Air pollution exposure monitoring and estimation Part VI.‡ Ambient exposure of adults in an industrialised region† Jocelyne Clench-Aas,* Alena Bartonova, Knut E. Grønskei, Leif O. Hagen, Ole-Anders Braathen and Sam-Erik Walker Norwegian Institute for Air Research, PO Box 100, N-2027 Kjeller, Norway. E-mail: jocelyne.clench-aas@nilu.no; Fax: +47 63 89 80 50; Tel:+47 63 89 80 00 Received 7th April 1999, Accepted 25th June 1999 This paper presents methodology and results of a dynamic individual air pollution exposure model (DINEX) that calculates the hourly exposure for each adult in a panel study.Each of over 260 participants, through the use of a diary, provided information used in the model to calculate his/her personal, individualised exposure. The participants filled out the diary daily, hour by hour, over two, two month periods.The exposure assessment model coupled the diary information and results of an indoor/outdoor measurement program, with the results of dispersion modelling on an hourly basis for an industrial area in Norway. The estimated air pollution concentrations from the dispersion model, based on continuous meteorological measurements, were calibrated with air pollutant concentrations measured continuously.in the region in the period after the field study was com- Introduction pleted (1988). It is methodologically challenging to assess the health impact of exposure to concentrations of air contaminants, especially those lower than air quality guidelines. The eVect of exposure to ambient air pollution needs to be Materials and methods quantified and separated from other known factors that influ- Exposure estimating ence health status, such as age, smoking habits, nutrition, preexisting disease and/or genetic constitution.The measurement Exposure estimating must account for people’s movements in or estimate needs to describe the fluctuations in concentrations areas varying substantially in pollution concentrations.individuals are exposed to, as a result of their movements in The spatial distribution of concentrations served as a basis diVerent microenvironments.1 for estimating each individual’s exposure. An exposure assess- Grenland, the study area, is a heavily industrialised area in ment model estimates exposure for each compound, for each southern Norway.Pollutants here originate from several geo- hour and for each participant, using information collected graphically distinct sources. This allows them to vary indepen- from the diary on where each individual was at given time dently of each other, facilitating their individual identification points. and quantification. Grenland is well known for its industrial An ambient pollution dispersion model for the entire geohaze leading to reduced visibility.Industrial emissions of graphic area combined information on emissions, with inforhydrochloric acid, ammonia and chlorine lead to the formation mation on meteorological conditions in the study area. Such of this haze, even though humidity is not high enough to lead a model estimated hourly concentrations of the diVerent to normal fog.This phenomenon occurs mainly on warm compounds in a square kilometre grid.1,2 The model estimated summer days in conjunction with the land–sea breeze. In the outdoor concentrations of pollution at each individual’s home, winter, industrial haze does not usually occur. place of employment or places visited. This method is described Typically important contaminants in the study region are: in more detail elsewhere.3 sulfur dioxide, nitrogen oxides, carbon monoxide, chloride, People generally spend so much of their time indoors, that ammonium, hydrocarbons, photochemical oxidants such as it is of importance to know indoor air quality. It is important ozone and peroxiacetylnitrate (PAN), polycyclic aromatic to quantify how much of outdoor air pollution penetrates into hydrocarbons (PAH), and traces of chlorinated organic com- the home, and to describe possible indoor sources of air pounds such as dibenzofurans and dioxins.pollution. Norwegian homes do not use gas cooking or heating, This paper presents a dynamic model used to estimate and therefore do not possess this important indoor source of exposure to a series of air pollutants in a panel investigation, nitrogen oxides.Tobacco smoke is the single most important designed to identify the compound or components, if any, factor for indoor pollution in Norwegian homes. Indoor responsible for adverse short-term health eVects. The dynamic measurements at home were made at selected homes, and individual pollution exposure model (DINEX) has also been algorithms derived from these measurements were used in the used in other studies.This paper is presented mainly to discuss individual exposure estimating. the method, since major improvements in pollution abatement An exposure assessment model estimated each individual’s and the closing of one of the factories has reduced pollution pollution exposure for each hour.For each of the pollutant compounds, the model took into account the following major elements: (1) geographic location, (2) proximity to traYc, †Presented at AIRMON ’99, Geilo, Norway, February 10–14, 1999. ‡For Part V, see ref. 14. (3) being indoors or outdoors, (4) shopping, and (5) travelling. J. Environ. Monit., 1999, 1, 341–347 341The dispersion model were available. Emissions were assessed for elemental chlorine and model calculations were performed for Cl2 and did not Grenland (see Fig. 1) lies in a long and relatively wide valley account for chemical reactions. opening to the sea. It is a highly industrialised locality of However, emissions were not known for all compounds. 16×23 km2. The topography combined with climate creates Ozone concentrations within the model area were assessed local temperature inversions in the winter, with poor atmos- from background concentrations, ozone depletion by NOx and pheric dispersion conditions, often leading to higher concen- ozone formation based on the dissociation of NO2.Longtrations of several types of pollutants. The land–sea breeze range transport that may at times contribute around 30% of primarily influences pollutant dispersion from most industrial the concentrations of NOx was also accounted for.sites in the valley in the summer and inversions in the winter. For sulfates, nitrates and suspended particles, emissions In the summer, the land–sea breeze leads the wind into the were again not known, and an appropriate model did not valley during the day and out to sea at night.exist. Therefore, the model for suspended particles distributed The main sources of air pollution in Grenland are local measured total suspended particles over a twelve hour period emissions from industry, vehicular traYc, domestic heating according to continuous measurements of visibility. and boat traYc. Fig. 2 gives an overview of the relative importance of various For modelling, the entire Grenland area was divided into a sources of emissions.Information on emissions from industry square kilometre grid system. Each participant’s home and and from both boat and car traYc was collected immediately work/school address was coded to the nearest km2. The code prior to the field study. EVorts were made to collect emissions also indicated proximity to a major road. on an hourly basis.The dispersion model (EPISODE) estimated the concen- A separate model using traYc counts estimated pollution tration of each compound in each grid square (km2) using concentrations along the major roads for diVerent hours of meteorological conditions measured continuously at two sta- the day and under diVerent meteorological conditions.tions. Measured ambient pollution concentrations at 5 stations For a more complete discussion of the elements included in were used to control the concentration field.1,2 The dispersion the dispersion model, the uncertainties in the model and tests model usually functions by dispersing known emissions at of the validation of the model, see other articles in this series.2,3 specified locations.For SO2, and NOx , emission inventories Air quality measurements Outdoor measurements. The principal compounds measured were sulfur dioxide (SO2), nitrogen oxide (NO), nitrogen oxides (NOx), ozone (O3), sulfates (SO4), nitrates (NO3), particulate matter (PM2.5) and pollen. Nine air quality stations in the area measured air quality and meteorological parameters during the two investigation periods.Fig. 1 shows the measuring sites for air quality and meteorological parameters. Indoor measurements. People in general spend most of their time indoors. If in addition windows and doors are closed, air quality indoors can be substantially diVerent from outdoors. Opening of windows for ventilation can for some compounds, influence indoor air quality.Using simultaneous measurements made indoors and outdoors, a set of algorithms was developed to estimate pollution concentrations indoors. Most notably, exposure to suspended particles was increased when people smoked or were exposed to passive smoking, or nearby traYc. Fig. 2 Average total emission intensity (Q) of SO2 , total suspended Fig. 1 Location of stations for measuring air quality and meteorologi- particles (TSP) and NOx from diVerent source groups in winter and summer.cal conditions in the Grenland area. 342 J. Environ. Monit., 1999, 1, 341–347Simultaneous indoor and outdoor measurements for a Table 2 summarises the maximum values of diVerent air pollution components measured outdoors during the period three day period for 15 typical homes, determined indoor concentrations due to penetration into the home, and/or January to June 1988.Table 3 provides the algorithms used for indoor concentrations based on the measurement pro- indoor sources (e.g. smoking). Measurements were made for three periods per day in each home, for both a winter and gramme inside selected homes. summer period.4 Estimated exposure to diVerent air pollution components in Data collected from each individual Grenland Over 260 people participated in the study both in winter and Based on information reported in the diary and summarised summer.Each person filled out a diary, once a day on an in the previous section, each participant’s exposure to each hourly basis. Each person was to indicate when and how long component was estimated each hour.It was necessary, in he/she was at home, at work or school, or visiting other places order to use these exposure estimates further in assessing the (providing address), outdoors, or indoors, with or without health eVects of air pollution, to know how contaminants open windows. Each address was coded to the nearest square correlated with each other.It was an a priori assumption that kilometre in addition to noting closeness to major traYc the unique geographical features of the region should result arteries. in a relatively independent distribution of the contaminants TraYc is one of the major pollution sources that contributes that should allow one to distinguish between the eVects of to exposure to NOx and suspended particles.Therefore, it is each component. As can be seen in the figures in Part I of also necessary to know when people are travelling and how this series,3 dispersion model results demonstrated that this much traYc they are encountering. The participants indicated assumption was correct. how many minutes they were travelling in dense, medium or Interrelations between exposure estimates for individual little traYc.These terms were defined. components are generally stronger in summer than in winter Each person was also to indicate how many minutes he/she but often of the same general dimension. Important exceptions was shopping either in downtown Skien or Porsgrunn (the are relative humidity and sulfates (slight negative correlation two major towns in the area) or other places.Skien had at in the winter, yet a stronger positive correlation in the summer). that time an outdoor shopping area that was essentially free O3 correlates negatively with all compounds other than sulfates of traYc, whereas a major road crossed Porsgrunn. Therefore, in the winter, whereas it correlates positively with all comthe model used an average of the squares that represent the pounds other than CO and chlorine in the summer.shopping area in Skien and an average of the km2 in downtown Table 4 indicates where the correlation exceeded 0.25, which Porsgrunn, plus an additional factor for vehicular traYc for is a value chosen to represent a meaningful correlation. With estimating concentrations when shopping in Porsgrunn. our amount of data, 0.001 significance level is reached for correlation coeYcient values under 0.1.Table 5 summarises participants’ exposure. The table shows Results percentiles of exposure estimates and maximum estimated Time-use information provided by the diary exposure for winter and summer. These estimated concentrations could be described as a Results of studies of the health eVect of air pollution carried function of various parameters. out in diVerent countries where individual exposure has not Changes in temperature and humidity during day and night been measured should be compared with care. Even though are especially noticeable in the summer, and aVect some of pollution levels may be lower in one country, exposure may the contaminants giving them marked daily variations.Human be higher due to cultural diVerences in ventilation of homes activities that tend to occur at routine times during the day, (i.e. sleeping with windows open) and amount of time spent that is, driving to work, working, etc. can also aVect exposure outdoors. This section summarises such features for the studied to contaminants. Norwegian population. Fig. 3 shows changes in exposure to the gaseous and particu- There is a sharp contrast in time spent with the window late air contaminants as a function of hour whereas Fig. 4 open, closed or outdoors between winter and summer shows exposure day by day during the investigation. Exposure (Table 1). Even in the winter, people spend an average of 17% to NO is higher in the winter than in the summer and exposure of their time in rooms with the window open and 3% of their to NO2 slightly lower in the winter.Concentrations of NO, time outdoors. In the summer, time spent outdoors can be as and to a lesser degree NO2, show the typical peaks associated high as 20%. Children are more outdoors than adults. Women with exposure to traYc pollution. These peaks are especially are outdoors less than men, but have the window open more noticeable in the afternoon rush hour and are slightly higher than men.Assuming that a yearly average is a direct mean of during the winter than the summer. Exposure to SO2 is only the summer and winter values given in Table 1, Norwegians slightly higher in the winter than in the summer and does not seem to be outdoors more than in most other countries.5–11 vary markedly with time of day.Exposure to O3 is, as expected, Time spent outdoors in this region does not diVer from values much higher in the summer than in the winter and shows a estimated for Norway12 as a whole. pronounced daily variation in the winter but not in the Time spent travelling is stable in both winter and summer. summer. Exposure to nitrates and sulfates is lower in the No comparison is available for time spent indoors with or winter than in the summer.without the window open, since few studies have segregated There were only minor diVerences in exposure between the these parameters. Otherwise it is evident that features of this two sexes. However, exposure to all parameters was either population reflect a society dominated by working in the equal or slightly higher in men, except for suspended particles.factories, with for example shift work influencing wake-up Men had slightly higher levels of exposure to the nitrogen time. oxide components. This is especially noticeable in younger men, possibly due to more exposure to traYc pollution. Both Measured ambient and indoor concentrations older men and women have higher exposure to ozone, due to more time spent outdoors.The higher exposure to suspended Due to a mild winter, pollution concentrations varied considerably. Increases in sulfates and suspended particles accompanied particles in women is especially evident in younger women. Younger men, however, have also higher levels of exposure the two coldest periods that occurred that winter.J. Environ. Monit., 1999, 1, 341–347 343Table 1 Time spent in diVerent microenvironments by age group of participants. Percent of registered wake-up and go to sleep time in the diVerent participant subgroups Winter Summer Adult Adult Adult Adult women men Children women men Children Type of location (% of time) Home 74.6 67.4 68.6 68.8 62.7 66.4 At work/school/kindergarten 9.7 17.4 12.4 8.7 15.2 10.4 Other places 15.7 15.2 19.0 22.5 22.1 23.2 Indoors window closed 76.5 76.3 81.2 45.5 44.0 54.9 Indoors window open 19.2 16.3 9.9 38.1 33.8 18.9 Outdoors 1.1 3.8 6.5 12.9 18.6 23.0 Travelling whole hour 3.2 3.6 2.4 3.6 3.6 3.2 Number of minutes traveling Dense traYc 3.9 6.9 6.9 6.1 8.0 6.2 Medium traYc 19.4 27.6 16.6 20.3 27.9 19.6 Light traYc 20.9 26.9 19.5 21.1 22.7 18.7 Total daily shopping 26.4 23.2 15.2 27.3 21.8 22.4 Type of activity (% of time) Sleeping 35.3 33.8 42.4 33.8 32.7 40.7 Daily activity 63.4 64.6 54.8 65.1 65.4 57.8 Hard work/training 1.2 1.5 2.8 1.0 2.0 1.7 Wake up time (% of registered cases) 06.00–07.00 0.9 3.3 0.7 0.9 4.2 0.1 07.00–08.00 12.4 25.7 4.1 15.1 26.6 2.2 08.00–09.00 26.6 28.6 46.2 29.3 30.5 39.7 09.00–10.00 24.8 17.7 24.7 27.0 19.5 32.2 10.00–11.00 18.0 11.8 12.2 14.9 11.0 15.1 11.00–12.00 6.3 4.0 6.7 5.0 2.7 5.6 12.00–13.00 2.4 1.5 2.2 1.7 1.1 2.3 Sleep time (% of registered cases) 19.00–20.00 0.1 0.1 0.3 0.3 0.0 0.1 20.00–21.00 0.2 0.2 13.1 0.2 0.1 4.2 21.00–22.00 1.4 2.3 37.0 0.6 1.1 30.8 22.00–23.00 11.8 16.7 20.0 7.1 13.8 23.5 23.00–24.00 35.9 38.8 17.7 34.6 39.1 23.8 24.00–01.00 33.6 28.2 7.6 42.8 35.7 14.2 01.00–02.00 5.2 3.7 0.5 5.3 3.8 0.1 02.00–03.00 2.3 2.0 0.6 2.3 1.4 0.8 Table 2 Summary of maximum values of diVerent air pollution components during the period January to June 1988 Averaging City hall G.Stangs gt Skien Component time/h A° s Herre Frednes Klyve Porsgrunn Nenset Skien Fire st. Kongensgt SO2/mg m-3 1 147 338 474 203 872 2027 24 32 23 37 55 26 63 134 320 121 NOx/mg m-3 1 296 761 326 820 463 551 24 110 320 104 273 167 229 NO2/mg m-3 1 192 119 191 125 102 121 24 84 70 75 61 47 59 90 Haze/10-6 m-1 1 764 1061 572 24 116 71 58 O3/mg m-3 1 185 150 8 179 141 Suspended particles/mg m-3 12 69 89 74 93 94 SO4/mg m-3 12 16.7 16.2 17.8 16.3 15.3 NO3/mg m-3 12 10.7 9.8 12.7 6.4 5.9 Cl2/mg m-3 12 6.6 4.7 3.3 4.6 5.0 Soot/mg m-3 24 31 30 79 104 Lead/mg m-3 24 1.21 NH3/mg m-3 24 9.6 NH4/mg m-3 24 8.7 5.3 Formaldehyde/mg m-3 24 0.7 than older men.This is due to higher exposure to tobacco more marked in daily smokers. There is a mean diVerence of 32 mg m-3 exposure to suspended particles in the winter, smoke in the young as opposed to the elderly, and in women as opposed to men, as seen in the data.between non-smokers and those who smoke every day; whereas that diVerence is only 10 mg m-3 in the summer. This is a Non-smokers are slightly more exposed in the winter than in the summer to particles. The diVerence between winter and direct reflection of time spent outdoors or with the window open in the summer as opposed to the winter. summer is more noticeable in occasional smokers and even 344 J.Environ. Monit., 1999, 1, 341–347Table 3 Relationships between indoor (Ci) and outdoor (Co) concentrations of selected pollutants in Norwegian homes in mg m3 Winter Summer Compound 0800–2000 2000–0800 0800–2000 2000–0800 SO2 Ci=0.49Co+5.05 NO2 Ci=0.21Co+10.5a Ci=0.28Co+6.3 Ci=0.34Co+9.55 Ci=0.56Co+7.5 PM2.5 Ci=0.73Co-0.75b Ci=0.7Co- 3.0c Ci=0.75Co+0.25d Ci=0.72Co+5.00e SO4 Ci=0.73Co+0.32 Ci=0.70Co-0.23 Ci=0.75Co+0.43 Ci=0.72Co-0.26 CO Ci=0.7Co O3 Ci=0.2Co aFor NO2 the time intervals are 1600–2400 and 0000–1600.bFor PM2.5 the values here are for non-smoking homes, the constant becomes additive and increases to 36.75 and 97.75 for homes with smoking of 1 to 10 cigarettes per day and homes with smoking of more than 10 cigarettes per day respectively.cFor PM2.5 the values here are for non-smoking homes, the constant increases to 23.0 and 60.5 for homes with smoking of 1 to 10 cigarettes per day and homes with smoking of more than 10 cigarettes per day respectively. dFor PM2.5 the values here are for non-smoking homes, the constant increases to 8.25 and 45.3 for homes with smoking of 1 to 10 cigarettes per day and homes with smoking of more than 10 cigarettes per day respectively.eFor PM2.5 the values here are for non-smoking homes, the constant increases to 23.0 and 60.5 for homes with smoking of 1 to 10 cigarettes per day and homes with smoking of more than 10 cigarettes per day respectively. Table 4 Mean weighted Pearson correlation coeYcients for the log-transformed air pollution exposure data for winter (under the diagonal ) and summer, (above the diagonal ) (only values over 0.25 listed ).Each season is based on approximately 260 individuals, with 1000 registered hours each Relative SO2 NOx NO2 O3 PM2.5 Cl2 SO4 NO3 CO Temperature humidity SO2 — 0.78 0.79 0.62 0.44 0.38 NOx 0.67 — 0.99 0.28 0.71 0.53 0.43 0.91 NO2 0.67 0.89 — 0.31 0.71 0.53 0.43 0.73 O3 — 0.35 0.31 -0.35 0.43 -0.40 S PM2.5 0.58 0.57 0.56 — 0.29 0.60 0.45 0.60 u Cl2 — m SO4 0.44 0.35 0.37 0.35 — 0.62 0.33 m NO3 0.40 0.35 0.26 0.43 — e CO 0.83 0.68 0.47 — r Temperature 0.33 — -0.53 Relative humidity -0.39 0.28 0.26 — Winter Table 5 Percentiles of the calculated exposure to air pollutants for showing the need for accounting for these facts when investigastudy population ting the eVects of air pollution on health.The diary method used in this investigation has proven itself Randomly selected 95% as a feasible basis for estimating personal exposure for an study population Median quantile Maximum investigation of short-term health eVects. It allows one to Wintera measure more individuals over a longer time period, divided Sulfur dioxide/mg m-3 9 22 900 into shorter time intervals.This method allows one to estimate Nitrogen dioxide/mg m-3 17 85 3065 more pollution compounds than an ideal method using port- Ozone/mg m-3 4 41 93 able pollution measuring equipment could possibly do. Particles (fine fr.)/mg m-3 17 108 581 However, one can argue that a combination of both these Sulfates/mg m-3 2 8 17 approaches would be advantageous.This was carried out in a Nitrate/mg m-3 0 2 8 Chlorine/mg m-3 0 7 297 later investigation in a region with traYc pollution.13 The exposure assessment model DINEX, used here, allowed Summer combining concentrations of pollution measured at fixed site Sulfur dioxide/mg m-3 7 19 1414 Nitrogen dioxide/mg m-3 16 55 2313 stations with individual diVerences in behaviour and lifestyle, Ozone/mg m-3 24 112 185 thus creating a personalised and hopefully more exact expo- Particles (fine fr.)/mg m-3 11 53 614 sure estimate.Sulfates/mg m-3 2 10 15 Nitrate/mg m-3 0 2 9 Chlorine/mg m-3 0 1 55 Acknowledgements Birch pollen/pollen m-3 0 47 833 Grass pollen/pollen m-3 0 19 2185 The work reported in this paper was carried out under contract from the Norwegian Ministry of the Environment, the aNumber of hours registered: Winter: 354 735; summer: 304 697.Norwegian State Pollution Control Authority and the Royal Norwegian Council for Scientific and Industrial Research who jointly appointed a board that had the administrative and Discussion co-ordinating responsibility for the investigation. We would like to thank these board members, Sigurd Hagen (Chairman), The most important sources in the area are industrial and high pollution concentrations occur only sporadically (prob- Lasse Hansen, Erik Dybing and Sverre Langa°rd for their constant and helpful contribution.We would also like to thank lems in operation of the plants). High exposure mainly occurs for one compound at a time.Odd F.Skogvold, whose earlier research in the area was the inspiration for the study. His constant help and advice during Variations in lifestyle (time spent outdoors, keeping windows open) etc. were large between seasons and population groups, all phases of the investigation was of utmost importance to us. J. Environ. Monit., 1999, 1, 341–347 345Fig. 3 Variations in concentrations of exposure to the gaseous contaminants (NO, NO2, O3 and SO2) and particulate contaminants (PM2.5, sulfate and nitrate) as a function of time of day and season.Fig. 4 Concentrations of mean daily exposure of the entire study population to the gaseous and particulate matter (PM2.5) contaminants as a function of day of study. Day 1=January 2, 1988. 346 J. Environ. Monit., 1999, 1, 341–3471990), Industrial Institute for Economics and Social Research, We would also like to thank the National Institute of Public Stockholm, 1990.Health, the Norwegian Computing Centre, the Telemark 7 M. D. Koontz and J. P. Robinson, Environ. Monit. Assess., 1982, Central Hospital and the local division of the State Pollution 2, 197. Control Authority for their expertise and help. 8 R. Letz, B. Ryan and J. D. Spengler, Environ. Monit. Assess., 1984, 4, 351. 9 J. J. Quackenboss, J. D. Spengler, M. S. Kanarek, R. Letz and References C. P. DuVy, Environ. Sci. Technol., 1986, 20, 775. 10 M. Schwab, A. McDermott and J. D. Spengler, Environ. Int., 1 K. E. Grønskei, S. E.Walker and F. Gram, Atmos. Environ., 1993, 1992, 18, 173. 27B, No. 1, 105. 11 A. Szalai, P. E. Converse, P. Feldheim, E. K. Scheuch and 2 S. E. Walker, L. H. Slørdal, C. Guerreiro, F. Gram and K. E. P. J. Stone, The use of time: daily activities of urban and suburban Grønskei, J. Environ. Monit., 1999, 1, 321. population in twelve countries, Mouton, The Hague, 1972. 3 J. Clench-Aas, A. Bartonova, T. Bøhler, K. E. Grønskei and S. 12 Central Bureau of Statistics, The time budget surveys 1970–90, Larssen, J. Environ. Monit., 1999, 1, 313. Oslo, 1992. 4 O. A. Braathen, Air pollution and short-term health eVects in an 13 S. Larssen, D. Tønnesen, J. Clench-Aas, M. J. Aarnes and K. industrialized area in Norway – Relationships between indoor and Arnesen, Sci. Total. Environ., 1993, 134(1–3), 51. outdoor concentrations of air pollutants (in Norwegian), 14 A. Bartonova, J. Clench-Aas, F. Gram, K. E. Grønskei, C. Norwegian Institute for Air Research (NILU OR 8/91), Guerreiro, S. Larssen, D. A. Tønnesen and S.-E. Walker, Lillestrøm, Norway, 1991. J. Environ. Monit., 1999, 1, 337. 5 F. S. Chapin, Jr., Human activity patterns in the city. Things people do in time and space,Wiley, New York, 1974. 6 F. T. Juster and F. P. StaVord, Working Paper Series (NO. 258, Paper 9/02781E J. Environ. Monit., 1999, 1, 341–347 347