Regime-dependent sensitivity of the atmospheric electric field (potential gradient) to anthropogenic air pollution in São Paulo, Brazil

Study Summary

The study analyzes the sensitivity of the atmospheric electric field, measured via the potential gradient (PG), to anthropogenic air pollution in the megacity of São Paulo during the 2018–2024 period, using a long-term dataset under strict fair weather conditions (no precipitation and satellite-based local cloud cover filtering < 20%). Before evaluating correlations with pollutants, the authors build a robust electrical climatology of the urban atmospheric electric field, demonstrating that its diurnal cycle clearly differs from the Carnegie curve in both phase and shape. The atmospheric electric field in São Paulo presents a strong diurnal modulation, with morning maxima and an afternoon depression associated with convective mixing, as well as a greater relative amplitude in winter, when more stable atmospheric conditions prevail.

The correlation analysis reveals a reproducible hierarchy in the electric field–pollutant coupling. Primary combustion gases (NOₓ, NO, and CO) show the strongest associations with the atmospheric electric field, with daily median coefficients around 0.6 under stable conditions. In contrast, PM₁₀ presents a weaker, more variable relationship that is dependent on the dynamical regime. This hierarchy is maintained in both strictly simultaneous subsets and expanded sets, confirming it is not a statistical artifact. Furthermore, the study demonstrates that the coupling is strongly dependent on the diurnal cycle: correlations are high during the night and early morning, when the boundary layer is stable and ventilation is weak, but they decrease markedly in the afternoon due to increased turbulent mixing.

An important contribution of the work is the analysis of time lags. Primary gases (NOₓ and CO) present maximum correlation with no lag, while secondary and particulate species (NO₂ and PM₁₀) show a positive lag close to one hour, suggesting a more immediate electrical response to primary emissions and a more delayed alignment for secondary or particulate processes. Likewise, the study quantifies the modulation by wind speed within the accepted fair weather range: although wind does not directly dominate the variability of the electric field, it systematically weakens the electric field–pollutant coupling as ventilation increases, with the most abrupt effect observed for PM₁₀.

The analysis of anthropogenic perturbations reinforces the physical interpretation. The 2018 truck drivers' strike did not produce a persistent signal in the electric field, likely due to its limited duration. In contrast, during the COVID-19 lockdown in 2020, a sustained reduction in the atmospheric electric field is observed under fair weather conditions, maintaining the diurnal structure but shifting the entire profile toward lower values, providing evidence of sensitivity to a significant decrease in emissions.

Finally, using multiple linear regression and Random Forest models, the predictive capacity of the atmospheric electric field to estimate CO, NO₂, and PM₁₀ concentrations is evaluated. Under stable nocturnal conditions, the models reach R² values close to 0.5, while performance decreases when considering the full diurnal cycle. The similarity between the linear and nonlinear models in the stable regime suggests that the physical coupling is predominantly linear when atmospheric mixing is suppressed.

Overall, the work concludes that the urban electric field should not be interpreted as a universal proxy for pollution, but rather as an electrical indicator dependent on the dynamic regime of the atmospheric boundary layer, whose sensitivity to anthropogenic emissions preferentially emerges under stable conditions and weak ventilation in a megacity environment.

How to cite this work: