Urban PM2.5 concentration monitoring: A review of recent advances in ground-based, satellite, model, and machine learning integration

Urban aerosols, especially fine particulate matter (PM_2.5), significantly affect public health and environmental quality. Accurate high-resolution monitoring of PM_2.5 is essential for exposure assessment, regulatory enforcement, and policy development. This review synthesizes recent advances in the integration of ground-based observations, satellite remote sensing, Chemical Transport Models (CTMs), and Machine Learning (ML) techniques for characterizing the spatio-temporal distribution of urban aerosols. Ground-based sensors provide accurate surface-level measurements but lack broad spatial coverage. In contrast, satellite-retrieved Aerosol Optical Depth (AOD), proxy to retrieve PM_2.5 concentration at surface, offers extensive coverage, but with limitations related to cloud cover and temporal resolution. CTMs provide continuous 3D aerosol fields, though their accuracy is limited by uncertainties in emissions and meteorology. ML algorithms effectively integrate these heterogeneous data sources, capture complex nonlinear relationships, and improve PM_2.5 predictions. Case studies from multiple global regions demonstrate that integrated approaches achieve high accuracy (cross-validated $R^2\approx 0.80$, Root Mean Square Error 2.5–$3.0\mu g$/m${}^3$, Mean Absolute Error 2.1–$2.7\mu g$/m${}^3$), enabling daily exposure estimates at fine spatial scales. These synergistic methods are increasingly being used in air quality policies, health risk assessments, and regulatory frameworks. Future directions include the development of physics-informed ML models, the deployment of Internet of Things (IoT)-enabled sensor networks, and the establishment of standardized uncertainty quantification frameworks. This review is intended for researchers and policy makers seeking a state-of-the-art perspective on urban aerosol monitoring.