Day 2 DQO Training Course
Module 7
The EPA 7-Step DQO Process
Step 7 - Optimize Sample Design
Presenter:
Sebastian Tindall
(70 minutes)
1 of 86
Terminal Course Objective
To be able to use the output from the previous
DQO Process steps to select sampling and
analysis designs and understand design
alternatives presented to you for a specific
project
2 of 86
Step 7: Optimize Sample Design
Step 1: State the Problem
Step 2: Identify Decisions
Step 3: Identify Inputs
Step 4: Specify Boundaries
Step 5: Define Decision Rules
Step Objective:
Identify the most resource
effective data collection
and analysis design that
satisfies the DQOs
specified in the preceding 6
steps
Step 6: Specify Error Tolerances
Step 7: Optimize Sample Design
3 of 86
Step 7- Optimize Sample Design
Information IN
From Previous Step
Decision Error
Tolerances
Gray Region
Actions
Information OUT
Review DQO outputs from Steps 1-6 to
be sure they are internally consistent
To Next Step
Develop alternative sample designs
For each design option, select needed
mathematical expressions
Select the optimal sample size that satisfies the
DQOs for each data collection design option
Go back to
Steps 1- 6
and revisit
decisions.
Check if number of samples exceeds
project resource constraints
Yes
No
Optimal
Sample
Design
4 of 86
Step 7- Optimize Sample Design
Information IN
From Previous Step
Decision Error
Tolerances
Gray Region
Actions
Information OUT
Review DQO outputs from Steps 1-6 to
be sure they are internally consistent
To Next Step
Develop alternative sample designs
The outputs should provide
information on the context
of,needed
requirements for, and
For each design option, select
constraints on data collection
mathematical expressions
design.
Select the optimal sample size that satisfies the
DQOs for each data collection design option
Go back to
Steps 1- 6
and revisit
decisions.
Check if number of samples exceeds
project resource constraints
Yes
No
Optimal
Sample
Design
5 of 86
Step 7- Optimize Sample Design
Information IN
From Previous Step
Decision Error
Tolerances
Gray Region
Go back to
Steps 1- 6
and revisit
decisions.
Actions
Review DQO outputs from Steps 1-6 to
be sure they are internally consistent
Information OUT
To Next Step
Develop alternative sample designs
For each design option, select needed
mathematical expressions
Based
the DQO
Select the optimal sample size
that on
satisfies
the outputs
fromdesign
Steps option
1-6, for each
DQOs for each data collection
decision rule develop one or
more sample designs to be
Check if number of samples
exceeds
considered
and evaluated in
project resource constraints
Step 7.
Optimal
Sample
No
Yes
Design
6 of 86
Step 7- Optimize Sample Design
Information IN
From Previous Step
Decision Error
Tolerances
Gray Region
Go back to
Steps 1- 6
and revisit
decisions.
Actions
Review DQO outputs from Steps 1-6 to
be sure they are internally consistent
Information OUT
To Next Step
Develop alternative sample designs
For each design option, select needed
For each option, pay close attention
mathematical expressions
to the Step 4 outputs defining the
population to be represented
Select the optimal samplewith
size the
thatdata:
satisfies the
DQOs for each data collection
design
optionmethod
• Sample
collection
• Sample mass size
• Sampleexceeds
particle size
Check if number of samples
Etc.
project resource•constraints
Optimal
Sample
No
Yes
Design
7 of 86
Step 7- Optimize Sample Design
Information IN
From Previous Step
Decision Error
Tolerances
Gray Region
Go back to
Steps 1- 6
and revisit
decisions.
Actions
Information OUT
Review DQO outputs from Steps 1-6 to
be sure they are internally consistent
To Next Step
Develop alternative sample designs
For each design option, select needed
mathematical expressions
Remember:
Sampling Uncertainty is decreased
Select the optimal samplewhen
size that
satisfies
the is increased.
sampling
density
DQOs for each data collection design option
Check if number of samples exceeds
project resource constraints
Yes
No
Optimal
Sample
Design
8 of 86
Types of Designs
Simple Random
 Systematic Grid with random start
 Geometric Probability or “Hot Spot” Sampling
 Stratified Random

– Stratified Simple Random
– Stratified Systematic Grid with random start
Statistical Methods for Environmental Pollution Monitoring,
Richard O. Gilbert, 1987
9 of 86
Simple Random
Definition- choice of sampling location or
time is random
 Assumptions

– Every portion of the population has equal
chance of being sampled

Limitation-may not cover area
10 of 86
Simple Random

To generate a simple random design:
– Either grid the site - set up equal lateral
triangles or equal side rectangles and number
each grid, use a random number generator to
pick the grids from which to collect samples
– Randomly select x, y, z coordinates, go to the
random coordinates and collect samples
11 of 86
Example - Simple Random Using
Coordinates
N
147'
168'
- A R andom ly Selected Sam pling L ocation
- D enotes random length & w idth C oordinates w alked-off by
sam pling team
12 of 86
Systematic Grid, Random Start


Definition-taking measurements at locations or times
according to spatial or temporal pattern (e.g.,
equidistant intervals along a line or grid pattern)
Assumptions
– Good for estimating means, totals and patterns of
contamination
– Improved coverage of area
13 of 86
Systematic Grid, Random Start
(cont.)

Limitations
– Biased results can occur if assumed pattern of
contamination does not match the actual pattern
of contamination
– Inaccurate if have serial correlation

NPDES outfall
– Periodic recurring release; time dependent

Groundwater:
– seasonal recurrence; water-level dependence
14 of 86
Systematic Grid, Random Start
(cont.)
Hot spot
c
Remember:
Start at random location
Move in a pre-selected pattern across
the site, making measurements
at each point
N
- Randomly Selected Starting
Location
- Equally Spaced Sampling
Locations
15 of 86
Geometric Probability or HotSpot Sampling
Uses squares, triangles, or rectangles to
determine whether hot spots exist
 Finds hot spot, but may not estimate the
mean with adequate confidence

16 of 86
Geometric Probability or HotSpot Sampling (cont.)

Number of samples is calculated based on probability
of finding hot area or geometric probability

Assumptions
– Target hot spot has circular or elliptical shape
– Samples are taken on square, rectangular or triangular grid
– Definition of what concentration/activity defines hot spot
is unambiguous
17 of 86
Geometric Probability or HotSpot Sampling (cont.)

Limitations
– Not appropriate for hot spots that are not elliptical
– Not appropriate if cannot define what is hot or the
likely size of hot spot
18 of 86
Example Grid for Hot-Spot Sampling
19 of 86
Geometric Probability or HotSpot Sampling (cont.)

In order to use this approach the decision
makers MUST
– Define the size of the hot spot they wish to find
– Provide rationale for specifying that size.
– Define what constitutes HOT (e.g., what
concentration is HOT)
– Define the effect of that HOT spot on achieving
the release criteria
20 of 86
Stratified Random
Definition-divide population into strata and
collect samples in each strata randomly
 Attributes

–
–
–
–

Provides excellent coverage of area
Need process knowledge to create strata
Yields more precise estimate of mean
Typically more efficient then simple random
Limitations
– Need process knowledge
21 of 86
Example - Stratified Simple
Random
Strata 1
Strata 2
22 of 86
Sampling Approaches

Sampling Approach 1
– Simple Random
– Traditional fixed laboratory analyses

Sampling Approach 2
–
–
–
–
Systematic Grid
Field analytical measurements
Computer simulations
Dynamic work plan
23 of 86
Approach 1 Sample Design
CS
Plan View
Former Pad
Location
Buffer
Zone
Runoff
Zone
0
0
50
15
100
30
150 ft
46 m
24 of 86
Design Approaches
Approach 1
Collect samples using Simple Random design.
Use predominantly fixed traditional laboratory
analyses and specify the method specific details
at the beginning of DQO and do not change
measurement objectives as more information is
obtained
25 of 86
Step 7- Optimize Sample Design
Information IN
From Previous Step
Decision Error
Tolerances
Gray Region
Go back to
Steps 1- 6
and revisit
decisions.
Actions
Review DQO outputs from Steps 1-6 to
be sure they are internally consistent
Information OUT
To Next Step
Develop alternative sample designs
For each design option, select needed
mathematical expressions
Select the optimal sample size that satisfies the
DQOs for each data collection design option
1. Statistical Method/Sample Size Formula
2. Cost Function
Check if number of samples exceeds
project resource constraints
Optimal
Sample
No
Yes
Design
26 of 86
Step 7- Optimize Sample Design
Information IN
From Previous Step
Decision Error
Tolerances
Gray Region
Actions
Review DQO outputs from Steps 1-6 to
be sure they are internally consistent
Information OUT
To Next Step
Develop alternative sample designs
For each design option, select needed
mathematical expressions
Select the optimal sample size that satisfies the
DQOs for each
data
collection design option
1. Statistical Method/Sample
Size
Formula
Define suggested method(s) for testing the statistical hypothesis
and define sample sizeCheck
formula(e)
that corresponds
to the method(s).
if number
of samples exceeds
Go back to
project resource constraints
Steps 1- 6
Optimal
and revisit
Sample
No
Yes
decisions.
Design
27 of 86
Step 7- Optimize Sample Design
Information IN
From Previous Step
Decision Error
Tolerances
Gray Region
Actions
Information OUT
Review DQO outputs from Steps 1-6 to
be sure they are internally consistent
To Next Step
Develop alternative sample designs
For each design option, select needed
mathematical expressions
Select the optimal sample size that satisfies the
DQOs
for each data collection design option
Perform a preliminary
DQA:
• Generate frequency distribution histogram(s) for each population
• Select one or more statistical
methodsofthat
will address
the PSQs
Check if number
samples
exceeds
back
• Go
List
thetoassumptions for choosing
these statistical
methods
project resource
constraints
• Steps
List 1the6appropriate formula for calculating the number of
and
revisitn
samples,
No
Yes
decisions.
Optimal
Sample
Design
28 of 86
CS
Histogram
4
Frequency
Frequency
3
2
1
3
2
1
0
0
0
7
14
21
28
0
35
95
5
10
5
11
U Concentration
Pb Concentration (m g/kg)
3
Frequency
4
Frequency
85
3
2
1
2
1
0
0
0
1.7
3.4
5.1
TPH Concentration
6.8
0
0.8
1.6
2.4
3.2
Arochlor 1260
29 of 86
3 Approaches for Calculating n
Normal approach
 Skewed approach
 FAM/DWP approach

– Badly skewed or for all distributions use
computer simulation approach
• e.g., Monte Carlo
30 of 86
Begin With the Decision in Mind
Contaminant Concentrations in the Spatial
Distribution of the Population
Population Frequency Distribution
Correct Equation for n (Statistical Method)
Data
• field
• onsite
methods
• traditional
laboratory
, , , 
Alternative Sample Designs
Optimal Sampling Design
How Many Samples
do I Need?
The end
31 of 86
Logic to Assess Distribution and
Calculate Number of Samples
Is frequency
distribution from
each population
symmetrical or
approximately
symmetrical?
Yes
Symmetrical
Use equations based on
symmetrical distribution.
No
Option 1
Skewed
Calculate the number of
samples based on skewed
distributions (e.g.,
nonparametric tests such
as WSR or WRS)
Option 2
Badly Skewed
Badly skewed or for any
distribution, use computer
simulations
(e.g.,Monte Carlo) to perform
calculations to estimate the
number of samples
32 of 86
Normal Approach
CS
Due to using only five samples for initial
distribution assessment, one cannot infer a
‘normal’ frequency distribution
Reject the ‘Normal’ Approach and
Examine ‘Non-Normal’
or ‘Skewed’ Approach
33 of 86
Logic to Assess Distribution and
Calculate Number of Samples
Is frequency
distribution from
each population
symmetrical or
approximately
symmetrical?
Yes
Symmetrical
Use equations based on
symmetrical distribution.
No
Option 1
Skewed
Calculate the number of
samples based on skewed
distributions (e.g.,
nonparametric tests such
as WSR or WRS)
Option 2
Badly Skewed
Badly skewed or for any
distribution, use computer
simulations
(e.g.,Monte Carlo) to perform
calculations to estimate the
number of samples
34 of 86
Step 7- Optimize Sample Design
Information
IN
Using
Actions
the formulae appropriate
to these methods, calculate
From Previous
the Step
number ofReview
samplesDQO
required,
varying
 for1-6
a given
outputs
from,
Steps
to .
Repeat the same
process
new s. consistent
be sure
theyusing
are internally
Information OUT
To Next Step
Review
Decision
Errorall of calculated sample sizes and along with
Develop
sample
their corresponding
levelsalternative
of , , and
. designs
Tolerances
Select those sample
have
acceptable
of
Forsizes
eachthat
design
option,
selectlevels
needed
, , and  associated mathematical
with them. expressions
Gray Region
Select the optimal sample size that satisfies the
DQOs for each data collection design option
Go back to
Steps 1- 6
and revisit
decisions.
Check if number of samples exceeds
project resource constraints
Yes
No
Optimal
Sample
Design
35 of 86
CS
Pb, U, TPH (DRO/GRO)
Because there were multiple COPCs with
varied standard deviations, action limits and
LBGRs, separate tables for varying alpha,
beta, and (LBGR) delta were calculated
 For the U, Pb, and TPH, the largest number
of samples for a given alpha, beta and delta
are presented in the following table

36 of 86
Pb, U, TPH Based on Non-Parametric Test
CS
Sample Sizes Based on Varying Error Tolerances and LBGR
Lead, Uranium, and TPH
Mistakenly Concluding < Action Level
s = 10.5 (U)
Sample size formula: n  1.16
 = 0.01
( 1  1  ) 2 S 2
() 2
 = 0.05

 = 0.10
 21
2
Mistakenly
Concluding
> Action
Level
Width of the Gray Region, () = 240 – 229.5 = 10.5 (total error estimate)
= 0.10
19
12
9
= 0.20
15
9
7
= 0.30
13
8
5
Mistakenly
Concluding
> Action
Level
Width of the Gray Region, () = 240 – 192 = 48 (20% of action level)
= 0.10
4
3
2
= 0.20
4
2
2
= 0.30
4
2
2
Mistakenly
Concluding
> Action
Level
Width of the Gray Region, () = 240 – 120 = 120 (50% of action level)
= 0.10
4
2
2
= 0.20
4
2
1
= 0.30
4
2
1
37 of 86
Aroclor 1260- Non-Parametric Test


CS
For PCBs, the Aroclor 1260 has the greatest
variance and using the standard deviation results
in a wide gray region
The following table presents the variation of
alpha, beta and deltas for Aroclor 1260
38 of 86
Aroclor 1260- Non-Parametric Test
CS
Sample Sizes Based on Varying Error Tolerances and LBGR
PCBs (based on Aroclor 1260)
Mistakenly Concluding < Action Level
s = 0.88 (A-1260)
Sample size formula: n  1.16
 = 0.01
( 1  1  ) 2 S 2
() 2
 = 0.05

 = 0.10
 21
2
Mistakenly
Concluding
> Action
Level
Width of the Gray Region, () = 1 – 0.12 = 0.88 (total error estimate)a
= 0.10
19
12
9
= 0.20
15
9
7
= 0.30
13
8
5
Mistakenly
Concluding
> Action
Level
Width of the Gray Region, () = 1 – 0.80 = 0.20 (20% of action limit)
= 0.10
296
194
149
= 0.20
229
141
103
= 0.30
186
108
75
Mistakenly
Concluding
> Action
Level
Width of the Gray Region, () = 1 - 0.50 = 0.50 (50% of action limit)
= 0.10
50
33
25
= 0.20
40
24
18
= 0.30
33
19
13
a
The total error estimate for subsurface concentrations exceeds the action limit, thus inappropriately moving
the LBGR below zero. Only the surface concentration error estimate is considered here for that reason.
39 of 86
Step 7- Optimize Sample Design
Information IN
From Previous Step
Decision Error
Tolerances
Gray Region
Actions
Review DQO outputs from Steps 1-6 to
be sure they are internally consistent
Information OUT
To Next Step
Develop alternative sample designs
For each design option, select needed
mathematical expressions
Select the optimal sample size that satisfies the
2. Cost Function
DQOssample
for each
datadevelop
collection
design option
For each selected
size,
a cost
function that relates the number of samples
to the total cost of
sampling
and analysis.
Check
if number
of samples exceeds
Go back to
project resource constraints
Steps 1- 6
and revisit
No
Yes
decisions.
Optimal
Sample
Design
40 of 86
Step 7- Optimize Sample Design
Information IN
From Previous Step
Decision Error
Tolerances
Gray Region
Actions
Review DQO outputs from Steps 1-6 to
be sure they are internally consistent
Information OUT
To Next Step
Develop alternative sample designs
For each design option, select needed
mathematical expressions
Select
the function,
optimal sample
size thatunit
satisfies
the
In order to develop
the cost
the aggregate
cost per
DQOs for This
each is
data
design option
sample must be determined.
thecollection
cost of collecting
one
sample and conducting all the required analyses for a given
decision rule.
Check if number of samples exceeds
Go back to
project resource constraints
Steps 1- 6
and revisit
No
Yes
decisions.
Optimal
Sample
Design
41 of 86
Aggregate Unit Sampling and Analysis
Cost
j
AUSCA$ = USC$ +
 USA$
i
i=1
Where (here):
USC$ = Unit Sample Collection Cost
USA$ = Unit Sample Analysis Cost
AUSCA$ = Aggregate Unit Sample Collection and Analysis Cost
j = Number of analytical methods planned
42 of 86
CS
Approach 1 Sampling Design (cont.)
Surface Soils S&A Costs
Lab Analytical Cost Without
PCBs
Pb by ICP/AES
U by ICP/AES
TPH (GRO) by GC
TPH (DRO) by GC
Total USA$
Unit Sample Collection Cost
AUSCA$ = USC$ + total USA$
Unit Sample
Analysis Cost
$35
$65
$65
$85
$250
$50
$300
43 of 86
CS
Approach 1 Sampling Design (cont)
Sub-surface Soils S&A Costs
44 of 86
CS
Approach 1 Sampling Design (cont.)
Surface Soils
Lab Analytical Costs for PCBs
by GC
Polychlorinated biphenyls
Total USA$
Unit Sample Collection Cost
AUSCA$ = USC$ + total USA$
$ 150.00
$ 150.00
$ 50.00
$ 200.00
45 of 86
Approach 1 Sampling Design (cont.)
CS
Sub-surface Soils
Lab Analytical Costs for PCBs
by GC
Polychlorinated biphenyls
Total USA$
Unit Sample Collection Cost
AUSCA$ = USC$ + total USA$
$ 150.00
$ 150.00
$ 100.00
$ 250.00
46 of 86
Step 7- Optimize Sample Design
Information IN
Actions
Merge the selected sample size outputs
with
From
Step Unit Sample Collection and
thePrevious
Aggregate
Review DQO outputs from Steps 1-6 to
Analysis cost output.
be sure they are internally consistent
Information OUT
To Next Step
Decision
Error
This results
in a table that shows the product
Develop
alternative
sample designs
ofTolerances
each selected sample
size and
the AUSCA$.
This table is used to present
the project
For each
design option, select needed
managers and decision makers
with a range
mathematical
expressions
Gray
Region
of analytical costs and the resulting
uncertainties.
Select the optimal sample size that satisfies the
DQOs for each data collection design option
From the Check
table, select
if number of samples exceeds
Go back to the optimal sample size that
project resource constraints
Steps 1- 6 meets the project budget
and revisit and uncertainty requirements.
No
Yes
decisions.
Optimal
Sample
Design
47 of 86
SHOW EXCEL File
48 of 86
Approach 1 Based Sampling Design

CS
Design for Pb, U, TPH
– Alpha = 0.05; Beta = 0.2; Delta = total error
– The decision makers agreed on collection of 9 surface samples for Pb, U and
TPH (GRO & DRO) from each of the two surface strata, for a total of 18
samples using a stratified random design
– For the sub-surface, 9 borings/probes will be made in each of the two
subsurface stratum at random locations; one sample will be collected at a
random depth down to 10 feet from each boring, to assess migration through
the vadose zone, for a total of 18 samples

Design for PCBs
– Alpha = 0.05; Beta = 0.20; Delta = 0.50 (50% of the AL)
– The decision makers agreed on collection of 24 surface samples from each of
the two surface strata; total of 48 samples using a stratified random design
– For the sub-surface, 24 borings/probes will be collected from each of the two
subsurface stratum at random locations, collected at a random depth down to
10 feet for a total of 48 samples
49 of 86
Approach 1 Sample Locations
CS
(Surface Strata)
Plan View
Former Pad
Location
Buffer
Zone
Runoff
Zone
0
0
50
15
100
30
150 ft
46 m
50 of 86
Approach 1 Sampling Design (cont.)
CS
51 of 86
CS
Remediation Costs*
DR
Description/Depth
#
1a Pad & Run-off Zone, 0-6”
1b Buffer Zone (excluding
Pad and Run-off area), 0-6”
2a Pad & Run-off Zone,
6”-10”
2b Buffer Zone (excluding
Pad and Run-off area),
6”-10”
Cost*
Area
(ft2)
12,272
42,884
Volume
(yd3)
227
794
$45,400
$158,800
12,272
4,318
$863,600
42,884
15,089
$3,017,800
* Assume $200 per yd3 for all COPCs
*Does not include layback area
52 of 86
Approach 1 Based Sampling Design

CS
Compare Approach 1 costs versus remediation
costs
– Approach 1 S&A costs
• $11,700 (Pb, U, TPH) + $21,600 (PCBs) = $33,300
– Remediation costs
• Cost to remediate surface soil under footprint of pad and
buffer area: $204,200
• Cost to remediate subsurface soil under footprint of pad and
buffer area: $3,881,400
53 of 86
Step 7- Optimize Sample Design
Information IN
From Previous Step
Decision Error
Tolerances
Gray Region
Actions
Information OUT
Review DQO outputs from Steps 1-6 to
be sure they are internally consistent
To Next Step
sample
designs
If noDevelop
sample alternative
design meets
the error
tolerances
within the budget: relax one or more of the
constraints
or request
more funding,
etc.
For each
design option,
select needed
mathematical expressions
Select the optimal sample size that satisfies the
DQOs for each data collection design option
Go back to
Steps 1- 6
and revisit
decisions.
Check if number of samples exceeds
project resource constraints
Yes
No
Optimal
Sample
Design
54 of 86
Design Approaches
Approach 2:
Dynamic Work Plan (DWP) & Field Analytical
Methods (FAMs)

Manage uncertainty by increasing sample density
by using field analytical measurements

Use DWP to allow more field decisions to meet the
measurement objectives and allow the objectives to
be refined in the field using DWP
55 of 86
Approach 2 Sampling Design

CS
Phase 1: Pb, U, TPH, PCBs
– Perform field analysis of the four strata on-site using XRF
(Pb & U), on-site GC (TPH), and Immunoassay (PCBs)
methods. Take into account the chance of false positives at
the low detection levels
– This will produce a worse-case distributions that will be
used to calculate the number of confirmatory samples for
laboratory analysis for the surface and below grade strata
56 of 86
CS
Approach 2 Sampling Design (cont.)

Phase 1: Pb, U, TPH, PCBs
– Provide detailed SOPs for performance of FAMs: XRF,
GC, & Immunoassay analysis
– Divide both surface strata into triangular grids
– Use systematic sampling, w/random start (RS), to locate
sample points; sample in center of each grid

Pad & Run-off zone





CSM expects contamination more likely here
10 ft equilateral triangle: 43.35 ft2
Pad + Run-off zone = 12,272 ft2
283 sample points
Buffer area:



Also 283 sample points
CSM expects contamination less likely here
Thus, grid triangle has larger area
57 of 86
CS
Approach 2 Sampling Design (cont.)

Phase 1: Pb, U, TPH, PCBs
– Sub-surface strata: Pad & Run-off zone






Use Direct Push Technology (DPT) to collect
Push at all surface sample points > ALs
Minimum sample locations: 40 (+ 10 >ALs) = ~50
Collect sub-surface samples every 3 feet
50 X 3 = 150 sub-surface samples in this strata
Use systematic sampling, w/random start (RS), to locate sample
points
– Buffer area





CSM expects contamination less likely here
Thus, fewer sample points
Same >ALs rationale as above
50 X 3 = 150 sub-surface samples in Buffer area
Use systematic sampling, w/RS, to locate sample points
58 of 86
Stratified Systematic Grid with Random Start
CS
(Surface Strata)
Runoff Zone
(Stratum 1)
N
Footprint of
Concrete Pad
(Stratum 1)
Buffer Zone
(Stratum 2)
Not to scale
Squares will be adjusted according to Step 7 design
59 of 86
Approach 2 Sampling Design (cont.)

CS
Phase 2: Pb, U, TPH, PCBs
– Evaluate the FAM results and construct FDs for each COPC
– Using Monte Carlo method, evaluate the alpha, beta and
delta and resulting n based on the XRF, on-site GC, and
Immunoassay data and select a value (worst case) for n to
confirm the FAM data, using traditional laboratory analysis
for each of the four strata
– For this Case Study, we will assume that number came out
to be 9 per strata or 36 confirmatory lab samples
60 of 86
Approach 2 Sampling Design (cont.)
CS
Surface Soils SC&SA Costs
U by Field XRF
Pb by Field XRF
TPH (GRO) on-site GC
TPH (DRO) on-site GC
PCBs by IMA kits
Total USA$
Unit Sample Collection Cost
AUSCA$ = USC$ + total USA$
$1.5
$1.5
$25
$25
$50
$103
$25
$128
61 of 86
Approach 2 Sampling Design (cont)
CS
Sub-surface Soils SC&SA Costs
U by Field XRF
Pb by Field XRF
TPH (GRO) on-site GC
TPH (DRO) on-site GC
PCBs by IMA kits
Total USA$
Unit Sample Collection Cost
AUSCA$ = USC$ + total USA$
$1.5
$1.5
$25
$25
$50
$103
$50
$153
62 of 86
Approach 2 Sampling Design (cont.)
COPC (Method)
U and Pb (XRF); TPH
(On-site GC); PCBs
(IMA kits); Surface
Soils (strata 1 & 2)
U and Pb (XRF); TPH
(On-site GC); PCBs
(IMA kits); SubSurface Soils (strata
1&2)
Total
Number of Samples, n
CS
Total SC&SA
AUSCA$ Cost
566
$128
$72,448
300
$153
$45,900
$118,348
63 of 86
Approach 2 Sampling Design (cont.)
CS
Confirmatory Traditional Laboratory Analyses Costs
Total Sampling
and Analytical
Number of Samples, n AUSCA$ Cost
Subtotal - Onsite
$118,348
Pb, U, TPH, PCBs Surface
18
$250
$4,500
Pb, U, TPH, PCBs Suburface
18
$275
$4,950
Subtotal - Lab
$9,450
Total Costs On-site and Lab Methods
$127,798
64 of 86
Approach 2 Sampling Design (cont.)

CS
Evaluate costs of Approach 2 vs. remediation
costs
– Sampling and analysis (S&A) costs $127,798
– Original budget for S&A $45,000
– Remediation cost
• Cost to remediate surface soil under footprint of pad and
buffer area: $204,200
• Cost to remediate subsurface soil under footprint of pad
and buffer area: $3,881,400
65 of 86
Approach 2 Sampling Design (cont.)
CS
Comparison
Costsn
# S amples,
n # S amples,
On-Site
Off-Site Total Sampling
(Surface /
(Surface / and Analytical
Approach S ub-surface) S ub-surface) Cost
1
none
66/66
$33,300
2
566/300
18/18
$127,798
Remediation Costs:
•Surface $204,200
•Sub-surface - $3,881,400
66 of 86
A Visual Decision Strategy
Start
Get
Data
Visual DQA
Visual Fit
Visual Test
Clean
Do
Data Check PDF
Dirty
Hypothesis
Data
Stop
Test
Fit Data
Need
More
Data
Get
Get
x, y
n
Sampling
Sample
Locations
Size
VSP
VESA
67 of 86
Approach 2b Sampling & Lab
Analyses


n=m*k
Select k of specified Mass/diameter3
– FE²  22.5 * d³ / M (to control sampling error)


Prepare m multi-increment samples for lab analysis
Perform lab analyses on m samples
Remember:
Sampling Uncertainty is decreased
when sampling density is increased
68 of 86
Collect
“n”
samples
Approach 2b Sampling & Lab
Analyses
n =m*k
Group
into “k”
Combine “k”
into “m”
composites
k=3
k=3
m=2
Laboratory
Remember;
we want the
AVERAGE
over the
Decision Unit
69 of 86
Approach 2b Sampling Design

CS
Phase 2b: Pb, U, TPH, PCBs
– Let n = 283 (for each Surface strata); n = 150 (for each Subsurface strata)
– Select appropriate values for m and k, based on cost and
managing uncertainty



k = 3 (Surface); k = 3 (Sub-surface); add $5 to SC cost
m = 94 (each Surface strata); Total = 188 Surface samples
m = 50 (each Sub-surface strata); Total = 100 Sub-surface samples
– Perform field analysis of the four strata on-site using XRF
(Pb & U), on-site GC (TPH), and Immunoassay (PCBs)
methods.
– Again, this will produce worse-case distributions that will be
used to evaluate  and  errors and to calculate the number
of confirmatory samples for laboratory analysis for the
surface and below grade strata; Still assume 36 total
70 of 86
Stratified Systematic Grid with Random Start
CS
(Surface Strata)
Runoff Zone
(Stratum 1)
N
Footprint of
Concrete Pad
(Stratum 1)
Buffer Zone
(Stratum 2)
Not to scale
Squares will be adjusted according to Step 7 design
71 of 86
Approach 2b Sampling Design (cont.)
SC Costs: Surface
U and Pb (XRF); TPH
(On-site GC); PCBs
(IMA kits); Surface
Soils (strata 1 & 2)
564
30
$16,920
188
$103
$19,364
SC Costs: Sub-Surface
U and Pb (XRF); TPH
(On-site GC); PCBs
(IMA kits); SubSurface Soils (strata
1&2)
Total
300
$55
$16,500
100
$128
$12,800
$65,584
CS
72 of 86
Approach 2b Sampling Design (cont.)
CS
Confirmatory Traditional Laboratory Analyses Costs
Total Sampling
and Analytical
Number of Samples, n AUSCA$ Cost
Subtotal - Onsite
$65,584
Pb, U, TPH, PCBs Surface
18
$250
$4,500
Pb, U, TPH, PCBs Suburface
18
$275
$4,950
Subtotal - Lab
$9,450
Total Costs On-site and Lab Methods
$75,034
73 of 86
Approach 2b Sampling Design (cont.)

CS
Evaluate costs of Approach 2b vs. remediation
costs
– Sampling and analysis (S&A) costs $75,034
– Original budget for S&A $45,000
– Remediation cost
• Cost to remediate surface soil under footprint of pad and
buffer area: $204,200
• Cost to remediate subsurface soil under footprint of pad
and buffer area: $3,881,400
74 of 86
Approach 2b Sampling Design (cont.)
Approach
1
2
3
CS
# S amples, n # S amples, n
On-S ite
Off-S ite
Total S ampling
(S urface /
(S urface / and Analytical
S ub-surface) S ub-surface) Cost
none
66/66
$33,300
566/300
18/18
$127,798
188/100
18/18
$75,034
Remediation Costs:
•Surface $204,200
•Sub-surface - $3,881,400
75 of 86
CS
Approach 2b Was Selected
Most Cost-Effective
and
Best Management of Uncertainty
76 of 86
CS
QC and Analysis Details
Used in All Approaches





Measure both gasoline & diesel range fractions
(GRO/DRO)
Ship & process all samples in one batch to
decrease cost.
QC defined per SW 846 [1 MS/MSD, 1 method
blank, 1 equipment blank (if equipment is reused),
1 trip blank for GRO only].
Cool GRO/DRO to 4°C, +/- 2°C.
QAP written and approved before implementation.
77 of 86
Iterative Process
Steps 1- 6
Step 7
Optimal Design
78 of 86
Step 7- Optimize Sample Design
Information IN
Actions
Information OUT
From Previous Step
Review DQO outputs from Steps 1-6 to
sure they are
internally
consistent
Justification for be
a judgmental
sampling
design
To Next Step
Timeframe
Decision•Error
• QualitativeDevelop
consequences
of ansample
inadequate
sampling
alternative
designs
Tolerances
design (low, moderate, severe)
• Re-sampling access after decision has been made
For
each design option, select needed
(accessible or
inaccessible)
mathematical expressions
Gray Region
Select the optimal sample size that satisfies the
DQOs for each data collection design option
Go back to
Steps 1- 6
and revisit
decisions.
Check if number of samples exceeds
project resource constraints
Yes
No
Optimal
Sample
Design
79 of 86
WARNING!!
If a judgmental design is selected in lieu of a statistical design the
following disclaimer must be stated in the DQO Summary
Report:
“Results from a judgmental sampling design can only
be used to make decisions about the locations from
which the samples were taken and cannot be
generalized or extrapolated to any other facility or
population, and error analysis cannot be performed
on the resulting data. Thus, using judgmental
designs prohibits any assessment of uncertainty in
the decisions.”
80 of 86
Step 7- Optimize Sample Design
Information IN
From Previous Step
Decision Error
Tolerances
Gray Region
Actions
Information OUT
Review DQO outputs from Steps 1-6 to
be sure they are internally consistent
To Next Step
Develop alternative sampleThe
designs
output is the most
resource-effective design for
the study
that is expected to
For each design option, select
needed
achieve the DQOs.
mathematical expressions
Select the optimal sample size that satisfies the
DQOs for each data collection design option
Go back to
Steps 1- 6
and revisit
decisions.
Check if number of samples exceeds
project resource constraints
Yes
No
Optimal
Sample
Design
81 of 86
Data Quality Assessment
Step 1: Review DQOs and Sampling Design
 Step 2: Conduct Preliminary Data Review
 Step 3: Select the Statistical Test
 Step 4: Verify the Assumptions of the Test
 Step 5: Draw Conclusions From the Data

Guidance for Data Quality Assessment,
EPA QA/G9, 2000
82 of 86
Summary
To succeed in a systematic planning process for
environmental decision making, you need
Statistical Support:
One or more qualified statisticians, experienced in
environmental data collection designs and statistical
data quality assessments of such designs.
83 of 86
Summary (cont.)


Going through the 7-Step DQO Process will
ensure a defensible and cost effective sampling
program
In order for the 7-Step DQO Process to be
effective:
– Senior management MUST provide support
– Inputs must be based on comprehensive scoping and
maximum participation/contributions by decision
makers
– Sample design must be based on the severity of the
consequences of decision error
– Uncertainty must be identified and quantified
84 of 86
Step 7- Optimize Sample Design
Information IN
From Previous Step
Decision Error
Tolerances
Gray Region
Actions
Information OUT
Review DQO outputs from Steps 1-6 to
be sure they are internally consistent
To Next Step
Develop alternative sample designs
For each design option, select needed
mathematical expressions
Select the optimal sample size that satisfies the
DQOs for each data collection design option
Go back to
Steps 1- 6
and revisit
decisions.
Check if number of samples exceeds
project resource constraints
Yes
No
Optimal
Sample
Design
85 of 86
End of Module 7
Thank you
86 of 86