Tropospheric NO2: Anthropogenic Influence, Global Trends, Satellite Data, and Machine Learning Application

Abstract

Nitrogen dioxide (NO2) is a critical air pollutant that has significant health and environmental impacts. Tropospheric NO2 refers specifically to the vertical column density of NO2, which is measured by satellites and serves as an indicator of anthropogenic NO2 sources. This pollutant is frequently assessed using satellite data owing to limitations in local monitoring. This investigation employs the Spectral Angle Mapper (SAM), a geometric machine-learning model, given its advantages in simplicity and computational efficiency, and OMI satellite measurements to carry out spatially supervised classification of tropospheric NO2 global patterns from 2005 to 2021. This study identifies four typical trends across developed urban centers, examining correlations with population growth, economic factors, and air quality policies. The results demonstrated regional variations, with a general downward trend in North America, Europe, and parts of Asia, underscoring the efficacy of stricter emission controls. However, upward trends persist in some Asian regions, reflecting varying policy implementations. This study revealed a pivotal inflection point around 2013, marking a shift in global NO2 dynamics. Although policies have led to improved air quality in some regions, achieving absolute decoupling of economic growth from NO2 emissions remains challenging. The COVID-19 pandemic has also exerted a significant influence, temporarily reducing emissions due to economic slowdowns. Overall, the SAM model effectively delineated NO2 patterns and provided insights for future policy and emission control strategies.