An optimized approach for estimating benzene in ambient air within an air quality monitoring network.

Benzene is a carcinogenic air pollutant for which European legislation has set an annual limit and criteria for the number of fixed monitoring sites within air quality networks (AQMN). However, due to the limited number of fixed sites for benzene measurement, exposure data are lacking. Considering the relationship between benzene levels and other variables monitored within an AQMN, such as NO, O, temperature, solar radiation, and accumulated precipitation, this study proposes an approach for estimating benzene air concentrations from the related variables. Using the data of the aforementioned variables from 23 fixed stations during 2016-2017, the proposed approach was able to forecast benzene concentration for 2018 with high confidence, providing enriched data on benzene exposure and its trends. Moreover, the spatial distribution of the estimated versus the most representative benzene levels was quite similar. Finally, an artificial neural network identified the most representative fixed benzene monitoring sites within the AQMN.