ndicators, Tracers and Surrogates - Why Use Them, Probability Analysis, Definitions and Examples”

Indicators, Tracers and Surrogates
Why Use Them, Probability Analysis, Definitions and Examples
AEHS, March 21st, 2017
Presented by Chris Lutes

2
A Key Term: Reasonable Maximum Exposure: Risk
Assessment Guidance for Superfund Part A:
• 6.4.1 QUANTIFYING THE REASONABLE MAXIMUM EXPOSURE
• “There are three categories of variables that are used to estimate intake….exposure
concentrations…..exposure frequency and duration……averaging time……
• Each intake variable in the equation has a range of values. For Superfund exposure assessments,
intake variable values for a given pathway should be selected so that the combination of all intake
variables results in an estimate of the reasonable maximum exposure for that pathway. As
defined previously, the reasonable maximum exposure (RME) is the maximum exposure that is
reasonably expected to occur at a site. Under this approach, some intake variables may not be at
their individual maximum values but when in combination with other variables will result in
estimates of the RME.
• Exposure concentration. The concentration term in the intake equation is the arithmetic average
of the concentration that is contacted over the exposure period. Although this concentration does
not reflect the maximum concentration that could be contacted at any one time, it is regarded as
a reasonable estimate of the concentration likely to be contacted over time. This is because in
most situations, assuming long-term contact with the maximum concentration is not reasonable.
…..Because of the uncertainty associated with any estimate of exposure concentration, the upper
confidence limit (i.e., the 95 percent upper confidence limit) on the arithmetic average will be
used for this variable.”
Quotes from OSWER 9285.701A; July 1989

3
Why Do We Need Indicators, Tracers and
Surrogates to Guide Sampling?
EPA 2015 VI guide definition: reasonable maximum exposure (RME)
A semi-quantitative term, referring to the lower portion of the high end
of the exposure distribution; conceptually, above the 90th percentile
exposure but less than the 98th percentile exposure.
Here we discuss the exposure concentration, but don’t forget that the
exposure frequency and exposure duration are also part of RME.
Observing the reasonable maximum exposure concentration with
random sampling is hard!

4
28
58
22
45
15
31
11
23
0
20
40
60
80
100
120
140
160
0.88 0.9 0.92 0.94 0.96 0.98 1
RequiredNumberofSamplestoObserveeRMEOnce
Percentile Defined as RME = Chance of Not Seeing RME With One Sample
Required Number of Unguided Random Samples Per
Location/Zone to Observe RME Once at Various Confidence Levels
5% Prob. Of Underestimating RME
This analysis is just the
mathematics of probability. No
assumptions about the
distribution have been made,
only the assumption of random
independent sampling. RME
defined as percentile.

5
58
28
13
8
6 4 3 2 2 10
10
20
30
40
50
60
70
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Surrogate Guided True Positive Rate = Chance of Seeing RME (Here defined as
95th Percentile with One Guided Sample)
Required Number of Surrogate Guided Samples Per Location/Zone
to Observe RME with 5% Probabilty of Underestimating
A surrogate does not need to be
perfect to be very helpful. It
just needs to “load the dice” by
significantly increasing the odds
of observing a sample toward
the top of the VOC distribution.
Note the 0.05 “guided
true positive” is a guide
no better then chance

6
Definition: Indicator
Metrics that can indicate elevated
potential for chlorinated VOC exposures.
Think:
• An indicator in chemistry is a substance whose
color varies with acidity or alkalinity.
• An indicator in biology is a species an animal or
plant species that can be used to infer
conditions of a particular habitat. For example
amphibians are sensitive to environmental
stress.
For VI season and differential pressure may be
indicators.
Images reprinted from: http://sciencefair.math.iit.edu/techniques/PHTesting/PHTesting.jpg
http://news.psu.edu/story/140702/2002/05/01/research/lessons-toads

7
Definition: Tracer
Easily observable substances that move
physically along with the target
compounds of interest for VI
Think:
• In hydrogeology and medicine an
identifiable substance, such as a dye or
radioactive isotope, that can be
followed through the course of a
mechanical, chemical, or biological
process.
For VI radon is a tracer of processes
across the slab.
Photos reprinted from: http://newsroom.unl.edu/announce/files/file9957.png
https://meyercancer.weill.cornell.edu/news/2016-01-25/tracer-treatment-radiopharmaceuticals-home-hard-detect-cancers

8
Definition: Surrogate
Metrics with a quantitative relationship
to the target compound for VI,
sufficiently accurate to be a substitute.
“In the context of environmental
microbiology and health risk assessment,
we have defined surrogates as organisms,
particles, or substances used to study the
fate of a pathogen in a specific
environment…. The use of surrogates in
these scenarios can allow quantification
of the degree of exposure” (Sinclair, 2012)
Sinclair, Ryan G., Joan B. Rose, Syed A. Hashsham, Charles P. Gerba, and Charles N. Haas.
"Criteria for selection of surrogates used to study the fate and control of pathogens in the
environment." Applied and environmental microbiology 78, no. 6 (2012): 1969-1977.
Images from:
https://water.me.vccs.edu/course
s/ENV295Micro/lesson9b.htm
https://images.tandf.co.uk/comm
on/jackets/amazon/978041921/9
780419218708.jpg

9
An Analogy
TO-15 total cost per
sample/data point >$350 leads
to very small sample sizes (i.e.
three sampling rounds)
A complex population
influenced by multiple
processes;
Rich information content
From a subsample, n=3
How much can you tell?
Could statistical analysis
alone help?
=

10
But Wait, I’ve Always Heard “Only EPA Standard
Method Results Can be Used in Risk Assessment”
Key elements of an exposure assessment include:
• “ Investigate patterns of exposure (e.g., frequency and duration)….
• Consider variability in exposures and appropriate exposure
distributions (e.g., the use of Monte Carlo or kriging).
• Establish descriptors of exposure, generally including estimates for
“average” and “high-end” exposures, as well as susceptible populations
or life stages. “
From US EPA Risk Assessment Forum (2014) “Framework for Human
Health Risk Assessment to Inform Decision Making” EPA/100/R-14/001.
Does the current approach do that well?

11
EPA Guidance for Data Usability in Risk Assessment
(1991)
• “Options for combining environmental analytical data
of varying levels of quality from different sources and
incorporating them into the risk assessment….
• Use data from different data sources together to
balance turnaround time, quatity (sic) of data, and
cost…..
• Three types of analytical data sources can be used
during the RI to acquire analytical data for a risk
assessment. These include field screening, field
analyses, and fixed laboratory analyses.

12
US EPA: Guidelines for Exposure Assessment (1992)
• “Exposure or dose profiles describe the exposure concentration or dose as a
function of time.. Such profiles are very important for use in risk assessment
where the severity of effect is dependent on the pattern by which the exposure
occurs rather than the total (integrated) exposure…..”
• “ Level of Detail: How accurate must the exposure or dose estimate be to
achieve the purpose? How detailed must the assessment be to properly account
for the biological link between exposure, dose, effect, and risk, if necessary?
How is the depth of the assessment limited by resources (time and money), and
what is the most effective use of those resources in terms of level of detail of
the various parts of the assessment?”
• “Surrogate data - Substitute data or measurements on one substance used to
estimate analogous or corresponding values of another substance.”
• Surrogate data is suggested as one of several methods to be used in addressing
data gaps.

13
Examples of Use of Indicators and Surrogates in Other
EPA Human Health Risk Management Programs
• Date of house construction, plumbing permits, plumbing codes etc. used as
surrogate for maximum lead risk in selecting sampling sites for lead in drinking
water EPA 816-R-10-004 March 2010
• “Total coliforms are a group of related bacteria that are (with few exceptions)
not harmful to humans. A variety of bacteria, parasites, and viruses, known as
pathogens, can potentially cause health problems ... EPA considers total
coliforms a useful indicator of other pathogens for drinking water. Total
coliforms are used to determine the adequacy of water treatment and the
integrity of the distribution system.” https://www.epa.gov/dwreginfo/revised-total-coliform-rule-and-
total-coliform-rule
• Total coliform, E coli or coliphages are used as a surrogates for a wide range of
pathogens in water. A range of surrogates of different specificity has been used as
technology evolves. https://www.epa.gov/sites/production/files/2015-10/documents/webinar-10-15-2015-
coliphage.pdf
• Good combustion practices (time, temperature) are used as indicators of
control of air toxics emissions https://www3.epa.gov/airquality/combustion/docs/20110221mboilersfs.pdf;
See also 40CFR Part 62, Subpart JJJ

14
Examples of Indicator, Tracer and
Surrogate Applications – Indianapolis
Applied R&D Projects
Spatial example from Wheeler Building –
coauthor Ron Mosley, Dale Greenwell
(EPA); Robert Uppencamp (ARCADIS)
Temporal examples from Indianapolis
Duplex – coauthors John Zimmerman and
Brian Schumacher (US EPA NERL_; Brian
Cosky (ARCADIS), Robert Truesdale, Brenda
Munoz and Robert Norberg (RTI
International)
420 Not
Heated
422
Heated

15
Example: Randomized Trial of Radon Tracer for
Spatial Sampling Application
• Fifty locations within the Wheeler
complex were screened for radon
• Then two subsets of these sample
locations were selected for
passive VOC sampling, one
randomly and the other based on
the radon.
• The upstairs radon guided
samples were significantly higher
in trichloroethene (TCE) than the
randomly selected locations
• t-test shows that the two mean
concentration values for the
guided and randomly collected
data are statistically different for
the data collected at the 95
percent confidence level.
0
5
10
15
20
25
Unit 106
(Column in
Kitchen)
2nd Floor,
On North
Wall, East
Side of
Atrium
Unit 134
(Column in
Center of
Unit)
Unit 224
(TV Stand
near
Exterior
Wall)
On
Papertowel
Dispenser
in Women’s
Restroom
Unit 154
(Shelf
Between
Bed and
Door)
Unit 106
(Column in
Kitchen)
Unit 148
(Shelf in
Kitchen)
On Coat
Rack in
Theater
Prop Room
Shelf at SE
Corner of
South File
Storage
Room
On Gate to
South
Overhead
Door in
Theater
Between
Basement
& 1s t Floor
on West
Stairs
Radon(pCi/L)andVOCs(µg/m3)
Sampling Locations
Results forUpstairs Sampling Locations
Radon
TCE
PCE
RADON GUIDEDRANDOM
Lutes, C.C., R. Uppencamp, L. Abreu , C. Singer, R. Mosley and D.Greenwell.
“Radon Tracer as a Multipurpose Tool to Enhance Vapor Intrusion
Assessment and Mitigation” poster presentation at AWMA Specialty
Conference: Vapor Intrusion 2010, September 28-30, 2010, Chicago, IL.
Available at
http://events.awma.org/education/Posters/Final/Lutes_RadonPoster.pdf

16
Introduction to Time Series Methods Used in
Indianapolis Project
• Sequential observations in time series are, in general, time-correlated
and thus, not independent of each other
• In this analysis used only consecutive, evenly spaced observations [i.e.,
daily (GC average) or weekly (passive) observations].
• Auto-correlation observed = the values at one point in the time series
are determined or strongly influenced by values at a previous time.
• Given the temporal resolution and length of available data sets we
expect to observe causes of change in indoor air concentration from
week to week and season to season; not year to year, and not within
the diurnal cycle.
• Time series is the statistically most correct way to analyze closely
spaced observations, but it isn’t likely to be a frequently used tool at
“practical” sites

17
Visualization of Time Series Data Fit – Change in
Exterior Temperature vs. Change in PCE
WarmingCooling
Decreasing
PCE
Increasing
PCE
Read more in “Simple, Efficient,
and Rapid Methods to
Determine the Potential for
Vapor Intrusion into the Home:
Temporal Trends, Vapor
Intrusion Forecasting, Sampling
Strategies, and Contaminant
Migration Routes.” U.S.
Environmental Protection
Agency, Washington, DC,
EPA/600/R-15/070, 2015.
http://cfpub.epa.gov/si/si_publi
c_record_report.cfm?dirEntryId
=309644

18
Example Time Series Results
Interpretation
Using on-line GC Data for basement PCE Dec 2012 – March 2013, mitigation
off, Exterior Temperature in F, the following equation was significant at the
99% confidence level:
Ct – C(t-1) = 0.748 + -0.078*Tt + 0.057*T(t-1)
C= Concentration T = Exterior temperature t= time
Note red and green coefficients similar but opposite in sign.
Example A: falling temp: 20 F today; 40 F yesterday, on average PCE
concentration increased 1.5 µg/m3.
Example B: rising temp: 20 F today, 10 F yesterday, on average PCE
concentration decreased 0.2 µg/m3

19
Key Time Series Conclusions from Indianapolis
• The change in the differential temperature and thus stack effect
strength was more important than the absolute value of the
differential temperature. Indoor air concentrations of VOCs are
expected to be high when the weather is getting colder, but would
not necessarily be expected to be as high during a period of
sustained cold weather. Not all winter sampling times are
equivalent.
• The most consistent relationship for barometric pressure is that an
elevated (greater than 30 inches) and/or rising barometric
pressure is associated with increasing vapor intrusion.
• No statistically significant effect of rain was seen.
• Increasing radon as a predictor was found to be statistically
significant at the 1% level and to predict 40-60% of the variability
in indoor air VOC concentrations.

20
Temperature Differential – Another Potential
Surrogate
Time Series Analysis Results
• The strength of the stack effect
predicted by the temperature
differential between the 422
basement south and outdoors
was significant at the 1% level.
• Increasing strength of the stack
effect was associated with higher
VOC concentrations, not primarily
high values in and of themselves
• Data from EPA/600/R-15/070 |
October 2015 ‘Third Indy Report’
422 Basement North Temperature (C)
422BasementSouthWeekly[PCE]
PCE Data from Indy Test House: Jan 2011 to Feb2012 (includes
locally weighted scatterplot smoothing line [blue], with a 95%
confidence interval [shaded])

21
Visualization of Time Series Data:
Change in indoor radon vs. change in indoor PCE
Increasing PCE
Decreasing
Radon
Decreasing PCE
Increasing Radon

22
What if This Time Series Stuff is Too Complicated for
Me?….What If I Just Look at the Thermometer or the
Radon Instrument Today and Use that To Guide
Sampling?
Image reprinted from www.homedepot.com and http://www.twu.edu/dsc/thermometer.jpg

New Analysis of ASU Sun Devil Manor
Data

ASU Sun Devil Manor Data Reprise
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
170 270 370 470 570 670
Sampling Day
Sun Devil Manor 24 hr Ave TCE (ppbv) vs Radon
24 h Average Radon in Indoor Air [pCi/L] 24 h Average TCE in Indoor Air [ppbv] -

SDM TCE vs Rn
R² = 0.37
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
24hrAveTCE(ppbv)
24 hr Ave Rn (pCi/L)
24hr Ave TCE vs 24 hr Ave Rn (SDM)

SDM TCE vs Differential Temperature
R² = 0.1607
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
-10 -5 0 5 10 15 20 25 30 35
24hrTCE(ppbv)
Differential T (C)
24 h Average TCE in Indoor Air [ppbv] vs Differential T (SDM)

SDM Differential Temperature vs Rn
R² = 0.2326
-10
-5
0
5
10
15
20
25
30
35
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
24hrDifferentialT
24 hr Rn (pCi/L)
24 hr Differential Temperature [°C] vs 24 hr Rn (SDM)

A Different Indicator Approach
 What if we use Radon or Differential
Temperature as an Indicator?
 What criteria work (at least for SDM)?
 How does this help us with the number of
samples required to identify the RME?
 *Note that this approach is very likely to seriously
overestimate the long term average exposure
concentration

Rn Indicator (>90th%) approach for RME
40% True Positives
60% False Positives

Differential T Indicator (>90th%) approach
for RME
Differential T>90th
percentile, TCE>95th
percentile
34% True
Positives
66% False
Positives

How Many Samples Are Needed?
58
28
13
8
6 4 3 2 2 10
10
20
30
40
50
60
70
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Surrogate Guided True Positive Rate = Chance of Seeing RME (Here defined as
95th Percentile with One Guided Sample)

95 UCL of Mean vs Number of Samples

New SDM Sampling Simulation Results &
Statistical Applications
(N=593 24 hr TCE points, 5000 simulations of random seasonal sampling)
Probability of 1 or more indoor air sample exceeding the Target Concentration
and
95UCL of Mean compared to percentile of total dataset for various sampling strategies
Seasons Sampled
4
season
s
Summer&
Winter
2
Winter
3
Winter
4
Winter
5
Winter
6
Winter
7
Winter
8
Winter
Total Samples 4 2 2 3 4 5 6 7 8
1 or more sample>90th% 34% 27% 47% 62% 73% 81% 86% 90% 93%
1 or more sample>95th% 19% 11% 26% 36% 45% 52% 59% 65% 70%
95 UCL OF MEAN=% OF
DISTRIBUTION 90TH%
>95.5
%
>95TH
%
>95TH
%
>94.5T
H%
>94TH
%
>94TH
%

Implications
 Indicators can provide guide for WHEN to
sample
 Indicators can limit the NUMBER of
samples
 Statistics for limited number of indicator
guided samples can provide a protective
estimate of the RME

Outline
Practical Methods for Using Indicators, Tracers, and Surrogates in Vapor
Intrusion Pathway Assessment
Presented by Chase Holton
AEHS, March 21st, 2017

38
Outline
1. Introduction
2. Indicators, Tracers, and Surrogates for
Vapor Intrusion Pathway Assessment
3. Related Guidance and Supporting
Information
4. Conclusions

39
1. Introduction
• 1.0 Introduction
– Describe general vapor intrusion (VI) pathway, issues associated with conventional
assessment approaches, long-term stewardship
– Define worst case conditions (spatially and temporally) based on current state-of-the-
science
– Describe current state-of-the-science of conventional sampling approaches, as well as
real-time/continuous monitoring approaches
– Short summary to define/differentiate indicators, tracers, and surrogates; describe
rationale for using alternative methods and need for document
• 1.1 Objectives
– Provide simple, efficient, and economically viable approaches for supporting VI
investigations (as a line of evidence), assessing the potential for VI,
estimating/projecting/forecasting potential risks/exposure,
– Provide insight on when conventional approaches should be applied
• 1.2 Document Scope
– Companion to statistical toolbox document
– Applicability to different building types, exposure scenarios

40
2. Indicators, Tracers, and Surrogates for Vapor
• 2.1 Characteristics of Potential Indicators, Tracers, and Surrogates
– Define/differentiate indicators, tracers, and surrogates using examples
from related fields to explain differences (examples shown in presentation
outline)
– Provide list of indicators, tracers, and surrogates included in past studies
(e.g., ~80 studied at USEPA research house in Indianapolis, not all
significant); highlight those that have shown greatest utility in VI
assessments
– Highlight best practices for use of radon, SF6, differential temperature,
differential pressure, and other methods/parameters that have shown
greatest utility in VI assessments

41
Intrusion Pathway Assessment (cont’d)
• 2.2 Uses and Utility within the Screening Process
– Define the screening process; building/site selection, prioritization, VI
susceptibility; discuss in context of conventional approaches
– Identify indicators, tracers, and/or surrogates applicable to screening
process and relative benefit (e.g., lower cost, higher confidence)
• 2.3 Uses and Utility within Detailed Assessments
– Define detailed VI assessments and provide a short summary of
conventional approaches
– Discuss approaches/tools/methods where indicators, tracers, and/or
surrogates can be applied and the relative value compared to other lines
of evidence.

42
• 2.4 Uses and Utility within Post-Mitigation/Remediation Phase
– Define mitigation/remediation, institutional controls, long-term
stewardship and provide a short summary of conventional approaches
– Identify indicators, tracers, and/or surrogates that can be applied to post-
mitigation (e.g. VIMS OMM) and long-term stewardship approaches

43
3. Related Guidance and Supporting Information
• 3.1 Incorporating Indicators, Tracers, and Surrogates into the Risk
Assessment Process
– Discuss purpose of risk assessments, estimation and projection of
risks/exposure, and how other approaches can be applied
– Discuss EPA guidance on data quality and usability requirements for risk
assessment and exposure assessment
– Present examples where indicators, tracers, and/or surrogates have been
used in HHRA, site management, decision-making

44
4. Conclusions
• Current assessment approaches rely on small number of point-
in-time samples analyzed by off-site laboratories, can result in
false-negatives; costly if implemented frequently
• Currently available continuous/real-time analytical instruments
and sensors have some limitations (costs, selectivity, sensitivity,
accuracy)
• Previous applications have demonstrated that certain variables
(e.g., differential temperature, radon) can be used to understand
variability, appropriate sampling periods, etc.
• Other USEPA guidance and/or studies supports use of indicators,
tracers, and/or surrogates for protection of public health

45
What Can You Potentially Do With a Good Tracer
or Surrogate?
• Guide sampling times and locations
• Find soil gas entry points to buildings
• Track mitigation system effectiveness
• Distinguish subslab from indoor sources of VOCs

46
Practical Aspects – Availability of Baseline
Comparison Data
• Differential temperature – almost always available, because
– Inside temperature often as easy as “where do you set your thermostat?”
– Historical outside temperature data for thousands of locations cataloged
either for specific period or normal values
https://www.ncdc.noaa.gov/cdo-web/
https://www.wunderground.com/history/ https://www.ncdc.noaa.gov/cdo-
web/datatools/selectlocation
• Radon – baseline comparison very building dependent, so needs to be
acquired
• Differential pressure:
o Beneficial to get some background data
o But upper end of range fairly similar for many structures (5 to 20 Pa) so may
be able to make useful decisions with limited background.
Temperature logger image reprinted from https://www.microdaq.com

47
Practical Aspects – Capital Cost of Monitoring
Equipment (not including data transmission)
• Differential temperature
o Range is from free to <$100 per
location
• Radon – as low as $129 for a device
that reads out the average of 48
hours to 7 days $250 for four hours
time resolution. Some have audible
alarms.
• Differential pressure - $405 for
resolution to 0.002” w.c. (0.5 Pa) Image reprinted from
www.homedepot.com

48
Practical Aspects – Advanced Predictability
• Temperature – readily
available forecasts,
probably sufficient short
term accuracy for purpose
• Radon – not easily
forecasted
• Differential pressure – may
be partially forecastable if in
a given structure controlled
by differential temperature
rather then wind Graphic reprinted from:
http://blog.extension.uga.edu/climate/2
015/07/when-weather-apps-go-bad/

Statistical Toolbox for RCRA
CVI Decisions
Lead in Slides

50
Introduction: Chlorinated VOCs in Indoor Air
• Challenging to assess at ‘low’ levels
• Difficult to predict
• Concentrations result from complex interaction of many variables
• Data-rich residential studies show significant variability:
– Spatial (x, y, z) – across neighborhood and within building scales
– Temporal (t) – diurnal, seasonal and climatic scales
Graphic adapted from:
http://www.nature.com/ncomms/2014/140708/ncomms5
344/images/ncomms5344-f2.jpg

51
Why Do Vapor Intrusion Plots Have Scatter?
Many Variables Involved – Radon Experience
“This paper identified about thirteen factors that can affect radon:
…soil moisture content, soil permeability, wind, temperature,
barometric pressure, rainfall, frozen ground, snow cover, earth
tides, atmospheric tides, occupancy factors, season and time of
day.
….. Four factors that influence radon concentrations indoors are
properties of the building material and ground; building
construction; meteorological conditions; and occupant activities.”
Lewis & Houle, A Living Radon Reference Manual (2009)
For mechanistic reasons it is likely that VOC VI is more variable and complex then radon
VI.

52
Current Practice
• Heavy reliance on extractive samples for laboratory analysis
– Costly (>$350 per sample) & disruptive to measure
– Only provide single ‘Points of Evidence’ (at a specific location &
time period) within spatial and temporal distributions
• The accuracy of inferences/extrapolations between & beyond the
‘points of evidence’ (samples) have rarely been tested (but ‘data-rich’
studies informative)
• Statistics is the scientific tool to provide defensible inferences from the
sample to the population. But statistical methods are rarely applied to
VI even though the variability in VI is more than for groundwater
• Analysis of performance of typical sampling strategies suggests safety
factor of at least 3X may be needed (Holton, 2013; Weinberg, 2014)

53
Reprise from Last Session
Why Study Statistics?
If we:
• don’t have guidance
on when to sample;
• don’t have the basis
for an assumption
about the distribution,
• don’t want to use an
additional safety factor
and
• do want to see a single
example of the RME
We will be sampling a
long time!
58
28
13
8
6 4 3 2 2 10
10
20
30
40
50
60
70
0 0.2 0.4 0.6 0.8 1
Surrogate Guided True Positive Rate = Chance of Seeing RME
(Here defined as 95th Percentile with One Guided Sample)

54
Occupational Methods for Estimating Exposure a/
• Identify similar exposure groups (SEGs)
– Based on similarity/frequency of tasks and types of materials/processes
• Number of samples needed for baseline assessment
– Based on statistical estimates/methods and SEG profile
– General AIHA recommendation: 6 – 10 samples where variability is expected to
be low to moderate
– Fewer than 6 samples can be used with caution; 3 is generally a minimum
a/ AIHA. 2015. A Strategy for Assessing and Managing Occupational Exposure. 4th Edition.
Ratio True
95th/OEL
Low Variability
(GSD=1.5)
GSD = 2
Moderate Variability
(GSD=2.5)
GSD = 3
High Variability
(GSD=3.5)
0.75 53 138 231 326 418
0.5 13 30 47 65 82
0.25 6 10 16 20 25
0.1 4 6 8 10 12
Sample Size Estimates (95% confidence; Power of 90%)
If goal is
for 95th to
be 10X
<OEL
GSDs at Industrial Buildings
Shown/Expected

55
Occupational Approach to Exposure Assessment:
Exceedance Fraction
• The exceedance fraction (f) is the probability that a concentration in a distribution is higher than
the screening level (AIHA, 2015).
• One property of a log-normal distribution is that the exceedance fraction can be related to the
mean, and more specifically the ratio of the screening level (SL) to the mean,
where Z is the value from a normal distribution table corresponding to the desired exceedance
fraction (Rappaport, 1991). For planning purposes, Z is used in lieu of the standard deviation
which generally is not known prior to sampling.
• This can provide a risk manager an indication of how far below a SL the mean concentration
needs to be to keep exposures below the desired exceedance fraction. For example if the
exceedance fraction is 5% (0.05), the mean should be maintained ~4-fold below the screening
level.
• How much sampling is needed to evaluate this situation with confidence? That will depend on
the variability of the distribution of concentrations in air in addition to the ratio of the screening
level to the mean concentration.
Rappaport, S.M. 1991. Assessment of Long-Term Exposure to Toxic Substances in Air. Annals of Occupational
Hygiene. 35(1): 61-121. See also Lowe, AEHS Conference 2016,

56
Jeff Kurtz notes that this material is his personal
opinion from experience at CDOT, Redfield and other
sites
Next Slide from Paul Johnson 2013• For full presentation see:
https://iavi.rti.org/attachments/WorkshopsAndConferences/05_
Johnson_03-19-13.pdf

57
Daily Average Concentration Data Set*
Fall FallSummer SummerWinter Spring Winter
August 14, 2012Aug 15, 2010
(08:00)
Spring Spring Summer
25 of 723 days (3.5%)
contribute 50% of
total exposure over
this time frame
0.01
0.1
1
10
-180 -120 -60 0 60 120 180 240 300 360 420 480 540 600 660 720
TCEinIndoorAir[ppbv]
Time [d]
Daily Average Concentrations Average (0.078 ppbv)
Median (<0.01 ppbv) 50% of Exposure (25 days >0.6 ppbv)

58
Statistical Toolbox Outline – Basic Questions
• Is VI pathway complete & significant at a site (that has GW
conc. >levels of concern & undergoing GW remediation)?
• Given that GW remediation can extend for decades, how can
we ensure, with a known degree of confidence, that VI
exposures are under control during the cleanup period?
Objectives
• Provide sound statistical methodologies to determine
appropriate frequencies & approaches for verifying protection
from CVI over long-term periods (with a known degree of
confidence)
• primary applicability is to RCRA VI decisions, but could be useful
for other types of VI sites

59
Organization and Scope
• Follows 2009 RCRA Statistical Analysis of Groundwater Monitoring Data at
RCRA Facilities (Unified Guidance) and supporting ITRC document, but applies
applicable methods to VI, structured by same or similar study questions.
• Also draws on other technical resources when Unified Guidance does not
address an important topic. Include the NIOSH Occupational Exposure
Sampling Strategy Manual, AIHA “A Strategy for Assessing and Monitoring
Occupational Exposures”, and ITRC’s “Groundwater Statistics and Monitoring
Compliance: Statistical Tools for the Project Life Cycle”. The web link to some
very useful and relevant Superfund Groundwater guidance for groundwater
statistics and their Excel-based groundwater statistics tool is:
• https://www.epa.gov/superfund/superfund-groundwater-groundwater-
response-completion
• Limits and Applicability: Not comprehensive, but does focus on many
decisions/questions a RCRA practitioner is likely to encounter during an
extended groundwater cleanup period at typical CVI sites.

60
Statistical Toolbox for RCRA VI – Considerations
• Background comparison
• Comparison to Screening Level
• Population Distribution
• Seasonality
• Trends
• Required sampling frequency
• Closure
Comparisons to RCRA/CERCLA
Groundwater Statistical Tools








61
ARE THE INDOOR AIR CONCENTRATIONS ABOVE
BACKGROUND?
• Start with lit values (mean or %); but background indoor air is
highly variable
• Are there indoor VOC sources? If so, can they be removed?
• Are indoor air concentrations above outdoor air concentrations?
• Are indoor air concentrations above background? - Probably a
“bright line” test against the selected background value.
• NAVFAC User’s Guide UG2-2091-ENV “Interim Final: Guidance
for Environmental Background Analysis; Volume IV: Vapor
Intrusion Pathway” has useful material on lines of evidence
evaluation and statistical tests for this purpose.

62
ARE INDOOR AIR CONCENTRATIONS ABOVE OR BELOW
A CRITERION?
• What is the criterion, ACUTE OR CHRONIC exposure?
• Specify certainty: e.g., max, 95th percentile , or 95% UCL of
mean? Selected certainty dependent on number of samples.
Maximum or 95th percentile typically used for small number of
samples.
• When will criteria be met? (base on rate of change over time)
• AIHA exposure assessment strategy, similar exposure group
concept & exceedance fraction concept from Rappaport provide
useful tools.
• Controls & engineering modeling can constrain variability of
indoor air due to VI.

63
IS THERE SEASONALITY?
•Most VI is seasonal, but not all; need to account for
when it is.
• Seasonality cannot be captured by the simple concept
of “winter worst”. Evidence suggests that “getting
colder” can be more important than being cold, in
some cases. Evidence shows that, even within one city,
buildings have very different seasonal patterns from
each other. Evidence also shows that southern climates
are likely different then northern US climates.
•Extreme values can’t be removed if using max or 95th
percentile

64
IS THERE A TREND? WHAT IS DIRECTION AND RATE OF
CHANGE OVER TIME?
• Trends can be used to predict future exposure, remediation
success, etc.
• Must account for seasonality, which can invalidate a trend test
• Start with scatter plot, linear regression
• Steady increase = increasing VI
• Confidence bands usually increase over (future) time unless GW
is being remediated & source term is decreasing, which is likely
to be common for RCRA sites under MNA
• Once trend is detected, rate of change over time can be
determined
• Trends & rates can be determined at the VOC source, indoor air,
or in between (subslab)

65
IS THE SAMPLING FREQUENCY APPROPRIATE
(TEMPORAL OPTIMIZATION)?
• Optimization and design of the monitoring program must assure
sample independence while covering site sufficiently & collecting
adequate data over an appropriate time period for proposed statistical
evaluations.
• In early stages, statistical design options should be considered such
that adequate number of samples are collected.
• For sites with existing long term monitoring data, sampling frequency
can often be reduced while still providing adequate data
• Required sampling frequency can be evaluated with statistical methods
that assess redundancy of sample results.
• How often do I need to sample to capture a trend?
• Cost effective sampling (CES) - rate of change (linear regression) vs
Trend (Mann-Kendal)

66
IS THE SPATIAL COVERAGE APPROPRIATE (SPATIAL
OPTIMIZATION)?
• Optimization and design of monitoring program must assure
sample independence while providing adequate spatial coverage
of site. Optimization can lead to decreasing or increasing number
of sampling points. Expected to result in decrease in locations
over time with MNA.
• Are sample locations redundant or should new locations be
added?
• Geostatistics, other spatial statistical methods applicable. –
unlikely to be applicable to indoor air, but might apply to GW,
external soil gas, and subslab soil gas.

Panel discussion – draft toolbox outlines – specific questions
1. We will begin the panel session with a review of the existing
outlines and any panel comments.
a) Indicators Toolbox Outline _ from Chase Holton Presentation Slides
b) Statistics Tool box Outline _ from Jeff Kurtz Presentation Slides

68
Indicators, Tracers and Surrogates Toolbox
1.0 Introduction
1.1 Objectives
1.2 Document Scope
2.1 Characteristics of Potential Indicators, Tracers, and Surrogates
2.2 Uses and Utility within the Screening Process
2.3 Uses and Utility within Detailed Assessments
2.4 Uses and Utility within Post-Mitigation/Remediation Phase
3.1 Incorporating Indicators, Tracers, and Surrogates into the Risk
Assessment Process
4.0 Conclusions

69
Statistical Toolbox for RCRA VI – Considerations
• Background comparison
• Comparison to Screening Level
• Population Distribution
• Seasonality
• Trends
• Required sampling frequency
• Closure
Comparisons to RCRA/CERCLA
Groundwater Statistical Tools








Panel discussion – draft toolbox outlines – general questions
1. Are there any community benefits or concerns for the Statistics or
Indicators approaches? How should they be addressed?
2. Would each product (Statistics or Indicators Toolboxes) be useful?
a) What are the best parts?
b) What would you add or change?
c) What would you not include?
3. Do you think indicators, tracers, and/or surrogates have promise
for greater use in the VI field? Why or why not?
4. What is your view on the utility of statistical methods for VI
decision making? At what stages of the project/types of sites?
5. What is your opinion on the directed sampling approach and its
ability to efficiently capture RME?
a) Is it statistically valid?
b) Where should it not be applied?

ndicators, Tracers and Surrogates - Why Use Them, Probability Analysis, Definitions and Examples”

More Related Content

Similar to ndicators, Tracers and Surrogates - Why Use Them, Probability Analysis, Definitions and Examples” (20)

More from Chris Lutes (20)

Recently uploaded (20)

ndicators, Tracers and Surrogates - Why Use Them, Probability Analysis, Definitions and Examples”