Correlation analysis of life cycle impact assessment methods and their impact categories in the food sector: representativeness and predictability of impact indicators

In the past three decades, a variety of life cycle impact assessment methods have been developed and applied to industrial sectors, but the challenge of applying the assessments is still posed. Although former practitioners have attempted to screen impact categories and indicators from the existing methods so as to represent sustainability, the studies have been rarely discussed within the food sector. Therefore, this present study aims to investigate the representativeness and predictability of impact indicators in the food sector, based on a correlation analysis between many indicators.

This study employs nine methods from life cycle impact assessment methods 2.1.1 package belonging to openLCA 1.10.3 software (the open source software updated in 2020, GreenDelta, Berlin). One hundred twenty-nine impact indicators in the methods are calculated by the LCA software for 1008 food products in the French database AGRIBALYSE v3.0.1, which are considered in the correlation analysis between the indicators using the statistical analysis, R 3.6.1.

The results indicate that since some of the life cycle impact assessment methods have close correlations with other methods and their indicator combinations do not show dependencies, such methods could represent the investigated impact categories. In contrast, since some of the methods are not observed close correlations with other methods and their indicator combinations show considerable dependencies, such methods might not represent the investigated impact categories providing an inefficient analysis. The results also show that, on the one hand, strong correlations are observed between the investigated indicators, which reveals the representativeness and predictability between the impact indicators and may reduce the number of the existing indicators, and on the other hand close correlations are not found between the indicators, but it might lead to the emergence of a life cycle impact assessment method from combining the impact indicators in the food sector.

The research findings might lead to the representativeness and predictability between the impact indicators in the food sector while screening impact indicators from the existing methods. However, this present study has some limitations of which, firstly, the analysis should be demonstrated by applying more widely for other sectors and additionally a comprehensive investigation needs to be conducted for all of the existing impact assessment methods. Therefore, a further research should focus on performing the statistical analysis in other industrial sectors and needs to consult with all of the existing impact assessment methods to strengthen the robustness of the research findings.