Quantifying wintertime O3 and NOx formation with relevance vector machines

This paper uses a machine learning model called a relevance vector machine (RVM) to quantify ozone (O3) and nitrogen oxides (NOx) formation under wintertime conditions. Field study measurements were based on previous work described by Olson et al. (2019), where continuous measurements were reported from a wintertime field study in Utah. RVMs were formulated using either O3 or nitrogen dioxide (NO2) as the output variable. Values of the correlation coefficient (r2) between predicted and measured concentrations were 0.944 for O3 and 0.931 for NO2. RVMs are constructed from the observed measurements and result in sparse model formulations, meaning that only a subset of the data is used to approximate the entire dataset. For this study, the RVM with O3 as the output variable used only 20% of the measurement data while the RVM with NO2 used 16%. RVMs were then used as a predictive model to assess the importance of individual precursors. Using O3 as the output variable, increases in three species resulted in increased O3 concentrations: hydrogen peroxide (H2O2), dinitrogen pentoxide (N2O5), and molecular chlorine (Cl2). For the two termination products measured during the study, nitric acid (HNO3) and formic acid (CH2O2), no change in O3 concentration was observed. Using NO2 as the output variable, only increases in N2O5 resulted in increased NO2 concentrations.

Automated summary

This paper uses a relevance vector machine (RVM) model to quantify ozone (O3) and nitrogen oxides (NOx) formation under wintertime conditions. The RVMs have high correlation coefficients with measured concentrations and are constructed from observed data, using only a subset to approximate the entire dataset. The RVMs were used as a predictive model to assess the importance of individual precursors in forming O3 and NO2.