Background for spatial differentiation in LCA impact assessment

If nitrogen levels exceed the optimum for the typical species in a given ecosystem, the growth of those species start to decline whereas growth is stimulated for other species better able to benefit from the increased availability of nitrogen. Downing et al. (1993) suggest critical levels of nitrogen in soil at which changes of vegetation composition are induced. These critical levels are used in RAINS to model critical nitrogen depositions or loads above which these changes of vegetation are expected.

Similar to what has been done for acidification, critical loads for eutrophication of terrestrial ecosystems have been estimated for the whole of Europe. Ecosystems covered are mainly forests, but for some countries also heathland and grassland. The critical loads are per grid-element summarised in a cumulative distribution curve for eutrophication (see Figure 4.4). While only few grid-elements contain less than 10 ecosystems, there are several with more than 10,000 ecosystems per grid-element (the highest number being 53,000 in a grid-element shared by the Netherlands, Belgium and Germany). (Posch et al. 1997) The full critical load database in the RAINS 7.2 version was based on almost 700,000 ecosystems (the most recent version relates to over 1.4 million ecosystems; Posch et al. 2002) ^[27]. Section 3.3.3 gives more details about the background of critical loads for acidification. See Downing et al. (1993) and Posch et al. (1995, 1997, 1999 and 2001) for more details about critical loads for eutrophication.

4.3 Calculation procedure

The total deposition per grid-element (as represented in Figure 4.2) can be compared with the cumulative critical load distribution curve for that grid-element (as in Figure 4.4) to establish the area of ecosystems for which critical loads are exceeded (unprotected ecosystems or UES in RAINS terminology). Proceeding in this manner grid-element by grid-element, the total area of unprotected ecosystem over the whole European domain can be calculated. Figure 4.5 represents the results of this exercise spatially resolved over Europe for the emission situation in 1990.

Figure 4.4. Typical cumulative distribution curve of critical load for eutrophication for a grid element

The total area of unprotected ecosystems can be calculated basically for every desired emission situation. This has been used to establish factors that express the area of ecosystems that becomes unprotected against eutrophication as result of small changes in the reference emission situation for 1990. The mathematical derivation of those eutrophication factors is given in Section 4.4, but the calculation procedure will be exemplified here.

Figure 4.5 shows the total area of unprotected ecosystem in the whole European domain caused by the European emission level of 1990. This area covered to 76.96 mln ha. We can calculate the area of unprotected ecosystem for exactly the same emission situation as in 1990, except that the total nitrogen oxide emission of 300 kton by Finland is reduced by 10% to 270 kton. The area of unprotected ecosystem for this new situation amounts to 76.62 million ha. The difference between the old reference and the new emission situations is 340,000 ha. This means that as a result of the 10% reduction in emission, the deposition of N will no longer exceed the critical load for 340,000 ha of ecosystem. Figure 4.6 shows where in Europe the reduction in Finnish nitrogen oxide emission leads to a reduction in the area of unprotected ecosystem. The change in total area of unprotected ecosystem in Europe divided by the change in Finnish emissions gives the characterisation factor representing the change in unprotected ecosystem area per unit change of nitrogen dioxide emission in Finland (340,000 ha/30 kton = 11,000 ha/kton = 0.11 m²/g).

Click here to see the Figure.

Figure 4.5. Percentage of ecosystem unprotected Figure 4.6. Percentage of ecosystem eutrophication by emissions in becoming protected by reduction 1990 from all European countries of the total nitrogen dioxide emission from Finland by 10%

In the same way, eutrophication factors have been established for nitrogen oxide emissions as well as for ammonia emissions for all European regions. These factors have been calculated for the emission scenarios for 1990 and 2010. The emission scenarios and the calculated factors for terrestrial eutrophication are shown in Table 4.1 and 4.2.

The factors for terrestrial eutrophication are thus calculated in exactly the same way as the acidification factors in Chapter 3. Section 3.4 describes in more detail the followed calculation procedure including the mathematical terms.

4.4 Results

The established factors for terrestrial eutrophication are given in Table 4.1 (for 1990) and in Table 4.2 (for 2010), and aggregated to provide "site-generic" factors in Table 4.5. There are many similarities but also some notable differences to the acidification factors (see Section 3.4 and 3.5).

Similar to the acidification factors, both the 1990 and 2010 factors for terrestrial eutrophication show large differences among regions. The Scandinavian regions and Baltic regions also here have rather high factors due to their high sensitivity, whereas factors for the western and mid-European regions are usually moderate to low. In contrast to the acidification factors, however, the terrestrial eutrophication factors are moderate and even high in several cases for many eastern and southern European regions. Whereas the calcareous soils in southern Europe adequately buffer acidification, it has little influence on the sensitivity of these soils to terrestrial eutrophication.

Table 4.5. The mean, standard deviation, minimum and maximum values of the terrestrial eutrophication factors for nitrogen oxide and ammonia, for different selections from the 44 European regions (calculated as simple means and standard deviations and weighted for the national emissions). Total Europe covers all 44 European regions. EU+2 consist of Austria, Belgium, Denmark, Finland, France, Germany (old and new), Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain, Sweden, Switzerland and United Kingdom. EU consists of the same population minus Norway and Switzerland. East Europe covers Albania, Belarus, Bosnia-Herzegovina, Bulgaria, Croatia, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Macedonia, Moldavia, Poland, Romania, Russia (four regions), Slovakia, Slovenia, Ukraine and Yugoslavia.

Click here to see the Figure.

The nitrogen oxide and ammonia factors for terrestrial eutrophication are for most regions considerably higher than the respective factors for acidification. Since both impacts are expressed in the same unit, this suggests soils to be more sensitive for eutrophication than for acidification from nitrogen deposition. This makes sense, since depositions do not contribute to acidification as long as the nitrogen is assimilated by the ecosystem. However, this assimilation of nitrogen is precisely the ecosystem growth discussed in Section 4.3.3 that can lead to a shift in species composition if the supply of nitrogen exceeds the critical load for eutrophication.

Though the nitrogen content of one kg ammonia is 2.7 times higher than for one kg nitrogen oxide, the ratio between the factor for ammonia and for nitrogen oxide is considerably higher for all regions. Due to the less widespread dispersion of ammonia, reductions in emission obviously results in larger deposition reduction locally, which then leads to a stronger increase in the area of ecosystems that becomes protected by the emission reduction.

The application of the site-dependent factors will not be discussed here. Instead, the reader is referred to a detailed description and discussion in the Guidance Document (Hauschild and Potting 2003) based on this Technical Report.

4.5 Discussion

The comments and reflections on the acidification factors in Section 3.4 and 3.5 in Chapter 3 also apply to the terrestrial eutrophication factors discussed here. Chapter 3 is with minor modifications reproduced from the article of Potting et al. (1998). This article and other work by Potting and colleagues on spatial issues in LCA inspired similar research activity elsewhere, and some of the developments since 1998 are discussed here.

Krewitt et al. ( 2001) used a different model, the EcoSense model, to calculate the same type of characterisation factors as Potting et al. (1998a,b). However, the effect data for acidification and eutrophication in EcoSense are of tentative quality and the characterisation factors in Krewitt et al. (2001) for these impact categories were later withdrawn (Krewitt 2003). They are therefore not further discussed here.

Huijbregts et al. (2000) used the RAINS-model to calculate acidification and terrestrial eutrophication factors, but their definition of the category indicator differs slightly from the one used here and in Chapter 3 in that they do not cover modelling of threshold or critical load exceedance in terms of the area of unprotected ecosystems. Instead, they quantify deposition – that is, increase of exposure – divided by the critical load in grid-elements. Two sets of these so-called Hazard Indices are calculated. The first set covers all ecosystems in the model domain and is independent of the emission levels in the reference situation. The second set covers only those ecosystems that are unprotected or exposed above their critical loads and does depend on the – deposition levels determined by the – emission levels in the reference situation. The reference emission levels and subsequent deposition levels are used to select the ecosystems to be taken into account (in the "Only Above" indices), but they do not play a role as such in the calculation of the hazard indices from Huijbregts et al. (2000). The "Only Above" indices are calculated for 1995 emission levels and the estimated 2000 emission levels

The acidification and eutrophication indices or factors of Huijbregts et al. (2000) show considerable, but smaller variation than in the acidification and eutrophication factors in this Technical Report. In a comparison of the 2010 acidification factors for sulphur dioxide, nitrogen dioxide and ammonia with those presented in Chapter 3, Huijbregts et al. (2000) found a poor correlation for the "Above and Below" indices (r2 = 0.01 - 0.33), but a moderate correlation for the "Only Above" ones (r2 = 0.41 - 0.67). Potting (2004) pointed out that the variation in factors seems to become larger as the category indicator is defined closer to the endpoint. This might explain the bad correlation between the factors of Huijbregts et al. and those in this Technical Report. Another reason may be that the estimated emission levels in Potting et al. differ slightly from those in Huijbregts et al., whereas the latter also used a newer set of Critical Load data. It would be useful to compare both indicators on the basis of factors that are calculated with the same input data and the same model version.

Goedkoop and Spriensma (1999) in their Eco-indicator 1999 defined their category indicator closer to the endpoint by following a damage approach. They quantified the potentially disappeared fraction of targeted vascular plant species as a function of the deposition of acidifying and eutrophying emissions. The resulting from-midpoint-to-endpoint-factor combines two impact categories. It is based on the Nature Planner from the National Institute of Public Health and the Environment (RIVM) which only covers the Dutch territory. The model of Goedkoop and Spriensma therefore does not allow the calculation of spatially differentiated acidification-eutrophication factors. Since their work, a European database with damage functions has become available (Bakkenes et al. 2002). It would be interesting to connect this database to the RAINS-model. This would facilitate damage factors that are both site-dependent and are based on a full emission-fate-exposure-effect(-damage) modelling.

Site-dependent characterisation factors for acidification and terrestrial eutrophication have also been calculated for North America with federal states as the source regions (Bare et al. 2003). These factors represent site-dependent fate factors that express what share of an emission deposits on land.

The calculation of the site-generic factors presented in Table 4.5 need some clarification and discussion,. A site-generic terrestrial eutrophication factor becomes necessary if the geographical location of a process is unknown, or known only at an aggregated level (for instance located in West or East Europe). In such a case, one would like to assess the eutrophying impact of the process by means of a factor based on the average of the factors for the regions where the process may be located. However, regions can vary considerably in economic activity. It would therefore be fair to weigh the average for the total regional activities or total regional emissions related to the process. Also this information may be unknown and the best choice is then to use one of the European averages from Table 4.5.

For different selections of the 44 European regions, the average, standard deviation, minimum and maximum factor for terrestrial eutrophication has been determined. The site-generic factors in Table 4.5 have been calculated as simple means of the national (regional) factors of Table 4.1 and Table 4.2, and as means weighed by the country or region's total emission. The European averages or site-generic factors in Table 4.5 show, compared to the regional or site-dependent factors on which they are based, a moderate variation in the average factors for clearly different selections of regions. Considering that the standard deviations have the same size as the means, there is not significant difference between the site-generic factors but it is clear that aggregating at this level means that a large spatially determined uncertainty is hidden in the factors.

Whether or not to perform a site-dependent impact assessment is still an unsolved point of discussion in life cycle assessment. A considerable group of practitioners would rather avoid the – limited – data gathering required in addition to present typical life cycle assessment. Even though the only additional required data, the geographical site of emission, is in general already provided by current inventory analysis. However, the geographical region of a process may sometimes really be unknown. This is the case in life cycle assessments dealing with processes taking place in the – far – future or for products consisting of materials taken from the spot market (though an average as described above may be appropriate in the latter case). There are also applications where even the effort of limited additional data gathering might exceed the resources of the practitioner ^[28]. One of the averages from Table 4.5 (e.g. the one for total Europe) can then be chosen as the "site-generic" characterisation factor. Given the large ranges and standard deviations in these European averages or site-generic characterisation factors, a sensitivity analysis with the regional or site-dependent characterisation factors is highly recommended in order to evaluate whether suggested improvement options are indeed significant. The procedure for such a sensitivity analysis is described in detail in the Guidance Document of Hauschild and Potting (2003).

4.6 Conclusions and recommendations

In the same way as for acidification, characterisation and normalisation factors for terrestrial eutrophication have been established for 44 regions in Europe to facilitate site-dependent LCIA of terrestrial eutrophication. The eutrophication factors relate the region of emission to the impact on its deposition areas. The conclusions about the acidification factors established in Chapter 3 basically also apply for the terrestrial eutrophication factors and are not repeated here.

Average factors, standard deviation, minimum and maximum factors for terrestrial eutrophication have been determined for different selections of the 44 European regions. Results show a moderate variation in the average factors for clearly different selections of regions (West and East Europe), but the variation between rather similar selections of regions is small (EU and EU+2). The exact selection of regions becomes thus less important as a rough indication for the unknown area where a process is expected to take place. Whereas even a rough indication of the process' location does result in clearly different averages, the large range and even large standard deviations prevent the differences from being significant.

There are applications where even the effort of limited additional data gathering might exceed the resources of the practitioner. One of the averages from Table 4.5 (f.i. the one for total Europe) can than be chosen as a sort of "site-generic" characterisation factor. Given the large ranges and standard deviations in these averages, a sensitivity analysis with the regional characterisation factors is highly recommended in order to evaluate whether suggested improvement options are indeed significant. The procedure for such sensitivity analysis is described in detail in the Guidance Document of Hauschild and Potting (2003).

4.6.1 Acknowledgement

Research for this work was also supported by a Training and Mobility grant from the European Commission, and by the Danish Agency for Industry and Trade under the EUREKA project LCAGAPS.

4.7 References

Alcamo, J., R. Shaw and L. Hordijk (eds.). The RAINS model of acidification. Science and strategies in Europe. Dordrecht (the Netherlands), Kluwer Academic Publishers, 1990.

Amann, M., I. Bertok, J. Cofala, F. Gyarfas, C. Heyes, Z. Klimont and W. Schöpp. Cost-effective control of acidification and ground-level ozone. Second interim report to European Commission, DG-X1. Laxenburg (Austria), International Institute for Applied Systems Analysis (IIASA), 1996.

Amann, M., M. Baldi, C. Heyes, Z. Klimont and W. Schöpp. Integrated assessment of emission control scenarios, including the impact of troposheric ozone. Water, air and soil pollution, Vol. 85 (1995), Issue 4, pp2595-2600.

Bakkenes, M., R. Alkemade, F. Ihle, R. Leemans and J.B. Latour. Assessing effects of forecasting climate change on the diversity and distribution of European higher plants for 2050. Global Change Biology, Vol. 2002, Issue 8, pp390-407.

Bare J.C., Norris G.A., Pennington D.W., McKone T. TRACI: The US EPA's tool for the reduction and assessment of chemical and other environmental impacts. Journal of Industrial Ecology, Vol. 6 (2003), Issue 3/4, pp49-78.

Barrett, K. and E. Berge (eds.). Transboundary air pollution in Europe. Part 1: Estimated dispersion of acidifying agents and of near surface ozone. EMEP MSC-W status report (research report no. 321996). Oslo (Norway), Norwegian Meteorological Institute, 1996.

Berdowski, J.J.M. and W.J. Jonker. Emissions in the Netherlands. Industrial sectors, regions and individual substances (1992 and estimates for 1993; Publication series emission registration no. 21; in Dutch). The Hague (the Netherlands), Ministry of Housing, Spatial Planning and Environment, 1994.

Chardon, W.J. The role of the soil in the phosphorous cycling In: Weidema, B.P. and M.J.G. Meusen (eds.) Agricultural data for life cycle assessment. Volume 1 and 2. The Hague (the Netherlands), Agricultural Economics Research Institute (LEI), 2000,p13-24.

Downing, R.J., J.P. Hettelingh and P.A.M. de Smet (Eds). Calculation and mapping of critical thresholds in Europe. CCE status report 1993. Bilthoven (the Netherlands), Co-ordination Centre for Effects (CCE) at the National Institute of Public Health and the Environment (RIVM), 1993.

Finnveden, G., Y. Andersson-Sköld, M.-O. Samuelsson, L. Zetterberg and L.-G Lindfors. Classification (impact analysis) in connection with life cycle assessment. In: Product life cycle assessment. Principles and methodology (Nord 1992:2). Copenhagen (Denmark), Nordic Council of Ministers, 1992, pp172-231.

Finnveden, G. and J. Potting. Eutrophication as an impact category. Int. Journal of Life Cycle Assessment, Vol, 4 (1999), Issue 6, pp311-314.

Goedkoop, M. and R. Spriensma. The Eco-indicator 99. A damage oriented method for life cycle impact assessment. Methodology report and annex. Amersfoort (the Netherlands), Pré consultants, 1999.

Hauschild, M. and J. Potting. Spatial differentiation in life cycle impact assessment – the EDIP2003 methodology. Guideline from the Danish Environmental Protection Agency, Copenhagen, 2003.

Heijungs, R. G. Huppes, R.M. Lankreijer, H.A. Udo de Haes, A. Wegener Sleeswijk, A.M.M. Ansems, P.G. Eggels, R. van Duin and H.P. de Goede. Environmental life cycle assessment of products. Guide and backgrounds (ISBN 90-5191-064-9). Leiden (the Netherlands), Centre of Environmental Science of Leiden University, 1992.

Huijbregts, M., W. Schöpp, E. Verkuijlen, R. Heijungs and L. Reijnders. Spatially explicit characterisation of acidifying and eutrophying air pollution in life-cycle assessment. Journal of Industrial Ecology, Vol. 4 (2000), Issue 3, pp125-142.

Krewitt, W. A. Trukenmüller, T. M. Bachmann and T. Heck. Country specific damage factors for air pollution; A step towards site-dependent life cycle assessment. Int. Journal of Life Cycle Assessment, Vol. 6 (2001), Issue 4, pp199-210.

Krewitt, W. Personal email communication in September 2003 to the SETAC Work Group of Impact Assessment concerning the effect assessment of eutrophication and acidification in EcoSense in relation to the characterisation factors in Krewitt et al. 2001.

Lindfors, LG., M. Alkemark, C. Oscarsson and C. Spännar. A manual for the calculation of ecoprofiles intended for third party certified environmental product performance declarations. Stockholm, Swedish Environmental Institute (IVL), 1998.

Lindfors, L-G., K. Christiansen, L. Hoffman, Y. Virtanen, V. Juntilla, O-J. Hanssen, A. Rønning, T. Ekvall and G. Finnveden. Nordic guidelines on life cycle assessment (Nord 1995-20). Copenhagen (Denmark), Nordic Council of Ministers, 1995

Posch, M, P.A.M. de Smet, J.P. Hettelingh, R.J. Downing (eds.). Calculation and mapping of critical thresholds in Europe. CCE status report 1995. Bilthoven (the Netherlands), Co-ordination Centre for Effects (CCE) at the National Institute of Public Health and the Environment (RIVM), 1995.

Posch, M., J.P. Hettelingh, R.J. Downing, and P.A.M. de Smet (Eds). Calculation and mapping of critical thresholds in Europe. CCE status report 1997. Bilthoven (the Netherlands), Co-ordination Centre for Effects (CCE) at the National Institute of Public Health and the Environment (RIVM), 1997.

Posch, M, P.A.M. de Smet, J.P. Hettelingh, R.J. Downing (eds.). Calculation and mapping of critical thresholds in Europe. CCE status report 1999. Bilthoven (the Netherlands), Co-ordination Centre for Effects (CCE) at the National Institute of Public Health and the Environment (RIVM), 1999.

Posch, M, P.A.M. de Smet, J.P. Hettelingh, R.J. Downing (eds.). Modelling and mapping of critical thresholds in Europe. CCE status report 2001. Bilthoven (the Netherlands), Co-ordination Centre for Effects (CCE) at the National Institute of Public Health and the Environment (RIVM), 2001.

Potting J. Acidification and terrestrial eutrophication. Comparison of different levels of sophistication. Int J. LCA. Accepted, 2004.

Potting, J. and K. Blok. 1995. Life cycle assessment of four types of floor covering. Journal of Cleaner Production, Vol. 3 (1995), Issue 4, pp201-213.

Potting, J. and K. Blok. Spatial aspects of life cycle impact assessment. In: Udo de Haes, H.A., A.A. Jensen, W. Klöpffer and L-G. Lindfors (eds.). Integrating impact assessment into LCA. Proceedings of the LCA symposium held at the fourth SETAC-Europe Congress, 11-14 April 1994 at the Free University of Brussels in Belgium. Brussels (Belgium), Society of Environmental Toxicology and Chemistry – Europe, 1994.

Potting, J., W. Schöpp, K. Blok and M. Hauschild. Comparison of the acidifying impact from emissions with different regional origin in life-cycle assessment. Journal of Hazardous materials, Vol. 1998a, Issue 61, pp155-162.

Potting, J., W. Schöpp, K. Blok and M. Hauschild. Site-dependent life impact cycle assessment of acidifciation. Journal of Industrial Ecology, Vol. 2 (1998b), Issue 2, pp.63-87.

Wenzel, H., Hauschild M.Z. and Alting, L.: Environmental assessment of products. Vol. 1 - Methodology, tools, techniques and case studies, 544 pp. Chapman & Hall, United Kingdom, ISBN 0 412 80800 5, Kluwer Academic Publishers, Hingham, MA. USA., 1997.

5 Aquatic eutrophication

5.1 Introduction
5.2 Typical life cycle impact assessment
5.3 Effect
     5.3.1 Relevant nutrients
5.4 Ecological effects
     5.4.1 Concept of the limiting nutrient
5.5 The CARMEN model
     5.5.1 Emission estimates
     5.5.2 Fate and exposure assessment
5.6 Calculation procedure
5.7 Results
5.8 Discussion
     5.8.1 Application of the eutrophication factors
     5.8.2 Proper inventory data
     5.8.3 Biological oxygen demand
     5.8.4 Related research activities
5.9 Conclusions
5.10 References
Annex 5.1: The river catchments modelled in CARMEN
Annex 5.2: The coastal seas modelled in CARMEN

Authors: José Potting ^[29], Arthur H.W. Beusen ^[30], Henriette Øllgaard ^[31],
Ole Christian Hansen31 , Bronno de Haan ^[32] and Michael Hauschild ^[33]

5.1 Introduction

Although still of good quality in some regions, European inland waters (fresh waters like rivers and lakes) as well as coastal seas are threatened by a multitude of human activities. The growing number of both users and uses ^[34] has increased the exploitative pressure on the European waters. This gives more and more conflicts between the various uses and users, and also between man and organisms living in or near those waters. A decline in quantity and quality of the water available for human uses, as well as of the ecological quality of aquatic ecosystems may be the result.

A release of pollutants to water may have considerable local impact (in case of a point source), but its main impact seems its contribution to pollution accumulation by other sources upstream and downstream in rivers and finally in coastal waters. Water quality, and eutrophication or nutrient enrichment as part of that, has therewith become a regional topic. Kristensen and Hansen (1994) provide an interesting and comprehensive overview of the state of the environment in European rivers and lakes. EEA (1998) gives a short overview for both marine and inland waters.

The focus of this chapter is on life cycle assessment of long-range nutrient enrichment of marine and inland waters. The presently typical way to assess eutrophication in life cycle assessment does not account for spatial variation in emissions, exposure and effect (Section 5.2). This chapter will explore the differences in eutrophying impact from nutrients released at different geographical locations and by different source categories. Section 5.4 describes the CARMEN-model, a model for integrated assessment of eutrophication, which is used to establish spatial resolved exposure factors. The CARMEN-model does not cover an assessment of ecological effect, but Section 5.3 gives a brief review on the issue. Section 5.5 describes how the CARMEN model is used to establish spatial resolved eutrophication factors. The results are presented in Section 5.6 and discussed in Section 5.7. Section 5.8 draws the main conclusions.

5.2 Typical life cycle impact assessment

The eutrophying impact usually characterised in life cycle impact assessment relates implicitly to eutrophication of aquatic ecosystems. This follows from the modelling of impact that in life cycle assessment typically takes its bases in the composition of aquatic biomass (Wenzel et al. 1997, Heijungs et al. 1992, Lindfors et al. 1995):

An increased input of nutrients into an aquatic ecosystem may lead to an increased production of biomass, primarily phytoplankton (see also Section 5.3). The typical composition of phytoplankton, as suggested by Redfield et al. (1993) ^[35] and commonly used in life cycle impact assessment, consists of carbon, nitrogen and phosphorus in a ratio of 106:16:1. This ratio suggest that 106 mole carbon, 16 mole nitrogen and 1 mole phosphorus is needed for the production of 1 mole of phytoplankton.

In life cycle impact assessment of eutrophication, the Redfield ratio is used the other way round to express the contribution of emissions of nitrogen and/or phosphorus to biomass production in terms of the equivalent emission of a reference substance. The reference substance differs between methods, but these methods are further nearly similar. Table 5.1 lists equivalency factors as proposed by Wenzel et al. (1997).

Table 5.1. Equivalency factors for nutrient enrichment from Wenzel et al. (1997).

Wenzel et al. (1997) suggest aggregation of the different nitrogen substances separate from those of phosphorus to allow site-considerations in the subsequent weighing of the accumulated substances (see Table 5.1 and Formula 5.1). This separate characterisation of phosphorus and nitrogen does anticipate inland freshwaters typically to be phosphorus limited while marine waters are typically limited by nitrogen.

However, Wenzel et al. do also provide factors to add together nitrogen and phosphorus (see Table 5.1 and Formula 5.1):

Click here to see the Figure.

Wenzel et al. (1997) do not distinguish between differences in eutrophying impact from emissions to air, water and soil, and also not distinguish between eutrophication of terrestrial and aquatic ecosystems. Terrestrial ecosystems get eutrophied mainly by airborne emissions ^[36], ^[37], however, whereas waterborne nutrients from waste water and originating from agriculture dominate eutrophication of aquatic ecosystems (though stagnant water can be polluted by air depositions as well). The agricultural nutrient data in life cycle inventory usually refer to the amount after plant uptake and binding. This amount can leave the topsoil and may transport to surface water depending on environmental conditions (like soil type or net precipitation). Wenzel et al. (1997) does also not consider variation in environmental conditions. They suppose all types of emissions to fully contribute to eutrophication in general.

Similar to Finnveden et al. (1992), we consider eutrophication of terrestrial and aquatic ecosystems as two separate impact categories. This chapter deals with eutrophication of aquatic ecosystems (Chapter 4 addresses terrestrial eutrophication). This chapter focuses on the geographical differences in transport of nutrients through the environment to inland waters and marine waters. The biological aspects of eutrophication will be described first, however.

5.3 Effect

The term "eutrophic" is a biological one and literally means "rich of nutrients", whereas the term "eutrophication" refers to "the process of becoming rich of nutrients" (Dikke van Dalen 1994, Kristensen and Hansen 1994) ^[38]. This section first elaborates on the nutrients relevant in eutrophication (Section 5.3.1), then describes the potential ecological effects from eutrophication (Section 5.3.2), and finally addresses which nutrients may limit biomass growth (Section 5.3.3).

5.3.1 Relevant nutrients

The production of biomass in aquatic ecosystems is governed at any time and place by numerous factors simultaneously. However, the overall productivity under otherwise comparable conditions is largely determined by those factors that limit production over a substantial length of time during the main growth period. This long-term limitation is most often – though not always – due to one nutrient being poorly available in the given ecosystem itself and/or not present in sufficient quantities in natural supplies from outside.

The concept of "the limiting nutrient" is essential when discussing nutrient enrichment or eutrophication. It states that one nutrient is limiting the growth of primary producers (plants) and thereby indirectly affects the ecosystem and that there is an excess of all other nutrients. An additional amount of the limiting nutrient will lead to increased growth, however, additional amounts of other nutrients will not since they are already in excess. (Finnveden and Potting 1999)

Those nutrients most likely to act as limiting factors can be identified by ranking the amount and proportions of all critical elements in aquatic ecosystems against their concentration and proportion in the – unpolluted – water. Hydrogen, oxygen and most salts (like calcium, magnesium, potassium, sodium and chloride) are usually available in abundance and can therefore be eliminated. Carbon dioxide, (bi)carbonates and sulphate are generally available in excess compared to nitrogen and phosphorus in both freshwaters and marine waters. This makes nitrogen and phosphorus the primary candidates for "chronic" nutrient limitation ^[39].

Both nitrogen and phosphorus exist in many different chemical forms in the aquatic environment. Impact and bio-availability of each form may be different, but they have the ability to interchange. The transformations between the several forms are site-specific and continuous (known as the nitrogen and phosphorus cycle). All forms of nitrogen and phosphorus forms are therefore relevant, and total nitrogen and phosphorus are a good expression for long-term eutrophication.

Trace elements may in rare cases be limiting. Primary producers as phytoplankton only need very small quantities of these elements (Hauschild 1998), however, and they are therefore not further addressed here.

5.4 Ecological effects

The supply of nutrients to water is under natural circumstances in balance with the subsequent growth of biomass such that a stable though variable society of plant and animals is ascertained. Such society is to some extent able to cope with variations in nutrient supplies. A too large nutrient increase pushes this stable society out of balance, however, and may through a chain of ecological effects provoke a shift of the biological structure. The effects are generally more apparent in standing or slowly moving waters than in (small) fast flowing rivers (Kristensen and Hansen 1994). The chain of ecological effects is illustrated in Figure 5.1:

Figure 5.1. Schematic representation of the ecological chain effects as a result of the increase with nutrients or eutrophication of lakes (reproduced from Kristensen and Hansen 1994). See below text for an explanation of the arrows.

Standing and shallow clear waters with submerged plants become dominated by phytoplankton after a high nutrient input (→1). Increasing growth of phytoplankton makes the water turbid (→2). The turbidity prevents the light to reach the water-bottom, which makes the submerged plants disappear (→3). The community fish also changes because the predatory fish are unable to see and catch the smaller fish (→4). The fish community becomes dominated by zooplankton-eating species that are more tolerant to the turbidity (→5). Zooplankton eats phytoplankton and the decrease in zooplankton (→6) therefore results in a further increase of phytoplankton (→7). The excess phytoplankton dies and sinks to the bottom (→8). The decay of phytoplankton uses oxygen (→9). The decrease of oxygen leads to further fish kills and disappearance of bottom fauna (→10), and may also evoke a release of phosphorus from the lake bottom (→11). The released phosphorus may be used for a new increase of phytoplankton (→12) after which a new cycle of phytoplankton dying and sinking starts.

Topography and the physical/chemical nature of the water influence the impact of nutrients on aquatic biomass growth. It is therefore difficult to define generic nutrient levels above which this growth may become problematic. Kristensen and Hansen (1994) give water quality indications for inland waters and EEA (1998) for marine waters (see Table 5.2).

Table 5.2. Water quality categories abstracted from EEA (1998) and Kristensen and Hansen (1994). These levels have to be taken with caution since threshold values are influenced by topography and physical/chemical nature of the water. The nutrient levels refer to the extent of anthropogenic origin rather than to ecological effect.

Based on EEA (1998)

Based on Kristensen and Hansen (1994)

Drinking Water Directive (80/778/EEC)

Kristensen and Hansen (1994) collected and analysed monitoring data about – amongst others – nitrogen and phosphorus concentrations from 800 stations in more than 550 European rivers and 1500 lakes. The annual mean concentrations in the majority of the rivers range from levels of moderate to heavy pollution. Only rivers in northern Europe seem hardly disturbed. Particularly rivers in western Europe show high nitrogen and phosphorus levels. The situation looks somewhat better for European lakes, though the same trends can be observed here.

The water quality indications in Table 5.2 relate nutrient levels to the extent of anthropogenic origin rather than to the occurrence of ecological effects caused by biomass growth. Kristensen and Hansen (1994) also describe the biological state or rivers in a number of European countries. Their results are given in Table 5.3 and confirm the trends shown by the monitored data as described in the previous paragraph.

EEA (1998) describe the water quality in European seas. They gives annual mean nitrogen and phosphorus concentrations at a number of sample points in west and north European seas and qualitative information about southern seas.

Obviously, nutrient concentrations are high and water quality is often poor/bad close to estuaries of rivers (southern part of the North sea, Irish Channel, Waddenzee). The Mediterranean Sea is one of world's most oligotrophic (nutrient-poor) seas. However, eutrophication problems occur in semi-enclosed bays (mainly due to untreated sewage) and uncontrolled expansions of fish farming (Eastern Mediterranean). The northern and west coast of the Adriatic Sea, that receives the nutrient loads of the River Po, is the most endangered in the Mediterranean area. The long residence time of the water in the Black Sea makes this sea highly sensitive to eutrophication. The water quality in this sea and the Sea of Azov is bad, probably due to high inputs from the Danube, Dnieper and Dniester.

Table 5.3. Percentage of river reaches in various European countries classified as being of good, fair, poor or bad quality. Of good quality are river reaches with nutrient-poor water, low levels of organic matter, saturated with dissolved oxygen, rich invertebrate fauna, and suitable spawning ground for salmonid fish. River reaches with moderate organic pollution and nutrient content, good oxygen conditions, rich flora and fauna, large fish population are classified as fair. Poor quality river reaches have heavy organic pollution, usually low oxygen concentrations, locally anaerobic sediment, occasional mass occurrence of organism insensitive to oxygen depletion, small or absent fish population, periodic fish kill. Of bad quality are those rivers with excessive organic pollution, prolonged periods of very low oxygen concentration or total deoxygenation, anaerobic sediment, severe toxic input, devoid of fish. (Kristensen and Hansen 1994)

5.4.1 Concept of the limiting nutrient

The concept of "the limiting nutrient" is essential when discussing nutrient enrichment or eutrophication. Generally seawaters are regarded to be "chronic" limited by nitrogen, whereas lakes and larger, slowly flowing rivers are limited by phosphorus. Estuaries can be limited both by nitrogen and phosphorus, whereas small streams and mountainous lakes are usually not limited by nitrogen and phosphorus, but by factors as light and temperature (Miljøstyrelsen 1991, Blau and Seneviratne 1995).

The concept of "the limiting nutrient" is a simplification because the limiting nutrient may change over seasons, and also over the years due to earlier loading. (Kristensen and Hansen 1994, Finnveden and Potting 2000) All nutrients can therefore potentially be used for biomass growth, though this maximum will for nitrogen usually not be the case in inland waters and for phosphorus not in marine waters.

5.5 The CARMEN model

CARMEN is an acronym for Cause effect Relation Model to support Environmental Negotiations. It is an integrated assessment model to analyse and evaluate strategies to reduce nutrient loading of inland waters ^[40] and coastal seas in Europe. The model does actually not contain an assessment of ecological effects, but calculates the change in nutrient loads in ground water, inland waters (river catchments) and coastal seas from changes in nutrient emissions and supplies (i.e. the causes). CARMEN, version 1.0 is used in this study.

The causes modelled by CARMEN are atmospheric deposition of nitrogen on soil and coastal seas, phosphorus and nitrogen supply to agricultural soils, and phosphorus and nitrogen in municipal wastewater (see Figure 5.2 and Section 5.4.1). CARMEN models the transport of nutrient by rivers to sea relatively straightforward, but the transport of nutrients from agricultural supply and atmospheric deposition through groundwater drainage and surface runoff, to surface water is modelled spatially resolved over 124320 grid-elements of 10*10 minutes (roughly 100-250km², depending on the longitude and latitude location of the grid-element). (see Section 5.4.2)

Meinardi et al. (1994a,b), Klepper et al. (1995) and Beusen (not published) describe the CARMEN model in detail, but a short overview is given here about estimates of emissions (Section 5.4.1) and transport via groundwater drainage and surface runoff (5.4.2) of nutrients as modelled in CARMEN.

5.5.1 Emission estimates

Main sources for nitrogen in the aquatic environment are – via groundwater drainage and surface runoff – the use of fertiliser and manure in agriculture (diffuse sources), and municipal wastewater and – too a lesser extent – industrial effluent (mainly point sources). Deposition of airborne nitrogen is of some relevance for marine waters, but has usually minor importance for inland waters (except for the indirect contribution via groundwater drainage and surface runoff). Natural nitrogen can be important in regions with low population density. (EEA 1998, Kristensen and Hansen 1994) Figure 5.3 gives an overview of the relative importance of the different sources.

Most of the phosphorus loading of inland waters is attributable to discharge from point sources, especially municipal wastewater and – too a lesser extent – industrial effluent. The application of phosphorus in agriculture may also contribute to inland waters through topsoil erosion. Phosphorus will usually not reach the ground water due to strong adsorption to the soil, though it exceptionally occurs after excessive loading on very poor soils (EEA 1998, Klepper et al. 1995, Kristensen and Hansen 1994)

Figure 5.2. Main sources for nitrogen (continuous arrow) and phosphorus (dashed arrow) to soil, groundwater, surface waters and coastal seas addressed in the CARMEN model (Beusen not published).

CARMEN models three main sources for nitrogen and phosphorus to surface water (see also Figure 5.2): agriculture, municipal wastewater and atmospheric deposition (only for nitrogen). The several nitrogen and phosphorus supplies have been allocated to each grid-element on the basis of the distribution of land uses in the given grid-element (arable land, grassland, permanent crops, forest, urban area, inland waters, others).

5.5.2 Fate and exposure assessment

The CARMEN model distinguishes between sources directly and indirectly via inland waters releasing to coastal seas.

Oxygen depletion has been a severe problem for several years, but implementation of biological treatment of domestic and industrial wastewater has resulted in that many rivers and lakes are now fairly well oxygenated. However, low oxygen content is a usual phenomenon in the first stretches downstream of point sources for wastewater treatment. Under low oxygen conditions, nitrate is decomposed into nitrogen gas and oxygen used for the biomass decomposition. The nitrogen gas is released to the atmosphere. Removal of nitrogen is modelled for all waters the same in CARMEN by assuming a generic removal of 30% of the nitrogen input (i.e., 70% transports to sea).

Click here to see the Figure.

Figure 5.3. (Figure 9.11 and 9.14 from EEA 1998, p201-202)

Phosphorus on its turn may be used for biomass production or temporarily adsorbed in the phosphorus pool in the bottom sediments of lakes and rivers. All phosphorus will remain in the water, however, and in the end transport to marine waters. CARMEN therefore considers no removal of phosphorus (i.e. all of it transports to sea).

The strength of CARMEN is a detailed spatial resolved modelling of the transport of nutrients from agricultural supply and atmospheric deposition – through groundwater drainage and surface runoff or topsoil erosion – to surface water. The water flow is the main transport mechanism that brings nutrients – through deep groundwater drainage (nitrogen), runoff (nitrogen) or topsoil erosion (phosphorus) – to surface water. The net precipitation determines the amount of water draining to either groundwater or by surface runoff.

The ratio between deep groundwater recharge and surface runoff resulting from net precipitation is determined in CARMEN by:

Aquifer type (permeability of subsurface)
Texture of the topsoil (grain size)
Slope of land surface
Land use and land cover
Seasonal temperature

The flow of deep groundwater to surface water is in addition determined by:
Thickness of the unsatured zone
Groundwater flow or groundwater age
Aquifer type

Nitrogen transport through deep groundwater and surface runoff to surface water further depends on the amount leaching to the soil water at shallow depth after plant uptake and binding in the topsoil matrix. The share of nitrogen leaching is fixed for arable land by soil type and for agriculturally used grasslands, it is increasing with nitrogen dose for all soils (see also Section 5.7.2.).

Phosphorus concentrations in groundwater are usually negligible because of the strong adsorption to the soil. Phosphorus transport to surface water is dominated by erosion of sediment from the topsoil by surface water runoff. (Meinardi et al. 1994/1995)

Each of the above factors are modelled by CARMEN spatially resolved over 124320 grid-elements of 10*10 minutes (roughly 100-250km², depending on the longitude and latitude location of the grid-elements).

5.6 Calculation procedure

Site-dependent eutrophication factors have been established with help of the CARMEN model for 32 European countries. The factors relate the amount of nutrient released in a given country to its share to eutrophication of

European inland waters ^[41] and coastal seas (i.e., respectively the share of nutrient indirect to seas through inland waters and the share of nutrient indirectly and directly to seas). The factors quantify the maximum biomass growth these nutrients may contribute to in the receiving water. The factors do not anticipate the deterioration of the water quality as a result of this biomass growth.

The eutrophication factors per country were calculated by changing the total amount of either nitrogen or phosphorus from a given source category in one country (other emissions for all countries and other source categories remaining the same). Agricultural supplies of manure or fertiliser to soil, and wastewater releases have been considered as sources for both inland waters and seas. The calculations for coastal seas in addition also addressed atmospheric deposition as a nitrogen source.

For each source category, the change in eutrophying loads is calculated spatially resolved over 101 river catchments and 32 coastal seas (see Annex 5.1 and 5.2). Next, the loading increases by this change of one country are accumulated over all river catchments and seas to obtain the factors expressing the share eutrophying respectively inland waters and seas (in kg per kg released).

CARMEN models the contributions from nitrogen depositions to eutrophication of inland waters and seas from the actual deposition pattern in Europe. It does not contain the relationships between country emissions and depositions. Huijbregts and Seppälä (2000) recently published data about the nitrogen deposited on European seas as a ratio of the emission in their country of release. These data have been combined with calculations with the CARMEN model to arrive at the loading of separate seas. The data from Huibregts and Seppälä do not consider deposition outside the EMEP model domain, and have therefore been corrected with data from Barrett and Berge (1996) about the percentages exported outside the EMEP model domain.

The CARMEN model does not include an assessment whether nutrient loading actually results into biomass growth and what effect this has on the ecological quality of the water. The calculated factors represent the potentially maximum contribution to biomass growth. They are exposure factors rather than expressing ecological effects.

5.7 Results

The results of the calculations are abstracted in Table 5.4. Columns 2 to 7 in Table 5.4 give the share of nutrient from a given source category that contributes to eutrophication of inland waters. The last eight columns do the same for marine waters.

Table 5.4. The share of nitrogen and phosphorus eutrophying inland waters (Column 2 to 7) and marine waters (Column 8 to 15) for different source categories in different countries (Fert.=fertiliser, Man.=manure, WW=wastewater, NH₃=airborne NH₃ , NO_x=airborne NO_x). Hence that the factors for fertiliser and manure relate to nutrient after plant uptake (see Section 5.7.2 for how to arrive at the proper inventory data). There are a number of countries where eutrophication of marine waters by agriculture and wastewater is much smaller than their eutrophication of inland waters (underlined). These numbers relate to countries releasing on seas outside the model domain (they may be replaces with factors for inland waters as best estimate and by a factor 0.7 for nitrogen and a factor 1.0 for phosphor in wastewater). The out-crossed numbers are concluded to be wrong. For Denmark, the factor for marine waters may be lowered to 0.7 for nitrogen and to 1.0 in phosphor in wastewater. The other numbers being striketrough should be replaced by the site-generic factors. Both underlined and out-crossed numbers are excluded from the mean etc.

Hence the factors relating to application of fertiliser and manure refer to the amount of nutrient after plant uptake. This amount is available to leave the topsoil. It is a peculiarity of life cycle assessment to consider the topsoil as part of the economic system or technosphere and the data from inventory analysis should reflect the outputs from the technosphere to the ecosphere. One should thus be careful to apply the factors in Table 5.4 to the proper inventory data (see also Section 5.7.2).

The factors in Table 5.4 show eutrophication from fertiliser and manure basically to be similar. There is therefore no reason to employ different factors for these source categories in life cycle assessment.

There is neither a difference in the mean factors for agricultural eutrophication of inland waters and marine waters. The eutrophication factors from nitrogen ^[42] show notable differences for a number of individual countries (i.e. the Netherlands, the Scandinavian countries, the United Kingdom). The main cause is that along the relative long coastline, small water systems are discharging directly to marine waters. Table 5.4 shows the Swiss and Austrian factors for inland waters to deviate from the one for marine waters, while both countries do not have a direct connection to sea. The reason for this deviation is not clear.

There are also a number of countries where agricultural eutrophication of seas seems to be much smaller than their eutrophication of inland waters (i.e. Russia, Spain, Portugal, Iceland and Ireland). However, some rivers in these countries release on marine waters that are outside the model domain of CARMEN. The sea factors may here be corrected by replacing them for the inland water factors (which are expected to be a fair estimate).

Similar as for nitrogen from agriculture, differences exist between the factors for eutrophication of coastal seas and inland waters by wastewater. These factors express basically the share of nutrient (wastewater) discharged indirectly via inland waters or indirectly plus directly to marine waters. The factors for marine waters and inland waters can be at maximum 0.7 for nitrogen and 1.0 for phosphorus. Hence the Icelandic factors for inland waters are larger then these maximum values and thus probably false. The same applies for Denmark's factors for seas. Small exceedances (like 0.01) of the maximum values (0.7 or 1.00) are expected to be rounding errors.

Within source categories, the differences between countries are relatively small. There is a factor 3 for agricultural nitrogen and a factor 7 for phosphorus between lowest and highest factor. The nitrogen factors for the Netherlands and Denmark (both having intensive agriculture) are remarkable low. This is due to the fact that the groundwater recharge of both countries is relative large compared to surface runoff (ratio larger than one). This means that the main part of the nutrients will flow into the groundwater, and due to a time delay of more than 10 years and a mixing with unpolluted groundwater, only a small part of the nutrients will flow out, to the inland waters. The phosphorus factors for Austria and Switzerland are remarkable high. The slope seems here to have a considerable impact on surface runoff and erosion.

5.8 Discussion

Some remarks about the eutrophication factors have already been made in the previous section and will not be repeated here. However, a few additional aspects are discussed in more detail in the following.

5.8.1 Application of the eutrophication factors

The application of the eutrophication factors in Table 5.5 is very simple. An emission in the product's life cycle is multiplied with the eutrophication factor for that region and substance to derive the estimated eutrophication of that emission. The only additional data required, the country where an emission takes place ^[43], are in general already provided by current LCI.

The receiving waters can be inland waters on the one hand, and marine waters on the other hand. Both should be regarded in life cycle assessment as subcategories and thus reported separately (i.e. not to aggregate them). It is recommended to refrain from aggregating phosphorus and nitrogen, since nitrogen may at some point leave the aquatic system by nitrification, whereas phosphorus does basically not leave and remain to be available for potential eutrophication.

This leads thus all together to two subcategories (inland waters and marine waters) with each two impact indicators (aggregated phosphorus and aggregated nitrogen). The Guidance Document of Hauschild and Potting (2003) based on this Technical Report gives a detailed description of the application procedure for the aquatic eutrophication factors.

5.8.2 Proper inventory data

It is common practice in life cycle assessment to consider the topsoil of agricultural soils as part of the technosphere. The data in life cycle inventory for nutrient supply in agriculture therefore usually refer to the amount of nutrient available for leaving the topsoil after plant uptake and binding. (Weidema and Meusen 2000). The factors from Table 5.4 also refer to the amount of nutrient available for leaving the topsoil after plant uptake and binding. The eutrophication factors from Table 5.5 can therefore directly be multiplied with that type of inventory data.

There is also a number of practitioners who do not consider the topsoil as part of the technosphere or lack data for implementing that, and stop their inventory at the level of nutrient supply. Inventory data reflecting nutrient supplies should first be corrected for plant uptake and binding before applying the eutrophication factors from Table 5.4. The factors in Table 5.5 are based on the relationships employed by CARMEN and can be used for this correction:

Table 5.5. The factors expressing the relation between nutrient application and nutrients available for drainage by groundwater and surface runoff or erosion. The factors in this table can be multiplied with the factors in Table 5.4 to obtain the relation between nutrient application and eutrophication of inland waters and coastal seas.

According to Table 5.5, less then 25% of the nitrogen application and 10% of the phosphorus application leaches after plant uptake to soil water at shallow depth and is then available for transport through deep groundwater and surface runoff or erosion. This means that on average less then 12% of the supplied nitrogen and 5% of the supplied phosphorus will end up to be available for biomass growth in inland waters and finally coastal seas. This is in accordance with values given in the literature. (Weidema and Meeusen 2000)

A similar situation is at stake for the phosphorus and nitrogen emissions to water. The waterborne nitrogen and phosphorus emissions in life cycle inventory usually refer to the content in effluent after wastewater treatment (unless emissions are released actually directly to surface water). The factors from Table 5.4 refer to releases directly to surface water (after potential purification thus). The eutrophication factors from Table 5.5 can therefore directly be multiplied with this type of inventory data. In case the inventory data do refer to wastewater before purification, the factors in Table 5.6 can be used in connection with Table 5.4 to correct the inventory data.

Table 5.6. The expected concentrations (in mg/l) and removal efficiencies (in g/g) for nitrogen and phosphorus in waste after different types of waste water treatment (Hansen, 2000).

5.8.3 Biological oxygen demand

Decomposition of organic material used oxygen that is measured as chemical or biological oxygen demand (COD or BOD). Fish kill and disppearance of bottom fauna may occur after prolonged situations of oxygen depletion. This may be the result of prolonged and high nutrient loading. Some methods presently typical for life cycle assessment of aquatic eutrophication therefore suggest considering organic materials, as part of this impact category (Lindfors et al. 1995, Heijungs et al. 1992). This suggestion is not followed here, since organic material often contains nitrogen or phosphorus and is then reflected already in the inventory table as nitrogen and/or phosphorus emission. To also add organic materials to this impact category, would then lead to double counting. Furthermore, organic material is not essentiel for growth of the biomass. Emissions of organic material, which causes BOD/COD, may increase or strengthen the effects and the environmental impacts of nutrient enrichment, especially with regards to oxygen depletion. However, these environmental impact of BOD/COD substances are comparable to the secondary effects of nutrient enrichment.

5.8.4 Related research activities

During and after the work for this chapter, similar research activities were performed by other groups. Huijbregts and Seppälä (2000) used spatially resolved source-receptor matrices to calculate the share of atmospheric emission from source countries that deposits on marine waters. The rest deposits on land, but can transport through deep groundwater and surface runoff to surface water and marine water. The depositions of Huijbregts and Seppälä have been gratefully used here to calculate site-dependent factors that cover the direct and indirect contribution from the atmospheric emissions in all European countries to eutrophication of surface and marine waters.

In addition to Huijbregts and Seppälä (2000), Huijbregts and Seppälä (2001) established characterisation factors for aquatic eutrophication, covering fate and potential effect for the Netherlands, Europe and the world. The fate part of these factors seem – at a first sight – less sophisticated than what is proposed in this chapter, whereas the effect factor is similar to what is proposed here. A closer comparison is recommended, however.

Bare et al. (2003) calculated similar factors using a similar type of modelling as in this chapter for the federal states in North America. Due to the advanced state of the reporting of the Danish project, it was unfortunately not possible to make a closer comparison here.

5.9 Conclusions

Eutrophication factors have been established for 32 countries in Europe to facilitate site-dependent assessment of eutrophication in life cycle assessment (see Table 5.5). The factors relate the country of emission to the potential maximum contribution to biomass growth in the receiving waters.

The potential maximum will usually not be the actual contribution. Generally however, seawaters are regarded to be "chronic" limited by nitrogen, whereas lakes and larger, slowly flowing rivers are limited by phosphorus. Interpreting the results from impact assessment can use this and it is therefore recommended to refrain from aggregating nitrogen and phosphorus.

Receiving waters can be inland waters and coastal seas (directly or indirect by river discharge to seas). Inland waters and coastal waters are recommended as separate subcategories (aggregating would lead to double counting).

The established eutrophication factors are exposure factors rather than expressing the ecological effect of eutrophication and biomass growth. The present state-of-the art in integrated assessment modelling of aquatic eutrophication does not yet allow such effect assessment. The spatial resolved factors in Annex 5.3 to 5.9 may be used for a tentative interpretation in this direction.

5.10 References

Bare J.C., Norris G.A., Pennington D.W., McKone T. TRACI: The US EPA's tool for the reduction and assessment of chemical and other environmental impacts. Journal of Industrial Ecology., Vol. 6 (2003), Issue 3/4, pp49-78.

Barrett, K. and E. Berge (eds.). Transboundary air pollution in Europe. Part 1: Estimated dispersion of acidifying agents and of near surface ozone. EMEP MSC-W status report (research report no. 321996). Oslo (Norway), Norwegian Meteorological Institute, 1996.

Berdowski, J.J.M. and W.J. Jonker. Emissions in the Netherlands. Industrial sectors, regions and individual substances (1992 and estimates for 1993; Publication series emission registration no. 21; in Dutch). The Hague (the Netherlands), Ministry of Housing, Spatial Planning and Environment, 1994.

Beusen, A. User manual of CARMEN1 (Draft). Bilthoven (the Netherlands), National Institute of Public Health and Environmental Protection (RIVM), not published.

Blau, S. and S. Seneviratne. Acidifiction and eutrophication in life cycle assessments (LCAs). Zürich (Switzerland), Swiss Federal Institute of Technology Zurich, 1995.

EEA Europe's environment: The second assessment. Copenhagen (Denmark), European Environmental Agency,1998.

Finnveden, G., Y. Andersson-Sköld, M.-O. Samuelsson, L. Zetterberg and L.-G Lindfors. Classification (impact analysis) in connection with life cycle assessment. In: Product life cycle assessment. Principles and methodology (Nord 1992:2). Copenhagen (Denmark), Nordic Council of Ministers, 1992, pp172-231.

Finnveden, G. and J. Potting. Eutrophication as an impact category. Int. Journal of Life Cycle Assessment, Vol, 4 (1999), Issue 6, pp311-314.

Hansen, O.C: Ole Christian Hansen, unpublished work, 2000.

Heijungs, R., J. Guinée, g. Huppes, R.M. Lankreijer, H.A. Udo de Haes, A. Wegener Sleeswijk, A.M.M. Ansems, P.G. Eggels, R. Van Duin en H.P. de Goede. Environmental life cycle assessment of products. Guide and background (ISBN 90-5191-064-9). Leiden (the Netherlands), Centre of Environmental Science of Leiden Univerisity, 1992

Huijbregts, M. and J. Seppälä. Towards region-specific, European fate factors for airborne nitrogen compounds causing aquatic eutrophication. Int. Journal of Life Cycle Assessment, Vol. 5, Issue 2, pp65-67, 2000.

Huijbregts, M. and J. Seppälä. Life cycle impact assessment of pollutants causing aquatic eutrophication. Int. Journal of Life Cycle Assessment, Vol. 6, pp339-344, 2001.

Jaarsveld, J.A. van, and D. Onderdelinden. TREND: An analtical long-term deposition model for multi-scale purposes (report no. 228603009). Bilthoven (the Netherlands), National Institute of Public Health and Environmental Protection (RIVM), 1990.

Klepper, O., A.H.W. Beusen and C.R. Meinardi. Modelling the flow of nitrogen and phosphor in Europe: from loads to coastal seas (RIVM report 461501004). Bilthoven (the Netherlands), National Institute of Public Health and Environmental Protection (RIVM), 1995.

Kristensen, P. and H. O. Hansen. European rivers and lakes. Assessment of their environmental state (EEA environmental monographs 1). Copenhagen (Denmark), European Environmental Agency, 1994.

Lindfors, L-G, K. Christiansen, L. Hoffman, Y. Virtanen, V. Juntilla, O-J. Hanssen, A. Rønning, T. Ekval and G. Finnveden. Nordic Guidelines on life cycle assessment (Nord 1995; 20). Copenhagen (Denmark), Nordic Council of Ministers, 1995.

Meinardi, C.R., A.H.W. Beusen, M.S.J. Bollen and O. Klepper. Vulnerability to diffuse pollution of European soils and groundwater (RIVM report 461501002). Bilthoven (the Netherlands), National Institute of Public Health and Environmental Protection (RIVM), 1994a.

Meinardi, C.R., A.H.W. Beusen, O. Klepper and W.J. Willems. Nitrate contamination of European soils and groundwater (RIVM report 461501003). Bilthoven (the Netherlands), National Institute of Public Health and Environmental Protection (RIVM), 1994b.

Miljøstyrelsen. Environmental impacts of nutrient emissions in Denmark. Redegørelse fra Miljøstyrelsesn, no. 1). Danish Environmental Protection Agency, Copenhagen, 1991.

Paaby H, Møller F, Skop E, Jensen JJ, Hasler B Bruun H, Asman WAH. Omkostninger ved reduktion af næringsstofbelastningen af havområderne – Metode, model, analyse. Faglig rapport fra DMU, nr. 165. Danmarks Miljøundersøgelser, 1996.

Redfield, A.C., B.H. Ketchum and F.A. Richards. The influence of organisms on the composition of sea water. In: Proceedings of the 2nd International Water Pollution Conference in Tokyo. Oxford (United Kingdom), Permagon Press, 1993, pp215-243.

Samuelsson, M.-O. Life cycle assessment and eutrophication. A concept for calculation of the potential effects of nitrogen and phosphorus. Stockholm (Sweden), Swedish Environmental Research Institute (IVL), 1993.

Weidema, B.P. and M.J.G. Meeusen (eds.) Agricultural data for life cycle assessment. Volume 1 and 2. The Hague (the Netherlands), Agricultural Economics Research Institute (LEI), 2000.

Wenzel, H., M.Z. Hauschild and L. Alting: Environmental assessment of products. Vol. 1 - Methodology, tools, techniques and case studies, 544 pp. Chapman & Hall, United Kingdom, 1997, Kluwer Academic Publishers, Hingham, MA. USA. ISBN 0 412 80800 5.

Wither, P.J.A. Cycling and sources of phosphorus in agricultural systems and to the wider environment: a UK perspective. In: Weidema, B.P. and M.J.G. Meusen (eds.) Agricultural data for life cycle assessment. Volume 1 and 2. The Hague (the Netherlands), Agricultural Economics Research Institute (LEI), 2000, pp13-24.

Annex 5.1: The river catchments modelled in CARMEN

Annex 5.2: The coastal seas modelled in CARMEN

Click here to see the Figure.

1. Irish sea (eastern part)
2. Irish sea (St. George Channel)
3. Irish sea (western part)
4. Celtic sea
5. English Channel (western part)
6. English Channel (eastern part)
7. Golf of Biscay
8. Atlantic ocean (around Scotland)
9. North sea/Norwegian sea
10. North sea (northern part)
11. North sea (southern part)
12. Skagerrak
13. Kattegat
14. Øresund/Great and Small Bealt
15. Baltic sea (west from Gotland)
16. Baltic sea (below 15)
17. Baltic sea (east from Gotland)
18. Baltic sea (below 17)
19. Gulf of Riga
20. Gulf of Finland
21. Gulf of Bothnia (southern part)
22. Gulf of Bothnia (northern part)
23. Norwegian sea
24. Venice bay
25. Adriatic sea (northern part)
26. Adriatic sea (southern part)
27. Aegean sea (western part)
28. Black sea (northern part)
29. Sea of Azov
30. Black sea (middle part)
31. Black sea (south/eastern part)
32. Marmara sea
33. Aegean sea (eastern part)
34. Sea of Crete
35. Ballearic basin (northern part)
36. Gulf of Lion/Ligurian sea
37. Algero Provencal basin
38. Tyrrhenian basin (northern part
39. Tyrrhenian basin (southern part)
40. Balearic basin (southern part)
41. Black sea(deep water)

6 Photochemical ozone formation

6.1 Introduction
     6.1.1 Precursors and photo-oxidants
     6.1.2 Photochemical ozone formation scheme
6.2 Existing systems for characterisation of ozone formation
6.3 The Rains model
     6.3.1 Modelling ozone formation
     6.3.2 Exposure indices in RAINS
     6.3.3 Validation of the RAINS model
6.4 Spatially differentiated characterisation factors
     6.4.1 Sub categories of ozone formation
     6.4.2 Characteristics of the new characterisation factors
     6.4.3 Differentiation between individual VOCs
6.5 How to perform site-generic and site-dependent characterisation
6.6 Normalisation references
6.7 Interpretation and recommendations
6.8 References
Annex 6.1 Influence from meteorological conditions
Annex 6.2 POCP values for individual VOCs
Annex 6.3 Efficiency factors for individual VOCs and source-specified VOCs
Annex 6.4 Normalisation references for photochemical ozone formation

Authors: Michael Hauschild ^[44]Annemarie Bastrup-Birk ^[45] Ole Hertel45 Wolfgang Schöpp ^[46] José Potting44

6.1 Introduction

The concentration of ozone in the troposphere rises by about 1% a year over the Northern hemisphere. An important source of this ozone as well as of other reactive substances is the photochemical oxidation of volatile organic compounds and carbon monoxide in the presence of nitrogen oxides and sun light in the boundary layer of the troposphere. The troposphere is the lower stratum of the atmosphere, reaching from the surface of the earth to the tropopause 8-12 km above us and the boundary layer is the lower part of the troposphere (up to an altitude of around 1 km) where full mixing can be assumed and where the exposure of life to the photo oxidants takes place. Due to their reactive nature, the photo-oxidants are injurious to the health of living organisms, vegetation as well as animals and humans. Apart from the general increase in the tropospheric ozone concentration, smog-episodes can occur on a more local scale in cities given a combination of high emission levels and the right meteorological conditions. During smog-episodes, the concentrations of ozone and other photo-oxidants reach extreme levels causing immediate damage to human health.

Due to the local and discontinuous nature of photochemical smog-episodes, the focus in life cycle impact assessment is mainly on the more regional increase of tropospheric ozone.

6.1.1 Precursors and photo-oxidants

The precursors necessary for photochemical ozone formation are nitrogen oxides which have a kind of catalytic effect and volatile organic compounds which are oxidised during the process.

While NO_x, nitrogen oxides, represents the sum of nitric oxide NO and nitrogen dioxide NO₂, there is no universal definition of what a volatile organic compound, VOC is. In the current context, it is defined as an organic compound with a boiling point below 250C (WHO, 1989). For a VOC to contribute to photochemical ozone formation it must either contain hydrogen (i.e. it can not be fully substituted), or contain double bonds (i.e. be unsaturated) (Hauschild and Wenzel, 1998). Methane, which is a VOC according to the definition, is generally treated separately from the rest of the VOCs due on the one side to its very low photochemical reactivity (as reflected in its extremely long atmospheric residence time), and on the other side its large proportion of the total emission of volatile organic compounds. The abbreviation VOC is therefore generally – and also throughout this chapter - used to characterise the volatile organic compounds apart from methane (in some contexts also designated the NMVOCs).

While being an intermediate in the atmospheric oxidation of VOCs, carbon monoxide from man-made emissions also contributes to photochemical ozone formation.

The photo-oxidants comprise a variety of unstable, oxidising substances formed when VOCs react with oxygen compounds which are naturally present in the troposphere (the most important being the hydroxyl radical, OH). Among the photo-oxidants, the most prominent are ozone, O3, and peroxyacetyl nitrate, PAN, but intermediate products from the oxidation of VOCs also contribute to the injurious effects. It is thus not the VOCs per se which cause the environmental problems associated with photochemical ozone formation, but the products of their conversion. If a direct toxic effect of VOCs occurs, it is treated separately in life cycle impact assessment as a contribution to the impact categories human toxicity and ecotoxicity.

6.1.2 Photochemical ozone formation scheme

Some tropospheric ozone arises from the natural transport of ozone from the stratosphere, where it is formed by the action of energetic UV rays from the sun, but the rest is produced in the troposphere by the photochemical ozone formation. Schematically, the photochemical ozone formation can proceeds through the following four steps:

1. Reaction between VOCs or CO and OH to form peroxy radicals, ROO.

2. The peroxy radicals oxidise NO to NO₂.

3. NO₂ is split by sunlight with formation of NO and release of oxygen atoms.

4. Oxygen atoms react with molecular oxygen, O₂, to form ozone.

The reaction schemes are complex and difficult to simulate in a model. The heterogeneous spatial distribution of VOC and NO_x sources across Europe, and the many hundreds of chemical species involved, makes the photochemical formation of ozone on a regional scale highly non-linear.

Apart from the emitted quantity of VOC and NO_x, the following factors influence the ozone formation:

Meteorological conditions

Solar radiation is a driving force behind the photo oxidation but also wind speeds and directions (determining the dispersion of the air pollutants), temperature (influencing reaction kinetics) and precipitation patterns (influencing wet deposition and hence removal of pollutants from the boundary layer) are conditions that influence the photochemical ozone formation

Interaction between pollutants

The simultaneous concentration of the other relevant pollutants, NO_x and individual VOCs, can strongly influence the overall result. The dependence of ozone formation on the precursor emissions (NO_x, VOC) is therefore often far from linear.

In regions with relatively low emissions, the NO_x/VOC ratio is generally low, and ozone formation is limited by NO_x. In these regions, reduced NO_x emissions will thus reduce ozone formation while reductions in VOC emissions may have little influence on the ozone formation.

In regions with extremely high NO_x emissions, reduction in the emissions may actually lead to increased ozone formation. Here, the high concentrations of NO may lead to reduction of ozone (through oxidation of NO to NO₂). In addition, NO may reduce the concentration of hydroxyl radical which thus limits the potential for oxidation of VOCs. In this situation, reduced emissions of NO_x will therefore increase the availability of hydroxyl radical, increasing the oxidation of VOC and hence the ozone formation, unless VOC emissions are reduced at the same time. (Heyes et al., 1997a).

Biogenic emissions of VOCs

In Europe, the total natural emissions of biogenic VOCs is thought to be around 30 % of the total VOC emissions (Derwent et al., 1991) and this must be represented in a model for ozone formation. For the biogenic (natural) releases of NMVOC only little information is presently available, and often the empirical expressions derived by Lübkert and Schöpp (1989) to describe releases from forested areas are therefore applied.

Properties of the individual VOC

As mentioned earlier, the photochemical reactivity of methane is much lower than the reactivity of the VOCs. Among the VOCs there is also variation in the potential for formation of ozone as reflected in the individual photochemical ozone creation potentials (POCP) shown in Annex 6.2.

A ranking of the different groups of VOCs according to their ozone formation potential per unit weight results in the following sequence (Derwent and Jenkin, 1990):

• Alkenes (reactivity decreasing with chain length) and aromatics (reactivity increasing with the degree of alkyl substitution, decreasing with the length of the chain in the substituted alkyl group).
• Aldehydes (the strongest is formaldehyde; benzaldehyde has no or even a negative ozone formation potential).
• Ketones
• Alkanes (reactivity almost constant from C3 upwards), alcohols and esters (the more substituted the lower reactivity).
• Halocarbons (reactivity decreasing with the degree of halogen substitution).

6.2 Existing systems for characterisation of ozone formation

Figure 6.1 shows the cause-impact chain for photochemical ozone formation. The descriptors provided for each link of the chain are characteristics or parameters which influence the behaviour at that point in the chain.

Figure 6.1 Cause-impact chain for photochemical ozone formation.

The existing characterisation factors for photochemical ozone formation are all based on the photochemical ozone creation potential, POCP or possibly the Maximum Incremental Reactivity (MIR) of the substances. Both express the short-term contribution to ozone formation. MIR is primarily designed to reflect the contribution during photo-smog episodes while POCP may be calculated with time horizons from 1 to 9 days. While MIR is expressed directly as moles of ozone produced per mole of VOC emitted, POCP is expressed as an equivalent emission of the reference substance ethylene (C2H4) – a strong ozone formation gas.

In EDIP97, photochemical ozone formation is characterised using POCP-values for ozone formation over 4-9 days. Some spatial differentiation is built into the EDIP97 through a distinction according to the NO_x concentration in the emission area. There are thus two sets of characterisation factors – one for low NO_x regions and one for high NO_x regions. This is a rather rough distinction since emissions taking place in high NO_x regions may well be transported to low NO_x regions and oxidised there (and vice versa) but at that time it was considered by the authors to be the best representation of the spatial variability of photochemical ozone formation that would be feasible in life cycle impact assessment.

In addition to POCP values for around 90 individual VOCs, EDIP97 also provides aggregated POCP values for some common VOC-mixtures identified by their source rather than their composition. Furthermore, tools are provided for estimation of missing POCP values.

Both the POCP and the MIR address impacts at the level called fate-dispersion and distribution in Figure 6.1. They reflect the substance's intrinsic potential for ozone formation under a set of standard conditions regarding the presence of other gases and the meteorological conditions and they focus primarily on the ozone formation during the first days after the emission as relevant for the prediction of photo smog episodes.

The behaviour through the first links of the cause-impact chain from emission to the point where the POCP and MIR are modelled, depends to a large extent on the properties of the substance. Further along the chain, the conditions of the environment where the substance is emitted, dispersed, oxidised and where exposure of sensitive target organisms takes place become decisive. Hitherto, it has been seen as impossible to include these descriptors in the modelling of impact, even though they are known to be important – for photochemical ozone formation perhaps more important than the descriptors in the early parts of the cause-impact chain as illustrated by the spatially differentiated factors developed in Section 6.4. It has not been feasible to perform the required spatial differentiation in life cycle assessment, and therefore a low environmental relevance of the modelled impacts has been accepted.

For photochemical ozone formation, an additional problem about the existing characterisation models is their inability to represent the impact from NO_x. Both POCP and MIR have been developed for ranking of VOCs according to their ozone formation potential and the presence of NO_x is considered a prerequisite. Due to the lack of factors for NO_x, its contribution to ozone formation is neglected in current life cycle impact assessment and this is a serious flaw as seen from the normalisation references calculated in Section 6.6 showing that the contribution from NO_x could be around 2/3 of the total ozone formation in Europe.

It has been the purpose of the work performed under the Danish LCA methodology and consensus-creation project to include as much as possible of the later links of the cause-impact chain including this spatially determined variation. The aim has been to include exposure assessment in the characterisation without losing the feasibility in LCIA.

In the endeavour to meet these objectives, an integrated impact assessment model was found which supports quantification of the ozone formation potential of both VOC and NO_x in a site-dependent manner allowing the development of characterisation factors which include the later links in the cause-impact chain - both exposure assessment and parts of the target system description. This model is described in Section 6.3

6.3 The Rains model

The RAINS (Regional Air Pollution Information and Simulation) model is an integrated assessment model that combines information on regional emission levels with information on long-range atmospheric transport in order to estimate patterns of deposition and concentration for comparison with critical loads and thresholds for acidification, eutrophication and tropospheric ozone formation. One of the main objectives with the RAINS model is to establish the relationship between the site of an emission and the impact on its deposition areas. The RAINS model (version 6.2) has therefore been seen as an obvious choice to establish spatially differentiated characterisation factors for photochemical ozone formation in parallel to what has been done for acidification (Chapter 3) and terrestrial eutrophication (Chapter 4). Alcamo et al. (1990) and Amann et al. (1995) provide a detailed description of the general model and Section 3.3 gives a summary of those parts of RAINS which are relevant for its use in our context.

6.3.1 Modelling ozone formation

RAINS is intended to support the development of cost-effective European abatement strategies for different types of air pollution. For photochemical ozone, the focus is on the long-term ozone levels at the more regional scale. For this purpose, it is crucial with an integrated assessment that represents the relationship between source and receptor regions. For the modelling of ozone formation, RAINS does not apply a mechanistic model of the highly complex reaction schemes behind the formation of ozone and other photo-oxidants. Such models are used for calculation of POCPs but they will not be operational in an integrated assessment model. Instead, RAINS builds on a computationally efficient `reduced-form' model of ozone formation using statistical methods to summarise the response to emission changes given by a more complex reference model, (Heyes et al., 1997b).

In the `reduced-form' model as presented in Heyes et al. (1997a), the long-term mean ozone concentration in any one grid cell j is expressed as a function of the annual non methane VOC emission vi and the annual NO_x emission ni from emitter country i and the effective NO_x emission enj experienced in grid cell j over the period considered (enj is corrected for exchange with the free troposphere):

(6.1)

Click here to see the Figure.

where

M is the number of emitter countries considered

k_j includes the effects of background concentrations of O3 and its precursors and natural VOC emissions

a_ijv_i provides the linear country-to-grid contribution from VOC emissions in country i (allowing for meteorological effects)

b_ijn_i provides the linear country-to-grid contribution from NO_x emissions in country i (allowing for meteorological effects)

c_ijn_i² serves as a correction term to allow for non-linearities occurring close to countries with very high NO_x emissions

α_jen_j² represents the average non-linearity in the O3/NO_x relationship experienced along the trajectories arriving at grid cell j and any non-linear effects local to grid cell j

d_ijen_jv_i allows for interactions between NO_x and VOCs along the trajectories

The coefficients a_ij, b_ij, c_ij and α_j are estimated by linear regression using ni, vi and enj as variables in the more complex reference model.

Emission estimates

The resulting factors depend on the emission patterns of European countries. They may therefore vary in time, and in order to reveal temporal variation, the factors are calculated for 1990, 1995 and 2010. The underlying per country emissions of non methane VOCs and NO_x are based on the inventories of the UN ECE for 1990 and 1995 while the 2010 emission scenario is estimated based on the current reduction plans as expressed by the individual European countries.

Meteorological conditions

To reduce the influence of annual variations in meteorological conditions, the characterisation factors for each of the emission years 1990, 1995 and 2010 are derived as the average of five different calculations using the meteorological data for the years 1989, 1990, 1992, 1993 and 1994 respectively. Annex 6.1 illustrates the variation between the meteorological years for the calculation of AOT60 for the emission year 1995

6.3.2 Exposure indices in RAINS

The main target organisms of photo-oxidants are human beings and plants. In RAINS, exposure indices have been developed to quantify the exposure of vegetation as well as humans.

Vegetation exposure

Research on ozone-related vegetation damage makes it possible to define biologically meaningful, but simple, indices to characterise critical ozone exposure and to identify the critical levels of exposure above which by definition, adverse direct effects on receptors, such as certain plant species, may occur. Critical levels to protect vegetation are best established using long term exposure measures. Based on the scientific work on critical levels carried out under the UN/ECE Convention on Long-range Transboundary Air Pollution Working Group on Effects, a number of guideline values have been recommended by WHO (1997). For vegetation, the recommended threshold is 40 ppb and the cumulative exposure index AOT40 – exposure above the threshold of 40 ppb has been accepted as the best available exposure index for damage to crops and natural vegetation (Kärenlampi and Skarby, 1996).

The concept of accumulated excess ozone assumes a threshold below which no or small effects occur. The AOT40 uses hourly concentrations during daylight hours over a three-month period (growing season) for herbal plants and crops and a six-month growing period for trees. The AOT40 is calculated as the sum of the differences between the hourly ozone concentrations in ppb and 40 ppb for each hour when the concentration exceeds 40 ppb, using daylight hours only. It is measured as ppbhours or more commonly as ppmhours.

Under the UN/ECE Convention on Long-range Transboundary Air Pollution, the critical level for exposure of herbal plants and agricultural crops (relating to a five percent crop loss) has been set at an AOT40 of 3 ppmhours for the growing season and daylight hours, averaged over a 5-year period. For forest trees, a critical level has been set at 10 ppmhours for daylight hours, accumulated over a six month growing season and also averaged over five years (Heyes et al., 1997b). For the currently prevailing European ozone levels, the critical level for crops and natural vegetation is thus stricter than the critical level for forest trees. Therefore the critical level for vegetation exposure applied in RAINS, and hence also in the calculated characterisation factors for vegetation exposure, is the level for crops and natural vegetation.

To reduce the influence from fluctuations in meteorological conditions, the ozone exposure in each grid used for calculation of the indices is determined as an arithmetic average of exposures calculated using the meteorology for five separate years in the period 1989-1994.

The AOT40 is expressed in two different vegetation-related exposure indices:

a cumulative vegetation exposure index calculated as the excess AOT40 (i.e. the AOT40 in excess of the critical level of 3 ppmhours) multiplied by the area of ecosystems exposed to the excess concentration

an average vegetation exposure index reflecting the average excess AOT40 in a country.

Both indices are based on rural ozone concentrations (Heyes et al., 1997b)

Human exposure

Ozone reacts with macro-molecules in the tissue of the human air ways. Depending on respiration frequency and ambient ozone concentrations, ozone will react in different parts of the air ways. At high respiration frequency and low ambient concentrations, ozone will mainly react in the upper air ways, whereas high speed of respired air and high ozone concentrations lead to transport of ozone deep into the lungs. A detailed Danish review and evaluation of health effects related to traffic air pollution is found in Larsen et al. (1997).

For exposure of human beings, no thresholds for chronic exposure have been established. Following the revised WHO Air Quality Guidelines for Europe (WHO, 1997), a maximum eight-hour average concentration of 60 ppb is proposed as the long-term environmental objective for the EU ozone strategy. The maximum is calculated from running eight-hour averages of the one-hour mean concentrations. The ultimate goal would be to eliminate all excess of this criterion. The modelling of European abatement strategies for individual days over a multi-months period is a rather ambitious task and is not entirely feasible at the moment. In order to simplify the modelling task, the target of no-exceedance of the WHO criterion (60 ppb as maximum eight hours mean concentrations) has been converted into an AOT index for human health in parallel to the AOT40 index for vegetation damages for use in RAINS. The AOT60 is calculated in the same way but applying the threshold of 60 ppb. As a result, an AOT60 (i.e., the cumulative excess exposure over 60 ppb, for practical reasons over a six-month period) of zero is considered as equivalent to the full achievement of the WHO criterion. Any violation of this WHO guideline will consequently result in an AOT60 of larger than zero.

It is emphasised by the group behind the RAINS model that the interpretation of the AOT60 index should be cautious. Given the current knowledge on health effects it is not possible to link any AOT60 value larger than zero with a certain risk to human health (Heyes et al., 1997b).

Like the AOT40, the AOT60 is expressed in two different exposure indices:

• a cumulative index which reflects the total exposure of a population and is expressed in personppmhours. It is calculated as the product of the average exposure per person and the total population within the exposed area. The RAINS model calculates these indices on a grid basis (using gridded data on AOT60 and population). Next, these grid values are aggregated to the country level. The cumulative index in RAINS uses the AOT60 concentrations per grid, representing the rural ozone concentrations, and the total population per grid in 1990.
• an average indicator reflecting the average exposure of a person in a country calculated from the gridded data

Inaccuracies may occur for grids with major urban areas, where the rural ozone concentrations used for these analyses are lower than the concentrations occurring in the city plumes. The `average' indicator reflects the average exposure of a person in a country, calculated from gridded data.

It is important to stress that these indices may not be used to derive estimates of health damage, for which more detailed information is deemed necessary. In the context of this report, these indices provide relative measures to enable a comparison of different scenarios.

As for AOT40, ozone exposure in each grid cell used for calculation of the indices is an arithmetic average of exposures calculated using the meteorology for five separate years during the period 1989-1994.

The hourly mean tropospheric concentrations of ozone in Europe vary between 0 and 150 ppb, but generally lie in the range 10 to 40 ppb. The background concentrations in Denmark are around 35 ppb (Granby et al., 1998). In the northern part of Europe the concentrations are highest in the summer months (May-August). During this period the ozone concentrations thus frequently exceed thresholds for damages on vegetation and for human health in large parts of Europe.

6.3.3 Validation of the RAINS model

The performance of the reduced-form model for ozone formation has been validated in terms of its calculation of AOT40 against the results of the full EMEP model in Heyes et al., 1996. Over a realistic range of emissions, it was found to give a very good performance with results on mean ozone concentrations typically deviating less than 1.3% from the full model. A bias appears when the AOT40 approaches zero. In our use, only the excess of 3 ppmhours are counted, so this seems to be a minor problem.

The full model takes into account differences in the composition of the VOC emissions for different sectors. Such differences are not included in the reduced-form model and this is a source of deviation from the results of the full EMEP model

Overall the performance of the reduced-form model seems satisfactory for our purpose. The reader is referred to Heyes et al., 1996 for further details on the validation.

6.4 Spatially differentiated characterisation factors

The use of RAINS for calculation of spatially differentiated characterisation factors is described in detail in Chapter 3 for acidification. The procedure is similar for photochemical ozone formation with one main difference. For acidification, the factor for a country is determined by the change in impact caused by a marginal increment of that country's emissions. For photochemical ozone formation, the national factors have been determined through an analytical solution of the differential quotient of the functional expression for AOT40 or AOT60 respectively.

The functional form of the AOT40-model is the same as the form of the mean ozone concentration model in equation (6.1):

(6.2)

The variables and parameters are explained under equation (6.1).

Site-dependent characterisation factors

The resulting site-dependent characterisation factors are shown in Table 6.1 for exposure of vegetation and in Table 6.2 for exposure of humans.

Table 6.1. Site-dependent characterisation factors for exposure of vegetation to photochemical ozone expressed as cumulative AOT40 in exceedance of the critical level (m²ppmhours/g). Factors calculated for the emission patterns of 1990 and 1995 and the expected pattern of 2010.

Table 6.2. Site-dependent characterisation factors for exposure of humans to photochemical ozone expressed as cumulative AOT60 (persppmhours/g). Factors calculated for the emission patterns of 1990 and 1995 and the expected pattern of 2010.

Site-generic characterisation factors

From the site-dependent characterisation factors of Table 6.1 and 6.2 site-generic factors are calculated as the emission-weighted average over all the countries and regions. The site-generic factors are compatible with the site-dependent factors in the sense that they are calculated using the same model and therefore cover the same part of the causality chain underlying photochemical ozone formation impacts. The only difference is that the spatial variation in the included parameters is no longer represented. The site-generic factors can be used if spatial differentiation is unwanted or if the location of some of the processes in the product system is unknown or lies outside Europe. The standard deviation of the site-generic factors reflects the spatially determined variation underlying the site-generic factors.

Table 6.3. Site-generic characterisation factors for exposure of vegetation to photochemical ozone expressed as cumulative AOT40 in exceedance of the critical level (m²ppmhours/g). The factors are calculated as the emission-weighted means of site-dependent factors in Table 6.1 for the emission patterns of 1990 and 1995 and the expected pattern of 2010.

Table 6.4. Site-generic characterisation factors for exposure of humans to photochemical ozone expressed as cumulative AOT60 (persppmhours/g). The factors are calculated as the emission-weighted means of site-dependent factors in Table 6.2 for the emission patterns of 1990 and 1995 and the expected pattern of 2010.

6.4.1 Sub categories of ozone formation

Measures of the AOT40 and AOT60 are by no means proportional, and a closer study of the factors for vegetation exposure and for human exposure in Tables 6.1 and 6.2 reveals marked differences in the ranking and relative impact of different emission countries. It is therefore concluded that the exposure of humans and vegetation must be described by each their index and the impact category is split into two sub categories representing respectively:

- critical ozone exposure of vegetation

- critical ozone exposure of humans

Being defined earlier in the cause-impact chain, the EDIP97 impact category photochemical ozone formation covers all impacts from photo-oxidants. In addition to human health and vegetation impacts, also damage to man-made materials is thus represented by the POCP-based impact potential. None of the two sub categories introduced here explicitly covers material damage. Indeed, it is mentioned in Heyes et al., 1997b, that the UN/ECE has proposed a preliminary threshold of an annual mean ozone concentration of 20 ppb based on an acceptable deterioration rate for sensitive organic materials. Separate calculations for material damage require additional information of the distribution of sensitive materials over the grids of the RAINS model. This far, no model calculations have been performed for material damage but until that may happen we suggest to consider damage to materials represented under the human health exposure. Even though the suggested threshold for materials is different, the geographic distribution of sensitive man-made materials is likely to be well represented by the geographic population distribution as already included in the RAINS model.

6.4.2 Characteristics of the new characterisation factors

The site-dependent characterisation factors presented in Tables 6.1 and 6.2 and the site-generic factors in Tables 6.3 and 6.4 are calculated using a model that covers a larger part of the cause-impact chain in Figure 6.1. In contrast to the old factors that nearly exclusively represent substance-specific characteristics, the new factors also include the dispersion, the concentration increase and the background concentration in the receptor area, the population density or density of vegetation in the receptor area as well as some quantitative concentration-effect relationship for vegetation. The new factors thus eliminate a large part of the spatially determined uncertainty that is included in the EDIP97 factors.

Reflecting the actual emission patterns of European countries, the new factors are dependent on changes in these patterns. Therefore, the factors have been calculated for the emission patterns of 1990, 1995 and the expected pattern of 2010. For vegetation exposure, the temporal variability is negligible for most emitter countries but for human exposure, there are significant differences depending on the basis year. In the guideline, the factors based on 1995 emissions is suggested as the default but for emissions taking place in the future, e.g. from the late use stage or disposal stage of the product, the factors based on 2010 could be applied alternatively as part of the sensitivity analysis.

The new factors also include the contribution from NO_x. From Tables 6.3 and 6.4 it is clear that on a weight basis, the factor for NO_x is on average considerably larger than the factor for VOC, so this is also an important improvement compared to the EDIP97 factors.

On the other hand, the new characterisation factors do not allow the differentiation between individual VOCs. This aspect is discussed in the next section.

6.4.3 Differentiation between individual VOCs

As mentioned in Section 6.3.1 the ozone formation model applied in RAINS does not allow distinction between individual VOCs. This eliminates a source of variation in the factors but since the variation between the ozone formation potentials lies within a factor 2-4 for most VOCs, this potential error is small compared to the variation eliminated by including the contribution from NO_x. Nonetheless, it is suggested to introduce differentiation between the individual VOCs through an efficiency factor s representing the ozone formation potential of a specific VOC (s) relative to the average ozone formation potential of European VOC emissions

(6.3)

The European average POCP was determined for 1985 emissions at 0,40 in Hauschild and Wenzel (1998). Based on this figure and the POCP values applied for characterisation factors in EDIP97, the efficiency factor is calculated and tabulated in Annex 6.3 for the most common individual VOCs as well as for source-specified VOC-mixtures frequently encountered in life cycle inventories.

The RAINS model only considers non-methane VOCs and NO_x. In comparison, the POCP values used in EDIP97 and other life cycle impact assessment methodologies are also available for methane and carbon monoxide. To include these two substances which may be of importance for some processes, the following assumptions are made:

Including carbon monoxide

In addition to being a substance emitted from processes with incomplete combustion, carbon monoxide is also an intermediary product in the photochemical oxidation of some VOCs and its atmospheric lifetime is thus within the interval relevant for non-methane VOCs. It is therefore assumed that the dispersion and deposition pattern of carbon monoxide is well represented by the model applied for VOCs and in the calculation of photochemical ozone impact, carbon monoxide is treated as a VOC.

Including methane

In contrast, methane is an extremely long-lived organic compound with an estimated residence time in the atmosphere of around 10 years. This means that a globally uniform distribution of methane in the atmosphere must be expected and the contribution of methane to ozone formation is rather low at a regional level. In the calculation of characterisation factors it is suggested to base the characterisation factors for methane on the site-generic factors developed for VOCs and correct for the fact that due to the long lifetime of methane, a large part of the ozone formed will expose ocean areas and hence not contribute to exposure of vegetation or humans. A correction factor of 0,5 is proposed.

Table 6.5. Site-generic characterisation factors for methane proposed to represent the exposure of vegetation or humans to photochemical ozone. The factors are based on the VOC factors in Table 6.3 and 6.4

6.5 How to perform site-generic and site-dependent characterisation

The site-generic photochemical ozone impacts from a product are calculated applying the new site-generic characterisation factors of Table 6.3, 6.4 or 6.5 in the following formulas:

Click here to see the Figure.

Where:

sg POI(veg) is the site-generic photochemical ozone formation impact on vegetation expressed as area exposed above threshold (in m²ppmhours/f.u.)

sg POI(hum) is the site-generic photochemical ozone formation impact on human health expressed as persons exposed above threshold (in persppmhours/f.u.)

sg POF(veg)_VOC is the site-generic photochemical ozone formation factor from Table 6.3 that relates emission of VOCs or CO to the impact on vegetation in the deposition area (in m²ppmhours/g).

sg POF(veg)_NOx is the site-generic photochemical ozone formation factor for from Table 6.3 that relates emission of NO_x to the impacts on vegetation in the deposition area (in m²ppmhours/g).

sg POF(hum)_VOC is the site-generic photochemical ozone formation factor from Table 6.4 that relates emission of VOCs or CO to the impacts on human health in the deposition area (in persppmhours/g).

sg POF(hum)_NOx is the site-generic photochemical ozone formation factor from Table 6.4 that relates emission of NO_x to the impacts on human health in the deposition area (in persppmhours/g).

sg POF(veg)_CH4 is the site-generic photochemical ozone formation factor for from Table 6.5 that relates emission of CH4 to the impacts on vegetation in the deposition area (in m²ppmhours/g).

sg POF(hum)_CH4 is the site-generic photochemical ozone formation factor from Table 6.5 that relates emission of CH4 to the impacts on human health in the deposition area (in persppmhours/g).

is a substance-specific efficiency factor from Annex 6.3 expressing the ozone creation potential of the individual volatile organic compound or CO (s) or methane relative to the ozone creation potential of the European average VOC (dimensionless).

E is the emission of NO_x, CH4 or individual or source-specified VOC or CO (s) according to index (in g/f.u.)

The potential spatially determined variation of the site-generic photochemical ozone impacts, can be estimated from the standard deviation given in Table 6.3, 6.4 or 6.5 for each substance.

The site-dependent photochemical ozone impacts from a process in the life cycle of a product are calculated applying the new site-dependent characterisation factors of Table 6.1 and 6.2 in the following formulas:

Click here to see the Figure.

Where:

sd POI(veg)_p is the site-dependent photochemical ozone formation impact on vegetation expressed as area exposed above threshold by the selected process (p) (in m²ppmhours/f.u.)

sd POI(hum)_p is the site-dependent photochemical ozone formation impact on human health expressed as persons exposed above threshold by the selected process (p) (in persppmhours/f.u.)

sd POF(veg)_NOx,i is the site-dependent photochemical ozone formation factor from Table 6.1 that relates emission of NO_x from country or region (i), where the selected process (p) is located, to the impacts on vegetation in the deposition area (in m²ppmhours/g).

sd POF(veg)_VOC,i is the site-dependent photochemical ozone formation factor from Table 6.1 that relates emission of VOCs or CO from country or region (i), where the selected process (p) is located, to the impact on vegetation in the deposition area (in m²ppmhours/g).

sg POF(veg)_CH4 is the site-generic photochemical ozone formation factor CH4 from Table 6.4 that relates emission of CH4 to the impacts on vegetation in the deposition area (in m²ppmhours/g).

sd POF(hum)_NOx,i is the site-dependent photochemical ozone formation factor from Table 6.2 that relates emission of NO_x from country or region (i), where the selected process (p) is located, to the impacts on human health in the deposition area (in persppmhours/g).

sd POF(hum)_VOC,p is the site-dependent photochemical ozone formation factor from Table 6.2 that relates emission of VOCs or CO from country or region (i), where the selected process (p) is located, to the impacts on human health in the deposition area (in persppmhours/g).

sg POF(hum)_CH4 is the site-generic photochemical ozone formation factor from Table 6.5 that relates emission of CH4 to the impacts on human health in the deposition area (in persppmhours/g).

η is a substance-specific efficiency factor from Annex 6.3 expressing the ozone creation potential of the individual volatile organic compound or CO (s) or methane relative to the ozone creation potential of the European average VOC (dimensionless).

E_p is the emission of NO_x, CH4 or individual or source-specified VOC or CO (s), according to index, from process (p) (in g/f.u.)

Emissions from a non-European or unknown region may as a first approach be calculated using the site-generic factors from Table 6.3, 6.4 or 6.7. The standard deviations on the site-generic factors in these tables give a range of potential spatial variation for the application of the site-generic factor within Europe. Given the size of the variation in emissions and sensitivities within Europe, the site-dependent factor is expected to lie within this range for most regions also in the rest of the world. Expert judgement may be used in the interpretation to assess whether the factor for emissions from processes in non-European regions should be found in the lower or upper end of the range.

6.6 Normalisation references

The normalisation reference for photochemical ozone formation is calculated using the new characterisation factors on the national European emission inventories for the corresponding year. Three sets of references are calculated in Annex 6.4 and the reference based on 1995 factors and emissions is retained as the EDIP default reference for normalisation in life cycle impact assessment.

Table 6.6. Normalisation references for the two sub categories of photochemical ozone formation calculated for the reference yeras 1990, 1995 and 2010 in Annex 6.4 and expressed as impact per person per year (the person equivalent).

From Table 6.6 it is seen that the photochemical ozone impact was reduced from 1990 to 1995 and is expected to continue the trend towards 2010.

6.7 Interpretation and recommendations

The new photochemical ozone formation impact potentials are improved in two aspects compared to the impact potentials calculated using the EDIP97 characterisation factors or other systems based on the POCP or MIR approach; the environmental relevance is increased and a part of the spatial variation in sensitivity of the receiving environment is now taken into account.

Environmental relevance

The environmental relevance is increased because the exposure of the sensitive parts of the environment (vegetation or human beings) is included in the underlying model which now covers most of the causality chain towards the LCA protection areas: Ecosystem health and human health. This is particularly important when weighting factors based on the environmental relevance are used instead of the EDIP default weighting factors (which are based on political reduction targets that for photochemical ozone formation are also aiming for protection of vegetation and human health). In comparison, the EDIP97 factors only cover the potential for formation of ozone.

In addition, the contribution of NO_x is now included in the impact potentials. The significance of this for a specific product system depends on the quantities of NO_x and VOCs emitted. From the calculation of the normalisation references, it is known that on a European level, NO_x contributes around twice as much as VOC to photochemical ozone formation and on average the characterisation factor for NO_x is more than three times the characterisation factor of VOCs.

Spatial variation

The spatial variation in exposure for photochemical ozone formation can be large even at the very local scale. The variation in sensitivity between European regions are now presented on a national scale showing a factor 15-20 of difference between least and the most sensitive emission countries for exposure of vegetation and a factor of around 400 times of difference for exposure of humans (the latter reflecting the variation in population density in the deposition areas). This variation is hidden when the EDIP97 factors or similar are used for characterisation.

6.8 References

Alcamo, J., R. Shaw and L. Hordijk (eds.). The RAINS model of acidification. Science and strategies in Europe. Dordrecht (the Netherlands), Kluwer Academic Publishers, 1990.

Amann, M., M. Baldi, C. Heyes, Z. Klimont and W. Schöpp. Integrated assessment of emission control scenarios, including the impact of troposheric ozone. Water, air and soil pollution, 85(4), 2595-2600, 1995.

Andersson-Sköld, Y., Grennfelt, P., and Pleijel, K.: Photochemical ozone creation potentials: a study of different concepts. J. Air Waste Manage. Assoc. 42(9), 1152-1158, 1992.

Derwent, R.G., and Jenkin, M.E.: Hydrocarbon involvement in photochemical ozone formation in Europe. AERE R 13736, AEA Environment and Energy, Harwell Laboratory, Oxfordshire, UK, 1990.

Derwent, R.G., Grennfelt, G., and Hov, Ø.: Photochemical oxidants in the atmosphere. Nordic Council of Ministers, Nord 1991:7, Copenhagen, 1991.

Derwent, R.G., Jenkin, M.E., Saunders, S.M., et al.: Photochemical ozone creation potentials for organic compounds in northwest Europe calculated with a master chemical mechanism. ATMOS ENVIRON 32 (14-15): 2429-2441, 1998

Granby, K., Kemp, K., Palmgren, F., Hovmand, M. and Egeløv, A.: Measurements of Photochemical Pollutants. In: Fenger, J. (ed.): Photochemical Air Pollution. Danish Aspects. National Environmental Research Institute. - NERI Technical Report 199: 63-101, 1997.

Hauschild, M. and Wenzel, H.: Photochemical ozone formation as criterion the environmental assessment of products. In Hauschild, M.Z. and Wenzel, H.: Environmental assessment of products. Vol. 2 - Scientific background, 565 pp. Chapman & Hall, United Kingdom, ISBN 0412 80810 2, Kluwer Academic Publishers, Hingham, MA. USA., 1998.

Hettelingh J.-P., Posch M., de Smet P., Downing R.J.: The use of critical loads for

emission reduction agreements in Europe. Water, Air and Soil Pollution 85:2381-2388, 1995.

Heyes, C., Schöpp, W., Amann: A simplified model to predict long-term ozone concentrations in Europe. WP-96-12, International Institute for Applied Systems Analysis (IIASA), Laxenburg, Austria, 1996.

Heyes, C., Schöpp, W., Amann, M., Bertok, I., Cofala, J., Gyarfas, F., Klimont, Z., Makowski, M. and Shibayev, S.: A model for optimizing strategies for controlling ground-level ozone in Europe. Interim report IR-97-002/January, International Institute for Applied Systems Analysis (IIASA), Laxenburg, Austria, 1997a.

Heyes, C., Schöpp, W., Amann, M., Bertok, I., Cofala, J., Gyarfas, F., Klimont, Z., Makowski, M. and Shibayev, S.: Simultaneous optimization of abatement strategies for ground-level ozone and acidification. Interim report IR-97-090/December, International Institute for Applied Systems Analysis (IIASA), Laxenburg, Austria, 1997b.

Kärenlampi L. and Skarby L., (eds), 1996. Critical levels for Ozone in Europe: Testing and

Finalizing the Concepts. UN-ECE Workshop Report. Dept. of Ecology and Environmental Science, University of Kuopio, Finland, 1996.

Larsen, P.B., Larsen, J.C., Fenger, J., and Jensen, S.S.: Health evaluation of air pollution from road traffic (In Danish: Sundhedsmæssig vurdering af luftforurening fra vejtrafik). Miljøprojekt nr. 352, Danish Environmental Protection Agency, Copenhagen, 1997.

Lübkert, B. and Schöpp, W.: A model to calculate natural (volatile organic compounds) VOC emissions from forests in Europe, wp-89-82, 65 pp, w89082, IIASA, Laxenburg, Austria, 1989

Posch, M., Hettelingh, J.P., Downing, R.J. and P.A.M. de Smet (eds): Calculation and mapping of critical thresholds in Europe. CCE status report 1997. Co-ordination Centre for Effects (CCE) at the National Institute of Public Health and the Environment (RIVM), Bilthoven (the Netherlands), 1997.

Wenzel, H., Hauschild M.Z. and Alting, L.: Environmental assessment of products. Vol. 1 - Methodology, tools, techniques and case studies, 544 pp. Chapman & Hall, United Kingdom, ISBN 0 412 80800 5, Kluwer Academic Publishers, Hingham, MA. USA., 1997.

World Health Organization: Indoor air quality: organic pollutants. EURO Reports and Studies 111, WHO Regional Office for Europe, Copenhagen, 1989

WHO: Ozone. In: Air Quality Guidelines for Europe, pp 315-326. WHO Regional Publications, European Series No 23, Copenhagen, 1987

Annex 6.1 Influence from meteorological conditions

Calculation of AOT60 for individual European countries based on 1995 emissions and meteorological data for five different years.

Click here to see the Table.

Annex 6.2 POCP values for individual VOCs

POCP values from different sources (Derwent and Jenkin, 1990, Derwent et al., 1998, Andersson-Skjöld et al., 1992). EDIP97 applies the low NO_x values from Andersson-Skjöld et al., 1992 for emissions from low NO_x areas and the values from Derwent and Jenkin, 1990 for high NO_x areas. In EDIP97, all values are scaled down by a factor 100 so the reference substance ethylene has a POCP of 1,00.

Annex 6.3 Efficiency factors for individual VOCs and source-specified VOCs

The dimensionless efficiency factor is representing the efficiency of individual VOCs relative to the European average VOC in contributing to ozone formation. It is derived as the quotient between the respective POCP-factors for 4-9 days in high NO_x-areas (the EDIP97 characterisation factors for high NO_x-areas, Wenzel et al., 1997).

Annex 6.4 Normalisation references for photochemical ozone formation

Normalisation references for photochemical ozone formation calculated with the new characterisation factors and the national European emission inventories for each of the years 1990, 1995 and 2010.The reference based on 1995 factors and emissions is retained as default factor for normalisation in life cycle impact assessment.

Impacts on vegetation

Click here to see the Table.

7 Human toxicity

7.1 Introduction
7.2 Exposure assessment
     7.2.1 Introduction
     7.2.2 The need for spatial differentiation
     7.2.3 Human exposure from air emissions
     7.2.4 Identification of source types, and classification of processes
     7.2.5 Accumulated human exposure increase local to the source (from 0 to
     10km
     7.2.6 Accumulated human exposure increase regional to the source (from
     10km to several hundreds to thousand kilometres)
     7.2.7 Total increase of accumulated exposure from air emissions
7.3 Effect assessment (evalution of threshold exceedance)
     7.3.1 Introduction
     7.3.2 Need for evaluation of threshold exceedance
     7.3.3 Possibilities for including evaluation of threshold exceedance
     7.3.4 Chosen type of threshold values
     7.3.5 Chosen emission/exposure situations and substances
7.4 Conclusions
     7.4.1 Acknowledgement
7.5 References
Annex 7.1 Characterisation factors for human toxicity assessment from emissions to air
Annex 7.2 Characterisation factors for human toxicity assessment from emissions to water
Annex 7.3 Characterisation factors for human toxicity assessment from emissions to soil
Annex 7.4 Specifications for the Wind rose Model Interpreter
Annex 7.5 Addresses of reports about regional emissions and exposures
Annex 7.6 Exposure situations ”near or above threshold”

Authors: José Potting (editor) ^[47] Alfred Trukenmüller ^[48]Frans Møller Christensen ^[49]
Hans van Jaarsveld ^[50] Stig Irving Olsen ^[51] Michael Hauschild51

7.1 Introduction

Several sets of human toxicity factors are presently in use in life cycle assessment, most notably those of Guinée et al. (1996), as updated by Huijbregts (1999), the factors of Hauschild and Wenzel (1998) and of Hertwich et al. (2001). All these sets of human toxicity factors follow a framework similar to the one as described in the technical guidance documents on risk assessment released by the European Commission (EC 1994, 1996). These documents are written in close parallel with the enactment of European legislation on management of risks from pesticides, and from society's use of chemicals.

Figure 7.1. Steps in the risk management process (modified from Van Leeuwen and Hermens 1995)

Figure 7.1 gives a general scheme for the steps in the risk management as described in the technical guidance documents of the European Commission (EC 1994, 1996). The human toxicity factors of Guinée et al. (1996), Hauschild et al. (1998), Huijbregts (1999) and Hertwich et al (2001) take their basis in those steps encircled in Figure 7.1. First, the increase of environmental exposure or concentration is predicted for one unit of emission of a given substance (exposure assessment), and in parallel the no-effect-concentration or safe dose is predicted (effect assessment). Next, the predicted environmental concentration or exposure (PEC) is divided by the predicted no-effect-level (PNEC) or safe dose. In this way, characterisation factors are obtained following a framework (PEC/PNEC) similar to the one for risk characterisation (PEC/PNEC).

The exposure assessment underlying the characterisation factors for use in life cycle assessment calculates similar exposure increases for all releases of the same quantity and substance. Disregarded are the circumstances under which these emissions take place. This contradicts what we intuitively would expect. An example for atmospheric emissions may illustrate this: Exposure increases from an emission released at moderate height will close to the source be considerably higher than those from an elevated release, but lower than those from an emission released at ground level or indoors. The traditional characterisation factors also do not take into account spatial differences in for example atmospheric conditions and population densities between areas.

Section 7.2 evaluates the need for spatial differentiation in characterisation factors for human toxicity, and continues by exploring the feasibility of a framework ^[52] based upon Potting et al. (1999) to establish site-dependent factors that assesses the increases of human exposure from air emissions. The framework takes into account variation in release height, atmospheric conditions, and population densities in the receiving areas. The framework is used to calculate site-dependent factors that quantify the increase of accumulated human exposure from an emission at a given location. Those factors can be used as an exposure factor in combination with the existing characterisation factors for human toxicity from Wenzel et al. (1997). The Guidance Document of Hauschild and Potting (2003) to this Technical Report describes in its entirety how to apply the developed methodology.

In contrast to a risk characterisation as described in the technical guidance documents of the European Commission (EC 1994, 1996), the characterisation factors used in life cycle assessment do usually not account for background exposures. They are based on exposure increases rather than actual exposures (sum of the background exposure and the exposure increase). As a consequence, exceedance of the no-effect-levels is not taken into account in these characterisation factors. The predicted exposure increases (PEC) are only weighted with the relevant no-effect-concentration (NEC) for the given substances (PEC/NEC). This weighting is needed to aggregate exposures to different substances.

A lively debate is going on whether or not life cycle assessment should perform an evaluation of threshold exceedance. This issue is closer examined in Section 7.3 in which information is also provided for a selection of substances that may facilitate a qualitative threshold evaluation.

7.2 Exposure assessment ^[53]

7.2.1 Introduction

Human toxicity is a complex impact category. This complexity arises, amongst others, from the numerous possible routes through which an emission can lead to human exposure. Figure 7.2 gives a simplified overview. There are three main routes of human exposure to environmental pollutants:

Inhalation of air (part of the exposures via the environment in Figure 7.2),

Ingestion of food, water and sometimes even soil (all part of the exposures via the environment in Figure 7.2), and

Penetration of the skin after contact with polluted surfaces, air, and sometimes also soil or water (part of consumer exposures in Figure 7.2).

Figure 7.2. Overview of the main routes of human exposure to toxic substances (Van Leeuwen and Hermens 1996).

The exposure of humans to environmental pollutants usually takes place via more than one route at the same time (multi-route exposure), but one exposure route is often dominating over the other exposure routes. A typical life cycle assessment focuses on inhalation and ingestion, though skin exposure may be of major relevance within the product system some products (like cloths or cosmetics).

Hauschild and Wenzel (1998) provide characterisation factors for over 100 substances to characterise the toxic impact from emissions to air, water and soil. An emission can have a direct effect through exposure to the medium to which it is initially released. However, emissions have often also indirect effects through re-distribution of the substance to another medium than the one to which the substance was initially emitted. The factors of Hauschild and Wenzel (1998) distinguish therefore further between inhalation of air, and ingestion of water and soil. Table 7.1 gives an overview of the resulting nine different characterisation factors, while Annex 7.1, 7.2 and 7.3 list the factors per substance. Hauschild and Wenzel (1998) can be consulted for an extensive description of the backgrounds for these human toxicity factors.

Table 7.1: Overview of the characterisation factors provided by Hauschild and Wenzel (1998).

The factors of Hauschild and Wenzel (1998) are based on similar exposure increases for all releases of the same quantity and substance. As clarified in the introduction, this contradicts with what we intuitively expect. Section 7.2.2 evaluates the need for spatial differentiation in the several sub-categories. Section 7.2.3 up to and including Section 7.2.7 explore the feasibility of a framework based upon Potting et al. (1999) to establish site-dependent factors that assess the increases of human exposure from air emissions.

7.2.2 The need for spatial differentiation

The degree to which a source contributes to exposure depends to a large degree on the properties of the substances, the characteristics of the source and the characteristics of the receptor (see also Potting and Hauschild 1997). The characterisation factors of Hauschild and Wenzel (1998) account fully for substance information, but address only to a limited extent information about source and receptor. In addition, Hauschild and Wenzel (1998) do not cover variation in characteristics of sources and receptors. Guinée et al. (1996) allow for some spatial differentiation by distinguishing between exposures resulting from emissions to agricultural soil, industrial soil and to other soils. Huijbregts (1999) continues with this differentiation into soil types, and adds a differentiation into emissions to fresh water and seawater.

Differentiation between emissions to different water and soil types may be a useful supplement to the characterisation factors of Hauschild and Wenzel (1998). However, the characterisation factors of both Guinée et al. (1996) and Huijbregts (1999) integrate all types of exposure resulting from emission to the given medium into one characterisation factor. This makes them incompatible to use in combination with the dis-aggregated factors from Hauschild and Wenzel (1998).

In the discussions about priority setting for the research underlying Section 7.2, we estimated direct exposures through ingestion of both fresh water and seawater to be small compared to indirect exposures after redistribution (most of the industrialised countries after all purify their water before supplying it as drinking water). Further, we estimated indirect exposure through ingestion of food from agricultural soils by far dominant compared to indirect exposure through ingestion of food from other soils. Differentiation between different water and soil types was therefore not prioritised and is not further addressed in this chapter. Better foundation for refraining from further differentiation in soil and water types is recommended for future research. ^[54]

Distinction between geographic locations of emission is another type of differentiation discussed in priority setting for the research underlying this chapter. We considered this differentiation of major importance for exposures from atmospheric emissions (see Section 7.2.3 up to Section 7.2.7), but of less relevant for exposures from emissions to water and soil. The present state-of-the-art does moreover not allow spatial resolved modelling of direct exposures from emissions to water and soil, and indirect exposures form emissions to all media.

Deliberating the importance of the possible and/or relevant spatial differentiation in the several exposure routes, we decided to focus the research underlying Section 7.2 on spatial differentiation in direct human exposure from emissions to air. The results of this research are reported in Section 7.2.3 up to 7.2.7. However, better foundation for refraining from spatial differentiation of exposures from emissions to soil and water is recommended for future research.

7.2.3 Human exposure from air emissions ^[55]

Among the different exposure routes, inhalatory exposures have as a unique feature that they are ubiquitous and can thus not be avoided once a substance is present in air (Williams sine dato). Inhalatory exposures also result in most cases directly from emissions to air rather than being the result of re-distribution between the different environmental media. This chapter focuses only on inhalatory exposures from atmospheric releases.

Dispersion and dilution in air are quick processes and the concentration increases at ground level are close to the source very much influenced by wind speeds and source characteristics like height and dynamics of the release (see Figure 7.3). The effective- release height is decisive for the concentration increase at ground level. A moderately high release (25m) has its peak concentration increase within 0.5km from the source, while the peak concentration from a high release (>150m) is a hundred times lower and occurs within 5km. The concentration increase from an emission at ground level (<1m) has its peak within a few metres from the source and is here a thousand times higher than for a similar emission released at a height of 25 metres.

Close to the source, the concentration is governed by dilution and the concentration increases are here for all substances almost equal when released at similar release height. The release height becomes less important at increasing distance, however, and removal processes start to take over as is illustrated in Figure 7.4 for the short-lived substance hydrogen chloride and the long-lived substance benzene. Removal processes are largely substance dependent. They are the combined result of deposition and chemical transformations. The emissions of substances with different removal characteristics begin to show their own concentration pattern, and concentration patterns for the emissions of the same substance released at different heights start to converge. At large distances, the height of release becomes negligible and removal processes finally take fully over (see Figure 7.5).

Click here to see the Figure.

Figure 7.3. Concentration increase at ground level versus distance local to the source (from 0 to 5km) from an emission of one gram per second in the Netherlands. Concentration increases have been calculated with the OPS model (Van Jaarsveld 1990, Van Jaarsveld en de Leeuw 1993). The numbers on the y-axis have to be multiplied with the factor for the given release height in the legends to obtain the proper order of magnitude.

Click here to see the Figure.

Figure 7.4. Concentration increase at ground level versus distance semi-local to the source (from 5 to 50km) from an emission of one gram per second in the Netherlands. Concentration increases have been calculated with the OPS model (Van Jaarsveld 1990, Van Jaarsveld en de Leeuw 1993).

Click here to see the Figure.

Figure 7.5. Concentration increase at ground level versus distance local to the source (from 50km to several hundred to thousand kilometres) from an emission of one gram per second in the Netherlands. Concentration increases have been calculated with the OPS model (Van Jaarsveld 1990, Van Jaarsveld en de Leeuw 1993).

Atmospheric conditions also influence concentration increases. Wind velocity is of major importance close to the source, whereas precipitation (important for wet deposition) and hours of sunshine (important for photo-chemical transformation) become more important at larger distances. Annual mean atmospheric conditions differ from region to region. Notably wind speeds and precipitation tend to be high in maritime and low in continental climates.

The population densities in the area with increased concentrations are the last factor determining the increase of human exposure. Population densities vary considerably over Europe (see Figure 7.6). Western Europe is far more densely populated than North-East and Northern Europe. Neighbouring regions differ less dramatically. Population densities near to the source are of special interest since concentration increases are highest here. Built-up areas show large differences in population densities. (Tobler et al. 1995, Stanners and Bourdeau 1995, EEA1998)

Figure 7.6. Estimate of population densities for 1994 from Tobler et al. (1995). Locations of the Northern, Central, Southern European and maritime sites are indicated with capital letters.

Potting et al. (1999) presented a framework to establish site-dependent factors to be used in life cycle assessment that assess the accumulated human exposure increase from outdoor emissions by taking into account all factors described above. This framework could not be applied in its initial form and has therefore been adapted slightly for the research reported here. It consist of 5 steps:

1. Identification of source types, and classification of processes.

Consecutively, for each source type:

2. Estimation of human exposure increases local to the source (from 0 to 10km) as a function of the concentration increases and population densities.
3. Estimation of human exposure increases regional to the source (from 10 km to several hundred to thousand kilometres) as a function of the concentration increases and population densities.
4. Evaluation and quantification of the exposure situations (sum of background concentration plus concentration increase) as being above or below a threshold value.
5. Establishing site-dependent factors that quantify the contribution to human toxicity from air emissions integrated over the exposed area.

Section 7.2 focuses fully on assessment of exposure increase (Step 1 to 3), but refrains from evaluating whether the exposure situation is above or below a threshold value (step 4). The site-dependent factors that are established in step 5, therefore quantify accumulated exposure increase rather than human toxicity.

The site-dependent exposure factors are used in combination with the old characterisation factors in Annex 7.1-7.3

7.2.4 Identification of source types, and classification of processes

Near to a source, the release height is decisive for the concentration increases at ground level. The release height for transport will typically be near to ground level (<1m). The information about the release height of industrial processes is usually not available in life-cycle assessment and can vary considerably between sources. The types of industrial processes are on the other hand typically known in life cycle assessment since this is one of its basic informations. An interesting question is therefore whether it is possible to identify source types (or classes of release height) to which industrial processes can be allocated.

The Dutch Emission Registration (DER) maintains a rather unique database that contains detailed information about the water and air emissions from over 700 major industrial sources in the Netherlands. Amongst others, the data cover release height, emission type (substance, quantity, flow, temperature), process type and industrial sector. The information is provided mainly by industry itself on a voluntary basis, and to a limited extent also by authoritative bodies like those for water management. All relevant data are recorded for the individual emission points and the individual equipment within each company. (Berdowski et al. 1995)

Data from 1994 have been used to analyse whether a relation exists between the height of release points and their connected processes. The median and mean height of release have been determined for records belonging to each industrial sector and for all industrial sectors together. The Statistical Analysis System (SAS Software Release 6.11) was used to analyse the data. The results are presented in Table 7.2. Though these results are expected to give a good indication of release heights, they should only be taken as indicative due to data problems ^[56] and because they represent data from Dutch industry only. Particularly for countries outside Europe, North America and Japan, the release heights may be considerably lower.

Table 7.2: The median, mean and range of release height for each industrial sector in a database containing detailed information about the air emissions from over 700 major industrial sources in the Netherlands based on analyses of data from the Dutch Emission Registration.

Click here to see the Figure.

The overall mean height is 31m with a standard deviation of nearly the same size. However, the median counts only 20 metres due to the majority of individual sectors that have considerably lower means. The mean is pulled up by a few sectors with very high emission points: "Cokes and crude oil refinery/processes", "centralised electricity production & district heating", and "waste processing" (mainly incineration). Excluding these sectors gives a mean ^[57] height of release for the remaining observations of 25m.

The median, mean and range of release height for each industrial sector have been calculated including as well as excluding the own production of energy (steam and/or electricity). The influence of own energy production on the median and mean is for most sectors minor. For a few sectors, the high release heights disappear from the observations without dramatically changing the median and mean when own energy production is excluded. Only the mean height of release for the wood & furniture industry falls considerably (from 20m to 11m). However, we have to take into account the very small number of observations for this sector. The agreement between the mean and the median is for most sectors reasonably good. Also the standard deviations for most sectors are relatively small when the number of observations is reasonable.

Most sectors have a relatively low mean height of release: The "extraction of gas, minerals, otherwise" (15m), "textile industry" (17m), "leather preparation" (19m), "wood & furniture industry" (20m), "paper industry " (22m), "printing houses" (15m), "plastic and rubber processing" (15m), metal processing (13m) and the group of "remaining industrial sectors" (17m). The release heights of the percentiles in the higher end are usually also relatively low. Some sectors contain also a few observations of relative high releases, but these observations are in most cases related to own energy production. The "leather preparation" contains only three observations from which two were very low (6 and 7m) and one higher (45m).

The "food and tobacco industry" (25m), "chemical industry" (30m) and "glass, pottery, stone and cement industry" (31m) have moderate mean heights of release that are not much influenced by excluding own energy production. The production of sugar, fish flour & glue etc., grass and pulp drying and coffee roasting somewhat pull up the mean height for the "food and tobacco industry". The maximum release height of 100m relates to sugar production. The "chemical industry" contains production of many different chemicals. Most sub-sectors have moderate release heights, but those for the production of inorganic basic chemical and petrochemical products are somewhat higher, while the medians are higher and maximums even very high. The high to very high sources are in most cases related to processes involving some kind of combustion process.

The mean release height is relatively high for the "crude oil refinery and processing" (91m), but surprisingly enough rather modest for "electricity production and heating" and "waste processing (mainly incineration)". The mean for "electricity production and heating" increases considerably, however, when the release heights are weighted with the annual production of electricity (133m). Something similar is the case for the "crude oil refinery and processing" and "waste processing". Since these two sectors cover output of several different products, however, it was impossible to calculate a weighted mean. The release heights of "waste processing" are high for processes related to waste incineration (93m).

The above results show that the majority of Dutch industry has relatively moderate release heights. This is further underlined by the fact that these results relate to the 700 main industrial sources while the less important ones are not covered. There are only few industries with high sources.

The results refer to height of release rather than to "effective release height". The effective release height also includes the plume rise as a result of its heat content. It is actually the effective release height that is relevant for dispersion and dilution calculations. Based on this, three release heights have been chosen for which accumulated exposure increases will be calculated: 1m for traffic, 25m representing the majority of industry, and 150m being a reasonable estimate for high sources. All results here refer to Dutch industry that may not appropriately reflect industry in other countries due to differences in legislation and production capacities and technologies. But the three classes of release heights are expected to be relevant for Europe.

7.2.5 Accumulated human exposure increase local to the source (from 0 to 10km)

An atmospheric emission leaves its source in a concentrated form that is sometimes visible as a plume. The plume usually mixes with the surrounding air before it results in concentration increases at ground level. Wind speeds largely determine how fast the plume dilutes, whereas the release height also influences how fast the plume reaches ground level. This is typically modelled with a Gaussian plume approach (Harssema 1995). Atmospheric conditions differ from region to region. Wind speeds tend to be high in maritime and low in continental climates. Van Jaarsveld follows a Gaussian plume approach that applies region-dependent atmospheric conditions (1990 annual statistics mean) in his EUTREND model (Van Jaarsveld 1995, Van Jaarsveld and De Leeuw 1993, Van Jaarsveld et al. 1997). This model is used here to estimate concentration increases at short distances from the source (from 0 to 10km) for different release heights and for sources located in different climates:

• A maritime climate (approximated by atmospheric conditions in the Netherlands; 5 long., 52.25 lat.)
• Climate in North Europe (approximated by atmospheric condition in Finland; 25 long., 65 lat.)
• Climate in Central Europe (approximated by atmospheric conditions in Austria; 14 long., 47 lat.)
• Climate in South Europe (approximated by atmospheric conditions in Italy; 15 long., 42 lat.)

The accumulated exposure increase has been calculated for a long-lived substance (benzene; residence time of seven days) and a short-lived substance (hydrogen chloride; residence time of seven hours). These two substances have been selected because the residence time and therewith the accumulated exposure increase from substances will in general be in between those of hydrogen chloride and benzene. The source strength is kept at one gram per second continuous, but the consequence of the release heights as identified in Section 7.2.4 has been analysed (1m, 25m, and 150m). The accumulated human exposure increase from a release on location (i) at each distance is the product of concentration increase times population density integrated over the whole surface:

Click here to see the Figure.

The population density is assumed to be constant (one person·km^-2) in the areas local to the source. The result of the calculations is the increase of human exposure (in person·µg·m^-3) accumulated over the area from 0 to 10 km from the source. The courses of the accumulated human exposure increase versus distance are depicted in Figures 7.7 to 7.11, and abstracted in Table 7.3. Clear differences exist between the accumulated increase at varying release heights on the one hand, and climate regions on the other hand, as well as between the two substances hydrogen chloride and benzene.

Click here to see the Figure.

Figure 7.7. The increase of accumulated exposure versus distance local to the source (from 0 to 10 km) from one gram benzene released at 1m in four different climatological regions in Europe (population density is one person·km^-2).

Click here to see the Figure.

Figure 7.8. The increase of accumulated exposure versus distance local to the source (from 0 to 10 km) from one gram benzene released at 25m in four different climatological regions in Europe (population density is one person·km^-2).

Click here to see the Figure.

Figure 7.9. The increase of accumulated exposure versus distance local to the source (from 0 to 10 km) from one gram benzene and hydrogen chloride released at 150m in four different climatological regions in Europe (population density is one person·km^-2).

Click here to see the Figure.

Figure 7.10. The increase of accumulated exposure versus distance local to the source (from 0 to 10 km) from one gram hydrogen chloride released at 1m in four different climatological regions in Europe (population density is one person·km^-2).

Click here to see the Figure.

Figure 7.11.The increase of accumulated exposure versus distance local to the source (from 0 to 10 km) from one gram hydrogen chloride released at 25m in four different climatological regions in Europe (population density is one person·km^-2).

Click here to see the Figure.

Figure 7.11.The increase of accumulated exposure versus distance local to the source (from 0 to 10 km) from one gram hydrogen chloride released at 25m in four different climatological regions in Europe (population density is one person·km^-2).

Table 7.3. The increase of accumulated exposure from one gram of benzene and hydrogen chloride at different distances from the source (0.5km, 5km and 10km), and released at different heights (1m, 25m and 150m) and in different climate regions in Europe. The accumulated exposures are expressed as the proportion from the accumulated benzene exposure at a height of 25m at 10km distance (20·20km²) in South Europe (69.7 person·µg·m^-3). The population density is in all cases one person·km^-2.

There is a notable difference in accumulated exposure between the maritime and North European climate regions on the one hand, and the South and Central European climates on the other hand. The climate region becomes more important with lower release heights. This is caused by the considerable difference in wind velocities between the regions. Low wind velocities give slower dilution and subsequently higher concentrations than high wind speeds (direct effect). In addition, low wind velocities go together in a stable and neutral atmosphere with low mixing heights (indirect effect). Decreasing mixing height has an increasing effect on the concentration increase for low sources, but a decreasing effect on the concentration increases for high sources (since they release above mixing height). Wind velocities in the south and central climate regions are on average lower than those are in the maritime and northern climate regions. The direct and indirect effects of wind velocity therefore result in diverging concentrations for low sources, but in converging concentrations for high sources between climate regions. Atmospheric conditions also show variation within regions themselves, as well as over the seasons. However, the subsequent variation in accumulated exposure increase remains within 10% (based on calculations with a similar model that is not further reported here: Van Jaarsveld and De Leeuw 1993).

Though benzene and hydrogen chloride have similar accumulated increases at very short distances (< 0.5km), they already show a clear divergence within the first 10km for low release heights. This is mainly due to dry deposition which is considerably higher for hydrogen chloride than for benzene. Emission at 150m will often be above the mixing height and therefore result in increases of accumulated exposures that are nearly the same for both substances. Removal by dry deposition is still very small here since the concentrations at ground levels are small.

All accumulated exposures in Table 7.3 and Figure 7.7 to 7.11 relate to a population density of one person·km^-2. Tobler et al. (1995) give detailed data about population densities in the European grid. As can be seen from Figure 7.6, which represents this data at regional resolution, population densities vary considerably between European regions (from one to more than five hundred persons km^-2). Tobler et al. (1995), EEA (1998) and Stanners and Bourdeau (1995) show that large variation also exists within built-up areas. The population of big cities for instance is distributed rather unequally over the city-area, though generally concentrated in the city-centre. City-centres are mostly limited to an area of at maximum 10 10km (exceptions may be Berlin, London and Moscow). Typical population densities have been defined on the basis of data from Tobler et al. (1995) about Paris:

• 21,500 persons km^-2 in city-centres (based on Paris-city), and
• 900 person km^-2 in built-up areas in general (based on Paris-metropolitan area)

Hence, the total population of 2.2 million persons in Paris-city (roughly 10 10km²) exceeds by far the total population in most cities and smaller communities. This would even be the case if Paris-city had a population density of 900 person ;km^-2 (which would correspond to a total population of 90.000 persons). One thus has to be careful in choosing the population density within the first 10km (or 20 20km²) from the source. While the population density in Figure 7.6 reflects the urbanisation in the relevant region, often no information is available in life cycle assessment about the exact location of a source in relation to built-up areas. We therefore suggest to take the mean population density within the relevant region in Figure 7.6 as the one for those within the first 10km.

7.2.6 Accumulated human exposure increase regional to the source (from 10km to several hundreds to thousand kilometres)

At longer distances, the plume from a source in vertical direction is distributed equally in the mixing layer of the atmosphere. Trajectory or one-dimensional Lagrangian modelling is an often-used way to trace concentration increases resulting from substance transport and removal at long ranges (Seinfeld and Pandis 1998). Atmospheric conditions differ from region to region, and notably precipitation and wind speeds tend to be high in maritime and low in continental climates. The Wind rose Model Interpreter (WMI) of the integrated assessment model EcoSense (Krewitt et al. 1997) follows trajectory modelling based on region dependent atmospheric conditions (1990 annual statistics mean). The WMI-module is used here to assess long-range accumulated exposures (see Annex 7.4 for model specifications and adaptations).

Similar to the modelling of exposure local to the source, calculations have been performed for hydrogen chloride and for benzene at an emission rate of one gram per second continuous. At the regional scale, release height only has minor importance for a long-lived substance as benzene. However, it will considerably influence the removal by deposition and consequently the amount of a short-lived substance going into long range transport. The calculations for hydrogen chloride have therefore been performed for the release heights 1m, 25m and 150m by correcting for deposition in the source grid-square of respectively 38%, 30%, and 9% of the emissions (see Annex 7.4).

Since atmospheric conditions are region-dependent, the patterns of concentration increase will depend on the geographical location of sources. Also population density is region-dependent. The concentration increase in each 150 150km grid-square is multiplied with the total population in that grid-square (derived from the same data as used for Figure 7.6). Next, the products of concentration increase times population in each grid-square are accumulated over the grid (see Formula 7.1). The result is the accumulated exposure increase (in person µg m^-3) over a distance of several hundreds to thousands kilometres from a source located in the specified grid-square in the European grid. The procedure is repeated for every grid-square as the source grid-square. The results for hydrogen chloride are represented in Figure 7.12 and the results for benzene in Figure 7.13.

Figure 7.12. The increase of exposure accumulated (in person µg m^-3) over the total receiving area (from 10km to several hundred to thousand kilometres) posed by a release of one gram emission hydrogen chloride at 25m in the source grid-square. The mean is 2,460 person µg m^-3, the standard deviation is 1600 person µg m^-3 (both weighted for population density). The accumulated exposure is attributed to the respective source grid-square. The increase in accumulated exposure from a release at 150m can be obtained by multiplying with a factor 1.30 (s.d. 0.02). The accumulated exposure increase from a release at 1m can be obtained by multiplying with a factor 0.89 (s.d. 0.04).

Figure 7.13. The increase of exposure accumulated (in person µg m^-3) over the total receiving area (from 10km to several hundred to thousand kilometres) posed by a release of one gram benzene at 25m in the source grid-square. The mean is 50,000 person µg m^-3, and the standard deviation is 33,000 person µg m^-3 (both weighted for population density). The exposure increase is extrapolated to transport distances where all benzene is removed from the atmosphere (see Annex). The accumulated exposure is attributed to the respective source grid-square.

The patterns of accumulated exposure increase for hydrogen chloride in Figure 7.12 roughly reflect the pattern of population densities in Figure 7.6. The pattern in Figure 7.13 of accumulated exposure increase for benzene is much smoother. Also the variation of exposure increases over the grid is smaller for benzene than for hydrogen chloride. There is a difference of less than a factor 20 for benzene, but almost a factor 100 for hydrogen chloride between highest rating source grid (South-Eastern Netherlands) and the very low rating source grids (in some very sparse populated areas in the far North) of accumulated exposure increase. Accumulated increases are rather close between neighbouring grid-squares. However, the spatial differences are expected to become sharper when exceedance of threshold values would be taken into account (see also Section 7.3).

To closer examine the importance of spatial variation in atmospheric conditions and population density, accumulated human exposure increase has been plotted versus distance for seven sites. Four of these sites are similar to the ones selected in Section 7.2.5. The other three sites are the neighbouring grid-squares of the Dutch site. The four Dutch sites experience rather similar atmospheric conditions, but their populations vary considerably (largely explained by the land-water ratio in those grid-squares). The results are represented in Figure 7.14 and 7.15.

Click here to see the Figure.

Figure 7.14. Increase of accumulated exposure for one gram hydrogen chloride released at 25 m at the North, Central, South European and maritime site. To compare dependence of the increase of accumulated exposure increase on climate with dependence on local population, three Dutch sites with similar climate as the maritime site but different population densities have been included.

Click here to see the Figure.

Figure 7.15. Increase of accumulated exposure versus distance for one gram benzene released at 25m at the North, Central, South European and maritime site as a function of distance from the source. To compare dependence of the increase of accumulated human exposure on climate with dependence on local population, three Dutch sites with similar climate as the maritime site but different population densities have been included. Though the residence time of benzene corresponds to a source distance of about 3500km we present model results only up to the distance where the trajectories hit the nearest edge of the model domain.

The curves in Figure 7.14 and 7.15 are linearly interpolated due to the accumulated grid-squares each representing a discrete value (being the product of concentration increase and population). The marked sections in the hydrogen chloride diagram illustrate that grid-squares of 150 150km are rather coarse for exposure assessment of a short-lived substance. The model domain as a whole is on the other hand too small to trace a long-lived substance as benzene over its full atmospheric residence time. The residence time of benzene corresponds to a source distance of about 3500km, but the source distance of the nearest edge of the model domain is between 1600km (South and North Europe) and 2400km (Northern Netherlands) for the sites considered here. Due to the longer lifetime, accumulated exposure increase of benzene is less dependent on the population in the source grid-square and more on the regional population than for hydrogen chloride.

The curves for the short-lived hydrogen chloride in Figure 7.14 are dominated by the influence of population in the source grid-square (within 80 km from the source), and in the four nearest neighbour grids (step at 150km). The grid-square in the South-Eastern Netherlands extends into Belgium and Germany. It is one of the most densely populated grid-squares with almost 11 million persons. In spite of the high wind speeds that dilute concentrations, the accumulated human exposure increase for emissions in this grid is therefore larger than for the sites in other climate regions. The south-western (maritime) and north-eastern sites in the Netherlands are also densely populated (respectively 6.4 and 4.5 million persons). The next group of curves includes the grid-squares with smaller populations in the North-Western Netherlands and Austria (both 1.5 million persons) and Italy (0.9 million persons). The exposure increase in the North-West Netherlands and Italy source grid squares is almost equal due to high wind speeds in the Netherlands (7 m s-1) that reduce, and low wind speeds in Italy (4 m s-1) that enhance concentration increase. Austria shows a larger exposure increase than Italy, because wind speeds in Austria (5 m s-1) are slightly higher and the population is considerable higher than in Italy. The discussion in the previous section already has shown that atmospheric conditions at the North European site lead to low concentrations. However, in this diagram it is the very low population with 0,2 million persons in the source grid-square and 0.1-0.3 million persons in the adjacent grid-squares that leads to an extraordinarily low human exposure.

Accumulated exposure increase of benzene (Figure 7.15) shows to be much less related to the population in the source grid-square but more to regional population than the short-lived hydrogen chloride. Again, the highest accumulated exposure is caused by emissions from the South-Eastern Netherlands. The curve for emissions from the Central Europe grid-square (Austria) is almost catching up though, whereas the population in the source grid-square is by a factor of seven smaller. That is because the Central European site is situated between highly populated areas in the United Kingdom, the Benelux countries, France, Germany, Italy, Hungary, and Poland. The South-Eastern Netherlands has a rather peripheral location on the other hand. Even the South European site is ranking before the North-Western Dutch grid-square. The course of accumulated exposure increase from the four Dutch sites with similar regional populations but different populations in the source grid-square is comparable after several hundred kilometres.

Atmospheric conditions have a smaller influence on the pattern of accumulated human exposure increase than expected (not plotted). Even for the short-lived hydrogen chloride, which is rapidly scavenged by precipitation, the wind speed and precipitation show to be of minor importance for human exposure increase accumulated over long distances compared to population. This finding is stronger still for benzene whose small scavenging ratio renders wet deposition negligible and whose atmospheric fate is far less determined by local wind speeds, due to its long transport distances.

7.2.7 Total increase of accumulated exposure from air emissions

Total exposure increase from an emission is the sum of the accumulated human exposure increase local to the source (from 0 to 10km) and the exposure increase regional to the source (from 10km to several hundreds or thousands kilometres).

Where:

ΔAHE_s,i(......) = The factor that relates one gram of substance (s) released at a height (rh) by a source located in grid-square (i), to the increase in accumulated human exposure (in person µg m^-3). Local refers to increase within 0 to 10km from the source (20 20km), and regional refers to an increase over a distance of several hundreds to thousands of kilometres from the source.

ΔAHE(default) = The factor that relates one gram of substance (s) released at default height (25m) and under default atmospheric circumstances (South Europe), to the increase in accumulated human exposure (in person µg ;m^-3) within short distances from the source (<10km) (based on a population density of one person per km²).

F(rh,ac)_s,j=i = The factor modifying the default increase of accumulated human exposure according to the actual height of release (rh) and the actual atmospheric circumstances (ac) in the receiving area (j=i) local to the source, and the substance released (s)

PD_j=i = The factor modifying the default increase of accumulated human exposure according to the population density in the receiving area (j=i)

The accumulated human exposure increase over long distances from the source can be taken from Figure 7.12 for hydrogen chloride and from Figure 7.13 for benzene. The exposure increase of benzene is hardly influenced by release height and thus for all heights similar. For hydrogen chloride, the data have to be adjusted according to the factors for the relevant release heights that are mentioned in the caption to Figure 7.12.

The accumulated human exposure increase local to the source can be calculated from the default of 69.7 person µg m^-3 for a population density of one person km^-2 multiplied by the population density in the grid-square where the source is located. The default of 69.7 person µg m^-3 relates to the exposure increase from benzene and from a release height of 25m in the area of 0 to 10km from the source (20 20km). Table 7.3 provides factors to modify this exposure increase according to the substance being emitted (benzene or hydrogen chloride), the height of release (1m, 25m, 150m), and the climate region in which the source is located.

The results of applying Formula 7.2 and 7.3 to an emission of one gram hydrogen chloride and benzene are illustrated in Table 7.4. The table shows basically the same trends as discussed in the previous sections, though the dominance of the accumulated human exposure increase local to the source becomes very strong now for hydrogen chloride. Short-lived substances as hydrogen chloride have a large share of their impact within the first 10km from the source and this is obviously intensified for low heights of release and high population densities. The sensitivity for source height and population density local to the source makes the exposure increase from short-lived substances far more uncertain than the exposure increase from long-lived substances. As Table 7.4 illustrates, long-lived substances as benzene usually will have most of their impact regional to the source.

Table 7.4: The sum of the local and regional increase of exposure for one gram of benzene and hydrogen chloride release at 25m at seven sites in Europe.

The analysis of release height for different industrial processes as reported in Section 7.2.4 indicates that a release height of 25m can be taken as default for most processes (the majority of sources have a release height around 25m). It is suggested not to deviate from this default release height of 25m unless a deviating release height is strongly probable. Only release heights for energy production and few other industries (see Table 7.2) typically will be relative high compared to the overall mean. A release height of 1m is relevant for emissions from transport.

Population densities will usually be higher in build-up areas (900 person km^-2) and city-centres (21,500 person km^-2). This will especially have large influence on exposure increase when the source is located in the vicinity of a relatively large city and the peak of the concentration increase occurs in the middle of this city. However, the location of a source in relation to a built-up area will in general not be known in life cycle assessment. In addition, a population density of 900 person km^-2 over 10km² corresponds to a population of 90,000 persons. While a city of that size is as such not unusual, this will often already be reflected in the population density of the region where this city is located (see Figure 7.6). The proposed framework in this chapter therefore suggests taking the population density in the relevant region (see Figure 7.6) to quantify the exposure increase local to the source.

The number of toxic substances in a typical life-cycle inventory can be considerable. However, the spatial differentiated exposure factors in this chapter relate to two substances only: Hydrogen chloride and benzene. These two substances have been selected on the basis of their lifetimes. Hydrogen chloride represents substances with a very short lifetime, while benzene represents rather long-lived substances. The exposure increase from other substances will thus in general be in between the increase from hydrogen chloride and benzene. The results for these two substances can therefore with reasonable confidence be used to evaluate spatial variation in source locations.

Exposure increase within short distance from the source and over long distances has been quantified with two different models. The models have been calibrated to the extent possible such that input and model data were comparable. Nevertheless, both models are based on different mathematics. The results have been added together without extensively checking how fluently the exposure increase from the local model runs over into the one of the regional model. A rough comparison, which is not further reported here, has been made with results from the OPS-model (Van Jaarsveld 1990, Van Jaarsveld and De Leeuw 1993). This model is limited to source grid-squares in the Netherlands, but fluently integrates modelling of local and regional concentration increases. The comparison suggests that it is reasonable to sum local and regional exposure increases as done here. A better calibration is recommended, however, in combination with quantification of uncertainties related to each model as such and from putting them together.

The presented site-dependent factors cover accumulated exposure increase from a release, and disregard whether threshold values are exceeded as a result of the exposure. The spatial differences between the factors are expected to become more pronounced if exceedance of threshold values would be taken into account.

7.3 Effect assessment (evalution of threshold exceedance)

7.3.1 Introduction

The impact assessment phase in life cycle assessment initially emerged from the wish to aggregate the large amount of data from inventory analysis to a manageable amount of impact data. For most impact categories and also for human toxicity, rather simple modelling was used at first to establish characterisation factors. Those characterisation factors were all based on equivalency assessment on the basis of intrinsic substance characteristics. For human toxicity, the emission data from inventory were usually converted in dilution volumes by means of dividing emission data by some kind of threshold value. (Potting et al. 1999, Potting 2000)

No assessment of threshold exceedance was performed in this simple modelling since the available data did not allow such evaluation. Threshold information was used in toxicity assessment only to express the emission quantity of a given substance in the volume needed in the amount of receiving environment needed to dilute the emission until the threshold value was reached. The so calculated dilution volumes provided the possibility to aggregate substances with different toxic effects. (Potting et al. 1999, Potting 2000)

The basis of equivalency in this simple modelling was taken in the toxicity potential of each substance ^[58]. The impact from an emission quantity equal to the threshold value was put on one, and the impact from any deviating quantity was assessed as the ratio of that quantity divided by the threshold value for that substance. The underlying assumption was that the toxicity impact from an emission quantity at the threshold value for a substance has the same importance as the toxicity impact from the quantity at the threshold value of another substance. To put it more clearly: If the quantities of both substances are at their threshold value, the impacts from a neuro-toxic substance and an irritating substance are regarded as equally important ^[59] (Potting et al. 1999).

The initial human toxicity factors have later been replaced by factors based on more sophisticated modelling. Present typical toxicity factors as from Guinée et al. (1996), Hauschild and Wenzel (1998), Huijbregts (1999) and Hertwich et al. (2001) now also cover modelling of fate and exposure increase. Aggregation of calculated exposure increases from different substances is still based on threshold values. Where initially occupational standards were often used, however, it is now common practice to use no-effect-concentrations or safe doses.

As mentioned in Section 7.2, the presently typical methods for the assessment of human toxicity impact in life cycle assessment do still not take into account background exposures, since they calculate exposure increases rather than actual exposures (sum of background exposure and exposure increase). This section discusses the need for threshold evaluation, and provides information for a selection of substances to implement a qualitative evaluation in life cycle assessment.

7.3.2 Need for evaluation of threshold exceedance

There is ongoing discussion about whether and how to perform an evaluation of threshold exceedance in life cycle assessment. The discussion is relevant in particular with regard to human toxicity. Human toxicity assessment initially focused on quantifying the probability of surpassing threshold values (like no-effect-concentrations or regulatory standards) in the direct vicinity of emission sources. This focus was directly related to the perceived local character of human toxic impact. Contrary to the present situation, increased pollution levels could in general be traced back to a neighbouring single source (usually large point sources like industry of waste processing). The severity of human toxic and eco-toxic impact local to some sources moved environmental regulation to curb the most pressing situations and prevent similar ones in future. Since then, a large body of policy instruments has been implemented (like licenses, levies and subsidies, and anti-pollution taxes). Risky situations local to sources have in this way been prevented by keeping the emissions from the point source under control. Risk was interpreted as exposure levels above a given threshold. (Potting et al. 1999)

Meanwhile, the effective implementation of environmental policies in most of the industrialised countries has led to a situation where the emissions of toxic substances from large point sources in general have been reduced considerably. Large reductions have been achieved by add-on emission reducing technologies or by structural technological improvements in production processes. Risk from a single source to its local environment is nowadays prevented in most cases, and the total emissions from all large point sources together have been reduced substantially for many substances over the last decades. (Potting and Hauschild 1997, Wenzel et al. 1997, Hulskotte et al. 1997, EEA 1998)

In spite of the remarkable successes in emission reduction, management of risk from toxic substances is more topical than ever. As mentioned above, individual sources rarely by themselves cause an exceedance of threshold values in their local environment. However, the intricate net of sources and the dispersion of many substances over large distances have led to a widespread presence of a broad variety of different substances that result from many sources together (rather than from a single source). Even though the exposure to an individual substance may remain below its no-effect-concentration, the accumulated exposure to a cocktail of substances may pose a risk. The toxic impact of exposure to a mixture of different substances is unknown, but can be additive or even synergistic. In addition, whereas initially no-effect-concentrations were supposed to exist, there is nowadays growing evidence that a substance may have effects below its assumed no-effect-concentration. This concern is raised by a number of observations of effects on human beings and animals that are highly suspected to relate to such combined exposures. Examples of this are the reduced sperm quality in mens, the increase in testicle cancer and the relative increase in the frequency of female trout in some water bodies (Toppari et al. 1995) and the increase in asthma and allergy among humans in Western European countries.

In the present environmental situation, it has become relevant to consider exposure situations both above and below relevant threshold values, but still to evaluate exposures above threshold as being more severe compared to the ones below threshold. This was the main incentive to explore the possibilities for a qualitative evaluation of threshold exceedance in life cycle impact assessment as part of the Danish LCA-methodology development and consensus creation project.

7.3.3 Possibilities for including evaluation of threshold exceedance ^[60]

As demonstrated in Section 7.2.3, the concentration increase from an outdoor source may have a considerable contribution local to that source (from 0 to 10km), but it rapidly becomes smaller at longer distances. Neither close to the source nor at longer distances, however, will outdoor emissions nowadays be expected to cause increases that by themselves lift the ambient concentration from below to above threshold values (see also Section 7.2.3). The ambient concentration at a given location, the concentration that one inhales, is the sum of the existing background concentration and the concentration increase from the given source. Information about the background concentration is thus indispensable to perform an adequate threshold evaluation.

The background concentration at a given location is often the result of contributions from more sources together, while these sources can each be close to, or more distant from that location. Obviously, background concentrations in densely populated areas will in general be higher than in sparsely populated areas due to differences in societal and economic activity (i.e., source density). The risk for threshold value exceedance by a concentration increase will therefore also be higher in densely populated areas. At the same time, the accumulated exposure increases are obviously also higher for sources located in densely populated areas (exposure being the product of concentration times population).

If exceedance of threshold values had been taken into account, the combined effect of this would probably have resulted in more pronounced spatial differences in the site-dependent factors as established in Section 7.2.

Integrated assessment models like the RAINS model described and used in Chapters 3, 4 and 6 are an obvious way to combine the estimation of background concentrations from multiple sources and the concentration increases from an individual source, with population densities and with threshold evaluation. Whereas such models already exist for other impact categories, however, they are only of limited availability for the assessment of human toxicity. Important reasons are the infinite number of potential human toxic substances, their considerable differences in fate, and the differences in number and type of their possible sources.

Also for human toxicity assessment, integrated assessment models are expected to become more important. At present, developments take place in that direction for several groups of substances. However, the present state-of-the-art in modelling and data availability regarding toxic substances does not yet allow making such threshold evaluation at more than a qualitative level.

One solution is to refrain from inclusion of threshold information in the quantification of human toxicity impact. This assumes that similar exposure increases are equally important, regardless whether the background concentration (together with the concentration increase) is below or above the relevant no-effect-concentration.

An intermediate approach could be to consider background concentrations as default being below threshold values, unless they are with high probability close to, or above these values. Such intermediate approach is presented in the next sections. This part of the developed framework also covers indoor situations.

7.3.4 Chosen type of threshold values

One threshold figure for exposure, above which effects occur and below which no effects are seen, is a crude simplification of reality. Humans differ in terms of age, size, sex, race and other physiological factors, and consequently they react differently upon chemical exposure.

Established "threshold" levels are often based on results obtained from animal experiments at relatively high exposure levels. This implies respectively extrapolation from animal-to-human and from high-to-low dose levels when establishing the threshold levels. In some situations it is further necessary to extrapolate from short-to-long term exposure durations.

Threshold levels for the external environment are developed to protect even the more sensitive population groups. Therefore, exposure above the threshold levels will not necessarily result in massive toxicological effects.

On the other hand, caution should be taken as definitions of threshold levels for regulatory purposes are sometimes not only based on a toxicological/scientific background but also take on technological feasibility, costs of compliance, prevailing exposure levels, social, economic and cultural conditions (WHO, 1998).

WHO stresses that a distinction should be made between WHO guidelines - derived from purely epidemiological/toxicological data - and other "quality standards" (like regulatory thresholds) where the above societal factors may have influenced the levels (WHO, 1998). The same principle is applied by the EU, which operates with both EU Guidelines and EU limit values. The guidelines are based on evaluation of scientific data only and are the levels below which only insignificant effects may be expected, whereas the legally binding limit values may be the result of cost-benefit analyses (EEA, 1997).

The developed framework will - where possible – compare actual exposure levels with guidance values; i.e. the guidance values will be considered "threshold".

These "threshold levels" include air concentration levels and acceptable/tolerable daily intakes (ADI/TDI's) which are assumed to cause no significant acute and also no chronic effects after life-long exposure/intake.

The existence of a threshold - meaning that an exposure level exists below which no toxic effect is seen - is controversial. Some scientists argue that no substances have thresholds, whereas others argue that all substances have. However, the more general assumption is that some substances act via a non-threshold mechanism; especially genotoxic substances. Therefore, acceptable exposure levels for these substances are often expressed as a level, which implies a life long risk of 10-5 or 10-6 of obtaining cancer.

7.3.5 Chosen emission/exposure situations and substances

As argued in Section 7.3.2, single sources in themselves do usually not evoke exceedance of threshold values for human toxicity anymore. It was after all local problems that initiated the extensive body of process oriented environmental policy. For the purpose of impact assessment in LCA, it seems therefore fair to assume that environmental concentration levels resulting from one single source generally remains below the no-effect-level due to the process oriented policy measures. At least this is expected to be the case for the most developed countries. However, there are exceptions from this assumption.

The first exception concerns the zone very closely surrounding the discharge point where exceedance of the no-effect-level may still occur. In environmental regulation, the exceeding of the no-effect-level is accepted within a certain defined dilution zone. This zone is in general situated at, or immediately surrounding the domain of the releasing industry, and might thus be considered a part of the technosphere rather than the ecosphere. It can therefore be argued that LCA can disregard this exception from the marginality assumption in characterisation modelling. (Potting and Hauschild 1997)

The second exception concerns occupational exposures and exposures resembling occupational exposures. This type of exposure situations are subject of another part of the Danish LCA-methodology development and consensus creation project. (Schmidt et al. 2000) and not further addressed here.

Other human exposure situations, which typically might be near or above thresholds are listed below. The list is a further development of a list prepared at the first workshop of the Danish LCA-methodology development and consensus creation project carried out 15 May 1998.

• Urban areas, especially with high traffic intensity ^[61]
• Long range transport of some air pollutants which may elevate regional background concentrations considerably
• Areas around point sources ^[62]
  - facilities manufacturing or applying chemicals
  - gas stations/storage tanks
  - (hazardous) waste sites
  - disposal facilities (incineration plants and other waste processing facilities)
• Accidents (accidental risk is usually considered separately in LCA)
• Indoors
  - indoor climate after painting, gluing, VOC emission from building materials etc.
  - formation of unwanted chemicals (for instance NO_x from gas stoves and heating facilities)
• Direct consumer exposure to products; for instance cosmetics (containing heavy metals, preservatives, perfumes and oxidised tensides which may all cause allergic reactions), textiles (containing for instance azo dyes, some of which may provoke allergic reactions), toys (which have appeared to contain potentially carcinogenic and endocrine disrupting substances) and a variety of products containing chromium and/or nickel (both being allergens)

Inhalation will be the primary exposure route for many of the described exposure situations, but also indirect exposure via drinking water and foodstuffs can be relevant as well as direct skin contact (mainly allergic reactions in the consumer phase).

Table 7.5 Overview of situations where guidance values may be exceeded regularly.

¤ To be used with caution. Has been derived from the TDI assuming: 64 kg body weight, inspiration of 22 m³ per day and the same bioavailability/uptake via oral and inhalation exposure. Especially the latter assumption may be questioned.

* `bw': Abbreviation for `body weight'.

# In relation to LCA, attention must be paid to substances with non-threshold mechanisms, e.g. benzene and particles. For these substances, any elevation in exposure will result in an elevated risk. The recommended threshold levels are therefore less relevant in relation to site characterisation.

& For industrialised countries, regulation of point sources will often aim at protecting the surrounding area from above threshold exposure situations. This assumption may often be interpreted as default, but exceptions may occur.

It is beyond the scope of this methodology development and consensus project to outline all substance specific exposure situations that may be "near or above threshold". A number of substances have therefore been selected. These substances are those contributing significantly to the EDIP97 normalisation factor for human toxicity (Hauschild and Wenzel, 1998). They are supplemented by a number of other typical and well-known environmental pollutants known as potentially harmful to human health, including particles and some aromatic and/or halogenated substances. The prevailing exposure levels will be outlined and where possible, they will be quantified and compared to threshold levels. An extensive description for each substance is given in Annex 7.6, but the results are summarised in Table 7.5.

Background concentrations in the outdoor environment have high probability to be close to or above threshold values on a regional scale in industrialised regions with a highly intricate net of sources emitting the given substance. Maps of regional exposure levels over Europe have been found for mercury (Ryabonshapko et al. 1998), lead and cadmium (Pekar et al. 1998a), lindane, polychlorinated biphenyls and benzo(a)pyrene (Pekar et al. 1998b), photochemical oxidants (Simpson et al. 1997), and nitrogen and sulphur (EMEP/MSC-W 1998). Berdowski et al. (1997) give emission projections over the European grid for a number of persistent organic pollutants and heavy metals. Annex 7.5 gives an overview of addresses where the several reports can be obtained.

Annex 7.6 gives a more qualitative description or regional exposure levels together with the exposure levels in those other situations as identified above.

It is difficult to give general recommendations on whether or not a specific exposure situation encountered in LCA can be assessed to be near or above "threshold". However, table 7.5 presents a crude summary of the findings of the present paper. The table should not be regarded as bare facts, but as a best attempt to summarise the previous paragraphs, which are again based on the level and extent of the consulted literature. Furthermore, it should be observed that many of the data of the consulted references originate from the beginning of the 1990'ies, since when the situation has been rapidly changing due to for instance new abatement technology and changes in the traffic load.

7.4 Conclusions

This chapter explores the influence of spatial differences in atmospheric conditions and source characteristics (location, release height, location in relation to built-up area) on the accumulated human exposure increase and subsequent effect from atmospheric emissions. The spatial differences in exposure increase are thoroughly quantified for two substances, the effects of possible exceedance of threshold values are covered in a more qualitative way. Some remarks about the established site-dependent exposure factors and about threshold exceedance have already been made in the previous section. Here, some main conclusions are drawn.

There is a difference of less than a factor 20 for benzene, but almost a factor 100 for hydrogen chloride between highest rating (South-Eastern Netherlands) and low ratings (in very sparse populated areas in the far North) of the (regional) accumulated exposure increase. The spatial differences are expected to become sharper when exceedance of threshold values would be taken into account.

The population density shows to have the largest influence on the differences in regional exposure increase from sources at different locations. The importance of population density becomes stronger for a short-lived substance like hydrogen chloride that has its increase of exposure predominantly within the first hundred km from the source. Also variation in atmospheric conditions (region-dependent) and source height have an influence on the regional exposure increase. However, their impact is smaller compared to population density.

Accumulated exposure increase local to the source is very dominant for short-lived substances as hydrogen chloride. Their impact occurs largely within the first 10km from the source. The sensitivity for source height and population density local to the source makes the (total) exposure increase from these substances far more uncertain than the exposure increase from long-lived substances that have most of their impact regional to the source.

The majority of industrial sources appear to have a release height around 25m and it is therefore suggested to take this as default unless a deviating release height is strongly probable. Only release heights for energy production and few other industries will typically be higher (see Table 7.2), and those for transport will be typically lower compared to the overall mean.

It is suggested to take the population density in the relevant region (see Figure 5.4) to quantify the exposure increase local to the source (from 0 to 10km). Population densities will usually be higher in build-up areas (900 person km^-2) and city-centres (21000 person km^-2). However, the location of a source in relation to built-up area is in general not known in LCA.

The number of toxic substances in a typical life cycle inventory can be considerable, while the spatially differentiated exposure factors in this chapter relate to two substances only. The exposure increase from other substances will thus in general be in between the increase from hydrogen chloride and benzene. These two substances can therefore with reasonable confidence be used to evaluate the effect on accumulated human exposure increase from spatial variation in source locations.

It is difficult to give general recommendations on whether or not a specific exposure situation encountered in LCA can be assessed to be near or above "threshold". Table 7.5 presents a crude summary of the findings of the present paper, however, that can be used for a qualitative evaluation of possible exceedance of threshold values.

This chapter does unfortunately not result in a set of factors integrating exposure and subsequent effect (from threshold exceedance), that can be used to characterise the contributions of various substances and emission situations to human toxicity in life cycle assessment. The state-of-the-art in modelling human toxicity does not allow such integrated factors, however. The material produced in this chapter can merely be used for sensitivity analysis in life cycle assessment. Hauschild and Potting (2003) describe in detail how this can be done.

The work in this chapter focuses on direct exposure and effect from atmospheric emissions. Spatial differences with regard to other exposure routes were regarded less important and not addressed (see Section 7.2.2). Huijbregts et al. (2003) assessed in a recent study the effect of using different generic environmental conditions (typical conditions in Australia, Western Europe, and the United States of America). It was found that the uncertainty in fate and exposure factors for ecosystems and humans due to choices of different sets of environmental conditions is between a factor 2 and 10. Particularly, fate and exposure factors of emissions causing effects in fresh water ecosystems and effects on human health have relatively high uncertainty. This uncertainty is mainly caused by the continental difference in the average soil erosion rate, the dimensions of the fresh water and agricultural soil compartment, and the fraction of drinking water coming from ground water. These results encourage to closer look into spatial issues of these exposure routes as well.

7.4.1 Acknowledgement

The authors like to express their gratitude to Burkhard Wickert (IER, University of Stuttgart) for processing of the population data on the Geographical Information System.

The work for this chapter was also supported by a Training & Mobility grant from the European Commission, by means of the European Commission under the JOULE Program, and under the EUREKA project LCAGAPS by means of the Danish Agency for Industry and Trade.

7.5 References

ATSDR (1997a). Toxicological Profile for Benzene (Update). Agency for Toxic Substances and Disease Registry, Public Health Service, U.S. Department of Health and Human Services.

ATSDR (1997b). Toxicological Profile for Chlorinated dibenzo-p-dioxins. (Draft for Public comment). Agency for Toxic Substances and Disease Registry, Public Health Service, U.S. Department of Health and Human Services.

ATSDR (1997c). Toxicological Profile for Lead (Draft for Public comment). Agency for Toxic Substances and Disease Registry, Public Health Service, U.S. Department of Health and Human Services.

ATSDR (1997d). Toxicological Profile for Cadmium (Draft for Public comment). Agency for Toxic Substances and Disease Registry, Public Health Service, U.S. Department of Health and Human Services.

ATSDR (1997e) Toxicological Profile for Mercury (Draft for Public comment). Agency for Toxic Substances and Disease Registry, Public Health Service, U.S. Department of Health and Human Services.

ATSDR (1998). Toxicological Profile for Sulfur dioxide (Draft for Public comment). Agency for Toxic Substances and Disease Registry, Public Health Service, U.S. Department of Health and Human Services.

Berdowski, J.J.M., W.J. Jonker, P. Verhoeve. Emissions in the Netherlands. Industrial sectors and regions in 1993 and estimates for 1994. No. 27 in publication series emission registration (In Dutch). The Hague (the Netherlands), Dutch Ministry of Housing, Spatial planning and the Environment, 1995.

Berdowski, J.J.M., J. Baas, JP.J. Bloos, A.J.H. Visschedijk and P.Y.J. Zandveld. The European Atmospheric Emission Inventory of Heavy Metals and Persistent Organic Compounds. Forchungsbericht 104 02 672/03. Umweltbundesamtes and TNO Institute of Environmental Sciences, Energy Research and Process Innovation, 1997.

Berge, E. Six-hourly data in the EMEP 150 km grid. Written communication and data on magnetic tape. Oslo (Norway), EMEP Meteorological Synthesising Centre – West, 1994.

Derwent, R.G., and K. Nodop. Long-range transport and deposition of acidic nitrogen species in north-west Europe. Nature, Vol. 324 (1986), November Issue, pp. 356-358, 1986

EC. Risk assessment of existing substances. Technical guidance document in support of the commission regulation (EC) no. 1488/94 on risk assessment for existing substances in accordance with council regulation (EEC) no. 793/93 (XI/919/94-EN). Brussels (Belgium), Directorate General Environment, Nuclear Safety and Civil Protection from the European Commission, 1994.

EC. Technical guidance document in support of the commission directive 93/67/EEC on risk assessment for new notified substances and the commission regulation (EC) no. 1488/94 on risk assessment for existing substances. Ispra (Italy), European Bureau of Chemicals, 1996.

EEA. Europe's environment. Statistical compendium for the second assessment. Copenhagen (Denmark), European Environment Agency, 1998.

EEA (1997). European Environmental Agency. Air pollution in Europe 1997. EEA Environmental Monograph No. 4. Copenhagen. ISBN 92-9167-059-6.

ETCAE (1997). Corinair 94. Summary report. Final Version. Report to the European Environment Agency from the European Topic Centre on Air Emissions.

European Council (1998). Common position (EU) No. 40/98 outlined by the council 16 June 1998 regarding limitation of the emission of volatile organic compounds from the application of organic solvent in certain activities and facilities.

Guinée, J., R. Heijungs, L. van Oers, D. van de Meent, T. Vermeire and M. Rikken. LCA impact assessment of toxic releases. Generic modelling of fate, exposure and effect for ecosystems and human beings with data for about 100 chemicals (series of publication concerning product policy, no. 1996/21). The Hague (the Netherlands), Dutch Ministry of Housing, Spatial planning and the Environment, 1996.

Harssema, H. Calculating and measuring air quality (H100-204). Wageningen (the Netherlands), Wageningen Agricultural University, 1995.

Hauschild, M.Z. and H. Wenzel: Environmental assessment of products. Vol. 2 - Scientific background, 565 pp. Chapman & Hall, United Kingdom, 1998, Kluwer Academic Publishers, Hingham, MA. USA. ISBN 0412 80810 2

Hauschild, M. and J. Potting. Spatial differentiation in life cycle impact assessment – the EDIP2003 methodology. Guideline from the Danish Environmental Protection Agency, Copenhagen, 2003.

Hertwich E.G., S.F. Mateles, W.S. Pease, T.E. McKone. Human toxicity potentials for life-cycle assessment and toxics release inventory risk screening. Environmental Toxicology and Chemistry, 20:928-939, 2001.

Huijbregts, M.A.J. Priority assessment of toxic substances in the frame of LCA. Development and application of the multi-media fate, exposure and effect model USES-LCA. Amsterdam (the Netherlands), Interfaculty dept. of Environmental Science, 1999.

Huijbregts, M.A.J., S. Lundi, T.E. McKone and D. van de Meent. Geographical scenario uncertainty in generic fate and exposure factors of toxic pollutants for life-cycle impact assessment. Chemosphere, Vol. 51, issue 6, pp. 501–508.

Hulskotte, J.H.J., G.P.J. Draaijers, J.J.M. Berdowski, W.L.M. Smeets and C.S.M. Olsthoorn. Recalculations of emissions to air for the period 1990-1996 (in Dutch). Publication series Emission Registration no. 41. The Hague (the Netherlands), Dutch Ministry of Housing, Spatial planning and the Environment, 1997.

Jensen B, Wolkoff P (1996). VOCBASE - odour and mocous membrane irritation thresholds and other physico-chemical properties. 2.1. National Institute of Occupational Health, Copenhagen.

Jensen B, Wolkoff P (1997). VOCBASE – a database of odour thresholds, irritation of mucous membranes and chemical parameters for volatile organic compounds (in Danish). Dansk Kemi; 5: 16-17, 20-26.

Krewitt, W., P. Mayerhofer, R. Friedrich, A. Trukenmüller, T. Heck, A. Greßmann, F. Raptis, F. Kaspar, J. Sachau, K. Rennings, J. Diekmann, B. Praetorius. ExternE – Externalities of energy. National implementation in Germany (EUR 18271). Directorate-General XII for Science, Research and Development of the European Commission, 1997.

Krüger, O., J.P. Tuovinen. The effect of variable sub-grid deposition factors on the results of the Lagrangian long-range transport model of EMEP. Atmospheric Environment, Vol. 31 (1997), Issue 24, pp.4199-4209.

Metcalfe, S.E., J.D. Whyatt, R.G. Derwent. A comparison of model and observed network estimates of sulphur deposition across Great Britain for 1990 and its likely source attribution. Quarterly Journal of the Royal Meteorological Society, Vol. 121 (1995), Issue 526 (July), Part B, pp.1387-1411.

NYSDEC (1995). Ambient air monitoring report for volatile organic compounds. Division of air resources - Bureau of Air Quality Surveillance. Summary of Toxic Monitoring Data from 1990 to 1993. New York State Department of Environmental Conservation. Reports can be obtained from: Bureau Director - BAQS, 50 Wolf Road, Albany, NY 12233-3256. Information in this paper has been retrieved from the Internet: http://www.dec.ny.us/website/pollution/voc_rpt.html.

Potting J, Hauschild M (1997). Predicted environmental impact and expected occurrence of actual environmental impact. Part II: Spatial differentiation in life-cycle assessment via the site-dependent characterisation of environmental impact from emissions. Int. J. LCA, 4: 209-16.

Potting, J., M. Hauschild, and H. Wenzel. "Less is better"and "only above threshold"; Two incompatible paradigms for human toxicity in life-cycle assessment? International Journal of Life Cycle Assessment, Vol. 4 (1999), Issue 1, pp.16-24.

Potting, J. A. Trukenmüller, H. van Jaarsveld and M. Hauschild. Human exposure from air emissions. In: Potting, J. Spatial differentiation in life cycle impact assessment. Ph.D. thesis. Utrecht (the Netherlands), 2000.

Potting, J. Spatial differentiation in life cycle impact assessment. Ph.D. thesis. Utrecht (the Netherlands), 2000.

Schmidt, A., Rasmussen, P. B., Andreasen, J., Fløe, T. and K. E. Poulsen. Integration of the working environmennt in life cycle assessments (LCA) – a new methodology. Guidelines from the Danish Environmental Protection Agency No. xxx 2000, Copenhagen (2000)

Seinfeld, J.H. and S.N. Pandis. Atmospheric chemistry and physics: From air pollution to climate change. New York (United States of America), John Wiley & Sons, 1998.

Stanners, D. and P. Bourdeau. Europe's environment. The dobris assessment. Copenhagen (Denmark), European Environment Agency, 1995.

Tobler, W., U. Deichmann, J. Gottsegen and K. Maloy. The Global Demography Project (technical Report TR-95-6 [online]). Santa Barbara (United States of America), dept. of Geography, University of California of the National Centre for Geographic Information and Analysis, 1995 [cited 11 August 1999].

Available from Internet http://www.ciesin.org/datasets/gpw/globldem.doc.html.

Toppari, J., J.C.Larsen, P. Christiansen, A. Giwercman, P. Grandjean, L.J. Guillette, B. Jégou, T.K. Jensen, P. Jouannet, N. Keiding, H. Leffers, J.A. McLachlan, O. Meyer, J. Müller, E. Rajpert-DeMeyts, T. Scheike, R. Shapre, J. Sumpter and N.E. Skakkebæk. Male reproductive health and environmental chemicals with estrogenic effects (Miljøprojekt no. 290). Danish Environmental Protection Agency, Copenhagen, 1995.

Trukenmüller, A. Using WMI: Implementing atmospheric trajectory models with the Wind rose Model Interpreter. Stuttgart (Germany), IER of the University of Stuttgart, 1998.

van der Ven B (1995). Default values. A discussion paper for the meeting of the SETAC workgroup "Data handling" on Sep. 4th 1995 at Gothenborg.

Van Jaarsveld, H. An operational atmospheric transport model for priority substances; specification and instructions for use. Report number 222501002. Bilthoven (the Netherlands), RIVM, 1990.

Van Jaarsveld, J.A. Modelling the long-term atmospheric behaviour of pollutants on various spatial scales. Utrecht (the Netherlands), PhD thesis, Utrecht University, 1995.

Van Jaarsveld J. A., W. A. J. van Pul and F. A. A. M de Leeuw. Modelling transport and deposition of persistent organic pollutants in the European region. Atmospheric Environment Vol. 1997, Issue 31, pp. 1011-1024

Van Jaarsveld, J.A. and F.A.A.M de Leeuw. OPS: An operational atmospheric transport model for priority substances. Environmental Software, Vol. 1993, Issue 8, pp. 91-100.

Van Leeuwen and J.L.M. Hermens. Risk assessment of chemicals: An introduction (ISBN 0-7923-3740-9). Dordrecht (the Netherlands), Kluwer Academic Publishers, 1995/1996.

Wenzel, H., Hauschild M.Z. and Alting, L.: Environmental assessment of products. Vol. 1 - Methodology, tools, techniques and case studies, 544 pp. Chapman & Hall, United Kingdom, ISBN 0 412 80800 5, Kluwer Academic Publishers, Hingham, MA. USA., 1997.

Wolkoff P, Clausen P A, Nielsen G D (1998). Volatile organic compounds (VOCS) in the indoor environment. State-of-the-science report. AMI Dokumentation 2. Arbejdsmiljøinstituttet, Copenhagen. In Danish.

Wolkoff P, Nielsen P A (1993). Statens Byggeforskningsinstitut. Indeklimamærkning af byggevarer. Del 2: Faglig og teknisk dokumentation af en prototypeordning. SBI-rapport 233. Statens Byggeforskningsinstitut. In Danish.

WHO/UNEP (1992). Urban air pollution in Megacities of the World. World Health Organisation, United Nation Environment Programme, Blackwell, Oxford.

WHO (1992). Environmental Health Criteria 134. Cadmium. International Programme on Chemical Safety (IPCS), World Health Organisation, Geneva.

WHO (1993). Environmental Health Criteria 150. Benzene. International Programme on Chemical Safety (IPCS), World Health Organisation, Geneva.

WHO (1994a). Environmental Health Criteria 163. Chloroform. International Programme on Chemical Safety (IPCS), World Health Organisation, Geneva.

WHO (1994b). Environmental Health Criteria 170. Assessing human health risks of chemicals: Derivation of guidance values for health-based exposure limits. International Programme on Chemical Safety (IPCS), World Health Organisation, Geneva.

WHO (1995). Environmental Health Criteria 165. Lead. International Programme on Chemical Safety (IPCS), World Health Organisation, Geneva.

WHO (1997a). Environmental Health Criteria 188. Nitrogen Oxides (second edition). International programme on chemical safety (IPCS), World Health Organisation, Geneva.

WHO (1997b). Environmental Health Criteria 195. Hexachlorbenzene. International Programme on Chemical Safety (IPCS), World Health Organisation, Geneva.

WHO (1998). Air Quality Guidelines for Europe, World Health Organisation, Regional Office for Europe, Copenhagen, WHO Regional Publications, European Series. (NB! Information for this paper has been retrieved via the Internet addresses: http://www.who.dk and http://www.who.org/peh/specprg.htm.)

Williams, M. (ed.). Exposure assessment. Report number 1. Cost 613/2 Report series on air pollution epidemiology. Directorate-General XII for Science, Research and Development of the Commission of the European Communities.

Annex 7.1 Characterisation factors for human toxicity assessment from emissions to air

Annex 7.2 Characterisation factors for human toxicity assessment from emissions to water

Annex 7.3 Characterisation factors for human toxicity assessment from emissions to soil