Evaluation and calibration of MERRA-2 and CAMS reanalysis for PM2.5 in a semi-urbanized area in the south of the Amazon

Abstract

Air pollution has significant implications for the climate and poses irreversible risks to human health. The Amazon region of Brazil is severely affected by biomass burning (BB) emissions, yet air quality monitoring remains highly inadequate. Given the scarcity of surface-based observations, reanalysis models have become essential tools for assessing air pollution. Although MERRA-2 and CAMS PM_2.5 products are widely utilized, their validation and comprehensive evaluation for the Amazon Basin remain limited. This study assesses the performance of these products in a semi-urbanized region in the southern Amazon. The calibrated time series was employed to analyze PM_2.5 concentrations from 2003 to 2023. Our results showed satisfactory performance of both products for the 24-h averages of PM_2.5, with linear correlations above 0.76. However, it was found that both products overestimate surface concentrations. MERRA-2 performed better, with approximately 30% lower bias than CAMS. Time series analysis showed that the study area is strongly impacted by emissions BB in the dry period, mainly in August and September. Furthermore, our findings indicate a positive trend in increasing PM_2.5 concentrations, with a notable rise observed since 2014. The average PM_2.5 levels frequently exceed the daily air quality guidelines established by the WHO in 2021. It has been estimated that the population of this region is exposed to concentrations above 15 μg.m⁻³, on average, more than 30 days per year. Our results contribute to the evaluation of MERRA-2 and CAMS products for Amazon and provide a corrected estimate for surface PM_2.5. Recent concerns about air quality and the implementation of new surface monitoring networks may improve the evaluation of reanalysis products. In the short term, the need for this information makes our assessments indispensable.