Lessons LearnedLessons Learned

Multi Incremental Multi Incremental SamplingSampling

Alaska Forum on the Environment

February, 2009Alaska Department of Environmental

Conservation

Sampling Theory ReviewSampling Theory Review

Heterogeneity – The RuleHeterogeneity – The Rule

Impossible to sample the entire Impossible to sample the entire populationpopulation

Statistical methods must be used to Statistical methods must be used to determine a determine a representativerepresentative meanmean

Goal is to minimize sampling errorGoal is to minimize sampling error

Sampling ErrorSampling Error Compositional HeterogeneityCompositional Heterogeneity

Contributes to fundamental error (FE) - result Contributes to fundamental error (FE) - result of not representing proportional of not representing proportional concentrations of all of the particles in the concentrations of all of the particles in the population. population.

Distributional HeterogeneityDistributional Heterogeneity Contributes to grouping and segregation Contributes to grouping and segregation

error (GSE) – result of not collecting enough error (GSE) – result of not collecting enough random increments in enough locations to random increments in enough locations to capture spatial variability. capture spatial variability.

To minimize To minimize fundamental fundamental

error…error… collect enough collect enough

mass.mass.

To minimize grouping To minimize grouping and segregation and segregation error…error…

collect from many collect from many random locations.random locations.

Fundamental Error EquationFundamental Error Equation

m

dFE

)(20 3

WhereFE = Sampling fundamental error

20 = Sampling constant d = maximum particle size (centimeters)m = sample mass (grams)

GoalGoal

Maintain FE at 15% or lessMaintain FE at 15% or less

At least 30 g of sample analyzedAt least 30 g of sample analyzed 2 mm soil fraction2 mm soil fraction Grinding required for smaller Grinding required for smaller

sample size sample size

Composite or MI?Composite or MI? MI uses a defined decision unitMI uses a defined decision unit

BUTBUT Composite sampling does not Composite sampling does not

consider the decision unitconsider the decision unit

MI attempts to control FE and GSEMI attempts to control FE and GSE

BUTBUT Compositing is a simple combination Compositing is a simple combination

of discrete samples and does not of discrete samples and does not control FE or GSEcontrol FE or GSE

Decision Unit IdentificationDecision Unit Identification The area or volume in question The area or volume in question

(i.e. contaminated zone)(i.e. contaminated zone)

Systematic planning - Thorough documentation Systematic planning - Thorough documentation when setting decision unit boundarieswhen setting decision unit boundaries

Potential “dilution” effect and hot spot removal Potential “dilution” effect and hot spot removal must be consideredmust be considered

Decision units must be approved by DEC Decision units must be approved by DEC

Sampling LocationsSampling Locations

Increments collected from multiple Increments collected from multiple random locationsrandom locations

Different types of random sampling Different types of random sampling techniquestechniques Systematic random preferredSystematic random preferred

Sample depth considerationsSample depth considerations Sampling from the excavator bucketSampling from the excavator bucket

Current Procedure – Current Procedure – Non-VolatilesNon-Volatiles

Lab must meet MI-specific requirementsLab must meet MI-specific requirements Scoop at least 30 - 60 g into appropriate Scoop at least 30 - 60 g into appropriate

container from each random increment locationcontainer from each random increment location Sieve now or bag and sieve laterSieve now or bag and sieve later Sub-sample in field or labSub-sample in field or lab Approx. 500 – 1,000 g should be available after Approx. 500 – 1,000 g should be available after

sievingsieving Spread evenly and divide into sections (~30)Spread evenly and divide into sections (~30) One small scoop (about 1-2 g) from each One small scoop (about 1-2 g) from each

section into a 2-4 oz sample jar section into a 2-4 oz sample jar

Current Procedure - Current Procedure - VolatilesVolatiles

Volatile sample containersVolatile sample containers Small spoon, spatulaSmall spoon, spatula No sievingNo sieving Sample increments deposited into Sample increments deposited into

methanol at a minimum 1:1 ratiomethanol at a minimum 1:1 ratio Remove large clumps or rocks Remove large clumps or rocks 2-5 g from each increment location2-5 g from each increment location

QA/QCQA/QC

Triplicates collected to determine Triplicates collected to determine Relative Standard Deviation (RSD)Relative Standard Deviation (RSD)

Multiple, similar decision units may have Multiple, similar decision units may have a reduced triplicate sampling frequencya reduced triplicate sampling frequency

Do not collect triplicates from co-located Do not collect triplicates from co-located or adjacent locationsor adjacent locations

RSD is a measure of data precision expressed in RSD is a measure of data precision expressed in percentpercent Indication of representativeness of MI sampling of Indication of representativeness of MI sampling of

decision unitdecision unit

30% or less required30% or less required At RSDs >35%, the data distribution starts to become At RSDs >35%, the data distribution starts to become

non-normal and the confidence in the non-normal and the confidence in the representativeness on the MI sample results diminishes.representativeness on the MI sample results diminishes.

s

RSD100

95% UCL95% UCL

95 % UCL must be calculated for all 95 % UCL must be calculated for all decision unitsdecision units

Only the 95% UCL will be used to Only the 95% UCL will be used to evaluate the decision units evaluate the decision units

n

tsUCL %95

MI Sampling Projects ReviewMI Sampling Projects Review

ADEC Developed Draft Guidance in ADEC Developed Draft Guidance in March 2007March 2007

Approximately 40 projects have been Approximately 40 projects have been proposed using MI sampling since proposed using MI sampling since then with approximately 20 projects then with approximately 20 projects accomplished using MI samplingaccomplished using MI sampling

ADEC plans on updating the ADEC plans on updating the guidance in the near futureguidance in the near future

Lessons LearnedLessons Learned

  • Sample DryingSample Drying• Sample grindingSample grinding• SW 846SW 846• VOCsVOCs• Decision UnitsDecision Units• SievingSieving• Risk Assessment and ITRCRisk Assessment and ITRC

Sample DryingSample Drying

Sieving wet samples can be difficult and Sieving wet samples can be difficult and might leave material behindmight leave material behind

Based on limited information, drying Based on limited information, drying samples for semi-volatile and non-volatile samples for semi-volatile and non-volatile analyses has not shown a significant analyses has not shown a significant decrease in contaminant concentrations decrease in contaminant concentrations ((e.g. e.g. weathered DRO)weathered DRO)

Contact ADEC if sample drying will affect Contact ADEC if sample drying will affect holding timesholding times

Sample GrindingSample Grinding

Grinding may be required for Grinding may be required for samples to be analyzed for metals or samples to be analyzed for metals or any other analytes where the any other analytes where the analytical sample size is smallanalytical sample size is small

Some out of state labs are offering Some out of state labs are offering grinding and MI prepgrinding and MI prep

Likely to become more common as Likely to become more common as more MI samples are collectedmore MI samples are collected

SW 846- General Test SW 846- General Test MethodsMethods

EPA Method 8330B- ExplosivesEPA Method 8330B- Explosives

““Various studies have shown that Various studies have shown that concentrations of energetic residues concentrations of energetic residues at military training ranges that were at military training ranges that were measured using the procedures in measured using the procedures in 8330B 8330B

(MI Sampling) were statistically more (MI Sampling) were statistically more representative relative to traditional representative relative to traditional sampling and analytical protocols”sampling and analytical protocols”11

VOC’sVOC’s

Using a spoon or spatula with wide Using a spoon or spatula with wide mouth jar results in loss of volatilesmouth jar results in loss of volatiles

Updated guidance will recommend Updated guidance will recommend using an Encore using an Encore TM TM Sampler or other Sampler or other similar coring device that will reduce similar coring device that will reduce the loss of volatiles and a narrow the loss of volatiles and a narrow mouth jarmouth jar

Decision UnitsDecision Units Should include only the release area, Should include only the release area,

if knownif known

Alternative decision units may be Alternative decision units may be proposed, if impacted area is not proposed, if impacted area is not known or has been reworkedknown or has been reworked

Decision units must be clearly Decision units must be clearly identified in a work plan and must be identified in a work plan and must be approved by ADECapproved by ADEC

SievingSieving

Highly organic soil types such as peat Highly organic soil types such as peat are not conducive to sieving, therefore, are not conducive to sieving, therefore, MI sampling is not appropriate without MI sampling is not appropriate without alternate sample collection and alternate sample collection and preparation procedurespreparation procedures

Contact ADEC for additional information Contact ADEC for additional information on MI sampling for this matrixon MI sampling for this matrix

Risk AssessmentRisk Assessment

EPA Region X Risk Assessment EPA Region X Risk Assessment Conference in 2008 included a Conference in 2008 included a presentation on MI sampling presentation on MI sampling MI sampling may be accepted for use in MI sampling may be accepted for use in

Risk Assessments in the futureRisk Assessments in the future

Contact ADEC to discuss how this Contact ADEC to discuss how this might impact your Risk Assessmentmight impact your Risk Assessment

ITRC WorkgroupITRC Workgroup

Will begin working on a guidance Will begin working on a guidance document for MI sampling in 2009document for MI sampling in 2009

For Best Results, Use MI For Best Results, Use MI Sampling…Sampling…

To find the mean concentration of a To find the mean concentration of a contaminant in surface soil that is contaminant in surface soil that is conducive to MI sample preparation conducive to MI sample preparation and analysisand analysis

Where the decision unit is easily Where the decision unit is easily identifiedidentified

When you’ve considered the potential When you’ve considered the potential outcome and are comfortable with it.outcome and are comfortable with it.

Consider another sampling Consider another sampling method when…method when…

Soil type is not conducive to MI Soil type is not conducive to MI sample preparation and analysessample preparation and analyses

Identifying hot spotsIdentifying hot spots Delineating the extent of Delineating the extent of

contaminationcontamination Determining the maximum Determining the maximum

concentrationconcentration

Questions and DiscussionQuestions and Discussion

ReferencesReferences

1- DoD Environmental Data Quality 1- DoD Environmental Data Quality Workgroup, Workgroup, Guide for Implementing Guide for Implementing EPA SW-846 Method 8330B EPA SW-846 Method 8330B July 7, July 7, 2008.2008.