Comparison of PM2.5 around 1893 elementary schools and kindergartens in Tehran over different time windows

Abstract

We employed a land use regression (LUR) model to estimate ambient fine particulate matter (PM2.5) concentrations around elementary schools and kindergartens across Tehran, utilizing 138 predictor variables within buffers ranging from 100 to 2000 m. Among these, nine variables predicted the annual ambient PM2.5 concentration around elementary schools and kindergartens. The model demonstrated acceptable performance, as indicated by the magnitude of the coefficients of determination (R2 and adjusted R2) and validation metrics such as K-fold cross-validation (K-foldCV) and leave-one-out cross-validation (LOOCV). R2, adjusted R2, K-foldCV R2 and LOOCV R2 were 0.74 and 0.68, 0.68, and 0.55, respectively. The predictor variables included green space, population density, the distance to secondary roads, water channels, fuel/gas stations, main squares, the number of parking lots and mosques. There is a substantial spatial inequality in annual concentration of ambient PM2.5 across Tehran as nearly all of the schools situated in the north experienced lower levels (< 35 μg/m3) compared with those in the south (> 40 μg/m3). This pattern observed for the Kindergartens across Tehran. Our findings highlight the importance of infrastructure design changes, such as expanding green spaces and relocating parking lots, to enhance air quality around schools and kindergartens.