Attributing Chemiluminescence NO2 Measurement Interference to Oxidized Reactive Nitrogen Chemistry in Urban Atmospheres: Its Implications for Photochemical Pollution Assessment

Abstract Nitrogen dioxide (NO2) is a key precursor to ozone and aerosols; therefore, it has been widely monitored in air quality networks using the chemiluminescence (CL) method. However, interference from oxidized nitrogen species (NOz) in CL-based measurements can lead to NO2 overestimation, introducing uncertainty into atmospheric composition assessments. Using 2 years of parallel CL and cavity-attenuated phase shift (CAPS) measurements at an urban Seoul, we quantified mean interference (z* = NO2_CL – NO2_CAPS) of 3.0 ppbv, with larger values in winter than in summer. We developed machine learning models that reproduced z* well (r > 0.7). Explainable AI results indicated that model inference process was broadly consistent with heterogeneous NOz formation in winter and photochemical NOz production in summer. Applying our models to global megacities suggests regime-dependent interference of about 10–20% of reported NO2. This study provides a framework for estimating NOz interference in CL-based NO2 measurements.