Strategic environmental assessment methodologies—applications within the energy sector

Environmental Impact Assessment Review 23 (2003) 91 – 123 www.elsevier.com/locate/eiar

Strategic environmental assessment methodologies—applications within the energy sector Go¨ran Finnvedena,*, Ma˚ns Nilssonb, Jessica Johanssona, ˚ sa Perssonb, A ˚ sa Moberga,c, Tomas Carlssond A a

Environmental Strategies Research Group (fms), Swedish Defence Research Agency, PO Box 2142, SE 103 14 Stockholm, Sweden b Stockholm Environment Institute (SEI), P.O. Box 2142, SE 103 14 Stockholm, Sweden c Environmental Strategies Research Group (fms), Department of Systems Ecology, Stockholm University, P.O. Box 2142, 103 14 Stockholm, Sweden d Department of Energetic Materials, Swedish Defence Research Agency, SE 147 25 Tumba, Sweden Received 1 February 2002; received in revised form 1 August 2002; accepted 1 September 2002

Abstract Strategic Environmental Assessment (SEA) is a procedural tool and within the framework of SEA, several different types of analytical tools can be used in the assessment. Several analytical tools are presented and their relation to SEA is discussed including methods for future studies, Life Cycle Assessment, Risk Assessment, Economic Valuation and Multi-Attribute Approaches. A framework for the integration of some analytical tools in the SEA process is suggested. It is noted that the available analytical tools primarily cover some types of environmental impacts related to emissions of pollutants. Tools covering impacts on ecosystems and landscapes are more limited. The relation between application and choice of analytical tools is discussed. It is suggested that SEAs used to support a choice between different alternatives require more quantitative methods, whereas SEAs used to identify critical aspects and suggest mitigation strategies can suffice with more qualitative methods. The possible and desired degree of sitespecificity in the assessment can also influence the choice of methods. It is also suggested that values and world views can be of importance for judging whether different types of tools and results are meaningful and useful. Since values and world views differ between

* Corresponding author. Tel.: +46-8-402-38-27; fax: +46-8-402-38-01. E-mail address: [email protected] (G. Finnveden). 0195-9255/02/$ – see front matter D 2002 Elsevier Science Inc. All rights reserved. PII: S 0 1 9 5 - 9 2 5 5 ( 0 2 ) 0 0 0 8 9 - 6

92

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

different stakeholders, consultation and understanding are important to ensure credibility and relevance. D 2002 Elsevier Science Inc. All rights reserved. Keywords: Strategic environmental assessment; Applications; Energy sector

1. Introduction 1.1. Background The main purpose of strategic environmental assessment (SEA) is to facilitate early and systematic consideration of potential environmental impacts in strategic decision-making (Therivel and Partidario, 1996; Partidario, 1999). It is intended to be used on policies, plans and programmes. The growing significance of SEA as a form of support to decision-making is manifested by the recent EC directive (2001/42/EC) on the assessment of environmental effects from certain plans and programmes (Feldmann et al., 2001). However, a number of challenges need to be overcome for SEA to be an effective tool. In order to be effective, a number of criteria need to be met. The International Association for Impact Assessment (IAIA) has published the IAIA principles that stipulate best practice for EIA (IAIA, 1999). The principles are: rigorous, practical, relevant, cost-effective, efficient, focused, adaptive, participative, interdisciplinary, credible, integrated, transparent, and systematic. While established for EIA, they are of key relevance also for SEA and in a workshop hosted by the Federal Ministry for the Environment Nature Conservation and Nuclear Safety (2001), these principles were adapted towards SEA. A number of publications have been concerned with how to design an SEA process that can be integrated with the decision-making process (e.g. European Commission, 1994; Therivel and Brown, 1999; Naturva˚rdsverket, 2000; ANSEA Project, 2002). Slightly different steps are defined in different sources, although the main features remain the same. The following steps are identified here, also based on (Nilsson et al., 2001): 1. 2. 3. 4.

Definition of objectives. Formulation of alternatives. Scenario analysis. Environmental analysis (including the use of objective and acceptable aggregated indicators, based on more traditional natural sciences). 5. Valuation (including the use of controversial aggregation methods, and political and ethical values). 6. Conclusions, review of quality/follow up measures, etc.

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

93

In addition, consultation and public participation are important aspects of the SEA process and should take place at several occasions in the process. Similarly, careful analysis of the uncertainties involved in the assessment, through methods such as sensitivity analysis should be applied throughout various stages of the SEA. A legal basis for undertaking SEA will be developed in Sweden within the next few years when the EC directive is incorporated into national legislation. The directive is, however, limited in scope. It applies to plans and programmes, and modifications of them, that are subject to preparation and/or adoption by an authority at national, regional and local level, and that are required by legislative, regulatory or administrative provisions (Article 2). Thus, it does not apply to policies, to the private sector or to plans and programmes that are not formally required. It is currently not clear which, if any, applications within the energy sectors will require an SEA according to the directive (Adolfsson, Swedish EPA, personal communication) and there is also a lack of methodological guidelines for this application. SEAs can be useful and effective for a number of applications where SEAs are not formally required. For example, SEA could be a useful tool for energy policies developed at national level (Nilsson et al., 2001; Noble and Storey, 2001). The national energy policy programmes established by the Swedish Parliament are normally based on major impact analyses prepared by the Government Commissions (for example SOU, 1995, p. 139, preceding the 1997 policy programme). However, some decisions are also taken in between these major programmes. At the local level the Swedish municipalities are required by law to carry out energy plans (SFS, 1977, p. 439) as a means to promote efficient use of energy and take action for a secure and sufficient energy supply. The plan shall concern the supply, distribution and use of energy within the municipality and it shall contain an analysis of what impacts these activities will have. The municipal energy plans are not legally binding, however. On a local level, SEA could also be useful for other purposes. The municipalities still have many important roles in the energy arena, for example as owners of energy companies and a large real estate stock, as environmental and planning authorities and as providers of information to the public. Within companies, SEA can support internal decision-making in several ways, including to minimise future environmentally related risk and associated economic costs and to gain competitive advantage. There is currently a lack of methods for this purpose (Bardouille, 2001) and SEA can possibly be useful within this context. Another possible application area is to use results from SEA as arguments in a public debate. A third application area can be to use a non-site-specific SEA in a tiered approach with a site-specific project EIA at a lower level. Also NGOs can use SEAs to develop and support sustainability arguments and positions in their campaigns and other activities. However, above all it should be used to improve decisions towards sustainable solutions.

94

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

SEA can be viewed as consisting of three components: institutional arrangements, procedure, and methods (Kørnøv, personal communication). While much of the SEA literature is focused on issues surrounding institutional arrangements (ANSEA Project, 2002) and process aspects (see, for instance, Therivel and Partidario, 1996), challenges related to what methods and analytical tools to use in SEA remain and need more attention. Noble and Storey (2001) address this question by developing a framework focused on multi-criteria decision-making. Questions remain however, when it comes to the environmental analysis that can support those methods. As a procedural tool, SEA can include a number of different analytical tools (Wrisberg et al., 2000). (The focus of procedural tools is on procedures to guide the process to reach and implement environmental decisions whereas analytical tools are modelling the system in a quantitative or qualitative way aiming at providing technical information for a better decision (ibid.). In this paper we use the words ‘‘tools’’ and ‘‘methods’’ as synonyms.) However, appropriate methods need to be established. For instance, SEA guidance often refers to Environmental Impact Assessment (EIA)-type analyses but it is often difficult to use the methods associated with project EIA in SEA because they are adjusted for site-specific information and local impacts whereas SEA often is not site-specific and can often be primarily concerned with cumulative and indirect impacts (e.g. Petts, 1999b). The lack of methodological guidance for SEA also acts as a barrier to the implementation of SEA in general and the European directive on SEA (European Parliament, 2001) in particular. 1.2. Aim The aim of this study is to examine how various analytical tools can be used within the SEA process, especially in the following steps: Scenario Analysis, Environmental Analysis and Valuation. Examples include Economic Valuation methods, Life Cycle Assessment (LCA) and Risk Assessment (RA) (see also Petts, 1999a). Relations between SEA applications and choice of methods are also discussed. The applications of particular interest in this study are within the energy sector in Sweden. The results and the discussion are however of relevance also for other applications. It should be noted that the study is not comprehensive in scope in so far as not all possible tools are discussed. It goes through a limited set of quantitative tools. There are, in addition to this, several quantitative tools that have not been discussed. There are also several qualitative tools that could be useful in an SEA context, such as various types of group decision making and preferenceindifference models (Noble and Storey, 2001). The tools that are discussed have been selected because they are established methodologies that are well-known, they have been developed, tested and applied in the energy sector, at policy, plan, and programme levels, and there is data available from these tools for testing the methods in a pilot study.

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

95

2. Describing methods A large number of methods and tools for assessing environmental impacts are available (Baumann and Cowell, 1999; Moberg et al., 1999; Petts, 1999b; Wrisberg et al., 2000). We need to characterise different methods in order to better understand their interrelationships and the appropriateness of different methods in different applications. In this section some frameworks for describing methods and their relations are presented. Firstly, there are several analytical features of the methods that need to be considered when several methods are used in combination (Finnveden and Moberg, submitted for publication): 





 



Degree of site-specificity. Some methods are generally site-specific (e.g. local air quality models), whereas others are generally site-independent (e.g. traditional Life Cycle Assessment methodology further described below). Between these two extremes, there may be a continuum with different types of site-dependency (Potting, 2000). Degree of time-specificity. In parallel to site-specificity, we can distinguish between time-specific, time-independent and different types of time-dependency (Potting, 2000). Type of comparison. Most methods include some sort of comparison, either between different alternatives, or within a studied system or against a reference. Degree of quantification. System boundaries, which are largely determined by the object of study. Distinctions can be made between different types of objects, for example: chemical substances, products and functions, companies and organisations, nations and regions, sectors, projects, and policies, plans and programmes (Finnveden and Moberg, submitted for publication). Types of impacts and effects considered.

Differences between methods with regard to these aspects can determine if and how different methods can be used in the context of an SEA. Secondly, the information produced by different methods should be classified. A useful categorisation developed for environmental and sustainable development indicators is the DPSIR (Driving forces, Pressure, State, Impact, Response) model. This model represents a systems analysis view of the interaction between the human and environmental systems (Smeets and Weterings, 1999). Example of indicators from the energy sector in this model are: 

Driving force-space used in residential living. Pressure-emissions from electricity production.  State-ambient concentrations of pollutants. 

96  

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

Impact-cases of cancer due to pollutants. Response-fraction of renewable fuels.

(In this paper, environmental changes or effects are used as encompassing terms covering changes of pressure, state and impacts.) The following sections will introduce and discuss the methods in relation to these criteria.

3. Methods 3.1. Future studies Since SEA is a process for facilitating and improving strategic decisionmaking processes much of what is handled concerns the future. Therefore it is often necessary to include some idea of the future in SEA. Future scenarios are often used in SEAs (see e.g. case studies in Therivel and Partidaro, 1996; Noble, 2000; Noble and Storey, 2001). However, much less is written on the appropriate methods for future studies within SEA. Scenario analysis is mentioned in Dom (1999) as an often-used tool for evaluating future impacts within SEA, in Naturva˚rdsverket (2000) scenario analysis is used as an example of methods for SEA and the use of forecast and backcasts (defined below) in SEA is discussed by Noble (2000). There are several different approaches for studying the future (Dreborg, 2001). Forecasts try to indicate a probable future and they are often based on trends and mechanisms that can be seen in past years. Such trends and mechanisms are then, more or less directly, extrapolated into the future-giving a forecast. As development and change are a constant part of society, reliable forecasts are useful mainly for the shorter term and for well-defined areas (Ho¨jer and Mattsson, 2000; Rescher, 1998; Makridakis et al., 1998). Conditional forecasts are based partly on extrapolations of historical data or mechanisms and partly on assumptions or results from scenarios. Conditional forecasts can then result in a number of scenarios based on some conditional assumptions. Scenarios are helpful when there is a significant qualitative uncertainty about the future. External scenarios means that the studied scenarios are dependent on factors which cannot be controlled by the user of the scenarios, but which are still relevant for the same. Scenarios present possible futures. External scenarios may be used in scenario planning to find strategies that are robust across a range of possible futures (see e.g. van der Heijden, 1996). If the future is considered to be possible to influence in a significant way by the user of the scenarios, policy scenarios may be suitable to use. In this case the user will act, not only react. In back-casting the future goal is first set and then images of what this preferable future could be/look like is made. Then in a second step, the ways to get there are described. For a more extensive presentation of back-casting, see e.g. Dreborg (1996).

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

97

Modelling can be used to support the different approaches to future studies. Modelling is often based on assumptions made from historical trends, but new mechanisms may also be used. An advantage of modelling is that quantitative information on, for example, future energy use and fuel mix can often be produced. A disadvantage with complex models is the possible loss of transparency. There is also a risk that models based on assumptions and mechanisms that describe the current situation are used for long-term future studies, where these assumptions and mechanisms cannot be presumed. Dynamic modelling is frequently used in energy systems studies. One example of such models is MARKAL which is a macro-level model. It has for example been used to study the effects of increased cooperation and cross-border energy-related trade in the Nordic countries (Unger, 2000). Another example is MARTES, a model used for local district heating systems (Olofsson, 2001). 3.2. Life cycle assessment Life Cycle Assessment (LCA) is a tool to assess the environmental impacts and resources used throughout a product’s life from raw material acquisition through production use and disposal. An ISO standard has been developed for LCA providing a framework, terminology and some methodological choices (ISO, 1997, 1998, 1999). Initiatives have also been taken to develop best available practice (Udo de Haes et al., 1999a,b, in press). The basis for the calculations is the functional unit to which all inputs and outputs are related. An example of a functional unit is 1 kW h electricity or 1 MJ heat. When different alternatives are compared, the functional unit is the basis for the comparison. According to the ISO-standard, an LCA is divided into four phases: 1. Goal and scope definition. 2. Inventory analysis, where inputs and outputs to and from the systems are identified and quantified. 3. Life cycle impact assessment (LCIA), aimed at understanding and evaluating the magnitude and significance of the potential environmental impacts of a product system. This phase is further divided into three mandatory elements. 3.1. Selection of impact categories, indicators for the categories and models to quantify the contributions of different inputs and emissions to the impact categories. 3.2. Assignment of the inventory data to the impact categories (classification). 3.3. Quantification of the contributions from the product system to the chosen impact categories (characterisation). 4. Interpretation, where the findings of either the inventory analysis or both the inventory analysis and the life cycle impact assessment phases are combined in line with the defined goal and scope of the study.

98

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

In principle, LCA is a comprehensive environmental assessment. In practice, not all types of environmental effects are equally well covered (Finnveden, 2000). Effects associated with land use are traditionally difficult to assess, although there has been a considerable methodological development during the last years (Lindeijer et al., in press). Toxicological effects are often only included with data gaps. Effects associated with radiation, accidents and disamenities are typically not covered at all. The impacts typically best covered in a traditional LCA are environmental impacts from emissions to air, such as global warming and acidification, and use of energy resources. LCA is traditionally a site- and time-independent tool. In a traditional LCA, no consideration is given to when and where emissions are taking place (Udo de Haes, 1996) and characterisation factors used for quantifying the contribution to different impact categories do not have any site-dependent information resulting in a site-generic result. This is mainly for two reasons. The first is a practical reason; it is not practically possible to gather site-specific information for all places included in an LCA. The second is more theoretical. In an LCA, not all emissions are considered. Only the emissions that are allocated to the functional unit are considered. There is however a trend towards making LCA more sitedependent if not site-specific (Huijbregts, 2001; Krewitt et al., 2001; Nigge, 2001a,b; Potting, 2000; Spadaro and Rabl, 1999). Introducing some typical environments and emissions situations does this. For example, emissions at low height in an urban environment are differentiated from emissions at high elevation in rural areas. Impact assessment factors are then calculated for these typical environments. Site-dependent characterisation factors can thus be calculated and used, resulting in a site-dependent characterisation. Besides the mandatory elements within LCIA, there are also some elements described as optional (ISO, 1999), e.g. weighting that aims at converting and possibly aggregating indicator results across impact categories resulting in a single result. Methods for weighting include both economic valuation methods and multi-attribute approaches, further discussed below (Finnveden et al., 2001). Within the DPSIR-framework for indicators discussed above, LCA can produce information on pressure and impact levels. Results on emissions are on a pressure-level and depending on how far in the environmental mechanism the impact assessment modelling goes, the resulting indicators may be described as pressure- or impact-indicators. LCA does however not describe a state. This is because only the fraction of the emissions which are related to the functional unit is considered, not the whole picture. In order to calculate a state, the whole picture including background emissions needs to be considered which is not done in an LCA. 3.3. Environmentally extended input/output analysis Input –Output Analysis (IOA) is a well-established analytical tool within economics and systems of national accounts (Miller and Blair, 1985) using a

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

99

nation or a region as the object of the study. The input – output matrices describe trade between industries. By performing an input/output analysis a calculation of the sectors or industries involved in the production of a product or service going to final demand can be calculated. IOA can be applied to include environmental impacts by adding emissions coefficients to the monetary IOAs (Lave et al., 1995; Joshi, 2000). IOA are typically applied using retrospective data and methodology for accounting purposes. Environmentally extended IOA is a quantitative method. It presents results for broadly defined sectors or products groups within a nation or a region. It is siteindependent within the nation or region. The emissions most often included are the traditional air pollutants, but in some applications more pollutants have been included giving results similar to LCAs. 3.4. Risk assessment of chemicals and accidents Risk assessment is a broad term covering many different types of assessments. Already the word ‘‘risk’’ is problematic (Hansson, 1999). Here, a distinction is made between risk assessment of chemical substances and risk assessment of accidents. The latter may include environmental aspects. Risk assessment of accidents concerns unplanned incidents, e.g. explosions or fires. This is typically in contrast to risk assessment of chemicals, where dispersion of chemicals is often planned and forms part of its use. Methods and protocols for risk assessment of chemicals have been developed in several international fora, e.g. EU and OECD (Eduljee, 1999). In the risk assessment of chemicals, an exposure assessment including a description of the nature and size of exposed targets, as well as magnitude and duration of exposure, is combined with an effect assessment (Eduljee, 1999; KemI, 1995; Olsen et al., 2001). The exposure assessment is done using some kind of model. Broadly speaking, two classes of models can be distinguished for toxic chemicals (Hertwich et al., in press): multi-media fate and exposure models that take into account the fate of pollutants across medium boundaries and model multipathway exposure routes, and spatially explicit single-medium models which can take into account dispersion and reactions. Multi-media models are for example used in the EUSES, the software which in practice is used for carrying out risk assessment of chemicals within the European regulation (Olsen et al., 2001). A single-medium model, Ecosense, is further described below in the section on the Impact Pathway Approach. In accident risk assessment, accident consequences and their frequency are estimated. The assessment is usually divided into three parts: hazard identification, consequence analysis and frequency estimation. For the hazard identification various methodologies have been developed to aid system experts to identify hazards in their expert domain, e.g. HAZOP (among others reviewed by Khan and Abbasi, 1998) and Safety Function Analysis (Harms-Ringdal, 2002). For the consequence analysis, several methods to calculate consequences due to explo-

100

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

sions, fires and releases of toxic chemicals are available (AIChemE, 1998). For the frequency estimation, historical records, fault-trees and event-trees are used. Risk assessment of accidents is typically done prospectively for different types of projects (Ra¨ddningsverket, 2000), and it is typically site-specific. Risk assessment of chemical substances can be site-specific but also more siteindependent for a region or a nation. It typically includes all emissions of the substance within the geographical boundary or from a particular project or plant. Comparisons can either be made between different alternatives (which alternative poses the greatest risk?) or against a standard (is the risk acceptable or not?). Comparisons can also be made internally within a system to identify the greatest risk. Risk assessment of chemicals is typically done quantitatively. Risk assessment of accidents can be made both quantitatively or qualitatively. 3.5. Impact pathway approach The Impact Pathway Approach (IPA) can be regarded as a special case of a risk assessment approach that is of particular importance for the environmental assessment of different energy systems. In the IPA the analytical sequence ‘economic activities>emissions>dispersion>concentrations>dose>impact’ is handled systematically. The ExternE project (Commission for the European Communities, 1995, 1999) and the URBAIR project (Jitendra et al., 1997) provides a methodology. Fig. 1 outlines the general stages in the impact pathway approach. Emissions factors for various energy systems are used to model the pollution load. Ambient concentrations depend not only on emissions but on background concentrations and topography and meteorology. Exposure assessment is of particular importance when studying health impacts. It serves to estimate how large a share of the population is exposed to different concentrations of pollutants. Impacts can be divided into impacts on health, materials (corrosion), ecological systems, forests, and agriculture (Commission for the European Communities, 1999). The impact assessment is based on combining information on the exposed receptor population and the concentration with dose – response

Fig. 1. The impact pathway approach.

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

101

relationships for various impacts and pollutants. The theoretical and statistical underpinnings of dose – response relationships are described in Calthrop and Maddison (1996) and health impacts of air pollution for the most common pollutants are thoroughly documented in the literature (Ostro, 1994; Commission for the European Communities, 1995; Dockery and Pope, 1994; Dockery et al., 1993). IPA is a site- and time-specific approach. It can however also be used for more generic applications by performing calculations for typical conditions. Also by applying typical conditions, site-dependent characterisation factors can be calculated which could be used in LCA as discussed above. IPA is a quantitative method that can be used for comparisons, either between different alternatives or within a studied system. Calculated ambient concentrations can also be compared to reference values. IPA is data demanding and therefore mostly applicable to conventional air pollutants (such as particulates, SOx, NOx, ozone, and CO). Within the DPSIR framework, IPA can be used to calculate indicators on Pressure, State and Impact levels. 3.6. Ecological impact assessment For ecological impact assessment we have not found any clear-cut tools that would be obvious to use within an SEA for the energy sector. Tools for assessing ecological impacts are currently being developed within the context of the Convention on Biodiversity (SBSTTA, 2001). Some guidance on how to deal with nature conservation in SEA is provided by Therivel and Thompson (1996). For impacts occurring at specified sites methods may be adopted from project EIA (see e.g. Wathern, 1999). In SEA the areas affected will typically be larger and the detailed assessment methods may have to be adjusted to a coarser resolution. In SEA more emphasis may be put on landscape ecological issues. For many impacts in SEA for the energy sector the specific site will not be known and here the assessment will have to rely on some sort of classification of affected landscape types and a general estimation of what effect a certain activity will have in different ecosystems. A promising approach is to use indicator species (see e.g. Treweek et al., 1998; Mo¨rtberg and Balfors, 2000; Dı´az et al., 2001). The issue of assessing impacts resulting from land use in a non-site specific context has also been dealt with within the field of LCA. For a recent review of the methods proposed, see Lindeijer et al. (in press). Weidema and Lindeijer (2001) suggest a method for quantified assessment of the physical impacts of land use in terms of indicators for biogeochemical substance and energy cycles, ecosystem productivity, biodiversity, cultural value and migration and dispersal. The data presented for the method are at biome level and the results from using it would accordingly be very crude. Supplying more refined data for the areas where most of the impacts occur could enhance the method. One problem that this method displays is the difficulty in determining how different

102

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

dimensions of land use impacts can be aggregated. This may be a common problem to all quantified ecological assessments as long as more than one indicator is chosen. 3.7. Multiple attribute analysis Multiple attribute analysis (MAA) (also referred to as multicriteria, multiobjective or multiattribute trade-off or utility analysis in the literature) aims to improve decision-making by making choices about conflicting or multiple objectives explicit, rational and efficient. It can help to structure the decisionprocess, display trade-offs among criteria, help people apply value judgments concerning trade-offs, help people make consistent evaluations of risk and uncertainty, and facilitate negotiation (Hobbs and Meier, 2000). The application of MAA usually follows a sequence of steps and hence shows many similarities with the SEA as a whole. It includes pre-decision issues such as problem definition, selection of attributes, alternatives selection, and quantification of impacts to various attributes. These steps emphasize the generation of trade-off information. These elements of MAA and experiences gained in the field can be useful in several stages of the SEA. Later stages are more specific for MAA, such as the trade-off analysis and evaluation steps that lead to the actual choices and decisions (Hobbs and Meier, 2000) and as such match closely with the valuation stage of the typical SEA. Decision research has accumulated a wide array of methods and concepts for MAA, including, for instance, ‘Analytical Hierarchy Process’ (Saaty and Vargas, 1994) and ‘Value-Focused Thinking’ (Keeney, 1992). The choice of MAA method is in itself a multiple attribute problem (Patton and Sawicki, 1993). One has to decide on the use of weights, rating systems and aggregation measures. The choice of attributes, or criteria, is a crucial aspect and also a difficulty in the approach that has great importance in the SEA method. As Keeney (1992) points out, the key to a successful model is the sets of objectives and the attributes that measure them. Most decision literature pays significant attention to this problem (Miser and Quade, 1985; Keeney, 1992; Vries, 1999). A key question with particular importance in the context of aggregation regards the weighting or non-weighting of criteria (Vries, 1999). People and decision-makers often value certain attributes more than others. The introduction of weights to criteria recognizes the relative importance of different criteria. Weights can be arrived at through observer-derived indirect methods, based on observed choices, and direct methods, where decision-makers are asked directly to assign numerical values. (Keeney and Raiffa, 1976; Hobbs and Meier, 2000). Matrix display systems are important for the MAA problem and representation. The common approach in such a system is to indicate the criteria on one axis (such as environmental quality objectives) and the alternatives on the other (such as different policy options) and create a scorecard or goals achievement matrix.

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

103

This is not just applicable in expert analysis. It has also been applied in stakeholder participation settings by, for instance, Bots and Hulshof (2000), in which an impact matrix is established as a first stage in the group decisionmaking process. This particular work was done in the health sector, but the approach could just as well be applied in the energy sector. The matrices can display data at different levels of aggregation. In value-laden public decisionmaking that is typical for SEA processes, a more disaggregated approach tends to be more acceptable and useful (Patton and Sawicki, 1993). In the SEA context matrix display systems can help in avoiding a need for producing single summary values or indices that are supposed to capture different environmental dimensions. Aside from matrix display systems, multiple attributes can be displayed in a diamond model (Nilsson, 1997), or in a value path (Hobbs and Meier, 2000). MAA is a very flexible family of methods. It can be applied for all kinds of impacts, be made site-/time-specific or not, and quantitatively as well as qualitatively. The relationship to the criteria in terms of pros and cons of the alternatives might not be possible to quantify, but the assessor or decision-maker would still be able to rank the alternatives in terms of achievement or in other ways display the best or worst alternative and hence provide a graspable imagery of the results of the MAA. One should be aware that there can be significant disagreement between methods as well as disagreements among individuals and one would normally expect great differences in results from the MAA. However, the purpose of the MAA is not to come up with one answer but also to be a learning process that exposes interest groups to different views, and forces people to think about the problems at hand (Hobbs and Meier, 2000). 3.8. Environmental objectives An SEA should be conducted as a means to achieve environmental and other objectives through the PPP proposal. Keeney (1992) suggests that objectives, which essentially articulate values, can be identified for example by using existing sets of strategic and generic objectives. In Sweden, a set of 15 national environmental objectives have been established by Parliament (Table 1) as guiding principles for environmental policy (Government Bill, 2000). In the relatively few regional and sectoral SEAs that have been undertaken to date in Sweden, the use of the national objectives has in some way been a common feature (Boverket och Naturva˚rdsverket, 2000). Although plans and policies in the energy sector have a more profound influence on some of these objectives, all of them can be affected by different energy policies. The 15 national objectives are quality objectives that specify the conditions, or states, which actions should be directed towards and the general time frame is one generation (Government Bill, 2000). Since they are of a qualitative rather than quantitative character, a valuation against these could only give indications whether the PPP would contribute positively or negatively towards achievement, and possibly the strength of that contribution (in qualitative terms).

104

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

Table 1 National environmental objectives in Sweden 1. Reduced climate impact 2. Clean air 3. Natural acidification only 4. A non-toxic environment 5. A protective ozone layer 6. A safe radiation environment 7. Zero eutrophication 8. Flourishing lakes and streams 9. Good quality groundwater 10. A balanced marine environment, flourishing coastal areas and archipelagos 11. Thriving wetlands 12. Healthy forests 13. A varied agricultural landscape 14. A magnificent mountain landscape 15. A good built environment Source: Government Bill (2000), The Swedish Environmental Objectives—Interim Targets and Action Strategies.

The Government has also proposed a more detailed framework with interim targets and action strategies for each objective (e.g. Government Bill, 2000). While the objectives specify desirable states only, the targets concern pressures, states, impacts and responses. Targets for environmental pressures and responses are by far the most common ones. Examples of targets for some of the objectives are presented in Table 2. It can be noted that there is a mixture of quantitative and qualitative targets. In general, targets related to objectives 1 – 6 are more quantitative. Having accepted the objectives as guiding principles or criteria for the SEA, they can be used in several places in the process: 1. Definition of objectives in the PPP—the objectives establish the general environmental standards which all PPPs should be directed towards. 2. Identification of impacts—the objectives can be used as a checklist when doing an inventory of potential impacts (see e.g. Nilsson et al., 2001). 3. Choice of indicators—several sets of indicators have been developed specifically for or with the environmental objectives in mind, and these can provide baseline data and measurement methodologies (Government Bill, 2000, pp. 223– 227; SOU, 2000, pp. 723 – 742). 4. Valuation—assuming that the parliamentary established objectives represent public values, they provide a comprehensive and solid base for valuation. Having chosen a set of objectives and/or targets and assembled the required information, the actual valuation, in terms of measuring attributes, can then take place in a simple matrix with more or less space for comments. In the matrix, the objectives are on one axis and the different options on the other. Valuation results

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

105

Table 2 Some environmental objectives and targets of relevance for the energy sector (examples only) Objectives

Interim targets

Clean air

2005: annual average of 5 mg SO2 m 3 in municipalities; 2010: annual average of 20 mg NO2 m 3 and hourly average of 100 mg NO2 m 3 in most places; 2010: ground-level ozone not exceeding 120 mg m 3 as an 8-h average; 2010: national emission of VOCs, excluding CH4, reduced to 241,000 tons 2010: not more than 5% of all lakes and 15% of the total length of running water affected by anthropogenic acidification; 2010: reversed trend of increased acidification of forests and recovery under way; 2010: national atmospheric emission of SO2 reduced to 60,000 tons; 2010: national atmospheric emission of NOx reduced to 148,000 tons 2010: spatial and community planning based on programmes and strategies for promoting more efficient energy use, the use of renewable energy sources and the development of production plants for district heating, solar energy, bio fuels and wind power; 2010: environmental impact made by energy use in residential and commercial buildings decreasing and lower than in 1995, through improving energy efficiency and eventually reducing use

Natural acidification only

A good built environment

Source: Government Bill (2000).

can be reported per target or objective, and, in the first case, a summarising assessment can also be made for the headline objectives if appropriate. The result is thus a valuation per objective. If a further aggregation is required, this has to be done using some MAA method, as discussed above. A valuation against environmental objectives can work with qualitative information as well as quantitative data. It can be site-/time-specific or unspecific, and provides for the assessment of a comprehensive set of impacts. 3.9. Economic valuation In the valuation stage of the SEA, the achievement of multiple objectives and the trade-offs between different objectives need to be analysed, processed and interpreted. At this stage, economic valuation methods can be useful. There is an extensive literature on economic valuation methods, e.g. Markandya and Richardon (1992), Viscusi (1997), Leksell (1998) and Commission for the European Communities (1995). However, economic valuation is controversial and has been criticized both conceptually and in applications and this is further discussed below. At the core of economic valuation in the SEA context is the divergence between social and private cost in the market as a result of a certain activity, i.e. a negative environmental externality (Bojo¨ et al., 1992; Naturva˚rdsverket, 1997; Begg et al., 1987). Economists distinguish between use values and non-use values. Use values are those values that are related to human production and consumption patterns and they can be direct, indirect or option values. Sometimes market prices exist for

106

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

externalities relating to use values. For example, the effect of reduced crop productivity due to air pollution can be estimated by the market value of crops lost. For other such externalities no market prices exist, however, for example cough episodes and uneasiness as a result of urban air pollution (Commission for the European Communities, 1995). Non-use values are of a more philosophical kind and they can be either existence values or bequest values. Non-use values are usually only possible to estimate through artificial market methods. Valuation approaches can thus be grouped into those that use conventional markets, implicit markets or artificial markets (Bojo¨ et al., 1992). Conventional market valuations include analysis of changes in production, changes in earnings, replacement cost and defensive expenditure. Implicit market valuations study the revealed preferences from actual consumer behaviour and choices. These include wage-risk approaches, travel cost approaches, land and property value or hedonic pricing approach. Artificial market valuations include measurements of consumer preferences in hypothetical situations, with Willingness-to-pay (WTP) or Willingness-to-accept (WTA) measures. These are sometimes referred to as direct methods (Asian Development Bank, 1996). In the SEA context, it is doubtful whether resources will be available to carry out valuation studies. Instead, the assessor will probably have to work with the benefits transfer approach, where one utilises results and data from existing studies and adjusts them to the decision situation. One can adjust at the impact quantification stage or in the valuation stage for differences in receptor qualifiers such as structure, density, income, and behaviour. 3.10. Surveys In many cases it will not be possible to model or calculate the environmental consequences in quantitative terms, or even to say something meaningful in qualitative terms. This situation is expected to arise when we are assessing effects on landscapes and cultural values that are highly related to people’s perceptions and preferences, i.e. non-use values, such as for instance changes in the visual situation in archipelagic or mountainous regions and in cultural rural landscapes. In those cases, it can be meaningful to apply expressed-preference survey methods to get an idea of people’s perception of these consequences. An example of such a method that is commonly used in economic research is the contingent valuation (CV) method, a stated preference method (or artificial market valuation method) where people are asked to express their willingness to pay for certain environmental features (see Section 3.9). There are also methods that do not directly aim to get economic values, as CV methods do. Opinion and attitudinal surveys is one such family of tools (Gregory, 1998). They can be used to compare the relative importance of different environmental values or to examine how people balance environmental protection with their economic and social needs and wants.

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

107

There are numerous survey techniques that draw upon various disciplinary bases. These include public-value surveys or constructive multi-attribute methods, where the problem is deconstructed, analysed and recomposed; and decision-pathway surveys, where we draw out attitudes through a set of linked questions (Gregory, 1998; Keeney and Raiffa, 1976). Another set of survey methods emphasise small group elicitations, in-depth interviews and narratives of individuals (Dahinden et al., 2000). Apart from public values and attitudes, surveys have also been used to solicit expert opinion on different alternatives. The Delphi technique consists of an iterative questionnaire process, in which the expert panel’s first round of responses are summarized statistically and fed back in order to give the opportunity for a second round of revised responses (see Dalkey, 1969; Ott, 1978; Sobral et al., 1981). Apart from the above-mentioned substantive argument for eliciting public opinions, i.e. to obtain better information on consequences, there are several good reasons to elicit public information in the assessment. The normative argument is that the decision-makers should obtain a meaningful input and participation from the affected population. This argument is embraced in legislation. The argument is that a meaningful participation is likely to increase acceptance for decisions and reduce the risk of conflict as a result of the process (Stern and Fineberg, 1996). Similar to MAA, surveys in general are a broadly applicable method that can be used for any type of impact. 3.11. Valuation methods based on mass, energy and area Partly as reactions to the use of risk assessment and economic assessments, alternative approaches for valuation have been developed focusing on the inputs of the studied system. This can be motivated because inputs are often known with greater certainty, whereas outputs and impacts of those, which are the subject of risk assessment and economic valuations, are normally more uncertain. Although these methods have not been primarily developed for SEA, they could be used as valuation methods within SEA, providing an alternative weighting system to the economic valuation. Material Flow Analysis (MFA) is a family of different methods (Bringezu et al., 1997). A common feature is the focus on material flows, especially on the input side. Two types are briefly mentioned here: Total Material Requirement (TMR) and Material Intensity Per Unit Service (MIPS). TMR and related concepts such as Direct Material Input (DMI) and Direct Material Consumption (DMC) normally have a nation as their principal object of study. TMR aims at calculating all material inputs to the society, including both direct and hidden inputs (Adriaansee et al., 1997) whereas DMI and DMC focus on the direct inputs, excluding hidden flows such as the overburden from mining operations. The approach has developed during the 1990s and has mainly been used in retrospective studies. Instrumental MIPS is similar to TMR, but in this case the

108

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

object is a product or a service (Spangenberg et al., 1999). MIPS is similar to LCA but focuses on the material inputs. Energy analysis has many similarities with bulk-MFA methods. They focus on the inputs in physical measures and they may be used as evaluation methods for different types of objects. There are several different types of energy measures such as exergy (which can be defined as a measure of available energy; Szargut et al., 1988) and emergy (which is a historic energy measure describing assimilated energy from energy, material, information and labour; Odum, 1996). Ecological footprint is also an evaluation method which in principle can be applied to different types of objects, although it has mainly been used on regions, nations and projects such as aquaculture. The results are presented in terms of area used. The focus is on the area necessary for different types of activities, but the indirect area which could be used for assimilating different types of emissions is also included (Wackernagel and Rees, 1996).

4. Integrating tools in SEA 4.1. Some aspects of SEA as a tool 4.1.1. System boundaries The choices regarding system boundaries can influence the choice of methods and are therefore discussed briefly here. System boundaries can exist in several dimensions: 1. 2. 3. 4.

Geographical Geographical Geographical Geographical

(and (and (and (and

time) time) time) time)

boundaries boundaries boundaries boundaries

for for for for

the activity. emissions and use of resources. impacts. other activities.

In the scoping stage of SEA, choices have to be made in all these dimensions. As an example, consider an energy policy in Sweden during a certain time period (1). This policy will influence emissions not only in Sweden but also in other countries, where for example different fuels or materials are produced. The policy can also lead to emissions far in the future. For example, emissions from landfills and contaminated ground can prevail for very long time periods after the activity has finished (2). The impacts will also occur in different areas because of regional and global transports of pollutants. If the pollutants are persistent, the impacts can occur long after the emissions occurred (3). A new energy policy in Sweden may also influence the energy system in other Nordic countries as well as in Europe. In parallel, even after an activity has finished, it can have an impact on other activities (4). Another type of system boundary relates to the life-cycle of products, fuels and materials. If a life-cycle perspective is used, environmental changes and

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

109

effects from raw material acquisition, via production and use, to waste disposal should be considered. In relation to energy applications, this means for example that not only the production of electricity should be included but also the production of fuels, and the disposal of waste materials. The first of the geographical and time boundaries is necessary as a part of the definition of the PPP under study. Choices about the other boundaries are not analytical but rather political choices. If certain geographical and time boundaries are used for the impacts considered, it means that impacts occurring outside the geographical boundary and impacts on future generations can be neglected. However, in relation to the basic recognition of the international and intergenerational importance of environmental protection we argue for broad system boundaries and a life-cycle perspective to activities and pressures at least as a starting point. 4.1.2. Types of environmental changes and effects and assessment objects Another important choice has to be made concerning which environmental changes and effects to consider in the assessment. One starting point can be a comprehensive list of environmental objectives. Another choice has also to be made concerning how to define the environmental effects and changes within the DPSIR model. This is partly a choice influenced by world views and values and further discussed below. We suggest that a comprehensive list of environmental effects and objectives should be considered at the start of the assessment. An example is the list of Swedish environmental objectives that can be used as a checklist in the early stages of the assessment, and later be narrowed down in relation to what is considered most important. The narrowing down should however be made in a structured way and the reasons should be documented. 4.1.3. Site-specificity The degree of site- and time-specificity may vary in different applications of SEA. Often, the sites of relevance may not be determined at the strategic level studied in SEA, in which case a site-specific analysis will be difficult. Furthermore, if a life-cycle perspective is used, a site-specific assessment of all the sites involved often proves unfeasible. Therefore, many SEAs may find themselves in a non-site-specific context. However, this may vary with the level of decisionmaking. In local or regional level processes, the site context can be of high relevance and more site-dependent information can be used compared to a national level. The degree of site-specificity is also connected to the choice of indicators. In general, indicators on an impact level require a higher degree of site-specificity than indicators on a pressure level. 4.2. A framework of analytical tools The examination of the different methods in Section 3 reveals different qualities. Table 3 summarises some of the findings in a few key dimensions.

110

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

Table 3 Key qualities of different methods in relation to SEA

Future studies Life cycle assessment Risk assessment Input/output analysis Impact pathway approach Ecological impacts Multiple attribute analysis Environmental objectives Economic valuation Surveys Mass/energy valuation Ecological footprints

Site- and time-specificity

Degree of quantification

DPSIR

variable low variable low high high variable variable variable variable variable variable

variable high high high high low variable variable high variable high high

PI I P PSI SI PSI PSI PSI PSI P P

Fig. 2 shows in which steps of the SEA process (as defined in Section 1) the different tools discussed above can be used. The environmental analysis step has been further divided into three substeps: – A detailed description of the systems that the PPP may affect. (For the energy sector this may require a detailed description of energy system including fuel mixes, but also the use of other products and materials, for example in relation to energy efficiency measures.) – Identification of environmental interventions (emissions, extractions of resources, land use, etc.). – Analysis of environmental change due to the interventions. Future studies can be used at several steps in the SEA process. The outcome of future studies in SEA can either be a set of alternatives which all lead towards the predefined goal (as in back-casting studies) or a set of future scenarios in which the different alternatives can be placed. If a back-casting approach is used, a more comprehensive environmental assessment should be performed since normally only some environmental aspects are used in the goal description. As shown in Fig. 2, back-casting can be used in the formulation of alternatives. Other types of future studies will be useful primarily in the scenario analysis. Concerning when to use which future study approach, only tentative guidelines can be presented here. In general it can be concluded that if the decision process concerns short term issues, in areas where trends can be assumed to be stable and not desirable or possible to change, forecasts (or conditional forecasts) may be preferable. If this is not the case then some kind of scenario approach may be more appropriate. An important aspect in the choice of scenario approach is the extent to which the decision-maker can influence the future. If the decisionmaker has limited power external scenarios are appropriate, while normative

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

111

Fig. 2. Relationship between steps of the SEA process and different tools.

policy scenarios are more appropriate when the future can be influenced through strategic action. Futures studies may also be used within the environmental analysis step of the SEA (not shown in Fig. 2). The consequences of assessed alternatives depend on future environmental baseline situations. Future studies could also be of relevance in the valuation step. This is because future societal values may be different from current ones. Several analytical tools can be used for the identification and description of environmental interventions, including life-cycle inventory analysis data and methodology, different types of checklists (possibly based on environmental objectives), accident-related risk assessment and environmentally extended input/ output analysis (IOA). Since the industries and product groups are defined rather broadly in IOA, it may be too blunt a tool for different fuels or fuel mixes. It may however be useful to provide information on indirect interventions from other industries as a consequence of changes within the energy sector.

112

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

For the analysis of environmental change, several tools can be used. The differences between the tools concern both the degree of site-specificity and what types of environmental effects and changes are considered in the DPSIR model. Characterisation methods developed and used for LCA can be used both with generic (site-independent) characterisation factors and with site-dependent characterisation factors if such are available. Furthermore, risk assessment methods in general and the Impact Pathway Approach in particular may be used at this stage. Concerning the valuation of environmental impacts, a valuation against environmental objectives producing a multi-dimensional result can be used. If a one-dimensional result is desired several types of aggregation methods may be used: multi-attribute approaches, economic valuation, surveys, LCA weighting methods and valuation using energy, mass or area. Fig. 2 gives a presentation of which tools can be used at which step in the SEA process. It is also of interest to consider what type of information different tools provide and how they can interact. This is illustrated in Fig. 3. Future studies will, together with the formulation of alternatives, result in a description of the technical system in each alternative. This information will typically have some site-specificity, e.g. concerning which nation or region the PPP is developed for. The information can be qualitative or quantitative. In the former case, only a qualitative environmental analysis is feasible. If the description is quantitative, a choice can be made to take either a qualitative or a quantitative path. Different paths can be chosen for different types of environmental impacts. Along the qualitative path, checklists can be used resulting in qualitative information about environmental impacts. A qualitative multi-dimensional valuation against environmental objectives can then be made. Along the quantitative path, a switch to the qualitative path can be made at any time, as indicated at some places in Fig. 3. It is more complicated to go from the qualitative to the quantitative path. By using survey methods it is however possible to ask people (laymen, experts or some other group) to value the qualitative descriptions of the environmental impacts. If people are asked to give monetary values, it will be an economic valuation. People can also be asked to give ‘‘points’’ resulting in a non-monetary measure as a part of a MAA method. If the quantitative path is taken, Life Cycle Inventory data can be used. It provides data on environmental emissions, resource extractions, etc., per functional unit, e.g. MJ fuel or kg of material. In a PPP, different alternatives can possibly provide different functions and this is a difference between SEA and LCA. When the life cycle inventory data is combined with the description of the technical system, the result is quantitative information about emissions, resource extraction etc. for each alternative, often with some site-specific information. The information available at this stage can either be taken directly to a valuation step (not shown in Fig. 3) or as indicated in Fig. 3 it can be used for

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

113

Fig. 3. Information provided by different tools and their possible interaction.

further processing in the environmental analysis. In Fig. 3, three different paths are suggested: traditional LCA characterisation, site-dependent LCA-characterisation or risk assessment including air quality modelling as in the Impact Pathway Approach. Traditional LCA characterisation will result in quantitative

114

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

information without site-specific information. The site-dependent LCA characterisation and the risk assessment approaches can result in quantitative information with some site-dependent information. The LCA-approaches can calculate results on a pressure- or impact-level within the DPSIR-model. They cannot, however, calculate environmental states (such as a concentration level). This is in contrast to the risk assessment approach which can produce information at both pressure, state and impact levels. After this step, a choice can be made as to whether a valuation against environmental objectives is to be made, if the information should be taken to another type of valuation or if the process is stopped at this stage and conclusions are drawn with the available information. If a valuation against environmental objectives is made, the LCA characterisation approaches cannot be used if the objectives are expressed as desirable states. If environmental state indicators are of interest, a risk assessment approach is required in the earlier steps. As an alternative to a valuation against environmental objectives, methods such as multi-attribute approaches, economic valuation, and valuation using energy, mass or area, can be used. All these methods can produce a onedimensional quantitative result, as opposed to a multi-dimensional qualitative result. The last step in Fig. 3 is to draw conclusions and formulate recommendations. This should be made in relation to the aims of the SEA as formulated earlier in the process. Not shown in Fig. 3 is the possibility to stop at almost any step in the process and try to draw conclusions. It is therefore not necessary to follow a path all the way. At this stage it is important to use all types of information produced in the process. Thus, even if a quantitative, one-dimensional valuation method is used, conclusions should also be based on results earlier in the process to ensure that no relevant information is lost. 4.3. Some comments on the framework The framework as outlined above is open to many choices, which are largely determined by what is regarded as possible and what is regarded as a meaningful result. Although partly scientific issues, the answers to such questions are partly determined by values and world views. A few examples are used to illustrate these discussions. One example concerns the choice of which types of impacts to consider. This can be illustrated by controversies around toxic compounds (Tukker, 1999). Different stakeholders (e.g. industry, governmental agencies and NGOs) may act in different frames or paradigms which determine what is considered as a meaningful object to study. For example, industries may act within a ‘‘risk assessment frame’’ where risk assessments are considered useful and meaningful and management should be based on results from such studies (ibid.). NGOs may on the other hand act within a ‘‘phase-out frame’’ where risk assessments are considered too uncertain because science does not have enough knowledge and

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

115

management must be based on a precautionary principle (ibid.). Analysis based on material and substance flows may be regarded as more appropriate and meaningful under a ‘‘phase-out frame’’. Within a ‘‘risk assessment frame’’ indicators on an impact level are typically regarded as relevant and meaningful. The precautionary principle is by some looked upon as unscientific, while others see it as no less scientific than other principles (Sandin et al., 2002). Within a ‘‘phase-out frame’’, indicators on a pressure level, or even a driving force level may be regarded as more meaningful. Another example is whether it is meaningful and useful to produce a onedimensional result using a valuation method. It is clear that the use of valuation methods involves different types of values, not only in relation to how different aspects are valued against each other, but also in relation to what type of valuation method should be used and also whether a valuation method should be used at all (Finnveden, 1997). An illustration is the controversies around economic valuation (Georgescu-Roegen, 1971; Daly and Cobb, 1990; Hobbs and Meier, 2000). Another illustration is whether methods based on mass (Kleijn, 2001), energy or area produce meaningful results. The framework suggests that several types of tools are to be used. It is however important to note that this does not imply that the analyses are made specifically for the SEA. It may be the case that results from earlier studies can be used in the SEA. For example, LCAs can be time-consuming to perform, but especially in the energy sector there are several studies that have already been made. There is also software available with databases that can be used (e.g. Rice et al., 1997; Jo¨nbrink et al., 2000). Other tools in the framework, for example surveys, are generally case-specific. It is important to note that the quantitative road cannot be used for all types of environmental impacts. In relation to the Swedish environmental objectives, it can be noted that typical LCAs, and similar tools, can provide information mainly related to objectives 1 –5 and 7 in Table 1. The other objectives are more related to impacts on ecosystems and landscapes and therefore more difficult to handle. The lack of appropriate methods for Ecological Impact Assessment within the framework of SEA was pointed out above. 4.4. The framework in relation to application It is clear that the application can and should determine the SEA process. It is however not clear how the application influences the choice of analytical tools and methods. This is partly because the applications can be described in many dimensions. In this section we will tentatively discuss some possible influences on the choice of analytical tools. 4.4.1. Applications in different sectors Although the focus of this study is on the Swedish energy sector, it can be noted that large parts of the framework are also valid for other sectors and other

116

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

countries. In other sectors, the focus of the assessment may change from air emissions. It is however suggested that for most sectors, most types of environmental impacts are of importance. For example, in the energy sector, not only air emissions are of relevance, but also water emissions (e.g. from waste water from air pollution control devices, landfilling of combustion residues and mining residues, and oil leakage) and direct impacts on ecosystems and landscape (e.g. from production of bio fuels, mining operations, hydro power and wind power). One possible difference concerns the site-specificity. For some sectors, for example traffic or physical planning, already at a PPP level, a higher degree of site-specificity may be possible and desirable. The influence of the site-specificity is further discussed below. 4.4.2. Applications by different actors SEA can be applied at national, regional and local levels. The degree of site-specificity may change at these different levels. On a more local level, both the possibilities and need for site-dependent assessments may increase. Values and world views may change between different actors, e.g. governmental agencies, industries and NGOs and this may also influence the choice of methods. Resources in terms of time and funding may also change between different actors. Besides these aspects (further discussed below), it is suggested that the framework and choice of analytical methods is not influenced by the actor. 4.4.3. Functions of the SEA An SEA can have several functions, such as supporting a choice between two or several alternatives or identifying critical aspects of studied alternative(s) and suggest mitigation strategies. Within the same SEA process, both these functions can be relevant. These different functions are related to the required degree of quantification. If the intended application is an identification of critical aspects in order to suggest a mitigation strategy, a qualitative approach is often sufficient. Qualitatively it is often possible to determine if something is ‘‘critical’’ or ‘‘significant’’. However, if the objective is to support a choice between two or several alternatives, the quantitative requirements typically increase. This is because if a trade-off has to be made between two important and critical aspects, a quantification of how severe the critical aspects are is often necessary. 4.4.4. Possible and desirable degree of site-specificity As noted above, the possible degree of site-specificity may vary between different applications. It is typically higher at local levels and in applications where land use is at the centre of the assessment. The desirable degree of sitespecificity is also related to world views and scientific paradigms. In the ‘‘risk assessment’’ paradigm it is generally considered important to do a site-specific assessment in order to assess the risks. In a ‘‘phase-out’’, or ‘‘strict-control

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

117

frame’’, it may be enough to consider the environmental pressures as discussed above. Related to this topic is also the discussion on choice of indicators within the DPSIR-model which is also connected to values and world views as discussed above. 4.4.5. Values and world views It is interesting to note that several of the application dependencies discussed above boil down to a dependency on values and world views which can have a decisive influence on the choice of analytical tools and methods. Since different stakeholders have different values and world views, consultation and understanding are important to ensure credibility and relevance. In some cases it may be useful to use several different approaches and produce several different types of results. These can then form a basis for further discussions. In summary it is suggested that SEAs used to support a choice between different alternatives require more quantitative methods, whereas SEAs used to identify critical aspects and suggest mitigation strategies can be made with more qualitative methods. The possible and desirable degree of site-specificity in the assessment can also influence the choice of methods. It is also suggested that values and world views can be of importance for judging whether different types of tools and results are meaningful and useful.

5. Conclusions The aim of this article is to analyse how various analytical tools can facilitate and enhance the SEA process, particularly in relation to energy sector SEA. It is found that several existing tools can contribute, either by focusing mainly on the identification and modelling of environmental change (e.g. LCA, RA, future studies) or by focusing mainly on the valuation stage (e.g. MAA methods, economic valuation methods, surveys). Based on the tools examined here, it appears that finding useful tools for analysing ecosystem and landscape impacts is more challenging than tools for analysing emissions of pollutants, at least in the energy sector context. Based on the examination of the selected analytical tools according to a set of analytical features, an integrative framework of methods for SEA is proposed (see Figs. 2 and 3). Three conclusions can be drawn from this exercise. First, the key factors influencing the choice of analytical tools are the definition of system boundaries, the amount and types of environmental changes included in the assessment, the degree of site-specificity desired, the degree of quantification desired, the degree of aggregation of results desired, and the preference of information type according to the DPSIR-model. Second, and in addition to these factors, the preferred function of the SEA also influences the choice. It is argued that to support a choice between two or more alternatives quantitative results may be needed, while a SEA with the purpose to identify critical aspects of

118

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

alternative(s) and suggest mitigation strategies might do with qualitative results. Lastly, it was suggested that underlying the choices that shape the use of methods in a SEA is the world view and assumptions of the assessor, i.e. considered relevant information. The next step will be to test the framework of analytical tools in a SEA on a Swedish energy sector PPP.

Acknowledgements Financial support from the Swedish National Energy Administration is gratefully acknowledged.

References Adriaansee A, Bringezu S, Hammond A, Morigutchi Y, Rodenburg E, Rogich D, Schu¨tz H. Resource flows: the material basis of industrial economies. Washington, DC: World Resources Institute; 1997. AIChemE. Guidelines for chemical process quantitative risk assessment. New York: Center for Chemical Process Safety of the American Institute of Chemical Engineers; 1998. ANSEA Project. Towards an analytical strategic environmental assessment new concepts in strategic environmental assessment—towards better decision-making. ANSEA Project Dissemination Document, Working Paper 28.2002, Milano: FEEM; 2002. Asian Development Bank. Economic evaluation of environmental impacts: a workbook. Manila: ADB; 1996. Bardouille P. A Framework for Sustainable Strategic Energy Company Investment Analysis and Decision-Support. PhD Thesis. Lund: Environmental and Energy System Studies, Lund Institute of Technology; 2001. Baumann H, Cowell S. An evaluative framework for conceptual and analytical approaches used in environmental management. Greener management international. J Corp Environ Strategy Practice 1999;(26):(26):109 – 22. Begg D, Fischer S, Dornbusch R. Economics. London: McGraw-Hill; 1987. Bojo¨ J, Ma¨ler K-G, Unemo L. Environment and development: an economic approach. Dordrecht: Kluwer Academic Publishing; 1992. Bots P, Hulshof J. Designing multi-criteria decision analysis processes for priority setting in health policy. J Multi-Criteria Decis Anal 2000;9(26):56 – 75. Boverket och Naturva˚rdsverket. SMB och o¨versiktlig fysisk planering. Karlskrona och Stockholm: Boverket och Naturva˚rdsverket; 2000 [in Swedish]. Bringezu S, Fischer-Kowalski M, Kleijn R, Palm V, editors. Analysis for action: support for policy towards sustainability by material flow accounting. Proceedings of the ConAccount Conference 11 – 12 Sept. Wuppertal, Germany: Wuppertal Institute; 1997. Calthrop E, Maddison D. The dose – response function approach to modelling the health effects of air pollution. Energy Policy 1996;24(7):599 – 607. Commission for the European Communities. ExternE—externalities of energy. Methodology, vol. 2. Luxembourg: Office for Official Publications of the European Commission; 1995. Commission for the European Communities. ExternE—externalities of energy. Methodology 1998 update, vol. 7. Luxembourg: Office for Official Publications of the European Commission; 1999.

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

119

Dahinden U, Querol C, Ja¨ger J, Nilsson M. Exploring the use of climate models in participatory integrated assessment—experiences and recommendations for further steps. Integrated Assessment 2000;1(7):253 – 66. Dalkey N. The Delphi method: an experimental study of group opinion. Santa Monica: The Rand Corporation; 1969. Daly HE, Cobb Jr JB. For the common good: redirecting the economy towards community, the environment, and a sustainable future. London: Green Print; 1990. Dı´az M, Illera JC, Hedo D. Strategic environmental assessment of plans and programs: a methodology for estimating effects on biodiversity. Environ Manag 2001;28(2):267 – 79. Dockery D, Pope A. Acute respiratory effects of particulate air pollution. Annu Rev Public Health 1994;15(2):107 – 32. Dockery D, Pope A, Xu X, Spengler J, Ware J, Fay M, et al. An association between air pollution and mortality in six US cities. N Engl J Med 1993;329(24). Dom A. Environmental impact assessment of road and rail infrastructure. In: Petts J, editor. Handbook of environmental impact assessment. Environmental impact assessment in practice: impact and limitations, vol. 2. London: Blackwell; 1999. Dreborg K-H. Essence of Backcasting. Futures 1996;28(9):813 – 28. Dreborg K-H. Tre fo¨rha˚llningssa¨tt till framtiden: backcasting i ett vidare perspektiv. Technol, Soc, Environ 2001;3(9):75 – 95 [in Swedish]. Eduljee G. Risk assessment. In: Petts J, editor. Handbook of environmental impact assessment, vol. 1. London: Blackwell; 1999. p. 374 – 404. European Parliament and Council of the European Union. On the assessment of the effects of certain plans and programmes on the environment. 2001 [C5-0118/2001]. European Commission. Strategic environmental assessment: existing methodology. Luxembourg: European Commission; 1994. Federal Ministry for the Environment Nature Conservation and Nuclear Safety. Workshop on Strategic Environmental Assessment (SEA) in the cooperation with developing and transition countries, Berlin, November 26th – 27th 2001. Berlin: Nature Conservation and Nuclear Safety, Federal Ministry for the Environment; 2001. Feldmann L, Vanderhaegen M, Pirotte C. The EU’s SEA Directive: status and links to integration and sustainable development. Environ Impact Asses Rev 2001;21(3):203 – 22. Finnveden G. Valuation methods within LCA—where are the values? Int J LCA 1997;2(3):163 – 9. Finnveden G. On the limitations of life cycle assessment and environmental systems analysis tools in general. Int J LCA 2000;5(3):229 – 38. ˚ . Environmental systems analysis tools—an overview in relation to decision Finnveden G, Moberg A context. 2001 [submitted for publication]. Georgescu-Roegen N. Entropy law and the economic process. Cambridge, MA: Harvard Univ. Press; 1971. Government Bill. Svenska miljo¨ma˚l—delma˚l och a˚tga¨rdsstrategier; 2000/01:130 [in Swedish]. Gregory R. Identifying environmental values. In: Dale VH, English MR, editors. Tools to aid environmental decision making. New York, NY: Springer; 1998. Hansson SO. A philosophical perspective on risk. Ambio 1999;28(3):539 – 42. Harms-Ringdal L. Safety analysis: principles and practice in occupational safety. London: Taylor & Francis; 2002. Hertwich E, Jolliet O, Pennington DW, Hauschild M, Schulze C, Krewitt W, et al. Fate and exposure assessment in the life cycle impact assessment of toxic chemicals. In: Udo de Haes HA, Jolliet O, Finnveden G, Goedkoop M, Hauschild M, Hertwich E, Hofstetter P, Klo¨pfer W, Krewitt W, Lindeijer E, Mu¨ller-Wenk R, Olsen SI, Pennington DW, Potting J, Steen B, editors. Towards best practice in Life Cycle Impact Assessment—report of the second SETACV-Europe working group on Life Cycle Impact Assessment. Pensacola, FL: SETAC; 2002 (in press). Hobbs BF, Meier P. Energy decisions and the environment. Dordrecht: Kluwer Academic Publishing; 2000.

120

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

Ho¨jer M, Mattsson L-G. Determinism and backcasting in future studies. Futures 2000;32(3):613 – 34. Huijbregts MAJ. Uncertainty and variability in environmental life-cycle assessment. PhD Thesis. Amsterdam: University of Amsterdam; 2001. IAIA. Principles of environmental impact assessment best practice. Fargo, ND: International Association for Impact Assessment; 1999. ISO. Environmental management-life cycle assessment-principles and framework. International Standard ISO 14040. Geneva: International Organisation for Standardisation; 1997. ISO. Environmental management-life cycle assessment-goal and scope definition and inventory analysis. International Standard ISO 14041. Geneva: International Organisation for Standardisation; 1998. ISO. Environmental management-life cycle assessment-life cycle impact assessment. International Standard ISO 14042. Geneva: International Organisation for Standardisation; 1999. Jitendra S, Nagpal B, Brandon C, editors. Urban air quality management strategy in Asia: a guidebook. Washington, DC: World Bank; 1997. Jo¨nbrink AK, Wolf-Watz C, Erixon M, Olsson P, Walle´n E. LCA software survey. Mo¨lndal: IVF Research Publication 00824, IVF (The Swedish Institute of Production Engineering Research); 2000. Joshi S. Product environmental life-cycle assessment using input – output techniques. J Ind Ecol 2000;3(2 and 3):95 – 120. Keeney RL. Value-focused thinking: a path to creative decision-making. Cambridge: Harvard Univ Press; 1992. Keeney R, Raiffa H. Decisions with multiple objectives. Cambridge: Cambridge Univ Press; 1976. KemI. Riskbedo¨mning och Riskhantering inom Kemikaliekontrollen. Stockholm: Rapport fra˚n Kemikalieinspektionen; 1995 [11/95 PrintGraf, 1995 in Swedish]. Khan FI, Abbasi SA. Techniques and methodologies for risk analysis in chemical process industries. J Loss Prev Process Ind 1998;11(2 and 3):261 – 77. Kleijn R. Adding it all up. The sense and non-sense of bulk-MFA. J Ind Ecol 2001;4(2):7 – 8. Krewitt W, Trukenmu¨ller A, Bachmann TM, Heck T. Country-specific damage factors for air pollutants. A step towards site dependent life cycle impact assessment. Int J LCA 2001;6(2):199 – 210. Lave LB, Cobas-Flores E, Hendrickson CT, McMichael FC. Using input – output analysis to estimate economy-wide discharges. Environ Sci Technol 1995;29(9):420A – 6A. Leksell I. Metoder att ge en samha¨llsekonomisk va¨rdering av luftfo¨roreningars ha¨lsoeffekter (arbetsmaterial). Go¨teborg: Go¨teborg University, Institutionen fo¨r tilla¨mpad miljo¨vetenskap; 1998. Lindeijer E, Mu¨ller-Wenk R, Steen B. Impact assessment of resources and land use. In: Udo de Haes HA, Jolliet O, Finnveden G, Goedkoop M, Hauschild M, Hertwich E, Hofstetter P, Klo¨pfer W, Krewitt W, Lindeijer E, Mu¨ller-Wenk R, Olsen SI, Pennington DW, Potting J, Steen B, editors. Towards best practice in Life Cycle Impact Assessment—report of the second SETACV-Europe working group on Life Cycle Impact Assessment. Pensacola, FL: SETAC; 2002 (in press). Makridakis S, Wheelright SC, Hyndman RJ. Forecasting. Methods and applications. New York: Wiley; 1998. Markandya A, Richardon J. The earthscan reader in environmental economics. London: Earthscan; 1992. Miller RE, Blair P, editors. Input-ouput analysis. Englewood Cliffs: Prentice Hall; 1985. Miser H, Quade E, editors. Handbook of systems analysis: overview of uses, procedures, applications and practices. Chichester: Wiley; 1985. ˚ , Finnveden G, Johansson J, Steen P. Miljo¨systemanalytiska verktyg—en introduktion Moberg A med koppling till beslutssituationer. Stockholm: AFR Rapport 251, AFN, Naturva˚rdsverket; 1999 [in Swedish]. Mo¨rtberg U, Balfors B. Landscape ecological assessment in an urbanising environment. Proceedings of Conference on how to Integrate Environmental Aspects into Spatial Planning, Stockholm. 2000;66 – 77.

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

121

Naturva˚rdsverket. Miljo¨skatter i Sverige: ekonomiska styrmedel i miljo¨politiken. Stockholm: Naturva˚rdsverket Fo¨rlag; 1997 [in Swedish]. Naturva˚rdsverket. Strategiska miljo¨bedo¨mningar: ett anva¨ndbart instrument i miljo¨arbetet. Stockholm: Rapport 5109, Naturva˚rdsverket, 2000 [in Swedish]. Nigge K-M. General spatial classes for human health impacts: Part 1. Methodology. Int J LCA 2001a;6(9):257 – 64. Nigge K-M. General spatial classes for human health impacts: Part 2. Application in a life cycle assessment of natural gas vehicles. Int J LCA 2001b;6(9):334 – 8. Nilsson M. Approaches to an earth audit. Nairobi: SEI/Earth Council/UNEP; 1997. ˚ . Naturgasutbyggnad i Sverige-metod fo¨r strategisk Nilsson M, Finnveden G, Johansson J, Moberg A miljo¨bedo¨mning inom energisektorn. Stockholm: Report 5161, Swedish EPA; 2001 [in Swedish]. Noble B. Strategic environmental assessment: what is it and what makes it strategic? J Environ Assess Policy Manag 2000;2(2):203 – 24. Noble B, Storey K. Towards a structured approach for strategic environmental assessment. J Environ Assess Policy Manag 2001;3(4):483 – 508. Odum HT. Environmental accounting. Emergy and environmental decision-making. New York: Wiley; 1996. Olofsson M. Linking the analysis of waste management systems and energy systems. Licenciate thesis. Go¨teborg: Energy Systems Technology Divisions, Chalmers University of Technology; 2001. Olsen SI, Christensen FM, Hauschild M, Pedersen F, Larsen HF, To¨rslo¨v J. Life cycle impact assessment and risk assessment of chemicals — a methodological comparison. Environ Impact Assess Rev 2001;21(4):385 – 404. Ostro B. Estimating the health effects of air pollution: a method with an application to Jakarta. The World Bank, policy research working paper, WPS 1301. Washington, DC: World Bank; 1994. Ott WR. Environmental indices: theory and practice. Ann Arbor: Ann Arbor Science; 1978. Partidario M. Strategic environmental assessment-principles and potential. In: Petts J, editor. Handbook of environmental impact assessment. Environmental impact assessment-process, methods and potential, vol. 1. Oxford: Blackwell; 1999. Patton CV, Sawicki DS. Basic methods of policy analysis and planning. Englewood Cliffs, NJ: Prentice Hall; 1993. Petts J. Environmental impact assessment versus other environmental management decision tools. In: Petts J, editor. Handbook of environmental impact assessment. Environmental impact assessment: process, methods and potential, vol. 1. London: Blackwell; 1999a. Petts J, editor. Handbook of environmental impact assessment. London: Blackwell; 1999b. Potting J. Spatial differentiation in life cycle impact assessment. PhD thesis. Utrecht: Department of Science, Technology and Society, University of Utrecht; 2000. Ra¨ddningsverket. Olycksrisker och MKB. Karlstad: Swedish Rescue Services Agency, 2000 [in Swedish]. Rescher N. Predicting the future. An introduction to the theory of forecasting Albany: State University of New York Press; 1998. Rice G, Clift R, Burns R. LCA software review. Comparison of currently available European LCA software. Int J LCA 1997;2(4):53 – 9. Sandin P, Peterson M, Hansson SO, Rude´n C, Juthe A. Five charges against the precautionary principle. J Risk Res 2002;5(4):287 – 99. Saaty T, Vargas L. Decision making in economic, political, social and technological environments with the analytic hierarchy process. Pittsburgh, PA: Rws Publishers; 1994. SBSTTA. Indicators and environmental impact assessment, UNEP/CBD/SBSTTA/7/13. Secretariat of the Convention on Biological Diversity, United Nations Environment Programe; 2001. SFS. Lag (1977:439) om kommunal energiplanering. The Swedish Parliament; 1977:439 [in Swedish]. Smeets E, Weterings R. Environmental indicators: typology and overview. Copenhagen: European Environment Agency; 1999.

122

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

Sobral MM, Hipel KW, Farquhar GJ. A multi-criteria model for solid waste management. J Environ Manag 1981;12(2):97 – 100. SOU. Omsta¨llning av energisystemet. Stockholm: Statens Offentliga Utredningar, Swedish Government; 1995:139 [in Swedish]. SOU. Framtidens miljo¨ - allas va˚rt ansvar. Statens Offentliga Utredningar. Slutbeta¨nkande fra˚n Miljo¨ma˚lskommitte´n; 2000:52 [in Swedish]. Spadaro JV, Rabl A. Estimates of real damage from air pollution: site dependence and simple impact indices for LCA. Int J LCA 1999;(4):(4):229 – 43. Spangenberg JH, Hinterberger F, Moll S, Schu¨tz H. Material flow analysis, TMR and the mipsconcept: a contribution to the development of indicators for measuring changes in consumption and production patterns. Department for Material Flows and Structural Change. Wuppertal: Wuppertal Institute for Environment, Climate and Energy; 1999. Stern PC, Fineberg HV, editors. Understanding risk. Informing decisions in a democratic society. Washington, DC: National Academy Press; 1996. Szargut J, Morris DR, Steward FR. Exergy analysis of thermal, chemical and metallurgical processes: hemisphere. New York: Hemisphere Publishing Corporation; 1988. Therivel R, Brown A. Methods of strategic environmental assessment. In: Petts J, editor. Handbook of environmental impact assessment. Environmental impact assessment—process, methods and potential, vol. 1. Oxford: Blackwell; 1999. Therivel R, Partidario MR. The practice of strategic environmental assessment. London: Earthscan; 1996. Therivel R, Thompson S. Strategic environmental assessment and nature conservation. Oxford: English Nature; 1996. Treweek JR, Hankard P, Roy DB, Arnold H, Thompson S. Scope for strategic ecological assessment of trunk-road development in England with respect to potential impacts on lowland heathland, the Dartford warbler (Sylvia undata) and the sand lizard (Lacerta agilis). J Environ Manag 1998;53(4):147 – 63. Tukker A. Frames in the toxicity controversy. Dordrecht: Kluwer, Academic Press; 1999. Udo de Haes HA, editor. Towards a methodology for life cycle impact assessment. Brussels: SETACEurope; 1996. Udo de Haes HA, Jolliet O, Finnveden G, Goedkoop M, Hauschild M, Hertwich E, Hofstetter P, Klo¨pfer W, Krewitt W, Lindeijer E, Mu¨ller-Wenk R, Olsen SI, Pennington DW, Potting J, Steen B, editors. Towards best practice in Life Cycle Impact Assessment—report of the second SETACV-Europe working group on Life Cycle Impact Assessment. Pensacola, FL: SETAC; 2002 (in press). Udo de Haes HA, Jolliet O, Finnveden G, Hauschild M, Krewitt W, Mu¨ller-Wenk R. Best available practice regarding impact categories and category indicators in Life Cycle Impact Assessment, background document for the second working group on Life Cycle Impact Assessment of SETACEurope. Part 1. Int J LCA 1999a;4(4):66 – 74. Udo de Haes HA, Jolliet O, Finnveden G, Hauschild M, Krewitt W, Mu¨ller-Wenk R. Best available practice regarding impact categories and category indicators in life cycle impact assessment, background document for the second working group on life cycle impact Assessment of SETAC-Europe. Part 2. Int J LCA 1999b;4(4):167 – 74. Unger T. Common CO2 Action and Cross-border Energy Trade—Modeling the Nordic Energy System. Paper I. Licentiate Thesis. Go¨teborg: Energy Systems Technology Division, Chalmers University of Technology; 2000. van der Heijden K. Scenarios. The art of strategic conversation. Chichester: Wiley; 1996. Viscusi K. Special issue. J Risk Uncertain 1997;15(2). Vries MSd. Calculated choices in policy-making: the theory and practice of impact assessment. London: MacMillan; 1999. Wackernagel M, Rees W. Our ecological footprint. Reducing human impact on the earth. British Colombia: New Society Publishers; 1996.

G. Finnveden et al. / Environmental Impact Assessment Review 23 (2003) 91–123

123

Wathern P. Ecological impact assessment. In: Petts J, editor. Handbook of environmental impact assessment. London: Blackwell; 1999. Weidema B, Lindeijer E. Physical impacts of land use in product life cycle assessment. Final report of the EURENVIRON-LCAGAPS sub-project on land use. Lyngby: Department of Manufacturing Engineering and Management, Technical University of Denmark; 2001. Wrisberg N, Udo de Haes HA, Triebswetter U, Eder P, Clift R. Analytical tools for environmental design and management in a systems perspective. Report to European Union. Leiden: CML, Leiden University; 2000.