incremental sampling methodology – an innovative approach to soil sampling

Presented by Shannon H. McDonald, P.G.August 4, 2010

Soil sampling approach designed to obtain contaminant concentrations more representative of your exposure unit/domain

Collected discrete samplesInfluenced by sampler biasNot scientifically reproducible or, therefore,

defensibleExpensive to collect sufficient samples to

adequately represent a siteMay miss ‘hot spots’Often biased upward or downward

For Risk Assessments, commonly used either maximum concentration or 95% UCL concentration as representative concentration in exposure domain

Theory of particulate sampling originally developed by geologist Pierre Gy for mining industryDesigned to address “Seven Sampling Errors”

FUNDAMENTAL ERRORAddress by collecting and analyzing sufficient sample

mass

SEGREGATION ERRORAddress by collecting sample increments randomlyCollect samples in sufficient locations to capture spatial

variability

Formally developed:◦By EPA◦November 2006◦Method 8330B◦ For nitroaromatics (energetics and explosives)◦Military munitions sites

Commonly “Multi-Increment™ Sampling” “Multi-Increment” is trademarked by

Envirostat, Inc. Incremental Sampling Methodology? Began to be applied to other types of

contaminantsSemi-volatile organic compoundsMetalsPesticidesPCBs

1. Identify Decision Unit (DU)• Generally not the entire site; subdivide site

into several Decision Units• Size will depend on goal of investigation• Important not to bias or dilute

• Useless to draw around small source area• Should not include large areas known to be

uncontaminated (“dilution effect”)

2. Grid Decision Unit• Need sufficient increments to not “miss”

contaminationo Typically 30 to 50 increments (guidance and

theory)• Addresses distributional heterogeneity• “Systematic random” approach reduces bias

2. Grid Decision Unit (cont.)• Collect one sample increment per grid cell• Collect increments from same relative location

within each grid cell• Start at random point in first grid cell

• Skipped increments?• Because “systematic random”, SKIP (don’t

move) inaccessible increments (due to buildings, pavement, standing water, etc.)

• Otherwise, introduce bias• For example:

3. Collect same volume of soil at each increment• Work backward to calculate how much soil

you will need at the end• Depends on analyses to be performed and

how many replicate samples you plan to collect (discussed later)

• For a site with lead and PAH contamination, TTL collected about 120 grams (4 ounces) of soil per increment

3. Collect same volume of soil at each increment (cont.)• Can use soil probe, hand auger, geoprobe,

etc.• Collect from specified depth at each

increment location• Place each increment in intermediate

container together• For example:

4. Sieve entire sample• Typically use #10 sieve (200-mesh)• Remove fraction of sample >2 mm in diameter

(trash, gravel, etc.)• Sieve resultant “soil” into decontaminated 2nd

container (bucket?)• May require air-drying (no elevated temps!)

first• TTL used large stainless steel trays to air-dry no

more than 2 hours (SVOCs, metals, etc.)

4. Sieve sample (cont.)

4. Grind sample?• Some guidance suggests grinding sample

• Obtains uniform small particle size• Can further reduce Fundamental Sampling Error

• Depends on analytes• ADEM does not use grinding step

5. Subdivide sample• Spread sample out in thin layer (1/4-inch) on

decontaminated surface

5. Subdivide sample (cont.)• Grid into between 30 and 50 sections

5. Subdivide sample (cont.)• Collect equal amounts of soil from each

gridded section• Scale• Stainless steel scoop – FLAT (to scoop to

bottom of sample; collect all particle sizes)

5. Subdivide sample (cont.)• Place one scoop from each grid section into

sample container• Number of containers to fill varies by state

• Alaska and Hawaii recommend triplicate samples• Alabama recommends minimum of TEN ultimate

samples for each analyte

5. Subdivide sample (cont.)• After filling all sample containers:

Relative standard deviation?◦Some guidance suggests calculating the RSD of

analytical results Quality control measure Ideally <30% Indicates data distribution

Calculate 95% Upper Confidence Limit• Ideally requires 10 to 15 concentration measurements• Ensures representative concentration

Ongoing investigation/remediation projectHigh concentrations of PAHs and metalsApproximate 85-acre parcelDivided into 17 decision unitsUsing ISM on each unit, have been able to

receive CLOCs without further remediation on 12 zonesTwo zones pending CLOCsThree zones require more remediation

Need careful planning before going out into the field

Coordinate with regulators before beginning to understand expectations

Alaska – Draft Guidance – March 2009 - http://www.dec.state.ak.us/spar/csp/guidance/multi_increment.pdf

ITRC Methodology Resources and Links – http://www.itrcweb.org/teamresources_79.asp

Hawaii Technical Guidance Manual - Nov. 2009 - http://www.hawaiidoh.org/tgm.aspx?p=0402a.aspx

USACE Interim Guidance – July 2009 - http://www.hnd.usace.army.mil/oew/policy/IntGuidRegs/IGD%209-02v2.pdf

Questions / Discussion???

Shannon H. McDonald, P.G.

TTL, Inc.

4154 Lomac Street

Montgomery, Alabama 36106

(334)244-0766

smcdonald@ttlusa.com