v>EPA
Unittd States
Environmental Protection
Agency
Office of
Solid Waste and
Emergency Response
.07A-
DIRECTIVE NUMBER: 9
9355. £-
TITLE: Data Quality Objectives Development
Guidance for Uncontrolled Hazardous
Waste Site Remedial Response Activities
APPROVAL DATE:
EFFECTIVE DATE:
ORIGINATING OFFICE:
D FINAL
£3 DRAFT
STATUS: Revision
REFERENCE (other documents):
OS WER OS WER OS WER
VE DIRECTIVE DIRECTIVE Dl
-------
£EPA
United States Environmental Protection Agency
Washington. DC 20460
OSWER Directive Initiation Request
Interim Directive Number
9355.07A
Originator Information
Name of Contact Person
Randall Kaltreider
Mail Code
WH-548E
Telephone Number
382-2448
Approved for Review
OUST
OWPE
D AA-OSWER
Title
Data Quality Objectives Development Guidance for Uncontrolled Hazardous Waste Site
Remedial Response Activities.
Summary of Directive
Provide guidance in the development of data quality objectives (DQO's) for remedial
response activities. DQOs are qualitative and quantitative statements specifyinq-
the data required to support remedial response decisions. DQOs are established
>rior to data collection to address site-specific requirements and are based on the
intended uses of the data. The DQO quidance consists of two documents.
0 DQO Development Guidance for Uncontrolled Hazardous Waste Site Remedial
Response Activities - Provides qeneral discussion on the DQO process
addressinq both analytical and samplinq considerations.
0 DQO Development Example for Uncontrolled Hazardous Waste Site Remedial Response
Activities - Scenario 1 - Presents a site-specific application of the DQO
process for RI/FS activities at a site with contaminated soils and qround
water.
Key words: DQOs, RI/FS, RD/RA, Samplinq and Analysis, QA/QC
Type of Directive (Manual. Policy Directive, Announcement, etc.)
Guidance document
Status
1*1 Draft
D Final
New
I I Revision
Does this Directive Supersede Previous Directive(s)? [XJ Yes | | No Does It Supplement Previous Directive(s)? |x Yes I I No
If "Yes" to Either Question. What Directive (number, title) 9355.05C
9355.07 9355.0-6B
Review Plan
D AA-OSWER D OUST
0 OERR E OWPE
H] QSW ID Regions
La OECM
Q OGC
[3 OPPE
Lo Other (Specify)
CAMS, ATSDR, ASTSWMO,
USAGE, REM contractors
This Request Meets OSWER Directives System Format
Signawfe of Leld Office Directives Officer
Date
Signature of OSWER Directives Officer
Date
EPA Form 1315-17 (10-85)
-------
OSWER Directive 9355.0-7A
DRAFT
DATA QUALITY OBJECTIVES
DEVELOPMENT OT11NCE FOR
UNCONTROLLED HAZARDOUS WASTE SITE
REMEDIAL RESPONSE ACTIVITIES
Office of Emergency and Remedial Response
Office of Waste Programs Enforcement
Office of Solid Waste and Emergency Response
U.S. Environmental Protection Agency
401 M Street, SW
Washington, DC 20460
October 17, 1986
-------
OSWER Directive 9355.0-7A
TABLE OF CONTENTS
1.0 INTRODUCTION 1-1
1.1 PURPOSE 1-2
1.2 DATA QUALITY OBJECTIVE POLICY BACKGROUND 1-4
1.3 FORMAT 1-4
2.0 DATA QUALITY OBJECTIVE DEVELOPMENT PROCESS 2-1
2.1 DQO STAGES 2-1
2.1.1 STAGE 1 - IDENTIFY DECISION TYPES 2-1
2.1.2 STAGE 2 - IDENTIFY DATA USES/NEEDS 2-3
2.1.3 STAGE 3 - DESIGN DATA COLLECTION PROGRAM 2-3
2.2 RI/FS PROCESS 2-3
2.2.1 GENERAL APPROACH 2-3
2.2.2 PHASED RI/FS APPROACH 2-5
2.3 REMEDIAL DESIGN 2-7
2.4 REMEDIAL ACTION 2-7
2.5 DATA QUALITY OBJECTIVES DOCUMENTATION 2-8
2.6 REFERENCES 2-8
3.0 RI/FS DQO STAGE 1 - IDENTIFY DECISION TYPES 3-1
3.1 IDENTIFY AND INVOLVE DATA USERS 3-1
3.1.1 DECISION MAKER'S ROLE 3-3
3.1.2 DATA USERS' ROLE 3-3
3.2 EVALUATE AVAILABLE INFORMATION 3-6
3.2.1 DESCRIBE CURRENT SITUATION 3-6
3.2.2 REVIEW AVAILABLE DATA 3-8
3.2.3 ASSESS ADEQUACY OF DATA 3-10
3.3 DEVELOP CONCEPTUAL MODEL 3-11
3.3.1 EVALUATION MODELS 3-13
3.3.2 COMPUTER MODELS 3-16
3.4 SPECIFY OBJECTIVES/DECISIONS 3-18
3.4.1 DECISION TYPES 3-18
3.4.2 DETERMINE NEED FOR ADDITIONAL DATA 3-20
3.5 REFERENCES 3-24
4.0 RI/FS STAGE 2 IDENTIFY DATA USES/NEEDS 4-1
4.1 IDENTIFY DATA USES 4-3
4.1.1 DATA USE CATEGORIES 4-4
4.1.2 RI/FS USES 4-8
4.2 IDENTIFY DATA TYPES 4-9
4.3 IDENTIFY DATA QUALITY NEEDS 4-12
4.3.1 DATA QUALITY FACTORS 4-12
4.3.2 COST ANALYSIS OF ALTERNATIVES 4-21
4.4 IDENTIFY DATA QUANTITY NEEDS 4-22
4.5 EVALUATE SAMPLING AND ANALYSIS OPTIONS 4-24
4.5.1 SAMPLING AND ANALYSIS COMPONENTS 4-25
AW3E-7 ii
-------
OSWER Directive 9355.0-7A
TABLE OF CONTENTS
(Continued)
4.5.2 SAMPLING AND ANALYSIS APPROACH (PHASING) 4-26
4.5.3 RESOURCE CONSIDERATIONS 4-30
4.6 REVIEW PARCC PARAMETER INFORMATION 4-32
4.6.1 PRECISION 4-33
4.6.2 ACCURACY 4-33
4.6.3 REPRESENTATIVENESS 4-34
4.6.4 COMPLETENESS 4-35
4.6.5 COMPARABILITY 4-35
4.7 UTILIZING PARCC PARAMETER INFORMATION 4-36
4.8 REFERENCES 4-37
5.0 RI/FS STAGE 3 - DESIGN DATA COLLECTION PROGRAM 5-1
5.1 ASSEMBLE DATA COLLECTION COMPONENTS 5-1
5.2 DEVELOP DATA COLLECTION DOCUMENTATION 5-1
5.2.1 SAMPLING AND ANALYSIS PLANS 5-3
5.2.2 WORK PLANS 5-5
5.2.3 ENFORCEMENT CONCERNS 5-6
5.3 REFERENCES 5-8
6.0 REMEDIAL DESIGN (RESERVED) 6-1
7.0 REMEDIAL ACTION (RESERVED) 7-1
8.0 STATISTICAL CONSIDERATIONS 8-1
8.1 DETERMINATION OF NUMBER OF SAMPLES 8-1
8.2 DETERMINATION OF TOTAL UNCERTAINTY 8-6
8.2.1 UNCERTAINTY ASSOCIATED WITH ONE SAMPLE ANALYSIS 8-7
8.2.2 TOTAL UNCERTAINTY ESTIMATES WHEN MANY DATA ARE
AVAILABLE 8-12
8.2.3 TOTAL UNCERTAINTY WHEN MANY DATA ARE AVAILABLE
AND LABORATORY UNCERTAINTY IS KNOWN 8-12
8.3 PROBABILITY OF LOCATING A CONTAMINATED ZONE 8-12
8.4 CONFIDENCE LIMITS ON ESTIMATES OF MEAN
CONTAMINATION 8-15
8.5 GEOSTATISTICS 8-19
8.5.1 LOCAL ESTIMATION OF CONTAMINATION 8-22
8.5.2 LOCAL ESTIMATION OF PROBABILITY 8-24
8.6 REFERENCES 8-26
9.0 ANALYTICAL CONSIDERATIONS 9-1
9.1 ANALYTICAL SUPPORT LEVELS 9-1
9.1.1 LEVEL V ANALYTICAL SUPPORT - NON-STANDARD METHODS .. 9-2
9.1.2 LEVEL IV ANALYTICAL SUPPORT - CONTRACT LABORATORY
PROGRAM (CLP) ROUTINE ANALYTICAL SERVICES (RAS) -
INVITATION FOR BIDS (IFB) 9-4
9.1.3 LEVEL III ANALYTICAL SUPPORT - LABORATORY ANALYSIS.. 9-6
AW3E-7 iii
-------
OSWER Directive 9355.0-7A
TABLE OF CONTENTS
(Continued)
9.1.4 LEVEL II ANALYTICAL SUPPORT- FIELD ANALYSIS 9-8
9.1.5 LEVEL I ANALYTICAL SUPPORT - FIELD SCREENING 9-15
9.2 ANALYTICAL FACTORS 9-16
9.2.1 ANALYTICAL QUALITY CONTROL 9-17
9.2.2 INSTRUMENTATION OPTIONS 9-18
9.2.3 -MEDIA VARIABILITY 9-20
9.2.4 METHOD DETECTION LIMIT 9-21
9.2.5 MATRIX EFFECTS 9-22
9.2.6 TENTATIVELY IDENTIFIED ORGANIC COMPOUNDS (TICS) 9-23
9.2.7 DATA QUALIFIERS 9-23
9.3 ANALYTICAL UNCERTAINTY 9-24
9.3.1 LEVEL IV 9-25
9.3.2 LEVEL III 9-26
9.3.3 LEVEL II 9-27
9.3.4 LEVEL I 9-27
9.4 REFERENCES 9-27
10.0 SAMPLING CONSIDERATIONS 10-1
10.1 SAMPLING STRATEGY 10-1
10.2 SAMPLING PROGRESSION 10-2
10.2.1 REVIEW OF EXISTING INFORMATION/DATA 10-4
10.2.2 REMOTE SENSING 10-4
10.2.3 FIELD SCREENING 10-7
10.2.4 INTRUSIVE SAMPLING 10-7
10.2.5 PILOT STUDIES 10-9
10.3 SOURCES OF VARIABILITY 10-10
10.3.1 SAMPLING/HANDLING VARIABILITY 10-10
10.3.2 TEMPORAL VARIABILITY 10-13
10.3.3 SPATIAL VARIABILITY 10-18
10.4 SAMPLE TYPES 10-20
10.4.1 MEDIA VS. WASTE SAMPLES 10-20
10.4.2 COMPOSITE VS. GRAB SAMPLES 10-22
10.4.3 RANDOM VS. NON RANDOM SAMPLING 10-23
10.4.4 BIASED VS. UNBIASED SAMPLING 10-25
10.5 SAMPLING PATTERN 10-27
10.5.1 GRID SYSTEMS 10-28
10.5.2 STRATIFICATION 10-33
10.5.3 GRID SPACING 10-34
10.6 QUALITY CONTROL SAMPLES 10-35
10.6.1 BACKGROUND SAMPLES • 10-35
10.6.2 CRITICAL SAMPLES 10-36
10.6.3 COLLOCATED AND REPLICATE SAMPLES 10-37
10.6.4 SPLIT SAMPLES 10-38
10.6.5 TRIP AND FIELD BLANKS 10-39
10.6.6 MATRIX SPIKES 10-41
10.7 REFERENCES 10-42
AW3E-7 iv
-------
OSWER Directive 9355.0-7A
TABLE OF CONTENTS
(Continued)
APPENDIX A REVIEW OF QAMS DQO CHECKLIST
APPENDIX B POTENTIALLY APPLICABLE OR RELEVANT AND APPROPRIATE REQUIREMENTS
APPENDIX C HISTORICAL PRECISION AND ACCURACY DATA CLASSIFIED BY MEDIA
BY ANALYTICAL LEVEL
APPENDIX D RCRA APPENDIX VIII CLP HSL COMPARISON
APPENDIX E CONTRACT REQUIRED DETECTION LIMITS FOR HSL ANALYSES USING
CLP IFB PROCEDURES
AW3E-7
-------
OSWER Directive 9355.0-7A
LIST OF FIGURES
Figure Page
2-1 DQO Three-Stage Process 2-2
2-2 Phased RI/FS Approach and the DQO Process 2-6
3-1 DQO Stage 1 Elements 3-2
3-2 Decision Makers Data Users Interaction 3-4
3-3 Elements of a Conceptual Evaluation Model 3-12
3-4 Example Conceptual Model Illustration 3-14
3-5 Relationship of Risk and Data Quality/Quantity 3-21
3-6 Relationship of Data Available for Making a Decision and Risk . 3-23
4-1 DQO Stage 2 Elements 4-2
4-2 Sample Type Specification Logic Diagram 4-10
4-3 Integration of Analytical Support Levels 4-30
5-1 Stage 3 Elements Design Data Collection Program 5-2
8-1 Normally Distributed Data 8-3
8-2 Contaminant Concentration/Distance Relationship 8-5
8-2 Example Probability that Reported Value Exceeds Action Level
Versus Percentage Recovery Plot (RSD=30%) 8-11
8-4 Hit & Miss Example 8-16
8-8 Example of Kriging 8-23
8-9 Probability Map 8-25
10-1 Approximate Ranges of Applicability of VOC Removal Techniques . 10-11
10-2 Relative Rise in Perched Water Table and Ground Water Table
in Response to Rainfall Event 10-17
10-3 Square Grid System 10-30
10-4 Triangular Grid System 10-31
10-5 Modified Grid System to Account for Directional Correlation ... 10-32
AW3E-7 vi
-------
OSWER Directive 9355.0-7A
LIST OF TABLES
Table Page
3-1 Generic RI/FS Objectives 3-19
4-1 Data Uses 4-5
4-2 DQO Summary Form 4-6
4-3 Summary of Analytical Levels 4-15
4-4 Appropriate Analytical Levels 4-16
5-1 Quality Assurance Project Plan Elements 5-4
9-1 RI/FS Analytical Levels 9-3
9-2 Comparison of PCB Screening Results 9-13
9-3 X-Ray Flourescence Analyzer Raw Sample Split Data 9-14
10-1 Conventional Water Quality Parameters for Treatability Studies . 10-12
10-2 Guidelines for Minimum QA/QC Samples for Field Sampling Programs 10-40
AW3E-7 vii
-------
OSWER Directive 9355.0-7A
PREFACE
This document provides guidance in the development of data quality
objectives (DQOs) for remedial response activities under the Comprehensive
Environmental Response, Compensation and Liability Act of 1980 (CERCLA).
This document, Data Quality Objectives Development Guidance for
Uncontrolled Hazardous Waste Site Remedial Response Activities, is intended
to guide the user through the process of developing data quality objectives
(DQOs) for site-specific remedial activities. Remedial response activities
include remedial investigations (RI), feasibility studies (FS), remedial
design (RD), and remedial action (RA). This guidance manual should be used
in conjunction with the Data Quality Objectives Development Example for
Uncontrolled Hazardous Waste Site Remedial Response Activities (Example
Scenario I - RI/FS Activities at a Site With Contaminated Soils And Ground
Water) which provides an outline of how the DQO process is applied to a
hypothetical site situation. Additional example case studies will be
developed in the future, addressing different site scenarios.
This is one of a series of guidance documents prepared in accordance with
the National Oil and Hazardous Substance Pollution Contingency Plan (NCP)
final rule, published in the Federal Register November 20, 1985 and
effective February 18, 1986. These guidance documents have been prepared
under the direction of the Office of Solid Waste And Emergency Response
(OSWER). The guidance document series includes the following titles:
o Guidance on Remedial Investigations Under CERCLA (EPA 540/G-85/002)
o Guidance on Feasibility Studies Under CERCLA (EPA 540/G-85/003)
o Superfund Remedial Design and Remedial Action Guidance (OSWER
Directive 9355.0-4A)
o Quality Assurance/Field Operation Methods Manual (Draft, March 1986)
o Superfund Public Health Evaluation Manual (OSWER Directive 9285.4-1)
o Superfund Exposure Assessment Manual (OSWER Directive 9285.5-1)
Guidance on Remedial investigations Under CERCLA and Guidance on
Feasibility Studies Under CERCLA were both issued in June 1985. These
documents provide direction for the planning and execution of RI/FS
projects consistent with legislation and site-specific requirements. The
Superfund Remedial Design and Remedial Action Guidance provides guidance in
the planning, administration, and management of remedial design and
remedial action at Superfund sites. The draft Quality Assurance/Field
Operations Method Manual presents detailed descriptions of the mechanics of
data and information collection during the RI/FS process. The Superfund
Public Health Evaluation Manual establishes a framework for analyzing
public health risks associated with hazardous waste sites during the RI/FS
process.
AW3E-7 viii
-------
OSWER Directive 9355.0-7A
Collectively, these documents provide guidance for the development and
performance of technically sound and cost-effective remedial response
activities which will support the program goals of both the Office of
Emergency and Remedial Response (OERR) and the Office of Waste Programs
Enforcement (OWPE). These documents are also available for use by state
agencies and private parties conducting remedial actions to ensure that
their activities are consistent with the intent of CERCLA.
AW3E-7 ix
-------
OSWER Directive 9355.0-7A
ACKNOWLEDGMENTS
This document was developed for the Office of Solid Waste and Emergency
Response (OSWER) by a task force composed of the following individuals:
Randall Kaltreider (Hazardous Site Control Division, OERR)
Linda Boornazian (CERCLA Enforcement Division, OWPE)
Andrew Szilagyi (Camp Dresser & McKee Inc.)
Jeffery Sullivan (Camp Dresser & McKee Inc.)
RoseMary Ellersick (Camp Dresser & McKee Inc.)
Tom Pedersen (Camp Dresser & McKee Inc.)
James Occhialini (Camp Dresser & McKee Inc.)
Dennis Gagne (Region I, Waste Management Division)
Bill Coakley (Region 2, Environmental Services Division)
Edward Shoener (Region 3, Hazardous Waste Management Division)
Diane Moshman (Region 5, Waste Management Division)
Steve Lemons (Region 6, Environmental Services Division)
Bill Bunn (Region 7, Environmental Services Division)
Mike Carter (Hazardous Response Support Division, OERR)
Duane Geuder (Hazardous Response Support Division, OERR)
Michael Kosakowski (CERCLA Enforcement Division, OWPE)
Dennisse Beauchamp (CERCLA Enforcement Division, OWPE)
Gary Liberson (Lloyd Associates)
Paul Clay (NUS Corporation)
Craig Zamuda (Policy Analysis Staff, OERR)
John Warren (Statistical Policy Branch, OPPE)
Wendy Sydow (Camp Dresser & McKee Inc.)
Helpful suggestions and comments on the draft document were provided by the
following as well as other EPA and contractor staff.
David F. Doyle (Camp Dresser & McKee Inc.)
Dean Neptune (QAMS)
Gene Brantly (RTI)
Daniel Michael (RTI)
AW3E-7
-------
OSWER Directive 9355.0-7A
LIST OF ACRONYMS
ATSDR Agency for Toxic Substances and Disease Registry
CERCLA Comprehensive Environmental Response, Compensation, and Liability
Act of 1980 (Superfund)
CDC Centers for Disease Control
CLP Contract Laboratory Program
DQO Data Quality Objective
EMSL-LV Environmental Monitoring and Support Laboratory - Las Vegas
ESD Environmental Services Division (of EPA)
FIT Field Investigation Team
FS Feasibility Study
HSL Hazardous Substance List
MDL Method Detection Limit
MBS National Bureau of Standards
NCP National Contingency Plan
NEIC National Enforcement Investigation Center
NPL National Priorities List
PARCC Precision, Accuracy/ Representativeness, Completeness,
Comparability
PRP Potentially Responsible Party
QAMS Quality Assurance Management Staff
QAPP Quality Assurance Program Plan
QAPjP Quality Assurance Project Plan
RA Remedial Action
RAS Routine Analytical Service
RD Remedial Design
RI Remedial Investigation
ROD Record of Decision
RPM Remedial Project Manager - federal official designated by EPA or
another lead agency to coordinate, monitor, or direct remedial or
other response activities under the NCP (Section 300.6)
RSCC Regional Sample Control Center
SAS Special Analytical Service
SMO Sample Management Office
SRM Standard Reference Materials
AW3E-7 xi
-------
Section 1.0
-------
OSWER Directive 9355.0-7A
1.0 INTRODUCTION
Data quality objectives (DQOs) are qualitative and quantitative statements
which outline the decision making process and specify the data required to
support Agency decisions during remedial response activities. Remedial
response activities include remedial investigations (RI), feasibility
studies (PS), remedial design (RD), and remedial actions (RA). Individual
site characteristics make it impossible to apply a generic set of DQOs to
all CERCLA activities; therefore, site-specific DQOs must be developed
based on the proposed end uses of the data from sampling and analytical
activities.
In order to ensure that the data generated during remedial response
activities are adequate to support decisions, a clear definition of the
decisions should be established early in the planning of remedial response
activities. These determinations are facilitated through the development
of data quality objectives (DQOs).
It is important to realize that DQOs are an integrated set of thought
processes which define data quality requirements based on the identified
end use of the data base. The DQO is not a separate deliverable. The
analysis of sampling and analytical options provided in this example
document will not appear explicitly in either the work plan or sampling and
analysis plan. However, the analysis presented in this example will occur
during project scoping, and meetings and phone conversations between the
RPM and data users. The rationale behind the selection of a particular
sampling and analysis option will appear in meeting minutes or internal
memos which will become part of the project file. The result of the DQO
process will be a well thought out sampling and analysis plan which details
the chosen sampling and analysis option.
DQOs are established prior to data collection and are critical in
developing a sampling and analytical plan (S&A Plan) consistent with CERCLA
program objectives. DQOs are developed to address the specific
requirements of individual sites and are based on the intended uses of the
data. Through implementation of the DQO process it is possible to
AW3D-4 1-1
-------
OSWER Directive 9355.0-7A
calculate the level of uncertainty associated with the data collected
during remedial response activities. It is important to note that this
calculation can only be done accurately with a large (i.e. approximately 20
data points) existing data base. With the limited amount of data usually
available for Superfund sites (especially at the start of the RI), the
calculation of the level of uncertainty can only be done at the conclusion
of data analysis. The level of uncertainty can then be used in making
decisions regarding site remediation.
\
Data quality objectives should be specified for each data collection
activity associated with a remedial response. The majority of data
collection activities will be undertaken during a remedial investigation
(RI) but additional data needs may be identified during the feasibility
study (FS), remedial design (RD), and remedial action (RA).
All investigation activities should be conducted and documented such that
sufficient data of known quality are collected to make sound decisions
concerning remedial action selection. This applies to fund-lead, federal
or state enforcement-lead, and potentially responsible party-lead projects.
1.1 PURPOSE
The purpose of this guidance document is to identify the framework and
process by which DQOs are developed and the individuals responsible for
development of DQOs. The document also identifies the relationship of the
DQO guidance to other guidance documents and the timeframe in which DQOs
are developed.
The variable nature of remedial response activities precludes development
of generic DQOs for use throughout the Superfund program. This document is
intended to guide the user through the process of DQO development. Each
site will have a unique history, data availability, site characteristics,
public and institutional considerations, and other factors. Therefore, a
unique set of DQOs must be developed for each site. Investigators are
expected to take advantage of previous experiece and data collected through
work on sites with similar media or contamination problems.
AW3D-4 1-2
-------
OSWER Directive 9355.0-7A
There are a number of factors that lead to variations among site-specific
projects. Project teams are made up of individuals with varying
backgrounds and levels of expertise; and each site presents a unique set of
conditions. This DQO guidance acts as a supplement to the existing
guidance by providing procedures for determining a quantifiable degree of
certainty which can be used in making site-specific decisions.
The DQO process ensures that a plan is developed describing the level and
extent of sampling and analysis required to produce data adequate for the
evaluation of remedial alternatives, design of the selected alternative,
and verification of remedial effectiveness. In actual practice to date,
projects conducted under CERCLA have complied with the intent of the DQO
process. DQOs have been incorporated as parts of sampling and analytical
plans, quality assurance project plans or work plans. The purpose of this
guidance is to provide a more formal approach to S&A plan development and
to improve the overall quality and cost effectiveness of data collection
and analysis activities.
Adherence to the guidance for DQO development presented in this document
should not require additional paperwork. Rather, the guidance builds upon
the already established process of development of work plans, quality
assurance project plans and sampling and analysis plans.
1.2 DATA QUALITY OBJECTIVE POLICY BACKGROUND
Mr. Alvin Aim, then Deputy Administrator of the EPA, in his memorandum of
May 24, 1984 to the Assistant Administrators (AAs), stated that one of the
most important steps in assuring the quality of environmental data is
development of DQOs. He requested active participation of the AAs in the
development of DQOs during the stages in which policy and guidance is
crucial, and asked for identification and scheduling of significant,
ongoing environmental data collection activities. The Quality Assurance
Management Staff (QAMS) issued guidance to assist the Agency in development
of DQOs in October 1984. A checklist for DQO review was then issued in a
memorandum from Stan Blacker on April 3, 1985. Appendix A includes a
AW3D-4 1-3
-------
OSWER Directive 9355.0-7A
comparison of this checklist with this DQO guidance document. Additional
guidance on the development of DQOs, specifically related to Stages 1 and
2 of the process, was provided in a draft document issued by QAMS March 17,
1986.
The approach to developing and implementing DQOs for remedial response
activities has been established by a DQO Task Force comprising technical
personnel from EPA Headquarters (OERR and OWPE), Regions 1, 2, 3, 5, 6 and
7; and EPA remedial contractors. The methodology used by the DQO Task Force
involved applying the guidance provided by QAMS to the remedial response
process. The efforts of the Task Force included identifying the elements
of the DQO process within existing planning documents and organizing them
into a formal implementation approach. The DQO development process
presented in this document is based on the best available information but
may be revised as additional information becomes available.
1.3 FORMAT
The Data Quality Objective Development Guidance Document for Uncontrolled
Hazardous Waste Site Remedial Response Activities includes the following
sections:
1.0 Introduction
2.0 DQO Development Process - provides an overview of the
process for developing DQOs and a discussion of how DQO
development relates to the remedial response program.
3.0 RI/FS-DQO Stage 1 - provides a description of Stage 1 of
the DQO process including identification and involvement
of data users, development of a conceptual site model
and defining decision types that will be made during the
RI/FS process.
4.0 RI/FS-DQO Stage 2 - describes approaches for determining
data needs and uses and for establishing criteria for
decisions and outlines methods by which analytical and
sampling options are identified and selected. The
sampling and analytical requirements or goals, such as
percentage recovery, are established during Stage 2.
AW3D-4 1-4
-------
OSWER Directive 9355.0-7A
5.0 RI/FS-DQO Stage 3 - describes the approach for
assembling sampling and analytical components into an
overall sampling design and identifies the documentation
required for a sampling and analytical program.
6.0 Remedial Design - Reserved
7.0 Remedial Action - Reserved
8.0 Statistical Considerations - provides a description of
some statistical approaches which may be applied during
the course of remedial action program. Scenarios are
presented to illustrate the applicability of statistical
techniques. References cited provide additional details
on statistical methods for interested readers.
9.0 Analytical Considerations - describes the various
options that are available for use in analyzing samples
obtained at uncontrolled hazardous waste sites. The
discussion is directed toward identification of
screening techniques and analytical approaches which
will result in expediting site investigations and reduce
the costs associated with analysis of samples.
10.0 Sampling Considerations - provides discussion of
sampling rationale related to the DQO development
process. The discussion is limited to.general issues
which should be considered when designing a sampling
program for remedial action programs. Details on
sampling mechanics and standard operating procedures are
documented in cited references.
Appendices to the DQO document provide information on the QAMs DQO
checklist, established criteria for RI/FS activities, and CLP performance
criteria.
As discussed in Section 1.0, the DQO process is specified for each data
collection activity, and therefore specific sections of this Manual are
applicable to specific components of the remedial response process.
As such, Sections 1, 2, 8, 9 and 10 are applicable to all remedial response
activities while Sections 3, 4 and 5 apply to the RI/FS process. Sections
6 and 7 are forthcoming and will provide guidance for the application of
DQOs to Remedial Design Activities (Section 6) and to Remedial Actions
(Section 7).
AW3D-4 1-5
-------
OSWER Directive 9355.0-7A
i
A companion to this guidance is the Data Quality Objectives Development
Example for uncontrolled Hazardous Waste Site Remedial Response Activities
(Example Scenario I) (EPA 1986) which provides an example case study of
implementation of the DQO process.
Additional example case studies will be developed in the future addressing
different site scenaries.
AW3D-4 1-6
-------
Section 2.0
-------
OSWER Directive 9355.0-7A
2.0 DATA QUALITY OBJECTIVE DEVELOPMENT PROCESS
Data quality objectives are identified during the course of the project
scoping and during development of sampling and analysis plans. DQOs are
established to ensure that the data collected are sufficient and of
adequate quality for their intended uses. Data collected and analyzed in
conformance with the DQO process described in this document can be used in
assessing the uncertainty associated with each remedial response
alternative.
2.1 DQO STAGES
Data Quality Objectives are developed through a three-stage process as
illustrated in Figure 2-1. Although the three stages are identified and
discussed sequentially in this guidance document, they should be undertaken
in an interactive and iterative manner whereby all the elements of the DQO
process are continually reviewed and applied during the remedial response
program. As such, the DQO process is applied and the resultant S&A plan is
developed at the onset of a remedial response project and revised or
expanded as needed based upon the results of each data collection activity.
This process is illustrated in the companion volume, Data Quality Objective
Development Example for Uncontrolled Hazardous Waste Site Remedial Response
Activities (Example Scenario I — RI/FS Activities at a Site with
Contaminated Soils and Ground Water (EPA 1986). It is anticipated that
several DQO Example Manuals addressing a range of different site conditions
(landfills, lagoons, etc.) will be developed in the future.
2.1.1 STAGE 1 - IDENTIFY DECISION TYPES
Stage 1 of the DQO process provides the foundation for Stages 2 and 3.
Stage 1 is undertaken to define the types of decisions which will be made.
In Stage 1, all available information on the site is compiled and analyzed
to develop a conceptual model understanding of the site. " This model
describes suspected sources, contaminant pathways, and potential receptors.
The model facilitates identification of decisions which must be made and
deficiencies in the existing information. Stage 1 activities include
AW3D-6 2-1
-------
STAGE 1
IDENTIFY DECISION TYPES
• IDENTIFY & INVOLVE DATA USERS
• EVALUATE AVAILABLE DATA
• DEVELOP CONCEPTUAL MODEL
• SPECIFY OBJECTIVES/DECISIONS
STAGE 2
IDENTIFY DATA USES/NEEDS
• IDENTIFY DATA USES
• IDENTIFY DATA TYPES
• IDENTIFY DATA QUALITY NEEDS
• IDENTIFY DATA QUANTITY NEEDS
• EVALUATE SAMPLING/ANALYSIS OPTIONS
• REVIEW PARCC PARAMETERS
i
STAGE 3
DESIGN DATA COLLECTION PROGRAM
• ASSEMBLE DATA COLLECTION COMPONENTS
• DEVELOP DATA COLLECTION DOCUMENTATION
FIGURE 2-1
DQO THREE-STAGE PROCESS
-------
OSWER Directive 9355.0-7A
defining program objectives and identifying and involving end-users
of the data. Stage 1 results in the specification of the decision making
process and forming an understanding of why new data are needed.
2.1.2 STAGE 2 - IDENTIFY DATA USES/NEEDS
Stage 2 results in the stipulation of the criteria for determining data
adequacy. This stage involves specifying the level of data certainty
sufficient to meet the objectives specified in Stage 1. Stage 2 includes
selection of the sampling approaches and the analytical options for the
site, including evaluation of multiple-option approaches to effect more
timely or cost-effective data collection and evaluation.
2.1.3 STAGE 3 - DESIGN DATA COLLECTION PROGRAM
Stage 3 results in the specification of the methods by which sufficient
data of acceptable quality and quantity will be obtained to make decisions.
This information is provided in documents such as the S&A plan or work
plan.
2.2 RI/FS PROCESS
2.2.1 GENERAL APPROACH
The overall objective of an RI/FS is to determine the nature and extent of
the threat posed by the release of hazardous substances and to evaluate
proposed remedies. The ultimate goal is to select a cost-effective
remedial alternative which mitigates threats to and provides protection of
public health, welfare, and the environment, consistent with the NCP.
RIs encompass the data gathering activities undertaken to determine the
degree and extent of contamination associated with an uncontrolled
hazardous waste site. The data obtained are used in the identification,
screening, and evaluation of remedial alternatives. The objective of the
AW3D-6 2-3
-------
OSWER Directive 9355.0-7A
remedial investigation is to collect the necessary and sufficient data to
determine the distribution and migration of contaminants; identify cleanup
criteria; and identify and support the remedial alternative technical
feasibility evaluation, public health evaluation, environmental assessment,
and cost analysis.
Remedial investigations are conducted concurrently with the feasibility
study (FS) in an iterative process wherein data are evaluated with respect
to their application to remedial alternatives. Data collection and
evaluation during the remedial investigation is performed only to the
extent needed to identify and evaluate remedial alternatives. The remedial
investigation must provide data to demonstrate the need for remedial
action, determine the extent of remedial action required, and evaluate the
feasibility of the potential remedial alternatives.
Feasibility studies entail development, screening, and evaluation of
remedial alternatives in a systematic manner. The objectives of the
FS are to develop and evaluate the remedial action alternatives with
respect to technical, public health, environmental, institutional, and cost
considerations. In order to ensure that adequate and sufficient data are
collected for performance of the FS, site managers must continually
coordinate the evaluation of data collected during the RI.
Although the major data collection process occurs during the RI, both the
RI and FS objectives must be taken into consideration during the scoping
process. The RI/FS typically addresses data collection and site
characterization from the perspective of contaminant source and contaminant
migration pathways. Once pathways are established and human and
environmental receptors are identified, further data collection efforts can
be directed toward evaluating the potential impact upon receptors, and for
use in evaluating potential remedial technologies and alternatives.
The DQO process is part of the RI/FS process currently in place as outlined
in EPA's Guidance on Remedial Investigations Under CERCLA (EPA 1985a) and
Guidance on Feasibility Studies Under CERCLA (EPA 1985b).
AW3D-6 2-4
-------
OSWER Directive 9355.0-7A
Through the process of developing DQOs, a series of statements and
definitions of the types, quantity and quality of data will be developed.
The purpose (or use) for which the data collection activity is being
undertaken will also be specified. The DQO process does not require that
any deliverables beyond those currently utilized in the RI/FS program be
developed. Rather, it provides guidance on the method for determining
appropriate means of identifying sampling/analysis options and
identification of the level of detail which should be developed to support
data collection activities.
2.2.2 PHASED RI/FS APPROACH
RI/FSs are undertaken at sites which have potential contaminant problems
confirmed through preliminary assessments (PA), site investigations (SI),
and hazard ranking system (HRS) scoring. Although data are collected
during the performance of PAs, Sis, and HRS scoring, these data are
generally limited in nature and may not provide adequate information on the
spatial distribution of contaminants at the site on which to base a
complete RI/FS.
The amount and quality of data required to support selection of a remedial
alternative will vary by site. In most situations it may not be possible
to identify all data needs during the initial scoping activities. Rather,
data needs will become more clearly defined as additional data are obtained
and evaluated. By separating the remedial investigation into distinct
phases, data can be collected and evaluated sequentially with a refinement
or redefinition of data collection needs at the completion of each phase.
Figure 2-2 illustrates the phased RI/FS approach.
It is seldom possible to identify fully all the data needed to complete an
RI/FS at the outset of the scoping process. For complex sites the phased
approach allows for added control of investigative activities when compared
to a singular sampling/analysis event. Through the application of the DQO
process to a phased investigation the usability of the data is improved,
thereby improving the cost effectiveness of the investigation.
AW3D-6 2-5
-------
INITIATION OF
RI/FS
FIGURE 2-2
PHASED RI/FS APPROACH AND THE DQO PROCESS
-------
OSWER Directive 9355.0-7A
2.3 REMEDIAL DESIGN
Following selection of a remedy (based on the RI/FS) and Record of Decision
(ROD)/Enforcement Decision Document (EDO) approved by the designated EPA
official, action must be taken to initiate design activities. The remedial
design encompasses the preparation of the final construction plans and
specifications to accomplish the remedial action alternative as defined in
the ROD/EDO.
Remedial actions, especially those involving onsite treatment or disposal
of contaminated wastes, may require additional field data collection
activities during the remedial design phase to supplement the technical
data available from the RI/FS, so that optimum methods and associated costs
may be determined or fine tuned and refined.
Data collected during the RD should refine cost estimates to the +15/-10%
range. The type of data required at this late stage of the remedial
response process varies depending on the type of remedies. For soil
excavation, a good estimate of contaminated soil volume is needed; for
treatment options, a refined estimate of the physical/chemical waste
character may be required.
The DQO process as mentioned earlier is equally applicable to data
collection activities performed during the RD as during the RI/FS. The
practical application of DQOs to RD activities will be described in future
updates to this document.
2.4 REMEDIAL ACTION
Following completion and approval of the RD package, action must be taken
to initiate remedial action (RA) activities. RA activities entail the
actual implementation of the preferred alternative selected in the ROD/EDO.
As with the RD, additional data collection activities may have to be
conducted during the RA, and the DQO process utilized. Data collected
AW3D-6 2-7
-------
OSWER Directive 9355.0-7A
during the RA is used to evaluate the progress of the RA and to verify at
the end of the RA that the set performance criteria were achieved.
2.5 DMA QUALITY OBJECTIVES DOCUMENTATION
The DQO development process is initiated during the scoping process and is
completed in conjunction with the development of a sampling and analysis
(S&A) plan for each project phase. The various stages of the DQO
development process are interactive in nature. As additional details
regarding the site are discovered during the scoping process the decisions
which will be made during the project are further refined. This refinement
of decisions allows for further specification of data needs and for design
of the data collection program.
It should be noted that as the DQO process continues, the scoping of the
project will become refined. It may be determined that additional decision
types are needed (Stage 1), or that data collection activities previously
identified should be modified (Stage 2 and Stage 3) as a result of
evaluation of data (Stage 1) collected during earlier phases of the RI.
Development of DQOs in a formal manner ensures that the appropriate data
are obtained in order to meet the objectives of the RI/FS, RD or RA.
Documentation of DQOs can be provided primarily in the sampling and
analysis plan (which includes QAPjP elements), with summary information
listed in the work plan.
2.6 REFERENCES
U.S. Environmental Protection Agency. 1985a. Guidance on Remedial
Investigations Under CERCLA. Office of Emergency and Remedial Response,
Office of Waste Programs Enforcement, Office of Solid Waste and Emergency
Response, Washington, DC. Office of Research and Development,
Cincinnati, Ohio. EPA/540/G-85/002. June.
. 1985b. Guidance on Feasibility Studies Under CERCLA. Office of
Emergency and Remedial Response, Office of Waste Programs Enforcement,
Office of Solid Waste and Emergency Response, Washington, DC. Office of
Research and Development, Cincinnati, Ohio. EPA/540/G-85/003. June.
AW3D-6 2-8
-------
Section 3.0
-------
OSWER Directive 9355.0-7A
3.0 RI/FS DQO STAGE 1 - IDENTIFY DECISION TYPES
Stage 1 of the DQO process is undertaken to identify the individuals
responsible for decisions, to identify and involve data users and to define
the types of decisions which will be made as part of each site specific
RI/FS. The actual decisions are made following collection and evaluation
of data at various points during the RI/FS. The general decision types are
identified early in Stage 1 to ensure that an investigative approach to
which specified degrees of certainty can be assigned is established early
in the planning process.
The major elements of Stage 1 include:
o Identify and involve data users
o Evaluate available information
o Develop conceptual model
o Specify RI/FS objectives and decisions
Stage 1 of the DQO process is an inherent part of the project scoping
process. The thought process by which a work plan is developed naturally
encompasses the Stage 1 DQO elements. Figure 3-1 provides a simplified
illustration of the Stage 1 elements. Although the elements of Stage 1 can
be thought of as distinct steps, they in fact constitute a continuous
thought process. The elements of Stage 1 are combined to identify the
types of decisions which will be made during the RI/FS process.
3.1 IDENTIFY AND INVOLVE DATA USERS
Data quality objectives are developed through a process which requires the
involvement of the data users early in the planning of the remedial
activities. Because of the interdisciplinary nature of remedial
activities, it is important that the appropriate technical expertise is
identified and involved in the DQO development process.
AW3D-5 3-1
-------
IDENTIFY & INVOLVE DATA USERS
EVALUATE
AVAILABLE DATA
DEVELOP CONCEPTUAL MODEL
- CONTAMINANT SOURCES
- MIGRATION PATHWAYS
- POTENTIAL RECEPTORS
- CONTAMINANTS OF CONCERN
SPECIFY OBJECTIVES/DECISIONS
FIGURE 3-1
DQO STAGE 1 ELEMENTS
-------
OSWER Directive 9355.0-7A
3.1.1 DECISION MAKER'S ROLE
The key RI/FS decision is remedy selection (i.e. ROD/EDO signature). For
the majority of RI/FS projects, remedy selection is delegated to the
Regional Administrator (RA) who is guided by the NCP and program policy.
Program Management responsibilities are delegated to the Waste Management
Division Director and managers, with project specific management and
oversight assigned to Remedial Project Managers (RPMs).
The EPA RPM is the designated decision maker for the DQO development
process. In this role, the RPM is responsible for coordinating the DQO
development process, and overseeing remedial contractors, state officials,
or private parties conducting the RI/FS to assure consistency with program
policy/guidance with direction provided by Regional Management.
For federal lead projects, the RPM delegates a certain extent of site
specific decision making to the remedial contractor's site manager
performing the RI/FS. For state lead or private party lead projects, the
RPM delegates a certain extent of site specific decision making to the
state project manager or private party project manager, who in turn
delegates some site specific decision making to their contractor's site
manager. However, the RPM should be in continual communication with the
federal remedial contractor, state project manager, or private party
project manager, so key decisions to be made can be concurred upon.
3.1.2 DATA USERS' ROLE
The interactions of decision makers and various data users during the DQO
devlopment process is illustrated in Figure 3-2 and discussed below.
Primary Data Users
Primary data users are those individuals involved in ongoing RI/FS
activities. These activities include RI/FS planning and implementation,
project management and oversight, site specific decision making, and DQO
devopment. For federal lead projects, this includes the RPM and the
AW3D-5 3-3
-------
DIVISION MGMT
• PROGRAM & PROJECT
OVERSIGHT
• ROD/SETTLEMENT
RECOMMENDATIONS TO RA
RA/AA
• ROD/SETTLEMENT
DECISIONS
ENFORCEMENT
• NEGOTIATIONS
STATE
• REMEDY
CONCURRENCE
DQO
DECISION MAKER
RPM
PRIMARY DATA USERS
RPM
CONTRACTOR'S SITE MANAGER & STAFF
STATE PROJECT MANAGER
(STATE LEAD PROJECT)
PRIVATE PARTY PROJECT MANAGER
(PRIVATE PARTY LEAD)
ATSDR
• HEALTH
ASSESSMENTS
CORPS
RD/RA ACTIVITIES
(FEDERAL LEAD PROJEC i
SECONDARY USERS
(ENF/STATE/ATSDR/CORPS)
•GENERIC DATA NEEDS
ATSDR PAPER
STATE STANDARDS
CHAIN OF CUSTODY
• SITE SPECIFIC DATA NEEDS
SPECIAL PATHWAY INFO
PRP IDENTIFICATION
RD/RA NEEDS
TECHNICAL SUPPORT
(ESD/OTHERS/TAC TEAM)
• UPON REQUEST BY RPM
SAMPLING/ANALYTICAL
SUPPORT
REGULATORY REQUIREMENTS
PROJ REVIEW/AUDIT
(ESD/ORC/HQ)
• QA INTEGRITY
• COMPLIANCE WITH POLICY
•>• INPUT
OUTPUT
FIGURE 3-2
DECISION MAKER DATA USERS INTERACTION
-------
OSWER Directive 9355.0-7A
remedial contractor's site manager and staff. For state lead or private
party lead projects, this includes the state or private party manager and
their contractor's site manager/staff, along with the RPM.
The contractor's site manager has the primary responsibility for
incorporating DQOs into planning and implementation activities. The
contractor site manager must have a basic understanding of the site under
consideration and the pertinent issues at hand. This understanding serves
as the basis for identification of the appropriate contractor staff to be
involved in the process. The contractor site manager must be cognizant of
the range of issues pertaining to the site and be comfortable with the
objectives of the RI/FS (and RD/RA as the remedial process progresses).
This comfort factor can only be gained through a thorough review of
available information, familiarization with the site and its environs
through a site inspection, and discussion of the situation with appropriate
primary data users.
The contractor site manager must identify the appropriate contractor
technical staff which need to be involved in the RI/FS, based upon the
overall problems posed by the site at initiation of the project. For
example, if ground water contamination is of concern, geologists/
hydrogeologists and water supply or water treatment engineers may be
involved at a minimum. If surface water contamination is of concern,
aquatic biologists, limnologists and water resource engineers may be
involved. The involvement of chemists in the RI/FS process is essential
since in all RI/FS projects samples are likely to be analyzed. Analytical
chemists can assist in specifying the types of analyses to be used and the
limitations of the particular techniques or methods, individuals familiar
with the interactions of chemicals in the environment, such as geochemists,
soil scientists, and chemists must also be involved so that the data
generated can be used to assess its impact on the environment. Geostati-
sticians can provide assistance in evaluating spatially distributed data.
Toxicologists and individuals familiar with risk assessments should also be
involved early in the scoping process in order to ensure that appropriate
consideration is given to potential migration pathways, receptors and
contaminants of concern.
AW3D-5 3-5
-------
OSWER Directive 9355.0-7A
Secondary Data Users
Secondary data users include all individuals or parties that rely on RI/FS
project specific outputs to support their programmatic activities.
Secondary data users provide input to the decision maker and primary data
users during the DQO development process through generic data needs and, on
occasion, site specific data needs. Depending on project lead, secondary
data users may include the State, Enforcement personnel, ATSDR, Corps of
Engineers (Corps), etc.
Technical Support and Project Review/Audit
At the request of the RPM, technical specialists such as BSD personnel,
other EPA personnel, etc. may provide technical support related to project
specific sampling/analytical activities, regulatory requirements, etc.
Project review/audit personnel such as BSD, Office of Regional Counsel, and
EPA HQ help ensure QA program integrity and compliance with program policy.
3.2 EVALUATE AVAILABLE INFORMATION
A review and evaluation of the information available for a site is
undertaken as an initial step in the RI/FS process. This review provides
the foundation for additional on-site activities and serves as the database
for RI/FS scoping (preparation of the work plan). The evaluation of
available information can be summarized in a narrative report or in a file
and should contain an interpretation of the site conditions based upon a
review of existing information, and of the initial site inspection.
3.2.1 DESCRIBE CURRENT SITUATION
A narrative summary of the existing information on the site should be
assembled as an initial step in the RI/FS scoping process. Information
should be obtained from EPA technical and enforcement files, state/local
regulatory agency files, USGS files, and other relevant sources. Files
AW3D-5 3-6
-------
OSWER Directive 9355.0-7A
from potentially responsible parties (PRPs) should also be referenced when
available. A detailed list of potential data sources is contained in
Section 2.0 of the Guidance for Remedial Investigations Under CERCIA
(EPA 1985a).
During the initial evaluation of available information for an Rl, the data
are confirmed by on-site observations. The intent of evaluating and
confirming the available data is to develop an objective assessment of the
site conditions.
The goals of the initial site inspection are as follows:
o Utilizing field analytical procedures, obtain data on volatile
chemical contaminants, radioactivity, and explosivity hazards that
may be present on-site in order to determine appropriate health and
safety levels to be utilized.
o Determine if any conditions pose an imminent danger to public health
o Confirm the information contained in previous documents.
o Record observable data identified as missing in previous documents.
o Update site conditions if undocumented changes have occurred.
o Perform an inventory of possible off-site sources of contamination.
o Obtain data such as location of access routes, sampling points and
the site organization requirements for the field investigation.
o Determine the applicability and feasibility of proposed Rl
activities.
It is essential to have aerial photographs or a compiled map of the site
available for the initial site inspection. Maps should provide a scaled
compilation of the best available information on the site and include:
o Topography of the site
o Identification of pertinent physical site features, (e.g.,
buildings, water bodies, water courses, wetland areas,
access points, property boundaries, wooded or vegetated
areas)
AW3D-5 3-7
-------
OSWER Directive 9355.0-7A
o Delineation, to the extent possible, of the areas of waste
storage or contamination, both historic and existing
The compiled map and/or aerial photographs should be developed to include a
reasonable area outside of the legal site boundaries to provide an
understanding of land use on adjacent properties and to identify
potentially sensitive off-site receptors. The compiled map and/or aerial
photographs provide an efficient tool to confirm existing site conditions;
record field notes, direct reading instrumentation, measurements, locations
of key photographs, and observations; and identify potential future
sampling locations.
Tasks such as geophysical (magnetometer, EM) surveys, limited field
screening or limited field analysis may be performed during the initial
site inspection. This type of initial sampling may help determine the
variability of the material, provide background information, assess changes
in site conditions, or determine if site conditions have changed.
Confirmatory activities which may be undertaken during the initial site
inspection include locating, numbering and labeling, photographing,
securing, and recording the condition of on-site stored waste and ground
water monitoring wells; identifying the number of occupied residences in
the vicinity of the site; and determining the adequacy and condition of the
site security system.
3.2.2 REVIEW AVAILABLE DATA
For many sites, previous studies have provided useful information upon
which further investigations can be based. For each of the major areas in
the remedial action process, all available relevant information should be
compiled and organized in a manner to fulfill the goals of the activities
to be identified. The quality of the data developed through previous
efforts should be analyzed to ensure that it is truly useful. These
evaluations determine the uncertainty associated with the conclusions drawn
from the data.
A number of factors relate to the quality of data and its adequacy for use
in the RI/FS process, including the following considerations:
AW3D-5 3-8
-------
OSWER Directive 9355.0-7A
Age/comparabi 1 ity - How long ago were the da.ta collected? The
user must determine if the data are relevant or comparable to the
present situation. It is not unusual for 2 years to lapse between
the site investigation and initiation of the RI/FS.
Analytic Methods - Were the analytic methods used consistent with
present practices? Methods need not be identical in order to be
comparable. Higher detection limits do not necessarily imply that
the older data are inadequate.
Detection Limits - As implied above, care should be taken to
determine if the detection limits of the analytical tests were
sensitive to the standards and criteria used in evaluating data.
QA/QC - Determine the quality and usability of the existing data
by asking questions such as: Are the spike recoveries acceptable
for the intended use? Were the laboratory blanks contaminated?
Careful evaluation of QA/QC data is essential in determining the
comparability of data files. Intralaboratory bias data are
essential for this evaluation.
Methods used for sample collection are as important to consider as the
methods used for sample analysis. These considerations fall into two broad
categories: statistical and standard operating procedures (SOPs). The
statistical considerations relate to the representativeness of the data and
the level of confidence that may be placed in conclusions drawn from the
data. The SOPs identify well construction methods and other issues related
to sample collection. Following SOPs ensures sample integrity and data
comparability and reduces sampling and analytical error. Typical issues to
consider include the following:
Sample Objective - What was the purpose for collecting samples?
Were the samples collected using a random or non-random sampling
approach? Was the sampling plan adequate? (i.e., were a
sufficient number of samples collected?) Was the sampling plan
followed? Were there deviations from the sampling plan?
Sample Collection - Where and how were the samples collected?
What methods were used for sample collection? Was equipment
decontaminated appropriately prior to use? Did individuals
obtaining samples have proper training?
Chain of Custody - Were chain of custody procedures followed if
the samples were analyzed off-site? If not, this may not mean
that the information cannot be used. It merely implies that one
should discuss with legal staff the extent to which any
conclusions may depend on these data. If the data are critical in
the decision-making process, a determination should be made if the
data would be legally defensible.
AW3D-5 3-9
-------
OSWER Directive 9355.0-7A
Sample Preservation - Were the samples preserved properly? Poor
preservation may only mean that the results actually understate
the true extent of the contamination or if the samples were not
filtered but preserved with acids, an overestimation of metal
concentration may result.
Sample Shipment - How were the samples shipped? Were the organic
samples iced?This again relates to how the data can be
interpreted.
Holding Times - How long were the samples held before being
analyzed?As before, this could relate to the amount of
contaminant found. For example, when holding times are exceeded
for volatile organics, the likelihood of a change in concentration
increases.
If limited or no information exists on sample collection, preservation
techniques or holding times the data should be interpreted with caution.
3.2.3 ASSESS ADEQUACY OF DATA
The uncertainty associated with each data measurement activity should be
considered when data are evaluated. Although data may be validated
analytically, the level of precision of a particular data point may not
provide sufficient certainty for use in a decision.
It is important to recognize the distinction between uncertainty associated
with a measurement activity and uncertainty associated with a decision
during development of DQOs. The uncertainty associated with each
measurement activity is based upon the sampling and analytical variable.
The contractor's site manager should determine to what extent data are
valid for use prior to incorporating the data into risk assessments or
other evaluations.
Data validation activities will result in the identification of invalid
data and qualification of the usability of the remaining data. However, a
more detailed evaluation is necessary to assess whether the measurement
activity provided a true representation of the conditions as they exist at
the site. For example, data received from an analytical laboratory may
report a contaminant concentration of 35 ppm. It may be determined that
AW3D-5 3-10
-------
OSWER Directive 9355.0-7A
this sample is below the specified action level of 50 ppm. However, the
supporting analytical documentation may indicate that the true
concentration value may fall within a range + 70 percent of the reported
value. The true concentration of the sample may therefore range from 10.5
to 59.5 ppm (35 ppm + 70 percent). Therefore, it may not be possible to
state that value is above or below the action limit with any degree of
certainty when only one data point is available.
The decision made regarding the uncertainty associated with each data
differs from the uncertainty assigned to a decision regarding site
remediation. Decisions regarding site remediation are based upon a
compilation of all data points into a cohesive statement regarding, for
example, the areal extent of contamination. The data users can then
delineate areas requiring remediation based on specific action levels. The
uncertainty associated with the decision incorporates the uncertainty at
each data point as well as the entire area delineated. These types of
discussions can only be made within specific degrees of certainty if
detailed statistical evaluation of the data is undertaken. Details
regarding establishment of criteria and action levels are discussed in
(Stage 2) of this document.
3.3 DEVELOP CONCEPTUAL MODEL
Conceptual models are narrative descriptions of an uncontrolled hazardous
waste site and its environs which present hypotheses regarding the
contaminants present on site, their routes of migration, and their
potential impact on sensitive receptors. The hypotheses presented are
tested, redefined and modified during the course of the RI/FS. Figure 3-3
depicts the basic elements of a conceptual model for an uncontrolled
hazardous waste site. These elements are expanded upon to develop a
written description of the site and its environs based on available
information. Chapter 2 in Guidance on Remedial Investigations Under CERCIA
(EPA 1985a) provides additional details for development of the conceptual
model.
AW3D-5 3-11
-------
SOURCE
PATHWAY
VARIABLES
CONTAMINANTS
CONCENTRATION
TIME
MEDIA
RATE OF MIGRATION
TIME
TYPE
SENSITIVITY
TIME
HYPOTHESIS
TO BE
TESTED
SOURCE EXISTS
SOURCE CAN BE
CONTAINED
SOURCE CAN
BE REMOVED
AND DISPOSED
SOURCE CAN
BE TREATED
PATHWAY EXISTS
PATHWAY CAN
BE RENOVATED
PATHWAY CAN
BE ELIMINATED
RECEPTORS ARE NOT
IMPACTED BY MIGRATION
OF CONTAMINANTS
RECEPTOR CAN
BE RELOCATED
INSTITUTIONAL CONTROLS
CAN BE APPLIED
RECEPTORS CAN BE
PROTECTED
FIGURE 3-3
ELEMENTS OF A CONCEPTUAL EVALUATION MODEL
-------
OSWER Directive 9355.0-7A
3.3.1 EVALUATION MODELS
The conceptual model should be detailed enough to address all potential or
suspected sources of contaminant, types of contaminants and concentrations,
affected media, rates and routes of migration, and receptors. Graphical
depiction of potential routes of migration are useful for illustrating the
hypothesis under investigation. Figure 3-4 presents an illustration which
supports a narrative evaluation conceptual model.
The following are assessed during development of the conceptual model to
determine appropriate remedial and/or removal actions at a site:
o Population, environmental, and welfare concerns at risk
o Routes of exposure
o Spatial distribution of contaminants ,
o Atmospheric dispersion potential and proximity of targets (wind
rose)
o Amount, concentration, hazardous properties, environmental fate
(e.g., ability to bioaccumulate, persistence, volatility,
solubility, mobility, etc.), and form of the substance(s) present
o Hydrogeological factors (e.g., soil permeability, depth to saturate
zone, hydrologic gradients, proximity to a drinking water aquifer,
floodplains and wetlands proximity)
o Climate (rainfall, seasonal variations etc.)
o Extent to which the source can be adequately identified and
characterized
o Potential for reuse, recycling or treatment of substances at the
site
o Likelihood of future releases if the substances remain on-site
o Extent to which natural or man-made barriers currently contain the
substances and the adequacy of the barriers
o Assessment of the potential pathways of migration and a model of
such
AW3D-5 3-13
-------
VOLATILIZATION
SURFACE RUNOFF
POTENTIAL SOURCES
/ \
I J \
o ' A DRUMS
0
LAGOON
'•5
31=
cor
«
v • / — = —
PERCHED
WATER TABLE
CONTAMINATED
SOILS
MIGRATION
GLACIAL TILL
UNCONFINED AQUIFER
/
/ '
/
*
S ' s / / ~
'
/ f
/ BEDROCK
FIGURE 3-4
EXAMPLE CONCEPTUAL MODEL
ILLUSTRATION
-------
OSWER Directive 9355.0-7A
o Extent to which the substances have migrated or are expected to
migrate from their area of origin and whether migration poses a
threat to public health, welfare, or the environment
o Extent to which contamination levels exceed applicable or relevant
and appropriate federal public health or environmental standards and
criteria
o Contribution of the contamination to an air, land, or water
pollution problem
o Ability of responsible party to implement and maintain the remedy
until the threat is permanently abated
Numerous techniques are available to evaluate the fate and migration of
contaminants in environmental media. Soil contamination evaluation models
which take into consideration soil properties (texture, pH, permeability),
the characteristics of the contaminant of concern (Koc, solubility) and
environmental factors (temperature, precipitation) are useful in estimating
the movement of contaminants. These types of models are especially
meaningful when used to relate a 10" cancer risk at a receptor, for
instance, to an action level for remediation of soils on a site.
By determining the rate of migration of contaminants from a source to a
receptor by use of evaluation models a better understanding of the urgency
for implementation of remedial actions can be obtained.
The principal assumptions and calculation methods used to develop risk
assessments can also be utilized during the early phases of an RI to assist
in identifying data needs. Although it is not practical to assume that a
risk assessment can be performed during the scoping process, use of an
abbreviated approach which takes factors such as migration and
concentration of contaminants at potential receptors into consideration can
be of value in developing an evaluation model.
The process of evaluating data should be undertaken at the initiation of
any remedial action program and at each point within the program that
additional data are obtained. Additional data collected during the RI are
used to expand the conceptual model and determine if sufficient data of
AW3D-5 3-15
-------
OSWER Directive 9355.0-7A
adequate quality have been obtained to address the issues of concern. As
the data base for the site expands, the level of uncertainty associated
with making a decision should decrease. For example, at the initiation of
an RI, information may be available on the general constituents present in
a lagoon on an uncontrolled hazardous waste site. These data may indicate
that the material in the lagoon contains priority pollutants. Subsequent
sampling rounds may be undertaken to define the range of variability of
constituents within the lagoon in terms of total volume and parameters
which would effect disposal options. These additional data are then used
to evaluate disposal/treatment options and develop cost estimates for
removal.
3.3.2 COMPUTER MODELS
The more difficult and more common questions to be addressed during a
remedial action program deal with defining the extent of contamination,
setting action limits and establishing total uncertainties associated with
remedial options. These types of decisions generally require that data be
evaluated utilizing tools such as ground water contaminant migration
•simulation models, air quality models, and/or geostatistical methods.
These techniques allow for further evaluation of the data and provide the
decision maker with a data base upon which an uncertainty value associated
with removal of contaminated soils can be developed.
Ground water models include several levels of analysis: simple graphical
techniques, analytical solution techniques, and numerical solution
techniques. Using this broad definition of modeling, one of these
techniques is almost always applied to examine a ground water contamination
problem. Thus, the primary question becomes not when to use modeling, but
what level of analysis is required to meet the objectives of the study.
The role of modeling must be evaluated with respect to the entire site
investigation. The evaluation of small sites with relatively uniform
geology may be accomplished by the use of simple analytical models. This
AW3D-5 3-16
-------
OSWER Directive 9355.0-7A
is in contrast to larger sites with complex stratigraphy involving
contamination in multiple layers with variable aquifer parameters. This
complexity can only be represented by a sophisticated numerical model
involving a major effort by the hydrogeologist/modeler.
One of the most common misconceptions about ground water modeling and
geostatistical techniques is that they are applied only during the final
stages of an Rl, after all the data are collected. While model application
at the final stages of an investigation may provide useful information,
modeling techniques could be applied throughout the Rl. For example,
during the early stages of an Rl, the conceptual model can be used to guide
the data collection program. Sensitivity analyses can help identify the
types of data which need to be collected, as well as the most critical
locations for data collection. As data collection proceeds during a phased
Rl, or when a large amount of data exist from previous investigations,
models can be used to provide a consistent framework for organizing the
data. The need to compare model results with field data forces a
systematic categorization and review of existing data. During the latter
stages of an FS, models can be applied to predict the future behavior of a
ground water system under natural or artificial stresses, particularly
implementation of remedial actions such as varied pumping schemes.
The role of geostatistical techniques and ground water models is to aid
investigators in the formulation of appropriate questions concerning
planning and design problems, and to help in obtaining quantitative answers
of sufficient accuracy and detail to guide the decision maker. Models may
not provide precise answers to the questions which have been posed.
Rather, the model should be used to produce information needed to guide the
thinking underlying the decisions to be made. For systems of even
relatively low complexity, the number and nature of alternatives,
interactions, and responses which are possible far exceed the capability to
enumerate and evaluate completely. Use of geostatistics and ground water
models allow decisions to be based on objective data, not solely on
intuition or past experience.
AW3D-5 3-17
-------
OSWER Directive 9355.0-7A
Detailed discussions of computer modeling techniques are beyond the scope
of this guidance document; however, the decision maker should be aware of
these tools and determine if they should be incorporated as part of a data
evaluation system for the site. Likewise, detailed discussion of possible
geostatistical techniques is beyond the scope of this document. However,
an overview of statistical methods is provided in Section 8.0 for reference
purposes.
3.4 SPECIFY OBJECTIVES/DECISIONS
In a broad sense, the objective of remedial action program is to determine
the nature and extent of the threat posed by the release or threat of
release of hazardous substances and to select a cost effective remedial
action which can be implemented at the site to minimize the risk of
migration of or exposure to contaminants. Achieving this broad objective
requires that several complicated and interrelated activities be performed,
each having objectives, acceptable levels of uncertainty, and attendant
data quality requirements. The expression of these objectives in clear
precise decision statements is the first step toward the development of a
cost-effective program for collection of sufficient data for decision
making.
3.4.1 DECISION TYPES
Project objectives should address major areas of the remedial action
process. These include characterizing the site with respect to the
environmental setting, proximity and size of human population, and nature
of the problem; identifying potential remedies; and determining specific
performance levels of the potential remedies.
Data collected during the RI are used to support decisions regarding
remedial actions for the site. Specifying the objectives can be thought of
as identifying problems to be solved. The fact that most uncontrolled
hazardous wastes sites pose numerous problems with respect to contaminant
AW3D-5 3-18
-------
OSWER Directive 9355.0-7A
migration will result in development of objectives geared towards separate
media or sources. However, these objectives should be consistent with the
ultimate objective of selecting a remedial alternative(s) to address the
entire site.
Table 3-1 lists general RI/FS objectives. These objectives are somewhat
simplistic but will be used in the ensuing steps in the process to refine
DQOs.
Defining the types of decisions which will be made regarding remedial
actions for uncontrolled hazardous waste sites requires a clear
understanding of the problems posed by the site and awareness of the conse-
quences of making a wrong decision.
3.4.2 DETERMINE NEED FOR ADDITIONAL DATA
The consequences of a wrong decision regarding site remediation will vary
depending on the situation under consideration. For example, a decision
may be made not to implement a remedial alternative designed to mitigate
the migration of contaminants in ground water because the data indicate
that dispersion and degradation of the contaminants will reduce
concentrations to health-based levels within the area of attainment. If in
actuality the contaminants migrated beyond the waste management levels and
were encountered in the ground water system beyond the site, it may be
suggested that a wrong decision was made. The consequences of this wrong
decision made at a site where neighboring residents derive their water from
private water supply wells tapping the contaminated aquifer would be
different from the consequences of contamination of an aquifer which was
not used as a source of water supply. The consequences of a wrong decision
when individual water supply systems are contaminated would generally be
considered more serious than those associated with contamination of an
aquifer system not used as a water supply.
AW3D-5 3-19
-------
TABLE 3-1
General RI/FS Objectives
OSWER Directive 9355.0-7A
Objective
RI
Activity
FS
Activity
Determine presence or absence
of contaminants
- Determine types of contaminants
Determine quantities (concentrations)
of contaminants
Determine mechanism of contaminant
release to pathways
Determine direction of pathway(s)
transport
Determine boundaries of source(s)
and pathways
Determine environmental/public
health factors
Determine source/pathway contaminant
characteristics with respect to
mitigation (bench studies)
- Establish presence/absence of
contaminants at source and in
all pathways.
- Establish "nature" of contaminants
at source and in pathways; relate
contaminants to PRP-cost recovery
- Establish concentration gradients
Establish mechanics of source/
pathway(s) interface
Establish pathway(s)/transport
route(s), Identify potential
receptor(s)
Establish horizontal/vertical
boundaries of source(s) and
pathway(s) of contamination
Establish routes of exposure,
and environmental and public
health threat
Establish range of contaminants/
concentrations
Evaluate applicability of no action alternative
for source areas/pathways.
Evaluate environmental/public health threat;
identify applicable remedial technologies.
Evaluate costs to achieve applicable or
relevant and appropriate standards
Evaluate effectiveness of containment
technologies
Identify most effective points in
pathway to control transport of
contaminants
Evaluate costs to achieve relevant/applicable
standards; identify applicable remedial
technologies
Evaluate applicable standards or risk; identify
applicable remedial technologies
- Evaluate treatment schemes
Al • :-35
-------
OSWER Directive 9355.0-7A
The consequences of a wrong decision must be weighed for each major
decision to be made during the remedial action process. Where the
consequences of a wrong decision carry significant public health, safety or
environmental impacts, greater attention must be paid to obtaining the data
required to ensure that the decision is sound.
The information available for making a decision is related to the risk of
making a wrong decision and the significance of the consequences. As shown
in Figure 3-5, as the quantity and quality of data increase, the risk of
making a wrong decision based upon the information generally decreases.
This is not a true inverse relationship since at some point the collection
of additional data or improvement of data quality will not significantly
decrease the risk of making wrong decisions. This can best be expressed
graphically as shown in Figure 3-6. The risk of making a wrong decision
decreases as data quantity and quality increases, until it reaches a point
of diminishing returns, where additional data or increased quality of data
do not significantly reduce the risk of making a wrong decision.
It should also be noted that data quantity and data quality are independent
variables which must be considered jointly during assessment of the
consequences of making a wrong decision. Collecting increasing quantities
of data points which are of low quality may not add significantly to the
reduction of risk of making a wrong decision. Increasing the data quality
of a limited number of samples may not add significantly to the body of
knowledge to be used in making a decision.
As part of the development of the objectives for the RI/FS, the decision
making process should be outlined. Specific decisions that will be made,
when they will be made, and by whom they will be made are critical in the
outline development. Critical decisions need to be considered when
defining the data to be collected, the sampling and analytical methods, the
sensitivities of the methodologies, and the method detection limits. The
adequacy of the data which will be collected during the RI/FS to meet the
overall project objectives must therefore be evaluated in Stage 1 of the
DQO process.
AW3D-5 3-21
-------
INCREASING
RISK OF
MAKING WRONG
DECISIONS
INCREASING DATA QUALITY/QUANTITY
FIGURE 3-5
RELATIONSHIP OF RISK AND DATA
QUALITY/QUANTITY
-------
INCREASING
/.
I
DECREASING
X
INCREASING
DECREASING
FIGURE 3-6
RELATIONSHIP OF DATA AVAILABLE TO RISK
FOR MAKING A DECISION
3-23
-------
OSWER Directive 9355.0-7A
The following are some general questions which should be addressed when
determining if additional data are required.
o What are migration pathways?
o What are potential receptors?
o Are contaminants present above levels of concern at points of
receptors?
o Is background an appropriate comparison?
o Are contaminants above levels of concern as determined from
available standards or technical guidance? (e.g., What are
applicable/relevant standards? How clean is clean?)
o What are the three-dimensional (spatial) and time boundaries, of
contaminant above action levels?
o Are there migration concentration gradients that could be handled
separately?
o Are there any operable units that can be expedited in order to
protect public health and the environment (e.g., source control,
alternate water supply)?
o Which alternatives are feasible and sufficient to protect public
health and the environment.
o Is treatment a viable option? Should treatment tests or pilot
studies be conducted concurrent with the RI?
o Have sufficient data been collected so that cost estimates are
within the +50 percent to -30 percent range for RI/FS? Within +15
percent to -10 percent range for remedial design (RD)?
o Which alternative should be selected in accordance with NCP? Would
the remedy comply with other environmental laws?
The value of obtaining additional data or increasing data quality has
traditionally been based on professional judgment for RI/FS projects. The
intent of the DQO process is to provide a systematic approach for the
evaluation of the" risk associated with making a wrong decision and for
determining levels of uncertainty associated with decisions to provide a
framework for the RPM.
AW3D-5 3-24
-------
OSWER Directive 9355.0-7A
3.5 REFERENCES
Federal Register. 1985. National Oil and Hazardous Substances Pollution
Contingency Plan. Final Rule. Vol. 50, No. 224. November 20.
U.S. Environmental Protection Agency (EPA). 1985a. Guidance on Remedial
Investigations Under CERCLA. Office of Emergency and Remedial Response,
Office of Waste Programs Enforcement, Office of Solid Waste and Emergency
Response, Washington, D.C. Office of Research and Development,
Cincinnati, Ohio. EPA/540/G-85/002. June.
. 1985b. Guidance on Feasibility Studies Under CERCLA. Office of
Emergency and Remedial Response, Office of Waste Programs Enforcement,
Office of Solid Waste and Emergency Response, Washington, D.C. Office of
Research and Development, Cincinnati, Ohio. EPA/540/G-85/003. June.
AW3D-5 3-25
-------
Section 4.0
-------
OSWER Directive 9355.0-7A
4.0 RI/FS STAGE 2 - IDENTIFY DATA USES/NEEDS
Stage 2 of the DQO process is undertaken to define specific data uses, and
to specify the types of data needed to meet the project objectives.
Although data needs are identified generally during Stage I, it is during
Stage 2 that specific data uses are defined.
The major elements of Stage 2 of the DQO process, as identified in Figure
4-1, are:
o Identify data uses ,
o Identify data types
o Identify data quality needs
o Identify data quantity needs
o Evaluate sampling/analysis options
o Review PARCC parameters
Stage 2 begins after the completion of the site specific conceptual model
and the specification of the overall project objectives. The conceptual
model and the general decisions that need to be made become the basis for
determining data uses and data needs. The result of Stage 1 is a determi-
nation of the sufficiency or insufficiency of the existing data to meet the
project objectives. If the existing data are sufficient, there is no need
to collect additional data. If the data are insufficient, the types,
quality, and quantity of data which must be collected will be determined in
Stage 2.
The purpose or use to which the data will be applied is defined in terms of
the specific purpose of the project, whether it is problem definitions,
•alternative analyses, identification of PRPs, or design of remedial
actions. Identification of data needs occurs simultaneously with
identification of data uses since these factors are inseparable.
AW3D-7 4-1
-------
IDENTIFY
DATA USES
IDENTIFY DATA
TYPES
IDENTIFY
DATA QUALITY
NEEDS
IDENTIFY
DATA QUANTITY
NEEDS
EVALUATE
SAMPLING/ANALYTICAL
OPTIONS
REVIEW PARCC
PARAMETERS
FIGURE 4-1
DQO STAGE 2 ELEMENTS
-------
OSWER Directive 9355.0-7A
4.1 IDENTIFY DMA USES
Although the use of the data may appear to be relatively easy to define
during the scoping phase, it is important that a detailed evaluation of
data uses be undertaken to ensure that the amount of data are appropriate
for their intended uses.
For example, during the scoping process it may be determined that ground
water samples should be obtained in the vicinity of a site at which
contaminants have been encountered in the shallow ground water aquifer on
site. The homes in the largely rural area surrounding this hypothetical
site derive water from private wells which tap the bedrock aquifer. Based
upon the DQO approach, professional experience, and program guidelines
provided by the RPM, the contractor decides that ground water from the
bedrock aquifer should be sampled to determine if contaminants are present.
However, the more difficult questions to address during Stage 2 of the DQO
process include:
o How many samples are required?
o Where should samples be obtained?
o How many QA/QC samples are needed (field trip blanks, collocated
sample, field and laboratory duplicates, spikes)
o Will data obtained be used to determine if an alternative water
supply should be provided to affected homes?
o At what contaminant level are water supplies considered to be
affected?
o Will decisions be based upon results of data from analysis of
private water supply wells or from monitoring wells?
o If contaminants are not detected in private water supply wells but
are detected in monitoring wells, how will data be used to assess
risks to receptors?
As demonstrated, the list of questions which can be generated to evaluate a
simplistic problem in one medium can be quite extensive.
AW3D-7 4-3
-------
OSWER Directive 9355.0-7A
4.1.1 DATA USE CATEGORIES
To facilitate the sequential thought process required to effectively and
accurately answer these questions raised in Section 4.1, and thus develop
OCX) Stage 2 elements, Standard Forms DQO 1.001 and 1.002 are provided as
Tables 4-1 and 4-2. Whereas these forms are shown as tables in the
Guidance Manual, during the development of DQOs for an actual remedial
response activity, they would be completed as part of the sampling and
analysis plan. Not only can these forms be used to summarize all DQO
decisions, but also as a working checklist to assure that all of the
required elements have been addressed.
The intended uses for data to be collected during an RI/FS can be described
in general purpose categories. These categories represent generic uses but
vary on a site-by-site basis. Further, specific sites may require data for
purposes other than those described here. In this case, site-specific data
use categories will be identified. As discussed above Table 4-1 presents a
suggested format to be used in identifying data use. The categories listed
represent the most common RI/FS data uses. They do not represent different
data qualities, only different uses which may require data of a given
quality. In other words, data collected for a site at a given level of
quality may be used for different purposes. The data use categories are
briefly described below:
o Site Characterization - Data collected for site characterization
purposes are used to determine the nature and extent of
contamination at a site. This category is usually the one that
requires the most data collection. Site characterization data are
generated through the sampling and analysis of waste sources and
environmental media.
o Health and Safety - Data collected for health and safety purposes
are typically used to establish the level of protection needed for
investigators or workers at a site, and if there should be an
immediate concern for the population living within the site
vicinity. Standard practice is to collect baseline health and
safety data, followed by collecting data during any site activities
which involve disturbing baseline conditions (e.g., test-pitting,
well drilling). Health and safety data are generally collected
using real-time, direct-reading portable instruments such as a
photoionization meter.
AW3D-7 4-4
-------
S E
NAME
LOCATDN.
NUMBER _
PHASE
TABLE 4-1
DATA USES
EPAREGDN
RI1 RI2 RI3 ERA FS RD RA
DATE
CONTRACTOR-
SITE MANAGER
^"^x. DATA USE
MEDIA ^"\.
SOURCE SAMPLING
TYPE
SOIL SAMPLING
GROUND WATER SAMPLING
SURFACE WATER/SEDIMENT
SAMPLING
AIR SAMPLING
BIOLOGICAL SAMPLING
OTHER
SITE
CHARACTERIZATION
(INCLUDING
HEALTH &
SAFETY)
RISK
ASSESSMENT
EVALUATION OF
ALTERNATIVES
ENGNEERMG
DESIGN OF
ALTERNATIVES
^
MONITORING
DURING
REMEDIAL ACTION
•
V*
PRP
DETERMINATION
OTHER
NOTE: CHECK APPROPRIATE BOX (ES)
COM SFDQO 1.001
-------
TABLE 4-2
DQO SUMMARY FORM
1. SITE
NAME.
EPA
REGION
LOCATION.
NUMBER —
PHASE
RI1 RI2 RI3 ERA FS RD RA
(CIRCLEONE)
2. MEDIA
(CIRCLEONE)
SOU.
GW
SW/SED
AIR
BIO
OTHER
3. USE
(CIRCLEALL THAT
APPLY)
SITE
CHARAC.
(H&S)
RISK
ASSESS.
EVAL
ALTS.
ENGG
DESIGN
PRP
DETER
MONITORING
REMEDIAL
ACTION
OTHER
4. OBJECTIVE
5. SITE INFORMATION
AREA ,
DEPTH TO GROUND WATER
GROUND WATER USE.
SOIL TYPES
SENSITIVE RECEPTORS
6. DATA TYPES (CIRCLE APPROPRIATE DATA TYPES)
A. ANALYTICAL DATA
B. PHYSICAL DATA
PH
CONDUCTIVITY
VOA
ABN
TCLP
PESTICIDES TOX
PCB . TOC
METALS BTX
CYANIDE COD
PERMEABILITY
POROSITY
GRAIN SIZE
BULK DENSITY
HYDRAULIC HEAD
PENETRATION TEST
HARDNESS
7. SAMPLING METHOD (CIRCLE METHOD(S) TO BE USED)
ENVIRONMENTAL BIASED GRAB
SOURCE GRID COMPOSITE
NON- INTRUSIVE
INTRUSIVE
PHASED
8. ANALYTICAL LEVELS (INDICATELEVELfS) AND EQUIPMENT& METHODS)
LEVEL 1 FIELD SCREENING -EQUIPMENT
LEVEL 2 FIELD ANALYSIS - EQUIPMENT
LEVEL 3 NON-CLP LABORATORY-METHODS_
LEVEL 4 CLP/RAS - METHODS
LEVEL 5 NON STANDARD
9. SAMPLING PROCEDURES
BACKGROUND - 2 PER EVENT OR
CRITICAL (LIST)
PROCEDURES
10. QUALITY CONTROL SAMPLES (CONFIRM OR SET STANDARD)
A. FIELD
COLLOCATED-5% OR
REPLICATE- 5%OR
FIELD BLANK - 5% OR
TRIP BLANK - 1 PER DAY OR.
B. LABORATORY
REAGENT BLANK -1 PER ANALYSIS BATCH OR
REPLICATE -1 PER ANALYSIS BATCH OR
MATRIX SPIKE -1 PER ANALYSIS BATCH OR
OTHER
11. BUDGET REQUIREMENTS
BUDGET
STAFF
SCHEDULE
CONTRACTOR _
SITE MANAGER
PRIME CONTRACTOR
DATE
FOR DETAILS SEE SAMPLING & ANALYSIS PLAN
COM SF DQO 1.002
-------
OSWER Directive 9355.0-7A
Risk Assessment - Data collected for risk assessment purposes are
used to evaluate the threat posed by a site to public health and the
environment. Some of the data must be qualitative so that the
chemical/physical properties, toxicity, and persistence of
contaminants can be factored into the risk assessment. The data
must also be quantitative to the degree that they may be compared
with quantitative statements of health risk criteria (e.g., 10
lifetime cancer risk level). Therefore, a high level of data
certainty is necessary. Risk assessment data are generated through
the sampling and analysis of environmental and biological media,
particularly where the potential for human exposure is great.
The level of data quality required will be related to the precision
of the model. For models in which gross assumptions are made,
qualitative data may be adequate whereas quantitative data may be
required for use in sophisticated models.
Evaluation of Alternatives - Data collected for engineering purposes
are used to evaluate various remedial technologies. Engineering
data is collected in support of remedial alternative evaluation and
to develop cost estimates (+50 to -30 percent). This may involve
performing bench-scale or pilot scale studies to determine if a
particular process or material may be effective in mitigating site
contamination.
Engineering Design of Alternatives - Data collected during the RI/FS
can be used for engineering design purposes to develop a preliminary
data base in reference to the performance of various remedial
technologies. Data types collected during the RI/FS which are
applicable to the RD process include waste characterization and
preliminary volume estimates (these estimates usually need to be
refined further by additional data collection activities during the
RD/RA).
Monitoring During Remedial Action - During the implementation of the
remedial action, samples can be taken to assess the effectiveness of
the action. Based on the analysis of these samples, corrective
measures to improve the performance of the action may be taken.
PRP Determination - Data collected for this purpose are used to help
establish the liability at multiple-party sites. For known RPs,
data are used to link their wastes to those found on the site and to
pollutants released to the environment, and for unknown RPs, by
comparing the site wastes to pollutant profiles of known waste
streams. Data collected to document the nature and extent of
contamination, and to justify the Agency selection of the remedial
alternative as being consistent with the NCP, are also used for
injunctive actions, as well as for cost recovery.
AW3D-7 4-7
-------
OSWER Directive 9355.0-7A
The format presented in Table 4-1 can be used by contractor personnel
involved in RI/FS scoping to identify sampling and analysis needs. The
actual matrix used should be developed on a site specific basis by sampling
task and intended data use category.
Once the data use categories are listed, the intended uses must be
prioritized. Establishing an order of priority for the intended data uses
will help identify the most demanding use of each type of data, i.e., the
use requiring the highest level of confidence, and therefore the lowest
level of uncertainty. The data quality required will be a function of the
acceptable limits of uncertainty established by the decision maker. The
limits on uncertainty will drive the selection of both the analytical and
sampling approaches.
Prioritizing the intended data uses begins with examining the list of uses
for each data collection task and identifying the use which is most
important for meeting the RI/FS objectives. Uses having lesser importance
are then arranged in order, under the first priority use. This is
especially true when analytical turn-around requirements differ based on
schedule constraints.
When a secondary use requires data of a much higher quality and the number
of samples required is different than the primary data use, it may be more
advantageous to treat the two uses as separate activities by collecting two
different data sets. Consideration should be given to developing a phased
approach to the data collection, in which the design of each subsequent
data collection task for an intended use is built upon the results of the
preceding one.
4.1.2 RI/FS USES
During the evaluation of data uses, a review of the potential remedial
options which will be considered during the RI/FS must be undertaken.
AW3D-7 4-8
-------
OSWER Directive 9355.0-7A
To the extent that is both possible and appropriate, at least one
alternative shall be developed for each of the five categories listed
below, as specified in Section 300.68 of the National Contingency Plan:
o Alternatives for off-site treatment or disposal, as appropriate
o Alternatives that attain applicable or relevant and appropriate
federal public health and environmental requirements
o As appropriate, alternatives that exceed applicable or relevant and
appropriate federal public health or environmental requirements
o As appropriate, alternatives that dp not attain applicable or
relevant and appropriate federal public health and environmental
requirements but will reduce the likelihood of present or future
threat from the hazardous substances and that provide significant
protection to public health and welfare and the environment
o No action alternative.
For each of the appropriate action categories, the following information or
analysis should be considered during the DQO process:
o List of candidate remedial actions
o Method by which the initial alternatives will be screened, including
cost criteria, acceptable engineering practice criteria, and
effectiveness criteria
o Method by which the limited number of alternatives will be
evaluated, including refinement and specification of alternatives;
detailed cost estimation; and evaluation of engineering
implementation, reliability and constructibility
o Analysis methods to assess the extent to which the alternative is
expected to effectively prevent, mitigate, or minimize threats to,
and provide adequate protection of, public health, welfare, and the
environment
The remedial action process involves a number of data collection
activities, each having specific objectives. Since the objectives require
varying degrees of data quality, it is critical to identify the specific
use to which each set of data will be applied.
AW3D-7 4-9
-------
OSWER Directive 9355.0-7A
4.2 IDENTIFY DATA TYPES
Data use categories define the general purposes for which data will be
collected during the RI. By defining the intended uses for the data early
in the RI scoping process, a concise statement regarding the data types
which are needed can be developed. After identifying the data types and
uses, data quality needs can be defined, and a systematic evaluation of
sampling and analysis options can be performed.
Data types can be specified in broad groups initially, such as background
samples, or media samples. For example, environmental media samples can be
used to determine the extent of contamination at a specific site. The DQO
process requires that data types be specified to a continually more
detailed level to ensure that the data obtained are useful in meeting the
objectives of the RI/FS. Figure 4-2 illustrates the process of continual
refinement of data types for a hypothetical ground water contamination
problem. The process should be followed for each media of interest or each
source material. The result of completing the entire decision matrix is
the specification of the data type needed for each intended data use.
Since environmental media and source materials are interrelated at
uncontrolled hazardous waste sites, data types to be used to evaluate
ground water contamination may also be used to evaluate soil contamination.
By identifying data types by media, the decision maker and the data users
can discuss overlapping data needs to refine the scope of the RI.
The types of analyses which will be performed on each sample must be
determined while identifying data types. The analytical requirements are
dictated by the use of the data, which is ultimately driven by the remedial
alternative under consideration.
Evaluation of the limitations posed by various treatment/disposal options
allows one to develop a listing of analytical data types required during
the RI. For example, in order to adequately evaluate treatment/disposal
AW3D-7 4-10
-------
FIGURE 4-2
SAMPLE TYPE SPECIFICATION LOGIC DIAGRAM
-------
OSWER Directive 9355.0-7A
options for materials contained in a waste impoundment, its PCS content,
pH, halogen content, viscosity, as well as other parameters which would
influence its treatability or acceptability for disposal must be determined
during the RI.
The data types specified in Stage 2 should not be limited to chemical
analytical parameters, but should also include physical parameters such as
permeability and porosity, which are needed to evaluate contaminant
migration. The level of detail to which data types are defined during the
DQO process must be sufficient to allow for evaluation of sampling/analysis
options during subsequent stages of the DQO process.
4.3 IDENTIFY DATA QUALITY NEEDS
4.3.1 DATA QUALITY FACTORS
Consideration of data quality needs should begin with the identification of
data uses and data types. Important factors in defining data quality
include:
o Prioritized Data Uses
o Appropriate Analytical Levels
o Contaminants of Concern
o Level of Concern
o Required Detection Limit
o Critical Samples
These factors should be considered to define data quality needs in a
general way at the start of an RI/FS. As work proceeds and more data
become available, more precise statements can be made. These factors,
coupled with consideration of data quantity needs and an evaluation of
sampling and anlaysis options, lead to a more quantitative statement of
quality needs as Precision, Accuracy/ Representativeness, Completeness, and
Comparability (PARCC) parameters.
AW3D-7 4-12
-------
OSWER Directive 9355.0-7A
Data uses and their prioritization have been addressed in Section 4.1.1.
The remaining data quality factors are discussed in this section. PARCC
parameters will be reviewed in Section 4.1.6.
Appropriate Analytical Levels
The quality of measurement data needed is dependent on the end use of the
data but site conditions vary so much that guidelines applicable to generic
RI/FS data uses cannot readily be provided. In addition, there is little
or no information on many factors which critically affect data quality such
as: sample variability, sample container cleanliness, effect of different
sample collection and analytical preparation techniques, etc. Most
available measurement data quality information addresses only the
analytical technique. To provide some guidance, this section defines
analytical levels and then indicates the levels appropriate to different
generic RI/FS data uses. Section 9 of this document provides a more
detailed discussion of analytical considerations.
The analytical levels are arbitrarily defined as follows:
o Level I - field screening or analysis using portable instruments.
Results are often not compound specific and not quantitative but
results are available in real-time. It is the least costly of the
analytical options.
o Level II - field analyses using more sophisticated portable
analytical instruments; in some cases, the instruments may be set
up in a mobile laboratory onsite. There is a wide range in the
quality of data that can be generated. It depends on the use of
suitable calibration standards, reference materials, and sample
preparation equipment; and the training of the operator. Results
are available in real-time or several hours.
o Level III - all analyses performed in an offsite analytical
laboratory using methods other than CLP/RAS procedures (i.e. SW846
methods). The laboratory may or may not be a CLP laboratory.
o Level IV - CLP routine analytical services (RAS). All analyses are
performed in an offsite CLP analytical laboratory following CLP
protocols. Level IV is characterized by rigorous QA/QC protocols
and documentation.
AW3D-7 4-13
-------
OSWER Directive 9355.0-7A
Level V - analysis by non-standard (NS) methods. All analyses are
performed in an offsite analytical laboratory which may, or may
not, be a CLP laboratory. Method development or method
modification may be required for specific constituents or detection
limits. CLP special analytical services (SAS) are level NS.
In general, offsite analytical laboratories have a large amount of
analytical equipment and instrumentation enabling them to handle complex
sample matrices. With the exception of the CLP RAS, the laboratories can
use varied sample preparation and analysis techniques to optimize the
quality of the resulting data.
Levels III, IV and NS all incorporate some time lag between submission of
samples to the laboratory and receipt of results. Table 4-3 provides more
information on these analytical levels; Table 4-4 identifies appropriate
analytical levels for generic RI/FS data uses.
It can be seen from Table 4-4 that, for each generic data use, several
analytical levels may be appropriate. The decision maker needs further
criteria to select the most appropriate. Important criteria are the
contaminants of concern and the level of concern for each contaminant.
Contaminants of Concern
At some sites it may be clear which contaminants are of concern because
they have known impacts on human health. In such cases, the appropriate
health standards can be used to set action levels. Often a large number of
contaminants are found at a site. In such cases it is not feasible or
desirable to specify action levels for each observed contaminant. Rather,
a small number of indicator chemicals are selected and action levels are
determined for these chemicals. Indicator chemicals are the most toxic,
mobile, persistent, and frequently occurring contaminants found on site.
The process of selecting indicator contaminants is described in the
Superfund Public Health Evaluation Manual
(EPA 1985).
AW3D-7 4-14
-------
TABLE 4-3
SUMMARY OF ANALYTICAL LEVELS APPROPRIATE TO
DATA USES
DATA USES ANALYTICAL LEVEL
SITE CHARACTERIZATION
MONITORING DURING LEVEL 1
IMPLEMENTATION
SITECHARATERIZATION
EVALUATION OF ALTERNATIVES i R/FI II
ENGINEERING DESIGN
MONITORING DURING
IMPLEMENTATION
RISK ASSESSMENT
PRP DETERMINATION
SITE CHARACTERIZATION
EVALUATION OF ALTERNATIVES
ENQNEERNG DESIGN LEVEL Ml
MONITORING DURING
IMPLEMENTATION
RISK ASSESSMENT
PRP DETERMINATION I PVPI w
EVALUATION OF ALTERNATIVES LtvtLiv
ENGINEERING DESIGN
RISK ASSESSMENT LEVEL V
PRP DETERMINATION
TYPE OF ANALYSIS
- TOTAL ORGANIC/INORGANIC
VAPOR DETECTION USING
PORTABLE INSTRUMENTS
- VARIETY OF ORGANICS BY
GC; INORGANICS BY AA;
XRF
- TENTATIVE ID; ANALYTE-
SPECIFIC
- DETECTION LIMITS VARY
FROM LOW ppm TO LOW ppb
- ORGANICS/INORGANICS
USING EPA PROCEDURES
OTHER THAN CLP CAN BE
ANALYTE-SPECIFIC
- RCRA CHARACTERISTIC TESTS
- HSL ORGANICS/INORGANICS
BY GC/MS; AA; ICP
- LOW ppb DETECTION LIMIT
- NOI^CONVENTIAL
PARAMETERS
- METHOD-SPECIFIC
DETECTION LIMITS
-MODIFICATION OF
EXISTING METHODS
- APPENDIX 8 PARAMETERS
LIMITATIONS
- INSTRUMENTS RESPOND TO
NATURALLY-OCCURING
COMPOUNDS
- TENTATIVE ID
- TECHNIQUES/INSTRUMENTS
LIMITED MOSTLY TO
VOLATILES. METALS
- TENTATIVE ID IN SOME
CASES
-CAN PROVIDE DATA OF
SAME QUALITY AS
LEVELS IV. NS
- TENTATIVE IDENTIFICATON
OF NON-HSL PARAMETERS
- SOME TIME MAY BE REQUIRED
FOR VALIDATION OF PACKAGES
- MAY REQUIRE METHOD
DEVELOPMENT/MODIFICATION
-MECHANISM TO OBTAIN
SERVICES REQUIRES
SPECIAL LEAD TIME
DATA QUALITY
- IF INSTRUMENTS CALIBRATED
AND DATA INTERPRETED
CORRECTLY. CAN PROVIDE
INDICATION OF CONTAMINATION
- DEPENDENT ON QA/QC
STEPS EMPLOYED
- DATA TYPICALLY REPORTED
IN CONCENTRATION RANGES
- SIMILAR DETECTION
LIMITS TO CLP
- LESS RIGOROUS OA/QC
- GOAL IS DATA OF KNOWN
QUALITY
- RIGOROUS QA/QC
- METHOD-SPECIFIC
-------
TABLE 4-4
APPROPRIATE ANALYTICAL LEVELS - BY DATA USE
DATA USE
ANALYTICAL
LEVa
SITE
CHARACTERIZATION
(INCLUDING
HEALTH &
SAFETY)
RISK
ASSESSMENT
EVALUATION OF
ALTERNATIVES
ENGiNEERNG
DESIGN OF
REMEDIAL ACTION
MONITORING
DURING
IMPLEMENTATION
OF
REMEDIAL ACTION
PRP
DETERMINATION
OTHER
LEVEL 1
LEVEL II
s/
s/
LEVEL
LEVEL IV
N/
LEVEL V
OTHER
NOTE: CHECK APPROPRIATE BOX (ES)
COM SFDQO 1.001
-------
OSWER Directive 9355.0-7A
In either case, the contaminants of concern should be identified to assist
in determining data quality needs.
Levels of Concern
The level of concern specifies a concentration range above which some
action may need to be taken. The level of concern is intimately linked
with the action level.
An exact action level is not required before initiating an RI field
investigation; however, a rough estimate is necessary to ensure that the
chosen analytical methods are accurate at the level of concern. Also,
knowledge of the level of concern can influence the number of samples
required and the selection of analytical methods. For these reasons, an
acceptable range of values should be specified. As work on a site progress
and more data become available, the level of concern will be further
refined and incorporated into the ROD as an action level.
Determination of levels of concern is a site specific activity. The
decision maker and data users (toxicologists, geologists, and engineers)
must meet to determine the appropriate action level range for the site.
For sites with contaminated surface and ground water the following can
serve as a guideline to determining levels of concern.
o Maximum Contaminant Levels (MCLs) stated for 20 chemicals in the
Safe Drinking Water Act
o National Ambient Water Quality Criteria (NAWQC) developed for 95
chemicals in ambient water systems (non-drinking water) under the
Clean Water Act (Note that NAWQC have the force of law in only a few
states.)
o Health Advisories; Suggested No Adverse Response Levels (SNARLs)
developed for 22 contaminants in drinking water under the Safe
Drinking Water Act
- EPA potency factors (10 cancer risks) developed for suspected
o Critical toxicity values such as:
EPA potency fact
carcinogens, and
AW3D-7 4-17
-------
OSWER Directive 9355.0-7A
- EPA reference doses (acceptable doses) for non-carcinogens
o State water quality standards
Regulatory guidelines and criteria are generally lacking for soil and
sediment contamination. Therefore, the action level for these media must
be set specifically for each site. However, criteria have been set for
PCBs in TSCA and for dioxin.
Several tables are provided in Appendix B that summarize potentially
applicable or relevant and appropriate requirements and toxicity values.
The standards do not consider simultaneous exposure from multiple routes.
Standards may also be based on levels, durations, or frequencies of
exposure that differ from those at a specific site. The standards and
criteria that are used, especially when conducting public health
assessments, must correspond to the media for which they are developed. As
a result of the various technical aspects of standards development, some
concentration limits will require adjustment before being applied.
It should be noted that applicable or relevant and appropriate ambient
concentration limits are not available for all media for many chemicals
commonly found at Superfund sites. In addition, it is possible that there
will be overlaps in the applicable or relevant and appropriate
requirements, other criteria, and toxicity values developed for EPA's
Health Effects Assessments (HEAs). For these reasons, it will be necessary
to rank these values, when available. Often a large number of contaminants
are found at a site. In such cases it is not feasible or desirable to
specify action levels for each observed contaminant. Rather, a small
number of indicator chemicals are selected and levels of concern are
determined for these chemicals, indicator chemicals are the most toxic,
mobile, persistent, and frequently occurring contaminants found on site.
The process of selecting indicator contaminants is described in the
Superfund Public Health Evaluation Manual (EPA 1985).
In the listing of applicable standards which can be used for selecting
action levels, few standards are available for soil contamination.
Standards are not available because direct ingestion of contaminated soil
is not common and routes of exposure are site specific. Generally some
AW3D-7 4-18
-------
OSWER Directive 9355.0-7A
type of modeling may be required to specify the level of concern for soil.
The type of model selected for use will be based on the potential route of
exposure. If contaminated soil is carried in the air and inhaled by
receptors, air modeling may be required. If contaminants leach from soils
into ground water and are transported to receptor wells, a ground water
model may be required. These models are useful in assessing the potential
impact resulting from migration of on-site contaminants at a specified
level of concern to a receptor at a specified cancer risk level, for
instance. The available models are specified in the Superfund Exposure
Assessment Manual (EPA 1985).
i
In any numerical modeling procedures such as level of concern
determination, there are some input parameters which are unknown or
uncertain. Generally these parameters are determined in an iterative
procedure known as calibration or history matching. When beginning model
calibration, input parameters are assumed or estimated and the model is
run. By using the model to predict values at known data points and
computing the modeling errors, the predictive power of the model is
obtained. Based on the observed modeling errors, input parameters are then
refined so that the predictive power of the model increases.
When a model has several input parameters, the calibration process is
time consuming because there are an extremely large number of input
parameter combinations. In most modeling efforts it is impossible to test
every likely combination of input parameters. In these cases, the first
few calibration runs are performed to determine the input parameters which
when varied, produce the largest response in the model output values. The
procedure by which these parameters are determined is termed a sensitivity
analysis. Once a sensitivity analysis is performed, further calibration of
the model centers on varying the parameters identified"in the sensitivity
analysis while keeping the other parameters constant.
Model calibration is an inexact science which is best performed by experts
in the theory underlying the model. Calibrated models have no unique
solution; that is, there are many combinations of the input parameters
which will yield similar model outputs. Finally, models generally contain
AW3D-7 4-19
-------
OSWER Directive 9355.0-7A
assumptions and simplifications of the underlying mathematical theory which
allow them to be utilized on a computer. For these reasons models can make
severe prediction errors at unsampled locations and it is impossible to
predict or evaluate the effect that the accuracy and precision of the data
will have on the efficiency of the model. It is believed, however, that
the uncertainties associated with the choice of input parameters and the
non uniqueness of the model are at least an order of magnitude larger than
the precision of the data.
Detection Limit Requirements
The level of concern selected will have an immediate impact on data quality
requirements. The sampling and analysis methods selected must be capable
of accurate measurement at the level of concern. Since sampling accuracy
is hard to evaluate or control, it is extremely important that the
analytical technique chosen has a detection limit well below the level of
concern. This factor must be considered in evaluating analytical options.
Section 9.2.4 provides more detailed information on detection limits.
Appendix E lists CLP contractually required detection limits.
Critical Samples
Critical samples are those for which valid data must be obtained to satisfy
the objectives of the sampling and analysis task. The identification of
critical samples should be based on careful review of these objectives.
Critical data points should be identified in every completeness statement
developed during the DQO. An example of a critical data point may be an
upgradient well in a ground water contamination study or any other data
point considered vital to the decision making process. Critical data
points should be carefully considered when designing data collection
activity and every effort must be made to obtain valid data for these
samples. In some cases, taking critical samples in duplicate is
appropriate. It should be recognized that a common problem of any sampling
design is the loss of data during implementation of the design.
AW3D-7 4-20
-------
OSWER Directive 9355.0-7A
4.3.2 COST ANALYSIS OF ALTERNATIVES
The program goal for developing cost estimates in feasibility studies is to
estimate the cost of the selected remedial alternative to within +50
percent and -30 percent of the actual cost. This statement not only puts
requirements on the type and amount of data which must be collected during
the field investigation, it also requires the decision maker to consider
the range of potential remedial alternatives before planning the field
investigation. Thus, after developing the site conceptual model, the range
of potential remedial alternatives must be developed.
In situations where a possible alternative is source removal, the cost
criteria may be used to determine the number of data required. If the cost
of the remedial alternative is strictly proportional to the volume of
material removed, sufficient data must be obtained to determine the volume
of material to within +50 percent and -30 percent. Normally, however,
there is some uncertainty in the capital costs and the efficiency of the
treatment or removal procedure. For this reason, it may be necessary to
determine the uncertainty in the volume of contaminants much more
accurately.
In many cases the cost estimates developed for remedial actions have fallen
outside the established +50 percent to -30 percent range. The reasons for
these differences may not be solely attributed to cost estimating
deficiencies but to performance deficiencies. For instance, developing a
cost estimate for removal of 1000 yd of contaminated soils from a site can
be readily computed by utilizing unit cost estimates for excavation
($/yd3); trucking ($/yd3/taile); and disposal ($/yd3). Each of the unit
costs will have a certain degree of variability due to contractor competi-
tive bidding systems and field conditions which may affect excavation or
trucking. However, if all factors such as time required for excavation,
miles transported to disposal, and disposal costs are constant, it would be
relatively easy to develop a cost estimate within +50 percent to -30
percent of actual costs. The single most important factor in developing
the cost estimate for this example is the volume of soil to be removed.
AW3D-7 4-21
-------
OSWER Directive 9355.0-7A
During the excavation of the soil the contractor may determine that
contaminants exist outside the limits set for excavation and it may be
necessary to increase the volume of material to be removed. Such a
situation could result in a significant increase in site remediation costs.
This demonstrates the need to establish performance criteria and determine
a level of uncertainty associated with the amount of contaminated soil to
be removed.
4.4 IDENTIFY DATA QUANTITY NEEDS
The number of samples which should be collected can be determined using a
variety of approaches. The validity of the approach utilized is dependent
on the characteristics of the media under investigation and the assumptions
used to select sample locations. Experience and professional training are
the basis .for making initial determinations of where samples should be
collected and how many are required. Usually the greater the quantity of
data available for making a decision regarding sample numbers, the higher
the chances are that data will be obtained which address the project
objectives. In situations where data are not available or are limited in
nature, sampling should be undertaken in a phased approach to allow for
collection of initial samples to characterize the general conditions at the
site. These data then can be used to guide in selection of the appropriate
number of samples to be obtained in subsequent phases of the RI.
In the absence of available data, the data users and decision makers will
be required to develop a rationale for selecting sampling locations for the
initial sampling phases. Questions to be asked to guide the data users in
selecting appropriate sampling numbers and locations could include:
o Do source materials still exist on the soil surface?
o Is there evidence of soil disturbance or vegetative stress based
upon review of aerial photographs?
o Do geologic features in the area control ground water and surface
water flow patterns?
AW3D-7 4-22
-------
OSWER Directive 9355.0-7A
o Do site conditions favor surficial soil erosion or wind erosion?
o Are sensitive receptors located in the vicinity of the site?
These types of questions can be addressed in the absence of any analytical
data and will assist in identifying additional data needs. Subsequent
discussions may lead to the recommendation that geophysical surveys or
field screening be conducted in areas of soil disturbance during the
initial sampling phases. Collection of a limited number of samples from
identified source materials or pathways, such as streams, may also be
considered during the initial stage of the Rl. Limited air sampling may
also be warranted during the early stages of the RI to determine if organic
vapors or particulates could pose a problem to those performing subsequent
sampling phases or to nearby residents.
In situations where data are available, or as new data are added to the
site's data base, statistical techniques may be utilized in determining the
number of data required. A number of EPA documents provided detailed
guidance for applying statistical methods to estimate data quantity
requirements, including:
o Soil Sampling Quality Assurance User's Guide (EPA 1984)
o Sediment Sampling Quality Assurance User's Guide (EPA 1985)
The documents referenced in Section 8.0 provide more complete descriptions
of the applicability and approach for statistical techniques.
Following evaluation of the data as discussed in Stage 1, a determination
can be made as to whether the data provides adequate levels of confidence
to support a decision. If a higher degree of certainty in the data is
required (e.g., a more definitive statement regarding the extent of
contamination), then additional data should be obtained in subsequent
AW3D-7 4-23
-------
OSWER Directive 9355.0-7A
sampling phases. In all cases, the actual level of confidence which can be
attributed to a set of data can only be established following collection
and evaluation of data. Therefore, at the completion of each data
collection activity, the data should be evaluated.
Comparability in the data collection activity must take into consideration
whether the events are even comparable in the first place. An example
would be trying to compare data from the same aquifer in a high water and a
low water situation. This criterion is most important when conclusions are
being drawn from existing data. If an activity is being planned to augment
existing data, field conditions must be considered as well as sampling and
analytical techniques.
4.5 EVALUATE SAMPLING/ANALYSIS OPTIONS
Following the identification of data uses, data types, and data quality
needs, an evaluation of sampling and analysis options can be undertaken.
Numerous sampling and analysis options could be developed for any data
collection activity. The options will be dictated by the data types
needed. In addition, the sampling and analysis option evaluation must be
undertaken in a manner which ensures that data quality needs (Section 4.3)
and data quantity needs (Section 4.4) are addressed and balanced. The
evaluation of sampling and analysis options is an interactive and iterative
process which must not be undertaken independent from other elements in the
DQO process.
The evaluation of sampling and analysis options must be undertaken
considering the following factors:
o Sampling and analysis components
o Sampling and analysis approach (phasing)
o Resource constraints, schedule, seasonal and special requirements
AW3D-7 4-24
-------
OSWER Directive 9355.0-7A
4.5.1 SAMPLING AND ANALYSIS COMPONENTS
Evaluation of sampling and analysis options can only be undertaken after
all components or subsets of the sampling and analysis options are
identified. The components of a sampling and analysis plan include the
individual sample collection and analysis procedures which will result in
the data types specified. For example, in order to provide data on the
concentration of volatile organics in a monitoring well, sampling
procedures which will result in a representative sample and analytical
methods which yield the desired results must be identified.
It is critical that the contractor's site manager involve technical
personnel familiar with analytical techniques during this stage of the DQO
process.
Analytical approaches which should be considered consist of Levels I - V
(see page 4-13) which vary as to cost, time required for analysis, and the
quality of the resulting analytical data:
Section 9.0 of this document contains additional details on analytical
considerations and Section 10.0 provides discussion of sampling
considerations which should be addressed during the evaluation of sampling
and analysis components. Additional details are also contained in the
publications referenced in this document including Quality Assurance/Field
Operations Method Manual (EPA 1986).
All sampling activities should be conducted and documented such that
sufficient data are collected to make sound decisions concerning remedial
action selection. This applies to fund-lead, Federal or state enforce-
ment-lead, and potentially responsible party lead projects. The data
collection and documentation activities should be similar for all types of
remedial action project phases. In other words, if enough data are
collected using appropriate protocols, and the data are suffi-
ciently valid upon which to base a remedial action decision, then the
procedures and documentation should be sufficient to be admissible as
evidence in litigation.
AW3D-7 4-25
-------
OSWER Directive 9355.0-7A
Enforcement/cost recovery actions have at least one additional requirement.
This requirement is to identify viable PRPs. This may require additional
sample collection and more complex analysis. Regional enforcement
personnel should be consulted prior to the planning of sampling and
analytical activities to assure that their data needs will be met.
Distinction must also be made between civil and criminal cases, with the
latter usually having more stringent requirements.
4.5.2 SAMPLING AND ANALYSIS APPROACH (PHASING)
Data collection activities for each uncontrolled hazardous waste site must
be designed to ensure that the use of available and collected data is
maximized. Data collection activities must therefore be undertaken in a
manner which results in cost effective and usable data. Collection of data
beyond that needed to meet the RI/FS objectives wastes resources.
Collection of inadequate quality or insufficient data likewise results in
inefficient use of resources. Developing a sampling and analysis approach
which ensures that appropriate levels of data quantity and quality are
obtained may be accomplished by use of a phased RI approach and by the use
of field screening, field analysis or remote sensing approaches to direct
the data collection activities. By subdividing the data collection program
into a number of phases, the data can be obtained in a sequence which
allows it to be used to direct subsequent data collection activities.
The extended periods of time required for receipt of analytical data from
laboratories often results in delays in an RI program. By utilizing field
techniques for assessing contaminant concentrations or media
characteristics, the RI can proceed in a directed manner towards the
intended goal.
Direct reading instruments which should be considered for use during the
evaluation of a sampling/analysis approach include:
o Photoionization detectors (PIDs)
o Flame ionization detectors (FlDs)
AW3D-7 4-26
-------
OSWER Directive 9355.0-7A
o Hydrogen sulfide analyzers
o Hg vapor analyzers
o Respirable particulate meters
o Radiation meters
o Oxygen/explosimeters
o pH and conductivity meters
Other devices and field tests which allow for assessment of site conditions
without the need for laboratory support include:
o Oil/water interface units
o Slug tests
o Infiltrometers
These direct reading instruments can be taken into the field to obtain data
without extensive calibration procedures. Additional levels of
quantification can be obtained by the use of transportable instruments such
as gas chromatographs (GC), X-ray fluorescence or atomic adsorption
devices. For these instruments, however, calibration using known standards
must be developed prior to use in the field.
The extent to which these mechanisms can be used depends on their ability
to identify contaminant concentrations of concern. Analytical support can
be developed by integrating aspects of individual analytical support levels
into one cohesive analytical approach. This type of approach permits a
larger number of samples to be collected and analyzed cost effectively.
Conceptually, this approach can be thought of as a large "inverted funnel"
whereby large numbers of samples can be initially analyzed quickly and cost
effectively in the field with succeedingly smaller numbers of samples
analyzed further at a higher level of sophistication. This approach
combines the advantages of each level of analytical support and offsets
disadvantages. The use of less sophisticated techniques initially allows
for large numbers of samples to be screened quickly and at low cost. Next,
AW3D-7 4-27
-------
OSWER Directive 9355.0-7A
a proportion of these samples are analyzed by a more sophisticated
procedure to verify the results of the lower level analysis. If parameters
were selected for screening purposes, full analysis should be performed on
a percentage of samples to verify assumptions of chemicals present or of
concern. The type and design of this analytical approach is determined by
how the data will be used. By strategically selecting which samples are to
be analyzed at each level, a much higher degree of certainty can be
obtained for the overall data set without sacrificing either the quantity
of samples to be analyzed or the quality of data collected.
For example, consider a hazardous waste site where the soil is contaminated
with volatile organic compounds (VDCs). For this example, the objectives
of the sampling are to determine concentration of VOCs at site boundaries
and to assess the direct contact threat. It is assumed that a photoioni-
zation detector will detect contaminants at the levels of concern for this
example.
Based upon review of existing data, a sampling plan was developed which
calls for soil samples to be collected at locations determined using a
grid pattern. To illustrate this approach in general terms, data uses and
data quality are not specified in this example.
The analytical approach for this hypothetical situation is illustrated in
Figure 4-3 and summarized below:
o Sample from all locations are analyzed in real time using
photoionization field headspace techniques (Level I).
o Samples which register below the detection limits of this
instrumentation are considered clean for the purposes of this study.
o A selected number of the clean samples (for which nothing was
detected) and all of the dirty (contaminated) samples are analyzed
onsite using a portable gas chromatograph (Level II) to obtain
semiqualitative and semi quantitative results within a few days.
o A number of samples are selected for analysis by CLP Routine
Analytical Services (HAS) (Level IV) for the Hazardous Substance
List (HSL) compounds. Included in these samples are all samples
identified as critical data points (CDPs). This step provides
confirmation for all preceding work including verification that
AW3D-7 4-28
-------
DATA
QUALITY
O
1
CO
a
CO
fe
?
O
LJLJ
Q
Although not applicable to the example situation, level III support
is shown to indicate that this is a viable option for confirmatory
analyses.
COST
AND
TURNAROUND
TIME
FIGURE 4-3
INTEGRATION OF ANALYTICAL
SUPPORT LEVELS
-------
OSWER Directive 9355.0-7A
indicator parameters are representative of contaminants of concern
and are identified appropriately. The results of all split samples
analyzed by different levels are interpreted for quality control
purposes.
This approach can also be utilized in a time-phased manner, i.e., by using
the results of an initial sampling round with a lower level of analysis to
fine-tune the sampling approach for a subsequent sampling round using
higher level(s) of analytical support. Another approach involves complete
GC/MS analysis of the initial sampling round to identify the organic
compounds present followed by GC analysis of specific compound of interest
in later rounds. Gas chromatography with the appropriate detector can
provide lower cost analyses, often with lower detection limits and higher
precision and accuracy, than GC/MS. It is necessary, however, to verify by
GC/MS that interfering compounds are not present.
4.5.3 RESOURCE CONSIDERATIONS
The resources available for performance of a remedial action must be
evaluated during the scoping process. Within Stage 2 of DQO development,
the time required for obtaining data, the personnel resources and equipment
required, and the costs for data collection must be evaluated. This
evaluation is most effectively performed as sampling/analysis options are
identified.
The cost for analytical support varies considerably depending on the type
of analysis required. Schedule requirements which dictate the need for
rapid turnaround of data further escalate analytical costs. The CLP is one
source that provides analytical support to the superfund program. The cost
associated with CLP analysis of a single RAS sample for HSL Tasks 1 and 2
approaches $1,000.00. This significant cost must be taken into
consideration when evaluating analytical options in order to balance the
cost of analysis with the quality of data required. In many cases, CLP RAS
analysis may not be warranted and field screening and field analysis
techniques (including mobile labs) could provide the data required. In
other situations a total HSL analysis may not be required and savings in
analysis can be realized by use of CLP SAS services for specific compounds
of interest.
AW3D-7 4-30
-------
OSWER Directive 9355.0-7A
The cost associated with obtaining samples must also be considered during
the evaluation of sampling/analysis options. Cost savings can be achieved
by performing multiple media sampling activities simultaneously (e.g.,
sample ground water and surface water during same sampling event). The
sampling tasks which are conducted simultaneously should coincide with each
other with respect to the overall phasing approach developed for the Rl.
Technical staff resources must also be evaluated during the course of
identification of analytical/sampling options. Special training may be
required in order to undertake certain field sampling or laboratory
analysis tasks.
Development of a schedule for the sampling activities requires a
substantial effort in identifying critical path elements. The schedule
must be linked with the other components of the remedial action project
since many of these may proceed simultaneously. Critical path elements in
all sampling activities include the considerations outlined below:
° Site Access - Site access for each property involved in the site
investigation activities must be obtained. The site access
permission may be required for each sampling round or investigation
activity. Access may be denied at any time or restrictions may
prohibit planned activities. Additional information on site access
is contained in Chapter 6, Institutional Issues, of the Guidance on
Remedial Investigations Under CERCLA (EPA 1985).
o Weather - Weather can have a great impact on the schedule for field
activities. For example, geophysical studies such as seismic
refraction cannot be performed when there is a substantial layer of
snow or ground frost. Weather conditions may also delay or slow the
rate at which activities can be performed. For example,
temperatures may require the job be shut down for activities
requiring personnel protective equipment.
o Health and Safety - Health and safety requirements may increase the
time required to perform field activities. For example, limitations
of the bottled air supply required for Level B activities can
increase time required for field investigations by a factor of four.
o Subcontractor and Equipment Procurement - Procurement of
subcontractors and equipment can severely impact schedules.
Preparation of subcontractor technical specifications, evaluation of
bids, award of subcontract, and mobilization of the subcontractor
require significant time commitments. For example, the request for
AW3D-7 4-31
-------
OSWER Directive 9355.0-7A
bids may not yield an acceptable number of bids or a reasonable bid
may not be received, requiring the request for bid process to be
repeated. Equipment procurement may also delay project schedules
since much of the special equipment required for RIs is built to
order and is not available "off-the-shelf."
Sample Analysis and Data Validation - Sample analysis and data
validation must be completed before analysis of the data can be
performed. Adequate time must be provided in the schedule for data
analysis and data validation. Turnaround time for CLP analysis, for
example, ranges from 4 to 6 weeks excluding data validation. Data
validation generally requires 2-3 hours per sample for complete RAS
packages.
4.6 REVIEW PARCC PARAMETER INFORMATION
The PARCC (Precision, Accuracy, Representativeness, Completeness, and
Comparability) parameters are indicators of data quality. Ideally, the end
use of the measurement data should define the PARCC parameters that are
necessary to satisfy that end use. In the ideal situation, numerical
precision, accuracy, and completeness goals would be established and these
goals would aid in selecting the measurement methods to be used.
As noted earlier, RI/FS work does not fit this ideal situation. RI/FS
sites are so different from each other and information on overall
measurements (sampling .plus analysis) is so limited that it is not
practical to set PARCC goals at this time. Rather, the historical
precision and accuracy achieved by different analytical techniques should
be reviewed to aid in selection of the most appropriate technique.
To correlate with earlier information on analytical levels several tables
in Appendix C present precision and accuracy information for analytical
techniques classified by level. EPA will continue to make information of
this type available to data users so that, gradually, a data base of
numerical precision and accuracy requirements appropriate to different data
uses will develop.
AW3D-7 4-32
-------
OSWER Directive 9355.0-7A
4.6.1 PRECISION
Precision is a measure of the reproducibility of measurements under a given
set of conditions. Specifically, it is a quantitative measure of the
variability of a group of measurements compared to their average value.
Precision is usually stated in terms of standard deviation but other
estimates such as the coefficient of variation (relative standard
deviation), the range (maximum value minus minimum value), and the relative
range are common.
The overall precision of measurement data is a mixture of sampling and
analytical factors. Analytical precision is much easier to control and
quantify than sampling precision. There are more historical data related
to individual method performance and the "universe" is limited to the
samples received in the laboratory. In contrast, sampling precision is
unique to each site.
Sampling precision may be determined by collecting and analyzing collocated
or field replicate samples and then creating and analyzing laboratory
replicates from one or more of the field samples. The analysis results
from the collocated or field replicate samples provide data on overall
measurement precision; analysis results from the laboratory replicates
provide data on analytical precision. Subtracting the analytical precision
from the measurement precision defines the sampling precision.
4.6.2 ACCURACY
Accuracy is a measure of the bias in a measurement system; it is difficult
to measure for the entire data collection activity. Sources of error are
the sampling process, field contamination, preservation, handling, sample
matrix, sample preparation and analysis techniques. Sampling accuracy
may be assessed by evaluating the results of field/trip blanks, analytical
accuracy may be assessed through use of known and unknown QC samples and
matrix spikes.
AW3D-7 4-33
-------
OSWER Directive 9355.0-7A
As an example of how the sampling process can impact accuracy, consider the
collection of ground water samples for volatile organic analysis. In the
actual sampling, some portion of the volatile components may be lost.
There is no way to measure this loss easily. The sample could also be
subjected to contamination from a wide range of sources in the field and
laboratory. To check the system for contamination, trip and field blanks
can be used.
4.6.3 REPRESENTATIVENESS
Representativeness expresses the degree to which sample data accurately and
precisely represent a characteristic of a population, parameter variations
at a sampling point, or an environmental condition. Representativeness is
a qualitative parameter which is most concerned with the proper design of
the sampling program. The representativeness criterion is best satisfied
by making certain that sampling locations are selected properly and a
sufficient number of samples are collected.
Representativeness is addressed by describing sampling techniques and the
rationale used to select sampling locations. Sampling locations can be
biased (based on existing data, instrument surveys, observations, etc.) or
unbiased (completely random or stratified-random approaches). Either way,
the rationale used to determine sampling locations must be explicitly
explained. If a sampling grid is being utilized, it should be shown on a
map of the site. The type of sample, such as a grab or composite sample,
as well as the relevant standard operating procedure (SOP) for sample
collection, should be specified.
An example of the way representativeness is ensured in a sampling program
is the use of proper ground water sampling techniques. The SOPs for ground
water sampling require that a well be purged a certain number of well
volumes prior to sampling, to be certain that the sample is representative
of the underlying aquifer at a point in time.
AW3D-7 4-34
-------
OSWER Directive 9355.0-7A
Representativeness can be assessed by the use of collocated samples. By
definition, collocated samples are collected so that they are equally
representative of a given point in space and time. In this way, they
provide both precision and representativeness information.
4.6.4 COMPLETENESS
Completeness is defined as the percentage of measurements made which are
judged to be valid measurements. The completeness goal is essentially the
same for all data uses: that a sufficient amount of valid data be
generated. It is important that critical samples are identified and plans
made to achieve valid data for them.
Almost no historical data on the completeness achieved by individual
methods exists. However, the CLP data has been found to be 80-85% complete
on a nationwide basis. This fact can be extrapolated to indicate that
level III, IV and V analytical techniques will generate data that is
approximately 80% complete. Hence, ignoring other factors, 1.25 times more
data may be necessary. Levels I and II would be expected to have lower
completeness levels. However, since they are onsite measurement techniques
providing results in real-time or after minimal delay, invalid measurements
can be repeated easily. Thus, a high degree of completeness can be
achieved with these analytical levels.
4.6.5 COMPARABILITY
Comparability is a qualitative parameter expressing the confidence with
which one data set can be compared with another. The goal for all data
uses is the sample data are comparable with other measurement data for
comparable samples and sample conditions. This goal is achieved through
the use of standard, well-recognized techniques to collect and analyze
representative samples and report analytical results in appropriate units.
AW3D-7 4-35
-------
OSWER Directive 9355.0-7A
4.7 UTILIZING PARCC PARAMETER INFORMATION
In Stage 2 of the DQO process, the FARCC parameters should be considered in
evaluating sampling and analysis options. To the extent possible, they
should be defined as goals in the Stage 3 Data Collection Program. It must
be recognized, however, that data quality objectives can be developed for
RI/FS work without strictly defined PARCC goals.
Whenever measurement data are reviewed (in Stage 1 of the DQO process), the
PARCC parameters which were achieved should be included in the review. The
analytical laboratory should provide numerical precision and accuracy data;
Level 2 field analyses may also generate precison and accuracy data. The
data user should request this information if it is not provided.
Precision and accuracy data may be expressed in several ways and are best
evaluated by an analytical chemist or a statistician. Since the precision
data are quantifying the scatter of results about a mean value, a lower
precison value means less scatter. Accuracy is most frequently reported as
percent recovery, or percent bias. A 100 percent recovery indicates a
completely accurate measurement; the closer the recovery is to 100%, the
more accurate the measurement. Percent bias reports the difference of the
result from the true value. A completely accurate measurement would have
zero percent bias;-the lower the percent bias, the more accurate the
measurement.
The data user must keep the level of concern and the end use of the data in
mind when reviewing precision and accuracy information. In some cases,
even data of poor precision and/or accuracy may be useful. For example, if
all the results are far above the level of concern, the precision and
accuracy are much less important. However, close to the level of concern,
precision and accuracy are quite important and should be carefully
reviewed. If results have very good precision but poor accuracy, it may be
acceptable to correct the reported results using the percent recovery or
percent bias data. This judgment should be made by a data user with
appropriate technical expertise.
AW3D-7 4-36
-------
OSWER Directive 9355.0-7A
4.8 REFERENCES
U.S. Environmental Protection Agency (EPA). 1986. Quality Assurance/Field
Operations Method Manual (draft) March.
. 1983. Characterization of Hazardous Waste Sites - A Methods
Manual. Volume 1 - Site Investigations. NTIS PB84-126929.
ERA/600/4-84/075
. 1985. Guidance on Remedial Investigations Under CERCLA. Office of
Emergency and Remedial Response, Office of Waste Programs Enforcement,
Office of Solid Waste and Emergency Response, Washington, D.C. Office of
Research and Development, Cincinnati, Ohio. EPA/540/G85/002. June.
. 1985. Guidance on Feasibilility Studies Under CERCLA. Office of
Emergency and Remedial Response, Office of Waste Programs Enforcement,
Office of Solid Waste and Emergency Response, Washington, D.C. Office of
Research and Development, Cincinnati, Ohio. EPA/540/G85/003. June.
. 1985. Superfund Exposure Assessment Manual. Office of Solid Waste
and" Emergency Response. OSWER Directive 9285.5-1
. 1985. Sediment Sampling Quality Assurance User's Guide. EPA
600/4-85-048
. 1984. Superfund Public Health Evaluation Manual. Office of Solid
Waste and Emergency Response, OSWER Directive 9285.4-1
. 1984. Soil Sampling Quality Assurance User's Guide. EPA
~~6Uu/4-84-043
AW3D-7 4-37
-------
Section 5.0
-------
OSWER Directive 9355.0-7A
5.0 RI/FS STAGE 3 DESIGN DATA COLLECTION PROGRAM
Stage 3 of'the DQO process is undertaken to develop and assemble the
detailed data collection program for the remedial action project. Through
the process of addressing the elements identified in Stages 1 and 2, all
the components required for completion of Stage 3 should be available for
compilation.
Stage 3 of the DQO development process is undertaken to specify the
complete sampling and analysis approach required to meet the project
objectives as outlined in Figure 5-1.
5.1 ASSEMBLE DATA COLLECTION COMPONENTS
The data collection program should be assembled by the contractor's site
manager and staff in a coordinated manner in Stage 3. During Stage 2,
specific DQOs have been developed by media or sampling activity. These
DQOs should have been developed by contractor staff which have specialized
expertise in appropriate disciplines. The intent of Stage 3 is to compile
the information and DQOs developed for specific tasks into a comprehensive
data collection program.
The data collection program should be developed to account for all sampling
tasks and phases. During this process a detailed list of all samples to be
obtained should be assembled in a format which includes phase, media,
sample type, number of samples, sample location, analytical methods, and
QA/QC samples (type and number). In addition, a schedule for all sampling
activities should be developed in bar chart or critical path method format.
5.2 DEVELOP DATA COLLECTION DOCUMENTATION
The output of the DQO process is a well defined sampling and analysis (S&A)
plan with summary information provided in the work plan.
AW3E-2 5-1
-------
ASSEMBLE
DATA COLLECTION
COMPONENTS
DEVELOP DATA COLLECTION DOCUMENTATION
SAMPLING & ANALYSIS PLAN
Include QAPjP Elements
WORK PLAN
FIGURE 5-1
STAGE 3 ELEMENTS
DESIGN DATA COLLECTION PROGRAM
-------
OSWER Directive 9355.0-7A
Data collection documentation requirements vary on a regional basis within
the EPA. Region specific requirements for development of S&A plans and
work plans should be followed during the course of development of data
collection documentation. The DQO guidance provided in this document does
not require the submittal of deliverables in addition to those already
established in the regions. Rather, the DQO process provides a framework
to ensure that all the pertinent issues related to the collection of data
with known quality are addressed.
5.2.1 SAMPLING AND ANALYSIS PLANS
Section 300.68 of the NCP specifies that a written quality assurance/site
sampling plan must be prepared for all remedial investigation activities
which involve sampling. These plans should include the following:
o Description of the objectives of the sampling efforts with regard to
both the phase of the sampling and ultimate use of the data
o Sufficient specification of sampling protocol and procedures
o Sufficient sampling to adequately characterize the source of the
release, likely transport pathways, and/or potential receptor
exposure
o Specification of the types, locations, and frequency of samples
taken, taking into account the unique properties of the site,
including the appropriate hydrological, geological, hyrogeological,
physiographical and meteorological properties of the site
The S&A plan identifies the individuals responsible and the procedures for
field activities and sample analyses associated with remedial
investigations. Quality Assurance Project Plan (QAPjP) elements should be
addressed in the S&A plan. The standard elements of a QAPjP are listed in
Table 5-1. Details on preparation of QAPjPs are contained in Interim
Guidelines and Specification for Preparing Quality Assurance Project Plans
(EPA 1980).
The 16 points required to be addressed for each site in a QAPjP should not
be repeated in an S&A plan if the information has been documented
elsewhere. For example, if a project description (Section 3) is available
AW3E-2 5-3
-------
OSWER Directive 9355.0-7A
TABLE 5-1
QUALITY ASSURANCE PROJECT PLAN ELEMENTS
1 Title Page
Introduction
2 Table of Contents
3 Project Description
4 Project Organization
5 Quality Assurance Objectives
for Data Measurement
6 Sampling Procedure
7 Sample and Document Custody
Procedures
8 Calibration Procedures and Frequency
9 Analytical Procedures
10 Data Reduction, Validation
and Reporting
11 Internal Quality Control Checks
12 Performance and System Audits
13 Preventive Maintenance
14 Data Measurement Assessment
Procedures
15 Corrective Action
16 Quality Assurance Reports to
Management
-------
OSWER Directive 9355.0-7A
in the work plan it is acceptable to refer to this document rather than
repeat the information in the S&A plan. Sample and document custody
procedures (Section 7) of the QAPjP and calibration procedures and
frequency (Section 8) may be provided in standard operating procedures
which need not be repeated in the S&A plan but should be included by
reference. Other quality assurance issues which are program wide in
nature, such as internal quality control checks (Section 11), performance
and system audits (Section 12), corrective action (Section 15) and quality
assurance reports to management (Section 16), are generally specified in
the quality assurance program plan (QAPP).
Field investigation activities are generally undertaken in a phased
approach in which each phase reflects a further refinement of knowledge.
Therefore, separate sampling/analysis plans may be prepared for the
separate phases of a remedial investigation. For example, geophysical
investigations may be performed to select locations for installation of
monitoring wells. In such a case, a sampling plan should be prepared for
the geophysical investigations and, following evaluation of the data, a
-separate plan should be developed for installation of monitoring wells.
Additional plans for the subsequent phases of a remedial investigation may
be prepared at any time during the course of the project as the need for
additional field investigation is identified.
5.2.2 WORK PLANS
Work plans define the scope of services, level-of-effort, costs, and
schedule for performing the RI/FS. The scope of the sampling effort
depends on the quality of existing data, an understanding of the site
problems, an identification/ evaluation of the feasible remedial actions
and data necessary to support them, and enforcement needs.
Sampling should be undertaken only to the extent it is necessary and
sufficient to fulfill the requirements of subsequent remedial action
implementation and/or legal enforcement proceedings. Sampling efforts may
begin at different levels depending on (1) the understanding of the scope
of the problem at the site, and (2) whether or not there are enforcement
AW3E-2 5-5
-------
OSWER Directive 9355.0-7A
considerations. For example, surveying all areas of large and complex
sites in great detail might be an inefficient use of resources if initial
screening indicates the problem is confined to small sub-areas.
The work plan provides the general description of the activities to be
performed as part of the RI/FS. However, it does not contain the detailed
description of how a sample is obtained or an analysis performed. This
type of information is presented in the S&A plan. The level of detail to
be included in the work plan for the Rl phase is outlined below:
o How site mapping will be performed including survey limits
identified on map included in the draft work plan, the scale of the
plan to be produced, the horizontal and vertical control and
identification of significant site features
o Number of individuals to be involved in each field sampling task and
estimated duration in days
o Identification of geophysical survey areas or transects, soil boring
and test pit locations on the map provided in the draft work plan
o Number of samples to be obtained in the field including blanks and
duplicates and the location from which the samples will be obtained
illustrated on a map included in the draft work plan
o List of analyses to be performed
o Identification of pilot or bench-scale studies that will be
performed
This information is required as part of the work plan in order to establish
a basis for the schedule and cost estimate. Work plans prepared for a
phased Rl approach should be specific for the initial phase, and general
for subsequent phases, with subsequent phases well defined when the
previous phase is completed.
5.2.3 ENFORCEMENT CONCERNS
All RI/FS activities should be conducted and documented such that
sufficient data are collected to make sound decisions concerning remedial
action selection. This applies to fund-lead, and potentially responsible
party lead projects. The data collection and documentation activities
should be similar for all types of RI/FSs. In other words, if enough data
AW3E-2 5-6
-------
OSWER Directive 9355.0-7A
are collected using appropriate protocols, and the data are sufficiently
valid upon which to base a remedial action decision, then the procedures
and documentation should be sufficient to be admissible as evidence in
litigation.
The guidelines outlined below should be followed to assure that data
quality objectives are met:
o Appropriate plans (i.e., work plans, sampling and analysis plans,
QAPjP) should be developed to document intentions.
o Field notebooks should include information on field conditions,
sample location, sample number, collection time, sample description
(chain of custody forms or other mechanism can be used to record
this information).
o Personnel throughout the process from planning, sample collection,
analysis and decision making should have experience or be
sufficiently trained.
o Chain of custody must be documented with a chain of custody form for
samples taken offsite for analysis. This assures the decision maker
that the analysis given is actually for the sample collected and
that the sample has not been tampered with. If analysis is
performed onsite, documentation of the process for custody of
samples in field logs or other media is sufficient. The chain of
custody form is, however, not necessary.
o Methods used for sampling and analysis should be generally
considered valid from an engineering/scientific standpoint and be
consistent with standard analytical procedures. Methods utilized
should be referenced in the RI/FS report or other documents and a
statement given that protocols were followed. Any deviation from
the referenced method should be documented and explained.
o Documentation should be sufficient to allow the persons involved in
the site studies to reconstruct the work years later when the matter
is litigated. If the documentation is adequate, the defendents may
be convinced by the strength of the government's case not to contest
those particular points, and hence testimony by the government or
contractor employees may not be necessary.
o EPA's or the state's responsibility from a QA/QC standpoint is to
audit randomly some RI/FS field sampling, analysis (QA/QC) and data
validation to confirm that procedures utilized were sufficient.
o Actual samples, sample tags and sample bottles are not required to
be kept to prove that samples were taken and maintained. This is
the purpose of the chain of custody sheet, field notebook, or other
similar mechanisms.
AW3E-2 5-7
-------
OSWER Directive 9355.0-7A
The above requirements pertain to civil cases only. Criminal cases will
require additional documentation and/or materials.
5.3 REFERENCES
American Chemical Society. 1983. Principles of Environmental Analyses.
Analytical Chemistry 55:2210-2218.
ASTM. 1985. Quality Assurance for Environmental Measurements. ASTM
Special Tech Pub 867. J.K. Taylor and T.W. Stanley eds.
Camp Dresser & McKee Inc. 1985. Performance of Remedial Response
Activities at Uncontrolled Hazardous Waste Sites, Technical Operations
Manual. April.
Taylor, J.U. 1981. Quality Assurance of Chemical Measurements Analytical
Chemistry. Volume 53, No. 14. December.
U.S. EPA. 1980. Standard Operating Procedures and Quality Assurance
Manual-Draft. Region IV Surveillance and Analysis Division. -Water
Surveillance Branch. Athens, Georgia.
. 1980. Interim Guidelines and Specifications for Preparing Quality
Assurance project Plans. QAMS. EPA-600/4-83-004. NTIS PB83-170514.
. 1981. Work Plan Development Technical Methods for Investigating
Sites Containing Hazardous Substances. Technical Monograph No. 6.
. 1982. Draft Guidance For Preparation of Combined Work/Quality
Assurance Project Plans for Water Monitoring. November 15.
. 1984. Soil Sampling Quality Assurance User's Guide. EPA
~50"0~/4-84-043.
. 1984. Memorandum from Stanley Blacker, about QAMS Checklist for DQO
Review.
. 1984. Quality Assurance Management and Special Studies Staff,
Calculation of Precision, Bias, and Method Detection Limit for Chemical
and Physical Measurements. March 30.
. 1984. Soil Sampling Quality Assurance User's Guide. EMSL-LV
~6W/4-84-043. May.
. 1985. Guidance on Remedial Investigations Under CERCLA Hazardous
Waste Engineering Research Laboratory Office of Emergency and Remedial
Response and Office of Waste Programs Enforcement.
. 1985. Sediment Sampling Quality Assurance User's Guide.
Environmental Research Center Cooperative Agreement CR 810550.01
AW3E-2 5-8
-------
OSWER Directive 9355.0-7A
. 1985. Construction Quality Assurance for Hazardous Waste Land
"Disposal Facilities Public Comment Draft. EPA/530-SW-85-021
. 1985. Draft DQO Report for SI Superfund Process. March.
. 1985. Protection of Public Water Supplies from Groundwater
Contamination. EPA/625/4-85/016.
. 1985. RCRA. Groundwater Monitoring Technical Enforcement Guidance
Document. Office of Waste Programs Enforcement, Office of Solid Waste and
Emergency Response.
. 1985. Draft Superfund Public Health Evaluation Manual. OSWER
Directive 9285.4-1. December.
AW3E-2 5-9
-------
Section 6.0
-------
RESERVED
OSWER Directive 9355.0-7A
6.0 REMEDIAL DESIGN
AW3E-12
6-1
-------
Section 7.0
-------
OSWER Directive 9355.0-7A
7.0 REMEDIAL ACTION
RESERVED
AW3E-13 7-1
-------
Section 8.0
-------
8.0 STATISTICAL CONSIDERATIONS
Statistical techniques can be utilized to evaluate environmental data
and to assist in designing appropriate sampling plans based on the data.
Statistical techniques may be applied during RI/FS, RD, and RA activities,
and become more meaningful as additional data are obtained.
Statistical considerations come into play in Stages 1, 2, and 3 of the DQO
process. In Stage 1 the existing data are compiled and evaluated and
statistical techniques can be used to evaluate the comparability of
different sets of existing data and to evaluate the need to obtain
additional data. In Stage 2 data quality and quantity needs can be stated
in terms of confidence limits or within other statistical framework. After
Stage 3, statistics can be used to evaluate newly acquired data and to
assess uncertainty in various decisions.
This section provides discussions of various statistical approaches which
may be appropriate for remedial action programs. The discussions are based
upon hypothetical scenarios which have or might occur at hazardous waste
sites and links available statistical methods to potential applications.
The scenarios presented are not the only situations in which statistics can
be applied, but they provide an indication of the information that can be
obtained from statistical methods. Additional discussion of statistical
techniques are provided in the references cited at the end of this section.
8.1 DETERMINATION OF NUMBER OF SAMPLES
The technical objectives, budget, and schedule of the program must be
considered when determining the number of samples to be obtained. Numerous
factors must be considered in the evaluation in an iterative manner with
the types of analyses as well as the constituents to be evaluated being of
primary importance.
AW3E-4 8-1
-------
OSWER Directive 9355.0-7A
Statistical approaches can be used to determine the number of samples which
are required in order to generate data which have specified limits of
confidence. However, application of statistical formulas to determine the
number of samples requires that previously obtained information on the
environmental media under consideration be available for evaluation.
Basic t-test methods for determining the number of samples required to
achieve a specified precision and confidence level have been utilized with
varying degrees of success in the development of data quantity
requirements. These techniques provide an estimate of the number of
samples required to achieve a specified level of precision and confidence.
However, in order to apply these equations, the mean of the population and
the standard deviation of the population must be known. Therefore, these
equations cannot be applied unless data are available for calculating means
and standard deviations. Furthermore, t statistics are based on
assumptions that the data are normally distributed and independent.
The number of data points required to characterize a population to a
specified confidence level within a specified confidence interval is
related to the variability of the population. The number of samples
required to characterize populations which have relatively low variability
(e.g., waste streams resulting from a treatment process) that would, in
most cases, be expected to be less than those required to characterize a
population in which the variability is higher (e.g., contaminants present
in the soil) for the same level of confidence and similar confidence
intervals.
The number of data required to characterize the population is further
influenced by the distribution frequency of the data. The analysis of
samples which are normally distributed would be expected to fall under a
bell shaped curve as shown in Figure 8-1. The majority of measurements
would be expected to fall near the average measurements for normally
distributed populations. Measurement of environmental media at
uncontrolled hazardous waste sites do not generally fall within the
category of normally distributed data. However, if a ground water sample
obtained from a hazardous waste site was divided into a number of
AW3E-4 8-2
-------
*— ACCEPTANCE REGION
20.2
30.0
(a) Acceptance region for H: p - 30.0
39.8
FIGURE 8-1
NORMALLY DISTRIBUTED DATA
-------
OSWER Directive 9355.0-7A
subsamples and these subsamples were each handled and analyzed in a
controlled laboratory setting, then the resulting data would be expected to
be normally distributed about an average value.
Environmental media populations in some cases may be log-normally
distributed. In these situations, repetitive measurements of the media
would fall under a skewed curve. The repetitive measurement of the
permeability of soil samples would result in a log-normal distribution of
data, for example.
Comparison of ground water quality analytical results or soil permeability
values obtained at any one location with the results of other samples
obtained at separate locations may not, however, be represented as normal
or log-normal populations. More frequently, the measurements are dependent
on one another and may vary over space due to factors such as those which
would influence migration of contaminants in ground water, or natural soil
forming processes which would affect permeability.
in a simplistic sense, as contaminants migrate from a source, the
concentration of the contaminants would be expected to decrease as the
distance from the source increases. Figure 8-2 illustrates this concept in
two dimensions. The concentration of the contaminants at any point in this
system is related to other (dependent) factors such as flow gradient,
dispersion coefficients and attenuation/biodegradation rates. As stated,
this is a simplistic model, and caution should be used when applying it to
actual site conditions.
The analytical values obtained from samples over the area of contamination
would range from the maximum level encountered at the source to lowest
level at the furthest distance from the source. The plotting of these data
would not yield a normal distribution or log-normal distribution.
Therefore, statistical techniques which require data which are normally
distributed and independent would not be valid for estimating the number of
samples to be obtained. Geostatistical techniques which take into
consideration the spatial dependence of these data are, therefore, more
appropriate for use in these situations.
AW3E-4 8-4
-------
GROUND WATER
FLOW
SOURCE 1000 ppm
1ppm
FIGURE 8-2
CONTAMINANT CONCENTRATION/
DISTANCE RELATIONSHIP
-------
OSWER Directive 9355.0-7A
The information available for review during project scoping of an RI may
not be adequate for purposes of applying statistical formulas. In these
types of situations it may be necessary to gather preliminary data in order
to determine the appropriate number of samples to be obtained during the
RI. This additional information may be obtained using a phased RI approach
wherein the initial investigative phase provides the data necessary to
determine the number of samples which would be required during the RI to
provide statistically valid results.
8.2 DETERMINATION OF TOTAL UNCERTAINTY
The sources of variability influencing the total uncertainty can be broken
down into:
o Laboratory variability
o Sampling variability
o Variability of the contaminant plume
Laboratory variability is the most commonly discussed source of
variability. For this discussion, laboratory variability will include all
factors which cause a laboratory to report a contaminant concentration
which is different from the actual concentration. Sampling variability
includes all errors introduced by random infrequent events which can affect
one or all of the samples. Sources of sampling error include cross
contamination and incorrect sampling procedures. Variability of the
contaminant plume is caused by physical and chemical processes, such as
dispersion and adsorption, which cause the chemical concentration within a
plume to vary as a function of location in space. Because of this
variability, contaminant concentrations of water samples taken from the
same well over time may be different even if laboratory and sampling
variability are zero.
AW3E-4 8-6
-------
OSWER Directive 9355.0-7A
8.2.1 UNCERTAINTY ASSOCIATED WITH ONE SAMPLE ANALYSIS
Total uncertainty is a function of the actual contaminant concentration,
the analytical method, the sample matrix, the sampling procedures, and
contaminant variability in space and time. Ideally, information should be
gathered concerning all variables contributing to the total uncertainty;
however, in most instances, data will not be gathered in sufficient amounts
to perform such an analysis. The following discussion will address
examples for which total uncertainty might be of interest. The information
required and the techniques used to obtain total uncertainty will be
discussed for each case.
In this example a single water sample from a potentially contaminated well
is sent to a laboratory for analysis. The laboratory reports a contaminant
concentration for the sample. Of concern is the uncertainty associated
with the reported value since the uncertainty will indicate the risk
associated with the decision to shut down the well.
If only one sample is submitted to the laboratory, it will be impossible to
specify the sampling or contaminant plume variability since these values
are site specific. It will also be impossible to determine the site
specific laboratory uncertainty due to matrix effects; however, if
sufficient historical information concerning laboratory methods is
available, an estimate of the laboratory uncertainty is possible. Assuming
that no large errors are made during the sampling of the well, the
laboratory uncertainty will accurately describe the uncertainty associated
with the particular sample.
The uncertainty associated with a reported concentration value can be
determined if historical information necessary to develop the bivariate
distribution of actual and reported values is available. A bivariate
distribution specifies the probability that a pair of random variables will
take on a particular pair of realizations. In this application, the
AW3E-4 8-7
-------
OSWER Directive 9355.0-7A
bivariate distribution of interest is the distribution of actual and
reported values. The bivariate distribution of actual and reported values
will specify, for example, the probability that the lab will report 5 ppm
when the actual concentration is 6 ppm.
Given the bivariate distribution of actual and reported values, the
conditional distribution of actual values for a given reported value can be
determined. As an example of a conditional distribution, consider a case
where a spiked sample of known concentration is submitted to a lab several
times; the lab is likely to report a different concentration value for each
spike. In this case, the distribution of the reported values is the
conditional distribution of reported values for the particular actual
value. Any conditional distribution of interest can easily be determined
once the bivariate distribution is known. Specifically, the conditional
distribution of actual values for any reported value can be determined.
The mean of the conditional distribution is related to accuracy and the
variance is related to precision. By determining the conditional
distribution for a large number of possible reported values, the regression
of actual or reported values can be obtained. Once the regression function
is available, the information contained in the curve can be summarized in a
table. The table will identify, for a specific method, the best estimate
of the actual value for any reported value. A second table can be
developed which specifies the confidence limits for any particular reported
value. These confidence limits will vary as a function of the reported
value.
If only the accuracy and precision of the statistical method are known, an
estimate of the confidence limits can be obtained. This method will not be
as accurate at the previously described method but the results will give a
rough idea of the uncertainties. To apply this method the accuracy and
precision of the statistical method must be known and it must be assumed
(incorrectly) that accuracy and precision are independent of concentration.
Finally the distribution of analytical errors must be assumed to be normal.
Given these assumptions, confidence statements can be made concerning the
value observed at a particular location.
AW3E-4 8-8
-------
OSWER Directive 9355.0-7A
The DQO with regard to each sample is to obtain .data of known quality.
This concept is encompassed in the total uncertainty, U, associated in a
single lab measurement. Any single measurement varies as a function of the
accuracy and precision of the analytic test. Both of these measures vary
as a function of the true concentration in the sample and sample matrix, as
well as the other contaminants which make up the sample.
Even with these variations, chemists use two measures to describe the
performance of an analytic test, P, percentage recovery, and relative
standard deviation (RSD). A rough measure of U for a single sample is:
U= (P+ 2xRSD)
It is important to note that this uncertainty is a function of the analytic
test and no additional sampling will reduce the uncertainty values for
single samples below this level. However, through the careful choice of
QA/QC samples, adequate estimates of P and RSD can be obtained. P is the
most significant variable to determine whether a sample is below an action
level.
Consider the following example: A site has a true concentration of 120 ppm
and the recovery is 70 percent, well within the CLP acceptable window of
recoveries. Historically, the RSD for for this hypothetical compound has
been 10 percent.
U= (.7+ 2x.l)
U= 0.5 to 0.9
The action level is 110 ppm for this hypothetical compound. U describes
the likely range of reoported values on a percentage basis. Given U, the
95 percent confidence level interval for the reported values is
0.5 x 120 to 0.9 x 120
60 to 108
AW3E-4 8-9
-------
OSWER Directive 9355.0-7A
Thus, the confidence region misses the action leyel and misses the true
concentration.
The only way to solve this problem is to correct samples by percentage
recovery as follows:
U - [1 + (2 x 0.1)]/0.7)
= (1 + 0.286)
U = 0.714 to 1.286
In making confidence statements concerning a particular reported value, the
accuracy of the method must be considered. Figure 8-3 demonstrates the
effects of not accounting for accuracy as measured by percent recovery.
The curves in this figure give the probability that a sample which in
actuality contains contaminant concentrations in excess of the action level
will have reported concentrations above the action level if percent
recovery is not considered. Percent recovery is a numerical representation
of the inaccuracy of an analytical procedure. Percent recovery is
generally determined by adding a known quantity of an analyte (spike) to an
environmental sample. The concentration value reported by the laboratory
divided by the known concentration of the sample expressed in percent is
known as percent recovery.
Percent recovery is not a constant. Not only is percent recovery a
function of concentration, it is also a non-repeatable measurement. That
is, if several replicates of a sample are spiked and analyzed, several
different percent recovery values will be obtained. These percent recovery
values may vary considerably. Based on the uncertainty in the percent
recovery, it is not recommended that analytical results be systematically
corrected for percent recovery. Correction for percent recovery can
introduce signficant errors if performed indiscriminantly. Hence,
correction for percent recovery should only be performed after detailed
discussions with an analytical chemist. The chemist will determine if the
recovery problem is due to a chemical change or a failure to recover the
total pollutant present.
AW3E-4 8-10
-------
UJ >
H£
tt _J
£z
ui O
Jen
h Q
> UJ
I- UJ
5^
52
CD UJ
RECOVERY (%)
FIGURE 8-3
EXAMPLE PROBABILITY THAT REPORTED
VALUE EXCEEDS ACTION LEVEL VERSUS PERCENTAGE
RECOVERY PLOT (RSD = 30%)
-------
OSWER Directive 9355.0-7A
8.2.2 TOTAL UNCERTAINTY ESTIMATES WHEN MANY DATA ARE AVAILABLE
For sample matrices other than water, development of laboratory correction
tables may be impossible due to matrix effects. However, if a sufficient
number of samples are obtained, statistics can be used to directly
determine total uncertainty. Statistics examine the variability within a
set of data. If there are no systematic errors, the variation observed in
a set of data is the sum of laboratory, sampling, and contaminant plume
variation. Thus, any statistical determinations of uncertainty will be
measures of total uncertainty. If statistics will be used to measure total
uncertainty, the sampling plan must include at least 10 split samples to
measure laboratory and sampling uncertainty. At least 20 additional
samples are required to measure the spatial variability of the
contamination.
8.2.3 TOTAL UNCERTAINTY WHEN MANY DATA ARE AVAILABLE AND LABORATORY
UNCERTAINTY IS KNOWN
When laboratory uncertainty has been determined and many data are available
total uncertainty can be made. In this case, the estimate of total
uncertainty can be estimated by evaluating laboratory uncertainty and other
types of uncertainty. The laboratory uncertainty can be used to replace
each reported concentration by the conditional distribution of actual
values. Thus, each reported value can be transformed into a distribution
of values. Statistics can be applied to then utilize the distribution of
values at each sample location to estimate total uncertainty at any
particular unsampled point or over any region of the site. This type of
conditional estimate of total uncertainty can be used when the uncertainty
at a specific point is required.
8.3 PROBABILITY OF LOCATING A CONTAMINATED ZONE
At sites or portions of sites where soil contamination is suspected but no
definite sources have been identified, an objective of the remedial
investigation might be to determine if soil contamination is present.
AW3E-4 8-12
-------
OSWER Directive 9355.0-7A
Important decisions facing the site manager are how many samples must be
taken to investigate the potentially contaminated area and where the
samples will be located.
In certain given situations geophysical surveys can be utilized in
determining the location of contaminated zones. Geophysics can effectively
be used to determine the locations of certain ground water plumes (such as
hydrocarbon plumes) and concentrations of buried metalic objects (drums and
tanks).
The following discussion concerning the probability of locating a
contaminated zone is applicable to geophysical methods as wells as to
standard sampling technologies.
The decision maker must determine, in Stages 1 and 2 of the DQO process,
the acceptable probability of not finding an existing contaminated zone in
the suspected area. For instance, it might be determined that a 20 percent
chance of missing a 100 ft by 100 ft contaminated zone is acceptable but
only a 5 percent chance of missing a 200 ft by 200 ft zone is acceptable.
This probability value provides the basis for using statistics to determine
how many samples are required. Statistics will be used to determine the
number and location of data required to lower the probability of missing an
existing contaminated zone to a value less than the acceptable predefined
value.
The acceptable probability of missing a contaminated zone must be
established by the decision maker working in concert with the data users.
Individuals involved in developing risk assessments may provide meaningful
inputs into determination of the appropriate probability values to be
utilized.
The statistical method applied in this instance involves geometric
probabilities. That is, the probability of not identifying a contaminated
zone is related to the area or volume of the contaminated zone and the
spatial location of the samples. To apply this method, the following
assumptions are required:
AW3E-4 8-13
-------
OSWER Directive 9355.0-7A
o The shape and size of the contaminated zone must be known at least
approximately. This known shape will be termed the target.
o Any sample located within the contaminated zone will identify the
contamination.
These assumptions are not severe and should be met in practice.
If in addition to the above assumptions, data are located on a perfectly
regular grid and the target is circular, the probability of hitting the
target for a given grid size is shown by the following: (Gilbert, 1982).
Probability of a Hit G/ f
0.8 1.13
0.9 1.01
0.95 0.94
0.99 0.86
Where A- is the diameter of the target and G is the linear grid spacing.
If data are not regularly located or the target is not circular, a
simulation procedure is used.
The procedure used for the hit and miss analysis includes the following
steps:
o Simulate a contaminated zone or target.
o Randomly locate the target within the site.
o Determine if any sample locations fall within the boundaries of the
target. If so score a hit.
o The simulation is repeated several hundred times using a computer
program and the number of hits and misses are recorded.
The probability of locating the contaminated zone is equal to the total
number of hits divided by the total number of simulations.
AW3E-4 8-14
-------
OSWER Directive 9355.0-7A
Figure 8-4 illustrates a hit and miss approach for two simulated
contaminated zones. The method is flexible so various different sample
numbers and locations and various different target sizes can be quickly
examined. By varying the number of samples for a fixed target, the number
of samples required to lower the risk of missing the contamination to an
acceptable level can be determined. Thus, this method allows determination.
of both the number and location of samples necessary to satisfy DQOs.
8.4 CONFIDENCE LIMITS ON ESTIMATES OF MEAN CONTAMINATION
At sites where contamination is known to exist, a parameter of interest is
the mean contaminant concentration over the contaminated area. Mean
contaminant concentrations are important when evaluating contaminants
contained within a confined area such as a lagoon. In this case, the mean
contaminant concentration determines the total amount of contaminants
contained in the lagoon. To assess various remedial alternatives it is
important to know the maximum quantity of contaminants present in the
lagoon. Confidence limits can be used to state the probable range of total
contaminants contained in the lagoon.
Confidence limits are a means of placing uncertainty limits on the data.
Confidence limits can be used to replace qualitative statements such as
"the data at site A are more precise than the data at site B" and ambiguous
quantitative statements such as "the data are accurate to within ± 20
percent" by precise probabilistic statements of uncertainty such as "The
chance the 10 to 15 ppm interval includes the unknown mean is 95 percent."
In the previous statement, the values 10 and 15 ppm are the confidence
limits for the 95 percent confidence interval. As the number of data
increases, the confidence interval narrows and the confidence in any value
derived from the data increases. Thus, if the desired degree of certainty
is known at the beginning of the project, confidence limits can be used to
determine the number of data required.
The number of available data points will determine the method which can be
used to calculate confidence limits. If there are less than eight existing
data, confidence limits cannot be reliably determined. If there are
AW3E-4 8-15
-------
B
NOTE: This tlgura Illuatrataa two poaaftala simui«tlon« of i circular targat tor a flsad
Ml of dan locations. Tht uppar llgura (a) Illuatrata* a hit whlla tha lowar
fIgura (b) Hluatrataa a mm
FIGURE 8-4
HIT & MISS EXAMPLE
-------
OSWER Directive 9355.0-7A
between 8 and 30 existing data, confidence limits can be determined using a
simple non-parametric technique. If more than 30 data are available, more
sophisticated methods can be used.
The problem of estimating confidence limits about the true mean is commonly
encountered. It can be shown that when very specific conditions hold, a
simple function of the sampling parameters follows a Student's t distribu-
tion. When the t distribution is valid, the mean and variance of the data
are calculated and the t statistic for the confidence interval of interest
is obtained from a table. A formula which directly determines the number
of data is then applied.
Because of the simplicity of this formula, the t statistic is widely used.
The t statistic appears in several documents discussing the determination
of data requirements at hazardous waste sites (EPA, 1984; EPA, 1985). It
is extremely important to realize that the conditions necessary to apply
the t statistic are generally not met at hazardous waste sites. The
conditions which must be met to apply the t statistic are:
o The sample mean must follow a normal distribution.
o The sample variance must follow a Chi squared distribution.
o The sample mean and variance must be independent.
o The data must be independent.
o The data must be identically distributed.
These conditions do not hold at the great majority of hazardous waste
sites, so any confidence limits calculated based on the t distribution will
be very inaccurate. For this reason, the t distribution should not be used
to evaluate hazardous waste sites unless the aforementioned conditions are
met.
For small data sets, the distribution of the sample mean is unknown and
site dependent. Because the shape of the distribution is unknown,
estimation of confidence limits on the population mean require the
AW3E-4 8-17
-------
OSWER Directive 9355.0-7A
application of non-parametric or distribution free statistical methods.
Non-parametric techniques can be applied regardless of the shape of
underlying distribution. One well known non-parametric technique is
Chebyshev's inequality which states:
P[ X - u < ko-//h] > (1-1A2)
Where: X is, in this case, the sample mean
u is the population mean
o~ is the population variance
k is some value >1
n is the number of data
This inequality states that, regardless of the distribution of X, the total
probability lying in the tails of the distribution does not exceed 1/k2•
This method can be used to determine confidence limits on the population
mean or to determine the number of data required to reduce the confidence
interval to an acceptable value.
Chebyshev's inequality can be used to solve for the number of samples which
should be collected provided the variance of the data. If it is assumed
that the data are independent and a 95 percent confidence limit the
equation can be solved as follows:
P[ X - u < Z //n] > 0.95
For normally distributed data a Z value of 1.96 is obtained from
statistical tables for the 95 percent confidence limit. The normally
distributed data can be related to number of samples needed (n2) when no
distributional assumptions are made by equating the two inequalities as
follows:
(1-1/K2) = 0.95
K - 4.472
AW3E-4 8-18
-------
OSWER Directive 9355.0-7A
Foe normal distribution:
P[ X-u < 1.96 / n] > 0.95
For unknown distribution:
P[ X-u < 4.472 / n2] > 0.95
Solving these inequalities yields:
n~ => 5.206 n.
Therefore, approximately five times as many samples would be required if an
assumption of normally distributed data cannot be made.
Chebyshev's inequality can be applied in all cases, regardless of the par-
ticular distributional properties of an individual site; however, it is a
conservative method which will often predict that very large numbers of
data are required to reach a predefined level of precision. If more than
30 data are available, a more exact technique, based on geostatistics, can
be applied.
8.5 GEOSTKriSTICS
Geostatistics, or more formally, the theory of regionalized variables, is
similar to classical statistics in many ways. However, it differs with
respect to basic assumptions regarding mutual independence of data.
Classical statistics assumes that data are mutually independent, that is,
that one data point is not related to another. Geostatistics recognizes
that data are related by spatial proximity as governed by physical
processes; thus, geostatistics can be used to estimate mean values at
unsampled points. Information at one particular point in space can be used
to impart information concerning the contaminant level at a location 5 or
10 ft away from the sampled point. Because knowledge of the contaminant
level at one point in space provides information concerning the contaminant
level at a second point in space, the data are spatially correlated.
Geostatistical tools measure and exploit the correlation between data to
estimate contaminant concentrations and determine the uncertainty
associated with the estimate.
AW3E-4 8-19
-------
OSWER Directive 9355.0-7A
The geostatistical estimate of a mean value is optimally obtained using a
method known as kriging. Associated with this estimate is the kriging
variance, which is a measure of the uncertainty of the estimate.
Geostatistics can be used to determine the variance of errors associated
with any estimate. In particular, geostatistics can be used to determine
the variance of errors associated with estimating the true mean contaminant
concentration by the mean of the available data. The detailed derivation
of the error variance is given in Journal and Huijbregts (1978). An intui-
tive discussion of the method is provided here.
An estimate of the true mean site contamination can be determined from an
average of the available data. The estimate is not, in general, equal to
the true mean so an error is made. The error of estimation is defined as
the estimated mean less the true mean. The particular error observed is
one realization of the error random variable. The distribution of the
error variable is unknown, but it is known that the mean of the error
distribution is zero since only unbiased estimators will be used. The
variance of the error distribution can be determined using geostatistics.
The variance estimate requires knowledge of the average correlation between
the data and the average correlation between the data and the volume
defining the site. Determination of these quantities requires a model of
correlation at the site. This correlation model is provided by the
experimental variogram determined from the data. The experimental
variogram is defined as follows:
g(h) - 1 £(z(Xi + h) - z(Xi))2
n(h)
Where: n(h) is the number of data separated by distance h
z(x.) is the contamination observed at location x.
z(x.+h) is the contamination observed at location x.+h
g(h) is the experimental variogram for distance h
AW3E-4 8-20
-------
OSWER Directive 9355.0-7A
By varying h, a model of the variogram versus h can be developed and
applied to determine the variance of errors.
To this point, the mean and variance of the distribution of errors have
been discussed. The remaining parameter of interest is the shape of the
distribution of errors. As the number of data used to estimate the true
mean increases, the distribution of errors becomes more and more like a
normal distribution. This is not a theoretical result but a practical
observation from applications in the mining industry. Given that the
errors follow an approximately normal distribution, the confidence limits
can be determined by the following procedure.
o Define the level of confidence required.
o Find the standard normal variate corresponding to this probability
in a normal table.
o Apply the following formula:
Z-ys
-------
OSWER Directive 9355.0-7A
8.5.1 LOCAL ESTIMATION OF CONTAMINATION
In many instances, the contamination at a particular point within the site
is of interest. For example, consider a site with a known source of
contamination. Available information indicates that contaminants are
migrating toward the western edge of the site. An objective of the RI
might be to determine the western extent of contaminant migration.
Geostatisties can be used to determine the likely extent of contamination.
This information will greatly aid in choosing data locations.
A second example where local estimation is important is in determining
optimal contours for a variable. For instance, in many enforcement cases
an accurate determination of the ground water gradient is required to
correctly identify potentially responsible parties. Water levels are
measured in wells which are separated by varying distances. The heads
between wells are unknown. To create a contour map of head, estimates of
heads between wells must be obtained. To ensure that the estimated heads
and associated contour lines are as accurate as possible the heads at
unsampled points should be estimated optimally using geostatistics.
Geostatistics can be used to address problems presented in the previous
scenarios. The geostatistical technique which will be applied is known as
kriging. Kriging, which is similar to multiple regression, determines an
optimal estimate of a variable at any particular location in space.
Associated with this estimate is an associated qualitative measure of
uncertainty known as the kriging variance.
To apply kriging, a model of the correlation between data is required.
This model is obtained by modeling the experimental variogram of the data.
An adequate correlation model cannot be developed if less than 30 data are
available so the use of kriging should be restricted to sites where more
than 30 data are available or will be obtained.
An example of the use of kriging to optimally estimate the concentration of
lead in soil surrounding a smelter is shown in Figure 8-5.
AW3E-4 8-22
-------
N
1000
(ft.)
750
500
250
250
500
750
1000(«.)
Not*: Contour map of lead concantration in soil surrounding a smaltar. Contours
ar* based on estimates of soil lead conctntralion (in ppm) determined by
kriging.
•
FIGURE 8-5
EXAMPLE OF KRIGING
-------
OSWER Directive 9355.0-7A
8.5.2 LOCAL ESTIMATION OF PROBABILITY
At soil contamination sites where a fixed cleanup criterion has been set,
geostatistics can be used to estimate the risk associated with not removing
any particular quantity of soil. Geostatistics can be used to quantify the
probability of exceeding this criteria and to develop probability contour
maps. This map can be used in conjunction with the acceptable uncertainty
determined during Stages 1 and 2 of the DQO process to define what volume
of soil must be removed.
To determine the probability of exceeding a given value at an unsampled
point it is necessary to estimate the entire contaminant distribution at
that point. Given this distribution, the probability that the contaminant
concentration exceeds any value of interest can be determined.
An example of a probability map is provided in Figure 8-6. In this
example, lead contamination has been found in soil surrounding a lead
smelter. It has been determined that all soil in excess of 1000 ppm will
be removed as part of the remedial action. The probability map gives the
likelihood of exceeding 1000 ppm, at each point in the site. If through
the DQO process 30 percent had been determined, as an acceptable
probability of exceeding 1000 ppm then all soil within the 30 percent
contour would be removed. The remaining soil would have, at most, a 30
percent chance of exceeding 1000 ppm. If a different acceptable
probability was defined, the volume of soil removed would be defined by the
particular contour. This method provides an objective method for
determining the volume of soil to be removed.
Techniques for estimating local probability distributions include indicator
kriging, probability kriging, and multivariate gaussian kriging. (Journal
1983; COM 1986; Verly 1983; and Isaakes 1983). These techniques are known
as non-linear estimators and are related to but are more complex than
kriging. Application of these techniques requires approximately 50 data.
This number of data is required because non-linear estimators require an
accurate and detailed model of the correlation structure of the data.
AW3E-4 8-24
-------
_ I 1 t I I I I I I I II I I I I I I.
NOTE: Probabilities of exceeding 1000 ppm soil lead concentration near a smelter.
Material with probabilities exceeding the acceptable risk defined in the DQO
process will be removed as part of the remedial action.
FIGURE 8-6
PROBABILITY MAP
-------
OSWER Directive 9355.0-7A
An important feature of non-linear estimators is that any uncertainty in
the data values, stemming from laboratory or sampling errors can easily be
incorporated into the estimate. Since non-linear estimators can be used to
estimate the mean or variance at a point or over a region, these techniques
provide a means of including uncertainty in any regional or local estimate
of the mean. The uncertainty associated with these estimates will include
the uncertainty present in the data.
8.6 REFERENCES
Addiscott, T.M., and R.J. Wagenet. 1985. A Simple Method for Combining Soil
Properties that Show Variability. Soil Science Society of America
Journal. 49: 1365-1369.
Box, G.E.P., W. Hunter, and J.S. Hunter. 1978. Statistics for
Experimenters: An Introduction to Design, Data Analysis, and Model
Building. John Wiley and Sons. New York.
Camp Dresser & McKee Inc. (COM). 1986. Probability of Locating
Contaminated Zones at the North Cavalcade Site, COM Internal
Correspondence, J. Sullivan.
EPA. 1984. A Soil Sampling Quality Assurance User's Guide EPA 600/4-84-043
EPA. 1985. Sediment Sampling Quality Assurance User's Guide EPA
600/4-85-048
Flatman, G.T. 1985 Design of Soil Sampling Program: Statistical
Considerations Draft.
Flatman, G.T. and A.A. Yfantis. 1984. Geostatistical Strategy for Soil
Sampling: The Survey and the Census. Environmental Monitoring and
Assessment 4:335-349.
Gilbert, 1982. Some Statistical Aspects of Finding Hot Spots and Buried
Radioactivity. TRAN-STAT Statistics for Environmental Studies, Batelle
Institute, Richland Washington, No. 19.
Isaaks, E., 1984. Risk Qualified Mappings for Hazardous Waste Sites: A
case study in distribution free geostatisties, unpublished master's
thesis, Stanford University.
Journel, A.G., 1983. Non Parametric Estimation of Spatial Distributions,
Jurnal of Mathematical Geology, Vol. 15, No. 3, pp. 445-468.
Journel, A.G. and Ch.J. Huijbregts.1978. Mining Geostatistics. Academic
Press, London.
Klusman, R.W. 1985. Sample Design and Analysis for Regional Geochemical
Studies. Journal of Environmental Quality. 14:369-375
AW3E-4 8-26
-------
OSWER Directive 9355.0-7A
Ripley, B. 1982. Spatial Statistics. John Wiley & Sons, New York.
Russo, D. 1984. Design of an Optimal Sampling Network for Estimating the
Variogram. Soil Science Society of America Journal 48 (4): 708-716
Sullivan, J. 1984. Conditional Recovery Estimation through Probability
Kriging - Theory and Practice, in Geostatistics for Natural Resource
Characterization, Reidel, Dordrecht, Holland.
Verly, G. 1983. The Multigaussian Approach and its Application to the
Estimation of Local Reserves, Journal of Mathematical Geology, Vol. 15,
No. 2, pp. 263-290.
Yost, R.S., G. Uehara and R.L. Fox. 1982. Geostatistical Analysis of Soil
Chemical Properties of Large Land Areas. I. Semi-variograms. Soil Science
Society of America Journal 46(5):1028-1032
Yost, R.S., G. Uehara and R.L. Fox. 1982. Geostatistical Analysis of Soil
Chemical Properties of Large Land Areas. II. Kriging. Soil Science
Society of America Journal 46(5):1033-1037
Zirschky, J., Deary, G.P., Gilbert, R.O., Middlebrooks, E.J. 1985 Spatial
Estimation of Hazardous Waste Site Data. In: Journal of Environmental
Engineering. Vol.111, No.6 pp.777-787.
AW3E-4 8-27
-------
Section 9.0
-------
OSWER Directive 9355.0-7A
9.0 ANALYTICAL CONSIDERATIONS
Analytical methods must be evaluated during the development of site
specific data quality objectives. The specific parameters for which the
analytical method is valid, its limitations, and any special considerations
(such as sample preparation) which will affect the resulting data quality
must be understood in order to select appropriate analytical methods for
specific uses.
This section provides an overview of the analytical considerations which
should be taken into account during DQO development. Analytical
considerations must be evaluated concurrently with statistical and sampling
considerations in order to ensure that a consistent approach is proposed
and that the data quality objective ultimately established can be attained.
9.1 ANALYTICAL SUPPORT LEVELS
The analytical options available to support remedial investigation/
feasibility study activities are presented in five general levels. These
levels are distinguished by the types of technology and documentation used,
and their degree of sophistication as follows:
o LEVEL V - Non-standard methods. Analyses which may require method
modification and/or development.
o LEVEL IV - CLP Routine Analytical Services (RAS). This level is
characterized by rigorous QA/QC protocols and documentation and
provides qualitative and quantitative analytical data. Some regions
have obtained similar support via their own regional laboratories,
university laboratories, or other commercial laboratories.
o LEVEL III - Laboratory analysis using methods other than the CLP
RAS.This level is used primarily in support of engineering studies
using standard EPA approved procedures.
o LEVEL II - Field analysis. This level is characterized by the use
of portable analytical instruments which can be used on-site, or in
mobile laboratories stationed near a site (close-support labs).
Depending upon the types of contaminants, sample matrix, and
personnel skills, qualitative and quantitative data can be obtained.
AW3E-3 9-1
-------
OSWER Directive 9355.0-7A
o LEVEL I - Field screening. This level is characterized by the use
of portable instruments which can provide real-time data to assist
in the optimization of sampling point locations and for health and
safety support. Data can be generated regarding the presence or
absence of certain contaminants (especially volatiles) at sampling
locations. Essentially nonqualitative; quantitative only for total
organics.
Table 9-1 provides a summary of the analytical levels, their applicability,
and limitations. Within each level, different procedures may be used to
produce different quality data to some extent. For example, Level II
encompasses both mobile laboratory procedures and less sophisticated
"tailgate" operations which may produce data of different quality.
9.1.1 LEVEL V ANALYTICAL SUPPORT - NON-STANDARD METHODS
The objective of non-standard analytical support is to provide the RI/FS
process with data that cannot be obtained through standard avenues of
analytical support. Analytical support of this type may involve the
research, development and documentation of a method, or more typically, the
modification of an existing method. EMSL, Las Vegas can be consulted for
protocol availability, modification, or development. Level NS methods are
available through CLP Special Analytical Services (SAS), university
laboratories, commercial laboratories, National Enforcement Investigation
Center, and Environmental Services Division. The types of analyses
available through Level NS support may ultimately be technology-limited.
Level V includes the modification of existing methods for lower detection
limits or for verification of tentatively identified compounds (TIC). TICs
are defined as the 30 non-HSL peaks of greatest apparent concentration
under the CLP RAS procedures. A more detailed discussion of TICs is
presented in Section 9.2.6. Level V support is used when "off the shelf"
procedures listed in other levels will not provide the needed data, when
analytical standard operating procedures (SOPs) are not available, or
non-standard techniques are required.
AW3E-3 9-2
-------
TABLE 9-1 SUMMARY OF ANALYTICAL LEVELS FOR RI/FS
Document No. 9555.0-7A
Option
Level V
Leve) IV
Level III
Level II
Level 1
Type of Analysis
- Non-conventlal
parameters
- Method-specific
detection Holts
- Modification of
existing methods
- Appendix 8 parameters
- TIC
- HSL Organ Ics/ Inorganics
by GC/MS; AA; ICP.
- Organlcs/lnorganlcs
using EPA procedures
other than RAS can be
analyte-speclf Ic
- RCRA characteristic
tests
- Variety of organ Ics by
GC; Inorganics by AA;
XRF
- Tentative ID; analyte-
speclf Ic
- Detection Units vary
from low ppm to to* ppb
- Portable/nob) le
1 nstrumentat Ion
- Total organic vapor
detection using
portable Instruments
Uses
- Conf Irnatlonal
- Toxicology
- Site-specific
conditions/parameters
- RCRA compliance
- Confirmations!
- Toxicology
- A 1 1 other program
Activities
- Conf Irmatlonal but Kith
less documentation
- Presence or absence of
contaminants
- Engineering uses
- Screening
- Presence or absence of
contaminants
- Relative concentrations
- Engineering
- Screening
- Assist In Identifying
sample locations
- Field screening
Limitations
- Requires method
deve lopment/nod 1 f 1 ca-
tlon
- Mechanism to obtain
services requires
special leodtlme
- Calibration standards
may not be read 1 ly
aval lable
- Tentative Identifica-
tion of non-HSL
parameters
- Some time Is Required
for Validation of
packages
- Methods may vary
- Tentative 10
- Techniques/Instruments
limited
- Instruments respond to
naturally-occurring
compounds
Data Quality Cost Time
- Method-specific - Initially high. If - Entries refer to
method development all types of
Is required. analysis listed.
No specific time/
cost requirements
can be specified.
In general the
time frame can
range from a few
weeks to signifi-
cantly longer If
method development
Is needed.
- Rigorous QA/QC - Jl.OOO/Sample for -Contractually.
- Standard Methods organ Ics 30-40 days
- »200/Sample for -Shorter turnaround
™tals time possible
through SAS
request
- Similar detection - S960/Sample for - 14 days
limits to CLP organ Ics
- Less rigorous QA/QC - 1200/Sample for
metals
- Dependent on QA/QC - StS-40/Sample - Real-time to
steps employed several hours
- Data typically reported
In concentration ranges
- If Instruments call- - Negligible, If - Real-time
brated and data capital costs
Interpreted correctly, excluded
pH, conductivity,
salinity, 00
- Health and safety
can provide Indication
of contamination
-------
OSWER Directive 9355.0-7A
The analysis of samples for the RCRA modified Appendix VIII list of
contaminants could currently be considered a Level NS application. The
modified Appendix VIII list contains 92 organic compounds that are not a
part of the Hazardous Substances List (HSL) and therefore are not normally
tested for, samples obtained from CERCLA sites. Appendix D of this
document contains a preliminary feasibility study performed to address the
applicability of using or modifying existing analytical procedures for
Appendix VIII analysis.
Level V poses limitations to implementation because the amount of lead-
time for start-up may be significant, and the analyses may be "one-of-a-
kind" applications of the method, resulting in a lack of comparability of
the data. The unit costs for Level V sample analysis are dependent on the
analysis requested. Generally, initial unit costs will be high, reflecting
the costs of becoming familiar with the method. If the method is used for
other projects or sites, unit costs may decrease with the demand, and the
method may become standard. The amount of documentation available for
Level V analytical support will vary depending on the sophistication of the
technology used. If method development is required, this information
should be requested and reviewed by the user.
Accuracy and precision information is generally not available for Level NS
due to the type of analytical support it provides. Specific data quality
information can best be obtained on a case-by-case basis by reviewing the
laboratory's method development work prior to submitting samples. If the
specific support required represents a modification of an existing
procedure or protocol, data quality information can be extrapolated from
the existing procedure.
9.1.2 LEVEL IV ANALYTICAL SUPPORT - CONTRACT LABORATORY PROGRAM (CLP)
ROUTINE ANALYTICAL SERVICES (RAS)
A high level of quality assurance and documentation has been incorporated
in all aspects of program activities. The CLP RAS provides for analyses of
all types of media for Hazardous Substance List (HSL) organic compounds and
priority pollutant inorganic compounds. These services are available
AW3E-3 9-4
-------
OSWER Directive 9355.0-7A
through CLP RAS and regional EPA ESD laboratories. Level IV analyses are
currently used for most RI/FS activities. However, the use of Level IV
data may not be required for many RI/FS purposes. Level IV analyses are
typically used for confirmation of lower level data, risk assessment, and
to obtain highly documented data.
CLP RAS generated data have the following properties:
o Confirmed identification and quantitation of compounds (for HSL
parameters only unless otherwise specified) to the detection
specified in the IFB.
o Tentative identification of a contractually-specified number (30) of
non-HSL parameters.
o Sufficient documentation to allow qualified personnel to review and
evaluate data quality.
o Uniform methods of analysis activities.
o Detection limits may not be sufficient for toxicological evaluations
o CLP support is one of the most expensive routine analytical services
available to the Superfund program, (e.g., RAS for organics is about
$l,000/sample. RAS for inorganic is about $200/sample).
o RAS is contractually operating on a 30-40 day turnaround although
delays can occur. Since demands fluctuate, space may be limited at
times for the Superfund program. In addition, data validation
usually takes 3-4 weeks after data is received.
The CLP RAS is very specific concerning the amount of laboratory
documentation that is supplied with every data package. The RAS
deliverables package contains information on initial and continuing
calibration, GC/MS tuning, surrogate percent recovery, and matrix spike
duplicates. In addition, hard copies are provided of reconstruction ion
chromatograms, GC chromatograms, and spectra for every sample and every
blank, standard, or spike run with a particular set of samples.
Documentation is also provided for blank analyses, internal chain of
custody and holding times.
AW3E-3 9-5
-------
OSWER Directive 9355.0-7A
The bias and precision of CLP analytical procedures can be assessed by
examining the performance of the laboratory in analyzing matrix spikes.
However, an indication of the performance of the laboratory is also
provided by the results of quarterly laboratory performance evaluation
samples. These evaluation samples are submitted blind so the laboratory
has no indication of the actual contaminant value. In contrast, the
laboratory knows the exact concentration of a matrix spike.
Historical CLP precision and accuracy data classified by media is presented
in Appendix C as Level IV. Each table is footnoted to show the source of
the precision and accuracy data and, to the extent possible, the type of QC
samples used, the numbers of data points, etc. Contract required detection
limits are presented in Appendix E.
9.1.3 LEVEL III ANALYTICAL SUPPORT - LABORATORY ANALYSIS
Level III analytical support is designed to provide laboratory analysis
using standard EPA approved procedures other than current CLP RAS. This
level is used to obtain similar analysis with less documentation.
Generally the analyses performed using Level III techniques are designed to
provide confirmed identification and quantification of organic and
inorganic compounds in water, sediment, and soil samples. These analyses
are available through commercial laboratories, ESD, CLP SAS, and the CLP
screening service (in development).
Level III provides data for site characterizations, environmental
monitoring, confirmation of field data and to support engineering studies
(e.g., design, modeling, and pilot/bench studies).
Level III laboratory analysis provides the following:
o Data to support engineering design parameters
o Data for use in evaluating the site for further action, e.g., to
determine extent of environmental contamination
o Data for use in risk assessments
AW3E-3 9-6
-------
OSWER Directive 9355.0-7A
o Rapid turnaround of data may be available
o Detection limits for presence or absence of compounds comparable to
Level IV
o Costs range from about $200/sample for inorganics to $960/samples,
for organics analysis. Turnaround time for Level III laboratory
analysis for organics is expected to be about 14-21 days.
Level III protocols all have built-in QA/QC, including calibration runs,
surrogate standards, etc. External QA, which is also used for the CLP, is
employed in the form of trip blanks, replicate and duplicate samples, and
blind spikes submitted with the samples.
The CLP screening service, which is currently under development, may be
more appropriate for hazardous waste analysis than the SW-846 methods, Test
Methods for Evaluating Solid Waste Physical/Chemical Methods, Second
Edition, (EPA 1982). This service would utilize CLP HAS methods with the
exception of the extensive documentation currently provided.
The type of laboratory support available under Level III ranges in
sophistication from GC/MS instrumentation to the measurement of water
quality parameters. The type and amount of documentation available depends
on the type of analysis requested. Data users should review a sample
report issued by the laboratory for the analysis requested to determine if
the degree of documentation supplied is adequate or whether additional
information must be requested. If the documentation is sufficient, Level
III could save time and cost.
Accuracy, precision and MDL information that is considered representative
of this level of analytical support was compiled from SW-846. This
information is provided in Appendix C. These procedures are applicable for
all sample matrices; however, the SW 846 information presented in Table
C-l-C was derived from the analysis of water and wastewater samples and
performance evaluation standards. Therefore, the criteria specified in
this table should be considered as "best case" information when non-aqueous
media samples are analyzed. Also, these data are presented irrespective of
the sample pretreatment or preconcentration techniques used. These
AW3E-3 9-7
-------
OSWER Directive 9355.0-7A
techniques may include liquid-liquid extraction (3520), acid/base-neutral
clean-up extraction (3530), soxhlet extraction (3540), sonication
extraction (3550), headspace (5020), and purge and trap (5030). They are
used in conjunction with the analytical procedures presented in SW 846.
9.1.4 LEVEL II ANALYTICAL SUPPORT - FIELD ANALYSIS
Level II analytical support is designed to provide real-time data for
ongoing field activities or when initial data will provide the basis for
seeking laboratory analytical support. There have also been a significant
number of instances where data derived from field analytical techniques
have been the sole basis for making decisions about site disposition or
health and safety. Level II analysis can also be utilized effectively when
a phased approach is used for field sampling. In a phased sampling effort,
the results of the first phase guide the development of subsequent phases,
and thus, real-time data are important.
Field analysis involves the use of portable or transportable instruments
which are based at or near a sampling site. Field analysis should not be
confused with the process of obtaining total organic readings using
portable meters.
Field analysis can provide data from the analysis of air, soil and water
samples for many Hazardous Substance List (HSL) organic compounds,
including volatiles, base neutral acid (BNA) extractable organics, and
pesticides/PCBs. Inorganic analysis can also be conducted using portable
atomic adsorption (AA) or other instruments.
The simplest type of field analysis is for volatile organic compounds.
Since the headspace analytical technique is used, the sample preparation is
minimal. Extractable organic and inorganic analyses require additional
time and equipment.
Level II analysis is used for onsite, real-time screening, baseline data
development, extent of contamination and onsite remedial activities.
AW3E-3 9-8
-------
OSWER Directive 9355.0-7A
Field analytical techniques provide the following:
o Rapidly available data for a variety of activities, including
hydrogeologic investigations (establish depth/concentration profiles
as wells are installed); cleanup operations (determine extent of
contaminated soil excavation); and health and safety (determine
nature and extent of release to ambient air).
o Detection limits for volatiles range from 0.5 ppb in air, 2-3 ppb in
water, and 10 ppb for soil. Detection limits for PCBs in soil are
about 1.0 ppm. Detection limits for extractable organic compounds
analyzed in mobile labs are in the vicinity of 10 ppb.
o Special applications - e.g., vadose zone monitoring.
o Volatile organic data can be used as early indicators or tracers of
off-site contaminant migration. Volatiles are the most mobile of
contaminants in all media, and are typically found at some
concentration at virtually all sites.
The ability to assess data quality for field activities is dependent upon
the QA/QC steps taken in the process (e.g., documentation of blank
injections, calibration standard runs, runs of qualitative standards
between samples, etc.).
If capital expenditures are excluded, the costs of field analysis are in
terms of personnel time in performing analyses, preparation/maintenance of
equipment, etc. Per sample costs for mobilizing and staffing a field
laboratory will decrease as the number of samples increases. Based on
limited data from Region I FIT experience, per-sample costs for volatile
and inorganic analyses are approximately $15. Per-sample costs for mobile
laboratory analyses may approach $100. Depending on the type of analysis,
time requirements per analysis range from 10 minutes to 1-2 hours.
Since Level II analyses are performed in the field, the amount and type of
documentation available will vary with the type of analysis and the
standard operating procedures used. Typically, a gas chromatograph
operated in the field provides the bulk of the analytical support at this
AW3E-3 9-9
-------
OSWER Directive 9355.0-7A
level. The documentation available utilizing this level of analytical
support would consist of the output of the strip chart recorder for all
samples/ standards/ and blanks analyzed. Field and analysis log books
would also be a source of additional documentation.
Data generated by Level II analysis are typically confirmed by submitting
some duplicate samples to CLP and/or a local laboratory. No statistical
methods are available to determine the exact number of samples to submit
for confirmation. Some guidance can however be provided.
Numerous factors have to be considered in choosing the number (or subset)
of samples to be submitted for confirmational purposes including:
o Total number of samples taken (i.e., when only a few samples are
taken, 100% confirmational analyses may be appropriate)
o Objective of sampling
o Data uses
o Method of analyses used
In general confirmational samples should include a subset (or all) of
designated critical samples, a subset of samples covering the entire range
of identified concentrations, and a subset of samples near the
(preliminary) action level and near the "0" concentration or not detectable
range.
An additional factor to consider is the measured precision of the field
instrument in use. When high precisions are measured less samples need to
be confirmed; if however a low precision is calculated, it is recommended
that analysis be suspended until the reason for the low precision is
determined. A qualified chemist should be contacted for input on
instrument calibration, and the utility of the analysis method with the
specific field conditions.
AW3E-3 9-10
-------
OSWER Directive 9355.0-7A
The data base for documenting accuracy, precision and MDL information for
Level II analyses is sparse. A number of factors have recently stimulated
an interest in the development of Level II methods. This activity is
centered primarily in various EPA Environmental Service Divisions (ESD) and
remedial contractors. There are two ongoing projects expected to
contribute significantly to the Level II data quality criteria data base.
These projects are an EPA Headquarters-directed compilation of all Level II
analytical methods currently used by Field Investigation Teams (FlTs) and
the operation of a mobile field analytical laboratory being directed by
EPA/ESD in Region IV. The Region IV project, in particular, holds the
promise of a significant contribution, since virtually all organic
Hazardous Substance List (HSL) parameters are being analyzed for. As these
data become available they will be incorporated into this document.
Examples of the performance Level II types of analysis are summarized
below.
Analysis of Polychlorinated Biphenyls (PCBs) in Soils and Sediments - The
EPA Region I Laboratory has developed a rapid method for the determination
of PCBs in soils and sediments that can be used in the field or in a
close-support laboratory setup. The procedure makes use of a simple
micro-extraction technique, with subsequent analysis by gas chromatography
using an electron capture detector (GC-ECD). Samples can be extracted in
less than 10 minutes and the GC analysis takes approximately 15 minutes.
By extracting the next samples to be analyzed while the previous sample is
being run on the GC, appoximately 25 samples can be run in a single 8-hour
day.
Percentage recovery of PCB Aroclor 1242 using this procedure ranges from
80-150 percent according to Region I estimates. It is theorized, but not
documented, that the extraction efficiency of this method decreases for the
heavier molecular weight aroclors. It is estimated that the recovery of
PCB Aroclor 1260 is approximately 50 percent. As such, the resulting data
has to be reviewed keeping these varying accuracies in mind.
AW3E-3 9-11
-------
OSWER Directive 9355.0-7A
Table 9-2 shows the results obtained by Region I using this procedure to
analyze a sediment reference material prepared by the Environmental
Monitoring Support Laboratory (EMSL) in Cincinnati, Ohio (Spittler 1986).
This procedure is referenced at the end of this section and should be
consulted for specific information regarding its use. It should be noted
that this procedure may be applicable to the analysis of other halogenated
compounds, such as chlorinated pesticides and pentachlorophenol. As is the
case with all Level II analyses, the procedure must be developed for
individual applications (sample matrix and analytes of interest) prior to
the actual analysis of investigative samples.
Lead Analysis of Soil Samples Using a Portable X-Ray Fluorescence Analyzer
Preliminary data have been obtained regarding the use of a portable X-ray
Fluorescence Analyzer (XRF) to analyze approximately 200 soil samples for
lead (COM 1986).
Two calibration curves were required for this analysis due to the wide
range of lead values encountered in the samples. A low curve covering the
0-1000 mgAg range, and a high curve covering the 1,000 to 17,900 mgAg
range were prepared. Sample preparation consisted of grinding the dried
sample with a mortar and pestle to less than approximately 100 mesh.
.Sample preparation averaged between 10 and 15 minutes per sample.
Initial indications are that sample preparation is critical in the use of
this procedure. On samples that had been ground to a homogeneous mixture,
acceptable precision and accuracy was obtained around a central value of
1000 mg/kg. However, the accuracy decreased near the detection limit and
at high concentrations.
Table 9-3 displays data obtained by the analysis of split samples by the
XRF procedure and CLP analysis.
AW3E-3 9-12
-------
OSWER Directive 9355.0-7A
TABLE 9-2
COMPARISON OF PCB SCREENING RESULTS
FOR AROCLOR 1242 WITH SRM
TRUE VALUE
Average Concentration
95% Confidence Interval
Standard Deviation
Cincinnati
True Value(ug/gm)
24.6
MDL - 51.6
11.1
Screening
Results (ug/gm)
22
14-35
8.4
Data were generated by three alternate methodologies
Source: EPA, 1984. Internal Memorandum from M. Lataille (Region I).
Provided by EMSL - Cincinnati. The acceptable range of values for this
QC sample is from the detection limit to 51.6 ug/gm
AW3F-12
-------
OSWER Directive 9355.0-7A
TABLE 9-3
X-RAY FLOURESCENCE ANALYZER
RAW SAMPLE SPLIT DATA
Lead Concentration (mg/kg)
Location
1
2
3
4
5
6
7
8
9
CLP RAS
492
5,600
94
18,200
3,770
718
90
9
1,800
XRF
517
9,450
347
21,694
7,890
879
208
298
5,540
RPD (%)
5.0
51.2
114.7
17.5
70.7
20.2
79.2
188.3
101.9
AW3F-13
-------
OSWER Directive 9355.0-7A
9.1.5 LEVEL I ANALYTICAL SUPPORT - FIELD SCREENING
The objective of Level I analysis is to generate data which are generally
used in refining sampling plans and determining the extent of
contamination. This information supplements background data and visual
evidence of contamination pathways. A second objective is to conserve
other analytical support resources. This type of support also provides
real time data for health and safety purposes. Additional data which can
effectively be obtained by Level I analyses include, pH, conductivity,
temperature, salinity, and dissolved oxygen.
Level I analyses are generally effective for total vapor readings using
portable photoionization or flame ionization meters which respond to a
variety of volatile inorganic and organic compounds. Detection is limited
to volatiles which have characteristics enabling them to be measured by the
respective instruments. These analyses are available through ESD or
remedial contractors.
Level I analysis provides data for onsite, real-time total vapor
measurement, evaluation of existing conditions, samples location
optimization, extent of contamination, and health and safety evaluations.
Data generated from Level I support are considered qualitative in nature.
Data generated from this type of analysis provide the following:
o Identification of soil, water, air and waste locations which have a
high likelihood of showing contamination through subsequent
analysis.
o Real-time data to be used for health and safety consideration during
site reconnaissance and subsequent intrusive activities.
o Quantitative data relative to a primary calibration standard if the
contaminant(s) being measured are unknown.
o Quantitative data if a contaminant is known and the instrument is
calibrated to that substance.
o Presence or absence of contamination.
AW3E-3 9-15
-------
OSWER Directive 9355.0-7A
Some instruments show a response to naturally occurring, non-hazardous
substances (methane) or other possible interferences. Data from
instruments may also be affected by weather and operator skill and
interpretive ability.
A hardcopy strip chart recorder output can be obtained for instrumentation
operated in the general total vapor survey mode but it is not common
practice. The most available form of documentation for this support level
is the field operator log book. Sample identification, location,
instrument reading, calibration and blank information is usually contained
in the field log book.
There are no data quality criteria specified for Level I, Field Screening
Support, because this level is characterized by the use of non-qualitative,
hand-held instrumentation (PID, FID). This instrumentation measures total
organic vapor concentrations only, and as such, is not conducive to the
generation of quantitative data. Specific information regarding individual
compound sensitivities and response factors can be obtained in the
manufacturer's owner's manual for specific instruments.
9.2 ANALYTICAL FACTORS
Other factors which may affect development of DQOs include the following:
o Analytical quality control
o Instrumentation options
o Media variability
o Method detection limit
o Matrix effects
o Tentatively identified organic compounds
o Data qualifiers
AW3E-3 9-16
-------
OSWER Directive 9355.0-7A
9.2.1 ANALYTICAL QUALITY CONTROL
The classification of analytical support into broad levels takes into
account internal laboratory quality assurance/quality control (QA/QC) in a
general manner only. Internal QA/QC refers to the surrogate and matrix
spikes, method blanks, and duplicate/replicate runs, among other laboratory
or field operation quality control. Within a given level of analytical
support, there may be differences in the way individual laboratories or
field operations approach internal QA/QC. For CLP Invitation for Bid (IFB)
RAS analytical support, the procedures are standardized and
contract-specified.
The methods and QA/QC for laboratories or field operations should be based
on the types of data needed for decision making. Documentation
requirements and appropriate QC requirements should also be required.
Performance of laboratories should be considered before selecting a lab.
Laboratory operation QA/QC plans should include calibration procedures,
frequency of laboratory blank and duplicate analysis, the use of surrogate
standards and spikes, and standard operating procedures. The
laboratory/field operation report format should also be evaluated in terms
of what information is reported along with the sample1 data (method blanks,
duplicates, spikes, etc.). At a minimum, method blank, internal
duplicate/replicate and matrix spike information should be reported along
with the sample data. Surrogate spike information should also be reported
for all GC/MS data.
When evaluating laboratory QA/QC, it is important for the reviewer to keep
the level of analytical support in perspective. These levels produce data
of different quality and documentation and should be reviewed with this in
mind. For example, it would be inappropriate to hold a screening
laboratory to CLP RAS standards, or expect a field screening operation to
have as rigorous QA/QC as a laboratory. Expectations such as these would
be inconsistent with the concept of classifying analytical support by the
quality of the data needed.
AW3E-3 9-17
-------
OSWER Directive 9355.0-7A
Once the required data quality for a given activity is established, the
data user must select the appropriate level of analytical support that will
supply data of the required quality. For example, an analytical level can
be flexible by specifying more or less Qft/QC- The cost and turnaround time
can be increased or decreased within a given level by adjusting the amount
of QA/QC. This reasoning results in a continuum of analytical support
services available to cover a wide spectrum of data quality requirements.
9.2.2 INSTRUMENTATION OPTIONS
In some cases, the decision maker may have the option of choosing between
similar analytical procedures for the analysis of a given parameter.
Although each procedure is an EPA approved method, the reason for the
equivalent procedures is that different analytical instrumentation is used
for each method. Although the results obtained are equivalent, there can
be subtle differences in the types of data produced by different
instrumentation. When choosing analytical procedures, consideration should
always be given to the instrumentation used in order to select the method
that will best satisfy the stated analytical requirements.
The most common example of equivalent procedures using different
instrumentation for the analysis of the same parameters is the gas
chromatography (GC) and gas chromatography/mass spectrometry (GC/MS)
procedures used for the analysis of organic compounds. For example,
consider the analysis of a water sample for volatile organics using each
type of instrumentation. Both GC and GC/MS analysis will yield qualitative
and quantitative results but there are trade-offs associated with each
technique. In deciding which analytical technique to use in the above
example, the major issues to consider are the confirmation of compound
identification, the detection limits achievable, and the cost of analysis.
In general, GC/MS analysis is superior to GC analysis in the area of
compound identification. This is because individual compounds are
identified by their mass spectra, which are plotted from a compound's ion
fragmentation pattern. These fragmentation patterns are specific (at least
AW3E-3 9-18
-------
OSWER Directive 9355.0-7A
for target compounds) and can be thought of as a compound's fingerprint.
This identification is further confirmed by the chromatographic retention
time. By use of the GC/MS system's spectral library, the tentative
identification of non-target compounds found in a sample can also be made
by running a library search.
Compound identification is determined by chromatographic retention time
alone in GC analysis. A second analytical column can be used to confirm an
identification (the confirmation column) but the primary method of
identification is still the retention time. Gas chromatographs are set up
with different detectors that are sensitive to different classes of
compounds which do provide some qualitative information. The presence of
non-target compounds can also be detected by GC, but the identification of
unknown chromatographic peaks is limited to trial and error. There is
nothing analogous to a GC/MS library search.
The GC has advantages over GC/MS analysis when the lowest possible
detection limits are the deciding factor. For volatile organic compounds
in water, the difference in detection limits can be as much as a factor of
100. Typically, GC methods report detection limits ranging from 0.1 to 1.0
ug/1; GC/MS detection limits typically range from 5.0 to 10.0 ug/1 for this
type of analysis. The GC/MS detection limits for specific compounds can be
improved by using the selective ion monitoring (SIM) technique.
GC analysis is usually less costly than GC/MS. The relative costs for each
type of analysis will vary from laboratory to laboratory, however,
depending on which type of instrumentation is more commonly used for a
given application by that laboratory. Also, the manner in which a gas
chromatograph is set up for analysis will influence cost. (GC analysis of
volatiles requires two separate detectors. If the two detectors are placed
in series, only one analysis is required versus two separate analyses if
the detectors are used individually.)
In deciding which type of instrumentation to use for volatile organic
analysis, the central issue is whether the increased confirmation of GC/MS
identification is worth the trade-off in detection limits. Often, the
AW3E-3 9-19
-------
OSWER Directive 9355.0-7A
identity of the contaminants is already known. In this ease, GC/MS
analysis is not providing any additional information other than confirming
what is already known. However, the increased sensitivity provided by GC
analysis can provide much usable information such as insight as to how
"clean" a water supply really is, a better indication of where the leading
edge of a ground water contaminant plume is, and the low analysis of
drinking water. As a rule of thumb, samples taken from unknown sites or
sources should always be analyzed by GC/MS, at least initially, to confirm
compound identification. Also, GC/MS is the method of choice when
non-target or a wide variety of different compounds are suspected. When
the application is more of a monitoring function, or whenever low level
analysis is required, GC procedures are the method of choice. It should be
noted that once non-target compounds are identified, they too can be
analyzed by GC.
Regarding the choice between GC versus GC/MS analysis for acid/base-neutral
extractable compounds (ABNs), pesticides, and PCBs, the choice is more
clear cut. There is no reason to opt for GC analysis of the ABN fraction
unless specific compound analysis is all that is desired. Pesticides and
PCBs are usually analyzed by GC procedures because of the greatly improved
detection limits. However, if positive values are detected and are above
GC/MS detection limits, the analysis is usually confirmed by GC/MS.
9.2.3 MEDIA VARIABILITY
Decision makers and data users should be aware that a great deal of
variability exists in regard to how a given analytical technique or method
responds to a given sample medium. Most of the analytical methods utilized
in support of RI/FS activities were developed, at least originally, for
aqueous samples and modified for use with other media with varying results.
Also, the quality control data published for most analytical methods
(concerning accuracy and precision information) were developed using
aqueous samples. The performance criteria published may not totally apply
to the use of the method with other sample media. When considering the
analysis of source materials, leachate or other complex matrices, qualified
analytical support personnel should be consulted to determine the most
appropriate analytical approach.
AW3E-3 9-20
-------
OSWER Directive 9355.0-7A
9.2.4 METHOD DETECTION LIMIT
Regardless of the specified method detection limit, the actual detection
limit reported may be sample specific. This is especially true of samples
having complex sample matrices. Also, if the concentration of a particular
sample constituent is so high that it requires dilution prior to analysis,
the resulting detection limit for that sample will be raised by the
dilution factor. For example, consider a sample being analyzed by GC/MS
for volatile organics. If the laboratory's normal detection limit for this
method is 5.0 ug/1, and the sample contains 20.0 mg/1 of benzene, the
sample will have to be diluted (say by at least a 1:10 ratio) and the
resulting detection limit will be 50.0 ug/1. In some cases, the laboratory
can analyze the same sample twice to obtain the specified detection limit
but this is not always possible, is not considered standard practice, and
would have to be specified prior to sample submittal. If several compounds
are present with an order of magnitude difference in concentrations, then
all the compounds may not be reported.
Another factor regarding detection limits is that data quality parameters
are usually concentration dependent. The standard error of the analytical
method being used increases as the concentration of the analyte of interest
decreases. The surest way for predicting what the accuracy and precision
will be for analyses at the detection limit is by generating QC data using
the detection limit concentration. In light of this decrease in the level
of certainty as the concentration decreases, the relationship between
action levels and detection limits should be considered carefully.
It is important to recognize that quantitative results reported at the
detection limit may not be reliable. If the action level of a contaminant
is 5.0 ug/1, an analytical method with a detection limit of 5.0 ug/1 may
not provide suitable data to meet the criteria. For example, the action
level for trichloroethene (TCE) as defined by the Safe Drinking Water Act
as a proposed Maximum Contaminant Level (MCL) is 5 ug/1. Analytical method
624 for volatile organics by GC/MS has a detection limit of 5 ug/1.
AW3E-3 9-21
-------
OSWER Directive 9355.0-7A
However as analytical techniques have an inherent inaccuracy, especially
near their detection limits, based on the objectives of the analysis,
method 624 may not be applicable.
When levels of interest are at or approaching MDL, caution must be used in
specifying precision in terms of a percentage. The use of percentages
distorts accuracy and precision information when relatively small numbers
are being compared. Accuracy and precision in terms of absolute values or
ranges may be more appropriate. For example, consider a precision
objective specifying that blind replicate samples must be within 50 percent
of each other. If the two replicate concentrations are 50 and 75 ug/1, the
use of this objective is reasonable. If the two replicate concentrations
are both in the 1-10 ug/1 range, the 50 percent objective would classify
these results as being outside of criteria, whereas in all probability the
replicate analyses show excellent precision. Caution must be used when
applying objectives expressed in percentages to numbers less than fifty.
If the precision objective is specified using a percentage, the working
range of the objective should also be specified.
9.2.5 MATRIX EFFECTS
A matrix effect is a phenomenon that occurs when the sample composition
interferes with the analysis of the analyte(s) of interest. This can bias
the sample result either in a positive or in a negative way, with the
negative bias being the most common.
The magnitude of a matrix effect is best assessed by the use of matrix
spikes. Matrix spikes supply percentage recovery information which
addresses the amount of bias present in the measurement system. This
information can be used to adjust reported concentrations by the
application of a correction factor based on percentage recovery. It is not
recommended that sample values actually be adjusted for percent recovery
unless a worst-case scenario is being developed. For non-aqueous matrices
(soils, sediments, leachate, solid wastes, etc.), this type of data should
be collected over the life of a project so that certain expectations about
the quality of data being produced can be developed.
AW3E-3 9-22
-------
OSWER Directive 9355.0-7A
9.2.6 TENTATIVELY IDENTIFIED ORGANIC COMPOUND (TIC)
Under the CLP RAS procedures, 30 non-HSL peaks present in the reconstructed
ion chromatogram are identified as tentatively identified compounds (TICs).
Other laboratories may not address TICs or have different reporting
criteria. If compounds of interest are tentatively identified by GC/MS and
are high in spectra matching criteria (above 90 percent match) and above
action levels, samples may be re-run against a standard in order to verify
the compound's identity. Chromatographic retention time consideration is
an important factor in assessing the probability of tentative
identification reliability. Approaches for providing more reliable
tentative identifications are under development.
9.2.7 DATA QUALIFIERS
When analytical data are validated, the analytical results and the
associated QA/QC information are reviewed using criteria specific to the
analysis performed. This review can range from superficial to very
rigorous, depending on the level of analytical support utilized and the
type of technical review requested by the data user.
Data qualifiers are commonly used during the data validation process to
classify sample data as to its conformance to QC requirements. The most
common qualifiers are listed below:
o A - Acceptable
o J - Estimate, qualitatively correct but quantitatively suspect
o R - Reject, data not suitable for any purpose
o U - Not detected at a specified detection limit (e.g., 10U)
Sample data can be qualified with a "J" or "R" for many different reasons.
Poor surrogate recovery, blank contamination, or calibration problems,
among other things, can cause sample data to be qualified. Whenever sample
data are qualified, the reasons for the qualification are stated in the
data validation report. Data users are reminded that data validation is
AW3E-3 9-23
-------
OSWER Directive 9355.0-7A
generally performed using strict analytical criteria which do not take the
sampling activity's DQOs into account. Data users should request that the
technical staff interpret the validation report according to the sampling
activity's objectives and data uses. For example, data qualified with a
"J" may be perfectly suitable for some data uses.
9.3 ANALYTICAL UNCERTAINTY
Analytical uncertainty can be estimated for each compound or element of
interest by method. Analytical uncertainty should be calculated for
non-standard (Level V) or field (Level I) methods when possible.
In order to determine total analytical uncertainty, the accuracy and
precision of the method must be known. Calculation of total uncertainty is
based on the generally accepted assumption that there is a direct
relationship between the accuracy and precision of environmental analyses
and the analyte concentration and matrix of the environmental sample. The
information required to develop meaningful calculations of analytical
uncertainty would include interlaboratory information for matrix spikes,
surrogate recoveries, duplicated and blind performance evaluation standards
for each compound analyzed for each analytical procedures as follows:
o Statistical Information - N, bias, RSD of percent recovery,
concentration of spike, and concentration of analyte
o Matrix - Air, aqueous, soil/sediment, leachate or source material
o Concentration Range - Liquids: 0-10 mg/1; 10-100 mg/1; 100-1000
mg/1 or >1000 mg/1. Solids: <1 mgAg; 10-1000 mg/kg; or >1000
mgAg
If the above listed information is available, analytical uncertainty could
be predicted for the majority of analyses conducted in support of remedial
actions.
(
For example; based on N number of interlaboratory spike recoveries of
benzene from ground water matrices, in the 0 to 10 ug/1 concentration range
using Method 624, the confidence interval at the 95 percent confidence
AW3E-3 9-24
-------
OSWER Directive 9355.0-7A
level can be stated. This statement would be further qualified based on
the number and types of laboratories, other types of performance evaluation
criteria, matrix strength, and other pertinent analytical information.
In the absence of the required information, uncertainty statements could
still be made using available information, but would be somewhat more
qualified. The confidence level associated with the analytical data will
be directly related to how closely the QA/QC data base used to develop the
uncertainty statement reflects the actual analytical conditions. In some
cases, a significant deviation between the analytical conditions associated
with the uncertainty statement and the actual analytical conditions could
produce significant distortions.
Given that the statistical data described above are not presently available
for all analytical support levels, the following represents an estimate of
the type of uncertainty statements that can be produced for analytical
support levels II, III, and IV, using available information.
9.3.1 LEVEL IV
Precision and bias data provided by the CLP FAS to be used in the
estimation of analytical confidence limits include:
o Interlaboratory volatile organic matrix spike duplicate data for
water and soil samples (N, percent RSD, percent RSD at 85th
Percentile)
o "Interlaboratory" surrogate recovery data of generated volatile
compounds from water and soil material (N, bias percent, percent
RSD)
o Interlaboratory performance evaluation standard data for volatile
and semi-volatile organic compounds in water and soil (N, bias
percent, percent RSD).
In all cases, the data base has been sanitized, i.e., outliers have been
removed. In the case of "interlaboratory" surrogate recoveries the data
base should be considered interlaboratory in the classic sense - same
sample submitted to a number of laboratories - but it is actually a close
approximation. The same chemical surrogates are added to samples in
AW3E-3 9-25
-------
OSWER Directive 9355.0-7A
individual laboratories but the laboratories are not recovering the
surrogate from the same matrix. In addition, recovery data should be
provided for the air, leachate and source material.
All of this information can still be used individually or in concert to
develop uncertainty statements but with some inherent limitations.
o The interlaboratory matrix spike data as provided do not stratify
the data with respect to concentration. Using this data would
require the implicit assumptions that matrix recovery is independent
of concentration and the laboratory in which the analysis was
conducted.
o The "interlaboratory" surrogate recovery data are generally for one
concentration range and as a result do not account for variability
of accuracy as a function of concentration, assume that all analytes
act as surrogate during the analytical process, and do not account
for interlaboratory variations associated with different matrices.
o Interlaboratory performance evaluation standard data can probably be
considered a "best case" for the development of uncertainty
statements (actual samples would have a greater degree of
uncertainty). The uncertainty associated with these data do not
account for true sample matrix effects, or a wide range of analyte
concentrations and as a result, the actual analytical uncertainty
could only be worse than that estimated using this data set. It
does have the advantage of being truly interlaboratory and blind
(sample concentrations not known by participating laboratories) and
should be a true measure of analytical uncertainty for the
concentration range and matrix analyzed.
The best estimate of analytical uncertainty would be a composite of the
uncertainty associated with matrix spike with the uncertainty associated
with performance evaluation standards (interlaboratory performance only).
9.3.2 LEVEL III
The available information to estimate uncertainty for Level III is the
accuracy and precision statements included with the individual EPA approved
procedures in SW-846. While this information is rarely stratified as to
matrix and concentration, it could serve as a starting point (best case)
from which the uncertainty associated with the actual analytical conditions
could be estimated.
AW3E-3 9-26
-------
OSWER Directive 9355.0-7A
9.3.3 LEVEL II
The roost important factor that influences the uncertainty associated with
Level II analyses is the skill of the analyst doing the work. Because the
procedures used are not formalized, a great deal of improvization usually
takes place. The inherent variability of the procedures themselves would
make the development of a centralized quality assurance data base tenuous.
The same reasoning would apply to making uncertainty predictions based on a
centralized data base.
9.3.4 LEVEL I
Level I analyses are qualitative, and therefore it is not possible to
quantify the uncertainty in these methods.
9.4 REFERENCES
Aleckson, K.A., J.W. Fowler and Y.T. Lee. 1986. inorganic Analytical
Methods Performance and Quality Control Considerations. In: Quality
Control in Remedial Site investigation: Hazardous Industrial Solid Waste
Testing Fifth Volume ASTM STP 925.
American Public Health Association, American Water Works Association, Water
Pollution Control Federation. 1975. Standard Methods for Examination of
Water and Wastewater, 14th Ed.
American Society for Testing Materials. 1976. Annual Book of ASTM
Standards, Part 31, "Water", Standard D3223-73, p. 343
Anderson, D.C., K.W. Brown and J. Green. 1981. Organic Leachate Effects on
the permeability of clay liners. National Conference on Management of
Uncontrolled Hazardous Waste Sites, pp 223-229. October 28-30, 1981
Washington, DC.
Bishop, J.N. 1971. Mercury in Sediments, Ontario Water Resources Comm.,
Toronto, Ontario, Canada.
Boston Society of Civil Engineers. 1985. Controlling Hazardous Wastes.
Lecture Series.
Brandenberger, H. and H. Bader, 1967. The Determination of Nanogram Levels
of Mercury in Solution by a Flameless Atomic Absorption Technique, Atomic
Absorption Newsletter (6), 101
CDM. 1986 Draft Memorandum re: XRF Field Analysis of Smuggler Mountain
Soil Samples from R. Chappel, R. Olsen to J. Hillman January 13, 1986
EPA Contract No. 68-01-6939 Document No. 149-WPl-EP-CCCU-l.
AW3E-3 9-27
-------
OSWER Directive 9355.0-7A
Federal Register, Organochlorine Pesticides and PCBs, Method 608;
2,3,7,8-TCDD, Method 613; Purgeables (Volatiles), Method 6224;
Base/Neutrals, Acids and Pesticides, Method 625; Vol. 44, No. 233,
Monday, December 3, 1979, pp. 69501, 69526, 69532 and 69540.
Flotard, R.D., M.T. Homshen, J.S. Wolff and J.M. Moore. 1986. Volatile
Organic Analytical Methods Performance and Quality Control
Considerations. In: Quality Control in Remedial Site Investigation:
Hazardous and Industrial Solid Waste Testing, Fifth Volume ASTM STP 925.
Garbarino, J.R. and H.E. Taylor, 1979. An Inductively-Coupled Plasma Atomic
Emission Spectrometric Method for Routine Water Quality Testing. Applied
Spectroscopy 33, (3)
Garner, F.C., M.T. Homsher and J.G. Pearson. 1986. Performance of USEPA
Method of Analysis of 2, 3, 7, 8 - Tetrachlorodibenzo-P Dioxin in Soils
and Sediments by Contractor Laboratories. In; Quality Control in
Remedial Site Investigation: Hazardous and Industrial Solid Waste
Testing Fifth Volume TSTM STP 925.
Goulden, P.O. and B.K. Aighan, 1970. An Automated Method for Determining
Mercury in Water. Technicon, Adv. in Auto. Analy. 2 317
Government Institutes. 1982. Superfund Comprehensive Environmental
Response, Compensation, and Liability Act of 1980. Third Edition.
Hatch, W.R. and W.L. Ott, 1968. Determination of Sub-Microgram Quantities
of Mercury by Atomic Absorption Spectrophotometry Analytical Chemistry
40: 2085.
Kopp, J.F., Longbottom, M.C. and Lobring, L.B. 1972. Cold Vapor Method for
Determining Mercury. AWWA, 64:20.
Martin, T.D. (EMSL/Cincinnati). Inductively Coupled Plasma - Atomic
Emission Spectrometric Method of Trace Elements Analysis of Water and
Waste, Method 200.7, Modified by CLP Inorganic Data/Protocol Review
Committee.
Martin, T.D., J.F. Kopp, and R.D. Ediger, 1975. Determining Selenium in
Water, Wastewater, Sediment and Sludge by Flameless Atomic Absorption
Spectroscopy. Atomic Absorption Newsletter 14: 109
Owerbach, D. 19 The Use of Cyanogen Iodide (CNI) as a Stabilizing Agent
for Silver in~~Photographic Processing Effluent Sample. Photographic
Technology Division, Eastman Kodak Company, Rochester, New York, 14650.
Shackelford, W.M., D.M. Cline, L. Faas, and G. Kurth, 1983. An Evaluation
of Automated Spectrum Matching for Survey Identification of Wastewater
Components by Gas Chromatography - Mass Spectrometry. Analytica Chimica
Acta.
Technicon Industrial Systems. 1980. Operation Manual for Technicon Auto
Analyzer 11C System. Technical Pub. #TA9-0460-00, Tarrytown, New York.
AW3E-3 9-28
-------
OSWER Directive 9355.0-7A
U.S. EPA. 197_ Handbook for Analytical Quality Control in Water and
Wastewater Laboratories, USEPA-600/4-79-019.
. 1973. Handbook for Monitoring Industrial Wastewater, USEPA
Technology Transfer.
. 1974. Methods for Chemical Analysis of Water and Wastewater, USEPA
Technology Transfer.
. 1979. Methods for Chemical Analysis of Water and Wastes.
600/4-79-20.
. 1981. EMSL, Users Guide for the Continuous Flow Analyzer Automation
System. Cincinnati, Ohio.
. 1982. Test Method for Evaluating Solid Waste. Physical/Chemical
Methods. SW-846. 2nd Edition.
. 1982. Office of Solid Waste and Emergency Response, Modification (By
Committee) of Method 3050. SW9846, 2nd Ed., Test Methods for Evaluating
Solid Waste. July.
. 1984. Soil Properties, Classification, and Hydraulic Conductivity
Testing. Draft-Technical Resource Document for Public Comment. SW-925.
. 1984. Solid Waste Leading Procedure. Draft-Technical Resource
Document for Public Comment. SW-924.
. 1984. Toxicity Characteristic Leaching Procedure (TCLP) - Draft
Method 13XX. TK0703.
. 1984. OERR. User's Guide to the Contract Laboratory Program.
. 1984. Test Methods for Evaluating Solid Waste, Physical/Chemical
Methods. SW-846.
. 1984. Calculation of Precision, Bias and Method Detection Limit for
Cfiemical and Physical Measurements. (QAMS Chapter 5.)
. 1986. Demonstration of a Technique for Estimating Detection Limits
with Specified Assurance Probabilities. Contract No. 68-01-6939.
Winefordner, J. D., Trace Analysis: Spectroscopic Methods for Elements.
Chemical Analysis, Vol. 46: 41-42.
Winge, R.K., V.J. Peterson, and V.A. Fassel, 1970 inductively Coupled
Plasma - Atomic Emission Spectroscopy Prominent Lines. EPA-600/4-79-017.
Wise, R.H., D.F. Bishop, R.T. Williams, and B.M. Austern, 19 Gel
Permeation Chromatography in the GC/MS Analysis of Organics in Sludges.
USEPA, Municipal Environmental Research Laboratory; Cincinnati, Ohio
45268.
AW3E-3 9-29
-------
OSWER Directive 9355.0-7A
Wolff, J.S., M.T. Homsher, R.D. Flotard and J.G. Pearson. 1986.
Semi-volatile Organic Analytical Considerations. In; Quality Control in
Remedial Site Investigation. Hazardous and Industrial Solid Waste
Testing, Fifth Volume, ASTM STP 925.
AW3E-3 9-30
-------
Section 10.0
-------
OSWER Directive 9355.0-7A
10.0 SAMPLING CONSIDERATIONS
The uncertainty introduced by sampling procedures must be considered during
the development of DQOs. Factors which affect sampling uncertainty include
sampling/handling variability and the variability of contaminants as a
function of location and time. The magnitude of each of these factors is
largely site specific. The site specific nature of sampling uncertainty
distinguishes sampling from analytical uncertainty, which is largely site
independent.
In this section, factors associated with sampling uncertainty are
discussed. Discussion of specific sampling methods is not provided.
Rather, the discussion centers on how and why each factor influences
sampling uncertainty and provides general guidance on sampling
considerations to be evaluated during DQO development. Rigid guidelines
for sampling design are not provided because of the site specific nature of
the sampling uncertainty.
10.1 SAMPLING STRATEGY
In designing a sampling plan there are a large number of factors which must
be considered. Some of these factors such as the physical characteristics
of the site (geology, hydrogeology, physiography) are unique to each site.
However, there also are several general factors which must be considered
for all sites^. The general factors include decisions addressed during the
DQO process such as:
o Will a phased approach be used?
o Will samples be collected for site characterization?
o Will samples be collected for confirmation purposes?
o Will grab or composite samples be collected?
o Will a grid system be used?
The importance of each of these factors varies from site to site, and
therefore must be analyzed individually.
AW3D-3 10-1
-------
OSWER Directive 9355.0-7A
Remedial investigations are undertaken to collect the data needed to
evaluate and select appropriate remedial actions to be implemented at a
site. The type of data collected will vary depending on the media of
interest. Although most RIs require that multi-media sampling be
performed, in some cases the sampling may be limited to one medium. The
level of detail to which an investigation is undertaken is influenced by
the amount of existing data and by the remedial actions which could
potentially be implemented at the site.
For sites at which a significant amount of data have been generated as a
result of preliminary assessments and site investigations a focused
approach to the RI can be developed. For sites for which little or no data
are available or data are inconclusive, a broader approach to site
investigations must be implemented.
In all cases, the RI should be directed towards providing the information
required to select and implement a remedial action. It may be quite
obvious at the outset of an RI/FS that the major source of contaminants at
a site is a lagoon, for instance. The RI should therefore be geared
towards identification of the characteristics of the wastes in the lagoon
which would govern the manner in which it could be disposed of (i.e., PCB
content, RCRA parameters etc.). The RI must also provide information
relative to the quantity of waste material which must be disposed of or
treated in order to allow for development of cost estimates during the
feasibility study.
10.2 SAMPLING PROGRESSION
Samples obtained at uncontrolled hazardous waste sites during a remedial
investigation are obtained in a progressive manner to allow for expansion
of the data base in a controlled manner. Due to the heterogeneity of
materials present and the variability of environmental conditions at sites,
it is difficult in most cases to develop a work plan which will encompass
all environmental measurement activities which may be undertaken in order
to satisfy the RI/FS and RD. Because of the variable nature of sites, RIs
must be designed in a manner which allows for flexibility and adjustment of
AW3D-3 10-2
-------
OSWER Directive 9355.0-7A
sampling approaches based on data which are continually obtained during
field investigations. This type of progressive sampling approach can be
accommodated by implementation of a phased RI/FS.
In the DQO process it may be necessary to identify a sampling approach
before sufficient information has been gathered to use statistical methods
as discussed in Section 8.0. In these cases, it may be beneficial to use a
phased data collection approach. In a phased approach, samples are
collected in a series of independent sampling events. The first phase may
be undertaken for site characterization purposes while subsequent phases
use the information generated by earlier phases to fill in data gaps. If a
mobile lab is utilized, phases may be continuous as results are analyzed
and data gaps are identified and filled. The DQO process applies to each
phase of an RI and for each sampling task. Initial sampling undertaken
during the first phase may not yield specific information since little or
no site specific data may be available. However, in subsequent phases of
the RI more data will be available for decision making.
A phased approach to sampling is, in most cases, a cost effective method
since areas of concern are identified in the early phases and are then
targeted for additional sampling. When sampling is performed in only one
phase, every conceivable target must be completely sampled. If one or
several of the targets prove to be uncontaminated, a large number of
unnecessary samples will have been taken.
Phased approaches must be developed on a site specific basis but generally
will follow sequentially from less intensive to progressively more
sophisticated field sampling and analysis programs as follows:
o Review of existing information/data
o Remote sensing
o Field screening
o Intrusive sampling
o Pilot studies
AW3D-3 10-3
-------
OSWER Directive 9355.0-7A
10.2.1 REVIEW OF EXISTING INFORMATION/DATA
All sources of available information should be obtained and reviewed during
the initial stages of the RI/FS work plan preparation process. It is
especially important to obtain and review data from any previous
investigations gathered in the National Priorities List (NPL) ranking
process, FIT and/or TAT team investigations, and other data gathering
activities conducted by the state or other parties. Detailed
discussions of the various data sources which should be accessed during
review of existing information are contained in the Remedial Investigation
Guidance Document (EPA, 1985).
10.2.2 REMOTE SENSING
Remote sensing is a term applied to methods used for the detection,
recognition, or evaluation of objects or conditions by means of distant
sensing or recording devices including aerial photography or satellite
imagery. Geophysical techniques are in fact remote sensing methods which
allow for the characterization of subsurface conditions without excavation.
For the purposes of this document, remote sensing will be used to denote
airborne data collection techniques and geophysical methods are those
applied at the earth's surface.
Remote sensing/geophysical techniques are used in RIs to obtain data on
site conditions while minimizing disruption of the site environs. These
techniques are used extensively for selection of sampling points and siting
for installation of in-situ monitoring devices such as ground water
monitoring wells.
Remote sensing and geophysical techniques can be applied as a tool in the
overall remedial investigation of a hazardous waste site. However, in no
instance should remote sensing or geophysical techniques be utilized in the
place of disruptive techniques for the confirmation of data obtained.
Remote sensing and geophysical techniques should be used in conjunction
with disruptive techniques to adequately characterize a site.
Remote sensing/geophysical techniques can be used to:
AW3D-3 10-4
-------
OSWER Directive 9355.0-7A
o Map geohydrologic features — both natural and those changed by man
o Map conductive and nonconductive contaminant plumes in both the
saturated and unsaturated zones
o Locate and define the horizontal and vertical extent of buried
materials
o Locate and define buried objects such as drums, tanks, pipes,
conduits, etc.
o Locate and define sources of contamination
Remote sensing/geophysical techniques provide data that are useful in the
performance of RIs. The information can be useful in development of DQOs
when generated early in the RI process. Remote sensing/geophysical
investigations should be used in the initial stages of RIs in order to gain
an overall sense of the site environs (aerial photographs) and subsurface
conditions. These techniques may also be utilized in the latter phase of
the RI to provide a means for extrapolation of data obtained from
disruptive techniques. For example, soil borings installed at a site may
reveal the presence of a clay lens over a portion of a site which could
affect ground water migration. Geophysical techniques could be used to
provide information on subsurface conditions between the soil borings and
allow for a determination of the nature of subsurface conditions. In the
absence of this information, an extrapolation of the soil strata between
the borings may result in an erroneous interpretation of subsurface
conditions.
When geophysical techniques will be utilized, the RI/FS process should be
phased to allow for interpretation of the remote sensing/geophysical data
and application of this information in the process for subsequent RI
phases.
Data obtained upon completion of each successive phase should be evaluated
in order to determine if the subsequent phase should be modified.
The results of geophysical investigation may provide information on the
suspected areal extent of a contaminant plume. This information can be
used in the DQO process for guiding selection of sampling points or
AW3D-3 10-5
-------
OSWER Directive 9355.0-7A
monitoring well installation locations. Rather than establishing a grid
system over the entire site, the results of the geophysical investigation
could be used to guide in the establishment of a targeted sampling grid.
Although this may introduce bias into the sampling approach from a
statistical perspective, the value of taking a completely unbiased approach
must be weighed against the cost, schedule and overall objectives of the
RI/FS. Likewise, selection of sampling locations at random for a site may
not take into account accessibility of areas to drilling rigs, for
instance. Modification of the sampling locations are necessitated by
safety considerations as well as practical considerations.
In establishing a grid system for sampling, remotely sensed information
such as aerial photographs are required at a minimum. This information is
useful in determining which areas may not be accessible for drill rigs or
not practical to sample because of other restrictions (e.g., rock outcrops,
water bodies). Final sampling locations are also modified in the field
based on the information available to the investigator. For example,
magnetometry or radar may indicate that buried drums are present in a
localized area. The investigator may therefore choose to relocate the soil
boring location based on safety considerations. Relocation of the sampling
point should be justified and should not be construed as diminishing the
quality of the data, since the investigator is relocating the sampling
point through a logical, thorough process.
Remote sensing and geophysical techniques can be considered as survey
methods to define areas in which investigations should be undertaken. The
limits of these areas (i.e., lateral and vertical extent) and the
characteristics of these areas (i.e., degree of contamination) must be
determined based on detailed sampling plans. Remote sensing/geophysical
investigation methods provide information necessary to develop a systematic
approach for sampling.
Remote sensing or geophysical techniques are generally not applied to sites
in a random manner. These techniques are generally used to characterize a
specific area of interest (i.e., historical aerial photographs of site to
identify areas used for waste disposal or aerial photos to define limits of
•^
AW3D-3 10-6
-------
OSWER Directive 9355.0-7A
a watershed). Geophysical techniques may be used to confirm areas with
suspected buried wastes. Those survey techniques are therefore applied to
areas based on an investigator's knowledge or bias regarding the site.
10.2.3 FIELD SCREENING
Field screening is primarily used to provide indications of contamination
(e.g., Level I & II - Field Screening). Thus, the decisions that will be
based on the results of this type of sampling are in many cases yes/no type
decisions. For instance, on the basis of soil gas sampling it may be
determined that contamination of a particular unconfined aquifer is
indicated and further direct sampling is warranted.
10.2.4 INTRUSIVE SAMPLING
Intrusive sampling includes all methods in which a physical sample from the
media of concern is obtained. Intrusively obtained samples are used to
obtain a numerical value for a physical or chemical measurement at a
particular point. Intrusive sampling provides much more exact information
concerning the concentration of contaminants or physical features than
non-intrusive remote sensing or field screening techniques.
All intrusive sampling methods are used to obtain physical samples of
material from the media of interest. Samples may be highly variable for
some sampling methods. For instance, when sampling a stream at a
particular location, the source of any contaminants found is uncertain.
Since the stream is a dynamic system, the concentration of contaminants is
expected to vary substantially over time and therefore a higher level of
variance is associated with this sample. Other examples of direct samples
with high variance are:
o Storm water runoff samples
o Cuttings sampled from a solid stem auger
o Ground water samples from wells with very large screened intervals
AW3D-3 10-7
-------
OSWER Directive 9355.0-7A
For each of these types of samples it is difficult to specify the exact
position from which any contamination originated. All of these sample
types will have large variability associated with them. These types of
direct samples are often used to investigate suspected contamination
pathways or to obtain background site information.
The highest level of confidence is associated with data obtained from
samples taken at precise locations in relatively stable media such as soil.
Such data are not expected to have high temporal variability. Examples of
data which would be expected to have low temporal variability include:
o Soil samples obtained from a split spoon
o Ground water samples from a monitoring well
o Surface water samples from a lined lagoon
o Sediment samples
This type of data is required if decisions requiring precise point
estimates of contamination will be made.
It should be noted, however, that over time contaminant concentration may
increase in any media as the source materials are subject to leaching.
Contaminant migration may in some cases approximate the behavior of a
wetting front, where a high concentration of contaminants analogous to the
advancement of a wetting front in soil. The concentration of contaminants
may be very low ahead of the wetting front and significantly greater at the
wetting front. In other cases, dispersion may be a major factor to
consider. In these cases, the contaminants may disperse through an aquifer
system at low levels. Soil contaminant evaluation models should be
utilized in evaluating these situations.
Intrusive samples are obtained to determine the contamination within the
site. These samples can be obtained to investigate suspected contaminant
pathways, identify contaminants present in the waste material, measure
contaminant concentrations for health and safety purposes, or examine the
contamination present in a municipal well. Intrusive sampling generally
AW3D-3 10-8
-------
OSWER Directive 9355.0-7A
entails different levels of data quality. For example, samples taken to
determine if a municipal well is contaminated require much higher data
quality than samples used to determine the range of contaminants present at
a site. Thus, careful analysis of the uses and purposes of intrusive
samples must be undertaken during the DQO process.
10.2.5 PILOT STUDIES
Pilot studies are undertaken to obtain data to assess the applicability of
various proposed alternatives for site remediation in a controlled manner.
in general, pilot studies can be designed to allow for control of variables
and thereby generate high quality data. Pilot studies include soil column
tests in which a known quantity of waste is applied to a known mass of soil
over a fixed period of time to evaluate the attenuation capacity of the
soil. Pilot testing could also entail isotherm testing to evaluate the
adsorption capacity of carbon for a specific compound present in ground
water associated with a site. In this type of approach, known quantities
of adsorbant are mixed with ground water containing various concentrations
of the contaminant of concern to construct isotherms.
Pilot studies can also be undertaken to evaluate the effectiveness of
various unit processes for treatment of a contaminant source at a site or
for developing data needed to optimize system design and operation.
Pilot treatability studies are undertaken to evaluate the technical
performance of a unit process or system. Optimization testing of the
system is generally undertaken to determine the most effective set of
operating parameters for the system. During optimization testing the
sensitivity of overall system performance for each parameter is evaluated.
The results of pilot treatability studies are used to develop design
criteria; develop cost estimates; and to identify any special management or
operational constraints which must be implemented in order to utilize the
system.
AW3D-3 10-9
-------
OSWER Directive 9355.0-7A
The approximate range of applicability of treatability procedures for
various contaminants is given in Figure 10-1. Water quality parameters
which should be determined to evaluate the potential effectiveness of
treatability schemes are given in Table 10-1.
10.3 SOURCES OF VARIABILITY
To determine the uncertainty associated with a decision, all sources of
variability must be identified and either measured or inferred. Other
important sources of variability are sampling/handling variability and the
variability of contaminants as a function of location and time. Of these
three sources of variability, the variability of the contaminants as a
function of location is expected to be the largest.
10.3.1 SAMPLING/HANDLING VARIABILITY
Sampling/handling variability is defined as any variability introduced by
the sampling and/or handling procedures, resulting in a contaminant
concentration in the sample that is different than the concentration in the
original media. Causes of sampling variability include incorrect sampling
procedures and cross contamination. Since most of the causes of
sampling/handling variability are related to errors in procedures,
measurement of sampling variability is difficult. The magnitude of
sampling variability can range from small to very large; however, if
correct sampling and handling procedures are followed, sampling variability
should be small compared to laboratory variability.
Sampling/handling variability can effectively be reduced by thoroughly
training sampling personnel and ensuring that all sampling is performed in
accordance with standard operating procedures (SOPs). SOPs are developed
to ensure that any samples collected are representative of the undisturbed
media of interest. By adhering to the SOPs, intra- and intersite
variability for a given sampling method are greatly reduced or eliminated.
AW3D-3 10-10
-------
OSWER Directive 9355.0-7A
LEGEND
I-
Air Stripping
Carbon Adsorption
1
Ozone/UV Radiation
I 1
Commercially applied
o
----- Potential extension
It
I I
Thin Film Evaporation
I ------ 1
Fractional
.1 Distillation i
Steam Stripping ,
• Wet Air Oxidation • i
0.01
0.05 0.1
05
Initial % Organic?
10
II
50 100
SOURCE: Allen. C.C. & B.C. Brancy, 1985. 'Techniques for Treating Hazardous Wastes to
Remove Volatile Organic Constituents." JAPCA, Vol. 35
No. 8. August 1985.
FIGURE 10-1
APPROXIMATE RANGES OF APPLICABILITY OF VOC
REMOVAL TECHNIQUES AS A FUNCTION OF
ORGANIC CONCENTRATION IN LIQUID WASTE STREAMS
-------
OSWER Directive 9355.0-7A
TABLE 10-1
CONVENTIONAL WATER QUALITY PARAMETERS
FOR TREATABILITY STUDIES
o Iron (mg/1)
o Manganese
o pH
o Total dissolved solids
o Total suspended solids
o Total organic carbon
o Total organic halides
o Hardness
o Alkalinity
o Organic color
o Chemical oxygen demand
o Filtered and unfiltered metals
AW3F-14 10_12
-------
OSWER Directive 9355.0-7A
Sampling/handling variability is expected to be largest for volatile
organic compounds. During sampling, care must be exercised to avoid
volatilizing these compounds. Samples must be kept cool and separated from
organic fumes during shipping and handling to avoid migration of organic
compounds out of or into the sample vial. Measurement of any migration of
organic compounds into the sample can be performed by shipping trip blanks
along with the samples.
Another important component of sampling/handling variability is cross
contamination. Cross contamination can be caused by improper
decontamination of sampling or downhole instruments. Cross contamination
can be greatly reduced or eliminated by following proper procedures. Cross
contamination can be identified through the use of field blanks.
10.3.2 TEMPORAL VARIABILITY
Many observed contaminant concentrations are dependent on time related
variables such as the time of day or season of the year. The important
variable linking concentration and time is often climatological (i.e.,
temperature or rainfall). Since the linking variables (temperature, for
instance) follow cyclical patterns over a day or year, time dependent
contaminant levels are also expected to follow cyclical patterns. To
obtain representative samples of time related variables, it is important to
identify the cyclical nature of the contaminant concentrations and to
sample at various phases of the cycle to obtain a representative sample.
Some general considerations related to influences of seasonality on data
for the following media are contained herein: air, surface water, ground
water and soil. Discussion of other environmental media such as biota are
not included.
Air - Assessment of atmospheric contamination at uncontrolled hazardous
waste sites is accomplished by use of real time monitoring devices or
collection of air samples in bags, of particulates on filters, or of
volatiles on sorbents to be analyzed in the laboratory. Seasonal
variations in climate as well as weather conditions at the time of sampling
may affect all of these sample types.
AW3D-3 10-13
-------
OSWER Directive 9355.0-7A
Volatile organic levels in the atmosphere would generally be expected to
rise as temperatures increase, provided the volatilization from the source
is not limited by some other factor. Temperature changes associated with
the seasons may have a pronounced impact on volatile organic levels
emanating from a source. Daily fluctuations in temperatures within any
season would also be expected to significantly affect volatilization rates.
In order to obtain meaningful data, it is necessary to establish what the
data will be used for. If the intent is to determine the risk associated
with volatile emissions to on-site workers, an evaluation of the conditions
over the work day within the season of interest may be adequate. If,
however, the potential risks to sensitive receptors are required, an
assessment which takes into consideration variations of conditions over the
night as well as diurnal fluctuations would be required.
J
Monitoring of air quality should be geared to provide information of use in
assessing site conditions. Air monitoring on clear, calm days may provide
data necessary to establish baseline conditions, however, it may not
provide accurate information relative to particulate contaminant dispersion
associated with windy conditions. Likewise, sampling for particulates when
a snow cover exists would not likely provide information useful to assess
offsite migration as a result of wind erosion.
Sampling during precipitation events can significantly affect data not only
because of its impact on air quality but also because of the impact
moisture has on most air sampling devices and real time monitors.
Precipitation washes particulates as well as volatiles from the air;
therefore, risks associated with particulates and volatiles would be less
during rain events. Precipitation adversely affects electronics of some
real time monitors, resulting in false readings. Rain also hinders the
collection efficiencies of sorbent tubes such as tenax and charcoal as well
as personal sampling pumps. Therefore, sampling of air during rain events
may not provide data of value in undertaking an RI.
AW3D-3 10-14
-------
OSWER Directive 9355.0-7A
Surface Water/Ground Water - Surface water and ground water quality vary
seasonally as influenced by the hydrologic cycle. Inputs to the cycle
which are associated with precipitation or snow melt output include
evapotranspiration and evaporation. Interconnections between surface
waters and ground waters also affect quality of these media. Contaminated
ground waters discharging to surface waters may adversely affect the
quality of the surface waters and vice versa. The degree to which surface
waters and ground water are related is dependent in part upon geologic
conditions. Where these media are intimately related, a change in the
quality of one may be manifest in the other in a very short time.
Where surface waters and ground water are in close communication, seasonal
changes in water quality may be closely correlated. In areas where ground
water is separated from surficial waters (i.e., confined aquifer), seasonal
changes may not be as significant.
Sampling of surface waters and ground waters must be undertaken in a manner
which will provide data representative of the media. Sampling of a stream
during low flow, for instance, may not provide a representative depiction
of the quality of the stream if an assessment of the impact of the flow on
downgradient receptors is required. In order to obtain data which can be
used in evaluating the potential impacts of a stream containing
contaminants on downstream receptors such as wetlands or a water supply,
data quality and quantity determinations over the range of seasonal
conditions should be obtained.
I
During high flow conditions, the concentration of contaminants in a stream
may be decreased as a result of dilution, or in some cases may be elevated
if contaminants are flushed from a source or erosion of contaminants
associated with soil results. Therefore, surface water conditions should
be monitored over a period of time adequate to account for seasonal
variations. Monitoring of meteorological conditions during the sampling
period is necessary in order to provide a means to interpret the data.
AW3D-3 10-15
-------
OSWER Directive 9355.0-7A
Surface water temperatures also vary seasonally and as such can influence
the chemical data obtained. In stagnant surface water systems, evaporation
can have a concentrating effect on chemicals, in addition, elevated
temperatures, within a certain range, can accelerate the rate of
biodegradation or volatilization. Therefore, organic levels may decrease
in some instances, only to rebound once biodegradation rates decline,
provided the contaminant source continues to discharge to the surface water
at a constant rate.
Variations in ground water quality as a result of seasonal changes may not
be as pronounced as those observed in surface water. For instance, ground
water temperatures encountered at depth do not vary significantly with
seasons or on a daily basis. Perched ground waters encountered near the
soil surface, however, do experience diurnal fluctuations in
characteristics, as well as being subject to seasonal influences. The
degree to which these variations are manifest is related in part to
topographic position. Figure 10-2 illustrates the relative rise in a
perched ground water table in response to a rainfall event. The perched
water table in the lower elevations will show a more pronounced response to
the rainfall event than those at upper elevations due to the contribution
of water from upslope areas. The confined aquifer underlying this area
does not respond to the event in the short term. The lag period during
which the changes in ground water elevations result is related to the
characteristics of the media in which the ground water occurs as well as
the intensity of the event.
Although ground water quality is subject to change over time, these changes
usually occur at a slower rate than observed in surface waters. The
quality of ground water is more closely associated with the characteristics
of its host rock or formation than with seasonal influences. Therefore,
ground water sampling programs generally do not include components to
evaluate seasonal variations. Nonetheless, time series analyses are
required in order to evaluate change as a result of contaminant
introduction to the system.
AW3D-3 10-16
-------
OSWER Di tec :lve !: 55.0-7A
^
••... >
-------
OSWER Directive 9355.0-7A
Soil - The influences of climate and weather on soil conditions are most
pronounced at the surface and become less apparent with increasing depth,
becoming negligible at depths greater than 20 feet. The rates at which
chemical compounds undergo biodegradation or volatilization will be
influenced by temperature and by the water content of the soil. Increases
in the water content of soil will result in decreases in the rate of
diffusion of gases to the soil surface and thus reduced volatilization
rates. Soil which becomes saturated may not be conducive to aerobic
degradation of compounds although anaerobic decomposition may take place.
These types of variations in soil conditions brought on by seasonal changes
are important with respect to remedial investigations. In developing data
quality objectives, efforts should be taken to ensure that the data
obtained are of value in evaluating alternatives. Volatilization of
organics from a soil sample obtained to assess the impacts of organics on
workers during excavation should be undertaken in a manner which would
simulate conditions expected during the period of planned excavation.
Samples taken during the winter should be allowed to equilibrate to a
temperature anticipated to be encountered during a summer excavation
program for instance.
In many of the examples of temporal variability which have been discussed,
contaminant concentrations vary over a yearly cycle. To measure the total
change in concentration over the period of the cycle, samples must be taken
over the course of a year. Such sampling does not fit conveniently into
the normal framework of an RI/FS. Schedule and budget constraints can
preclude the use of seasonal samples. In such cases, the occurrence of
seasonal variation should be noted and an estimate of the effects of the
seasonal variation should be made if possible.
10.3.3 SPATIAL VARIABILITY
Spatial variability describes the manner in which contaminants vary as a
function of location. Although this source of variability is normally not
considered explicitly, it is implicitly expected. It is intuitive that
soil samples taken at locations separated by 10 ft will have different
levels of contamination. The magnitude of the difference in contaminant
AW3D-3 10-18
-------
OSWER Directive 9355.0-7A
concentration of samples separated by a fixed distance is a measure of
spatial variability. The level of spatial variability is site and
contaminant specific. When spatial variability is high, a single sample is
likely to be unrepresentative of the average contaminant concentration in
the media surrounding the sample. Thus, although it is important to
recognize the nature of spatial variability at all times, it is crucial
when the properties observed in a single sample will be extrapolated to the
surrounding volume (i.e., when estimating mean site contamination).
The variability of contaminants in space tends to be a combination of two
components termed the trend component and the random component. The
variability seen in samples separated by large distances is due primarily
to the trend component. Trend describes the large scale changes in the
value of a variable over space so that a contour map is essentially a map
of trend. Superimposed on the trend is the random component of
variability. As the distance between a pair of sample locations decreases,
the importance of random variability increases.
As an example of why spatial variability contains both trend and random
components, consider the movement of contaminants in ground water. The
migration of ground water contaminants is driven primarily by the hydraulic
gradient present in the aquifer. The gradient causes the contaminants to
spread in one primary direction, thus causing a trend or gradient in
contaminant concentration. However, as individual contaminant particles
move, the possible pathways in the direction of the hydraulic gradient are
controlled by the configuration of the pore space in the aquifer. Thus,
individual particles tend to follow different complex flow paths which lead
to dispersion of the contaminant plume. Because of particle dispersion,
the contaminant concentration within the plume is much more variable than
the hydraulic gradient, which is the driving force behind the migration of
the particles. This variability introduced by particle dispersion is one
source of the random component of variability.
AW3D-3 10-19
-------
OSWER Directive 9355.0-7A
Spatial variability is a large portion of the total variability of
contaminants, so measurement of spatial variability is important in
determining DQOs. Measurement is possible through the use of
geostatistical tools such as the semi-variogram (see Section 8.0).
10.4 SAMPLE TYPES
During the DQO development process the decision maker and data users must
determine which types of samples should be obtained during the RI. The
types of samples required to characterize a site may differ from those
required to perform a pilot study. An evaluation of the intended use of
the data must be undertaken in order to ensure that the type of sample
obtained provides the necessary information to address the issues of
concern. In determining the types of samples which should be obtained the
following issues should be considered:
o Media vs. waste samples
o Grab vs. composite samples
o Filtered vs. unfiltered samples
o Random vs. non-random sampling
o Biased vs. unbiased sampling
10.4.1 MEDIA VS. WASTE SAMPLES
Media or environmental samples refer to sampling of air, water, soils, and
other environmental media to determine the extent of contamination. Waste
samples refer to the sampling of the actual wastes. Typically this will
mean drum, impoundment, tanks, or other areas where waste has been
suspected of accumulating.
AW3D-3 10-20
-------
OSWER Directive 9355.0-7A
Whether considering a site investigation (SI), remedial investigation (RI),
feasibility study (FS), or a removal activity, sampling will involve both
investigation of general environmental media and specific waste
accumulation areas. General questions regarding environmental media
include:
o Which media are contaminated? (air, water, soil, ground water,
biota)
o What is the average contamination?
o What is the total contamination? (mass, volume)
o What is the maximum contamination? (concentration)
o What area of the site is contaminated?
o What is the vertical and horizontal extent of contamination?
Each of these questions requires different sampling considerations.
Clearly, the techniques and specific data collection factors pertaining to
air will not be the same as those for ground water or any other media.
However, in general, in order to answer these or related questions the
collection information must be used on a sampling plan capable of being
used to draw conclusions regarding the general state of the environment at
the site.
Waste samples are those collected from drums, tanks, lagoons, pits, waste
piles, fresh spills, and other areas of waste accumulation. The specific
area or container being sampled differs from the media samples in two ways:
(1) the questions asked of the data and (2) the general characteristics of
the materials being sampled. The most common questions are concerned with
waste characterization:
o What contaminants are present?
o Do these contaminants exceed any criteria or standards?
AW3D-3 10-21
-------
OSWER Directive 9355.0-7A
The areas sampled differ from environmental samples in three ways: the
bounds of the contaminated area are better defined, the factors which might
affect the analysis are better understood than media testing, and waste
data tend to have greater variations than media samples.
With regard to the first point, tanks, drums, and even impoundments are
clearly defined by their boundaries. In environmental sampling this is not
the case. Always at issue in media sampling is how to define the area of
interest.
Furthermore, the analyst is better prepared for the variability in the
data. When sampling a lagoon, one knows to look for vertical and
horizontal variations in the waste. In fact, the vertical variations are
more likely. RCRA guidance for sampling impoundments and lagoons should be
carefully reviewed since these guidance documents imply that compositing
vertically within the waste area is an acceptable procedure.
Finally, waste areas can show dramatic variations when compared to
environmental sampling. In many instances, because of the high
concentrations and the large variations that are possible, composite
sampling may offer the best opportunity for characterizing the waste.
10.4.2 COMPOSITE VS. GRAB SAMPLES
Grab samples are discrete aliquots which are representative of a specific
location at a specific point in time. Composite samples represent the
mixing of a number of grab samples and typically represent an average
value. In the most common case, two or more grabs are actually added to
the same container, mixed, and then a single aliquot is taken from the
mixture. However, other forms of composite sampling might be from
radiation badges or body samples for lead readings. In both these cases,
the measurements would be over a number of hours and would not represent a
single sampling location or time.
When developing or reviewing a sampling plan, it is important to consider
the uses of grab and composite samples. Grab samples offer the most
information regarding variation of the contaminants. Grabs can be used to
AW3D-3 10-22
-------
OSWER Directive 9355.0-7A
analyze time, area/ sample collection errors and any other factors which
relate to time and point of collection. However, composite sampling offers
better estimators of average and total contamination at a site when
comparing the same number of grab samples analyzed as composite samples.
This occurs because in order to collect each composite, at least two grabs
were collected. Thus, the actual number of samples collected is at least
twice as many as a grab sample design.
In the previous discussion of waste samples, composite sampling was
recommended because this method tends to average data. The large
variations present in waste data are dampened by composite sampling, and
thus, better estimates of the amount of contaminants present can be
calculated.
10.4.3 RANDOM VS. NON-RANDOM SAMPLING
Random sampling is any method of choosing sampling locations which is based
on random chance probabilities when the probability of choosing a single
location is known. Non-random sampling is any other form of choosing
sampling locations. The purpose of random sampling is to collect
information which can be used to extrapolate or make inferences about the
general population being sampled. Non-random sampling may be chosen when
information regarding specific points is more important than general
inferences about the population at large.
There are many sampling schemes which might not appear random at first
reading, but as long as the above rules are followed, random sampling will
be achieved.
Three types of sampling typically used in analysis are:
o Simple random sampling
o Systematic sampling
o Stratified random sampling
AW3D-3 10-23
-------
OSWER Directive 9355.0-7A
How well one method reduces variation versus its cost becomes the basis of
preference. Sampling methods are always evaluated relative to their
performance versus cost.
Simple Random Sampling - This form of sampling is used when there is not
information available on the area or media being sampled which might help
the designer of the sampling protocol account for variation in the
population being sampled. Since information on the population being
sampled is necessary for calculating the sample size and for more intricate
sampling procedures, it is not uncommon for a pre-sample to be collected.
Often this is done using simple random sampling to collect a small amount
of information to make decisions regarding the type of data collection
program and the amount of data.
Systematic Sampling - Systematic sampling involves the collection of
samples at predetermined, regular intervals. A good example is soil
sampling where a grid is designed so that each point sampled is
equi-distant from its neighboring points.
Though this is a commonly employed sampling scheme, care must be taken to
avoid bias; this is no easy task. The nature of systematic sampling offers
the impression that all data are equally likely to be sampled and thus, the
possibility of introducing bias is small. Yet, this assumption is false.
For example, if there is a periodic variation in the material to be
sampled, systematic sampling may produce erroneous results.
Non-random sampling or judgment sampling can also, in some cases, be
systematic sampling. Consider the example of a treatment system which is
being tested for its efficiency- The design engineer might require that
the system be sampled at specific loadings which are based on design
performance relative to the material being treated. This type of sampling
is common when the engineer intends to model the results.
Stratified Random Sampling - Stratified sampling essentially involves the
division of the sample population into groups based on knowledge of sample
characteristics at these divisions. Stratified sampling requires more
AW3D-3 10-24
-------
OSWER Directive 9355.0-7A
planning than other types of data collection. The designer of the sample
must analyze data and background information made available from the
preliminary site survey, prior investigations conducted on site, and/or
experience with similar situations.
If applied correctly, stratified sampling can give sample designs which
have greater precision than simple random sampling for the same number of
samples. This is accomplished by selected divisions which maximize the
variation between divisions while at the same time reducing the variation
within each division. For example, if drums to be sampled at a site were
first grouped by the production line that produced the waste, then the
variation within each group might be smaller than between the production
line groups.
However, in many instances, precision may not be the only reason for
including specific sectors of the sampled media or waste area. As long as
the probability of picking any single division or strata is known, valid
estimators can be calculated from the design. The following example points
out the importance of knowing the probability of drawing a strata and the
breadth of the definition of random sampling.
A common situation is to sample a lagoon for its contents. In this example
we will assume that a visual inspection of the lagoon shows that one small
portion of the lagoon is black while the rest of the lagoon is green. An
acceptable random sample might be to divide the lagoon into equal grids the
size of the black area and randomly sample a group of grids with the
exception of the black one. The black grid is sampled with certainty,
sometimes called a judgment sample.
10.4.4 BIASED VS. UNBIASED SAMPLING
Biased sampling refers to a sampling scheme whose resulting data places
more undue emphasis on a single characteristic or factor of the problem.
Unbiased sampling refers to sampling methods which, though they may
overemphasize a particular characteristic, allow for estimates to be drawn
from the data which are representative of the population at large. These
AW3D-3 10-25
-------
OSWER Directive 9355.0-7A
tenns usually can be considered to be synonymous with random and non-random
sampling. However, any sampling program can create data which have an
inherent bias because of poor planning.
A statistician defines bias as the difference between expectation of an
estimator or procedure and the actual population characteristic being
estimated. An unbiased estimator is one whose expectation is the same as
the population characteristic being estimated; its bias is zero. In the
example above, the question which one might ask is whether the resulting
estimates will be unbiased. If the purpose of the sampling is to estimate
the total amount of a particular contaminant, then the sample is unbiased.
The analysis of the data will correct for the bias introduced by sampling
the black grid with certainty. The results of any chemical analysis are
averaged by weighting the randomly chosen samples by the total area of the
possible grids which could have been chosen, while the black grid is
weighted by its area. This process removes any bias in the estimate.
Biased sampling is undertaken when it is not possible to calculate
estimators which reflect the overall characteristics of the population.
Probably the most commonly used form of biased sampling is conducted during
the site investigation (SI) process. The purpose of the SI is to find out
whether any contamination is present. Thus, these studies are typically
conducted in ways that maximize the chance of analyzing samples which have
contamination above a particular criteria.
The use of direct reading instruments to screen samples is a good example
of such biased sampling. The samples which are finally analyzed using, for
example, GC/MS will represent higher contamination than might exist overall
at the site.
However, this type of sampling is correct for the SI. In the RI/FS, this
type of sampling might still be acceptable.in cases where design of a
treatment system might be dependent on the maximum treated load.
AW3D-3 10-26
-------
OSWER Directive 9355.0-7A
10.5 SAMPLING PATTERNS
Determining the number of data points required to address issues associated
with a hazardous waste site requires a multidisciplinary approach. Data
are acquired to address a range of concerns and answer a diverse set of
questions. Before attempting to determine how many data are required, the
purposes for collecting data must be clearly defined. For instance, at a
single site, soil samples may be taken to examine the characteristics of
the source, to identify and evaluate potential contaminant pathways, to
examine the magnitude and spatial extent of contamination, and to
investigate unsampled areas where contamination is suspected. The number
of data required to accomplish each of the above will vary. To
characterize the source, one or two composite samples may be required and
detailed statistical applications may not be appropriate in such cases.
However, when data are collected to describe the distribution of
contamination throughout the site or to investigate the possibility of
unsampled areas, statistics should be used to determine the number of data
required. Thus, in a well designed sampling plan the total number of data
collected will be the sum of data collected to investigate small, precisely
defined portions of the site, and the statistically determined number of
data required to fully characterize the site and its environs.
When acquiring data which will be used to make general inferences
concerning site characteristics, it is important that samples provide
complete coverage of the area of interest and that sample locations do not
introduce bias. Complete coverage is necessary to ensure that no areas of
contamination are missed. Bias in a data set causes the mean of the data
to be systematically different from the true mean. Bias is caused by any
systematic error in data location, such as clustering of data. When data
are clustered (located close together) some small portions of the site are
sampled more densely than the remainder of the site. The particular
contaminant value observed in the densely sampled area will be over
represented in the estimate of the sample mean. If, as is often the case,
samples are clustered in highly contaminated areas, the mean site
contaminant concentration will be overestimated.
AW3D-3 10-27
-------
OSWER Directive 9355.0-7A
Sampling patterns should be designed to minimize bias and provide complete
site coverage. The best sampling pattern for accomplishing both of these
goals is a regular grid. It can be shown theoretically (Ripley 1982), that
data taken on a regular grid will yield a more precise estimate of the mean
site contamination than data located according to any other procedure.
This fact combined with the superior coverage and non-biased property of
regular sampling make it the preferred sampling pattern when statistics
will be applied.
The use of an unbiased approach during the initial sampling phases is
recommended in order to ensure that no area of the site is overlooked in
sampling. Subsequent sampling phases should incorporate the information
resulting from the unbiased sampling which occurred during the initial
phases. The data should be used to identify areas in which additional
samples should be obtained and areas where no additional samples are
required. Introduction of bias during subsequent phases is justified in
these instances.
10.5.1 GRID SYSTEMS
Sampling grids consist of a network of crossing parallel lines which can be
used to identify locations at which samples will be obtained. Grids can be
superimposed on topographic maps, aerial photographs, site plans or other
graphical representations of the site to identify potential sampling
locations. These mapped points can then be accurately located in the field
using standard surveying techniques. Grids can also be generated in the
field from some arbitrary starting point and later tied down using
surveying techniques and transferred to a reference map.
Grid systems are used in developing systematic sampling plans in which
samples are located at consistent distances from one another. The most
elementary grid system is a straight line between two points on which
regularly spaced sampling locations are noted. This type of one-
dimensional sampling grid may be useful for sampling along a straight
AW3D-3 10-28
-------
OSWER Directive 9355.0-7A
drainage ditch or other man-made feature. The majority of environmental
sampling, however, requires a two-dimensional approach to sample location
identification.
Figure 10-3 presents a two dimensional square grid system for locating
sampling points. The grid is comprised of equidistant parallel lines at
right angles to each other. Figure 10-4 presents a two dimensional
triangular grid system comprised of equidistant parallel lines intersected
by lines drawn at 60° from vertical in both directions. Sampling generally
is undertaken at the intersection of the parallel lines which compose a
grid, although other approaches such as sampling in the center of each grid
box or obtaining a composite of samples within a grid box are also
acceptable. It may be appropriate to modify the grid system to account for
variations in concentration gradients as illustrated in Figure 8-5.
Grid systems can be applied in sampling most environmental media provided
the areal limits of the sampling media are established. The limits in
which the grid is applied are generally based upon previously obtained
knowledge of the site conditions, physical limitations, or property lines.
In the initial stages of developing a sampling plan, the grid is generally
superimposed on the site map or aerial photograph by some arbitrary means.
In order to ensure that the grid is placed on the site in an unbiased
manner, one grid point could be placed at a predetermined location (i.e, on
an existing bench mark) with grid lines oriented parallel to a north-south
line or some other predetermined orientation. Following placement of the
grid the sampling locations can be surveyed and "staked" (located) in the
field. In all cases the grid system established on a map must be field
verified. In most cases field verification may require that some of the
sampling locations be abandoned. For example, a grid system may be used to
identify surficial soil sampling locations. When field verified, some
sampling locations occur on bedrock outcrop where no soil exists. In these
cases alternate locations may be selected or the sample point eliminated.
AW3D-3 10-29
-------
SITE BOUNDARY
FIGURE 10-3
SQUARE GRID SYSTEM
[A = B]
-------
SITE BOUNDARY
FIGURE 10-4
TRIANGULAR GRID SYSTEM
[A = B = C < a = b = c = 60o]
-------
O
O
2
O
<
tr
&
^
FIGURE 10-5
MODIFIED GRID SYSTEM TO
ACCOUNT FOR DIRECTIONAL CORRELATION
-------
OSWER Directive 9355.0-7A
10.5.2 STRATIFICATION
Stratification refers to the process of locating samples within distinct
populations or strata. Commonly occurring strata are geological
formations, soil horizons, and visually different areas of contamination.
An example of visually different contamination is a lagoon containing a
zone of green liquid and a zone of black liquid. Typically the number of
samples taken within strata varies. For instance, more samples should be
taken from a visibly contaminated soil horizon than from an uncontaminated
soil horizon. By varying the number of samples in each strata based on
existing information or information obtained in the field, the sampling
program can concentrate on the most important aspects of the site.
Stratification is thus a valuable method for conserving resources.
Grid systems are generally applied to a site in the horizontal plane such
that sampling points can be identified using an X, Y; latitude, longitude;
or other coordinate system. The Z or elevation coordinate of each sample
must also be identified in sampling plans in order to fix the spatial
location of the sample. The Z component may be designated as a depth below
ground surface or relative elevation.
In situations where a determination of the areal extent of surficial soil
contamination exists at a site, it may be appropriate to obtain samples at
the surface of the soil. However, in most cases the extent of
contamination in the vertical component (i.e., depth below the soil
surface) must also be determined. Therefore, samples are generally
obtained at various depths at each X,Y location. Selection of the depth at
which samples are obtained should be based upon knowledge of the
characteristics of the environmental media under investigation. Outlined
below are a number of different situations in which the method used to
select the Z component may vary:
o Sediment samples obtained at the bottom of a lake or pond are
generally obtained at the X,Y coordinate designated on the grid
system. The Z component of the sample location is generally
determined in the field as a depth below the surface of the water.
These depths are later used to develop a bathymetric depiction of
the pond/lake bottom.
AW3D-3 10-33
-------
OSWER Directive 9355.0-7A
o Soil samples obtained over an area in which liquid contaminants have
been disposed of on the surface may be obtained at predetermined
depths to evaluate the extent of contamination. For example, at the
. surface and at 2-ft or other specified intervals extending to
bedrock or some other established depth.
o Soil samples may also be obtained at predetermined elevations where
the Z coordinate will be crucial in cleanup operations. For
instance, it may be appropriate to obtain samples at specified
elevation increments (i.e., at 2-ft increments from 100 ft msl
extending downward to 50 ft msl). This would provide consistency in
the Z coordinate plane and allow for correlation of data regardless
of changes in the surficial topography.
o Where physical conditions of the environmental media vary
significantly, the Z sampling coordinates may best be selected to
allow for characterization of the media. This is generally the most
appropriate approach for characterizing geologic deposits, in these
situations the Z location is selected to provide information on each
horizon, or strata of instance. This type of approach would allow
for selection of appropriate numbers of unconsolidated glacial till
samples as well as samples of the underlying bedrock regardless of
its orientation, for example.
The appropriate approach for stratification of samples is dependent in
large part on the objectives of the sampling program and characteristics of
the environmental media under investigation. The degree of stratification
(e.g., the vertical spacing of Z sample locations) is dependent on the
total number of samples that will be required to characterize the media and
more specifically each horizon or strata within the media under
investigation.
10.5.3 GRID SPACING
Spacings of grids are usually established to allow for sampling at each
grid intersection. These alternative sampling approaches can be used when
a low intensity investigation is used preliminary to more intensive
sampling to be performed following review of the data. For example, a grid
system may be placed over a site with grid lines spaced at 10-ft intervals.
During the preliminary investigation samples may be obtained at every tenth
intersection on every tenth grid line, thereby resulting in samples being
AW3D-3 10-34
-------
OSWER Directive 9355.0-7A
obtained at 100-ft spacings. Following review of the preliminary data,
intensive sampling may be warranted in a number of discrete locations on
the site. This intensive sampling may then be performed at the previously
established 10-ft grid intersections.
The distance between the grid lines will determine the number of
intersections and hence the number of potential sampling points within a
specified area. As the grid line spacings increase, the number of
potential sampling points will decrease for any given sampling area.
10.6 QUALITY CONTROL SAMPLES
Various types of samples may be obtained during a remedial investigation in
order to provide quality control information for interpretation of data
including:
o Background samples
o Critical samples
o Collocated and replicate samples
o Split samples
o Field and trip blanks
o Matrix spikes
In all cases QC samples must be submitted to the laboratory as blind
samples.
10.6.1 BACKGROUND SAMPLES
Inclusion of background samples in an RI sampling task must be taken into
consideration during the DQO process. Monitoring data as well as available
literature on natural background concentrations of chemicals in the area
should be collected, reviewed and/or verified to determine background
conditions. Background data should be defined as either natural or
anthropogenic chemical contamination resulting from a source or sources
other than the site undergoing assessment.
AW3D-3 10-35
-------
OSWER Directive 9355.0-7A
A background sample is one taken from media characteristic of the site but
outside the zone of contamination. At least two background samples should
be collected for each sampling event. A sampling event is a specific media
event over a specified period of time. For example, each quarterly ground
water sampling round would be considered a sampling event.
10.6.2 CRITICAL SAMPLES
Critical data points are sample locations for which valid data must be
obtained in order for the sampling event to be considered complete. An
example of a critical data point may be an upgradient well in a ground
water contamination study or any other data point considered vital to the
decision making process. Critical data points should be carefully
considered in the sampling plan design of a data collection activity and
every effort must be made to obtain valid data for these samples, in some
cases, taking critical data point samples in duplicate is appropriate.
A common problem of any sample design is the loss of data during
implementation of the design. Care must be taken to determine the set of
points for which data must be collected in order to analyze the results
accurately. The set of points which must be collected are called the
"critical points." Critical points may be defined in terms of the minimum
number of data points which must be collected and analyzed.
The identification of critical points can be based on statistical theory or
it may be based on careful review of the issues which the analysis will
address. In the lagoon sampling example, the single sample from the black
grid might be considered a critical point. The assumption in the sampling
plan was that this grid was different from the others. Therefore, while
the sampling plan might be able to yield acceptable results if a number of
the, assumed similar, green grid samples were lost, it seems unlikely that
the design would be acceptable without the black grid.
Critical data points should be identified in every completeness statement
developed during the DQO process.
AW3D-3 10-36
-------
OSWER Directive 9355.0-7A
10.6.3 COLLOCATED AND REPLICATE SAMPLES
Collocated samples are independent samples collected in such a manner that
they are equally representative of the parameter(s) of interest at a given
point in space and time. Examples of collocated samples include: samples
from two air quality analyzers sampling from a common sample manifold, two
water samples collected at essentially the same time and from the same
point in a lake, or in some cases side-by-side soil core samples.
Collocated samples, when collected, processed, and analyzed by the same
organization, provide intralaboratory precision information for the entire
measurement system including sample acquisition, homogeneity, handling,
shipping, storage, preparation and analysis. Collocated samples, when
collected, processed and analyzed by different organizations, provide
interlaboratory precision information for the entire measurement system.
Replicate samples are samples that have been divided into two or more
portions at some step in the measurement process. Each portion is then
carried through the remaining steps in the measurement process. A sample
may be replicated in the field or at different points in the analytical
process. For field replicated samples, precision information would be
gained on homogeneity (to a lesser extent than for collocated samples),
handling, shipping, storage, preparation, and analysis. For analytical
replicates, precision information would be gained on preparation and
analysis. Examples of field replicated samples include a soil core sample
that has been collected and poured into a common container for mixing
before being split and placed in individual sample containers.
Collocated samples can be used to estimate the overall precision of a data
collection activity. Sampling error can be estimated by the inclusion of
collocated and replicated versions of the same sample. If a significant
difference in precision between the two subsets is found, it may be
attributed to sampling error. As a data base on field sampling error is
accumulated, the magnitude of sampling error can be determined.
AW3D-3 10-37
-------
OSWER Directive 9355.0-7A
The use of both collocated and replicated samples in soil sampling is an
attempt to quantify the degree of error that can be attributed to the
sampling process. This approach is valid when the homogeneity of the
sample matrix is in question. Swiftly flowing streams or discharge pipes
would also fall into this category. The use of these two types of
duplicate samples and the frequency for their inclusion in an RI is
dependent on the sample matrix and the intended use of the data as
discussed in Stage 2 of the DQO process. More replicate samples are
generally taken when nonhomogeneity is expected. The inclusion of
collocated samples into a sampling program also depends on the sampling
method utilized. It may not be appropriate to collocate deep soil boring
samples, for example. In this case example, a field replicate would be
more appropriate than drilling two separate, side-by-side boreholes.
In summary, the following are suggested guidelines for the inclusion of
collocated and replicated samples in field programs:
o Ground and surface water - one out of every 20 investigative samples
should be collocated. Replicated samples could be substituted where
appropriate. These samples should be spread out over the sampling
event, preferably at least one for each day of sampling.
o Soil, sediments and solids - one out of every 20 investigative
samples should be field replicated or collocated. To estimate
sampling error, collocated and field replicated samples should be of
the same investigative sample. These samples should be spread out
over the sampling event, preferably one per each day of sampling.
10.6.4 SPLIT SAMPLES
Split samples are replicate samples divided into two portions, sent to
different laboratories, and subjected to the same environmental conditions
and steps in the measurement process. They serve as an oversight function
in assessing the analytical portion of the measurement system.
AW3D-3 10-38
-------
OSWER Directive 9355.0-7A
10.6.5 TRIP AND FIELD BLANKS
Trip blanks generally pertain to volatile organic samples only. Trip
blanks are prepared prior to the sampling event in the actual sample
containers and are kept with the investigative samples throughout the
sampling event. They are then packaged for shipment with the other samples
and sent for analysis. At no time after their preparation are the sample
containers opened before they reach the laboratory.
Field blanks are defined as samples which are obtained by running
analyte-free deionized water through sample collection equipment (bailer,
pump, auger, etc.) after decontamination, and placing it in the appropriate
sample containers for analysis. These samples will be used to determine if
decontamination procedures have been sufficient. Using the above
definition, soil field blanks could be called rinsate samples. These
should be included in a sampling program as appropriate.
The following guidelines for including blanks in sampling programs are
suggested.
o Ground and surface water - Field blanks should be submitted at the
rate of one field blank/matrix/per day or one for every 20
investigation samples, whichever results in fewer samples. Trip
blanks should be included at a frequency of one per day of sampling
or as appropriate.
o Soil sediments and solids - Rinsate samples should be submitted at
the rate of one for every 20 investigative samples for each matrix
being sampled or as appropriate. EMSL-LV is currently evaluating a
material which can be used as a soil field blank.
Guidelines for blank, duplicate, and background samples are provided in
Table 10-2. These guidelines serve as a starting point from which to
develop site-specific sampling plan QC sample numbers. In certain
instances, it may be appropriate to utilize known reference materials when
available for QC checking. The numbers and sources of reference materials
AW3D-3 10-39
-------
OSWER Directive 9355.0-7A
TABLE 10-2
GUIDELINES FOR MINIMUM QA/QC SAMPLES
FOR FIELD SAMPLING PROGRAMS
MEDIA
Aqueous
DUPLICATES
FIELD
COLLOCATED OR REPLICATE
one in twenty
FIELD
BLANK
one in
twenty
TRIP
BLANK
one per
day of
sampling
BACKGROUND
SAMPLE
min. of two
per sampling
event-media
INTER-LAB
SPLIT SAMPLE'
when required
to meet
objectives
Soil,
sediment
Air
Source
material
one in twenty
one in twenty
one in twenty
one in
twenty
not
available
not usually
required
one per
day of
sampling
min. of two
per sampling
event-media
min. of two
per sampling
event-media
when required
to meet
objectives
when required
to meet
objectives
when required
to meet
objectives
NOTE: This table is provided to serve as a guideline only; QA/QC sample requirements must be
developed on a site-specific basis. Laboratory blanks and spikes are method specific and are
not included in this table.
AW3F-15
-------
OSWER Directive 9355.0-7A
which would provide meaningful comparison and checks for media obtained
v
from hazardous waste sites are limited. It is advisable to consult with
analytical chemists regarding the appropriateness of use of reference
materials as a QC check.
10.6.6 MATRIX SPIKES
Many samples exhibit matrix effects, in which other sample components
interfere with the analysis of contaminants of interest. Matrix spikes
provide the best measurement of this effect. When done in the field,
immediately after collection, they also provide a measurement of sampling,
handling and preservation error. The field matrix spike does provide the
best overall assessment of accuracy for the entire measurement system, as
collocated samples do for precision assessment. However, there are some
serious issues regarding the field spiking of environmental samples that
must be considered. Field matrix spikes are generally not recommended
because of the high level of technical expertise required for proper use
and their sensitivity to environmental variables.
The major problems associated with field matrix spikes are due to the fact
that all spike recovery data must be interpreted very carefully. Spike
recoveries are subject to many competing factors, such as analyte
stability, holding time, and the sample matrix. Because of the inherent
variability associated with spike recoveries, the additional variability
introduced by spiking samples in the field can increase the overall
uncertainty associated with a data set rather than decrease it.
The two most important issues to address when considering field spiking as
an option are the source of the spiking material and the technical
capability of the person doing the spiking. Spiking materials that can be
used are Standard Reference Materials (SPMs), EPA quality control ampules,
or laboratory-prepared solutions made from pure compounds. SRMs are
stand-alone standards prepared by NBS that can be placed in the appropriate
sample containers and sent to the laboratory to be analyzed. The use of
certified standards such as SRMs solves the "traceability" issue concerning
the integrity of the blind standard and also does not require a skilled
AW3D-3 10-41
-------
OSWER Directive 9355.0-7A
technician to prepare the standard. However, because the SRM is a
stand-alone sample, it provides no information on the impact of the sample
matrix on the measurement system. An aliquot of an SRM can be used to
spike an environmental sample, but it would no longer be traceable and
would require a person skilled in the appropriate analytical techniques,
just as the use of quality control ampoules or laboratory-prepared spikes
do. the competence of the person doing the spiking is critical. The exact
amount of spiking material must be recorded for future use in assessing
recoveries. Errors in measurement of the spike or use of the wrong spiking
material will cause serious problems in interpreting the usability of the
data.
In summary, field matrix spikes are not recommended unless the appropriate
technical support is available. Absolute attention to all details is
required to obtain useful information from the procedure. If field matrix
spikes are used, the results should be compared with laboratory matrix
spike results.
10.7 REFERENCES
American Society of Agronomy. 1965. Methods of Soil Analyses Part 1 -
Physical and Mineralogical Properties Including Statistics of Measurement
and Sampling. C.A. Blacked. Agronomy Monograph No. 9. Madison,
Wisconsin.
Camp Dresser & McKee Inc. 1985. Site Investigation Procedures Manual -
Performance of Remedial Response Activities at Uncontrolled Hazardous
Waste Sites. Volumes I, II, III, IV.
Charlie, W.A., R.E. Wardwell, and O.B. Andersland. 1979. Leachate
Generation from sludge disposal area. ASCE Journal of Environmental
Engineering Division Vol. 10S EES:947-960.
Claassen, H.C. 1982. Guidelines and Techniques for Obtaining Water Samples
that Accurately Represent the Water Chemistry of an Aquifer. USGS Open
File Report 82-1024. Lakewood, CO.
Clay, P.F. and T.M. Spittler, 1982. The Use of Portable Instruments in
Hazardous Waste Site Characterizations, Proceedings of the Third National
Conference on Management of Uncontrolled Hazardous Waste Sites,
Washington, D.C.
Ecology and Environment, Inc. 1982. FIT Operations and Field Manual, HNu
Systems PI 101 Photoionization Detector and Century Systems Model OVA-128
Organic Vapor Analyzer, prepared for U.S. EPA, Washington, D.C.
AW3D-3 10-42
-------
OSWER Directive 9355.0-7A
Elans, R.B. 1986. Groundwater Monitoring Data Quality Objectives for
Remedial Site investigations. In: Quality Control in Remedial Site
Investigations: Hazardous and Industrial Solid Waste Testing Fifth
Volume ASTM STP 925.
Fuller, W.H., A. Amoozeyar-Fard, E.F. Niebla, as M. Boyle. 1980. Influence
of Leachate Quality on Soil Attenuation of Metals. In: Disposal of
Hazardous Wastes pp 108-117. EPA-600-9-80-010.
IT Corporation. 1984. A Sampling Strategy for Remedial Action at Hazardous
Waste Sites: Cleanup of Soil Contaminated by
Tetrachlorodibenzo-p-dioxin. Draft No.2. August 20, 1984
Jacot, Brian. 1983. OVA Field Screening at a Hazardous Waste Site.
Proceedings of the Fourth National Conference on Management of
Uncontrolled Hazardous Waste Sites. Washington, D.C.
Lockheed Engineering and Management Series Co., Inc. 1984.
Characterization of Hazardous Waste Sites, A Methods Manual. Volume 3 -
Available Laboratory Analytical Methods. May.
Mehran, M. et al. 1983. Delineation of Underground Hydrocarbon Leaks by
Organic Vapor Detection. Proceedings of the Fourth National Conference on
Management of Uncontrolled Hazardous Waste Sites. Washington, D.C.
Panaro, J.M. 1984. Air Monitoring and Data Interpretation During Remedial
Action at a Hazardous Waste Site. Proceedings of the Hazardous Waste and
Environmental Emergencies Conference. Houston, Texas.
Quimby, J.M. et al. 1982 Evaluation and Use of a Portable Gas Chromatograph
for Monitoring Hazardous Waste Sites. Proceedings of the Third National
Conference on Management of Uncontrolled Hazardous Waste Sites.
. Washington, D.C.
Raveh, A. and Y. Avnimelech. 1979. Leaching of pollutants from sanitary
landfill models. Journal WPCF 51(11):2705-2716.
Spittler, T.M. 1980. Use of Portable Organic Vapor Detectors for Hazardous
Waste Site Investigations. Second Oil and Hazardous Materials Spill
Conference and Exhibition. Philadelphia, Pennsylvania.
Spittler, T.M. et al. 1981. Ambient Monitoring for Specific Volatile
Organics Using a Sensitive Portable PID GC. Proceedings of the Second
National Conference on Management of Uncontrolled Hazardous Waste Sites.
Washington, D.C.
Spittler, T.M. 1983. Field Measurement of PCBs in Soil and Sediment Using a
Portable Gas Chromatograph. Proceedings of the Fourth National Conference
on Management of Uncontrolled Hazardous Waste Sites. Washington, D.C.
U.S. EPA. 1977. Procedures Manual for Groundwater Monitoring at Solid Waste
Disposal Facilities. EPA 530/SW-611.
AW3D-3 10-43
-------
OSWER Directive 9355.0-7A
U.S. EPA. 1980. Environmental Monitoring and Support Laboratory,
Cincinnati, Ohio, Interim Methods for the Sampling and Analysis of
Priority Pollutants in Sediments and Fish Tissue. October.
. 1981. EPA/DOT: Hazardous Waste Transportation Interface.
. 1981. Soil and Sediment Sampling Methods. Technical Methods for
Investigating Sites Containing Hazardous Substances. Technical Monograph
No. 17.
. Handbook for Sampling and Sample Preservation of Water and
Wastewater. EPA 600/4-82-029.
. 1983. Characterization of Hazardous Waste Sites—A Methods Manual,
Volume II, Available Sampling Methods. NTIS PB84-126929.
. 1983. Preparation of Soil Sampling Protocol: Techniques and
Strategies NTIS PB83-20-6979 EPA 600/4-83-020.
. 1984. Soil Sampling Quality Assurance User's Guide. EPA
~~B15u/4-84-043
. 1984. Vegetation Sampling: Techniques and Strategies LIP 83-159.
. 1984. Quality Assurance Handbook for Air Pollution Measurement
Systems. Volumes 1 and 2, EPA - 600/9-76-005. January 19.
. 1984. Standard Operating Safety Guides. Office of Emergency and
Remedial Response.
. 1985. Sediment Sampling Quality Assurance User's Guide.
. 1985. Characterization of Hazardous Waste Sites - A Methods Manual.
Volume 1 - Site Investigation. EPA/600/4-84/075.
. 1985. Field Standard Operating Procedures No.4: Site Entry. Office
of Emergency and Remedial Response.
. 1985. Field Standard Operating Procedures No.6: Work Zones, Office
of Emergency and Remedial Reponse.
. 1985. Field Standard Operating Procedures No.8: Air Surveillance.
Office of Emergency and Remedial Response.
. 1985. Field Standard Operating Procedures No.9: Preparation of a
Site Safety Plan. Office of Emergency and Remedial Response.
. 1985. Technical Background Document to Support Rulemaking Pursuant
to CERCLA Section 102. Volume 1 Contract No. 68-03-3182.
. 1986. Groundwater Monitoring Quality Assurance for RCRA April 1986
Second Draft.
AW3D-3 10-44
-------
Appendix A
-------
OSWER Directive 9355.0-7A
APPENDIX A
REVIEW OF QAMS DQO CHECKLIST
-------
OSWER Directive 9355.0-7A
APPENDIX A
REVIEW OF QAMS DQO CHECKLIST
In a memorandum dated April 3, 1983, Mr. Stanley Blacker, Director of the
Quality Assurance Management Staff (QAMS) issued a checklist to be used by
QAMS staff during their review of DQOs. The purpose of this appendix is to
review the QAMS checklist with respect to this RI/FS DQO guidance.
The QAMS checklist is designed for use in reviewing specific DQOs rather
than an approach to DQO development for a complex process such as an RI/FS.
This appendix presents a review of the checklist items, along with a
reference to the section where the item is addressed and/or a comment
regarding the applicability of the item to the RI/FS DQO process.
The RI/FS process involves multiple levels of data and data uses, and
culminates in a decision regarding the degree of remedial response to be
implemented for a site. Decisions are based on analytical and other
measurement data which are often integrated to interpret various aspects of
a site's characteristics. Thus, many different sets of DQOs may be
required for a given RI/FS.
AW3F-10 A-l
-------
OSWER Directive 9355.0-7A
APPENDIX A
SUMMARY OF DQO CHECKLIST ITEMS WITH RESPECT TO
RI/FS DQO APPLICABILITY
DQO CHECKLIST ITEM
A-l. The decision maker and
associated users are clearly
identified.
COMMENT RE; RI/FS DQO APPLICABIL3
The key RI/FS decision is remedy
selection (i.e., ROD/EDO
signature). For the majority of
RI/FS projects, remedy selection is
delegated to the Regional
Administrator (RA). Program
Management responsibilities are
delegated to the Waste Management
Division Director and Managers,
with project specific management
and oversight assigned to Remedial
Project Managers (RPM). In this
role the regional EPA RPM, is
responsible for coordinating the
DQO development process, and
overseeing remedial contractors,
state officials, or private parties
conducting the RI/FS. Associated
data users include primary,
secondary and technical support and
project review/audit personnel.
A-2. The decision maker and
associated data users have been
involved in the development of
DQOs.
B-la. A statement of the
decision(s) that depend(s) on
the results of this data
collection activity.
See Section 2.0, Stage 1 - Identify
& Involve Data Users.
The decision(s) that result from
the RI/FS process involve multiple
levels of data for multiple
purposes. See Section 3.0, Stage 1
- Specify RI/FS Objectives.
B-lb. If the data collection
activity is of an exploratory
nature and not formally linked
with a regulatory decision, then
the document should include a
clear explanation of the purpose
for which the environmental data
are intended.
See Section 4.0, Stage 2 Identify
Data Uses/Needs.
AW3F-10
A-2
-------
OSWER Directive 9355.0-7A
APPENDIX A
SUMMARY OF DQO CHECKLIST ITEMS WITH RESPECT TO
RI/FS DQO APPLICABILITY (continued)
DQO CHECKLIST ITEM
B-2. Statements of each specific
question that will be addressed
in the data collection activity
and the type of conclusion that
is anticipated as an appropriate
answer to each question. The
conclusions should depend only on
measurement data.
B-3. A clear statement of the
way in which each conclusion of
the study will be represented, in
terms of the results of
statistical calculations made
with the data.
B-4. Statements of the
acceptable levels of precision
and accuracy associated with each
of the conclusions depend on
measurement data.
B-5. A definition of the
population to which each of the
conclusions apply, including
definitions of all subpopulations
or strata.
B-6. Definitions of the
variables that will be measured.
B-7. The acceptable levels of
precision and accuracy for the
measurements to be made.
B-8. A flow chart or spread
sheet illustrating the
relationship between the measure-
ment data and each conclusion
that will be made with the data.
COMMENT RE; RI/FS DQO APPLICABILITY
See Section 4.0, Stage 2
See Section 4.0, Stage 2 - The
conclusions of an RI/FS study are
highly interdependent. The format
for data presentation will vary,
based upon data quantity. A
statistical approach may not be
feasible.
See Section 4.0, Stage 2.
See Section 4.0, Stage 2.
See Section 4.0, Stage 2.
See Section 4.0, Stage 2.
See Section 4.0, Stage 2.
AW3F-10
A-3
-------
Appendix B
-------
OSWER Directive 9355.0-7A
APPENDIX B
POTENTIALLY APPLICABLE OR
RELEVANT AND APPROPRIATE REQUIREMENTS
excerpt from National Contingency Plan final rule
Federal Register, Vol. 50, No. 224
November 20, 1985
-------
Federal Register / Vol. 30. No. 224 / Wednesday. November 20. 1985 / Rules and Regulations
Potentially Applicable or Relevant and
Appropriate Requirement*
;. EPA '* Office of Solid Waste
administers, inter alia, the Resource
Conservation and Recovery Act of 1978,
as amended (Pub, L 94-S8Q. SO Slat 95.
42 US.C 8X1 et seq.). Potentially
applicable or relevant requirements
pursuant to that Act are:
a. Open Dump Criteria—Pursuant to
RCRA Subtitle 0 criteria for
classification of solid waste disposal
facilities (40 CFR Part 257).
Not*.—Only rtltvint to noiihizardous
wtitit. ' ' .
b. In most situations Superfund
wastes will be handled in accordance
with RCRA Subtitle C requirements
governing standard* for owners and
operators of hazardous waste treatment
storage, and disposal fadl'Her 4OCFR
Part 284. for permitted facilities, and 40
CFR Put 285. for interim status
facilities. "-.
• Ground Water Protection (40 CFR
264.90-2d4.109).
• Ground Water Monitoring (40* CFR
265.90-263.94).
• Closure and Post Qosurt (40 CFR
110-284.120. 285.110-285.112).
• Containers (40 CFR 281170-284.178.
265.170-285.177).
• Tanks (40 CFR 284.190-284.3XX
285.190-265.199).
• Surface Impoundments (40 CFR
.. ..220-284.249. 28S.220-28SJOO).
• Waste Piles (40 CFR 284.250-
264.289.283.230-265.238).
• Land Treatment (40 CFR 284.270-
21.299. 265.270-265.282).
• Landfills (40 CFR 284.300-284.339.
265.300-265.316).
• Incinerators (40 CFR 284.340-
264.999.253J40-285J69).
• Dioxin-containir.g Wastes (50 FR
1978;. Includes the fisal rule for the
listing of dioxin containing waste.
2. EPA's Off in of Water administer*
several potentially applicable or
relevant and appropriate statutes and
regulations issued thereunder
a. Section 14.2 of the Public Health
Service Act as amended by the Safe
Drinking Water Act as amended (Pub. L.
< 23. 88 Stat 1680.42 U.S.C. 300f et .
sec..)
• Maximum Contaminant Levels (for
all sources of drinking water exposure).
; CFR 141.11-141.19).
• Underground Injection Control
Regulations. (40 CFR Parts 144.145.148,
and 147).
b. Clean Water Act as amended (Pub.
L 92-500.88 Stat 816.33 U.S.C 1251 •/.
tea.)
• Requirements established pursuant
to sections 301. 302. 303 (including State
water quality standards). 306.307.
(including Federal pretreatment
requirements for discharge into a
publicly owned treatment works), and
403 of the Clean Water Act. (40 CFR
Parts 131, 400-469).
c. Marine Protection. Research, and
Sanctuaries Act (33 U.S.C. 1401).
• Incineration at sea requirements.
(40 CFR Parts 220-225. 227.228. See also
40 CFR 125.120-125.124).
3. EPA's Office of Pesticides cad Toxic
Substances
Toxic Substances Control Act (M
U.S.C.2601). . , • .
• PCS Requirements Generally: 40
CFR Part 761; Manufacturing Processing.
Distribution in Commerce, and Use of
PCBs and PCB Items (40 CFR 761.20-
761.30): Markings af PCBs and FCB
Items (40 CFR 761.40-761.45): Storage
and Disposal (40 CFR 761^0-761.79).
Records and Reports (40 CFR 761.180-
761.185). See also 40 CFR 129.105. 750.
• Disposal of Waste Materiel
Containing TCOD. (40 CFR Parts
775.180-775.197).
4. £T.l '* Office of External Affein
• Section 404(b)(l) Guidelines for
Specification of Disposal Sites for
tJredged or Fill Material (40 CFR Part
230).
• Procedures for denial or Restriction
of Disposal Sites for Dredged Material
(S 404(c) Procedures. 40 CFR Part 231).
3. EPA's Office of Air and Radiation
administers several potentially
applicable or relevant and appropriate
statutes and regulations issued
thereunder
a. The Uranium Mill Tailings
Radiation Control Act of 1978 (42 U.S.C
2022).
• Uranium mill tailing rules—Health
and Environmental Protection Standards
for Uranium and Thorium Mill Tailings
,-13 CFR Part 192).
b. Clean Air Act (42 U.S.C 7401).
• National Ambient Air Quality
StardaHs for total suspended
particulates (40 CFR Parts 50.6-50.7).
• National Ambient Air Quality
Standards for otone (40 CFR 50.9).
• Standards for Protection Against
Radiation—high and low level
radioactive waste rule. (10 CFR Part 20).
See also 1C CFR Part* 1C. 40.60.61.72.
Ma 981.
• National Emission Standard for
Hazardous Air Pollutants for Asbestos.
(40 CFR 81.140-81.158). See also 40 CFR
427.110-427.118.783,
• National Emission Standard for
Hazardous Air Pollutants for
Radionuclides (40 CFR Part 61.10 CFR
20.l01-20.10a).
6. Other Federal Requirements.
a. OSHA requirements for workers
engaged in response activities are
codified under the Occupational Safety
and Health Act of 1970 (29 U.S.C 651).
The relevant regulatory requirements
are included under
• Occupational Safety and Health
Standards (General Industry Standards]
(29 CFR Part 1910).
• The Safety and Health Standards
for Federal Service Contracts (29 CFR
Part 1926).
• The Shipyard aad Longshore
Standards (29 CFR Parts 1915.1918).
• Recordkeeping. reporting, and
related regulations (29 CFR Pan 1904).
b. Historic Sites. Buildings, and
Antiquities Act (18 U.S.C 461).
c. National Historic Preservation Act
16 U.S.C 470. Compliance with NEPA
required pursuant to 7 CFR Part 650.
Protection of Archaeological Resources:
Uniform Regulations—Department of
Defense (32 CFR Part 229.229.4).
Department of the Interior (43 CFR Part
7.7.4).
d. O.OT. Rules for the Transportation
of Hazardous Materials. 49 CFR Parts
107.17l.l-171.500. Regulation of
activities in or affecting waters of the
United States pursuant to 33 CFR Parts
320-329. The following requirements are
also trisgsred by Fund-financed actions:
• Endangered Species Act of 1973.18
U.S.C 1531. (Generally. 50 CFR Parts 81.
225.402). Wild and Scenic Rivers Act 16
U.S.C 1271.
• Fish and Wildlife Coordination Act
18 U.S.C 681 note.
• Fish and Wildlife Improvement Act
of 1978. and Fish and Wildlife Act of
1958.16 U.S.C. 742a note.
• Fish and Wildlife Conservation Act
of 1980.16 U.S.C. 2901. (Generally. SO
CFR Part 83).
• Coastal Zone Management Act of
1972.16 U.S.C 1451. (Generally. IS CFR
Part ?30 and IS CFR 323.45 for Air ir.d
Water Pollution Control Requirements).
Other Federal Criteria. Advisories.
Guidance, aad Slate Standards To Be
Considered
;. Federal Criteria. Advisories and
Procedures
• Health Effects Assessments (HEAs).
• Recommended Maximum
Concentration Limits (RMCLa).
• Federal Water Quality Criteria
(1978, 1980,1984). Note: Federal Water
Quality Criteria an not legelly
enforceable. State water quality
standards are legally enforceable, and
an developed using appropriate aspects
-------
of Federal Water Quality Criteria. In
many cases. State water quality
standards do not include specific
numerical limitations on a large number
of priority pollutants. When neither
State standards nor MCL* exist for a
given pollutant Federal Water Quality
Criteria are pertinent and therefore are
to be considered.
• Pesticide registrations.
• Pesticide and food additive
tolerances and action levels. Note:
Germane portions of tolerances and
action levels may b« pertinent and
therefore are to b« considered in certain
situations. .. • - .
• Waste load allocation procedures.
EPA Office of Water. ..
• Federal sole source aquifer
requirements.
• Public health basis for the decision
to list pollutants as hazardous under
section 112 of the Clean Air Act...:
• EPA's Ground-water Protection ;. -•
Strategy. -. . •; • .. . . -.
• New Source Performance Standards
for Storage Vessels for Petroleum '-.
Liquids. . . .. .— • .-
• TSCA health data, .. --.
• Pesticide registration data.
• TSCA chemical advisories (2 or 3
issued to date). . ~ . '
• Advisories issued by FWS and
NWFS under the Fish end WUdltft
Coordination Act
• Executive Order* related to
Floodplains (11988) and Wetlands
(11990) as implemented by EPA's August
6.1985. Policy on Floodplains and .
Wetlands Assessment* for CERCLA
Actions.
• TSCA Compliance Program Policy.
• OSHA health and safety standards
that may be used to protect public
health (non-workplace).
• Health Advisories, EPA Office of
Water.
2. Stain Standards
• S:ate Requirements on Disposal and
Transport of Radioactive wastes.
• State Approval of Water Supply
System Additions or Developments.
• State Ground Water Withdrawal
Approvals. .
• Requirements of authorized
(Subtitle C of RCRA) State hazardous
waste programs.
• State Implementation Plans and
Delegated Programs Under Clean Air
Act
• All other State requirements, not
delegated through EPA authority.
• Appioved State NPDES programs
under the Clean Water Act
• Approved State UIC programs
under the Safe Drinking Water Act.
Note: Many other State and local
requirements co Jd be pertinent.
Forthcoming guidance w.ll include a
more comprehensive list
3. USEPARCRA Guidance Documents
• Draft Alternate Concentration
Limits (ACL) Guidance.
A. EPA's RCRA Design Guidelines
1. Surface Impoundments. Liners
Systems. Final Cover and Freeboard
Control.
2. Waste Pile Design—Liner Systems.
3. Land Treatment Units.
4. Landfill Design—Liner Systems and
Final Cover.
B. Permitting Guidance Manuals
1. Permit Applicant'* Guidance
Manual for Hazardous Waste Land
Treatment Storage, and Disposal
Facilities.
2. Permit Writer's Guidance Manual
for Hazardous Waste Land Treatment
Storage, and Disposal Facilities.
3. Permit Writer's Guidance Manual
for Subpart F.
4. Permit Applicant's Guidance
Manual for the General Facility
Standards.
3. Waste Analysis Plan Guidance
Manual
8. Permit Writer's Guidance Manual
for Hazardous Waste Tank*,
7. Model Permit Application for
Existing Incinerators.
8. Guidance Manual for Evaluating
Permit Applications for the Operation of
Hazardous Waste Incinerator Units.
9. A guide for Preparing. RCRA Permit
Applications for Existing Storage
Facilities.
10. Guidance Manual on Closure tnd
Post-Closure Interim Status Standards.
C. Technical Resource Documents
(TRDs)
(1) Evaluating Cover Systems for Solid
and Hazardous Waste.
(2) Hydrologic Simulation of Solid
Waste Disposal Sites.
(3) Landf:!! and Surface Impoundment
Performance Evaluation.
(4) Lining of Water Impoundment and
Disposal Facilities.
(3) Management of Hazardous Waste
Leachate.
(8) Guide to the Disposal of
Chemically Stabilized and Solidified
Waste.
(7) Closure of Hazardous Waste
Surface Impoundments.
(3) Hazardous Waste Land Treatment
(9) Soil Properties. Classification, and
Hydraulic Conductivity Testing.
D. Test Methods for Evaluating Solid
Waste '
(1) Solid Waste Leaching Proced...
Manual.
(2) Methods for the Prediction of
Leachate Plume Migration and Mixing.
(3) Hydrologic Evaluation of Landfill
Performance (HELP) Model Hydrolcgic
Simulation on Solid Waste Disposal
Sites.
(4) Procedures for Modeling Flew
Through Clay Linen to Determine
Required Liner Thickness.
(3) Test Methods for Evaluating Solid
Wastes.
(6) A Method for Determining the
Compatibility of Hazardous Wastes.
(7) Guidance Manual on Hazardous
Waste Compatibility.
4. USEPA Office of Water Guidance
Documents
A. Pretreatment Guidance Docu&.»..;s
% - •
(1) 3M(g) Guidance Document Revised
Pretreatment Guidelines (3 Volumes)
B. Water Quality Guidance Doc_.« '
(1) Ecological Evaluation of Propo—1
Discharge of Dredged Material into
Ocean Waters (1977)
(2) Technical Support Manual:
Waterbody Surveys and Assessmer
for Conducting Use Attainability
Analyses (1983) -
(3) Water-Related Environmental Fate
of 129 Priority Pollutants (1979)
(4) Water Quality Standards
Handbook (1983)
(3) Technical Support Document for
Water Quality-based Toxics Co..-ol.
C. NPDES Guidance Documents
(1) NPDES Best Management Practices
Guidance Manual (June 1981)
(2) Case studies on toxicity reduction
evaluation (May 1983).
D. Ground Water/UIC Guidance
Document
(1) Designation of a USDW
(2) Elements of Aquifer Identification
(3) Interim guidance foi public
participation
(4) Definition of major facilities
(5) Corrective action requirements
16) Requirements applicable to wells
injecting into, through or above an
aquifer which has been exempted
pursuant to |148.104('o)(4).
(7) Guidance for U!C implementation
on Indian lands.
3. USEPA Manuals from the Office of
Research and Development
(1) EW M8 methods—laboratory
analytic methods.
(2) Lab protocols developed pursuant
to Clean Water Act I 304(h).
-------
Appendix C
-------
APPENDIX C
HISTORICAL PRECISION AND ACCURACY DATA
CLASSIFIED BY MEDIA AND BY ANALYTICAL LEVELS
AW3F-32
-------
APPENDIX C CONTENTS
HISTORICAL PRECISION AND ACCURACY TABLES
Introduction
Table C-l-C
Table C-l-D
Water: Level III
Water: Level IV
Table C-2-A
Table C-2-B
Table C-2-C
Table C-2-D
Soil: Level I
Soil: Level II
Soil: Level III
Soil: Level IV
Table C-3-A
Table C-3-B
Table C-3-C
Air: Level I
Air: Level II
Air: Level III
Table C-4-C
Other Media: Level III
AW3F-32
-------
CWSER Directive 9355.0-7A
INTRODUCTION
The data in this Appendix have been compiled to assist the reader in
selecting an analytical method appropriate for each data use. The methods
are classified by media and by analtyical levels defined as follows:
o Level I - field screening or analysis using portable instruments.
Results are often not compound specific and not quantitative but
results are available in real-time.
o Level II - field analysis using more sophisticated portable
analytical instruments; in some cases, the instruments may be set up
in a mobile or onsite laboratory. There is a wide range in the
quality of data that can be generated. Quality depends on the use
of suitable calibration standards, reference materials, and sample
preparation equipment; and the training of the operator. Results
are available in real-time or several hours.
o Level III - all analyses performed in an offsite analytical
laboratory using standard, documented procedures. The laboratory
may or may not be a CLP laboratory.
o Level IV - CLP routine analytical services (RAS). All analyses are
performed in an offsite CLP analytical laboratory following CLP
protocols.
Precision and accuracy data are presented in tabular fashion. Footnotes to
each table cite the sources of the data and the concentration or
concentration range at which the precision and accuracy were determined.
When no concentration is cited no concentration information was available
in the source material.
Precision is a measure of the variability in repeated measurements of the
same sample compared to the average value. Precision is reported as %
Relative Standard Deviation (RSD). The lower the % RSD, the more precise
the data.
AW3D-8
-------
CWSER Directive 9355.0-7A
RSD is calculated for a pair of replicates using the following formula:
%RSD - [ 2 (X.J-X2 |/(X1+X2) ] (100//2)
where X* is measurement #1 of a replicate
X- is measurement #2 of a replicate
Accuracy is reported as % Bias; as % Bias approaches zero, accuracy
increases. Bias is calculated by the following formula:
% Bias = X-Y (100)
Y
where Y is the known concentration or true value
X is the reported concentration
Bias measures the systematic error within an analytical technique.
AW3D-8
-------
TABLE C-1-C: HISTORICAL PRECISION AND ACCURACY DATA/WATER a
LEVEL III ANALYTICAL TECHNIQUES - METHODS OTHER THAN CLP RAS METHODS
ANALYTES
BROMODICHLOROMETHANE
BROMOFORM
METHOD CONCENTRATION
(TECHNIQUE) RANGE
624
(GC/MS)
8240
(GC/MS)
624
(GC/MS)
501.1
(PURGE & TRAP GC/MS)
501.2
(EXTRACTION GC/MS)
624
(GC/MS)
501.1
(PURGE & TRAP GC/MS)
501.2
(EXTRACTION GC/MS)
11 ug/l
480 ug/l
5-100 ug/l
8 ug/l
480 ug/l
0.9 ug/l
550 ug/l
1.8 ug/l
170 ug/l
9 ug/l
400 ug/l
4.8 ug/l
550 ug/l
6 ug/l
170 ug/l
PRECISION
% RSD
16
21
21
28
18
66
34
61
23
32
30
44
41
14
15
ACCURACY
% BIAS
0
-16
12
-8.8
-6.7
0
-3.8
33
-19
-23
10
-27
7.5
-23
1.8
-------
TABLE C-1-C: HISTORICAL PRECISION AND ACCURACY DATA/WATER a
(continued)
LEVEL III ANALYTICAL TECHNIQUES - METHODS OTHER THAN CLP RAS METHODS
ANALYTES
CHLOROFORM
DIBROMOCHLOROMETHANE
DIOXIN
METHOD CONCENTRATION
(TECHNIQUE) RANGE
624
(GC/MS)
501.1
(PURGE & TRAP GC/MS)
501.2
(EXTRACTION GC/MS)
624
(GC/MS)
501.1
(PURGE & TRAP GC/MS)
501.2
(EXTRACTION GC/MS)
613
(GC/MS)
4.5 ug/l
300 ug/l
0.9 ug/l
550 ug/l
1.8 ug/l
170 ug/l
8.1 ug/l
360 ug/l
0.8 ug/l
550 ug/l
1.8 ug/l
170 ug/l
21 ng/l
202 ng/l
PRECISION
% RSD
31
14
64
14
68
26
13
19
35
36
37
13
25
21
ACCURACY
% BIAS
2.2
-0.6
44
-0.02
-39
-1.2
-3.1
10
-12.5
4.7
0
0.02
N.A.
N.A.
-------
• 'A 3. E C-1 -C: HIS O: CAL PRECISION AND ACC. RACY DA 'A/WATER a
(continued)
LEVEL III ANALYTICAL TECHNIQUES - METHODS OTHFR THAN CLP RAS METHODS
ANALYTES
METHYLENE CHLORIDE
TOLUENE
TRICHLOROETHENE
LEAD
METHOD
(TECHNIQUE!
624
(GC/MS)
624
(GC/MS)
8240
(GC/MS)
624
(GC/MS)
8240
(GC/MS)
200.7
(ICP)
239.1
(FLAME AA)
239.2
(FURNACE AA)
CONCENTRATION
RANGE
7.2 ug/l
480 ug/l
13.5 ug/l
600 ug/l
25 ug/l
75 ug/l
5.4 ug/l
360 ug/l
25 ug/l
75 ug/l
42 ug/l
47.7 ug/l
12 ug/l
105 ug/l
10 ug/l
234 ug/l
PRECISION
% RSD
78
52
19
31
19
48
39
24
34
5
5.9
6.7
53
19
ACCURANCY
% BIAS
-17
-25
15
-14
-10
44
-2.3
5
31
4.4
17
-1.9
-22
-3.1
a. Source: Draft Compendium of Information and Performance Data on Routinely Used Measurement Methods (RUMM) - Pilot Phase,
RTI/3087/03, prepared for EPA Quality Assurance Management Staff, January 1986. This document should be
consulted for more information on individual analytes.
-------
Document No. 9355.0-7A
TABLE C-l-C: HISTORICAL PRECISION AND ACCURACY DATA/WATER
(Continued)
LEVEL III ANALYTICAL TECHNIQUES - SW-846 METHODS
Method
Number
ORGANICS;
8010
8020
8030
8040
8060
8080
8090
8100
8120
8140
8150
8240
8250
8040
Method Name
•
•
Halogenated Volatile Organics
Aromatic Volatile Oranics
Acrolein, Acrylonitrile,
Acetonitrile
Phenols
Esters
Organochlorine Pesticides
and PCBs
Nitroaromatics and Cyclic
Ketones
Polynuclear Aromatic
Hydrocarbons
Chlorinated Hydrocarbons
Organophosphorous Pesticides
Chlorinated Herbicides
Volatile Organics
GC/MS Semivolatiles (Packed
Column)
GC/MS Semivolatiles
(Capillary)
Data
Source
SW 846
SW 846
SW 846
SW 846
EPA 606
SW 846
SW 846
SW 846
SW 846
SW 846
SW 846
Range of
Recovery (%)
75.1 - 106.1
77.0 - 120
96 - 107
41 - 86
82 - 94
86 - 97
63-71
NAb
76 - 99
56.5 - 120.7
NA
95 - 107
41 - 143
NA
Precision
(%)
2.0
9.4
5.6
7.9
1.3
1.3
3.1
NA
10
5.3
NA
9 -
20
NA
- 25.1
- 27.7
- 11.6
- 16.5
- 6.5
- 6.5
- 5.9
- 25
- 19.9
28
-145
MDL
(mg/1)
0.03 - 0.52
0.2 - 0.4
0.5 - 0.6
058 - 2.2
0.29 - 3.0
0.29 - 3.0
0.06/ND
NA
0.03 - 1.34
0.1 - 5.0
0.1 - 200
1.6 - 6.9
0.9 - 44
NA
-------
Document No. 9355.0-7A
TABLE C-l-C: HISTORICAL PRECISION AND ACCURACY DATA/WATER
(Continued)
LEVEL III ANALYTICAL TECHNIQUES - SW-846 METHODS
Method
Number
Method Name
Data
Source
Range of
Recovery (%)
Precision
(%)
MDL
(mg/1)
8310
INORGANICS;
7000 Series
7470
9010
9030
Polynuclear Aromalic SW 846 78 - 116
Hydrocarbons (HPLC)
(Capillary)
Metals (ICAP) EPA 200.7 NA
Metals (FLAME) 7000 Series EPA 200 NA
Metals (FLAME LESS/GF) EPA 200 NA
Metals (MERCURY) EPA 245.2 87 - 125
Cyanides EPA 335.2 85 - 102
Sulfides EPA 376.1 NA
7.3 - 12.9
3 - 21.9 (RSD)
NA
NA
0.9 - 4.0
0.2 - 15.2
NA
0.03 - 2.3
1.3 - 75 Mg/1
0.01 - 5
0.001 - 0.2 Mg/1
0.0002
0.02 Mg/1
1 Mg/1
a. For water only
b. NA Not Available
NOTES: Method Detection Limit (MDL) as listed on this table is the minimum concentration of a substance
that can be measured and reported with 99% confidence that the value is above zero.
Accuracy, presented as an average percent recovery, was determined from replicate (10-25) analyses
of water and wastewater samples fortified with known concentrations of the analyte of interest at
or near the detection limit. In most cases this was less than 10 times the MDL.
Precision data are used to measure the variability of these repetitive analyses reported as a
single standard deviation or, as a percentage of the recovery measurements. For presentation
purposes accuracy, precision and MDL information is presented as an average range of individual
values for every analyte covered by the procedure. If specific information on a particular
compound is required, the specific analytical method cited should be consulted.
-------
OSWER Directive 9355.07-A
TABLE C-l-D: HISTORICAL PRECISION AND ACCURACY DATA/WATER3
LEVEL IV ANALYTICAL TECHNIQUES - CLP RAS METHODS
ANALYTES
Volatilesb
Methylene chloride
1,1-Dichloroethene
1,1-Dichloroethane
Trans-1,2-Dichloroethene
Chloroform
1,2-Dichloroethane
1,1,1-Trichloroethane
Carbon Tetrachloride
1,1,2,2-Tetrachloroe thane
Bromodichloromethane
1,2-Dichloropropane
Trans-1,3-Dichloropropene
Trichloroethene
Dibroraochloromethane
1,1,2-Trichloroethane
Benzene
Cis-1,3-Dichloropropene
Bromofora
Tetrachloroethene
Toluene
Chlorobenzene
Ethyl Benzene
Semivolatiles
bis(2-Chloroethyl)ether
2-Chlorophenol
1,3-Dichlorobenzene
1,4-Dichlorobenzene
1,2-Dichlorobenzene
2-Methylphenol
bis( 2-Chloroisopropyl )ether
TECHNIQUE
Purge & Trap GC/MS
CONCENTRATION
RANGE
N.A.C
GC/MS
N.A.
PRECISION
% RSD
56
20
13
31
12
13
19
12
11
19
18
31
17
14
11
12
22
16
13
14
14
4
24
29
24
21
29
29
25
ACCURACY
% Bias
+36.6
-26.3
-46.4
-21.7
-21.1
+2.4
-41.0
-32.1
-5.8
-13.0
-12.9
-41.2
-22.8
-3.
-7.
-3.3
-35.5
-1-6.5
-42.5
-23.3
-15.9
-31.9
.3
.0
-16
-21
-48
-25
-28
-30
-22
-------
OSWER Directive 9355.07-A
TABLE C-l-D: HISTORICAL PRECISION AND ACCURACY DATA/WATER
LEVEL IV ANALYTICAL TECHNIQUES - CLP RAS METHODS
ANALYTES
Semivolatiles
4-Methylphenol
N-Ni t roso-di-n-propylamine
Nitrobenzene
Isophorone
2-Nitrophenol
bis (2-Chloroethoxy) me thane
2,4-Dichlorophenol
1,2,4-Trichlorobenzene
Naphthalene
4-Chlo ro-3-methylphenol
2,4,6-Trichlorophenol
2-Chloronapthalene
Acenapthene
2,4-Dinitrophenol
2,4-Dinitrotoluene
2,6-Dinitrotoluene
4-Chlorophenyl-phenylether
Fluorene
4,6-Dinitro-2-methylphenol
4-Bromophenyl-phenylether
Hexachlorobenzene
Pentachlorophenol .
Phenanthrene
Fluoranthene
Benzo(b)fluoranthene
Benzolajpyrene
TECHNIQUE
GC/MS
CONCENTRATION
RANGE
N.A.C
PRECISION
% RSD
33
31
32
23
30
34
29
30
44
26
25
24
28
24
34
25
34
25
30
32
36
31
21
42
39
42
ACCURACY
% Bias
-36
+0.3
-23
-8
-21
-2.6
-20
-47
-38
-32
-17
+3.4
-12
-23
-33
-48
+12
-24
-13
-0.1
-42
-24
-28
-15
-10
-29
-------
OSWER Directive 9355.07-A
TABLE C-l-D: HISTORICAL PRECISION AND ACCURACY DATA/WATER
(continued)
LEVEL IV ANALYTICAL TECHNIQUES - CLP RAS METHODS
ANALYTES
Metals
TECHNIQUE
CONCENTRATION
RANGE
Aluminum
Antimony
Arsenic
Barium
Beryllium
Cadmium
Calcium
Chromium
Cobalt
Copper
Iron
Lead
Magnesium
Manganese
Mercury
Nickel
Potassium
Selenium
Sodium
Thallium
Tin
Vanadium
Zinc
ICP
ICP
Furnace AA
ICP
ICP
ICP
ICP
ICP
ICP
ICP
ICP
Furnace AA
ICP
ICP
Cold Vapor
ICP
ICP
Furnace AA
ICP
Furnace AA
ICP
ICP
ICP
1000-3000 ug/1
180-600 ug/1
50-150
800-1500
30-45
25-50
1000-30000
50-150
200-1000
125-250
200-800
30
10000-40000
30-150
5-20
160
10000-20000
50
10000-45000
80-100
160
60-200
50-800
PRECISION
% RSD
9.1
11
9.4
6.8
15
12
6.0
9.8
6.
6.
.7
.7
10.4
32
6.6
6.2
18.8
9.0
16.
8.
8.7
17.2
N.A.
7.6
9.1
.2
.7
ACCURACY
% Bias
-4.3
-9.2
-8.3
-3.9
+3.7
-3.3
-1.6
-2.6
-2.9
-1.1
+6.5
-0.7
-2.5
-1.
-14.
-2.
-12.1
-5.7
-2.8
-4-.2
-2.5
-0.46
+3.0
.0
.4
.5
a. Source: Quality Control in Remedial Site Investigation: Hazardous and Industrial Solid Waste Testing, Fifth Volume,
ASTM STP 925, C.L. Perket, Ed., American Society for Testing Materials, Philadelphia, 1986.
b. Volatile precision and accuracy data from 26-34 laboratories' results on quarterly blind performance evaluation
samples; 29-152 data points for each compound.
c. N.A. «». Not Available.
d. Semivolatile Decision and accuracy data from 19 )5 preaward program data; 22-227 data points :or each compound.
e. Metals precisic i and accuracy data is based on performance evaluation san ) e results from 18 a x>rator es; number
of data coin :s is not- given.
-------
OSWER Directive 9355.0-7A
TABLE C-2-A: HISTORICAL PRECISION AND ACCURACY DATA/SOILS
LEVEL I FIELD SCREENING TECHNIQUES
MEASUREMENT
RESISTIVITY
TERRAIN
CONDUCTANCE
TERRAIN
CONDUCTANCE
Magnetic Field
Intensity
Subsurface
Lithology
Changes
Subsurface
Lithology
Changes
INSTRUMENT
(TECHNIQUE)
Bison 2390 T/R
(Resistivity meter)
EM 31
(conductivity)
EM 34-3
(conductivity)
EDA - Omni IV
(Magnetometer)
SIR-8
(Ground Penetrating
Radar)
EG+G 1225
(Seismograph)
INSTRUMENT
RANGE
0-1999
millivolts
0-1000
millimhos/meter
0-300
millimhos/meter
18000-110000
QdlDIDctS
1-81 dielectric
constant
0-2000
milliseconds
INSTRUMENT .
PRECISION
at 1% range setting,
0-5% of full scale
2% of full scale
2% of full scale
0.02 gamma
H
N/Ad
j
N/Ad
INSTRUMENT
O
ACCURACY
2% of measured
value
5% at 20 millimhos/Dneter
5% at 20 millimhos/meter
1 gamma at 50000 gammas
at 23oC
ri
N/A°
0.01%
-------
OSWER Directive 9355.0-7A
TABLE C-2-A: HISTORICAL PRECISION AND ACCURACY DATA/SOIL
(continued)
LEVEL I FIELD SCREENING TECHNIQUES
MEASUREMENT
TOTAL
VOLATILE
ORGANICS
INSTRUMENT FIELD SCREENING
(TECHNIQUE) RESULTS in ppm (X)
PHOTO VAC 11.4
( GC/Photoionization ) 22 . 0
56.0
139
70.0
24.9
60.0
6.6
12.1
8.7
CLP
RESULTS in ppm (Y)
26.9
32.8
129.7
228.0 & 258.0
126.7
2823.0
53.3
0.056
0.032
0.024
ACCURACY6
(% Bias)
-57.6
-32.9
-56.8
-42.8
-44.8
+99.1
+12.6
+116.9
+377.1
+361.5
a. Source: Manufacturers' manuals unless otherwise cited. Mention of specific models does not constitute
and endorsement of these instrument.
b. Precision refers to reproducibility of meter or instrument reading as cited in instrument specifications.
c. Accuracy refers to instrument specifications unless otherwise cited.
d. N.A. = not available.
e. Accuracy of PhotoVac field screening results calculated by assuming that CLP results on the same samples
were completely accurate. % Bias = 100 (X-Y). Source of these data is CDM project files.
Y
-------
TABLE C-2-B: HISTORICAL PRECISION AND ACCURACY DATA/SOIL
LEVEL II FIELD TECHNIQUES
ANALYTES INSTRUMENT FIELD RESULTS
(TECHNIQUE! IN ppm (x)
PCBs HNu 301 6.0
(GC/ELECTRON 6.0
CAPTURE) 6.0
9.0
13.0
14.0
14.0
21.0
35.0
41.0
48.0
50.0
65.0
67.0
92.0
95.0
11
202
269
286
1215
1647
3054
CLP RESULTS
IN ppm (y)
22.0
6.1
510.0
3.9
3.0
3.1
23.5
8.1
7.7
2.1
11.0
460.0
23.1
18.7
75.0
30.0
12.3
99.0
370.0
80.5
640.0
1040.0
9,300
ACCURACY b
% BIAS
-72.7
-1.6
-98.8
+56.7
+333.3
+351.6
-40.4
+159.3
354.5
+1,852
+336.3
-89.1
+181.4
+258.3
22.7
+216.7
-10.6
+ 104.0
-27.3
+255.3
+90.0
+58.4
-67.2.
a. Source: COM Project files.
b. Source: Accuracy calculated by assuming that CLP results on the same samples were completely accurate. % Bias = 100 (x'y)
V
-------
TABLE C-2-C: HISTORICAL PRECISION AND ACCURACY DATA/SOIL3
LEVEL III ANALYTICAL TECHNIQUES - METHODS OTHER THAN CLP RAS METHODS
ANALYTE METHOD
(TECHNIQUE)
DDXINS 8280
(HPLC/LRMS)
JAR EXTRACTION GC/MS
CONCENTRATION
RANGE
5 ppb
125 ppb
1 ppb
10 ppb
PRECISION
% RSD
6-30
3-10
20
10
ACCURACY
% BIAS
N.A.
N.A.
0
-18
a. Source: Draft Compendium of Information and Performance Data on Routinely Used Measurement Methods (RUMM) - Pilot Phase,
RTI/3087/03, prepared for EPA Quality Assurance Management Staff, January 1986. This document should be
consulted for more information on individual analytes.
-------
OSWER Directive 9355.07-A
TABLE C-2-D: HISTORICAL PRECISION AND ACCURACY DATA/SOILS*
LEVEL IV ANALYTICAL TECHNIQUES - CLP RAS METHODS
ANALYTES
Volatilesb
Chloroform
1,2-Dichloroethane
Dibromochloromethane
Benzene
Bromoform
2-Hexanone
Toluene
Chlorobenzene
Semivolatiles
1,4-Dichlorobenzene
Nitrobenzene
Isophorone
2-Nitrophenol
2,4-Dichlorophenol
1,2,4-Trichlorobenzene
Penta Chlorophenol
Pyrene
2-Nethylnaphthalene
bis-(2-EthylhexylJphthalate
Phenol
Acenaphthylene
Diethyphthalate
Dioxin
271,7,8-TCCD
TECHNIQUE
Purge & Trap GC/MS
CONCENTRATION
RANGE
N.A.C
GC/MS
N.A/
1-10 ugAg
PRECISION
% RSD
8.0
13.1
35.0
32.1
16.6
16.6
13.8
21.2
27
21
24
35
31
28
17
25
26
33
38
26
16
15
ACCURACY
% Bias
-0.1
+11.1
-12.0
-10.3
-12.1
-45.5
+13.7
+13.2
-51
-48
-47
-36
-59
-43
-48
-15
-42
-2
-27
-27
-20
-11.5
-------
OSWER Directive 9355.07-A
TABLE C-2-D:
LEVEL IV ANALYTICAL TECHNIQUES - CLP RAS METHODS
HISTORICAL PRECISION AND ACCURACY DATA/SOILS
(continued)
ANALYTES
TECHNIQUE
CONCENTRATION
RANGE
Metalsb
Aluminum
Cadmium
Calcium
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Mercury
Nickel
Tin
Zinc
ICP
ICP
ICP
ICP
ICP
ICP
Furnace AA
ICP
ICP
Cold Vapor
ICP
ICP
ICP
2-22600 ug/kg
5.5-20
2664-29000
8.5-29600
33-109
5028-113000
11.5-714
2428-7799
73.5-785
1.1-26.5
44-67
N.A.
19-1720
PRECISION
% RSD
14.4
33.2
N.A.
7.8
11.
10.
9.2
7.5
9.4
25.0
15.0
44.1
.2
.7
5.8
ACCURACY
% Bias
-78.8
+2.9
-4.2
-6.1
-2.5
-27.0
-2.2
-10.6
-15.1
-9.1
-17.
N.A.
-6.2
a. Source: Quality Control in Remedial Site Investigation: Hazardous and Industrial Solid Waste Testing, Fifth Volume,
ASTH STP 925, C.L. Perket, Ed., American Society for Testing Materials, Philadelphia, 1986.
b. Volatiles precision and accuracy data is based on 1985 preaward analysis results from laboratories awarded
contracts; 6-14 data points for each compound.
c. N.A. = Not Available.
d. Semivolatlies precision and accuracy data is based on 1985 preaward analysis results; 9-20 data points
for each compound.
e. Dioxin precision and accuracy data is based on results of four performance evaluation samples including
120 data points.
f. Metals precision and accuracy data is based on performance evaluation sample results from 18 laboratories;
number of data points is not < iven.
-------
OSWS* )irec:ive 9355.07-A
TABLE C-3-A: HISTORICAL PRECISION AND ACCURACY DATA/AIR
a
LEVEL I FIELD SCREENING TECHNI
:QUESb
ANALYTES
Organics
Organ! cs
Organics
Organics
INSTRUMENT
(TECHNIQUE)
Century OVA-128
(Flame lonization)
HNu PI-101
( Photoioni zation )
AID - 710
(Flame lonization)
PhotoVac
(GC-Photoion-
i zation)
INSTRUMENT
RANGE
0.1 - 1000 ppm
Methane
0.1 - 2000 ppm
Benzene
0.1 - 2000 ppm
Methane
N.A.
INSTRUMENT
SENSITIVITY
0.1 ppm Methane
0.1 ppm Benzene
0.1 ppm Methane
0.001 ppm
Benzene
INSTRUMENT
PRECISION
N.A.d
± 1% of full scale
deflection
N.A.d
N.A.d
a. Source: Manufacturers' manuals unless otherwise cited. Mention of specific models
does not constitute an endorsement of these instruments.
b. It is difficult to differentiate between Level I and Level II techniques and
instrumentation. Several instruments may be used at both levels.
c. Sensitivity and precision refer to instrument specifications.
d. N.A. = Not Available.
AW3F-29
-------
OSWER Directive 9355.07-A
• TABLE C-3-B: HISTORICAL PRECISION AND ACCURACY DATA/AIR
LEVEL II FIELD TECHNIQUES
ANALYTES
Organics
Compound-
Specific
Organics,
Compound-
Specific
Organics,
Compound-
Specific
Organics,
Compound-
Specific
Mercury
INSTRUMENT
(TECHNIQUE)
Mi ran IB
(Infrared)
Century OVA-128
(GC/Flame
lonization)
PhotoVac
(GC-Photo-
ionization)
SCENTOR
(Argon lonization
or Electron Capture)
Gold film Mercury
Analyzer
INSTRUMENT
RANGE
Compound Dependent,
0-2000 ppm
1-1000 ppm
Methane
N.A.
N.A.
N.A.
INSTRUMENT
SENSITIVITY
N.A.d
N.A.
0.001 ppm
Benzene
0.001 ppm
Benzene
less than
0.01 ppm
INSTRUMENT
PRECISION
N.A.1
N.A.
N.A.
N.A.
N.A.
a. Source: Manufacturers' manuals. Mention of specific models does not constitute an
endorsement of these instruments.
b. It is difficult to differentiate between Level I and Level II techniques and
instrumentation. Several instruments may be used at both levels.
c. Sensitivity and precision refer to instrument specifications.
d. N.A. = Not Available.
AW3F-29
-------
TABLE C-3-C: HISTORICAL PRECISION AND ACCURACY DATA/AIRC
LEVEL III ANALYTICAL TECHNIQUES - METHODS OTHER THAN CLP RAS METHODS
ANALYTES
METHOD CONCENTRATION
(TECHNIQUE) RANGE
CRYOGENIC TRAP/GC
TENAX GC/MS
3.9 ppb
93ppb
7.8 ug/m3
4.5 ug/m3
PRECISION
% RSD
4.0
5.1
11
21
ACCURACY
% BIAS
N.A.
N.A.
N.A.
N.A.
TOLUENE
10.8 ppb
5.11
N.A.
TRICHLOROETHENE
3.5 ppb
84 ppb
4.1
3.7
N.A.
N.A.
VINYL CHLORIDE
7.8 ppb
6.37
N.A.
LEAD
40 CFR 50, APP G
(FLAME AA)
0.6
8.01
ug/m3
ug/m3
8.6
3.9
0
-3.6
a. Source: Draft Compendium of Information and Performance Data on Routinely Used Measurement Methods (RUMM) - Pilot Phase,
RTI/3087/03, prepared for EPA Quality Assurance Management Staff, January 1986. This document should be
consulted for more information on individual analytes.
-------
TABLE C-4-C: HISTORICAL PRECISION AND ACCURACY DATA/OTHER MEDIA3
LEVEL III ANALYTICAL TECHNIQUES - METHODS OTHER THAN CLP RAS METHODS
ANALYTE METHOD CONCENTRATION PRECISION ACCURACY
(TECHNIQUE) MEDIUM RANGE % RSD % BIAS
LEAD 6010 .OILWASTE 1.0 mg/kg 3.1 -10
(ICP) -2.5 mg/kg 22 -20
SOLID WASTE 50 mg/kg 10 3.4
75 mg/kg 3.7 -0.8
SOLID SLUDGE 5 mg/kg 2 0
20 mg/kg 11 55
a. Source: Draft Compendium of Information and Performance Data on Routinely Used Measurement Methods (RUMM) - Pilot Phase,
RTI/3087/03, prepared for EPA Quality Assurance Management Staff, January 1986. This document should be
consulted for more information on individual analytes.
-------
Appendix D
-------
OSWER Directive 9355.0-7A
APPENDIX 0
MODIFIED RCRA APPENDIX VIM
-------
OSWER Directive 9355.0-7A
TABLE D-l
ORGANIC COMPOUNDS ON CLP/BSL
BUT NOT INCLUDED ON MODIFIED APPENDIX VIII
Common Name CAS RN
Acetone 67.64.1
Vinyl Acetate 108.05.4
2-Hexanone 591.78.6
Ethylbenzene 100.41.4
Styrene 100.42.5
Xylenes (Total) 1330-20-7
Benzyl Alcohol 100.51.6
Isopho rone 78.59.1
2-Nitrophenol 88.75.5
Benzoic Acid 65.85.0
2-Methylnaphthalene 91.57.6
2-Nitroaniline 88.74.4
3-Nitroaniline 99.09.2
Dibenzofuran 132.64.9
4,Chlorophenyl-phenylether 7005.72.3
Endrin Ketone 53494.70.5
Endosulfan Sulfate 1031.07.8
AW3F-21
-------
OSWER Directive 9355.0-7A
TABLE D-2
ORGANIC COMPOUNDS CN MODIFIED APPENDIX VIII LIST
BUT NOT INCLUDED ON CLP/HSL
Common Name
Acetonitrile
Acetophenone
2-Acetylaminofluorine
Acrolein
Acrylonitrile
Allyl Alcohol
4-Aminobiphenyl
Aramite
Benzenethiol
p-Benzoquinone
Bromoacetone
2-sec-butyl-4,6-dinitrophenol
Chlorobenzilate
2-chloro-l,3-butadiene
3-chloropropene
3-chloropropionitrile
Diallate
Dibenzo [a,e] pyrene
Dibenzo [a,h] pyrene
Dibenzo [a,i] pyrene
1,2-dibromo-3-chloropropane
1,2-dibromoethane
Dibromomethane
1,4-dichloro-2-butene
Dichlorodifluoromethane
2,6 Dichlorophenol
1,3-Dichloropronene
0,0-Diethyl 0-2-pyrazinyl
phosphorothioate
3,3-Dimethoxybenzidine
p-Dimethylaminozobenzene
7,12-Dimethylbenz[a)anthracene
3,3 -Dimethylbenzidine
alpha-Dimethylphenethylamine
1,4-Dioxane
Diphenylamine
1,2-Diphenylhydrazine
Di-n-propylnitrosamine
Disulfoton
Ethyl Cyanide
Ethylene Oxide
meta-dini trobenzene
Silvex
1,2,3-trichloropronene
Tris (2,3-dibromopropyl) phosphate
CAS RN
75.05.8
98.86.2
53.96.3
107.02.8
107.13.1
107.18.6
92.67.1
140.57.8
108.98.5
106.51.4
598.31.2
88.85.7
510.15.6
126.99.8
107.05.1
542.76.7
2303.16.4
192.65.4
189.64.0
189.55.9
96.12.8
106.93.4
74.95.3
764.41.0
75.71.8
87.65.0
542.75.6
297.97.2
119.90.4
60.11.7
57.97.6
119.93.7
122.09.8
123.91.1
122.39.4
122.66
621.64
298.04
107.12.0
75.21.8
100.25.4
93.72.1
96.18.4
126.72.7
.7
.7
.4
Class*
CLP/VOA
CLP/BNA
CLP/BNA
CLP/VOA
CLP/VOA
NRA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/BNA
NRA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/VOA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/VOA
CLP/VOA
CLP/VOA
CLP/VOA
CLP/VOA
CLP/BNA
CLP/VOA
NRA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/BNA
NRA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/VOA
CLP/VOA
NRA
CLP/BNA
NRA
CLP/VOA
CLP/BNA
AW3F-20/1
-------
Common Name
Phenacetin
N-Phenylthiourea
Phorate
Famphur
2-Picoline
Propanamide
2-Propyn-l-ol
Pyridine
Resorcinol
Safrole
1,2,4,5-Tetrachlorobenzene
1,1,1,2-Tetrachloroethane
2-Naphthylamine
N-Ni t rosodi-n-butylamine
N-Ni trosodiethylamine
N-Ni trosomethylethylamine
N-Nitrosomorpholine
N-Nitrosopiperdine
5-Ni t ro-o-toluidine
Parathion
Pentachlorobenzene
Pentachloroethane
Pentachloronitrobenzene
Kepone
Malonitrile
Methyacrylonitrile
Methapyrilene
3-Methylchloranthrene
4,4-Methylene-bis (2-chloroaniline)
Methylmethacrylate
Methylmethanesulfonate
Aldicarb
Methyl parathion
1,4 Naphthoquinone
1-Naphthylamine
2,3,4,6-Tetcachlorophenol
Tetraethyldithiopyrophosphate
Trichloromethanethal
Trichloromonofluoromethane
2,4,5-T
Ethyl Methacrylate
Isodrin
Hexachlorophene
Hexachloropropene
lodomethane
Isobutylalcohol
Isosafrole
OSWER Directive 9355.0-7A
TABLE D-2 (CONT'D)
CAS RN
a
62.44.2
103.85.5
298.02.2
52.85.7
109.06.8
23950.58.5
107.19.7
110.86.1
108.46.3
44.59.7
95.94.3
630.20.6
91.59.8
924.16.3
55.18.5
10595.95.6
59.89.2
100.75.4
99.44.8
56.38.2
608.93.5
76.01.7
82.68.8
143.50.0
109.77.3
126.98.7
91.80.5
56.49.5
101.14.4
80.62.6
66.27.3
116.06.3
298.00.0
130.15.4
134.32.7
58.90.2
3689.24.5
75.70.7
75.69.4
93.76.5
97.63.2
465.73.6
70.30.4
1888.71.7
74.88.4
78.33.1
120.58.1
Class
CLP/BNA
CLP/BNA
NRA
NRA
CLP/BNA
CLP/BNA
NRA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/VOA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/BNA
NRA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/PCB-PeSt
CLP/BNA
CLP/VQA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/BNA
CLP/BNA
NRA
CLP/BNA
CLP/BNA
CLP/BNA
NRA
CLP/BNA
CLP/VOA
NRA
CLP/BNA
CLP/PCB-Pest
CLP/BNA
CLP/VOA
CLP/VOA
CLP/VOA
NRA
AW3F-20/2
-------
OSWER Directive 9355.0-7A
TABLE D-2 (CCNT'D)
NOTES
aClass Abbreviations
NRA - Not readily analyzable using current CLP Procedures
CLP/VOA - Potentially analyzable using current CLP/HSL GC/MS Volatile
Organics Procedure
CLP/BNA - Potentially analyzable using current CLP/HSL Base/Neutral Acid
Extractable GC/MS Procedure
CLP/PCB-Pest - Potentially analyzable using current CLP/HSL PCB/Pesticide
GC Procedure
AW3F-20/3
-------
OSWER Directive 9355.0-7A
TABLE D-3
ORGANIC COMPOUNDS ON MODIFIED
APPENDIX VIII
LIST THAT ARE NOT READILY ANALYZABLE BY CURRENT
CLP/HSL PROCEDURES
Common Name CAS RN
Allyl alcohol 107.18.6
Bromoacetone 598.31.2
0,0-Diethyl-0-2-Pyrazinyl phosphorothioate 297.97.2
1,4 Dioxane 123.91.1
Ethylene Oxide 75.21.8
Silvex 93.72.1
Phorate 298.02.2
Famphur 52.85.7
2-Propyn-l-ol 107.19.7
Parathion 56.38.2
Methyl Parathion 298.00.0
Tetcaethyldithiopycophosphate 3689.24.5
2,4,5-T 93.76.5
Isosafrole 120.58.1
Class*
WS/NV
WR
OP
WS/NV
NR (VGA)
CH
OP
OP
WS/NV
OP
OP
WR
CH
D/H
NOTES
aClass Abbreviations
WS/NV - Water soluble, nonvolatile compound probability not amenable to
purge and trap or liquid/liquid extraction pretreatment.
WR - Water reactive, unanalyzable in aqueous matrix.
OP - Organophosphorous pesticide best analyzed by a modified SW-846,
Method 8140.
NR (VGA) - Not recoverable at 200 PPB using standard HSL/CLP volatile
organics procedures. May be more ameneable to head space analysis.
CH - Chlorinate herbicide, must be derivatized prior to analysis. Best
analyzed using modified SW-846 Method 8150.
D/H - Decomposes at conventional GC temperatures HLPC procedure may be
applicable. ~
AW3F-19/1
-------
Appendix E
-------
APPENDIX E
CONTRACT - REQUIRED DETECTION LIMITS
FOR HSL ANALYSIS
USING CLP IFB PROCEDURES
AW3F-28
-------
OSWER Directive 9355.0-7A
TABLE E-l
CLP VOLATILE ORGANIC CRDL
Target comoound name
till oromethane
Bromomethane
Vinyl Chloride
Chloroethane
Methyl ene Chloride
Acetone
Carbon Disulflde
1,1-Dichloroethene
1,1-01 chl oroethane
Trans-1 ,2-Oi chl oroethene
Chloroform
1,2-01 chl oroethane
2-Butanone
1 ,1 ,1-Tri chl oroethane
Carbon Tetrachloride
Vinyl Acetate
Bromodi chl oromethane
1,1,2 ,2-Tetrachl oroethane
1 ,2-Di chl oropropane
Trans-1 ,3-01 chl oropropene
Tri chl oroethene
D1 bromochl oromethane
1 ,1 ,2-Tr1 chl oroethane
Benzene •
C1 s-1 ,3-01 chl oropropene
2-Chloroethyl Vinyl Ether
Bromoform
4-Methyl -2-pentanone
2-Hexanone
Tetra chl oroethene
Toluene
Chlorobenzene
Ethyl Benzene
Styrene
Total Xylenes
cccc
SPCC
CCC
CCC
SPCC
CCC
SPCC
CCC
SPCC
CCC
SPCC
CCC
Low soil
CRDL.
10
10
10
10
5
10
5
5
5
5
5
5
10
5
5
10
5
5
5
5
5
5
5
5
5
10
5
10
10
5
5
5
5
5
5
Low water
CRDL,
ufl/L
10
10
10
10
5
10
5
5
5
5
5
5
10
5
5
10
5
5
5
5
5
5
5
5
5
10
5
10
10
5
5
5
5
5
5
CAS number
74-87-3
74-83-9
75-01-4
75-00-3
75-09-2
67-64-1
75-15-0
75-35-4
75-35-3
156-60-5
67-66-3
107-06-2
78-93-3
71-55-6
56-23-5
108-05-4
75-27-4
79-34-5
78-87-5
10061-02-6
79-01-6
124-48-1
79-00-5
71-43-2
10061-01-5
110-75-8
75-25-2
108-10-1
591-78-6
127-18-4
108-88-3
108-90-7
100-41-4
100-42-5
N.A.
aCRDL values obtained from the IFB WA85-0664 [7J.
DSystem Performance Check Compounds (SPCC) are used to check compound
instability and degradation in the GC/MS and to insure minimum average
response factors are met prior to the use of the calibration curve.
cColumn Check Compounds (CCC) are used to check the validity of the
initial calibration.
Note: Medium soil and water CRDLs are 100 times the low level CRDLs.
SOURCE: Flotard, R.D. et al 1986
AW3F-26
E-l
-------
OSWER Directive 9355.0-7A
TABLE E-2
CLP INORGANIC COMPOUND CRDL,
INSTRUMENT DETECTION LEVEL AND WAVELENGTH
El ement
Al
Sb
As
Ba
Be
Cd
Ca
Cr
Co
Cu
Fe
Pb
• Mg
Mn
Hg
N1
K
Se
Ag
Na
Tl
Sn
V
Zn
CRDL
ZOO
60
10
200
5
5
5000
10
50
25
100
5
5000
15
0.2
40
5000
5
10
5000
10
40
50
20
Method
ICP
ICP
FAA
ICP
ICP
ICP
ICP
ICP
ICP
ICP
ICP
ICP
ICP
ICP
CV
ICP
ICP
FAA
ICP
ICP
ICP
ICP
ICP
ICP
N
7
5
18
5
10
5
7
9
11
11
10
12
11
10
12
9
8
18
10
9
18
7
10
• o
IDL
Mean
70.7
42.3
4.6
22.1
2.3
4.0
529
5.8
11.4
9.7
27.4
2.3
385
5.2
0.2
17.8
668
2.8
5.4
756
4.3
23.8
13.1
8.3
IDL
Std Dev
59.3
11.3
2.3
31.7
1.7
1.1
472
2.9
8.5
6.5
20.9
1.2
449
4.6
0.1
10.1
444
1.3
2.7
864
2.4
8.4
10.0
' 6.3
Wave-
Length (nm)
309.3
217.6
.198.7
493.4
312.0
228.8
317.9
267.7
228.6
324.5
259.9
283.3
279.6
257.6
253.7
232.0
766.5
196.0
328.1
589.0
276.8
190.0
292.5
213.9
IDL - InstrumentDetectionunnt Cw9/L).
N - Number of laboratories using the most common wavelength.
CRDL - Contract Required Detection Limit (yg/L).
SOURCE: Aleckson, K.A. et al 1986.
AW3F-26 E-2
-------
OSWER Directive 9355.0-7A
TABLE E-3
CLP SEMI-VOLATILE HSL CXIMPOUNDS AND CRDL
Compound name
Phenol
b1s(2-Chloroethy!) ether
2-Chl orophenol
1 ,3-01 chl orobenzene
1 ,4-D1chl orobenzene
Benzyl alcohol
1 , 2-D1 chl orobenzene
2-Methyl phenol
bis(2-Chloro1sopropyl)ether
4-Methyl phenol
N-N1troso-d1-n-propylam1ne
Hexachloroethane
Nitrobenzene
Isophorone
2-N1trophenol
2, 4 -01 methyl phenol
Benzole add
b1 s(2-Chl oroethoxy )me thane
2 ,4-01 chl orophenol
1 , 2 ,4-Tr1 chl orobenzene
Naphthalene
4-Chl oroanHlne
Hexachl orobutadlene
4-Chl oro-3-methyl phenol
2-Methyl naphthal ene
Hexachl orocycl opentadl ene
2, 4, 6-Trl chl orophenol
2 , 4 , 5-Trl chl orophenol
2-Chl oronaphthal ene
2-N1troan1l1ne
Dimethyl phthal ate
Acenaphthyl ene
3-Nttroanlllne
Acenaphthene
2,4-D1n1trophenol
4-N1trophenol
Dlbenzofuran
2, 4-01 nltro toluene
2, 6-01 nltro toluene
D1 ethyl phthal ate
4-Chl orophenyl -phenyl ether
Fl uorene
4-N1troan1lfne
4 ,6-01 n1 tro-2-methyl phenol
SPCCa
or CCCb
CCC
ccc
SPCC
ccc
ccc
ccc
SPCC
ccc
ccc
SPCC
SPCC
LOW 5011
CRDL, ug/kg
330
330
330
330
330
330
330
330
330
330
330
330
330
330
330
330
1,600
- 330
330
330
330
330
330
330
330
330
330
1,600
330
1,600
330
330
1,600
330
1,600
1,600
330
330
330
330
330
330
1,600
1,600
Low Water
CRDL, ug/L
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
50
10
10
10
10
10
10
10
10
10
10
50
10
50
10
10
50
10
50
50
10
10
10
10
10
10
50
50
CAS
Number
108-95-2
111-44-4
95-57-8
541-73-1
106-46-7
100-51-6
. 95-50-1
95-48-7
39638-32-9
106-44-5
621-64-7
67-72-1
98-95-3
78-59-1
88-75-5
105-67-9
65-85-0
111-91-1
120-83-2
120-82-1
91-20-3
106-47-8
87-68-3
59-50-7
91-57-6
77-47-4
88-06-2
95-95-4
91-58-7
88-74-4
131-11-3
208-96-8
99-09-2
83-32-9
51-28-5
100-02-7
132-64-9
121-14-2
606-20-2
84-66-2
7005-72-3
86-73-7
100-01-6
534-52-1
AW3F-26
E-3
-------
OSWER Directive 9355.0-7A
TABLE E-3
CLP SEMI-VOLATILE HSL COMPOUNDS AND CRDL
(continued)
Compound name
N-N1 trosodl pheny 1 ami ne
4-Bromophenyl -phenyl ether
Hexachl orobenzene
Pentachlorophenol
Phenanthrene
Anthracene
01 -n-bu tyl phthal ate
Fluoranthene
Pyrene
Butyl benzyl phthal ate
3,3'-D1chlorobenz1d1ne
Benzo( a ) anthracene
b1s(2-Ethylhexyl ) phthal ate
Chrysene
D1-n-octyl phthal ate
Benzol b ) fl uoran thene
Benzo(k)fluoranthene
Benzol a jpyrene
Indeno(l,2,3-cd)pyrene
D1benz(a,h)anthracene
Benzo(g,h,1 Jperylene
aCCC-Cal1brat1on Check Comp
SPCC°
or CCCb
CCC
CCC
CCC
CCC
CCC
lound
Low Soil
CRDL, ug/kg
330
330
330
1.600
330
330
330
330
330
330
660
330
330
330
330
330
330
330
330
330
330
Low Water
CRDL, ug/L
10
10
10
50
10
10
10
10
10
10
20
10
10
10
10
10
10
10
10
10
10
CAS
Number
86-30-6
101-55-3
118-74-1
87-86-5
' 85-01-8
120-12-7
84-74-2
206-44-0
129-00-0
85-68-7
91-94-1
56-55-3
117-81-7
218-01-9
117-84-0
205-99-2
207-08-9
50-32-8
193-39-5
53-70-3
191-24-2
bSPCC-System Performance Check Compound
Note: Medium soil/sediment contract required detection limits are 60
times the Individual low soil/sediment CRDL and medium water
contract required detection limits are 100 times the Individual
low water CRDL.
SOURCE: Wolf, J.S. et al 1986.
AW3F-26 E-4
------- |