Abstract
Assessing children's exposure to air pollutants remains a critical challenge due to their dynamic daily routines and the complexity of pollutant behaviour across microenvironments. This study compared seven methodologies for estimating personal exposure to nitrogen dioxide (NO
), sulfur dioxide (SO
), and ozone (O
): passive samplers (PS), fixed-site monitoring (FSM), the CALPUFF and CMAQ models, and their indoor-outdoor (I/O) ratio counterparts. Measurements were conducted across four urban sites, integrating questionnaire-based activity patterns and pollutant-specific I/O ratios. Evaluation metrics included the fraction of predictions within a factor of two (FAC2), the normalized mean square error (NMSE), and a multicriteria decision matrix ranking each method by accuracy, data requirements, and operational feasibility. Results showed that no single method was universally optimal. FSM × I/O achieved the highest overall balance between accuracy and feasibility for primary traffic-related pollutants (NO
, SO
), while CMAQ×I/O best represented secondary, regionally driven pollutants (O
). CALPUFF captured near-road dispersion effectively but exhibited high sensitivity to local emissions and meteorology. Home environments dominated total exposure (82-91%), followed by schools (8-14%) and commuting pathways (<5%). The decision matrix confirmed FSM × I/O and CMAQ×I/O as the most reliable approaches for large-scale exposure assessments in indoor-dominated populations. These findings demonstrate the necessity of integrating I/O ratio and human activity data into exposure models and provide a transparent framework for selecting optimal methods according to pollutant behaviour, data availability, and research objectives.