Abstract
Metrics (potentials, potency factors, equivalency factors or characterization factors) are available to support the environmental comparison of alternatives in application domains like process design and product life-cycle assessment (LCA).  These metrics typically provide relative insights into the implicit concern associated with chemicals, emissions and resource consumption in the context of human health, ecological health and resource depletion.  The approaches used to derive the metrics range in their site-specificity, complexity, comprehensiveness, sophistication and uncertainty.  It is therefore often necessary to consider the use of more than one approach within the context of a given impact category to help support a decision.  In this paper we outline some of the strengths and weaknesses of available approaches in the commonly considered categories of global warming, stratospheric ozone depletion, tropospheric ozone (smog) creation, eutrophication/nutrification, acidification, toxicological impacts and resource depletion.

Introduction
Similar in concept to the comparison of process design alternatives on a site-specific (gate-to-gate) basis, although broader in scope, life-cycle assessment (LCA) is one framework for evaluating the inputs and emissions associated with all the stages in a product's life-cycle (from raw material acquisition, manufacturing, use and through to disposal).  Having established the product system boundaries and aims of the assessment (goal and scope definition), the associated resource consumption and emissions must be identified and tabulated (inventory analysis).  These inventory data are then considered according to the substance�fs implicit contribution within an "impact" category, such as global warming and ozone depletion.  Associated elements include:

• The selection of impact categories, metrics and models
• Assignment of the inventory analysis results to impact categories (classification)
• Calculation of category indicators using associated metrics (characterization factors) derived using characterization models (characterization)

The additional steps of calculating the magnitude of category indicator results relative to reference information for a given region or industrial sector (examples of normalization), sorting or ranking impact categories (grouping) and aggregating indicator results across impact categories (weighting, valuation or multi-objective decision making) remain controversial topics that may not be required in some case studies and cannot be addressed in sufficient detail in this paper.

In this paper we provide an overview of environmental comparison metrics (potentials, potency factors, equivalency factors or characterization factors) for the following "impact" categories: global warming, stratospheric ozone depletion, tropospheric ozone (smog) creation, eutrophication/nutrification, acidification, toxicological (carcinogenic and non-carcinogenic) impacts to humans, toxicological impacts to ecosystems and resource depletion.  Although these categories are commonly addressed, this list is not comprehensive.  Other categories include, but are not limited to, habitat alteration, changes in biodiversity, odor, noise and radiation.

A significant number of approaches are available to derive the comparison metrics.e.g.1, 2, 3  These approaches range in complexity (data intensity, knowledge requirements), comprehensiveness (breadth or scope of representation), sophistication (relevance to and depth of representation of the environmental mechanisms) and accuracy (uncertainty inherent to the model and associated with input data).  As a result, the selection of a methodology often remains subjective and strongly influenced by resource availability (in-house knowledge, input data availability, etc.).  No endorsement of the approaches outlined in this paper or suggestion that they are the "best available practice" for a given application is intended.

Emissions and resource consumption data are typically multiplied by environmental comparison metrics to help estimate their relative importance within a given impact category.  The results usually do not indicate that an actual impact will occur but often reflect relative differences in terms of implicit concern.  For example, the metrics can be based on implicit differences at a common midpoint in a cause-effect chain (environmental mechanism), as illustrated in Figure 1 for acidification.

Figure 1 Midpoints and endpoints in the simplified cause-effect chain for acidification

There is a tendency to define indicators at common midpoints to ensure simplicity and to minimize perceived uncertainty.  For example, comparison can be made in terms of radiative forcing and half-life differences in the context of global warming without the need to forecast specific endpoint effects.  Comparison at midpoints may not however always account for all factors in a cause-effect chain (as in Figure 1) and can result in a reduced ability to aggregate across impact categories.

Approaches used to derive endpoint metrics are typically more complex but have a number of potential advantages.  In addition to improved perceptions of "defensibility" and some opportunities to link emissions to observed effects, endpoint results can be more readily aggregated across certain impact categories.  For example, comparison can be performed in terms of endpoint effects to human health associated with ozone depletion, ozone creation, global warming, carcinogenic impacts, etc.e.g. 4  The resultant health effect measures often differ in the context of severity but these can be more readily combined using available valuation tools due to their "observable nature".  This approach provides a single human health comparison score.  Analogous techniques to derive single scores in terms of ecological health and resource depletion are however less developed.  Other disadvantages of endpoint methodologies can include reduced transparency, limitations in scope and significant uncertainty, particularly when predicting future damages.

Despite the limitations of the available comparison approaches addressed in this paper, the difficulty to compare design and product alternatives can be further compounded by factors related to the emissions and resource consumption (inventory) data supplied, including:

• the inventory data reflecting only a portion of the total consumption/emissions at given sites
• the often unknown and changing location of numerous sites in a life-cycle

In a gate-to-gate process design comparison and for some specific product life cycles such issues may be addressed and this facilitates the use of the site-specific approaches outlined.  In other applications, the generic metrics described will be more appropriate.  Site-specific insights will still however provide an important role in determining the uncertainties associated with more generic comparison approaches, although such uncertainties may be off-set by the large, sometimes worldwide, distribution of emission sites and resource sources in LCA.

Concluding remarks
Our ability to compare product and process alternatives using life cycle assessment (LCA), as well as in the context of more restricted scopes (e.g. site-specific or gate-to-gate), remains limited.  In this paper we briefly outlined some of the many considerations associated with available metrics that are used in relative "environmental" comparison applications in the context of resource depletion, human health and ecological impact categories.  Limitations are associated with the degree to which we can comprehend, predict and represent the complexities of our interactions with the environment, but also with the nature of inventory (emissions and resource consumption) data often presented for analysis.  It will sometimes be feasible to use site-specific methodologies, while in other cases it will be necessary or more appropriate to adopt a generic comparison basis.  In spite of such limitations, the high levels of uncertainty and the significant debates that ensue, decisions must be made to help in our continual effort to reduce the likelihood of negative consequences associated with resource consumption and chemical emissions.  Further detailed information of the discussions outlined in this paper can be found in the attached references and through ongoing activities like those of SETAC�fs Working Group on Life-Cycle Impact Assessment.e.g. 8, 9.