incremental sampling methodology - michigan · 2016. 7. 20. · ism overview . 8 ¤ ism provides an...

INCREMENTAL SAMPLING METHODOLOGY

Evaluating Risk by Design: The Importance of Representative Sampling for Managing Environmental Risk

John Bradley, Geologist and Alisa Lindsay, P.E. Michigan Department of Environmental Quality – Remediation and Redevelopment Division Incremental Sampling Technical and Program Support (TAPS) Team

2016 MECC

q Incremental Sampling Methodology (ISM)Overview

q Decision Making and Error Reduction by Designq Development of Decision Units (DUs)q Case Studiesq Resources and Contactsq Q&A

Agenda 2

Incremental Sampling Methodology 2016 MECC


Have you ever wondered if you… 3

¨ have taken enough samples?¨ have missed a “hot spot”?¨ have found yourself confused or arguing about

what the data means?¨ have been less than confident in the decisions

you or others are making (based on thesedata)?

¨ could have more confidence in your samplingplans? Your data? Your decisions?

q Yields representative, reproducible, and defensible data

q Reduces sampling and laboratory error by design

q Reduces chances of missing or underestimating significant contamination

q Simplifies statistical analysis of data

ISM Overview 4



The Nature of Nature 5

Soil is like a Bowl of Cheerios • Contaminant concentrations can vary

significantly (e.g. above and below screening levels) between co-located samples;

• Collection and combination of a large number of “increments” is required to obtain a representative sample (aka “incremental sampling”).

Compliments of Roger Brewer (DOH, Hawaii)

• Data Quality Objectives (USEPA 1987): “Any sample located within the contaminated zone will identify the contamination.”

• Methods for Evaluating the Attainment of Cleanup Standards (USEPA 1989): “When there is little distance between points it is expected that there will be little variability between points.”

Soil is like a Bowl of Fruit Loops

Investigation Objective: What is the average concentration of chemical “X” in this “Decision Unit” bowl of cereal?

What we thought in the 1980s: What we know in the 2010s: 6

Incremental Sampling Methodology


Where are you?? 7


ISM Overview 8

¨ ISM provides an accurate average concentration for anyconstituent in a given area.

¨ Discrete sampling: Calculation of the averageconcentration from multiple samples (data interpretation)n Incremental sampling: Measurement of the average

concentration from a single sample (direct result)¨ In the world of risk assessment, the concentration term

(i.e. cleanup criterion) is a reasonable estimate of theconcentration of a hazardous substance that a receptorcan be exposed to over time (i.e. average concentration)without adverse effect.


Discrete Sampling Incremental Sampling

9


ISM Overview: Process 10

¤ Part 1: Define Project Objectives and Develop a Sampling Plan

¤ Part 2: Field Implementation

¤ Part 3: Laboratory Preparation and Analysis



¤ Part 1: Define Project Objectives

n Develop Conceptual Site Model

n Identify Potential Receptors and Exposure Pathways

n Establish Data Quality Objectives (DQO) using asystematic planning process (i.e. U.S. EPA Seven StepDQO Process)

ISM Overview: Process

¨ Step 1: State Problem

¨ Step 2: Identify Decision

¨ Step 3: Identify Decision Inputs

¨ Step 4: Define Study Boundaries

¨ Step 5: Develop Decision Rules

¨ Step 6: Specify Performance Criteria

¨ Step 7: Optimize Design

Why

What

Where

How

12

¤ Part 1: Define Project Objectives (cont.)

U.S. EPA Seven Step DQO Process




¤ Part 2: Field Implementation ¤ Select increment locations using a Systematic

Random Process

¤ Collect multiple (>30 to 100) increments of uniform size from the entire decision unit

¤ Combine increments to yield one sample:

n Non-volatile organic compounds (VOCs): Allincrements into a single 1 to 2 kilogram sample

n VOCs: All increments (~5 to 10g each) into a singlesample bottle, preserved with methanol (~1 mlMeOH/1g of sample)



¤ Part 2: Field Implementation (cont.) ¤ Collecting a Non VOC Incremental Sample



¤ Part 2: Field Implementation (cont.) ¤ Collecting a VOC Incremental Sample



¤ Part 3: Laboratory Processing and Analysis

n Air drying and sieving entire sample

n Particle size reduction (grinding) of entire sample, if required by Sampling Plan.

n Specialized grinding equipment; may not be available at all laboratories.

n May not be appropriate in some situations, such as for malleable metals.

n Increment sub-sampling (>30 increments) to provide representative aliquot for analysis



¤ Part 3: Laboratory Processing and Analysis (cont.)

LAB

ISM Laboratory Processing is designed to retain

representation of the sample collected in the field

q Decisions and Confidence

q Data Quality

q Statistics

q Determining Error (Uncertainty)

Decision Making and Error Reduction by Design

21



Decision Making and Error Reduction by Design: Decisions and Confidence

22



23



Ø TThink of the number of samples you take…henumber of samples you take…

24



Think of the size of the samples you take…size of a sample…

25



Ø TThink of the mass of the samples you take……

26



q Site is 1 cubic yardØ One part in a millionnOne penny in $10,000n 32 seconds out of a yearnOne drop in a 10 gallon aquarium

q Site is 1 acre and 3 feet deepØ One part in a billionnOne drop in a 10,000 gallon

swimming pool

27



28



uninformed

29


Decision Making and Error Reduction by Design: Data Quality

30


Decision Making and Error Reduction by Design: Data Quality

¨ Excerpt from Part 201

14) Approval by the department of remedial action based on the categorical standard in subsection (1)(a) or (b) shall be granted only if the pertinent criteria are satisfied in the affected media. The department shall approve the use of probabilistic or statistical methods or other scientific methods of evaluating environmental data when determining compliance with a pertinent cleanup criterion if the methods are determined by the department to be reliable, scientifically valid, and best represent actual site conditions and exposure potential.

31


Decision Making and Error Reduction by Design: Statistics

32

¨ In the world of risk assessment, the concentration term (i.e. cleanup criterion) is a reasonable estimate of the concentration of a hazardous substance that a receptor can be exposed to over time (i.e. average concentration) without adverse effect.

¨ ISM provides an accurate average concentration for any constituent in a given area.

n Discrete Sampling: Calculation of the average concentration from multiple samples (data interpretation)

n Incremental Sampling: Measurement of the average concentration from a single sample (direct result)



Inferential Statistics to the Rescue!

33



34



¨ Descriptive vs. Inferential Statistics

¤ Descriptive Statistics is a more direct way to quantitativelydescribe our data.

¤ Inferential Statistics infers from the sample to the population based on the characteristics of the sample.

n Inferential Statistics is a model based on (and limited by) whatyou sampled, the number of samples taken, and an assumeddistribution.

n Inferential Statistics assumes that the sample is representativeof site variability.

35


Decision Making and Error Reduction by Design: Determining Error

It Depends on Where (and How) You Sample



Enter Error It Depends on Where (and How) You Sample



Enter Error Enter Uncertainty

It Depends on Where (and How) You Sample



Uncertainty (aka Error)



40

¨ Use replicate sampling and data to evaluate error

¤ ISM typically uses triplicate sampling to evaluate error

¨ Relates to DQO Step 6: Specify Performance Criteria (how good does my data have to be to be valid input for decision-making).

¨ Replicates (and measuring error) are most important when results are close to the decision criteria, when establishing criteria (i.e. site-specific background), or closing a site – when there is less tolerance for error or uncertainty.



41

¨ The number of replicates depends on the DQO (Step 6)

¨ The error is most often evaluated in terms of RelativeStandard Deviation (RSD) (Coefficient of Variance as a%) of triplicates of specific DU(s) – this is an evaluationof the reproducibility (precision).

¨ %RSD = Standard Deviation/Mean, as a %

¨ Typical %RSD for ISM (DQO Step 6)

¤ Lab replicates < 20%

¤ Field replicates < 30%

q Proper identification of DUs is instrumental in ensuring that the data collected can be used to meet investigative goals and make the correct risk-based decision (i.e DQO).

q A DU is typically the smallest exposure area (volume), also taking into account other factors and parameters.

q Relates to DQO Step 4: Define Study Boundaries and Step 7: Optimize Design

Development of Decision Units (DUs) 42




¨ Factors to consider:

¤ Site conditions

n Surface Cover

n Buildings

n Storm Water Runoff

¤ Past land uses

¤ Current land uses

¤ Future land uses: Known and potential

¤ Zoning



¨ Factors to consider (cont.):



¨ Factors to consider (cont.):

¤ Stakeholder needs

n Purpose - DQO

n Due Care

n Characterization/Background

n Investigation

n Remedial Design

n Verification/Closure/No Further Action

n Budget

n Schedule



¨ Parameters to consider:¤ Contaminant Source

¤ Exposure pathways

¤ Receptors

¤ Chemicals of Concern

¤ Site Heterogeneityn Soil Type(s)

n Lithology

¤ Statutory requirement

¤ Remedial goals



¨ Example: Former Shooting Range¤ Factors:

n Site condition: Sloped terrain, heavy vegetation

n Current land uses: Municipal wood chip storage

n Past land uses: Small arms shooting range

n Layout of targets and firing line unknown

n Primary firing direction believed to be to the north

n Consider penetration depth

n Future land uses: Unknown, owned by municipality,zoned industrial



¨ Example: Former Shooting Range (cont.)¤ Stakeholder needs

n Purpose: No Further Action

¤ Parameters to consider:

n Exposure pathways: Direct Contact

n Receptors: Workers

n Statutory requirement: Non-Residential

n Remedial Goal: Non-Residential Contact Criteria forLead (900 mg/kg)



¨ Example: Former Shooting Range (cont.)

Primary Shooting

Lane




¨ Example: Former Shooting Range (cont.)


q Ford Wixom Plant

q The Moors Golf Course

q Orchardview Property

Case Studies 52



Case Studies: Ford Wixom Plant 53

¨ Data Quality Objectives (DQOs)

¤ 1) State Problem: Staged soils may be contaminated

¤ 2) Identify the Decision: If the staged soils exceed the appropriate criteria, the soils cannot be used as backfill.

¤ 3) Identify Decision Inputs: Analytical soil concentration data, MDEQ Non-Residential Criteria

¤ 4) Define Study Boundaries: 650 CY of staged soils

¤ 5) Develop Decision Rule: If soil concentrations exceed Non-Residential Criteria, then the soil will be used as on-site backfill. Otherwise, soil will be transferred to an off-site disposal facility.



¨ DQO (cont.)

¤ 6) Specify Performance Criteria: %RSD between replicates <30%

¤ 7) Optimize Design

n The stockpile (650 CY) was considered one DU.

n Fifty (50) increments were collected; one (1) increment perapproximately every seven (7) bucket loads.

n Triplicate samples were collected to assess the precision of the data.

n Samples were analyzed for VOCs, PNAs, PCBs, and metals

n Analytical data compared to Non-Residential Criteria



¤ Results and Outcome

n Analytical data identified concentrations werebelow Non-Residential Criteria for all analyzedconstituents.

n %RSD < 14%

n Sample results were compared to Part 201 CleanupCriteria and regional background concentrations asapplicable. Concentrations did not exceed Non-Residential Criteria in the DU.

n The stockpile material will be used for backfill asneeded.


Case Studies: The Mines Golf Course 56

¨ DQO

¤ 1. State the Problem: Arsenic levels in soils mayexceed the MDEQ Direct Contact Criteria for Arsenic.

¤ 2. Identify the Decision: If surface soils pose an unacceptable risk to users, then additional work will be needed for approval of the Documentation of Due Care Compliance (DDCC).

¤ 3. Identify Decision Inputs: Soil analytical data

¤ 4. Define Study Boundaries: The portion of the golf course that was formerly orchard.



¨ DQO (cont.)

¤ 5. Develop Decision Rules

n If mean value for Arsenic is > Non-Residential DCC,additional work will be needed for approval of theDDCC.

n If < Non-Residential DCC but > Residential DCC,approve DDCC with controls for residential trespass.

n If < Residential DCC, the site is no longer a facility.

¤ 6. Specify Performance Criteria: %RSD between replicates < 30%



¨ DQO (cont.)

¤ 7. Optimize Design

n ISM will be used to estimate the average concentrationof Arsenic in surface soils.

n 30 increments will be collected in each DU to produce asample.

n Increment locations will be selected using a systematicrandom process.

n Triplicate samples will be collected in one DU (DU #3)to estimate precision (%RSD).


8.62 mg/kg 34.1 mg/kg

5.29 mg/kg

7.29 mg/kg

8.46 mg/kg

41.9 mg/kg

17.9 mg/kg

DISCRETE SOIL SAMPLING FOR ARSENIC

59


< Res DCC (7.6 mg/kg) >Res DCC but <Non Res DCC >Non-Res DCC (37 mg/kg)


14 mg/kg

4.6 mg/kg

9.2 mg/kg 11 mg/kg

8.5 mg/kg

7.7 mg/kg

8.1 mg/kg

17 mg/kg

INCREMENTAL SAMPLING FOR ARSENIC

60


< Res DCC (7.6 mg/kg) >Res DCC but <Non Res DCC >Non-Res DCC (37 mg/kg)



¨ Performance Criteria: %RSD ≤ 30

¤ Discrete Sampling

n No replicates sampled

n %RSD for discrete samples = 82.9%

¤ ISM

n Triplicate samples collected from DU #3

n Original: sample: 9.2 mg/kg

n Duplicate Sample: 11 mg/kg

n Triplicate Sample: 8.5 mg/kg

n % RSD = 13.5%



¨ Outcome

¤ The DDCC Plan was approved with additional signage and a clear visual natural demarcation of the boundary between the golf course and the adjacent residential properties.

¤ A Restrictive Covenant was placed on the property restricting site use to commercial or industrial uses, as well as restricting commercial uses that could result in a “residential type exposure” such as educational uses, parks, community gardens, outdoor athletic fields, playgrounds, and zoos.

¤ The No Further Action Report was approved.


Case Studies: Orchardview Property 63

¨ DQO

¤ 1) State Problem: Does Arsenic-impacted soil from aformer orchard pose a risk for redevelopment of the property for residential housing?

¤ 2) Identify Decision: Is any risk mitigation needed prior to redevelopment, to address any potential direct contact hazards to future residents, due to former use as an orchard?

¤ 3) Identify Decision Inputs: Soil analytical data

¤ 4) Define Study Boundary: Historical orchard, approximately 9.38 acres



¨ DQO (cont.)

¤ 5) Develop Decision Rules:

n If the average soil concentration of arsenic in any exposure DU is > Residential Direct Contact Criterion (RDCC) for Arsenic, then there is a potential hazard to future residents and appropriate exposure controls or other risk mitigation measure(s) would be required prior to redevelopment.

n If the average concentration of Arsenic in any exposure DU is < RDCC, then the soil in the exposure DU does not pose a hazard and no response activity is required.



¨ DQO (cont.)

¤ 6) Specify Performance Criteria: % RSD ≤ 30 for triplicates

¤ 7) Optimize Design

n Approximate ¼ acre DUs designated for the residential portions of the Property

n Other DU sizes based on uses such as a children’s play area and a storm water retention pond.

n ISM samples collected from 0-6 and 6-12 inches in each DU area (i.e. separate upper and lower DUs)

n Approximately 50 increments/DU

n Three sets of triplicate samples



¨ Results and Outcome

¤ Check Performance Criteria: Triplicate data

% RSD = 7.9%

% RSD = 3.7%

% RSD = 7.9%

¤ Discrete sampling underestimated Arsenic concentrations:

In areas previously considered homogeneous (historical pesticide spray areas), discrete sampling can underestimate concentrations.

Samples from five of the eight DUs that were previously designated as “clean” using discrete sampling actually exceeded RDCC for Arsenic. Two other DU samples were at or just below the Arsenic RDCC.



¨ Results and Outcome (cont.)

¤ The Traverse City Housing Commission was able to use theISM data to defend their decision to move forward with their Phase II construction project and make remedial and redevelopment decisions with confidence.

q Resources

q Local MDEQ Points of Contact

Resources and Contacts 70



Resources and Contacts 71

¨ Resources

¤ MDEQ – RRD Incremental SamplingMethodology and Applications Draft Resource Materials n http://www.michigan.gov/documents/deq/deq-

rrd-IncrementalSamplingReference-3-26-2015_485537_7.pdf

¤ Interstate Technology Regulatory Council (ITRC) Incremental Sampling Methodology (2012) n http://www.itrcweb.org/ism-1/

http://www.itrcweb.org/Team/Public?teamID=11



http://www.itrcweb.org/ism-1/

Resources and Contacts: Local MDEQ Points of Contact

¨ John Bradley, Superfund¨ Matt Baltusis, Superfund¨ Joe DeGrazia, SE Michigan

¨ Sheryl Doxtader, Jackson¨ Bill Harmon, Superfund¨ Paul Knoerr, Grand Rapids

¨ Kirby Shane, MDEQ Lab

¨ Alisa Lindsay, Kalamazoo¨ Dave Slayton, Hazardous

Waste¨ Rebecca Taylor, Lansing

¨ John Vanderhoof, Cadillac¨ Joe Walczak, Superfund¨ Eric Wildfang, Toxicology

¨ Patricia Williams, SaginawBay

72


QUESTIONS?

MICHIGAN DEPARTMENT OF ENVIRONMENTAL QUALITY 800-662-9278

www.michigan.gov/deq Sign up for email updates

Follow us on Twitter @MichiganDEQ

73

¨ John BradleySenior GeologistDEQ-Superfund Section Geology and Defense Site Management Unit [email protected]

¨ Alisa Lindsay, P.E.Senior Environmental Quality AnalystDEQ-Remediation and Redevelopment Division Kalamazoo District Office [email protected]

2016 MECC

incremental sampling methodology - michigan · 2016. 7. 20. · ism overview . 8 ¤ ism provides an...

Documents