminant concentrations through time, to assess temporal changes in the lateral and vertical
extent of contaminants in groundwater, and to assess the rate of removal of contaminant mass from
the subsurface.
                                          17

-------
American Lake
ป.              •U!*'W4&73
 *%  LC-111b02U  VUU.fH
    " - LX-1 10, f1*-3 "
      * „   eTix.? 11
 • FL3-C11   ป ป _   '
        -..  ปIW !"ซ
         - .^^^    *

              ^^\  ป '
                               1-15 2.3  *~ *\
                              12* 0 2U I .JI.C-16 9 3
                                                           • LC.132 MO  ,LC<6. ซ-
                                                                  ft      \ ซLC-19.. ITO--5?
                                                                                                 1240    7          '
                                                                                                 LX.16:150  •ซ  /Lr'137ซ 'Vf
                                                                                                 LX-19.
                                                                                                LC-136a;
                                                                                                 , LC-136b: 130-
                                                                                                     LX-17:2100
                                                                                       • LCป53 190       LC-64a: 290(
                                                                                                 • LC-51  160
                                                                                                                            Legend
                                                                                                                            Vashon Aquifer Wells
                                                                                                                               o    Extraction Well
                                                                                                                               •    Upper Vashon Monitoring Well
                                                                                                                               •    Lower Vashon Monitoring Well
                                                                                                                            [>..  '•,,-,] East Gate Disposal Yard
                                                                                                                            !      J Approximate extent of Upper Vashon TCE plume (5ug/L isopleth)
                                                                                                                                 ป Approxhrale direction ot groundivater flow In Vashon Aquifer
                                                                                                                                   Groundwater sink

                                                                                                                            (TCE plume boundaries from USAGE. 2001)
                                                                                                                                                                II
                                                                                                                                                               A
                                                                                                                                                0   500 1,000     2,000    3ฃ00
                                                                                                                                                                                eet
                                   Figure 3.1:  Features of Fort Lewis Logistics Center Area (after Parsons, 2003b)
                                                                                  18

-------
Two distinct  monitoring zones are recognized in the  groundwater system beneath  the Logistics
Center area.  Most groundwater monitoring wells are completed in the upper monitoring zone (the
"Upper Vashon" zone); relatively few monitoring wells  are completed in the lower monitoring zone
(the  "Lower  Vashon" zone).  An LTMO evaluation  of the groundwater extraction system and
associated monitoring network at the Logistics Center was completed by the Fort Lewis project team
in May 2001  (Appendices C and D); the refined monitoring program generated as a result of this
evaluation is  known as the  LOGRAM program.   Based on the results of the LOGRAM LTMO
evaluation, 24 monitoring wells were added to the Logistics Center monitoring program, and 11
previously  sampled monitoring wells  were  removed  from the program (a net  increase of 13
monitoring wells); sampling frequencies generally were reduced.   The revised Logistics Center
monitoring program (LOGRAM), which was initiated in December 2001, includes 72 wells -- 51
monitoring wells (29 wells sampled quarterly, 3 wells sampled semi-annually, and 19 wells sampled
annually), and 21  extraction  wells (6 wells sampled quarterly and 15 wells sampled annually).  The
reduction in sampling frequency at a number of wells produced a net reduction in the total number of
primary samples collected and analyzed per year, from 236 samples to 180 samples.  All  samples
from the monitoring and extraction wells are analyzed for volatile organic compounds (VOCs) using
U.S. EPA Method SW8260B.

3.2      RESULTS OF LTMO EVALUATION COMPLETED USING MAROS TOOL

Because  extensive  historical data  were not  available  for  the new   wells  installed  during
implementation of the current LOGRAM monitoring program, the MAROS  tool was used to evaluate
data from the 59 wells that  remained in the monitoring program in September 2001 (21  extraction
wells and  38 groundwater  monitoring  wells;  Appendix C)  included in  the original  monitoring
program, and was not used to evaluate the LOGRAM program. The detailed results of the MAROS
evaluation  of the groundwater monitoring program at the Fort Lewis Logistics Center area are
presented in Appendices C (Section C1.5) and D-l, and are summarized in this subsection.

Prior to  the  evaluation,  five wells that potentially would  provide  "redundant" information were
identified on the basis of qualitative considerations (Appendices C and D-l); these were not included
in the moment analysis or in the spatial evaluation.  Historic monitoring results from all monitoring
and extraction wells were included in the temporal evaluation.  However,  results from groundwater
extraction wells were not used in the spatial  evaluation; and the results from two monitoring wells
completed  in the  lower  part of the Lower  Vashon subunit also were excluded from the spatial
evaluation, because these two wells were considered to be within a different monitoring zone than the
other monitoring wells (Appendix D-l).

Application  of the  Mann-Kendall and  linear-regression  temporal  trend  evaluation  methods
(Appendices  B and C) indicated that the extent and concentrations of TCE  in groundwater at the
Logistics Center source area (the EGDY) probably are decreasing (GSI, 2003a). TCE concentrations
in groundwater at most of the extraction wells located northwest of the EGDY source area also are
probably decreasing.  The results of the moment analysis indicated that the location of the center of
mass of the plume has remained essentially unchanged, and that the extent of TCE in groundwater
has decreased over time, providing further evidence that the plume is stable under current conditions.
The evaluation of overall plume stability indicated that the extent of TCE in groundwater is stable or
decreasing, resulting in the recommendation that a monitoring strategy appropriate for a "Moderate"
design category be adopted (Appendices C and D).
                                          19

-------
 The results of detailed spatial analyses using the Delaunay method (Appendices C and D) indicated
 that 8 monitoring wells could be removed from the original monitoring program (which included 38
 monitoring wells) without significant loss of information.  However, the accompanying well-
 sufficiency analysis  indicated that there  is  a  high  degree of uncertainty  in  predicted TCE
 concentrations in six  areas within the network where the available historical sampling information
 may be inadequate; new monitoring wells were recommended for installation in these six areas (GSI,
 2003a).  These six locations recommended for  installation of new wells correspond to six wells that
 had been installed and were being monitored in conjunction with the LOGRAM program (Appendix
 C).  All groundwater extraction wells were recommended for retention in the refined monitoring
 program.  The results of the sampling-frequency optimization analysis completed using MAROS
 (Appendices C and D) indicated that most  wells in the monitoring network  could be sampled less
 frequently  than in the current (LOGRAM) monitoring program.  The results  of the data-sufficiency
 evaluation, completed using power-analysis methods, indicated that RAO concentrations of TCE in
 groundwater have nearly been achieved at the compliance boundary.

 The optimized monitoring program generated  using the MAROS tool includes 57 wells, with  19
 sampled quarterly, 2 sampled  semiannually, 30 sampled  annually, and  6  sampled  biennially
 (Appendices C and D). Adoption of the optimized program would result in collection and analysis of
.113 samples per year,  as compared with collection and analysis of 180 samples per year in the current
 LOGRAM monitoring program (Table 3.1) and 236 samples per year in  the original sampling
 program. Implementing these recommendations could lead to a 37-percent reduction in the number
 of samples collected and analyzed annually, as compared with the current LOGRAM program, or a
 52-percent reduction in the number of samples collected and analyzed, as compared with the original
 program (Table 3.1).  Assuming a cost per sample of $500 for collection and chemical analyses
 (based on information provided by the U.S.  Army Corps of Engineers [USAGE, 2001]), adoption of
 the monitoring program as optimized using the  MAROS tool is projected to  result in savings  of
 approximately $33,500  per year as compared with the  LOGRAM program  (Table  3.1).   (The
 estimated cost per sample is based on information provided by facility personnel in conjunction with
 efforts to estimate potential cost savings resulting from optimization of the monitoring program, and
 includes costs associated with sample collection and analysis, data compilation and reporting, and
 handling of materials generated as investigation-derived waste [IDW] during sample collection [e.g.,
 purge  water].)   The  optimized program remains  adequate  to  delineate the  extent of TCE  in
 groundwater, and to monitor changes in the plume over time (GSI, 2003a).

 3.3      RESULTS OF LTMO EVALUATION COMPLETED USING THREE-TIERED APPROACH

 The three-tiered approach  was used to evaluate the original monitoring  program at the Logistics
 Center area (which included 59 wells), and also was used to evaluate the current LOGRAM program
 (which includes 72 wells). Because extensive  historical data were not available for the new wells
 included in the  LOGRAM  program,  temporal  analyses were  not  used in evaluating  the new
 LOGRAM  wells - only qualitative and  spatial evaluations  of that program were completed for these
 wells,  and as a consequence, the results  of evaluation of the  two programs  are  not directly
 comparable.  The detailed results  of  the three-tiered evaluation of  the groundwater monitoring
 programs at the Fort Lewis Logistics Center area are presented in Appendices C (Section C1.6) and D
 (Appendix D-l), and are summarized in this subsection.
                                          20

-------
                    Table 3.1:  Results of Optimization Demonstrations at
                       Logistics Center Area Fort Lewis, Washington
Monitoring-Program Feature
Wells sampled quarterly
Wells sampled semi-annually
Wells sampled annually
Wells sampled biennially
Wells sampled every 3 years
Total wells included in LTM program
Total number of samples (per year)
Annual costh/ of LTM program
Monitoring
Original
(prior to
December
2001)
59
—
—
~
~
59
236
$118,000
Current
(LOGRAM,
after December
2001)
35
3
34
~
—
72
180
$90,000
ป Program"'
Original
Refined using
MAROS
19
2
30
6
-
57
113
$56,500
Refined using
3-Tiered
Approach
16
7
16
14
15
69
107
$53,500
  Details regarding site characteristics and the site-specific monitoring programs at the Logistics Center area, Fort Lewis,
  Washington, are presented in Appendices C and D-l.
w  Information regarding annual monitoring program costs was provided by facility personnel. Costs associated with
  monitoring include cost of sample collection, sample analyses, data compilation and reporting, and management of
  investigation-derived waste (e.g., purge water).

The primary COCs (TCE, PCE, cw-l,2-DCE, and VC)  were considered in the qualitative and
temporal stages of the three-tiered evaluation; however, because TCE has been the most frequently
detected COC in groundwater at the Fort Lewis Logistics Center area, the spatial-statistical stage of
the three-tiered evaluation of the monitoring program used only the results of analyses for TCE in
groundwater  samples.  Furthermore, because the Upper Vashon  and Lower Vashon subunits are
considered to be separate monitoring zones (Section 3.1), and the results of only a single water-
bearing unit or monitoring zone can be considered in the spatial-statistical evaluation, the spatial-
statistical  evaluation  was  conducted  using  the  sampling results  from  those  monitoring  wells
completed in the Upper Vashon subunit only.  Sampling results from groundwater extraction  wells
were not used  in  the  spatial-statistical  evaluation;  however,  sampling  results from  all  wells
(groundwater extraction wells,  and groundwater monitoring wells completed  in the Upper Vashon
and Lower Vashon subunits) were used in the qualitative and temporal evaluations.

The results of the three-tiered evaluation  indicated that 6 of the 72 existing wells could be  removed
from the LOGRAM groundwater LTM program with little  loss of information (Parsons, 2003b), but
also indicated that 2 existing wells that are not currently sampled should be included in the program,
and that one new well should be installed and monitored. A refined monitoring program (Appendices
C and D), consisting of 69 wells, with 16 wells sampled quarterly, 7 wells sampled semi-annually, 17
wells sampled annually, 14 wells sampled biennially, and 15 of the extraction wells sampled every 3
years (Table 3.1), would be adequate to  address the two primary objectives of monitoring. If this
refined monitoring program were adopted, 107 samples per year would be collected and analyzed, as
compared with the collection  and analysis  of 180  samples  per year in the current  LOGRAM
monitoring program and  236  samples per year in the original  sampling program.   This would
represent  a  40-percent reduction in the  number  of samples  collected  and analyzed annually, as
compared with the LOGRAM program, or a 55-percent reduction in the number of samples collected
and analyzed, as compared with the  original program. Assuming a  cost per sample  of $500 for
                                           21

-------
collection and chemical analyses, adoption of the monitoring program as optimized using the three-
tiered approach is projected to result in savings of approximately $36,500 per year as compared with
the LOGRAM program, or $64,500 per year as compared with the original monitoring program
(Table 3.1).  Additional cost savings potentially could be realized if groundwater samples collected
from select wells (e.g., upgradient wells, and wells along the lateral plume margins) were analyzed
for a short list of halogenated VOCs using U.S. EPA Method SW8021B instead of U S  EPA Method
SW8260B (Parsons, 2003b).
                                         22

-------
     4.0    SUMMARY OF DEMONSTRATIONS AT LONG PRAIRIE
        GROUNDWATER CONTAMINATION SUPERFUND SITE,
                                   MINNESOTA
An overview of features  pertinent to the groundwater monitoring program  at the Long Prairie
Groundwater Contamination Superfund Site, Minnesota (Long Prairie site) is provided in this section,
together with a summary of the results of the LTMO demonstrations. The features of the site, and of
the monitoring-program evaluations that were completed using the MAROS tool and the three-tiered
approach, are summarized in Appendix C, and are described in detail in Appendix D-2.

4.1     FEATURES OF LONG PRAIRIE SITE

The town of Long Prairie, Minnesota is a small farming community located on the east bank of the
Long Prairie River in central Minnesota. The Long Prairie site comprises a 0.16-acre source area of
contaminated soil that has generated a plume of dissolved CAHs in the drinking-water  aquifer
underlying the north-central part of town.  The source of contaminants in groundwater was a dry-
cleaning establishment, which operated from 1949 through 1984 in the town's commercial district.
Spent dry-cleaning solvents, primarily PCE, were discharged into the subsurface via a  french drain.
The  subsequent migration of contaminants through the vadose zone to groundwater produced a
dissolved CAM plume that has migrated to the north a distance of at least 3,600 feet from the source
area, extending beneath a residential neighborhood and to within 500 feet of the Long Prairie River.

The  plume of contaminated groundwater currently is  being  addressed by  extraction  of CAH-
contaminated groundwater via nine extraction wells, treatment of the extracted water, and discharge
of treated water to the Long Prairie River.  The performance  of the groundwater extraction system is
monitored by means of periodic sampling of monitoring wells and water-supply wells, and routine
operations and maintenance (O&M) monitoring of the extraction and  treatment systems.  The
program  that  was established to monitor the  concentrations and  extent  of contaminants  in
groundwater in the vicinity of, and  downgradient from the PCE source area, and to assess the
performance of the OU1 groundwater extraction, treatment, and discharge (ETD) system, was the
subject of the MAROS and three-tiered evaluations (Appendices C and D).

PCE and its daughter products TCE and cis-l,2-DCE are the primary COCs at  the Long Prairie site,
and have been detected through a volume of groundwater about 1,000 feet wide, which extended  (in
October 2002) from the source area, approximately 3,200 feet downgradient to the northwest (Figure
4.1).  VC also has been detected in groundwater samples, although at few locations  and at lower
concentrations than other CAHs.

Groundwater conditions are monitored periodically at the Long Prairie site, to evaluate whether the
groundwater ETD system is effectively preventing the continued migration of CAH contaminants in
groundwater to downgradient locations, and to confirm that contaminants are not migrating to the
water-supply wells of the municipality of Long Prairie.  Several of the monitoring locations include
wells installed in clusters, with each well in a  cluster completed at a different depth.  Groundwater
monitoring wells, extraction wells, and municipal water-supply wells are included in the monitoring
program.  A total  of 44 wells  in the Long  Prairie  area  were sampled during  the most recent
                                         23

-------
                                      '18A: ND
                                         : 38
   *
 WW1A: ND
  RW1B:
MW10A:
                  •MW5B: ND

                                :22

                        MW13• flow direction under non-
                                   pumping conditions (Barr, 2002)

                            Oct 2002 PCE concentrations in ug/L
                            ND = PCE not detected at well
                                                                                                                   I Feet
                                                                                                250    500         1,000
                                                                                                 1 inch equals 500 feet
                                                            N
                                                           A
     Figure 4.1: Features of Long Prairie Groundwater Contamination Superfund Site (after Parsons, 2003c)

                                                        24

-------
monitoring event (October 2002) for which sampling results are available.  Approximately one-half
of the wells sampled during October 2002 are sampled routinely in conjunction with the groundwater
monitoring program.  The "current" (2002) 27-well monitoring program  at the Long Prairie site
includes the 18 monitoring wells, 6 active groundwater extraction wells, and one inactive extraction
well sampled  during scheduled monitoring events  in 2000 and  2001,  together with two nearby
municipal-supply wells (Appendices C and D). All samples from the monitoring and extraction wells
are analyzed for VOCs using U.S. EPA Method SW8021B.

4.2       RESULTS OF LTMO EVALUATION COMPLETED USING MAROS TOOL

The detailed results of the MAROS evaluation of the groundwater monitoring program at the Long
Prairie site are presented in Appendix C (Section C2.6) and D (Appendix D-2), and are summarized
in this subsection.

Application of  the  Mann-Kendall  and  linear-regression  temporal  trend  evaluation  methods
(Appendices B and C) indicated that the extent and concentrations of PCE in groundwater at the Long
Prairie source area probably are decreasing  (GSI, 2003b).  PCE concentrations in groundwater at 24
of 27 wells downgradient of the source area also are probably decreasing under current conditions.
The results of the moment analysis indicated that the mass of PCE in groundwater is relatively stable,
and that although the location of the center of mass of the plume has moved downgradient over time,
the extent of PCE in groundwater has decreased through time. Overall, the results of trend analyses
and moment analyses indicated that the extent of PCE in groundwater is stable or decreasing,
resulting in a recommendation that a  monitoring strategy appropriate for a "Moderate"  design
category be adopted (Appendices C and D).

Seventeen of the 44 wells in the existing monitoring network were included in the detailed spatial
analysis (Appendices C and  D);  the results indicated that none of the  17 wells evaluated was
redundant. Other wells  in the monitoring network were examined qualitatively; and the results of
qualitative considerations (GSI, 2003b) indicated that nine monitoring wells could be removed from
the  monitoring network without significant loss of information. Using similar qualitative analyses,
three  extraction wells in the source area were  identified as candidates for removal  from  service,
because concentrations of COCs in effluent from these wells historically have been below reporting
limits (GSI, 2003b).  However, six wells that currently are not routinely sampled were recommended
for inclusion in the monitoring program. These changes in the monitoring network were projected to
have a negligible effect on the degree of characterization of the extent of PCE in groundwater. The
accompanying well-sufficiency analysis indicated that there is only a moderate degree of uncertainty
in predicted PCE concentrations throughout the network, so that no new monitoring wells were
recommended for installation (GSI, 2003b).  The results of the sampling-frequency optimization
analysis completed using MAROS  (Appendices C and D) indicated that most wells in the monitoring
network could be sampled less frequently than in the current monitoring program. The results of the
data-sufficiency evaluation, completed using power-analysis methods (Appendices B and C) suggest
that the monitoring program is adequate to evaluate the extent of PCE in groundwater  relative to
compliance points through time (GSI, 2003b).

The optimized monitoring program generated using the MAROS tool includes 32 wells, with  10
monitoring wells and 5 extraction  wells sampled annually, and 13 monitoring wells, two extraction
wells, and two municipal wells sampled biennially (Appendices C and D). Adoption of the optimized
program would result in  collection and analysis of 22 samples per year, as compared with collection
and analysis of 51 samples per year in  the current monitoring program (Table 4.1).   Implementing

                                          25

-------
 these recommendations could lead to a 51-percent reduction in the number of samples collected and
 analyzed annually, as compared with the current program. Assuming a cost per sample in the range
 of $100  to  $280 for collection and chemical  analyses, adoption  of the  monitoring program as
 optimized using the MAROS tool is projected to result in savings ranging from approximately $2,900
 to $8,120 per year.  (The estimated  range of costs per sample is based on information provided by
 facility personnel in conjunction with efforts  to estimate potential cost savings resulting  from
 optimization of the monitoring  program, and  includes costs associated with sample collection and
 analysis,  data compilation and reporting, and handling of IDW  [e.g., purge water].) The optimized
 program remains adequate to delineate the extent of COCs in groundwater, and to monitor changes in
 the plume over time (GSI, 2003b).


                    Table 4.1:  Results of Optimization Demonstrations at
             Long Prairie Groundwater Contamination Superfund Site, Minnesota
Monitoring-Program Feature
Wells sampled quarterly
Wells sampled semi-annually
Wells sampled annually
Wells sampled biennially
Total wells included in LTM program
Total number of samples (per year)
Annual costb/ of LTM program
Monitoring Program"7
Actual
(October 2002)
8
-
19
-
27
51
$14,280
Refined using
MAROS
—
—
16
16
32
22
$6,160
Refined using
3-Tiered Approach
2
6
14
4
26
36
$10,080
   Details regarding site characteristics and the site-specific monitoring programs at the Long Prairie Groundwater
   Contamination Superfund Site are presented in Appendices C and D-2.
   Information regarding annual monitoring program costs was provided by facility personnel. The cost of monitoring is
   assumed to be $280 dollars per sample; costs associated with monitoring include cost of sample collection, sample
   analyses, data compilation and reporting, and management of investigation-derived waste (e.g., purge water).
4.3
RESULTS OF LTMO EVALUATION COMPLETED USING THREE-TIERED APPROACH
The detailed results of the three-tiered evaluation of the groundwater monitoring program at the Long
Prairie site are presented in Appendices C (Section C2.6) and D (Appendix D-2), and are summarized
in this subsection.

The results of the three-tiered evaluation indicated that 18 of the 44 existing wells could be removed
from  the groundwater monitoring network with little loss of information (Parsons, 2003c).  The
results further  suggested that  the current monitoring program  (18 monitoring  wells, 6 active
extraction wells, one  inactive extraction well, and 2 municipal water-supply wells included in the
2002  sampling  program) could  be further refined by removing  4 of the 27 wells now in the LTM
program, and adding three wells not currently included in the program.  If this refined monitoring
program, consisting of 26 wells  (2  wells to be sampled quarterly,  6  wells to  be  sampled  semi-
annually, 14 wells to be sampled  annually, and 4 wells to be  sampled biennially) were adopted, an
average of 36 samples per year would be collected and analyzed, as compared with the collection and
analysis of 51  samples per  year  in  the current (2001/2002)  monitoring program (Table 4.1) - a
reduction of about 29 percent. Assuming a cost per sample ranging from $100 to $280 for collection
and chemical analyses, adoption  of the monitoring program as optimized using  the three-tiered
                                           26

-------
approach is projected to result in savings ranging from about $1,500 per year to about $4,200 per year
(Table 4.1), as compared with the current program (Parsons, 2003c).
                                            27

-------
  5.0    SUMMARY OF DEMONSTRATIONS AT McCLELLAN AFB OU
                                  D, CALIFORNIA


 An overview of features pertinent to the groundwater monitoring program at OU D, McClellan AFB,
 California, is provided in this  section, together with a  summary  of the results of the LTMO
 demonstrations.   The features of the site,  and of  the monitoring-program evaluations  that  were
 completed using the MAROS tool and the three-tiered approach, are summarized in Appendix C, and
 are described in detail in Appendix D-3.

 5.1      FEATURES OF MCCLELLAN AFB OU D

 The former McClellan AFB is located approximately 7 miles northeast of downtown Sacramento,
 California, and covers approximately 3,000 acres.  OU D consists  of contaminated groundwater
 beneath and downgradient from contaminant  source areas in the northwestern  part of McClellan
 AFB, and occupies approximately 192 acres.  Through most of its operational history, McClellan
 AFB was engaged in a wide variety of military/industrial operations involving the use, storage, and
 disposal  of hazardous  materials, including  industrial solvents, caustic  cleaners, electroplating
 chemicals, metals, polychlorinated biphenyls, low-level radioactive wastes, and a variety of fuel oils
 and lubricants.

 The  COCs in groundwater targeted by the current LTM program at  OU D are exclusively CAHs,
 including PCE, TCE, cw-l,2-DCE, and 1,2-dichloroethane  (DCA), with 1,1-DCA, 1,1-DCE, 1,1,1-
 TCA, and VC also detected, but at lower concentrations and/or lower  frequencies. Dissolved CAHs
 originating at sources near former disposal areas at OU D have migrated with regional groundwater
 flow to the south and southwest, and historically extended off-base, to the west of OU D.  Currently,
 VOCs (primarily TCE) are present in groundwater primarily in the central and southwestern parts of
 OU D (Figure 5.1). The remediation systems  currently operating to  address CAH contaminants in
 groundwater at OU D include a groundwater ETD system, and the associated monitoring network.

 In accordance with the requirements of the basewide groundwater monitoring plan, wells in the OU D
 area  are sampled during the first  quarter of each year.  In the OU D area, groundwater sampling is
 conducted to monitor areas where dissolved VOC concentrations exceed their respective maximum
 contaminant levels (MCLs) in monitoring zones A and B.  Groundwater  monitoring data also are
 used to evaluate contaminant mass-removal rates. Because the extent of COCs in groundwater at OU
 D is relatively well defined, and COCs appear to be contained  by the groundwater extraction system,
 the wells  associated with  the OU D  plume are  sampled  relatively infrequently (annually  or
biennially). Currently, 22 of the  32 wells that  monitor the upper part (Zone A) of the groundwater
 system at OU D are sampled biennially, and 10 are sampled annually. Twelve of the 13 wells that
monitor a deeper part (Zone B) of the groundwater system are sampled biennially, and the remaining
well  is sampled annually.  The  six extraction wells (EWs)  are  sampled  annually.  Historically,
however, the sampling schedule for wells at OU D was irregular,  so that some monitoring wells at
OU D have been sampled as few as five times  through the historic monitoring  from the monitoring
and extraction wells are analyzed for VOCs by U.S. EPA Method SW8260B.
                                         28

-------
                                                           MW-1027j
                                                          MW-102J
                                                          MW-1026'
                                           1Q02 5ug/LA-Zone TCE IsqDleth
                                           (based on non-lAB wells)
 Legend
OUD Area Well
  &  Zone A
  ฉ  Zone A (previously IAB)
 , ฉ  Zone B
j *, 0  Zone B (previously IAB)
  4-  Extraction Well
 CuTFent Sampling Frequency
  O  Annual  "1
                                                                       @  Biennial
                                                                      —fe GiouTndwatei Flow Direction
                                                                                        '-350j [
Figure 5.1:  Features of McClellan AFB OU D (after Parsons, 2003d)
                               29

-------
 5.2      RESULTS OF LTMO EVALUATION COMPLETED USING MAROS TOOL

 The detailed results of the MAROS evaluation of the groundwater monitoring program at McClellan
 AFB OU D are presented in Appendices C  (Section C3.5) and D-3, and are  summarized in this
 subsection.

 Application of the   Mann-Kendall  and  linear-regression  temporal trend evaluation  methods
 (Appendices B  and C) indicated that the extent and concentrations of TCE in groundwater at the OU
 D source area probably are decreasing (GSI, 2003c).  However, the absence of identifiable  trends in
 TCE concentrations at many  locations downgradient of the plume may be a consequence of less-
 frequent sampling in these areas than occurs near the OU D source area (GSI, 2003c).  The results of
 the moment analysis  indicated that  the mass of TCE in groundwater is  relatively stable, with
 occasional fluctuations suggesting increases or decreases in TCE mass.  The location of the  center of
 mass of the plume also appears  to  be relatively stable, with periodic  temporal fluctuations in
 concentrations tending to cause the center of TCE mass to appear to move in the upgradient or
 downgradient directions. The  lateral extent of TCE in groundwater has been variable, suggesting that
 TCE concentrations in wells used to evaluate conditions over large, off-axis areas of the plume have
 varied considerably through time, or that the wells have not been sampled consistently enough for a
 clear trend in TCE concentrations to emerge.  Temporal fluctuations in the apparent mass of TCE in
 groundwater (calculated  using the zero* moment), the center of mass of TCE (calculated using the
 first moment), and the lateral extent of TCE (calculated using the second moment) likely are due to
 long-term variability  in  locations  sampled,  resulting  from an inconsistent monitoring  program
 through time (GSI, 2003c). The evaluation of overall plume stability indicated that the extent of TCE
 in groundwater at OU D is  stable or slightly decreasing, resulting in a recommendation that a
 monitoring strategy appropriate for a "Moderate" design category be adopted (Appendices C and D).

 The results of the detailed spatial analysis, supplemented with a qualitative evaluation (Appendices C
 and  D), identified  five monitoring  wells as candidates for removal  from  the monitoring  network.
 Removal of the recommended five  wells would result in an 11 percent reduction in the number of
 wells in the monitoring network, with negligible effect on the degree of characterization of the extent
 of TCE  in groundwater.  The possibility of removing additional monitoring wells on the periphery of
 OU D also was examined qualitatively, and it was concluded (GSI, 2003c) that the decision to  stop
 sampling the periphery  wells should be made in accordance with non-statistical considerations,
 including regulatory requirements, community concerns, and/or public health issues. Non-statistical
 considerations  may indicate that continued sampling of the periphery wells is warranted.   The
 accompanying  well-sufficiency analysis indicated that there is only a low to moderate degree of
 uncertainty in predicted TCE concentrations throughout the network, so that no new monitoring wells
 were recommended for installation (GSI, 2003c). In nearly all instances, the results of the sampling-
 frequency optimization analyses at  McClellan AFB  OU D were adversely affected by the lack of
consistent temporal monitoring  data  (Appendices C and D).  Accordingly, all recommendations
generated by MAROS were examined qualitatively, after the temporal  statistical evaluations had been
completed, to generate recommendations regarding sampling frequency (GSI, 2003c).  The results of
the  data-sufficiency  evaluation, completed   using  power-analysis  methods,   indicate   that  the
monitoring program is more than sufficient to evaluate the extent of TCE in groundwater relative to
the compliance boundary through time, assuming continued operation of the extraction system (GSI,
2003c).

The optimized monitoring program  generated  using the MAROS tool includes 29 A-zone wells, 11
B-zone wells, and 6 groundwater extraction wells, with 11 monitoring wells and 6 extraction wells

                                          30

-------
sampled annually, and 29 monitoring wells sampled biennially (Appendices C and D).  Adoption of
the optimized program would result in collection and analysis of 32 samples per year, as compared
with collection and analysis of 34 samples per year in the current monitoring program (Table 5.1).
Implementing these recommendations could lead to an approximately 6-percent reduction, in  the
number of samples  collected and  analyzed  annually, as compared with the  current program.
Adoption of the monitoring program as optimized using the MAROS tool is projected (GSI, 2003c)
to result in savings of approximately $300 per year (Table 5.1).  (Estimated annual cost savings were
provided by facility personnel; however, specific information regarding the estimated annual cost of
the LTM program at McClellan AFB OU D, and the total cost per sample is not available; and the
means used to derive the  estimated cost  savings are uncertain.)  The optimized program remains
adequate to delineate the extent of COCs in groundwater, and to monitor changes in the condition of
the plume over time (GSI, 2003c).

  Table 5.1: Results of Optimization Demonstrations at McClellan AFB OU D, California
Monitoring-Program Feature
Wells sampled annually
Wells sampled biennially
Total wells in LTM program
Total number of samples (per year)
Annual costb/ of LTM program
Monitoring Program3'
Actual
(October 2002)
17
34
51
34
--
Refined using
MAROS
17
29
46
32
c/
Refined using
3-Tiered Approach
13
8
21
17
C/
^  Details regarding site characteristics and the site-specific monitoring programs at McClellan AFB OU D are
   presented in Appendices C and D-3.
w No information regarding annual monitoring program costs was provided by facility personnel.
c/ Total costs associated with refined monitoring programs cannot be estimated; no information available.
5.3
SUMMARY OF LTMO EVALUATION COMPLETED USING THREE-TIERED APPROACH
The detailed results of  the  three-tiered evaluation  of the  groundwater monitoring program at
McClellan AFB OU D are presented in Appendices C (Section C3.6) and D (Appendix D-3), and are
summarized in this subsection.

The results of the three-tiered evaluation (Parsons, 2003d) indicated that 30 of the 51 existing wells
could be  removed  from the groundwater monitoring  program with comparatively little loss of
information (Parsons, 2003d). Most of the wells recommended for removal from the monitoring
program are wells peripheral to the OU D plume, which also were identified as possible candidates
for removal during the MAROS evaluation. If this  refined monitoring program (Appendices C and
D), consisting of 21 wells (13 wells to be sampled  annually,  and 8 wells to be sampled biennially)
were adopted, an average of 17 samples per year would be collected and analyzed, as compared with
the collection and analysis of 34 samples per year in the current monitoring program - a reduction of
50 percent in the number of samples collected and analyzed annually, as compared with the current
program.  Although information regarding the annual costs  associated  with  the LTM program at
McClellan AFB OU D including the estimated total cost per sample  is not available, based on
analytical costs alone, and assuming a cost per sample of $150 for chemical analyses (analyses for
VOCs only), adoption of the monitoring program  as optimized using the three-tiered approach is
projected  to result  in  savings of about $2,550 per year as compared with the current  program
                                           31

-------
(Parsons, 2003d).  Additional cost savings could be realized if groundwater samples collected from
select wells (e.g., upgradient wells, and wells along the lateral plume margins) were analyzed for a
short list of halogenated VOCs using U.S. EPA Method SW8021B instead of U S  EPA Method
SW8260B (Parsons, 2003d).
                                         32

-------
            6.0    CONCLUSIONS AND RECOMMENDATIONS
A software tool (MAROS) developed for AFCEE, and a three-tiered approach applied by Parsons,
were used to evaluate and optimize groundwater monitoring programs at the Fort Lewis Logistics
Center, Washington, the Long Prairie Groundwater Contamination Superfund Site in Minnesota, and
OU D,  McClellan AFB,  California.   Although many of the basic assumptions and  techniques
underlying both optimization approaches  are  similar,  and  both approaches utilize qualitative,
temporal, and spatial analyses, there are  several differences in the details of implementation in the
two approaches, which can cause one optimization approach  (e.g., the  three-tiered  approach) to
generate results that are not completely consistent with the results obtained using the other approach
(e.g., MAROS). As a consequence of structural differences in approaches to the evaluation and
optimization of monitoring programs, the results generated by any. optimization approach should be
expected to differ slightly from the results generated by other approaches; however, the results of any
optimization approach  should be defensible, if the  decision logic on which the approach has been
based is sound.

6.1       SUMMARY OF RESULTS OF MAROS EVALUATIONS AND THREE-TIERED APPROACH

The results of the MAROS optimization and three-tiered evaluation of the monitoring program at the
Fort Lewis Logistics Center are summarized in Table 6.1. "Final" recommendations  for the entire
program could be developed by considering together the results of the three-tiered evaluation and of
the MAROS evaluation for each well.  Example composite recommendations are provided in Column
5 of Table 6.1.

        Table 6.1: Summary of Optimization of Groundwater Monitoring Program at
                            Fort Lewis Logistics Center Area"'
Well ID
Current11'
Sampling
Frequency
Recommendations
Generated Using
MAROS Tool
Recommendations
Generated Using
Three-Tiered
Approach
Example Composite0
Recommendations
                     Monitoring Wells Completed in Upper Vashon Subunit
FL2 (new"")
FL3 (new)
FL4B (new)
FL6 (new)
LC-03
LC-05
LC-06
LC-14a
LC-16(new)
LC-19a
LC-19b
LC-19c
LC-20 (new)
LC-24 (new)
Annual
Quarterly
Quarterly
Quarterly
Quarterly
Annual
Semi-Annual
Annual
Quarterly
Quarterly
..tf
—
Quarterly
Quarterly
Not Considered"'
Quarterly
Not Considered
Not Considered
Annual
Quarterly
Quarterly
Annual
Quarterly
Annual
Remove
Remove
Quarterly
Not Considered
Annual
Removef/
Biennial
Biennial
Biennial
Remove
Annual
Annual
Remove
Annual
Remove
Remove
Biennial
Biennial
Annual
Quarterly
Biennial
Biennial
Annual
Annual
Semi-Annual
Annual
Quarterly
Annual
Remove
Remove
Quarterly
Biennial
                                           33

-------
Table 6.1: Summary of Optimization of Groundwater Monitoring Program at
                   Fort Lewis Logistics Center Area
Well ID
Current
Sampling
Frequency
Recommendations
Generated Using
MAROS Tool
Recommendations
Generated Using
Three-Tiered
Approach
Example Composite
Recommendations
Monitoring Wells Completed in Upper Vashon Subunit (continued)
LC-26
LC-34 (new)
LC-41a
LC-44a
LC-49
LC-51
LC-53
LC-57 (new)
LC-61b(new)
LC-64a
LC-66a
LC-66b
LC-73a
LC-108
LC-132
LC-136a
LC-136b
LC-137a
LC-1375
LC-149c
LC-149d
LC-165
LC-167 (new)
LC-180
NEW-1 (new)
NEW-2 (new)
NEW-3 (new)
NEW-4 (new)
NEW-5 (new)
NEW-6 (new)
PA-381
PA-383
T-04
T-06 (new)
T-08
T-llb(new)
T-12b
T-13b
Annual
Quarterly
Annual
-
Annual
-
Annual
Quarterly
Quarterly
Quarterly
—
Annual
-
~
~
Quarterly
Annual
-
Quarterly
Annual
-
—
Quarterly
Annual
Not Considered
Quarterly
Remove
Semi-Annual
Remove
Quarterly
Not Considered
Not Considered
Quarterly
Remove
Annual
Biennial
Annual
Quarterly
Quarterly
Remove
Remove
Quarterly
Biennial
Remove
Biennial
Quarterly
Proposed for installation using 3-tiered
approach'1'
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
Annual
Annual
Annual
Quarterly
Semi-Annual
Quarterly
Quarterly
Semi-Annual
Not Considered
Not Considered
Quarterly
Not Considered
Quarterly
Not Considered
Annual
Biennial
Annual
Not Considered
Annual
Not Considered
Annual
Annual
Remove
Biennial
Annual
Remove
Annual
Remove
Annual
Biennial
Semi-Annual
Quarterly
Remove
Annual
Remove
Remove
Annual
Quarterly
Annual
Remove
Remove
Biennial
Biennial
Remove
Semi-Annual
Annual
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
Biennial
Biennial
Annual
Quarterly
Semi-Annual
Quarterly
Biennial
Semi-Annual
Annual
Biennial
Annual
Remove
Annual
Remove
Annual
Biennial
Semi-Annual
Quarterly
Remove
Annual
Remove
Remove
Annual
Quarterly
Annual
Remove
Quarterly
Biennial
Biennial
Biennial
Quarterly
Annual
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
Annual
Biennial
Annual
Quarterly
Semi-Annual
Quarterly
Biennial
Semi-Annual
                               34

-------
        Table 6.1: Summary of Optimization of Groundwater Monitoring Program at
                                 Fort Lewis Logistics Center Area



Well ID

Current
Sampling
Frequency

Recommendations
Generated Using
MAROS Tool
Recommendations
Generated Using
Three-Tiered
Approach
Monitoring Wells Completed in Lower Vashon Subunit
FL4a (new)
LC-41b (new)
LC-64b
LC-lllb
LC-116b
LC-122b
LC-128
LC-137c
MAMC 1
MAMC 6 (new)
T- 10 (new)
Quarterly
Quarterly
Annual
Annual
Annual
Annual
Annual
Annual
Quarterly
Quarterly
Quarterly
Not Considered
Not Considered
Annual
Biennial
Semi-Annual
Biennial
Annual
Annual
Not Considered
Not Considered
Not Considered
Biennial
Annual
Annual
Biennial
Annual
Remove
Annual
Annual
Quarterly
Quarterly
Semi-Annual


Example Composite
Recommendations

Biennial
Annual
Annual
Biennial
Annual
Biennial
Annual
Annual
Quarterly
Quarterly
Semi-Annual
Groundwater Extraction Wells
LX-1
LX-2
LX-3
LX-4
LX-5
LX-6
LX-7
LX-8
LX-9
LX-10
LX-11
LX-1 2
LX-1 3
LX-1 4
LX-1 5
LX-1 6
LX-1 7
LX-1 8
LX-1 9
LX-21
RW-1
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Quarterly
Quarterly
Quarterly
Quarterly
Annual
Quarterly
Every 3 years
Every 3 years
Every 3 years
Every 3 years
Every 3 years
Every 3 years
Every 3 years
Every 3 years
Every 3 years
Every 3 years
Every 3 years
Every 3 years
Every 3 years
Every 3 years
Every 3 years
Semi-Annual
Quarterly
Quarterly
Quarterly
Quarterly
Semi-Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
el

0
Information from GS1 (2003a) and Parsons (2003b).
"Current" monitoring program was initiated in December 2001 (Section 3.1).
"Composite"  recommendations generated  considering the  current monitoring  program,  and recommendations
generated by MAROS tool and three-tiered approach.
"new" =  the well was not included in the monitoring program prior to December 2001.
"Not Considered" = the well was not included in the MAROS evaluation.
"Remove" indicates that the well is recommended for removal from the monitoring program.
A dash (-) indicates that the well is not included in the current or refined monitoring program.
"Proposed for installation" indicates that a location for an additional monitoring well was identified on the basis of the
evaluation.
                                                 35

-------
A well was not selected for removal from the program in the example "composite" recommendations,
unless that well was recommended for removal in both the MAROS and three-tiered evaluations, or
unless that well was recommended for removal in one of the evaluations, and was not included in the
monitoring program that was initiated in December 2001. The frequency of sampling provided in the
"composite" recommendations was the frequency of sampling specified in  the recommendations
generated in the MAROS  and three-tiered evaluations, if those recommendations were in agreement.
If the frequencies recommended in the MAROS and three-tiered evaluations did not agree, but one of
the recommended frequencies was the same as the current sampling frequency, the current sampling
frequency was retained in the example "composite" recommendations.  If the frequency of sampling
at a particular well, specified in the recommendations generated in the three-tiered evaluation, did not
agree with the frequency of sampling at that well in the current monitoring program, and the MAROS
evaluation did not consider that well, the frequency of sampling recommended in the three-tiered
evaluation was specified in  the  "composite" recommendations.   If none  of the  current,  and
recommended, sampling frequencies  were in agreement, the intermediate sampling frequency  was
specified in the "composite" recommendations. This example represents a "conservative" approach
to LTMO for the program  at the Fort Lewis  Logistics Center  area,  because  it considers
recommendations generated  using two different approaches, in addition to giving weight to currently-
accepted monitoring practice  at the site, by  also considering  the current  monitoring  program.
Adoption  of the example  "composite" monitoring program would result in removal of eight wells
from  the  current  monitoring  program at the Fort Lewis  Logistics  Center area, together  with
adjustment of the frequency of sampling to less-frequent events at most locations. Of course, more
aggressive approaches to a "composite" optimization scheme also could be applied.

The results of the MAROS optimization and the three-tiered evaluation, including recommendations
for removal of wells and adjustments to sampling frequency, were fully consistent for approximately
40 percent of the wells in the Fort Lewis Logistics Center monitoring program.  (Wells that MAROS
did not consider are not included in this comparison.)

The results of the three-tiered evaluation and MAROS optimization of the monitoring program at the
Long Prairie Groundwater Contamination  Superfund Site are summarized in Table 6.2.  Example
composite recommendations also are provided in Column 5 of Table 6.2.
        Table 6.2:  Summary of Optimization of Groundwater Monitoring Program at
                 Long Prairie Groundwater Contamination Superfund Site8'
Well ID
Current1"7
Sampling
Frequency
Recommendations
Generated Using
MAROS Tool
Recommendations
Generated Using
Three-Tiered
Approach
Example Composite07
Recommendations
                                     Monitoring Wells
BAL2B
BAL2C
MW1A
MW1B
MW2A
MW2B
MW2C
MW3A
MW3B
d/
—
—
-
Annual
Annual
Annual
—
-
Biennial
Biennial
Remove
Biennial
. Remove
Annual
Annual
Remove
Biennial
Remove67
Remove
Remove
Remove
Remove
Annual
Remove
Remove
Remove
Remove
Remove
Remove
Remove
Remove
Annual
Annual
Remove
Remove
                                           36

-------
Table 6.2: Summary of Optimization of Groundwater Monitoring Program at
         Long Prairie Groundwater Contamination Superfund Site
Well ID
Current
Sampling
Frequency
Recommendations
Generated Using
MAROS Tool
Recommendations
Generated Using
Three-Tiered
Approach
Example Composite
Recommendations
                       Monitoring Wells (continued)
MW4A
MW4B
MW4C
MW5A
MW5B
MW6A
MW6B
MW6C
MW10A
MW11A
MW11B
MW11C
MW13C
MW14B
MW14C
MW15A
MW15B
MW16A
MW16B
MW17B
MW18A
MW18B
MW19B
	
Annual
Annual
—
—
Annual
Annual
Annual
Annual
—
Annual
Annual
—
Annual
Annual
Annual
Annual
—
Annual
Annual
—
—
Annual
Remove
Annual
Annual
Remove
Biennial
Remove
Annual
Annual
Annual
Remove
Biennial
Biennial
Biennial
Annual
Biennial
Biennial
Biennial
Remove
Annual
Annual
Remove
Biennial
Biennial
Remove
Annual
Annual
Remove
Annual
Remove
Annual
Annual
Annual
Remove
Biennial
Biennial
Biennial
Annual
Biennial
Biennial
Biennial
Remove
Annual
Annual
Remove
Biennial
Biennial
Remove
Annual
Annual
Remove
Biennial
Remove
Annual
Annual
Annual
Remove
Biennial
Biennial
Biennial
Annual
Biennial
Biennial
Biennial
Remove
Annual
Annual
Remove
Biennial
Biennial
                      Groundwater Extraction Wells
RW1A
RW1B
RW1C
RW3
RW4
RW5
RW6
RW7
RW8
RW9
RW7
RW8
RW9
—
—
—
Quarterly
Annual
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
Quarterly
Remove
Remove
Remove
Annual
Biennial
Annual
Annual
Annual
Annual
Biennial
Annual
Annual
Biennial
Remove
Remove
Remove
Annual
Biennial
Annual
Annual
Annual
Annual
Biennial
Annual
Annual
Biennial
Remove
Remove
Remove
Annual
Biennial
Annual
Annual
Annual
Annual
Biennial
Annual
Annual
Biennial
                                 37

-------
 a/

 b/
         Table 6.2: Summary of Optimization of Groundwater Monitoring Program at
                   Long Prairie Groundwater Contamination Superfund Site
Well ID
Current^
Sampling
Frequency
Recommendations
Generated Using
MAROS Tool
Recommendations
Generated Using
Three-Tiered
Approach
Example Composite
Recommendations
Municipal Water-Supply Wells
CW3
CW6
Quarterly
Quarterly
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Information from GSI (2003b) and Parsons (2003c).
"Current" monitoring program was in effect in 2002.
"Composite"  recommendations generated  considering the current monitoring  program,  and  recommendations
generated by MAROS tool and three-tiered approach.
A dash (—) indicates that the well is not included in the current monitoring program.
"Remove" indicates that the well is recommended for removal from the monitoring program.
The results of the MAROS optimization and the three-tiered evaluation, including recommendations
for removal of wells and adjustments to sampling  frequency, were fully consistent for nearly 90
percent of the wells in the monitoring program at the Long Prairie site.  Adoption of the  example
"composite" monitoring program would result in removal of  16 wells from the current monitoring
network at the Long Prairie site, together  with adjustment of the frequency of sampling to less-
frequent events at several locations.

The results of the three-tiered evaluation and MAROS optimization of the monitoring program at
McClellan AFB OU D are summarized in Table 6.3.  Example composite recommendations also are
provided in Column 5 of Table 6.3.
         Table 6.3:  Summary of Optimization of Groundwater Monitoring Program at
                                   McClellan AFB OU D"'
Well ID
Currentb/
Sampling
Frequency
Recommendations
Generated Using
MAROS Tool
Recommendations
Generated Using
Three-Tiered
Approach
Example Composite'
Recommendations
                                   Zone A Monitoring Wells
MW-10
MW-11
MW-12
MW-14
MW-15
MW-38D
MW-52
MW-53
MW-55
MW-70
Annual
Annual
Annual
Biennial
Annual
Annual
Biennial
Biennial
Biennial
Biennial
Annual
Annual
Annual
Remove*"
Annual
Annual
Biennial
Biennial
Biennial
Biennial
Annual
Annual
Annual
Biennial
Annual
Annual
Remove
Remove
Biennial
Remove
Annual
Annual
Annual
Biennial
Annual
Annual
Biennial
Biennial
Biennial
Biennial
                                             38

-------
Table 6.3: Summary of Optimization of Groundwater Monitoring Program at
                        McClellan AFB OU D
Well ID
Current
Sampling
Frequency
Recommendations
Generated Using
MAROS Tool
Recommendations
Generated Using
Three-Tiered
Approach
Example Composite
Recommendations
Zone A Monitoring Wells (continued)
MW-72
MW-74
MW-76
MW-88
MW-89
MW-90
MW-91
MW-92
MW-237
MW-240
MW-241
MW-242
MW-350
MW-351
MW-412
MW-458
MW-1004
MW-1026
MW-1041
MW-1042
MW-1064
MW-1073
Annual
Biennial
Annual
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Annual
Annual
Biennial
Annual
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Annual
Annual
Annual
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Remove
Annual
Biennial
Annual
Biennial
Biennial
Biennial
Biennial
Remove
Biennial
Biennial
Biennial
Remove
Remove
Annual
Remove
Biennial
Biennial
Remove
Remove
Remove
Remove
Remove
Remove
Remove
Remove
Remove
Remove
Remove
Remove
Remove
Remove
Remove
Remove
Annual
Annual
Annual
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Remove
Annual
Biennial
Annual
Biennial
Biennial
Biennial
Biennial
Remove
Biennial
Biennial
Biennial
                       Zone B Monitoring Wells
MW-19D
MW-51
MW-54
MW-57
MW-58
MW-59
MW-104
MW-1001
MW-1003
MW-1010
MW-1027
MW-1028
MW-1043
Biennial
Biennial
Annual
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Biennial
Annual
Biennial
Biennial
Biennial
Biennial
Biennial
Remove
Biennial
Biennial
Remove
Biennial
Biennial
Biennial
Annual
Remove
Biennial
Biennial
Remove
Remove
Remove
Remove
Biennial
Remove
Biennial
Biennial
Biennial
Annual
Biennial
Biennial
Biennial
Biennial
Biennial
Remove
Biennial
Biennial
Remove
Biennial
                                39

-------
         Table 6.3:  Summary of Optimization of Groundwater Monitoring Program at
                                   McCIellan AFB OU D
Well ID
Current
Sampling
Frequency
Recommendations
Generated Using
MAROS Tool
Recommendations
Generated Using
Three-Tiered
Approach
Example Composite
Recommendations
                                Groundwater Extraction Wells






a/
b/
c/
d/
EW-73
EW-83
EW-84
EW-85
EW-86
EW-87
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Annual
Information from GSI (2003c) and Parsons (2003d).
"Current" monitoring program was in effect in 2002.
"Composite" recommendations generated considering the current monitoring program, and recommendations
generated by MAROS tool and three-tiered approach.
"Remove" indicates that the well is recommended for removal from the monitoring program.
The results of the MAROS optimization and the three-tiered evaluation, including recommendations
for removal of wells and adjustments to sampling frequency, were fully consistent for approximately
50  percent of the wells  in the monitoring program at McCIellan AFB OU D.  Application of the
three-tiered approach to  the monitoring program generated considerably more recommendations for
well-removal from the program than did the  MAROS evaluation, primarily on the basis of the
qualitative evaluation, which recommended the removal of wells at the periphery of OU D, that
historically  have had  no  detections  (or  few detections at  low concentrations) of  COCs in
groundwater. Even though the example "composite" program represents a conservative approach to
program optimization, adoption  of the example "composite" monitoring program would result in
removal of four wells from the current monitoring program at OU D, together with adjustment of the
frequency of sampling to less-frequent events at several locations.

Application of the two approaches to the optimization of long-term monitoring programs at each of
the three case-study example sites generated recommendations for reductions in sampling frequency
and changes in the numbers and locations of monitoring points that are sampled.  Implementation of
the optimization recommendations could lead to reductions ranging from only a  few percent (using
MAROS at McCIellan AFB OU D) to more than 50 percent (using MAROS at the Long Prairie site
and the three-tiered approach at McCIellan AFB OU D) in the numbers of samples collected and
analyzed annually at particular sites.  The median recommended reduction in the annual number of
samples collected, generated during the optimization demonstration, was 39 percent.  Depending
upon the scale of the particular long-term monitoring program, and the nature of the optimization
recommendations, adoption  of an optimized monitoring program could lead to annual cost savings
ranging from a few hundred dollars  (using MAROS at McCIellan AFB OU D) to  approximately
$36,500 (using the three-tiered approach at the Fort Lewis Logistics Center Area).  The results of the
evaluations also demonstrate that each of the optimized monitoring programs remains  adequate to
address the primary objectives of monitoring.
                                           40

-------
6.2      OTHER ISSUES

The  procedures used in the LTMO  evaluations were  discussed  with  various  stakeholders (the
environmental coordinators, responsible parties, and regulatory-agency personnel) through the entire
course of the project.   After the evaluations had been completed, the  results were presented to
stakeholder groups at each facility.  Presenting the results to regulators at the three facilities raised
questions that had to do more  with the data quality objectives (DQOs) than with the approaches
themselves.  It became clear that every monitoring location that was recommended for removal, or
for a change in sampling frequency, had a non-quantifiable, subjective value that depended on the
person making the  optimization  decision.  Much discussion  revolved around the necessity of
monitoring to a degree sufficient to incontrovertibly document plume capture. Other questions were
raised regarding whether changes to monitoring  programs would require modifications to existing
Records of Decision (RODs).

Based on those discussions, it is clear that before any optimization recommendation is accepted, there
must be a careful and thorough presentation of the long-term groundwater monitoring DQOs from the
viewpoint of all the  stakeholders, followed by stakeholder agreement on DQOs, possibly  for every
groundwater monitoring location.  After the  objectives have been defined, and consensus has been
reached,  the results of the optimization analyses can be examined, and a decision made to  accept or
reject recommendations.  Note  that there may be intangible  costs associated with the development
and  presentation of recommendations  to reduce the spatial density or temporal frequency of
monitoring, including resistance of stakeholders and changes in public perception.

Depending upon the degree of  difficulty  in arriving  at stakeholder  concurrence with LTMO
recommendations, the tangible and intangible costs associated with conducting and implementing an
LTMO evaluation  may outweigh the dollar  cost  savings that might be realized from an optimized
program. This possibility must be addressed on a site-specific basis.

6.3      CONCLUSIONS

The  most significant advantage conferred by the  optimization approaches is the fact  that both
approaches apply consistent, well-documented procedures, which incorporate formal decision logic,
to the process of evaluating and  optimizing monitoring programs.  However, there  are certain
limitations to each approach to monitoring program optimization. The primary limitation of MAROS
is associated with the way in  which  the tool deals with COC concentrations that are below the
reporting limit - MAROS  assigns the value of the reporting limit (or some fraction thereof) to
samples  having a constituent concentration below the reporting limit (Appendix  B).  This can lead to
identification of spurious temporal trends in concentrations, or to incorrectly concluding that reported
concentrations are unstable through time.  Identification of spurious trends, in turn, will  affect the
recommendations regarding the optimal frequency of sampling. The primary limitation of the three-
tiered approach is that the spatial-statistical stage of the  evaluation generally is completed using
sampling results for only one constituent (Appendix B). The fact that the  spatial evaluation currently
is conducted in two spatial dimensions (rather than three) represents a limitation of both approaches.

For  either approach, the process of becoming familiar with the pertinent characteristics  of  a  site,
identifying those data appropriate for the intended application, and transferring those  data  to the
appropriate format (even if the data are available in an electronic database), can be  time-consuming
and  labor-intensive,  and represents a  significant  up-front investment of time and resources.  Both
approaches could benefit from  further development efforts to address these limitations;  continued
development of both approaches is contemplated or in progress.

                                             41

-------
 Experience obtained during the demonstrations indicates that although the MAROS tool is capable of
 being applied by an individual with little formal statistical  training,  interpretation of the  results
 generated by either approach  requires a relatively  sophisticated understanding of hydrogeology,
 statistics, and the processes governing the movement and fate of contaminants in the environment.
 The two approaches differ primarily in the procedures used to select a sampling frequency. MAROS
 utilizes a relatively rigorous, statistical approach based on identification of temporal trends in COC
 concentrations,  while the three-tiered approach depends  primarily upon qualitative considerations,
 applied using detailed knowledge of the local hydrogeologic system, with support from the results of
 the temporal and spatial-statistical evaluations.  However,  if the assumptions underlying the MAROS
 statistical approach are  violated (e.g., the number of separate  monitoring events is  not sufficient to
 identify  a  trend), application  of MAROS  to develop  recommendations  regarding monitoring
 frequency also will depend on qualitative considerations (e.g., GSI, 2003c). Both approaches use a
 ranking  approach to identify potentially-unnecessary  monitoring  locations,  although  the spatial-
 statistical procedures used to implement the ranking approach are somewhat different.

 In  general, the recommendations generated by MAROS regarding spatial redundancy and sampling
 frequency were  more conservative  than the  recommendations  generated during  the  three-tiered
 evaluation (e.g., MAROS may recommend semi-annual sampling at a particular monitoring location,
 while the three-tiered evaluation may recommend annual sampling at the same location).  In addition,
 the three-tiered  approach tends  to generate recommendations for removing a larger proportion  of
 wells from a monitoring program than does MAROS,  because the three-tiered approach considers the
 results of qualitative, temporal, and spatial analyses together to determine whether a particular well
 should be retained or removed from the monitoring program, while MAROS will recommend  a well
 for removal from the program only if it is classified as redundant for all COCs based on the results  of
 the spatial evaluation alone. It is possible that the more rigorous qualitative evaluation in the three-
 tiered approach justifies less-conservative recommendations than are generated using the MAROS
 approach.   For  example,  the  three-tiered evaluation generated a recommendation  for biennial
 sampling at well LC-149c in the optimized Fort Lewis Logistics Center monitoring program, because
 the qualitative review in the three-tiered evaluation identified well LC-149c as having no historical
 detections of COCs throughout a monitoring history comprising 24 sampling events.  By contrast, the
 temporal-statistical evaluation algorithm in  MAROS originally generated a  recommendation for
 annual sampling at that well.   (The recommendation for annual  sampling  later  was revised by
 applying qualitative considerations during subsequent stages of the MAROS  evaluation.)

 The general characteristics of  each of the  three  case-study  example  sites  addressed in  this
 demonstration project are similar, comprising chlorinated solvent contaminants in groundwater,
 occurring at relatively  shallow  depth in unconsolidated  sediments.   However, the  assumptions
 underlying the two approaches, and the procedures that are followed in  conducting  the evaluations,
 are applicable to a much broader  range of conditions (e.g., dissolved metals in groundwater, or
 contaminants in a fractured bedrock system). In summary, either the MAROS tool or the three-tiered
 approach can be used to generate sound and defensible recommendations for optimizing a long-term
 monitoring program, under a wide range of site conditions.

 Prior to initiating an LTMO evaluation, it is of critical importance that the  monitoring objectives of
the program to be optimized and the  DQOs for individual monitoring points be clearly articulated,
with all stakeholders agreeing to the stated objectives, decision  rules, and procedures, so that the
program can be optimized in terms of recognized objectives, using decision rules and procedures that
are acceptable to all stakeholders. The decisions regarding whether to conduct an LTMO evaluation,
which approach to use, and the degree of regulatory-agency involvement in the LTMO evaluation


                                            42

-------
and subsequent implementation of optimization recommendations, must be made on a site-specific
basis. Factors to be considered in deciding whether to proceed with an LTMO evaluation include:

   •   The projected level of effort necessary to conduct the evaluation;

   •   The resources available for the evaluation (e.g., quality and quantity of data, staff having the
       appropriate technical capabilities);

   •   The anticipated degree of difficulty in implementing optimization recommendations; and

   •   The potential benefits (e.g., cost savings) that could result from  an  optimized monitoring
       program.

Experience suggests that optimization of a monitoring program should be considered for most sites
where  the LTM programs are based on monitoring points  and/or sampling  frequencies that were
established during site characterization, or for sites where more than  about 50  samples are collected
and analyzed on an annual basis.  Because it is likely that monitoring programs can benefit from
periodic evaluation as environmental programs evolve, monitoring program optimization also should
be undertaken periodically, rather than being regarded as a one-time event.  Overall site conditions
should be relatively stable, with no large changes in remediation approaches occurring or anticipated.
For sites at which response decisions are being validated or refined (e.g., during periodic remedy-
performance reviews), optimization of the LTM program should be  postponed until adjustments to
the response have been implemented and evaluated. Successful application of either LTMO approach
to the site-specific evaluation of a monitoring program is directly dependent upon  the amount and
quality of the  available data - results from a  minimum of four to six separate sampling events are
necessary to support a temporal analysis, and results collected at a minimum of about six  (for a
MAROS evaluation) to 15 (for a three-tiered evaluation) separate monitoring points  are necessary to
support a spatial analysis.  It also  is necessary to develop an adequate CSM, describing site-specific
conditions (e.g., direction and rate of groundwater movement, locations of contaminant sources and
potential receptor exposure points) prior to applying either approach; the  extent of contaminants in
the subsurface at the site also must be adequately delineated before the monitoring  program can  be
optimized.

Typically, a program manager should anticipate incurring costs on the order of $6,000 to $10,000 to
complete an LTMO evaluation using one of the two approaches presented in this demonstration, at
the level of detail of the case-study examples used in the demonstration (Sections 3, 4, and 5; and
Appendices C and  D).  Consequently,  an  LTMO evaluation may be cost-prohibitive  for smaller
monitoring programs.   Assuming  a  payback period of three years, potential  cost  savings  of
approximately $2,000  to $3,300 per year must be realized if optimization of a monitoring program is
to be cost-effective.  Because the costs  associated with collection and analysis of a groundwater
sample (including prorated mobilization costs, and costs for field sampling,  management of water
produced during sampling, laboratory analyses, QA/QC, and reporting) using conventional sampling
technologies (bailer or purge pump) can range from about $200 per  sample to more than  $500 per
sample (U.S. Air Force, 2004), an LTMO evaluation that can be used to reduce the total number of
samples collected at a site by about 5 to 10 samples per annum should  be cost-effective.
                                             43

-------

-------
                               7.0   REFERENCES
Air Force Center for Environmental Excellence  (AFCEE).  2000.  Monitoring and Remediation
      Optimization System (MAROS) User's Guide, Beta Version 1.0. October.
AFCEE.  2002. Monitoring and Remediation Optimization System (MAROS) User's Guide, Version
      2.0.  October.
American Society of Civil Engineering (ASCE)  Task  Committee on Geostatistical Techniques in
      Hydrology.  1990a.  Review of geostatistics in geohydrology - I. Basic concepts. Journal of
      Hydraulic Engineering 116(5):612-632.
ASCE Task Committee on Geostatistical Techniques in Hydrology. 1990b. Review of geostatistics
      in geohydrology - II. Applications. Journal of Hydraulic Engineering 116(5):633-658.

Bartram, J., and R. Balance. 1996.  Water Quality Monitoring.  E&FN Spon. London.

Cameron, K., and P.  Hunter.  2002.  Optimization of LTM Networks  Using GTS:   Statistical
      Approaches  to  Spatial and Temporal Redundancy.   Online document available on the
      Worldwide Web at http://www.afcee.brooks.af.mil/er/rpo/ GTSOptPaper.pdf

Cameron, K.  and P. Hunter.  2002. Using spatial models and kriging techniques to optimize long-
      term ground-water monitoring networks: A case study. Envirometrics 13(5-6):629-656.

Cameron, K. and P. Hunter. 2004.  Optimizing LTM Networks with GTS:  Three New Case Studies,
      in  Accelerating  Site  Closeout, Improving  Performance, and  Reducing  Costs  Through
      Optimization --  Proceedings  of  the  Federal  Remediation  Technologies  Roundtable
      Optimization Conference. June 15-17. Dallas, Texas.
Clark, I.  1987. Practical Geostatistics.  Elsevier Applied Science, Inc. London.

Cieniawski, S.E., J.W. Eheart, and S. Ranjithan.   1995.   Using genetic algorithms to solve a
      multiobjective groundwater monitoring problem.  Water Resources Research 31 (2):399-409.

Davis, J.C.  1986.  Statistics and Data Analysis in Geology.  John Wiley & Sons, Inc.  New York,
      New York. 2nd ed.
Deutsch, C.V., and A.G. Journel. 1998. GSLIB - Geostatistical Software Library and User's Guide.
      Oxford University Press. New York, New York.  2nd ed.
Dresel,  E.P., and C. Murray.  1998.  Groundwater monitoring network design  using stochastic
      simulation. Geological Society of America Abstracts with Programs 30(7):181.
Englund, E., and A. Sparks. 1992.  GEO-EAS (GEOstatistical Environmental Assessment software),
      Program Version 1.2.1 and User's Guide.  US Environmental Protection Agency.  EPA/600/4-
      88/033a.
Environmental Systems Research  Institute, Inc. (ESRI).   2001.  ArcGIS Version  8 Software.
      Redlands, California.
Everett,  L.G.  1980.  Groundwater Monitoring - Guidelines and Methodology for Developing and
      Implementing a  Groundwater  Quality Monitoring Program.   General  Electric Company.
      Schenectady, New York.
                                            44

-------
Franke, O.L. (ed.)  1997.  Conceptual Frameworks for Ground-Water-Quality Monitoring.  Ground-
       Water  Focus Group of the Intergovernmental  Task Force on  Monitoring Water Quality.
       Denver, Colorado.  August.

Gibbons, R.D.  1994.  Statistical Methods for Groundwater Monitoring.  John Wiley & Sons, Inc.
       New York, New York.

Gibbons, R.D. and D.E. Coleman.  2002.  Statistical Methods for Detection and Quantification of
       Environmental Contamination. John Wiley & Sons, Inc. New York, New York.

Griffin, D.A.  1996. The Need for Spatial Statistics, in Arlinghaus, S.L., ed., Practical Handbook of
       Spatial Statistics. CRC Press LLC. Boca Raton, Florida.

Groundwater  Services, Inc. (GSI).  2003a.   MAROS 2.0 Application - Upper Aquifer  Monitoring
       Network Optimization, Fort Lewis Logistics Center, Pierce County, Washington. Final. April.

GSI.  2003b.  MAROS 2.0 Application - Upper Outwash Aquifer Monitoring Network Optimization,
       Long Prairie Site, Long Prairie, Minnesota. Final. May.

GSI.  2003c.  MAROS 2.0 Application - Upper Aquifer Monitoring Network Optimization, Operable
       Unit D, McClellan Air Force Base, California.  Final. April.

Hirsch, R.M., R.B. Alexander,  and R.A. Smith.  1991.  Selection of methods for the detection and
       estimation of trends in water quality. Water Resources Research 27(5):803-813.

Hudak, P.P. 2000. Effects of two-dimensional mass transport modeling on groundwater monitoring
       configurations.  Bulletin of Environmental Contamination and Toxicology 64( 1): 1 -7.

Hudak, P.P., H.A. Loaiciga, and F.A. Schoolmaster.  1993.  Application of geographic information
       systems to groundwater monitoring network design.  Water Resources Bulletin 29(3):383-390.

Kelly, B.P.  1996.  Design of a Monitoring  Well Network for the City of Independence, Missouri,
       Well Field Using Simulated Ground-Water Flow Paths and Travel Times.  US Geological
       Survey Water-Resources Investigation  Report 96-4264.

Lapham, W.W., F. D. Wilde, and M.T. Koterba.  1996. Guidelines and Standard Procedures for
      Studies  of Ground-Water Quality:    Selection and Installation of Wells,  and  Supporting
      Documentation. US Geological Survey Water-Resources Investigations Report  96-4233.

Ling, M., .S. Rifai, C.J. Newell, J.J. Aziz, and J.R. Gonzales.  2003.  Groundwater monitoring plans
       at small-scale  sites  -  an innovative  spatial  and  temporal  methodology.   Journal  of
      Environmental Monitoring 5( 1): 126-134.
Loaiciga, H.A., R.J.  Charbeneau, L.G.  Everett, G.E. Fogg,  B.F. Hobbs, and S.  Rouhani.   1992.
      Review of  ground-water quality  monitoring  network design.   Journal  of Hydraulic
      Engineering 118( 1): 11 -37.

Makeig, K.S.   1991.  Regulatory Mandates for Controls on Ground-Water Monitoring, in Nielsen,
      D.M., ed.  Practical Handbook of Ground-Water Monitoring. Lewis Publishers, Inc.  Chelsea,
      Michigan.
National Research Council  (NRC).   1993.  Alternatives for Groundwater Cleanup.   National
      Academy of Sciences, National Academy Press. Washington, D.C.
NRC.  1999.  Environmental Cleanup at Navy Facilities - Risk-Based Methods.  National Academy
      of Sciences, National Academy Press.  Washington, D.C.
                                            45

-------
Parsons Corporation (Parsons). 2000. Remedial Process Optimization Report for Operable Unit 1,
      Hill Air Force Base, Utah.  Final.  December.
Parsons.  2003a.  Comparative Evaluation of the Monitoring and Remediation Optimization System
      (MAROS) Tool. Draft. January.
Parsons.  2003b.  Three-Tiered Groundwater Monitoring Network Optimization Evaluation for the
      Fort Lewis Logistics Center, Washington.  Draft Final. May.

Parsons.  2003c. Three-Tiered Groundwater Monitoring Network Optimization Evaluation for Long
      Prairie Groundwater Contamination Superfund Site, Minnesota.  Draft Final. May.

Parsons.   2003d.  Three-Tiered Groundwater  Monitoring Network  Optimization Evaluation for
      Operable Unit D, McClellan Air Force Base, California.  Draft Final. May.

Reed, P.M. and B.S. Minsker.  2004.   Striking the balance:  Long term groundwater monitoring
      design for multiple, conflicting  objectives.   Journal  of Water  Resources  and Planning
      Management 130(2): 140-149.
Reed, P.M., B.S. Minsker, and D.E.  Goldberg.  2001.   A multiobjective approach to cost effective
      long-term groundwater monitoring using an elitist nondominated sorted genetic algorithm with
      historical data. Journal of Hydroinformatics 3:71-89.

Reed, P.M., B.S. Minsker, and D.E.  Goldberg.  2003.  Simplifying multiobjective optimization II:
      An automated design methodology for the nondominated sorted genetic algorithm.  Water
      Resources Research 39(7): 1196.
Reed, P.M., B.S.  Minsker,  and A.J.  Valocchi.   2000.   Cost-effective long-term  groundwater
      monitoring design using a genetic algorithm and global mass interpolation.  Water Resources
      Research 36(12):3731-3741.
Ridley, M.N., V.M. Johnson, and R.C. Tuckfield.  1995. Cost-Effective Sampling of Ground Water
      Monitoring  Wells.   Lawrence Livermore National Laboratory (LLNL).  Report UCRL-JC-
      118909.
Rock, N.M.S. 1988. Numerical Geology. Springer-Verlag. New York, New York.

Sheskin, D.J.  2000. Handbook of Parametric and Nonparametric Statistical Procedures. Chapman
      & Hall/CRC, Inc.  Boca Raton, Florida. 2nd ed.
Schock, S.C., E. Mehnert, and D.P. McKenna. 1989. Design of a Sampling System for Agricultural
      Chemicals  in  Rural, Private  Water-Supply Wells, in  Proceedings of the Third  National
      Outdoor  Action  Conference  on Aquifer Restoration,  Ground  Water Monitoring,   and
      Geophysical Methods.  May 22-25.  Orlando, Florida. National Water Well Association.
Sen, P.K.  1968.   Estimates of the regression coefficient based on Kendall's tau.  Journal of the
      American Statistical Association 63:1379-1389.
Storck, P., A.J.  Valocchi, and J.W. Eheart.   1995.   Optimal Location  of Monitoring Wells for
      Detection of Groundwater Contamination in Three-Dimensional Heterogeneous Aquifers, in
      Wagner, B.J., T.H. Illangasekare, and K.H. Jensen, eds., Models for Assessing and Monitoring
      Groundwater  Quality  -  Proceedings  of the  Prague  Conference  of the International
      Association ofHydrological Sciences.  Boulder, Colorado.  IAHS Publication No. 227.

Tuckfield, R.C., E.P. Shine, R.A. Hiergesell, M.E. Denham, S.  Reboul,  and C. Beardsley. 2001.
      Using Geoscience and  Geostatistics  to Optimize Groundwater Monitoring Networks at the
      Savannah River Site. U.S. Department of Energy Publication No. WSRC-MS-2001-00145.


                                            46

-------
U.S. Air  Force.   2004.   Comprehensive  Results Report for the Passive Diffusion  Sampler
      Demonstration. Draft. August.
U.S. Army Corps of Engineers (USAGE). 2001. Draft Logistics Center (FTLE-33) Remedial Action
      Monitoring Network Optimization Report. May.
U.S. Environmental Protection  Agency  (U.S. EPA).   1994a.   Guidance for the Data  Quality
      Objectives Process.   U.S. EPA  Office of Research and Development.   EPA QA/G-4.
      EPA/600/R-94/055. September.
U.S. EPA.  1994b.  Methods for Monitoring Pump-and-Treat Performance.  U.S. EPA Office of
      Research and Development. EPA/600/R-94/123.
U.S. EPA.  1996. Guidance for Data  Quality Assessment - Practical Methods for Data Analysis.
      U.S. EPA Office of Research and Development.  EPAQA/G-9.  EPA/600/R-96/084. July.
U.S. EPA.  2000. Drinking Water Standards and Health Advisories.  U.S. EPA  Office of Water.
      EPA822-B-00-001. Summer.
U.S. EPA. 2004.  Guidance for Monitoring at Hazardous Waste Sites - Framework for Monitoring
      Plan Development and Implementation.  U.S. EPA Office of Solid Waste and Emergency
      Response.  OSWER Directive No. 9355.4-28. January.
Ward, R.C., J.C.  Loftis, and G.B. McBride.  1990. Design of Water Quality Monitoring Systems.
      Van Nostrand Reinhold. New York, New York.
Wiedemeier, T.H. and P.E. Hass. 1999. Designing Monitoring Programs to Effectively Evaluate the
      Performance of Natural Attenuation. AFCEE. August.
Zhou, Y.  1996.  Sampling frequency for  monitoring the  actual state  of groundwater systems.
      Journal of Hydrology 180(3):301:318.
                                          47

-------

-------
                                                              Office of Solid Waste and
                                                              Emergency Response
542-R-04-001a
September 2004
www.clu-in.org
www.epa.gov/tio
U.S. EPA National Service Center
for Environmental Publications
P.O. Box 42419
Cincinnati, OH 45242-2419

-------