Page 293 - Materials Chemistry, Second Edition

11 Uncertainty Management and Sensitivity Analysis 279

straightforward to measure and well reproducible in a laboratory, or in other words
it is a precise indicator. It indicates the potential absorption of energy in molecules
in the atmosphere, but it does not inform us on its impact on the environment or
human health, or in other words it is not accurate regarding the goal of quantifying
potential environmental impacts. Most toxicity midpoint indicators, however,
quantify statistically how many disease cases (or affected species) may potentially
occur in a human (or ecosystem) population per mass emitted. Therefore, toxicity
indicators are much more representative regarding the consequences of a potential
impact than GWP, or in other words a toxicity midpoint indicator has a higher
environmental relevance than GWP and may thus actually be more accurate than
GWP, while being less precise. Only the inherent, and most likely often uncon-
scious, assumption of causal links between radiative forcing––increased tempera-
ture––melting polar caps––rising sea levels––more extreme weather events––loss
of agricultural yield––increased competition for food––starvation and possibly even
war––and thus effects on human health makes this indicator useful for LCA, but
does it make it actually less uncertain for indicating a potential environmental or
human health impact?
The argument of too high uncertainty of toxicity indicators thus refers to their
precision (reproducibility), but not necessarily to their accuracy (in representing
environmental impacts) and may hence be misleading. In addition, the spread
between the highest and the lowest values for an indicator may differ widely
between impact categories. Given that the toxicity-related impact categories cover
several thousand elementary flows (i.e. chemical emissions) with different envi-
ronmental mechanisms, related variability is higher by several orders of magnitude
than for impact categories only covering a handful of elementary flows (e.g. climate
change including *50 chemicals). An example of the relationship between
uncertainty around results for a single chemical and spread of results across
chemicals is given in Chap. 31, Fig. 31.7.
In LCA, uncertainty should always be referring to what a study aims to quantify.
The environmental relevance of indicators varies greatly among impact categories
and is also a source of uncertainty towards the conclusions of a study. Just because
this uncertainty is not quantified or even somewhat unconscious, that does not mean
that it is not present. Hence, a direct comparison of purely precision-related
uncertainty among midpoint indicators is not meaningful unless the compared
indicators have a similar level of accuracy (i.e. environmental relevance).
This brings us to another common misconception about the uncertainty of LCA
indicators, namely the choice of using midpoint or endpoint indicators. The typical
trade-off between both options is that a midpoint indicator result will be more
precise but less environmentally relevant, while it will be the opposite for an
endpoint indicator (i.e. less precise but more environmentally relevant). Therefore,
endpoint indicators are typically perceived as more uncertain based on their usually
lower precision (due to a larger number of choices and hypotheses involved in their
modelling compared to midpoint indicators). When considering environmental
relevance as a measure of accuracy and a type of uncertainty (as discussed further
below), it is important to keep in mind that midpoint indicators have a large portion