Avaliação de modelos de dispersão atmosférica no contexto de avaliação de impacto do ciclo de vida no Brasil

Abstract

Air pollution is one of the main environmental and public health challenges, with Particulate Matter (PM2.5) being one of the most critical pollutants due to its adverse impacts. This study investigated the modeling of PM2.5 concentrations in Brazil, considering regional specificities, aiming to improve Life Cycle Impact Assessment (LCIA). Different pollutant dispersion models were mapped, and the InMAP model was selected to simulate PM2.5 concentrations. The evaluated inventories covered five key sectors: agriculture, industry, transportation, biomass burning, and biogenic emissions. For the agricultural sector, the EDGAR inventory was used; in the industrial sector, the EDGAR, BRAIN, and Rey (2023) inventories were considered; and for the transportation sector, the EDGAR and BRAVES inventories were used. Biomass burning was represented by the FINN inventory, while biogenic emissions were represented by the MEGAN inventory. From these sources, six datasets (C_01 to C_06) were formulated and applied in the InMAP model, subsequently validated using the Root Mean Square Error (RMSE) against monitored concentration data from 31 stations in the state of São Paulo. The analysis revealed a significant lack of specific data in key sectors, such as agribusiness, highlighting the urgent need to develop more detailed emission inventories in Brazil. The BRAVES inventory stood out for its superior representativeness, being used in the three best-performing combinations of inventories, suggesting its suitability for the Brazilian context. The southern and southeastern regions of Brazil, especially urban areas such as the São Paulo metropolitan area, showed the highest PM2.5 concentrations, reflecting the strong influence of transportation and industrial emissions. Among the six evaluated inventory sets, C_03 was identified as the most robust. This set combined data from EDGAR for agriculture and industry, BRAVES for transportation, FINN for biomass burning, and MEGAN for biogenic emissions, presenting the lowest RMSE and the greatest consistency among the inventories used. The study emphasized the importance of uniformity in data collection to increase the accuracy of simulations, indicating that the adoption of data collection standards is essential to improving the modeling of emission inventories in Brazil.