This study aims to develop a robust machine learning framework for estimating ground-level nitrogen dioxide (NO₂) concentrations by integrating satellite remote sensing with meteorological reanalysis data, addressing critical needs in environmental monitoring and air quality assessment. While satellite-based column measurements provide extensive spatial coverage, their conversion to ground-level concentrations remains challenging due to complex atmospheric dynamics. This study provides the first dedicated assessment for Napoli Province, offering locally validated air quality estimation for this complex Mediterranean coastal-urban environment. A Random Forest regression model was developed using 11,325 quality-controlled observations from 16 monitoring stations. The model incorporated 12 predictor variables including TROPOMI tropospheric NO₂ column density, planetary boundary layer height, temperature, relative humidity, surface pressure, wind speed and direction, solar radiation, precipitation, elevation, and cyclical temporal features. The model achieved an R² of 0.71 and RMSE of 8.61 μg/m³ on the independent test set (n=2,832), demonstrating robust predictive capability. Ground-level NO₂ concentrations averaged 29.3 ± 16.1 μg/m³, with values ranging from 0 to 112.8 μg/m³. These results demonstrate the effectiveness of combining space-based observations with meteorological data for high-resolution air quality monitoring, providing a scalable approach for environmental situational awareness in regions with limited ground-based infrastructure.
Qamar et al. (Wed,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: