1、Designation: D 6312 98 (Reapproved 2005)Standard Guide forDeveloping Appropriate Statistical Approaches for Ground-Water Detection Monitoring Programs1This standard is issued under the fixed designation D 6312; the number immediately following the designation indicates the year oforiginal adoption o
2、r, in the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. Asuperscript epsilon (e) indicates an editorial change since the last revision or reapproval.1. Scope1.1 This guide covers the context of ground-water monitor-ing at waste disposal f
3、acilities. Regulations have requiredstatistical methods as the basis for investigating potentialenvironmental impact due to waste disposal facility operation.Owner/operators must perform a statistical analysis on aquarterly or semiannual basis.Astatistical test is performed oneach of many constituen
4、ts (for example, 10 to 50 or more) foreach of many wells (5 to 100 or more). The result is potentiallyhundreds, and in some cases, a thousand or more statisticalcomparisons performed on each monitoring event. Even if thefalse positive rate for a single test is small (for example, 1 %),the possibilit
5、y of failing at least one test on any monitoringevent is virtually guaranteed. This assumes you have done thecorrect statistic in the first place.1.2 This guide is intended to assist regulators and industryin developing statistically powerful ground-water monitoringprograms for waste disposal facili
6、ties. The purpose of thisguide is to detect a potential ground-water impact from thefacility at the earliest possible time while simultaneouslyminimizing the probability of falsely concluding that thefacility has impacted ground water when it has not.1.3 When applied inappropriately, existing regula
7、tion andguidance on statistical approaches to ground-water monitoringoften suffer from a lack of statistical clarity and often imple-ment methods that will either fail to detect contamination whenit is present (a false negative result) or conclude that the facilityhas impacted ground water when it h
8、as not (a false positive).Historical approaches to this problem have often sacrificed onetype of error to maintain control over the other. For example,some regulatory approaches err on the side of conservatism,keeping false negative rates near zero while false positive ratesapproach 100 %.1.4 The pu
9、rpose of this guide is to illustrate a statisticalground-water monitoring strategy that minimizes both falsenegative and false positive rates without sacrificing one for theother.1.5 This guide is applicable to statistical aspects of ground-water detection monitoring for hazardous and municipal soli
10、dwaste disposal facilities.1.6 It is of critical importance to realize that on the basis ofa statistical analysis alone, it can never be concluded that awaste disposal facility has impacted ground water. A statisti-cally significant exceedance over background levels indicatesthat the new measurement
11、 in a particular monitoring well for aparticular constituent is inconsistent with chance expectationsbased on the available sample of background measurements.1.7 Similarly, statistical methods can never overcome limi-tations of a groundwater monitoring network that might arisedue to poor site charac
12、terization, well installation and location,sampling, or analysis.1.8 It is noted that when justified, intra-well comparisonsare generally preferable to their inter-well counterparts becausethey completely eliminate the spatial component of variability.Due to the absence of spatial variability, the u
13、ncertainty inmeasured concentrations is decreased, making intra-well com-parisons more sensitive to real releases (that is, false negatives)and false positive results due to spatial variability are com-pletely eliminated.1.9 Finally, it should be noted that the statistical methodsdescribed here are
14、not the only valid methods for analysis ofground-water monitoring data. They are, however, currentlythe most useful from the perspective of balancing site-widefalse positive and false negative rates at nominal levels.Amorecomplete review of this topic and the associated literature ispresented by Gib
15、bons (1).21.10 The values stated in both inch-pound and SI units areto be regarded as the standard. The values given in parenthesesare for information only.1.11 This standard does not purport to address all of thesafety concerns, if any, associated with its use. It is theresponsibility of the user o
16、f this standard to establish appro-priate safety and health practices and determine the applica-bility of regulatory limitations prior to use.1.12 This guide offers an organized collection of informa-tion or a series of options and does not recommend a specificcourse of action. This document cannot
17、replace education or1This guide is under the jurisdiction ofASTM Committee D18 on Soil and Rockand is the direct responsibility of Subcommittee D18.21 on Ground Water andVadose Zone Investigations.Current edition approved Jan. 1, 2005. Published February 2005. Originallyapproved in 1998. Last previo
18、us edition approved in 1998 as D 631298.2The boldface numbers given in parentheses refer to a list of references at theend of the text.1Copyright ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959, United States.experience and should be used in conjunction with
19、professionaljudgment. Not all aspects of this guide may be applicable in allcircumstances. This ASTM standard is not intended to repre-sent or replace the standard of care by which the adequacy ofa given professional service must be judged, nor should thisdocument be applied without consideration of
20、 a projects manyunique aspects. The word “Standard” in the title of thisdocument means only that the document has been approvedthrough the ASTM consensus process.2. Terminology2.1 Definitions:2.1.1 assessment monitoring program, nground-watermonitoring that is intended to determine the nature and ex
21、tentof a potential site impact following a verified statisticallysignificant exceedance of the detection monitoring program.2.1.2 combined Shewhart (CUSUM) control chart, nastatistical method for intra-well comparisons that is sensitive toboth immediate and gradual releases.2.1.3 detection limit (DL
22、), nthe true concentration atwhich there is a specified level of confidence (for example,99 % confidence) that the analyte is present in the sample (2).2.1.4 detection monitoring program, nground-watermonitoring that is intended to detect a potential impact from afacility by testing for statisticall
23、y significant changes ingeochemistry in a downgradient monitoring well relative tobackground levels.2.1.5 intra-well comparisons, na comparison of one ormore new monitoring measurements to statistics computedfrom a sample of historical measurements from that same well.2.1.6 inter-well comparisons, n
24、a comparison of a newmonitoring measurement to statistics computed from a sampleof background measurements (for example, upgradient versusdowngradient comparisons).2.1.7 prediction interval or limit, na statistical estimate ofthe minimum or maximum concentration, or both, that willcontain the next s
25、eries of k measurements with a specified levelof confidence (for example, 99 % confidence) based on asample of n background measurements.2.1.8 quantification limit (QL), nthe concentration atwhich quantitative determinations of an analytes concentra-tion in the sample can be reliably made during rou
26、tinelaboratory operating conditions (3).2.2 Definitions of Terms Specific to This Standard:2.2.1 false negative rate, nin detection monitoring, therate at which the statistical procedure does not indicate possiblecontamination when contamination is present.2.2.2 false positive rate, nin detection mo
27、nitoring, therate at which the statistical procedure indicates possible con-tamination when none is present.2.2.3 nonparametric, adja term referring to a statisticaltechnique in which the distribution of the constituent in thepopulation is unknown and is not restricted to be of a specifiedform.2.2.4
28、 nonparametric prediction limit, nthe largest (orsecond largest) of n background samples. The confidence levelassociated with the nonparametric prediction limit is a functionof n and k.2.2.5 parametric, adja term referring to a statistical tech-nique in which the distribution of the constituent in t
29、hepopulation is assumed to be known.2.2.6 verification resample, nin the event of an initialstatistical exceedance, one (or more) new independent sampleis collected and analyzed for that well and constituent whichexceeded the original limit.2.3 Symbols:2.3.1 athe false positive rate for an individua
30、l compari-son (that is, one well and constituent).2.3.2 a*the site-wide false positive rate covering all wellsand constituents.2.3.3 kthe number of future comparisons for a singlemonitoring event (for example, the number of downgradientmonitoring wells multiplied by the number of constituents tobe m
31、onitored) for which statistics are to be computed.2.3.4 nthe number of background measurements.2.3.5 s2the true population variance of a constituent.2.3.6 sthe sample-based standard deviation of a constitu-ent computed from n background measurements.2.3.7 s2the sample-based variance of a constituent
32、 com-puted from n background measurements.2.3.8 the true population mean of a constituent.2.3.9 xthe sample-based mean or average concentrationof a constituent computed from n background measurements.3. Summary of Guide3.1 This guide is summarized in Fig. 1, which provides aflowchart illustrating th
33、e steps in developing a statisticalmonitoring plan. The monitoring plan is based either onbackground versus monitoring well comparisons (for example,upgradient versus downgradient comparisons or intra-wellcomparisons, or a combination of both). Fig. 1 illustrates thevarious decision points at which
34、the general comparativestrategy is selected (that is, upgradient background versusintra-well background) and how the statistical methods are tobe selected based on site-specific considerations. The statisticalmethods include parametric and nonparametric predictionlimits for background versus monitor
35、ing well comparisons andcombined Shewhart-CUSUM control charts for intra-wellcomparisons. Note that the background database is intended toexpand as new data become available during the course ofmonitoring.4. Significance and Use4.1 The principal use of this guide is in ground-waterdetection monitori
36、ng of hazardous and municipal solid wastedisposal facilities. There is considerable variability in the wayin which existing Guide USEPA regulation and guidance areinterpreted and practiced. Often, much of current practice leadsto statistical decision rules that lead to excessive false positiveor fal
37、se negative rates, or both. The significance of thisproposed guide is that it jointly minimizes false positive andfalse negative rates at nominal levels without sacrificing oneerror for another (while maintaining acceptable statisticalpower to detect actual impacts to ground-water quality (4).D 6312
38、 98 (2005)2FIG. 1 Development of a Statistical Detection Monitoring PlanD 6312 98 (2005)3FIG. 1 (continued)D 6312 98 (2005)4FIG. 1 (continued)D 6312 98 (2005)5FIG. 1 (continued)D 6312 98 (2005)6FIG. 1 (continued)D 6312 98 (2005)74.2 Using this guide, an owner/operator or regulatoryagency should be a
39、ble to develop a statistical detectionmonitoring program that will not falsely detect contaminationwhen it is absent and will not fail to detect contamination whenit is present.5. ProcedureNOTE 1In the following, an overview of the general procedure isdescribed with specific technical details descri
40、bed in Section 6.5.1 Detection Monitoring:5.1.1 Upgradient Versus Downgradient Comparisons:5.1.1.1 Detection frequency $50 %.5.1.1.2 If the constituent is normally distributed, compute anormal prediction limit (5) selecting the false positive ratebased on number of wells, constituents, and verificat
41、ionresamples (6) adjusting estimates of sample mean and variancefor nondetects.5.1.1.3 If the constituent is lognormally distributed, com-pute a lognormal prediction limit (7).5.1.1.4 If the constituent is neither normally nor lognor-mally distributed, compute a nonparametric prediction limit (7)unl
42、ess background is insufficient to achieve a 5 % site-widefalse positive rate. In this case, use a normal distribution untilsufficient background data are available (7).5.1.1.5 If the background detection frequency is greater thanzero but less than 50 %.5.1.1.6 Compute a nonparametric prediction limi
43、t and de-termine if the background sample size will provide adequateprotection from false positives.5.1.1.7 If insufficient data exist to provide a site-wide falsepositive rate of 5 %, more background data must be collected.5.1.1.8 As an alternative to 5.1.1.7 use a Poisson predictionlimit which can
44、 be computed from any available set ofbackground measurements regardless of the detection fre-quency (see 2.2.4 of Ref (4).5.1.1.9 If the background detection frequency equals zero,use the laboratory-specific QL (recommended) or limits re-quired by applicable regulatory agency (8).35.1.1.10 This onl
45、y applies for those wells and constituentsthat have at least 13 background samples. Thirteen samplesprovide a 99 % confidence nonparametric prediction limit withone resample for a single well and constituent (see Table 1).5.1.1.11 If less than 13 samples are available, more back-ground data must be
46、collected to use the nonparametricprediction limit.5.1.1.12 An alternative would be to use a Poisson predictionlimit that can be computed from four or more backgroundmeasurements regardless of the detection frequency and canadjust for multiple wells and constituents.5.1.1.13 If downgradient wells fa
47、il, determine cause.5.1.1.14 If the downgradient wells fail because of natural oroff-site causes, select constituents for intra-well comparisons(9).5.1.1.15 If site impacts are found, a site plan for assessmentmonitoring may be necessary (10).5.1.2 Intra-well Comparisons:5.1.2.1 For those facilities
48、 that either have no definablehydraulic gradient, have no existing contamination, have toofew background wells to meaningfully characterize spatialvariability (for example, a site with one upgradient well or afacility in which upgradient water quality is either inaccessibleor not representative of d
49、owngradient water quality), computeintra-well comparisons using combined Shewhart-CUSUMcontrol charts (9).45.1.2.2 For those wells and constituents that fail upgradientversus downgradient comparisons, compute combinedShewhart-CUSUM control charts. If no volatile organic com-pounds (VOCs) or hazardous metals are detected and no trendis detected in other indicator constituents, use intra-wellcomparisons for detection monitoring of those wells andconstituents.5.1.2.3 If data are all non-detects after 13 quarterly sam-pling events, use the QL as the n
