Federal Register Vol. 52, No. Monday, August 24, 1987 ...€¦ · 31948 HeinOnline -- 52 Fed. Reg. 31948 1987 This information is reproduced with permission from HeinOnline, under

Federal Register / Vol. 52, No. 163 / Monday, August 24, 1987 / Proposed Rules

ENVIRONMENTAL PROTECTIONAGENCY

40 CFR Part 264

[FRL-3213-1]

Statistical Methods for EvaluatingGround-Water Monitoring Data FromHazardous Waste Facilities

AGENCY: Environmental ProtectionAgency.ACTION: Proposed rule.

SUMMARY: EPA promulgated regulationsfor detecting contamination of groundwater at hazardous waste land disposalfacilities under the ResourceConservation and Recovery Act of 1976(RCRA). The statistical procedures usedto evaluate the presence of.contamination have been subject tocriticism and require improvement.Therefore, EPA today proposes to revisethese statistical procedures and requestscomments from the public to assist inthe regulatory development process.DATE: Written comments should be

/submitted on or before October 23, 1987.ADDRESSES: The original and two copiesof comments on this proposal should bemailed to the Docket Clerk, Office ofSolid Waste (WH-562), U.S.Environmental Protection Agency, 401 MStreet SW., Washington, DC 20460. Thedocket is open from 9:00 to 4:00 Mondaythrough Friday, except for Federalholidays. The docket number for thisrule is F-87-SGWP-FFFFF. The publicmust make an appointment to reviewdocket materials. Call (202) 475-9327 forappointments. The public may copy amaximum of 50 pages of material fromany one regulatory docket at no cost.Additional copies cost $.20/page.

The official docket for this regulation,including comments received by theAgency, is located in Room MLG 100,U.S. Environmental Protection Agency,401 M St. SW., Washington, DC 20460.FOR FURTHER INFORMATION CONTACT:For general information contact: RCRA/Superfund Hotline, Office of SolidWaste (WH-563C), U.S. EnvironmentalProtection Agency, 401 M Street SW.,Washington, DC 20460, telephone (800)424-9346, or (202) 382-3000. Fortechnical information contact JamesBrown, (202) 382-4658.SUPPLEMENTARY INFORMATION:

Preamble OutlineI. AuthorityII. Background

A. Concerns About Existing StandardsB. Suggested Changes Published in ANPRMC. Public Comments on ANPRM

Ill. Today's ProposalA. General Performance Standards

B. Basic Statistical Procedures andSampling Schemes

C. Effects of Proposed Changes on ExistingMonitoring Programs1. Detection Monitoring2. Compliance Monitoring

IV. Regulatory AnalysisA. State AuthorityB. Regulatory Impact AnalysisC. Regulatory Flexibility ActList of Subjects

I. Authority

These regulations are issued under theauthority of sections 1006, 2002(a), 3004,and 3005 of the Solid Waste DisposalAct (SWDA), as amended by theResource Conservation and RecoveryAct of 1976 (RCRA), as amended (42U.S.C. 6905, 6912(a), 6924, and 6925).

II. Background

Subtitle C of the ResourceConservation and Recovery Act of 1976(RCRA) creates a comprehensiveprogram for the safe management ofhazardous waste. Section 3004 of RCRArequires owners and operators offacilities that treat, store, or dispose ofhazardous waste to comply withstandards established by EPA that are"necessary to protect human health andthe environment." Section 3005 providesfor implementation of these standardsunder permits issued to owners andoperators by EPA or authorized States.Section 3005 also provides that ownersand operators of existing facilities thatapply for a permit and comply withapplicable notice requirements mayoperate until a permit determination ismade. These facilities are commonlyknown as "interim status" facilities.Owners and operators of interim statusfacilities also must comply withstandards set under section 3004.

EPA promulgated ground-watermonitoring and response standards forpermitted facilities in 1982 (47 FR 32274,July 26, 1982), codified in 40 CFR Part264, Subpart F. These standardsestablish programs for protecting groundwater from releases of hazardouswastes from treatment, storage, anddisposal units. Facility owners andoperators are required to sample groundwater at specified intervals and to use astatistical procedure to determinewhether or not hazardous wastes orconstituents from the facility arecontaminating ground water. Asexplained in more detail below, theSubpart F regulations EPA promulgatedin 1982 have generated criticism. EPA istoday proposing to amend theseregulations to respond to theseconcerns.

A. Concerns About Existing Standards

The current Part 264 regulationsprovide that the Cochran'sApproximation to the Behrens FisherStudent's t-test (CABF) or an alternativestatistical procedure approved by EPAbe used to determine whether there is astatistically significant exceedance ofbackground levels, or other allowablelevels, of specified hazardous wasteconstituents. Although the existing 40CFR Part 264 regulations have alwaysprovided latitude for the use of analternat e statistical procedure, concernshave been raised that the CABFstatistical procedure in the currentregulations may not be appropriate. Ithas been pointed out that: (1) Thereplicate sampling method is notappropriate for the CABF procedure, (2)the CABF procedure does notadequately consider the number ofcomparisons that must be made, and (3)the CABF does not control for seasonalvariation. Specifically, the concerns arethat the CABF procedure could result in"false positives" (Type I error), thusrequiring an owner or operatorunnecessarily to collect additionalground-water samples, to furthercharacterize ground-water quality, andto apply for a permit modification,which is then subject to EPA review. Inaddition, there is concern that the CABFmay result in "false negatives" (Type IIerror), i.e. instances where actualcontamination goes undetected. Thismay occur when the background data,which are used as the basis of thestatistical comparisons, are highlyvariable due to temporal, spatial,analytical, and sampling effects.

B. Suggested Changes Published inANPRM

As a result of these concerns, EPA isproposing to change both the statisticalprocedure and the samplingrequirements of the regulations, byrequiring that owners or operators morecompletely characterize thehydrogeology at the facility, andincluding in the regulations performancestandards which the statisticalprocedures and the sampling methodsmust meet. Statistical procedures andsampling methods meeting theseperformance standards.would have alow probability of indicatingcontamination when it is not presentand of failing to detect contaminationthat actually is present. The facilityowner or operator would have todemonstrate that a procedure isappropriate for the conditions at thatfacility and ensure that it meets theperformance standard outlined below.

31948

HeinOnline -- 52 Fed. Reg. 31948 1987

This information is reproduced with permission from HeinOnline, under contract to EPA. By including this material, EPA does not endorse HeinOnline.


In addition EPA has outlined severalsampling methods and types ofstatistical procedures that the Agencybelieves will meet the performancestandards.

EPA recognizes that the selection ofappropriate monitoring parameters isalso an essential part of a reliablestatistical evaluation. The Agencyaddressed this issue in a previousFederal Register notice (51 FR 26632;July 24, 1986).

In an advance notice of proposedrulemaking (ANPRM)} issued August 20,1986 (51 FR 29182), EPA solicitedinformation that would help evaluateapproaches to determining if a facility iscontaminating the ground water. TheAgency requested comments on thefollowing performance standards that itwas considering:

1. The procedure(s) and samplingrequirements must be protective ofhuman health and the environment.

2. The owner or operator mustdetermine the statistical distribution ofeach parameter or constituent listed inthe facility permit. The statisticalprocedure(s) must be appropriate for thedistribution. The owner or operatorcould demonstrate that the distributionsof constituents differ and, thus, morethan one procedure is needed at afacility.

3. The procedure(s) should have'a lowprobability of indicating contaminationwhen it is not present and of failing todetect contamination that is actuallythere. The owner or operator shouldconsider different numbers of samplepoints for different constituents orprocedures.

4. The procedure(s) should beappropriate for the hydrogeologic settingand the physical layout of the ground-water monitoring system.

5. The owner or operator shoulddescribe how observations below thedetection limit will be handled in theprocedure(s).

6. The owner or operator should,consider, or control for, seasonal andspatial variability and temporalcorrelation in developing theprocedure(s).

In addition, EPA identified certainstatistical procedures that wereexpected to meet these performancestandards and asked for available dataon the following:

1. How will the statistical proceduresmeeting the performance standardsperform in actual practice?

2. How sensitive will the proceduresbe to different distributions?

3. Are data available for EPA to use todetermine Type II error levels for theprocedures in light of the fact that TypeI error (indicating contamination when it

is not present) is crosely related to TypeII error (missing existing contamination].

4. Are there other statisticalprocedures or sampling requirementsthat minimize both Type I and Type IIerrors? Are there data showing thenumber of Type II errors expected underany alternate statistical procedure orsampling scheme?

5. Are there modeling or measurementtechniques that make it possible todetermine the flow path of the groundwater from an upgradient well to aparticular downgradient well, or toseveral adjacent downgradient wells?

6. Does transforming data to itslogarithm or square root improveconformance to assumptions of astatistical procedure or are thereappropriate procedures foruntransformed data?

7. If EPA uses a simple comparison ofmean concentrations rather than astatistical procedure, would, this haveacceptable Type I and Type 11 errorlevels?

8. What Type I and Type II errorlevels result for the identified;procedures when concentrations ofconstituents are below the detectionlimit? What error levels would result forother procedures?

9. Ground-water monitoring data maybe autocorrelated (i.e.,. variation inmonitoring parameters may becorrelated). Is there information on thedegree of autocorrelation at facilitiesand appropriate corrections such asadjusting the degrees of freedom ofstatistical tests or procedures that mightbe more appropriate for autocorrelated"data?

10. Are intra-well comparisonsappropriate for new facilities?

11. Is there available information thatcould be used to evaluate the frequencycontrol comparison and to determine anacceptable range for them?

C. Public Comments on ANPRM

No new data or information weresupplied by commenters in response tothese 11 specific questions. However,commenters expressed several concernswith the six proposed performancestandards. One concern was that someof the items in the ANPRM may notreally be performance standards forstatistical procedures. For example,numbers 1 and 4 are goals appropriateto the entire ground-water monitoringeffort, not just to determine theappropriateness of the statisticalprocedures. Numbers 2 and 5 werethought to be specific requirements thatthe owner/operator must. meet todevelop an acceptable procedure, ratherthan broad performance standards.Finally, the two Items that would

usually be construed as specifyingperformance of a statistical procedure,numbers 3 and 6, were thought to be toogeneral for a procedure to meet. Onecommenter suggested that the section onperformance standards be changed toevaluation criteria for determining theacceptability of a statistical test orprocedure.

A consensus of respondents said thatperformance standards should bespecific, yet flexible. Further. theyshould be site-specific to allow for theunique character of a site to beconsidered in defining the standard.

Another area of consensus was thatperformance standards should provide a"systems" approach to ground-watermonitoring. That is, the performancestandards should address the adequacyof the site characterization, the numberand siting of monitoring wells, thesampling frequency and technique, thelaboratory analytical methods, and thestatistical analysis of the data. All ofthese components interact to determinethe adequacy of a ground-watermonitoring methodology to protecthuman health and the environment.Several commenters felt that the lastperformance standard, pertaining toseasonal and spatial variability, should'be considered in developing thestatistical procedures and not justduring the data analysis. Consequently,EPA feels that a performance standardmust incorporate aspects of ground-water monitoring in addition to thestatistical techniques.

While some respondents called for theperformance standards to specifyrequirements, few suggested what thosespecific requirements should be. Theonly specific suggestion made was thatthe Type I error should be set at 0.01.However, the context of that commentmade it clear that 0.01 was merely inpreference to 0.05, and that selectionwas based on the desire to keep thefalse positive rate as low as possible.

One suggestion was that risk beconsidered in setting performancestandards and that the cost of meetingthe performance requirements should betaken into account. It was alsosuggested that the level of performanceto be required was a matter for policydecision by EPA. The required level ofperformance should be based on what isperceived as technically achievable. andwhich would result in an acceptable risklevel. The achievement of zero risk wasthought to be an unattainable, idealizedgoal.

In general, the respondents endorsedthe concept of a performance standard,but they provided little in the way ofconcrete suggestions as to what the

31949




performance standards should be. Theyraised many concerns, but suggestedfew answers. Some of the requestscontradict others, making it evident thatno single standard can satisfy. all of thecomments.

Il1. Today's Proposal

In today's proposal, EPA has electedto retain the idea of generalperformance requirements that theregulated community must meet. Thisproposal allows for flexibility indesigning statistical procedures to site-specific considerations. -

EPA has tried to bring a measure ofcertainty to today's regulations whileaccommodating the unique nature ofmany of the regulated units in question.Consistent with this general strategy,the Agency is establishing severaloptions for the sampling schemes andstatistical tests to be used in detectionmonitoring and, where appropriate, incompliance monitoring.

The Regional Administrator willspecify for each of the hazardousconstituents one or'more of thestatistical tests and sampling schemesdescribed in today's regulations. Indeciding which statistical test isappropriate, he will consider thetheoretical properties of the test, thedata available, and the hydrogeology ofthe site.

The Agency recognizes that there maybe situations where the statistical testsand sampling schemes specified may notbe appropriate. In such cases, it isnecessary to develop procedures thatare tailored to the specific performancestandards that every procedure mustmeet. Thus, today's regulations establishperformance standards for use by theRegional Administrator in determiningwhether an alternative procedure orscheme will be sufficiently protective ofhuman health and the environment.

A. General Performance Standards

EPA's basic concern in establishingtoday's performance standards forstatistical procedures is to achieve aproper balance between the risk that theprocedures will falsely indicate that aregulated unit is causing backgroundvalues or concentration limits to beexceeded (false positives) and the riskthat the procedures will fail to indicatethat background values or concentrationlimits are being exceeded (falsenegatives). Today's proposal is designedto address that concern directly. Thus,any statistical test or sampling scheme,whether specified in today's regulationsor an alternative to those specified,should meet the following performancestandards:

1. If the distributions differ amongconstituents, more than one proceduremay be needed. The owner or operatormust show that the normal distributionis not appropriate if using a statisticalprocedure which assumes the data arenot normally distributed. A "goodness offit" test should be used to demonstratethat the distribution assumptions are notviolated.

2. The statistical test is to beconducted separately for eachhazardous constituent. At each time thata test is done, the test for individualconstituents shall be done at a Type Ierror level no less than 0.01. A multiplecomparisons procedure may be used to -control the experimentwise error rate ata level no less than 0.05. However, theindividual well comparisons must havea Type I error no less than 0.01, whichmay make the experimentwise error rategreater than 0.05. The owner or operatormust evaluate the power and may berequired to increase the sample size toachieve an acceptable power level. Thesample size and sampling procedureshall be appropriate to the level of theType I and Type II errors and thedecision criteria.

3. The monitoring well system shouldbe in accordance with the naturalfeatures of the site. The owner oroperator must ensure that the number,locations, and depths of wells willdetect hazardous waste constituentsthat migrate from the wastemanagement area to the uppermostaquifer at the first sampling period aftersuch migration occurs.

4. The statistical procedure should beappropriate for the behavior of theparameters involved. It should includemethods for handling data below thelimit of detection. In cases where thereis a high proportion of values belowlimits of detection, the owner oroperator may demonstrate that analternative procedure is moreappropriate.

5. The statistical procedure shouldconsider, and if necessary control orcorrect for, seasonal and spatialvariability and temporal correlation inthe data.

In referring to "statistical procedures",EPA means to emphasize that theconcept of "statistical significance"must be reflected in several aspects ofthe monitoring program. This involves-not only the choice of a level ofsignificance, but also the choice of astatistical test, the samplingrequirements, the number of samples,and the frequency of sampling. Since allof these interact to determine the ability.of the procedure to detectcontamination, the statistical:procedures

must be evaluated in their entirety, notby individual components.

A set of specific numericalperformance standards that wouldachieve the proper balance betweenfalse positives and false negatives is notpossible due to site specific differences.The probability of correctly decidingthat a regulated unit is contaminatingground water (often expressed as thepower of a statistical test) cannot easilybe summarized by a single numberbecause the power of a test is related tothe magnitude of the difference between:two populations. Today's regulations donot attempt to express the idea of"exceeding background values orconcentration limits" in terms of anyminimum magnitude. This is becauseany statistically significant increase is acause for concern. A performancestandard related to the power of astatistical test would have to bespecified for every possible minimummagnitude that might be of concern.This is not feasible given the currentstate of knowledge about ground-watercontamination.

An alternative would be for EPA todecide what magnitude of increase it isconcerned about and to specify howpowerful the test would be for thatmagnitude of difference. However, theAgency does not believe it isappropriate to determine an amount ofcontamination that is acceptable foreach contaminant at each site. Also,there would remain the problem ofhaving to specify how powerful the testshould be for values above the minimumdifference of concern. EPA invitescomment on this issue.

B. Basic Statistical Procedures andSampling Schemes

Today's proposed regulations specifyfour types of statistical procedures todetect contamination in ground water.EPA believes that at least one of thesetypes of procedures will be appropriatefor a wide variety of situations. Toaddress situations where theseprocedures may not be appropriate, EPA'has placed a provision in today'sproposed regulations for the RegionalAdministrator to select an alternateprocedure. The suggested proceduresare based in part on suggestionsreceived in public comments.

1. A parametric analysis of variance(ANOVA) followed by multiplecomparison procedures to identifyspecific sources of difference. The.procedures will include estimation and'testing of the contrasts between themean of each downgradient well and theupgradient mean for each constituent.

31950



Federal' Register*/ Vol. 52, No. 163 / Monday, August 24, 1987 / Proposed Rules,

2. An analysis of variance (ANOVA}based on ranks followed'by multiplecomparison procedures to identifyspecific sources of difference. Theprocedure will include estimation andtesting of the contrasts between themedian of each downgradient well andthe median upgradient levels for eachconstituent.

3. A procedure in which a tolerance orprediction interval is established fromthe background (upgradient well) data,and the level of each constituent in eachdowngradient well is compared to itsupper tolerance or prediction limit.

4. A control chart approach whichwould give control limits for eachconstituent. If any downgradient wellhas a value outside the control limits forthat constituent, that would constitutestatistically significant evidence ofcontamination

5. Another-statisticat test procedurespecified by the Regional Administrator,.

,provided that it is protective of'humanhealth and the environment and meetsthe performance standards specified..

EPA has specified multiple statistical.and sampling procedures and' hasallowed for alternatives because no oneprocedure is appropriate for allcircumstances. EPA believes that thesuggested procedures are appropriatefor the site-specific- design and analysisof data from ground-water monitoringsystems, and they can account for moreof the site-specific factors thanCochran's Approximation to, the BehrensFisher Student's t-test (CABF) and theaccompanying sampling procedures inthe current regulation. The statistical'procedures specified in today'sregulations address the multiplecomparison problems and: provide fordocumenting and accounting for sourcesof natural variation.

EPA believes that the specified. tests.consider and control for naturaltemporal and spatial variation.Technical details and examples for eachspecified procedure will be included, in adraft guidance document which isanticipated to be available bySeptember, 1987. The decision on thenumber of wells needed in a monitoringsystem will be made on a site-specificbasis by the Regional Administrator andwill consider the statistical methodbeing used, the natural hydrogeology,and the sampling scheme.. The numberof wells must be sufficient to ensure ahigh probability of detecting ,contamination when it is present.. Todetermine which sampling scheme'should be used. the. RegionalAdministrator should consider existingdata and site characteristics includingthe possibility of trends and seasonality.The regulations establish three sampling

schemes and a provision for-analternative for use in detection andcompliance monitoring systems. Thesesampling schemes are:

1. Obtain a sequence of daily (ornearly daily) observations at least twicea year unless it is found that thisfrequency of sampling results inautocorrelation of the observations..

2. Obtain a sequence of weeklyobservations at least twice a year,provided that weekly observations arenot autocorrelated and that no seasonaleffects are present in the data.

3. Obtain monthly observationsprovided the data exhibit no seasonal'effects.

4. Use another' sampling scheduleappropriate to an alternative statistical'.procedure specified by the RegionalAdministrator. The' alternativeprocedure must be protective of humanhealth and the environment.

EPA believes that the above sampling'schemes will allow the use:of statisticalprocedures that will accurately defectcontamination. These different 'samplingscenarios were chosen to allow for theunique nature of the ground'watersystems beneath hazardous waste, sites.These sampling schemes will' giveproper-consideration to the temporalvariation of and correlation among- theground-water constituents. The specifiedprocedures require sampling data from

"upgradient wells, at the compliancepoint, and according to a specific testprotocol. The owner or operator should'use a background value determinedfrom data collected under one of thesescenarios if a test specified by theRegional Administrator' requires it, or-if'a concentration limit in compliancemonitoring is to be based' uponbackground data. A guidance documentunder development includes scenarios'for which each sampling scheme wouldbe appropriate

If the owner or operator uses astatistical method' based' on, adistribution other-than the-normal, hemust show the normal distribution is notappropriate. The same applies totransformations of the data. If the owneror operator desires to use atransformation, it must first be shownthat the untransformed data areinappropriate for-a normal theory test.There are several procedures for doingthis, some of which will be detailed inthe draft guidance document. Ifcontamination is detected by, one ofthese tests and the owner or operatorsuspects that the detection is an artifactcaused by some feature of the data otherthan contamination, the RegionalAdministrator may specify thatstatistical tests of trend, seasonalvariation, autocorrelation, or other

interfering aspects of the data beperformed in order to establish whetherthe result indicates genuine.contamination or-whether the resultarose from natural variation.

EPA recognizes that even where. thedistribution of a constituent is expectedto be normally distributed . there may besituations where the owner or operatorcan devise sampling procedures that aremore appropriate to the facility andwhich will provide reliable results.Therefore, today's regulations allow theRegional Administrator to approve such,procedures if he finds that theprocedures balance the risk of falsepositives and false negatives in atmanner comparable to that provided bythe above specified tests and that theymeet the. specified performancestandards. In. examining the:comparability of the suggestedprocedure, the Regional Administratorwill examine the ability of the procedureto provide a reasonable balancebetween' the risk of false positives andlfalse negatives. The Regional"Administrator will specify in the permitsuch things as the sampling frequencyand the sample size. for the alternativestatistical procedure.

The regulations indicate that the,procedure must provide reasonableconfidence that the migration ofhazardous constituents from a regulatedunit into and. through, the aquifer will' bedetected. (The reference to hazardousconstituents' does not mean that thisoption applies only to compliancemonitoring; the test also. applies tomonitoring- parameters and constituents.in the detection monitoring programsince they are surrogates indicating the'presence of hazardous constituents.)The protocols forthe specific tests,however, will be used as a generalbenchmark to define "reasonableconfidence" in the proposed' procedure.If the owner or operator shows that hissuggested testis, comparable' in itsresults to one of the specified- tests, thenit is likely to be acceptable under the"reasonable confidence" test. Theremay be situations, however; where itwill be difficult to directly compare theperformance of an alternative test to theprotocols for the specified. tests. In suchcases the alternative test will have-to, beevaluated on its own merits.

A situation that will' probably requirethe crafting of a specialized procedure isthe one in which.the background level ofa constituent either is below thedetection limit of the analytical methodsused or is recordedas a trace level ofthe.constituent..EPA believes thatappropriate statistical procedures canbe developed in such cases. EPA seeks

31951



31952 Federal- Register -/ Vol. 52, No. 163 / Monday, August 24, 1087 / Proposed Rules

comments from the public on whatmethods are available that rely onfacility-specific data and properlyapplied statistical methodologies.C. Effects of Proposed Changes onExisting Monitoring Programs

1. Detection Monitoring .

The detection monitoring programrelies on the finding of increases overbackground levels to define when aregulated unit is leaking. In addition, inmany situations, the concentration limitfor a particular hazardous constituentmay be set at the backgroundconcentration. Today's proposedregulations are designed to ensure thatreasonable methods are available forthe analysis of background groundwater quality and for determining if aspecific ground water standard isexceeded.

.The concentration of chemical-'constituents in ground water may..

fluctuate substantially over'time. Duringdifferent times of the year the rechargerates to ground water will vary,reflecting the differences in climate,rainfall, and other factors. Differingrecharge rates or other factors maycause seasonal variation in the .concentrations of chemical constituents-in ground water.

.EPA- believes that such variation inbackground concentrations should bedocumented ard considered in thestatistical test if it occurs. 'If such naturalvariation in background concentrationscan be documented, a statisticaltechnique that, accounts for suchvariation should be used, provided thatthis can be done without compromisingother regulatory objectives.

For detection monitoring, and, whereappropriate, for compliance monitoring,today's regulations provide threedifferent sampling schemes for gatheringground water data. The RegionalAdministrator will specify whichsampling scheme is to be used afterconsidering the data, the statistical 'testto be used, the presence of trends, andnaturaltemporal variation. Thesampling scheme should be designed sothat the majori components of temporalvariability can be characterized.

The different sampling proceduresincluded in today's regulations enablethe Regional Administrator to, specify asampling scheme that will allow fordocumentation of and adjustment for-natural temporal variabilities. In thepast, the tendency has been to abandonthe sampling scheme and to use analternate 'statistical approach when dataexhibited temporal variability. EPAbelieves that in most cases one of thelisted sampling schemes will still be

appropriate in this situation if acorrection factor or data transformationis applied.

-It is possible that hazardousconstituents could already be migratingdowngradient before a facility hasreceived a permit. EPA does not believe-it is-generally appropriate to allow suchcontamination to continue to migratewhile the owner or operator collectsbackground data for one year.Therefore, EPA will, whenever possible,rely on whatever reliable backgrounddata Is available to establishbackground values for the compliancemonitoring program. Our concern is thatdata collection should not cause a delayin characterizing background. Possiblemethods.which would avoid delays incharacterizing the site would besampling schemes 1 or 2.

Occasionally, additional backgroundsampling and analysis over time may beappropriate even where compliance.point concentrations exceed upgradientconcentrations, at a given point in time,if the Regional Administrator believes itreasonably possible that this differenceis-due to seasonal or spatial variation inground water quality. In this case, theRegional Administrator would considerwhether the rate of ground water flow(and any contamination) wassufficiently slow that additional time for.collection of ground water quality datawould not jeopardize the potential forsuccessful corrective action if it isdetermined to be necessary. TheRegional Administrator would not,however, allow time for additional data*gathering in cases where the initialdifference in compliance point andupgradient constituent concentration isabove potential seasonal variation.

The owner or operator who wants toaccount for seasonal variations in thebackground values has at least threeadditional options. One, he cananticipate the need for such data bycollecting upgradient data on AppendixIX constituents likely to be in leachatebefore the detection monitoring programindicates that leakage has occurred.Two, he may continue to collectbackground data after the compliancemonitoring program permit is issued.Three, he may use that data in making ademonstration that an apparent increaseover concentration limits in the groundwater protection standard was causedby contamination from other sources. He.may also use the data in seeking apermit modification to change thebackground values contained in thecompliance monitoring program.

Another issue in the establishment ofbackground for a constituent is thequestion of which wells should be usedIn the data base. In evaluating temporal

variability, the frequency of samplingand the statistical method used toanalyze .the sampling data should bedesigned and consideredsimultaneously.

One option, which EPA is consideringfor detection monitoring programs atnew facilities and possibly also atfacilities where it is known that groundwater is not contaminated, is to collectbackground data by monitoringdowngradient wells. In eachdowngradient well, ground-water.concentration levels would bemonitored and the resulting data wouldbe used to establish a background limitthat would be unique to eachdowngradient well. The advantages ofthis approach are that first, the influenceof spatial 'variability would be removed;second, a control chart class ofstatistical thresholds could bedeveloped for each downgradient well;and third, upgradient wells may not berequired.The disadvantages are thatfirst, there must be assurance that,contamination from the unit is not beingfactored into establishment of thebackground data base, and; second, ifthis method is used and contaminationunrelated to the regulated unit flowsunder the downgradient side of the unitafter establishment of the backgroundlimits; a false positive may result.

EPA believes that this method bestapplies to newer units that have had noopportunity to contaminate the groundwater and that are located in areas withlittle potential to be influenced fromexternal sources unrelated to the unit. Itis 'also clear that this method can onlybe applied the first time a facility is indetection monitoring. A facility whichbegins compliance monitoring andsubsequently 'returns to detectionmonitoring will not be allowed toestablish or reestablish backgroundlimits based on'data collected in thedowngradient wells. EPA seekscomment on the utility of allowingdowngradient wells to serve asbackground in these situations.

A second option, which EPA believes.is preferable in most situations, andwhich corresponds to the currentregulatory. approach, is to basebackground data on upgradient wells.Assuming these wells are properlyplaced andunaffected by the unit; they,should produce data that are not biasedby contamination from the unit.

2. Compliance Monitoring

The statistical procedure(s) that willbe-used in compliance monitoring willbe contingent upon the type ofconcentration limit used. If theconcentration limit-is based upon the



Federal Registe~r / Vol. 52, No. 163 / Monday, August 24, 1987 / Proposed Rules

background values, then one of thestatistical procedures specified-in theregulations may be used. If theconcentration limit is based upon amaximum concentration limit or analternate concentration limit, then theRegional Administrator will require anappropriate test. In many cases this isthe tolerance interval procedurespecified in the proposed regulations.

IV. Regulatory Analysis

A. State Authority

Under section 3006 of RCRA, EPAmay authorize qualified States toadminister and enforce their Statehazardous waste management programsin lieu of EPA operating the Federalprogram in those States. Authorization,either interim or final, may be granted toState programs that regulate theidentification, generation,transportation, or operation of facilitiesthat treat, store, or dispose of hazardouswaste. Upon authorization of the Stateprogram, EPA suspends operationwithin the States of those parts to theground-water monitoring requirementsfor land-based hazardous wastemanagement facilities applying for andoperating under permits. Since theground-water monitoring requirementsare not imposed under any of theamendments made by the Hazardousand Solid Waste Amendments of 1984,final rules modifying the statisticalprocedures would not take effectdirectly in all States under section3006(g). Rather, if EPA promulgates thisproposal, States that have been grantedfinal authorization will have to revisetheir programs to cover the additionalrequirements in today's announcement.Generally, these authorized Stateprograms must be revised within oneyear of the date of promulgation of suchstandards, or within two years if theState must amend or enact a statute inorder to make the required revision. See40 CFR 271.21. However, States mayalways impose requirements which aremore stringent or have greater coveragethan EPA's programs.

Regulations which are broader inscope, however, may not be enforced aspart of the federally-authorized RCRAprogram.

B. Regulatory Impact AnalysisExecutive Order 12291 (46 FR 13191,

February 9, 1981) requires that aregulatory agency determine whether anew regula tion will be "major" and if so,that a Regulatory Impact Analysis beconducted. A major rule is defined as aregulation that is likely to result in:

1. An annual effect on the economy of$100 million or more;

2. A major increase in costs or prices.for consumers, individual industries,Federal, State, or local governmentagencies or geographic regions; or

3. Significant adverse effects oncompetition, employment, investment,productivity, innovation, or the ability ofUnited States-based enterprises tocompete with foreign-based enterprisesin domestic or export markets.

Therefore, the Agency has determinedthat today's proposal is not a major rule.Today's action should produce a netdecrease in the cost of ground-watermonitoring at each facility. Thisproposal has been submitted to theOffice of Management and Budget(OMB) for review in accordance withExecutive Order 12291.

C. Regulatory Flexibilfty Act

Pursuant to the Regulatory FlexibilityAct, 5 U.S.C. 601 et seq., whenever anagency is required to publish a generalnotice of rulemaking for any proposed orfinal rule, it must prepare and makeavailable for public comment aregulatory flexibility analysis whichdescribes the impact of the rule on smallentities (i.e., small businesses, smallorganizations, and'small governmentaljurisdictions). The Administrator maycertify, however, that the rule will nothave a significant economic impact on asubstantial number of small entities. Asstated above, this proposal will have noadverse impacts on businesses of anysize. Accordingly, I hereby certify thatthis proposed regulation will not have asignificant economic impact on asubstantial number of small entities.This regulation therefore does notrequire a regulatory flexibility analysis.

List of Subjects in 40 CFR Part 264

Hazardous material, Reporting andrecordkeeping requirements, Wastetreatment and disposal, Ground water,Environmental monitoring.

Date: August 14, 1987.Lee M. Thomas,Administrator.

Therefore, it is proposed that 40 CFRChapter 1 be amended as follows:

PART 264-STANDARDS FOROWNERS AND OPERATORS OFHAZARDOUS WASTE TREATMENT,STORAGE, AND DISPOSALFACILITIES

1. The authority citation for Part 264continues to read as follows:

Authority: Secs. 1006,2002(a), 3004, and3005 of the Solid Waste Disposal Act, asamended by the Resource Conservation and:Recovery Act, as Amended (42 U-S.C. 6905,,6912(a), 6924, and 6925).

2. In § 264.91 by revising..paragraphs(a)(1) and (a)(2) to read as follows:

§ 264.91 'Required programs.

(a) * * *

(1) Whenever hazardous constituents'.under § 264.93 from a regulated unit aredetected at the compliance point under§ 264.95, the owner or operator mustinstitute a compliance monitoringprogram under § 264.99. Detected isdefined as statistically significantevidence of contamination as describedIn § 264.98(e);

(2) Whenever the ground-waterprotection standard under § 264.92 isexceeded, the owner or operator mustinstitute a corrective action programunder § 264.100. Exceeded is defined asstatistically significant evidence ofincreased contamination as described in§ 264.99(d);* * * * *

3. Section 264.92 is revised to read asfollows:

§ 264.92 Ground-water protectionstandard.

The owner or operator must complywith conditions specified in the facilitypermit that are designed to ensure thathazardous constituents under § 264.93detected in the ground water from aregulated unit do not exceed theconcentration limits under § 264.94 inthe uppermost aquifer underlying thewaste management area beyond thepoint of compliance under § 264.95during the compliance period under§ 264.96. The Regional Administratorwill establish this ground-waterprotection standard in the facility permitwhen hazardous constituents have beendetected in the ground water.

4. In § 264.97 by removing the word"and" from the end of (a)(1), adding(a)(1)(i) and (a)(3), revising paragraphs(g) and (h), and adding (i), (j), and (k) toread as follows:

§ 264.97 General ground-water monitoringrequirements.

(a) * *

(1) * * *

(i) A determination of backgroundquality may include sampling of wellsthat are not upgradient from the wastemanagement area where:

(A) Hydrologic conditions do notallow the owner or operator todetermine what wells are upgradient;and

(B) Sampling at other wells willprovide an indication of backgroundground-water. quality that is • -representative or more -representative

I31953-



Federal' Register 1 Vol. 52, No. 163 / Monday, August 24, 1987 1 Proposed Rules

than that provided by the upgradientwells; and

(3) Provide statistically significantevidence that hazardous waste orhazardous constituents that migratefrom the waste management area to theuppermost aquifer will be detected.

(g) In detection monitoring or whereappropriate in compliance monitoring,data on each hazardous constituentspecified in the permit will be collectedfrom background wells and wells at thecompliance point(s). The samplingprocedures and frequency must beprotective of human health and theenvironment. The samplingrequirements must ensure that thestatistical procedure has an acceptablylow probability of failing to detectcontamination. TheRegi6nal 'Administrator will specify one or moreof the following requirements in thepermit for each hazardous constituent tobe monitored for:

(1) Obtain a' sequence of daily (or neardaily) samples at least twice a year,unless it is found that this frequency ofsampling results in autocorrelation ofthe observations that cannot becorrected by the statistical procedureused.. (2) Obtain a sequence of weeklysamples it least twice a year, providedthat weekly observations are notautocorrelated and that no seasonaleffects are present in the data used ineach periodic comparison.

(3) Obtain monthly samples providedthe data exhibit no seasonal effects.

(4) Use an alternate samplingprocedure specified by the RegionalAdministrator under paragraph (i) ofthis section;

(h) Based on the factors in § 264.97(i),the Regional Administrator will specifyone of the following statisticalprocedures to be used in combinationwith the sampling requirements for eachhazardous constituent:

(1) A parametric analysis of variance(ANOVA) followed by multiplecomparisons procedures to identifystatistically significant evidence ofcontamination. The procedure mustinclude estimation and testing of thecontrasts between each downgradientwell's mean and the upgradient meanlevels for each constituent.

(2) An analysis of variance (ANOVA)based on ranks followed by multiplecomparisons procedures to identifystatistically significant evidence ofcontamination. The procedure mustinclude estimation and testing of thecontrasts between each downgradient

weirs median and the backgroundmedian levels for each constituent.

(3) A tolerance or prediction intervalprocedure in which a tolerance intervalfor each constituent is established fromthe distribution of the background data.and the level of each constituent in eachdowngradient well is compared to theupper tolerance or prediction limit.

(4) A control chart approach that givescontrol limits for each constituent.

(5) Another statistical test procedurethat is protective of human health andthe environment and meets theperformance standards specified in§ 264.97(i).

(i) The Regional Administrator canestablish an alternative samplingprocedure and statistical test forhazardous constituents that he finds willbe protective of human health and theenvironment. In establishing thatprocedure, the Regional Administratorwill consider the following factors:

(1) If the distributions for differentconstituents differ, more than oneprocedure may be needed. The owner oroperator must show that the normaldistribution is not appropriate if using a

* nonparametric or other methodology notrequiring an assumption of normality.For any statistic not based on a normaldistribution, a goodness of fit test shallbe conducted to demonstrate that thenormal distribution is not appropriate.Other tests shall be conducted todemonstrate that the assumptions .of thestatistic or distribution are not grosslyviolated.

(2) Each hazardous constituent is tobe tested for separately. At each timethat a test is done, the test for individualconstituents shall be done at a Type Ierror level no less than 0.01. A multiplecomparisons procedure may be used ata Type I experimentwise error rate noless than 0.05. The owner or operatormust evaluate the ability of the methodto detect contamination that is actuallypresent and may be required to increasethe sample size to achieve an acceptablepower level.

(3) The monitoring well system shouldbe consistent with § 264.97(a). Theowner or operator must ensure that thenumber, locations, and depths ofmonitoring wells will detect hazardousconstituents that migrate from the wastemanagement area to the uppermostaquifer.

(4) The statistical procedure should beappropriate for the behavior of theparameters involved. It should includemethods for handling data below thelimit of detection. The owner or operatorshould evaluate different ways ofdealing with values below the limit ofdetection and choose the one that is-most protective of human health and the

environment. In cases where there is ahigh proportion of values below limits ofdetection, the owner or operator maydemonstrate that an alternativeprocedure is more appropriate.

(5) The statistical procedure usedshould control for seasonal and spatialvariability and temporal correlation.

(j) Ground-water monitoring datacollected in accordance with paragraph(g) of this section including actual levelsof constituents must be maintained inthe facility operating record. TheRegional Administrator will specify inthe permit when the data must besubmitted for review.

(k) If contamination is detected byany of the statistical, tests, and theRegional Administrator or the owner oroperator suspects that the detection isan artifact caused by some feature ofthe data other than contamination, theRegional Administrator may specify thatstatistical tests of trend, seasonalvariation, autocorrelation, or otherinterfering aspects of the data be doneto establish whether the significantresult is indicative of detection ofcontamination or resulted from naturalvariation.

(5). In § 264.98 by removingparagraphs (i), (j) and (k), and byrevising paragraphs.(c), (d). (f), (g). and(h] to read as follows:

§ 264.98 Detection monitoring program.

(c) The owner or operator mustconduct a ground-water monitoringprogram for each parameter orconstituent in accordance with§ 264.97(g). The owner or operator mustmaintain a record of ground-water "analytical data as measured and in aform necessary for the determination ofstatistical significance under § 264.97(h).for the active life and post-closure careperiod of the facility.

(d) The Regional Administrator willspecify how often the owner or operatormust collect samples and conductstatistical tests to detect contamination.The frequencies for both will be at leastsemi-annually and will be consistentwith § 264.97(g) considering the size ofthe Type I or Type 11 errors. Whereappropriate, the comparison will bemade between background wells andwells at the point(s) of compliance.

(f) The owner or operator mustdetermine whether there is statisticallysignificant evidence of contaminationfor any parameter or constituentspecified in the permit pursuant toparagraph (a) of this section at a

31954




frequency specified under paragraph (d)of this section.

(1) In determining whetherstatistically significant evidence ofcontamination exists, the owner oroperator must use the procedure(s)specified in the permit under § 264.97(h).This procedure(s) must compare datacollected at the compliance point(s) tothe background water quality data.

(2) The owner or operator mustdetermine whether there is statisticallysignificant evidence of contamination ateach monitoring well at the compliancepoint within a reasonable period of timeafter completion of sampling. TheRegional Administrator will specify inthe facility permit what period of time isreasonable, after considering thecomplexity of the statistical test and theavailability of laboratory facilities toperform the analysis of ground-watersamples.

(g) If the owner or operatordetermines pursuant to paragraph (f) ofthis section that there is statisticallysignificant evidence of contaminationfor parameters or constituents specifiedpursuant to paragraph (a) of this sectionat any monitoring well at thecompliance point, he must:

(1) Notify the Regional Administratorof this finding in writing within sevendays. The notification must indicatewhat parameters or constituents haveshown statistically significant evidenceof contamination.

(2) Immediately sample the ground-water in all monitoring wells at thewaste management area of concern anddetermine if there is a statisticallysignificant difference between thecompliance and background levels forconcentration of all constituentsidentified in Appendix IX of Part 264.

(3) For any Appendix IX compoundsfor which there is a significantdifference, the owner or operator mayresample within one month and repeatthe Appendix IX analysis for thosecompounds detected. If the results of thesecond analysis confirm the initialresults then these constituents will formthe basis for compliance monitoring. Ifthey do not resample, the hazardousconstituents found during the initialAppendix IX analysis will form thebasis for compliance monitoring.

(4) Within 90 days, submit to theRegional Administrator an applicationfor a permit modification to establish acompliance monitoring program meetingthe requirements of § 264.99. Theapplication must include the followinginformation:

(i) An identification of theconcentration of any Appendix IXconstituent detected in the ground water

at each monitoring well at thecompliance point;

(ii) Any proposed changes to theground-water monitoring system at thefacility necessary to meet therequirements of § 264.99;

(iii) Any proposed additions orchanges to the monitoring frequency,sampling and analysis procedures ormethods, or statistical procedures usedat the facility necessary to meet therequirements of § 264.99;

(iv) For each hazardous constituentdetected at the compliance point, aproposed concentration limit under§ 264.94(a) (1) or (2), or a notice of intentto seek a variance under § 264.94(b); and

(5) Within 180 days, submit to theRegional Administrator:

(i) All data necessary to justify anyvariance sought under § 264.94(b); and

(ii) An engineering feasibility plan fora corrective action program necessary tomeet the requirements of § 264.100,unless:

(A) All hazardous constituentsidentified under paragraph (g)(2) of thissection are listed in Table 1 of § 264.94and their concentrations do not exceedthe respective values given in thatTable; or

(B) The owner or operator has soughta variance under § 264.94(b) for everyhazardous constituent identified underparagraph (g)(2) of this section.

(6) If the owner or operatordetermines, pursuant to paragraph (f) ofthis section, that there is a statisticallysignificant difference for parameters orconstituents specified pursuant toparagraph (a) of this section at anymonitoring well at the compliance point,he may demonstrate that a source otherthan a regulated unit caused thecontamination or that the contaminationresulted from error in sampling,analysis, or evaluation. While the owneror operator may make a demonstrationunder this paragraph in addition to, or inlieu of, submitting a permit modificationapplication under paragraph (g)(3) ofthis section, he is not relieved of therequirement to submit a permitmodification application within the timespecified in paragraph (g)(3) of thissection unless the demonstration madeunder this paragraph successfully showsthat a source other than a regulated unitcaused the increase, or that the increaseresulted from error in sampling,analysis, or evaluation. In making ademonstration under this paragraph, theowner or operator must:

(i) Notify the Regional Administratorin writing within seven days ofdetermining statistically significantevidence of contamination at thecompliance point that he intends to

make a demonstration under thisparagraph;

(ii) Within 90 days, submit a report tothe Regional Administrator whichdemonstrates that a source other than aregulated unit caused the contaminationor that the contamination resulted fromerror in sampling, analysis orevaluation;

(iii) Within 90 days, submit to theRegional Administrator an applicationfor a permit modification to make anyappropriate changes to the detectionmonitoring program at the facility; and

(iv) Continue to monitor in accordancewith the detection monitoring programestablished under this section.

(h) If the owner or operatordetermines that the detection monitorin gprogram no longer satisfies therequirements of this section, he must,within 90 days, submit an applicationfor a permit modification to make anyappropriate changes to the program.

6. In § 264.99 by revising paragraph(c), revising paragraphs (d), (f), and (g),removing paragraph (h), redesignatingparagraph (i) as (h), (j) as (i), and (k) as(j), revising the redesignated paragraphs(h) introductory text and (i) introductorytext, and removing paragraph (1) to readas follows:

§ 264.99 Compliance monitoring program.

(c) The Regional Administrator willspecify the sampling requirements andstatistical procedures appropriate forthe constituents and site, consistentwith § 264.97 (g) and (h).

(1) The owner or operator mustconduct a sampling program for eachparameter or constituent in accordancewith § 264.97(h).

(2) The owner or operator must recordground-water analytical data asmeasured and in a form necessary forthe determination of statisticalsignificance under § 264.97(h) for theactive life and post-closure care periodof the facility.

(d) The owner or operator mustdetermine whether there is statisticallysignificant evidence of increasedcontamination for any parameter orconstituent specified in the permit,pursuant to paragraph (a) of this section,at a frequency specified underparagraph (f) of this section.

(1) In determining whetherstatistically significant evidence ofincreased contamination exists, theowner or operator must use theprocedure(s) specified in the permit.under § 264.97(h). This procedure mustcompare data collected at thecompliance point(s) to a concentration

31955



31956 Federal Register j Vol. 52, No. 163 / Monday, August 24, 1987 / Proposed Rules

limit developed in accordance with§ 264.94.

(2) The owner or operator mustdetermine whether there is statisticallysignificant evidence of increasedcontamination at each monitoring. wellat the compliance point within areasonable time period after completionof sampling. The Regional Administratorwill specify that time period in thefacility permit, after considering thecomplexity of the statistical test and theavailability of laboratory facilities toperform the analysis of ground-watersamples.

(f) The Regional Administrator willspecify the frequencies for conductingstatistical tests to determine statisticallysignificant evidence of increasedcontamination. The frequencies will beat least semi-annually and will beconsistent with § 264.97(g) consideringthe size of the Type I and Type II errors.

(g) The owner or operator mustanalyze samples from all wells at thecompliance point of a regulated unit forall constituents contained in AppendixIX of Part 264 at least annually todetermine whether additional hazardousconstituents are present in the upper-most aquifer, pursuant to procedures in§ 264.98(f). If the owner or operator findsAppendix IX constituents in the ground-water that are not identified in thepermit, the owner or operator mayresample within one month and repeatthe Appendix IX analysis.

If the second analysis confirms thepresence of new constituents, the owneror operator must report theconcentration of these additionalconstituents to the RegionalAdministrator within seven days aftercompletion of the second analysis andadd them to the monitoring list

(h) If the owner or operatordetermines pursuant to paragraph (d) of

this section that any concentrationlimits under § 264.94 are being exceededat any monitoring well at the point ofcompliance, he must:

(i) If the owner or operatordetermines, pursuant to paragraph (d) ofthis section that the ground-waterconcentration limits under this sectionare being exceeded at any monitoringwell at the point of compliance, he maydemonstrate that a source other than aregulated unit caused the contaminationor that the contamination resulted fromerror in sampling, analysis, orevaluation. In making a demonstrationunder this paragraph, the owner oroperator must:

[FR Doc. 87-19185 Filed 8-21-87: 8:45 aml

BlLUNG CODE 6560-50-M



Documents

Federal Register Vol. 52, No. Monday, August 24, 1987 ...€¦ · 31948 HeinOnline -- 52 Fed. Reg. 31948 1987 This information is reproduced with permission from HeinOnline, under