ranking contaminated sites using a partial ordering method

776

Environmental Toxicology and Chemistry, Vol. 22, No. 4, pp. 776–783, 2003q 2003 SETAC

Printed in the USA0730-7268/03 $12.00 1 .00

RANKING CONTAMINATED SITES USING A PARTIAL ORDERING METHOD

TRINE S. JENSEN,* DORTE B. LERCHE, and PETER B. SøRENSENNational Environmental Research Institute, Department of Policy Analysis, Frederiksborgvej 399, 4000 Roskilde, Denmark

(Received 5 November 2001; Accepted 21 July 2002)

Abstract—In this project, we apply the method of partial ordering on the ranking of 74 contaminated sites located in the countyof West Zealand (Denmark). The method is based on the concept that the parameters are kept separated through the ranking analysis,and thus no weighing of the different parameter values is necessary. The ranking is displayed in a graphical form by the Hassediagram technique to ease the interpretation. A critical comparison is made of the ranking of contaminated sites by the partialordering method and an index function used by the county of West Zealand. Comparing the ranking by the partial ordering methodto the index function shows that the choice of score points and index function highly influences the ranking result, as only foursites are equally ranked. The importance of the parameters used to identify the environmental hazard of the contaminated sites isanalyzed in order to evaluate the influence of each parameter on the ranking. From among a total of six different parameters, twohave high influence, two medium, and two low because of both the construction of the scoring system and the characteristics ofthe data.

Keywords—Hasse diagram Waste disposal Priority list Priority criteria

INTRODUCTION

More than 99% of the water supply in Denmark is bygroundwater. Hence, groundwater protection is given a highpriority, and the main goal is to keep the groundwater of sucha high quality that it can be used as drinking water withoutsignificant treatment. This is reflected in the political objectivesconcerning the management of contaminated sites.

The number of known contaminated sites in Denmark ap-proximated 5,000 at the beginning of 2000, and this figure isestimated to reach 14,000 at the end of the national survey.The estimated total cost of the Danish cleanup program is morethan 1 billion EUR. Registration of contaminated sites is grow-ing faster than the budget of the cleanup activities, and hencepriority of both the investigation and the cleanup activitiesbecomes crucial in order to optimize the environmental effectof the economic investments. However, prioritizing the effortis a challenge to the decision makers because the various dataon the contaminated sites (geological information, soil char-acteristics, and type and distribution of the contaminants,among others) constitute a complex set of information.

In 1990 the Danish counties started systematically inves-tigating all sites that are being or have been used for industrialand municipal activities presenting a potential contaminationrisk. The main national objectives as defined in the Act of SoilContamination ([1]; http://147.29.40.90/pMAINRFpA578882395/459) are to clean up all contaminated sites that are at risk ofcontaminating the groundwater or that pose a health risk at res-idential areas or other sensitive areas (e.g., kindergartens).

The Danish counties have designated particularly valuablewater abstraction areas that cover 35% of the country. Des-ignated areas play a key role in the prioritization of contam-inated sites for investigation and remediation, as these areasare given a high political priority. A three-step priority scheme

* To whom correspondence may be addressed ([email protected]).Presented at the Organic Soil Contaminants Meeting, SETAC Eu-

rope, Copenhagen, Denmark, September 2–5, 2001.

is followed. At the first level, contaminated sites for investi-gations are identified among all potential hazardous sites iden-tified through historical mapping of the activities on the sites,and preliminary field investigations are performed. The secondlevel is a ranking of contaminated sites for extended investi-gations based on results from the field investigations. The thirdlevel is a ranking of contaminated sites for remediation basedon the extended field investigations.

In Denmark, the counties are responsible for identification,risk assessment, and remediation of soil contamination at theregional level. In support, the Danish Environmental Agencyhas developed a national ranking system of point-source con-tamination [2] in order to forward standardized national riskassessments procedures. This system is based on a scoringsystem where the site-specific parameters (groundwater gra-dient, toxicity of contaminants, and so on) are given scorepoints according to a classification system. The priority systemis developed to rank contaminated sites either due to a potentialrisk of groundwater pollution or to a potential risk due to thearea use. The disadvantage of this kind of scoring systems,which are also referred to as index functions, is that a linearrelation among the parameters is anticipated in order to obtaintotal score points for each contaminated site. Furthermore,scoring classes and score points have to be developed for eachparameter. In addition, the parameters are ascribed more orless subjective weights based on nonclassified information.

The county of West Zealand has extended the national rank-ing system and included new score points and parameter clas-ses because of shortcomings in the national system. However,the scoring system methodology generates a number of priorityproblems. For example, the relative ranking of waste disposalsites and contaminated industrial sites is difficult to interpretbecause of the scores on degradation potential. At waste dis-posal sites, the degradation potential is often enhanced becauseof a synergistic effect of the many compounds present, whereasat industrial sites relatively few compounds are usually present

Ranking contaminated sites using a partial ordering method Environ. Toxicol. Chem. 22, 2003 777

Fig. 1. Principle of linearization of partial ordering elements.

with little or no interaction effect. These findings are difficultto include in a scoring system because of difficulties in ex-pressing the findings in score points and classes. The scoringsystem is also based on many subjective decisions on the val-ues of the scores and the weights of the index function. Fur-thermore, the information of the original data is lost as scorepoints replace actual data and as transparency in the final rank-ing result is diminished.

A method not based on scoring systems and index functionsis the partial ordering method. This method has been developedand used in a series of different environmental problems [3–7]. In this project, we apply the method on the ranking of 74contaminated sites located in the county of West Zealand. Themethod is based on the concept that the parameters are keptseparated through the analysis, and thus no weighing of thedifferent parameters by score points is necessary. A criticalcomparison between the ranking by the partial ordering methodand the index function used by the county of West Zealand ismade. The ranking by the partial ordering method is madeaccording to the criteria identified in West Zealand’s rankingsystem and covers specifically the risk assessment of ground-water pollution. Four criteria form the basis for assessing therisk to the groundwater in the priority setting: geological andhydrological conditions (protection by soil layer and ground-water gradient), the harmful potential of polluting substances(toxicity), the potential for contaminant dispersal in ground-water (mobility and degradation), and the importance of thecontaminated sites as water abstraction areas, as designatedby the Danish counties. The importance of the parameters usedto identify the environmental hazard of the contaminated sitesis further analyzed in order to evaluate the influence of eachparameter on the ranking.

METHODOLOGY

The principle of the partial ordering method can be de-scribed as follows: If site 1 is described by the parameters (a1,b1, c1), for example, a1 5 toxicity, b1 5 mobility, and c1 5degradation, and site 2 by the parameters (a2, b2, c2), then site1 is ranked above site 2 if all the parameter values for site 1are above all the parameter values for site 2, that is, a1 . a2

and b1 . b2 and c1 . c2. If not, the sites are not comparable.If all parameters of site 1 are equal to site 2, the rank of bothsites is equivalent. All site combinations are analyzed, and arank is made when possible. Since all sites are not necessarilycomparable, the ranking becomes partial. The ranking is or-ganized in a network and can be drawn in a Hasse diagram[8] that links the ordered sites. It is important to emphasizethat no weighing is needed of the different parameters, as theyare kept separated through the analysis. The method can ranka substantial number of sites very quickly, and in general thismethod becomes more powerful for predictions as the numberof sites increases. A detailed description of the method andthe underlying mathematical consideration can be found in[9,10]. The applicability of the partial ordering results can beimproved by using the concept of linear extension. Linearextension is a methodology where the partial order is projectedinto a linear order [11,12]. To illustrate the principle, a simpleHasse diagram is drawn for five sites as shown in Figure 1.A partial ordered set always has at least one but often a largenumber of possible linear orders, where all elements (contam-inated sites) are compared to each other and where no conflictsexist between the individual linear order and the partial order.Such a range of possible linear orders is denoted linear ex-

tensions of the partial order. From the total set of linear ex-tensions, it is possible to calculate the probability for a specificelement to be placed at a specific rank and thus obtain onlyone linear extension. In Figure 1, it is seen that the probabilityfor b to be placed at rank level 4 is 2/3 and the probabilityfor rank level 3 is 1/3. The numbers in this table are integervalues, so a zero value indicates that it is completely impos-sible for a specific element to be placed at this rank level.Since the total set of linear extensions is often too large tohandle, even by computers, a method based on random linearextensions is applied [11]. In the present study, we furtherdevelop the concept of random linear extensions by assumingthat contaminated sites are chosen one at a time for remediationand then are deleted from the pollution register. This procedureis introduced in the linear extension procedure as a repetitiveprocedure where the site with the highest rank is removedbefore a new linear extension is obtained. This integrated ap-proach ensures that the most hazardous site is always chosenfirst when sites for further investigation and/or remediation areto be identified.

The importance of the influence of the parameters on theranking is also analyzed by the partial ordering method. Simplycounting the additional number of comparisons when a pa-rameter is removed does this. As such, the partial orderingmethod not only ranks contaminated sites but also identifiescontradictions in the criteria used to rank the sites and identifieswhich criteria are the most important.

DATA ON CONTAMINATED SITES

Appendix 1 shows data from 74 contaminated sites fromthe municipalities of Slagelse and Ringsted, West ZealandCounty. The contaminated sites are characterized by six cri-teria, each of which is described by a classification system asshown in Table 1. The most favorable parameter criteria aregiven the lowest value. However, the specific values do notinfluence the result since the partial ordering method comparesthe parameters according to the criteria higher than, lower than,or equal to. The parameters and classification are equal in thepartial ordering method and the scoring method except for oneparameter, the groundwater protection class, which is split intotwo classes in the partial ordering method. The classificationsystem and scoring point of the West Zealand priority systemare shown in Appendix 2. The two classes are split becausethe subjectivity of this class in the scoring system is remarkablebecause of the combination of two parameters in one class.The same argument can be used for the classification of thedegradation potential. However, it is chosen only to split thegroundwater protection class in order to illustrate the flexibilityof the partial ordering method.

The contaminated sites are identified both by a locality(Locality no.) and a Hasse diagram number (Hasse no.) running

778 Environ. Toxicol. Chem. 22, 2003 T.S. Jensen et al.

Table 1. Overview of the six ranking parameter classes used in thepartial ordering method

Classes

Water resource classesWater abstraction areas with limited interestValuable water abstraction areasParticularly valuable water abstraction areasWater abstraction areas for public water supply

1233

Groundwater protection by soil layerSoil layer protection .30 m claySoil layer protection 15–30 m claySoil layer protection ,15 m clay

123

Groundwater gradientUpward gradientUncertain gradientDownward gradient

123

MobilityLow mobilityMedium mobilityHigh mobility

123

ToxicityLow toxicityMedium toxicityHigh toxicity

123

Degradation potentialHigh aerobe/high anaerobe degradationHigh aerobe/medium anaerobe degradationHigh aerobe/low anaerobe degradationMedium aerobe/high anaerobe degradation

1242

Medium aerobe/medium anaerobe degradationMedium aerobe/low anaerobe degradationLow aerobe/high anaerobe degradationLow aerobe/medium anaerobe degradationLow aerobe/low anaerobe degradation

35456

Fig. 2. Hasse diagram of ranking of 74 contaminated sites in West ZealandCounty, Denmark. Sites having equivalent rankings are the following:{3;12;16;217;32;37;42;48;52;55;56;66;70;71}, {4;5;11;13;15;17;18;19;20;21;25;26;28;29;30;34;35;41}, {6;65;68}, {7;8;39;73}, {9;38;51}, {10;22;31;47;60;61;62;63}, {23;24}, {33;49;50;53}, {43;72}, {44;46;74},{67;69}. Dark circles are contaminated sites equally ranked by the par-tially ordering method and the index function, that is, {58;40;67;69}.

from 1 to 74 (see Appendix 1). The locality number refers toWest Zealand’s database ([13]; http://www.vestamt.dk) withinformation on the different sites. The Hasse diagram numberis used to identify the contaminated sites in the Hasse diagram(Fig. 2; see also Results and Discussion). Data on the com-pound specific parameters (toxicity, degradation, and mobility)are given for the most hazardous compound identified amongthe list of detected compounds at each site. The identificationis based on an evaluation of concentration and toxicity of thecompounds made by West Zealand County. In Appendix 1,the total groundwater score and the total number and type ofhazardous compounds are also listed. The total groundwaterscore from West Zealand’s priority system is used to comparethe ranking with the partial ordering method.

RESULTS AND DISCUSSION

Ranking by partial ordering method

In Figure 2, a Hasse diagram of the ranking of contaminatedsites is shown using the parameters in Table 1. The contam-inated sites are ordered in a network, and only the sites withconnecting lines are comparable. The partial ordering methodrecognizes that not all sites can be directly compared with allother sites in terms of environmental hazard and that, whenmany criteria are used, contradictions in the ranking of sitesexist. The partial ordering method is thus fundamentally dif-ferent in structure than an index function. The Hasse diagramshows that site 58 is at the top of the ranking and site 23 isat the bottom. To illustrate the general interpretation rule of

the diagram, it is seen that site 45 is ranked lower than site43 since 43 is above 44, which again is above 45. Some ofthe numbers in the diagram represent a group of contaminatedsites equally ranked. For example, site 43 also represents site72, site 67 also represents site 69, and so on. Thus, instead of74 sites, 22 ranking classes are obtained. A list of equivalentsites is shown in Figure 2. Compared to the index functionthat organizes the sites into 16 classes, that is, sites with equiv-alent total groundwater scores (Appendix 1), the ranking asdisplayed in the Hasse diagram organizes the contaminatedsites into more ranking classes. The more classes the sites areorganized into, the less subjective evaluations are introducedinto the final decision of which contaminated sites are to beprioritized for further investigation and remediation. For ex-ample, if site 67 is chosen, a subjective decision must be madeas to whether site 67 or site 69 should be chosen first sincethese two sites have equal score points.

Comparing the ranking by the partial ordering method andthe index function

To compare the ranking by the partial ordering method withthe index function, we linearize the partial ordering rankingby the random linear extension methodology [11]. The resultsare presented in Table 2, which shows that site 58 has thehighest probability of being ranked at the top followed by sites40, 67 (including 69), 43 (including 72), 6 (including 65 and68), and 14. Compared to the priority list of West ZealandCounty, the top scores, that is, those with highest total ground-water scores (Appendix 1), are 58, 40, 67 (including 59 and69), 54, and 6 (including 36, 51, 65, and 68). The rankingagrees on the top four priority sites (58, 40, 67, and 69) butdisagrees on the other sites as shown in Table 3. Sites 59, 54,36, and 51 of the top 11 priority on the ranking list of WestZealand County are not included in the top 10 priority sitesobtained by the linearized partial ordering method. This showsthat the choice of score values and type of index function, inthis case an addition function, highly influence the ranking


Table 2. Linear extension probabilities of the contaminated sites to be given highest priority in the ranking. The sites with the highest probabilitynumber are ranked at the top of the linearized priority list

Hasseno. Probability






58 1 4043

254

0.4540.2590.1420.121

6743

614

25459

0.2140.1690.1670.1500.1020.0910.089

436

143654

259

0.2080.1930.1590.1260.1090.1020.084

644143659

254

0.2150.1700.1550.1210.1150.1100.097

1444

936

25459

0.2070.1950.1480.1280.1190.0990.082

Table 3. List of top priority sites with the partially ordered method and the index function including sites with equivalent priority

Priority

1 2 3 4 5 6 7 8 9 10

Hasse diagramHasse no. (Fig. 1)Equivalent ranked sites

58 40 6769

4372

665, 68

14 44,46, 74

10, 22,31, 47,60, 61,62, 63

9,38, 51

36

Index functionHasse no. 58 40 59, 67,

6954 6, 36,

51, 65,68

9, 38,43, 72

14 7, 8, 10,22, 31, 39,44, 46, 47,60, 61, 62,63, 73, 74

results. To illustrate the twisting of information in the indexfunction compared to the partial ordering method, we analyzethe ranking of sites 59 and 43. Site 59 is given a high priorityin the index function but a low rank in the linearized partialordering method, where the site is not included in the top 23priority sites. On the other hand, site 43 is ranked as one ofthe five top sites using the linearized partial ordering methodbut only as one of the 15 top priority sites using the indexfunction. Sites 59 and 43 are not comparable in the partialordering ranking (Fig. 2). Comparing the values of the param-eters of site 43 and site 59 (Appendix 1), it is seen that thesetwo sites are different in the mobility and the degradationscores. The mobility score of site 43 is 0 compared to 12 forsite 59. The degradation score of site 43 is 8 and of site 59 is2. The total difference in groundwater scores adds up to 6,with the index function being additive. If another index func-tion were chosen, the score difference would be changed. Thedifference in the ranking can be explained by the probabilitydistribution of the individual sites to be ranked in one of the22 priority classes of the partial ordering method. The indi-vidual probability distributions of sites 40, 43, and 59 areshown in Figure 3. The probability distribution of site 40 isincluded only for further illustration. It is noticed that theprobability of site 58 to be ranked at the top is one (Table 2)and thus becomes zero for all other sites. Therefore, only 21priority classes are left for further priority distribution. It isseen (Fig. 3) that the probability of sites 40 and 43 to be rankedat the top level by the linearized partial ordering method ishigh for both sites, although it is higher for site 40. The prob-ability of site 59 being ranked at the top is low, and this sitehas a probability of being ranked in any of the remaining 21classes. Indeed, site 59 has only a few connections in the Hassediagram; therefore, as generally derived in [7], the probabilitydistribution of site 59 is smeared out over many ranks and isranked low in the partial ordering method.

In order to compare the risk potential of the two sites, thesite-specific characterization is evaluated. The most hazardouscompound identified at site 59 is phenol (see Appendix 1). Onthe one hand, the toxicity and mobility of phenol are high,contributing to an increased risk of contaminating the ground-water. On the other hand, the groundwater gradient is upward,the soil protection layer is above 30 m, and the degradationpotential of phenol is high at both aerobe and anaerobe con-ditions, all contributing to reduce the risk of groundwater pol-lution. The result of a risk assessment of the site could be notto remediate the site but to monitor the natural attenuation.Site 43 is contaminated with zinc (see Appendix 1). The tox-icity of zinc is high, and the degradation is classified as low,as metals do not degrade. This unfavorable situation is, how-ever, counteracted by a low mobility, an upward groundwatergradient, and a soil protection layer above 30 m. The resultof a risk assessment could be that no immediate risk exists tothe groundwater resource. However, since zinc is not degrad-able remediation is to be considered at some point. The soonerthe remediation is performed, the less soil is contaminatedbecause of the mobility of the contamination.

Influence of parameters on ranking

The influence of the different parameters on the rankinghas been evaluated, and the results are shown in Figure 4. Thehigher the number on the y axis, the more comparisons areadded to the partial order when the specific parameter is re-moved and thus the more significant is the parameter. It isseen that the toxicity does not influence the number of com-parisons significantly since only six more comparisons areadded if this parameter is not included. Likewise, the impor-tance of the classification of the water abstraction area is low(80 more comparisons). The groundwater gradient and the deg-radation potential have some influence on the ranking result(182 and 219 more comparisons, respectively), whereas the


Fig. 3. Probability distribution of contaminated sites 40, 43, and 59.

Fig. 4. Influence of the different parameters on the ranking of thecontaminated sites. Low numbers mean low influence. GW 5 ground-water.

mobility and the thickness of the soil layer highly influencethe ranking result (616 and 885 more comparisons, respec-tively).

The water abstraction class does not influence the rankingresults significantly since most of the contaminated sites in themunicipalities of Slagelse and Ringsted are located in the samewater abstraction area, that is, particularly valuable water ab-straction areas. However, the low influence of the toxicity ismore a function of the classification of this parameter. Thechoice of only three different score levels may be too aggre-gated for this parameter since a wide range of different com-pounds has been identified at the sites. Therefore, the indi-vidual score classes may consist of compounds as different as,for example, metals, fuel oil, and gasoline in the medium-toxicity class and phenol, atrazine, benzo[a]pyrene, and te-trachlorethylen in the high-toxicity class.

CONCLUSION

The use of the partial ordering method on the ranking ofcontaminated sites with data from the municipalities of Sla-gelse and Ringsted, West Zealand County, has been demon-strated. The partial ordering method recognizes that not allsites can be directly compared with all other sites because ofcontradictions in the data set; that is, if all sites were rankedby one parameter, the rank would be different compared to theranking by another parameter. Thus, the sites are partiallyranked. Presenting the ranking results in a Hasse diagram isa visual way of comparing contaminated sites based on many

parameters that may otherwise be confusing when displayedin table form. Contradictions in the ranking results also becomeclear through the graphical presentation. Although the methodis based on a partial ranking, it is possible to linearize thepartial ranking by the use of probability distribution of sitesto have the highest rank. This approach ensures that the mosthazardous site is always chosen first when sites for furtherinvestigation and remediation are to be identified. Comparingthe linearized ranking by the partial ordering method and theindex function shows that the choice of score values and typeof index function highly influences the ranking result, as onlyfour sites are equally ranked.

The partial ordering method not only ranks sites but alsoidentifies the most important parameters and thus adds valuableinformation to the ranking result. This information may alsocontribute to evaluation of the field investigation program. Asnew contaminated sites are located and field data are available,a new ranking can be easily performed. The classification ofthe individual parameters can also be changed as new infor-mation on the sites appears. Therefore, this method is excellentfor priority setting at different data aggregation and risk as-sessment levels.

The partial ordering method is thus dynamic and flexible,and a number of analyses can easily be performed with manydifferent criteria, for example, analysis with a less aggregatedclassification level of the toxicity, analysis without the lowinfluence parameters, the influence of regional priorities, in-dustrial activities, and so on.

The partial ordering method is a useful method for rankingcontaminated sites and has been shown to overcome many ofthe shortcomings of index functions in the ranking of contam-inated sites when using several incomparable parameters.Many of the subjective criteria introduced into the rankingprocedure by the choice of scores and type of index functionare avoided, and the final result does not lose transparencyand important information of the original data set. In addition,the relative influence of the different parameters does not be-come clear using an index function, and it is based on highlysubjective criteria when adding scores and score levels. It is,however, important to notice that the partial ordering methodis not a precise evaluation of the potential hazard posed bycontaminated sites. The method provide a useful tool when,for example, evaluating the relative hazards of sites describedby only a few parameters or in cases where the parameters’relative influence is not known or can be described by a moreprecise mathematical function. However, a more detailed dis-cussion of the possibilities and limitations of partial orderingand index functions is beyond the scope of this study and ispublished elsewhere [12].

Acknowledgement—We thank Jan H. Vestergaard, West ZealandCounty, for the access to the data on contaminated sites and InaNielsen, West Zealand County, for helpful assistance in the data ex-traction on contaminated sites from the database GeoEnviron and forvaluable discussions on the scoring system.

REFERENCES

1. Ministry of the Environment. 1999. Law on Soil Contamination(L 183). Legal Act 370. Denmark.

2. COWIconcult A/S and Nellemann, Nielsen & Rauschenberger A/S. 1995. Priority system of point source contamination. Soil andGroundwater Protection 19. Environmental Protection Agency,Copenhagen, Denmark.

3. Halfon E, Reggiani MG. 1986. On ranking chemicals for envi-ronmental hazard. Environ Sci Technol 23:1173–1179.


4. Halfon E. 1989. Comparison of an index function and a vectorialapproach method for ranking waste disposal sites. Environ SciTechnol 23:600–609.

5. Bruggemann R, Halfon E. 1997. Comparative analysis of near-shore contaminated sites in lake Ontario: Ranking for environ-mental hazard. J Environ Sci Health Part A 32:277–292.

6. Shima D, Mogensen BB, Sørensen PB. 2000. Validation of aranking model in relation to Danish monitoring data. NERI Tech-nical Report 318. Proceedings, 2nd Workshop, October 21, 1999,National Environmental Research Institute, Roskilde, Denmark,pp 145–161.

7. Bruggemann R, Halfon E, Welzl G, Voigt K, Steinberg CEW.2001. Applying the concept of partially ordered sets on the rank-ing of near-shore sediments by a battery of tests. J Chem InfComput Sci 41:918–925.

8. Hasse H. 1952. Uber die Klassenzahl abelscher Zahlkorper. Aka-demie-Verlag, Berlin, Germany.

9. Bruggemann R, Bucherl C, Pudenz S, Steinberg CEW. 1999.Application of the concept of partial order on comparative eval-uation of environmental chemicals. Acta Hydrochim Hydrobiol27:170–178.

10. Davey BA, Priestley HA. 1990. Introduction to Lattices andOrder. Cambridge University Press, Cambridge, UK.

11. Sørensen PB, Lerche DB, Carlsen L, Bruggemann R. 2001. Sta-tistically approach for estimating the total set of linear orders, Apossible way of analysing partial order sets. Proceedings, Work-shop on Order Theoretical Tools in Environmental Science andDecision Systems, November 6–7, 2000, Berichtes des IGB, Heft14. Berlin, Germany, pp 87–97.

12. Lerche D, Bruggemann R, Sørensen PB, Carlsen L, Nielsen OJ.2002. A comparison of partial order technique with three methodsof multi-criteria analysis for ranking of chemical substances. JChem Inf Comput Sci 42:1086–1098.

13. Government of Denmark. 2000. Database GeoEnviron. West Zea-land County, Denmark.


APPENDIX 1

Site-specific parameters and score points for contaminated sites in the municipalities of Slagelse and Ringsted, West Zealand County, Denmark.GWA class 5 water resource class; GWP class 5 groundwater protection class; total GW 5 total groundwater score points

Locality no. Hasse no.GWAclass

GWPclass Mobility Toxicity Degradation

TotalGW

No. ofcompound Compound Gradient Depth

333-G01-106329-I04-247333-G01-107333-I04-101333-I04-141

5840596769

1616161616

33322

1212121212

1010101010

85255

5855525252

46755

AtrazineBenzenePhenolBenzeneBenzene

33333

33322

333-E07-103329-A06-119329-I04-151333-C09-121333-G02-108

546

365165

1616161616

53123

1212121212

106

1066

63533

5149494949

46268

PerchlorenthyleneTolueneBenzeneTolueneToluene

23333

23123

333-I04-129329-C09-101329-I04-243333-A02-103333-I04-272

689

384372

1616161616

32233

121212

00

666

1010

33388

4946464646

66646

TolueneTolueneTolueneZincBenzo[a]pyrene

33333

32233

329-E11-116329-C03-101329-C04-102329-C09-102329-G01-110

1478

1022

1616161616

31133

61212

66

66666

43333

4443434343

110

713

KeroseneTolueneTolueneFuel oilFuel oil

33333

31133

329-I03-107329-I04-245333-A03-126333-A06-116333-A06-142

3139444647

1616161616

31223

612

006

66

1010

6

33883

4343434343

13431

Fuel oilTolueneCadmiumZincFuel oil

33333

31223

333-G01-108333-G01-116333-G01-117333-G01-129

60616263

16161616

3333

6666

6666

3333

43434343

5133

Fuel oilFuel oilFuel oilFuel oil

3333

3333

333-I04-308333-I04-312329-A03-102329-A03-113

7374

23

16161616

1212

12066

610

66

3883

43434240

6246

TolueneBenzo[a]pyreneArsenicFuel oil

3333

1212

329-E03-104329-F07-103329-G01-136329-I03-120329-I04-224

1216273237

1616161616

22222

66666

66666

33333

4040404040

21332

Fuel oilFuel oilFuel oilFuel oilFuel oil

33333

22222

329-I04-292333-C04-103333-D07-101333-F01-102333-G01-102

4248525556

1616161616

22222

66666

66666

33333

4040404040

32215


33333

22222

333-I03-105333-I04-168333-I04-233329-G01-112333-G01-144

6670712364

161616

816

22212

66660

6666

10

33334

4040403939

24121

Fuel oilFuel oilFuel oilFuel oilCoal tar

33333

22212

329-I04-103333-A06-109333-C09-110333-C09-113333-E03-107

3345495053

1616161616

54555

60666

610

666

38333

3838383838

39214

Fuel oilBenzo[a]pyreneFuel oilFuel oilFuel oil

22222

21222

329-A03-120329-A06-118329-C09-119329-E06-103329-F01-101

45

111315

1616161616

11111

66666

66666

33333

3737373737

47212


33333

11111

329-G01-102G29-G01-103329-G01-105329-G01-107329-G01-108

1718192021

1616161616

11111

66666

66666

33333

3737373737

42333


33333

11111

329-G01-117329-G01-124329-G01-138329-G01-139329-G02-110

2526282930

1616161616

11111

66666

66666

33333

3737373737

33342

Fuel oilFuel oilFuel oilFuel oilGasoline

33333

11111


APPENDIX 1

Continued

Locality no. Hasse no.GWAclass

GWPclass Mobility Toxicity Degradation

TotalGW

No. ofcompound Compound Gradient Depth

329-I04-123329-I04-127329-I04-248329-A02-106

343541

1

16161616

1111

6660

6666

3338

37373736

1224

GasolineFuel oilGasolineNickel

33

11

333-G01-104329-G01-116

5724

88

31

66

66

33

3529

32

Fuel oilFuel oil

APPENDIX 2

Classification score points for the different parameters used in the priority system of West Zealand County, Denmark. The total groundwaterscore is calculated by an index function based on a simple addition function

ranking contaminated sites using a partial ordering method

Documents