Green Heart Louisville: Intra-urban, hyperlocal land-use regression modeling of ultrafine particles

Exposure to ultrafine particles (UFP) is increasingly linked to adverse health outcomes. While nation-wide air monitoring networks in the United States do not measure UFP, small-scale measurements have revealed persistent patterns in urban UFP. This project maps hyperlocal UFP in a 12 km2 study area of a health effects study in Louisville, KY, through mobile measurements to elucidate the relationship between the urban landscape and UFP exposures. We measured UFP number concentration along all drivable streets ([~]340 km) during daytime and nighttime on both weekdays and weekends. After deconvoluting UFP levels to isolate local signals from neighborhood and urban signals, we fitted a land-use regression (LUR) model to explain differences in local UFP as a function of characteristics of the built and natural environment. Median UFP in the study domain was 6,850 #/cm3, which is comparable to urban background measured or estimated for other U.S. cities. UFP was higher during the weekend than on weekdays, potentially due to changes in local activity (e.g. increased restaurant hours) apparent at fine spatial scales. The final LUR model explained 61% of the spatial heterogeneity in log(UFP). Leave-one-area-out cross validation revealed overprediction in regions farther from highways and underprediction in regions with dense food service locations and major roads. This suggests that additional mobile measurements to capture longer-term, robust UFP may yield improved models.