Predição de ozônio troposférico no monitoramento da qualidade do ar

Abstract

Resulting of the chemical reaction between primary pollutants, tropospheric ozone (O3) causes extensive damage to human health, vegetation and animals due to its high oxidative power. Sensors for accurately measuring O3 are expensive, difficult to handle, and require regular maintenance, which renders its availability at air quality stations difficult, unlike those used to measure the elements that compose O3. In this paper, we use a forecasting time series method developed by the Facebook - Prophet model. The approach is to apply this model to real data collected during the period of 2018 to 2020 from an air quality station in Parana, Brazil, to predict O3 from meteorological variables combined with as few sensors of primary pollutants as possible but aiming for accurate results.