Quantifying uncertainty for AWARE characterization factors

Although it is not yet current practice in life cycle assessment, it is recommended thatimpact assessment methods be accompanied by their uncertainty data to better guidethe decision maker. This work uses the best available information to assess uncertaintyof the AWARE model for water scarcity and corresponding sensitivities of input param-eters. An uncertainty estimate for the AWARE characterization factors (CFs) is pro-vided via (1) arrays (5000 values per CF) with statistics, (2) dispersion analysis, and(3) distribution best fit and parameters. Results show that uncertainty, representedby the dispersion of the values, varies significantly around the world and tends to bemore important in regions of higher scarcity and low in most regions around the world(area based) in terms of absolute spread. Globally, values of 18.8 and 66.28 are foundfor the spread, represented by the interpercentile range (95%) and interquartile range(25–75%), respectively. The lognormal distribution shows the best fit for most regionsaround the world and could be used as a default distribution. Two parameters comeout as influential: actual water availability (because of precipitation uncertainty) andthe global hydrological model itself (because of the variability of results obtained fromdifferent models). When compared with uncertainty associated with spatio-temporalvariability, uncertainties found in this work are generally lower, and hence improv-ing resolution in water scarcity assessments (to monthly and watershed levels) shouldremain the priority. Finally, required data for software integration of AWARE uncer-tainty are provided. This article met the requirements for a Gold-Gold JIE data open-ness badge described at http://jie.click/badges.

Mots clés

• characterization factor
• impact method
• industrial ecology
• uncertainty analysis
• water scarcityfootprint