Comparative Analysis of Urban Air Quality Monitoring Techniques Across Cities

Thesis Abstract

Urban air quality monitoring has become a critical challenge for rapidly expanding cities, where inconsistent monitoring practices impede the comparability and reliability of pollution assessments across urban cores and peri-urban areas. This study addresses the problem of heterogeneity in monitoring techniques that undermines cross-city decision-making, policy evaluation, and public health risk assessment. The aim is to evaluate and compare the performance, data quality, and operational feasibility of diverse urban air quality monitoring techniques across multiple metropolitan contexts to identify best practices and scalable strategies. Specific objectives are (1) to inventory and categorize monitoring techniques—fixed-site reference stations, low-cost sensor networks, satellite-derived estimates, and mobile monitoring campaigns—across ten large cities; (2) to assess data quality attributes including accuracy, precision, spatial representativeness, and temporal resolution; (3) to evaluate the compatibility of data streams using inter-method agreement analysis and to quantify biases introduced by each technique; (4) to analyze the influence of urban form, meteorology, and traffic patterns on inter-method discrepancies; and (5) to develop a framework for integrative data fusion that improves city-scale air quality characterization. The research adopts a multi-method comparative design grounded in the theoretical framework of data fusion and quality assessment models, drawing on the Theory of Measurement Validity and the Data Fusion Theory to interpret method complementarities. The population comprises urban air quality monitoring initiatives in ten global cities with diverse geographic, climatic, and regulatory contexts. A stratified sampling approach targets sensor networks and data products from (a) fixed-site regulatory monitors (n?40 stations per city), (b) low-cost sensor networks deployed in selected neighborhoods (n?100–250 per city, depending on network scale), (c) satellite-based aerosol optical depth products and tropospheric column concentrations, and (d) mobile monitoring campaigns collecting spatially explicit measurements along representative transects. Data collection instruments include calibrated reference monitors, co-located low-cost sensors with traceable calibration protocols, high-resolution satellite data, and standardized mobile measurement kits. Instrument validity and reliability are ensured through cross-validation with collocated reference stations, calibration against gravimetric particulates, and inter-instrument reproducibility tests following ISO 17025-compatible procedures. Data analysis proceeds in several stages. Descriptive statistics summarize central tendency, dispersion, and data availability across methods. Inter-method agreement is quantified using intraclass correlation coefficients (ICCs) and Bland-Altman analyses for PM2.5, PM10, NO2, and O3 species. Regression-based calibration models (partial least squares and ridge regression) evaluate correction factors for low-cost sensors against reference data, while ANOVA and mixed-effects models assess the influence of city-level covariates such as urban density, green cover, and traffic intensity on measurement discrepancies. Spatial analyses utilize geostatistical techniques (ordinary kriging and co-kriging) to compare spatial fields derived from different methods, and data fusion is explored through ensemble learning approaches (stacked generalization) and Bayesian data fusion to produce harmonized city-wide exposure surfaces. Sensitivity analyses test the robustness of results to missing data, sensor drift, and temporal misalignment. Anticipated findings indicate that low-cost sensor networks, when properly calibrated and co-located with reference monitors, can yield spatially rich exposure estimates with acceptable accuracy for neighborhood-level assessments, yet exhibit systematic biases that vary by pollutant and meteorological conditions. Satellite-derived products provide robust regional context but require downscaling with local predictors to achieve city-scale relevance. Mobile campaigns capture hot spots and episodic events that fixed stations may miss, thereby improving spatial coverage. The study is expected to produce a validated integrative framework for data fusion that enhances comparability across cities, reducing uncertainty in cross-city policy evaluations and health impact assessments. The study contributes to knowledge by operationalizing a cross-city comparison of monitoring modalities, offering standardized metrics for data quality and compatibility, and proposing a scalable integrative framework for urban air quality assessment. Practical implications include guidance for city planners, air quality managers, and public health authorities on selecting monitoring configurations, calibrating emerging sensing technologies, and implementing harmonized data products for inter-city collaborations. The main conclusion anticipates that a hybrid monitoring strategy combining calibrated low-cost sensors, satellite-derived context, and targeted mobile measurements, all integrated through Bayesian data fusion, yields the most reliable cross-city air quality characterizations and informs more equitable, evidence-based urban air quality management. Recommendations emphasize investment in calibration protocols, open data sharing, and capacity-building to sustain harmonized air quality monitoring and cross-city comparability.