Uncertainty analyses in life cycle assessment (LCA) literature have focused primarily on the life cycle inventory (LCI) phase, but LCA experts generally agree that the life cycle impact assessment (LCIA) phase is likely to contribute even more to the overall uncertainty of an LCA result. The magnitude of perceived uncertainties in characterization relative to that in LCI, however, has not been examined in the literature. Here, we use the pedigree approach to gauge the perceived uncertainty in the characterization phase relative to the LCI phase. In addition, we evaluate the level of approval on the pedigree approach as a means to characterize uncertainty in LCA. Applying the Numeral Unit Spread Assessment Pedigree (NUSAP) approach to environmental risk assessment literature, we extracted the criteria for evaluating the uncertainty in the characterization phase. We used expert elicitation to identify a pool of experts and conducted a survey, to which 47 LCA practitioners from 12 countries responded. In order to reduce personal biases in perceived geometric standard deviation (GSD) values, we used two reference questions on weight and life expectancy at birth for calibration. Nearly half (49%) of respondents expressed their approval to the pedigree matrix approach as a means of characterizing uncertainties in LCA, and responses were highly sensitive to the respondent’s familiarity with the pedigree matrix. For instance, respondents who are highly familiar with the pedigree matrix were more polarized, with 15% and 19% of them expressing either strong approval or strong disapproval, respectively. Respondents less familiar with the pedigree approach were generally more favorable to its use. Compared with LCI, variability in characterization factors was influenced more strongly by geographical correlation and reliability of the underlying model, which showed 11 to 16% larger average GSDs when compared with the comparable criteria for LCI. Conversely, temporal correlation criterion was a less significant factor in characterization than in LCI. Overall, survey respondents viewed LCIA characterization as only marginally more uncertain than LCI, but with a wider variability in responses on characterization than LCI. This finding indicates the need for additional research to develop more thorough methods for characterizing uncertainties in life cycle impact assessment that are compatible with the uncertainty measures in LCI.

中文翻译：

---

LCA 中表征的感知不确定性：一项调查

生命周期评估 (LCA) 文献中的不确定性分析主要集中在生命周期清单 (LCI) 阶段，但 LCA 专家普遍同意，生命周期影响评估 (LCIA) 阶段可能对整个不确定性的影响更大。 LCA 结果。然而，相对于 LCI 中的特征而言，感知到的不确定性的程度尚未在文献中进行检查。在这里，我们使用谱系方法来衡量表征阶段相对于 LCI 阶段的感知不确定性。此外，我们评估了对谱系方法的认可程度，作为表征 LCA 不确定性的一种手段。将数字单位分布评估谱系 (NUSAP) 方法应用于环境风险评估文献，我们提取了评估表征阶段不确定性的标准。我们使用专家启发来确定专家库并进行了一项调查，来自 12 个国家/地区的 47 位 LCA 从业者对此做出了回应。为了减少个人对感知几何标准偏差 (GSD) 值的偏见，我们使用了两个关于出生时体重和预期寿命的参考问题进行校准。近一半 (49%) 的受访者表示赞同将谱系矩阵方法作为表征 LCA 中不确定性的一种手段，并且响应对受访者对谱系矩阵的熟悉程度高度敏感。例如，高度熟悉谱系矩阵的受访者更加两极化，分别有 15% 和 19% 的受访者表示强烈赞同或强烈反对。不太熟悉谱系方法的受访者通常更喜欢使用它。与 LCI 相比，特征因素的变异性受地理相关性和基础模型可靠性的影响更大，与 LCI 的可比标准相比，其平均 GSD 大 11% 至 16%。相反，与 LCI 相比，时间相关性标准在表征中是一个不太重要的因素。总体而言，调查受访者认为 LCIA 表征的不确定性仅略高于 LCI，但与 LCI 相比，表征响应的可变性更大。这一发现表明需要进行额外的研究，以开发更彻底的方法来表征生命周期影响评估中的不确定性，这些方法与 LCI 中的不确定性度量兼容。