||In the context of a global awareness of the climate change issue, carbon footprint has recently become extensively used not only as a marketing tool but also as a simple way to sensitize public opinion. However, limitations in its environmental representativeness might arise if one decides to expand the outlook to include other environmental impacts, which are commonly evaluated in Life Cycle Assessments (LCA): What are the relations between those impact results and the ones from carbon footprint when they are expressed with the same comparative unit? And, to what extent can we state that such product or service is “green”, without omitting a significant part of the environmental burden? This diptych explicitly constitutes the keystone of the present study.
Comparisons were performed through a broad panel of products/services tested in relation to their appearances in a widely used LCA-software database (over 537 processes, mainly from Eco-invent database). Two concrete cases were also built to illustrate the assessments of a family and an institution. The applied method allowed the objective quantification of the difference in magnitude occurring between results from both carbon footprint and LCA-based impact indicators. The objectiveness was achieved using the normalization step for each of the considered impacts, enabling results to be specified in the same unit (Person Equivalent). Relative deviations between normalized results of carbon footprint and those of other impacts were then calculated and used as the basis for the interpretation.
The impact assessments referred to three methodologies: the well-proved EDIP-methodology (EDIP2003, Hauschild et al., 2004), the recently-set USEtox-based methodology (toxicity-related impacts; Rosenbaum et al., 2008), and a new methodology to assess respiratory inorganics caused by inorganics (Humbert et al., 2009). All of them were updated with the latest set of characterization factors and with normalization references for the emission year 2004. Geographical scope was restricted to Europe.
Following this approach, the formation of categories of products/services with relatively similar properties turned out to comprise too large uncertainties due to the dispersion among processes within each of it. Consequently, results could not be exploitable at a high level of accuracy. A compromise between a satisfactory accuracy and a low uncertainty due to dispersion was then reached and privileged: it led to reasoning in terms of orders of magnitude rather than with exact values. Nevertheless, results were generally reported in a way so that the reader can easily get access to information about a particular sub-category of products/services and consult the divergences between impacts results in terms of both exact figures and orders of magnitude.
Furthermore, another source of methodological uncertainties still remained associated with these results, as interim USEtox-based toxicity characterization factors were used for metals (and also for organics regarding marine and terrestrial ecotoxicity). The substantial significance of metals in the assessments clearly emphasized the need for recommended factors. Their predominance also made the comprehensiveness of the inventories in Eco-invent questionable as one might suspect that this is the consequence of several missing organic substances, causing the whole characterization results for toxicity-related impacts to be an underestimation of the actual situation. Unfortunately, the same bias happens in the calculation of normalization references, and therefore, the combined effects arerendered difficult to foresee. It is however certain that this bias might influence the comparisons between the toxicity impact results and carbon footprint ones, as the comprehensiveness of the inventory in the latter is more extensive. Regardless, a bias in normalization due to metals was remediated for all toxicity-related impacts and a work was conducted to check out the USEtox-based ecotoxicity assessment (only freshwater is recommended for use; Rosenbaum et al., 2008) in opposition to the EDIP-based ones. A relatively consistent correlation was observed for aquatic ecotoxicity, with deviations of less than one order of magnitude between both. On the opposite, terrestrial ecotoxicity was credited with a too large dispersion to be exploitable.
Generally, results tended to prove that, in spite of less considered substances, carbon footprinting was close to the LCA-based global warming assessment, though divergences might rise whenever NMVOC showed a peculiarly significant contribution in the inventory. On a broader scale, carbon footprint results put into perspective with other LCA impact ones turned out to reveal that, in most cases, the first were (largely) dominated by toxicity-related impact results. Typically, one order of magnitude was obtained after partial remediation of the bias in normalization. Similar results were observed when considering the concrete assessments of a family and of an institution.
Thus, this study gives the evidence that performing a carbon footprint to assess climate change of a product/service overlooks all other environmental impacts, which are very likely to be relevant if one want to assess the whole environmental burden. This can lead to situations where carbon footprint is misused and inappropriately interpreted, such as a product presenting low CO2 emissions and improperly qualified as “green”, while its environmental burden still remains high due to the significance of other impacts (e.g. freshwater ecotoxicity). Therefore, stakeholders (politicians, companies, people) are recommended to be aware that, though carbon footprint is a first step towards a more “environmental friendly” policy, it does not represent the whole environmental burden and therefore its use must be controlled carefully to avoid misleading, particularly at the level of the public opinion.