A spatio-temporal land use regression model to assess street-level exposure to black carbon

Estimation of exposure to air pollution using land use regression (LUR) models often focuses on spatial variation in (annual) average concentration. However, temporal variability is known to be an important factor for exposure. To estimate the short-term street-level exposure to black carbon (BC), we build a spatio-temporal LUR model by including time-dependent variables as predictor variables. We developed and evaluated the model based on data from an opportunistic mobile monitoring campaign in which city employees measured black carbon (BC) during their surveillance tours. Exposure estimates based on the hourly LUR model are more accurate than those based on a fixed site monitoring station or on a spatial LUR model, and can be used to estimate exposure of cyclists or pedestrians to traffic-related pollution based on a GPS track. We demonstrate the potential of building a real-time dynamic pollution map based on unstructured opportunistic measurements to provide personalized exposure information