Street dust pollution by heavy metals: a geographically weighted regression approach in Mexico City

Street dust pollution by heavy metals has raised concerns because of its potentially harmful effects on the population. It has been suggested in the literature that the spatial distribution of heavy metals in street dust is associated with the urban environment. However, robust spatial econometric analyses have not been applied yet. The study of the spatial distribution of street dust load is also often overlooked. Thus, using previously collected data in Mexico City, a spatial econometric approach was applied to analyze the association between the built environment and street dust heavy metals pollution. Firstly, spatial clusters of street dust load were identified. Then, bivariate plots (street dust load vs. metal content) were analyzed for a broad set of metals. Log-transformed Ordinary Least Squares regression models were tested to make statistical inferences about built environment determinants of heavy metal concentrations in street dust. Finally, the non-stationary property of these regression coefficients was analyzed using geographically weighted regression models. One cluster of high dust load in the east and another with low dust load in the southwest were found. Traffic-related metals (Cr, Cu, Pb, and Zn) were identified in the bivariate plots with low R-2 and relatively low residual standard error. The Cu content in street dust had significant associations with several covariates. For example, it increased in areas with factories and high car use. For the rest of the metals (Cr, Pb, and Zn), spatial patterns of regression coefficients were found and interpreted in terms of pollution indicators.