Predicting personal noise exposure from travel behavior and activity patterns in Taiwan

Urban noise exposure varies across microenvironments, activities, and commuting modes, yet most studies rely on static assessments that overlook individual behavior. This study investigates the behavioral and spatial determinants of personal noise exposure and evaluates air pollution as a potential co-exposure factor. Participants were recruited from Taipei (urban, n = 26; training dataset) and Miaoli (rural, n = 9; validation dataset) between June and October 2021. Real-time noise data were collected using portable Class-2 sound meters, while participants recorded location, surroundings, activity, and sound awareness at 30-min intervals over one weekday and one weekend day. Air pollution data were linked using GPS coordinates from Taiwan’s Environmental Protection Administration. The 30-min equivalent noise levels ranged from 31.0 to 82.3 dBA in Taipei and 31.3 to 82.1 dBA in Miaoli. Among four predictive models, the behavioral model incorporating temporal and activity-based variables showed the best performance (adjusted R 2 = 0.564 training; 0.538 validation). Location (partial R 2 = 0.057–0.050), sound awareness (0.053–0.048), and transit activity (0.045–0.040) were key predictors. These findings highlight the importance of integrating behavioral and spatio-temporal data to better understand transport-related noise exposure.