Marginal and non-marginal approaches in characterization: how context and scale affect the selection of an adequate characterization model. The AWARE model example


LCA traditionally has been founded on the ceteris paribus principle, by which the assessed contribution is assumed not to affect the background state, i.e., being marginal. As LCA is increasingly used to assess interventions at larger scales (e.g., territory, sectors), it becomes necessary to provide adequate characterization factors. Applying this concept to the water scarcity footprint AWARE model, this paper has for main objective to provide guidance on the use of different characterization approaches; the resulting interpretation, including in relation to normalization; and the implication for decision making.
The specific case of AWARE is taken, and average factors are calculated by integrating the characterization factor’s equation of the AWARE model with respect to local water consumption, and dividing the total impacts obtained per each cubic meter consumed. The resulting average factors are applied (at the country scale) to European Union countries for the total water consumption, and the results are compared with the same assessment performed using the traditional marginal factors.
Results and discussion
Average CF at the watershed level for AWARE are provided for the country scale. Differences, sometimes significant, are observed between the two sets, with the average factors always being lower than (or equal to) the marginal ones. The rank correlation coefficient (correlation between the watershed values’ rank with both approaches) is of 0.965, and the mean difference coefficient is 0.16 (the larger the value, the more different the datasets, out of a maximum value of 2). For countries presenting areas with potentially more extreme water scarcity, the difference between the two normalization sets is higher, reflecting that there can be significant differences in applying the marginal or average CFs. A set of points for attention for methodological choices are presented and specific recommendations discussed from the perspective of the practitioner. In particular, by building on the shortcomings shown of marginal and average characterization factors, a broader application of LCIA is proposed to large-scale, non-marginal, and prospective assessments.
In conclusion, as goals and scopes of life-cycle-based studies are expanding, it is important to ensure that methodologies used reflect the new applications and the specific context for which LCA is needed. This paper provides the average CF for the AWARE model, which will now allow practitioners to assess water scarcity footprint of large interventions coherently, providing guidance on the implication of the selection of marginal or average CFs and the interpretation thereof. It also provides important guidance for practitioner to apply when using characterization factors of any methods in order to ensure coherence of their interpretation and consistency within their study.