Development of characterization factors for health effects of particulate matter in Brazil

Abstract

Particulate matter (PM) is an atmospheric pollutant that cause adverse health effects and Life Cycle Assessment (LCA) can support the management of PM emissions from production systems. However, there are currently no highly spatialized characterization factors (CFs) available to Brazil. Then, this study aimed to calculate CFs for PM and precursor gas emissions (NH3, SOx, NOx, VOC) for Brazil. For this, the first step was a critical review of existing models, which showed that the models with global coverage and availability of CFs for Brazil, are the most appropriate: Van Zelm et al. (2016), Fantke et al. (2017, 2019), and Oberschelp et al. (2020). However, limitations have been observed. Then, CFs for PM2.5 and precursor gases were calculated per state, mesoregion and nationally, using death results calculated using the InMAP (Intervention model for air pollution) and national emission inventories. At the state level, the highest CFs were observed for SOx (ranging from 1.18×10⁻¹ to 4.09×10⁻⁴ deaths/ton), followed by PM2.5 (9.49×10⁻² to 2.09×10⁻⁴), NH3 (6.38×10⁻² to 1.41×10⁻⁴), NOx (1.35×10⁻² to 5.42×10⁻⁴), and VOCs (2.61×10⁻³ to 1.32×10⁻⁵). The highest CFs were observed in the Southeast region, particularly in São Paulo and Rio de Janeiro. Finally, a sensitivity analysis verified the variation in results of an LCA case study concerning the choice of the characterization model, including the recommended models for Brazil and the CFs calculated in this research. The analysis used four milk production systems as a case study. Spearman’s correlation was calculated to verify the consistency of the milk systems’ ranking. Correlation analysis showed that regionalized CFs presented higher variation in the milk systems’ ranking. The CFs calculated by this research presents high correlation (> 0.9) with global CFs from other models. However, the CFs at state level resulted in low correlation (< 0.5) with all models, except with the state CFs from Oberschelp et al. (2020), the only model with CFs for Brazilian states, which presented correlation indicator equal to 0.9. The consistency between the CFs obtained in this study and those found in the literature reinforces the validity of the results and highlights the importance of using regionalized factors in LCA studies. The main limitations of the CFs are the effects equation based on data from the United States, incorporated into the global InMAP model, and the unavailability of these factors in LCA software. However, the results obtained contribute to advancing knowledge and refining the calculation of health impacts resulting from atmospheric pollutant emissions in Brazil. These contributions can benefit both academia and the business sector by providing data more aligned with national realities for impact management.