The correction of TEOM $${\hbox {PM}}_{10}$$ measurements at different monitoring sites and climates

Abstract

The reference method for PM 10 and PM 2.5 measurement in Europe is active sampling with subsequent gravimetric analysis. Also tapered element oscillating microbalance (TEOM) monitors are widely used across Europe for the assessment of particulate matter (PM), even though they are known to be biased due to the loss of semi-volatile materials attributed to sample heating. In our work we correct TEOM PM 10 measurements at eight different monitoring sites located in different surroundings and different climates across Slovenia to replicate reference measurements. We use simultaneous reference and TEOM measurements from a period of seven years (2011-2017) and assess the advantage of custom defined correction factors (with linear and orthogonal regression) compared to default correction factors. We further try to improve the corrections by adding meteorological parameters as inputs and training a linear model (lasso) and nonlinear model (random forest). Random forest and lasso models also enable us to evaluate the impact of different meteorological parameters or chemical compounds on TEOM measurements. Our results show that TEOM measurements can be efficiently corrected using correction factors defined with two linear regressions (for summer and winter) at most locations across continental Slovenia. Koper, which lies at the sea side, was the most problematic measurement site. There the measurements were the most affected by the meteorological situation, and they could not be successfully corrected. When examining the effect of the chemical composition at two different locations, levels of NH + 4 , NO − 3 , and organic carbon exhibited major impact on the discrepancies between reference and TEOM measurements.