Integrating Chemical Mechanisms and Feature Engineering in Machine Learning Models: A Novel Approach to Analyzing HONO Budget

Abstract

Nitrous acid (HONO) serves as the primary source of OH radicals in the atmosphere, exerting significant impacts on atmospheric secondary pollution. The heterogeneous reactions of NO 2 on surfaces and photolysis of particulate nitrate or adsorbed nitric acid are important sources of atmospheric HONO, yet the corresponding kinetic parameters based on laboratory investigations and field observations exhibit considerable variations. In this study, we developed an explainable machine learning model to analyze the HONO budget using two years of summer urban supersite observations. By integrating chemical mechanisms and feature engineering into our machine learning model, we assessed the contributions of different sources to HONO and inferred the kinetic parameters for the primary HONO formation pathways, thereby partially addressing the limitations associated with predetermined rate coefficients. Our findings revealed that the primary source of daytime HONO in the summer was the photolysis of nitric acid adsorbed on both aerosol and ground surfaces, accounting for over 40% of its unknown sources. This was followed by the photoenhanced heterogeneous conversion of NO 2 and the photolysis of particulate nitrate. Additionally, we derived the corresponding kinetic parameters, analyzed their influencing factors, and confirmed that machine learning methods hold great potential for the study of the HONO budget.