Machine learning-optimized interpretability analysis for identifying key drivers of NO₃ lifetime variability.

Area of Science:

• Atmospheric Chemistry
• Environmental Science
• Air Quality Research

Background:

• Nitrate radical (NO3) is a key nocturnal oxidant influencing atmospheric chemistry.
• Understanding NO3 lifetime is crucial for modeling atmospheric processes and air quality.
• Coastal regions experience complex interactions between anthropogenic emissions and natural air masses.

Purpose of the Study:

• To continuously measure nocturnal NO3 concentrations and determine its lifetime in Shanghai Dongtan Wetland Park.
• To identify the primary drivers influencing NO3 lifetime using a machine learning approach.
• To investigate how air mass origins affect NO3 loss mechanisms and lifetime.

Main Methods:

• Continuous measurement of NO3, NO2, and O3 from January 2022 to December 2023.
• Steady-state analysis to calculate NO3 lifetime.
• Application of machine learning-assisted SHAP interpretability for driver analysis.

Main Results:

• Nocturnal NO3 levels ranged from 4.7 to 300 pptv; >50% of NO3 lifetimes were shorter than 120 seconds.
• NO2, relative humidity (RH), and wind direction (WD) were identified as key drivers of NO3 lifetime.
• NO3 lifetime varied significantly with air mass origin, from seconds in polluted air to ~13 minutes in clean ocean air.

Conclusions:

• Air mass origin profoundly impacts NO3 loss mechanisms and lifetime at coastal sites.
• Reduced NO2, altered RH effects, and lower PM2.5 contribute to longer NO3 lifetime in clean air masses.
• This study provides critical insights into nitrogen chemistry dynamics influenced by emissions and transport in coastal environments.