Application of non-linear models to black carbon modelling at urban scale

Abstract

Black carbon (BC) is a health-relevant component of atmospheric particulate matter (PM), present in urban environments. BC is emitted through the incomplete combustion of carbonaceous material and it is typically associated with vehicle exhaust. BC measurements are not always available at urban scale due to the operational cost and complexity of the instrumentation. Therefore, it is advantageous to develop a mathematical model (or BC proxy) to estimate BC concentrations in urban air. This work presents the development and testing of a BC proxy based on a frequentist framework, Support Vector Regression (SVR), using observations of BC, particle mass and number concentrations (N), gaseous pollutants and meteorological variables from a reference air quality monitoring station in Barcelona (Spain) over a 2-year period (2018-2019). Two months of additional data were available from another site in Barcelona, for model validation. The BC concentrations estimated by the adaptive proxy showed a high degree of correlation with the measured BC concentrations (R2 = 0.838–0.878) with a relatively low error (RMSE = 0.27–0.47 μg/m3). Model performance was dependent on seasonality and time of the day, due to the influence of new particle formation events on the input variables. When validated at a different station, performance indicators showed a decrease (R2 = 0.719; RMSE = 1.02 μg/m3) but still an adequate correlation with BC observations. Due to its flexibility and reliability, it is concluded that the model can act as a virtual sensor to complement on-site measurements, for epidemiological and air quality research.