Is a Land Use Regression model capable of predicting cleanest home-to-school routes?

Introduction. In urban environments, the highest levels of air pollution usually occur during morning rush hours (MRH), while children go to school. Here, we study whether a Land Use Regression (LUR) model can be used as a tool to mitigate personal exposure of schoolchildren. Methods. A LUR model to study the spatial distribution of black carbon (BC) was developed based on 34 monitoring sites made available by citizens living in an elementary school catchment area in Milan, Italy. BC was monitored in Winter 2018 by means of micro-aethalometers at high temporal resolution (5 minutes). The model was validated using both leave-one-out cross validation (LOOcV) and personal exposure data collected in Winter 2019 by 43 schoolchildren. Results. BC concentrations during MRH in the school catchment area varied with mean ± SD of 4.3 ± 0.7 µg/m3. The LUR model (R2=0.65, LOOcV R2=0.51) shows that BC variability is well explained only by traffic predictors. Personal exposure varied widely across the entire monitoring period with mean ± SD of 9.0 ± 4.9 µg/m3 and an intraday variation of 3.4 ± 4.8 µg/m3. The comparison between estimates and measures showed good agreement (Pearson’s r=0.74, Lin’s Concordance Correlation Coefficient=0.6). However, the model tends to underestimate absolute concentrations by 29%. Conclusion. Our results suggest that a LUR model could be a valuable tool in the attempt to mitigate personal exposure of schoolchildren by identifying the cleanest routes to school.