Estimates of daily ground-level NO2 concentrations in China based on Random Forest model integrated K-means

Abstract

Nitrogen dioxide (NO2) is one of the most important atmospheric pollutants and the precursors of acid rain, tropospheric ozone, and atmospheric aerosols. However, due to the poor quality of source data and the computing power of the models, current ground-level NO2 concentration data lack either high-resolution coverage or full nation-wide coverage. This study estimates the ground-level NO2 concentration in China with national coverage at relatively high spatiotemporal resolution (0.25°; daily intervals) over the newest past 6 years (2013–2018). We developed an advanced model, named Random Forest model integrated K-means (RF-K), for the estimates with multi-source parameters. Besides meteorological parameters, satellite retrievals parameters, and anthropogenic emission inventories parameters, we also innovatively introduce socioeconomic parameters to assess the impact of human activities. Our results show that: (1) the RF-K model developed by us shows better prediction performance than others. (2) the annual average NO2 concentration of China showed a weak declining trend (-0.013±0.217 μgm−3yr−1) from 2013 to 2018, indicating that pollutant controlling targets had been achieved in China overall. By mapping daily nationwide ground-level NO2 concentrations, this study provides high-quality timely, and detailed data for air quality management and epidemiological analyses for China. The RF-K model can be used easily for other pollutants (e.g. SO2 and O3) considering that their ground-level concentrations can be estimated depending on the similar emitting sources and influence factors, and our model's input data sources also cover information on other pollutants.

Keywords

• Ground-level NO2 concentration
• China
• Random Forest model
• Multi-source big data
• Socio-economic parameters