Quantifying uncertainty for AWARE characterization factors

Although it is not yet current practice in life cycle assessment, it is recommended that impact assessment methods be accompanied by their uncertainty data to better guide the decision maker. This work uses the best available information to assess uncertainty of the AWARE model for water scarcity and corresponding sensitivities of input parameters. An uncertainty estimate for the AWARE characterization factors (CFs) is provided via (1) arrays (5000 values per CF) with statistics, (2) dispersion analysis, and (3) distribution best fit and parameters. Results show that uncertainty, represented by the dispersion of the values, varies significantly around the world and tends to be more 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 found for the spread, represented by the interpercentile range (95%) and interquartile range (25-75%), respectively. The lognormal distribution shows the best fit for most regions around the world and could be used as a default distribution. Two parameters come out as influential: actual water availability (because of precipitation uncertainty) and the global hydrological model itself (because of the variability of results obtained from different models). When compared with uncertainty associated with spatio-temporal variability, uncertainties found in this work are generally lower, and hence improving resolution in water scarcity assessments (to monthly and watershed levels) should remain the priority. Finally, required data for software integration of AWARE uncertainty are provided. This article met the requirements for a Gold-Gold JIE data openness badge described at http://jie.click/badges.