Developing High-resolution Air Quality Reanalysis Dataset over China for Years 2013-2018 Based on Ensemble Kalman Filter and Surface Observations from CNEMC

A six-year long high-resolution Chinese air quality reanalysis datasets (CAQRA) covering the period 2013-2018 has been developed in this study by assimilating over 1000 surface air quality monitoring sites from China National Environmental Monitoring Centre (CNEMC) based on the ensemble Kalman filter (EnKF) and the Nested Air Quality Prediction Modeling System (NAQPMS). This reanalysis provides the surface fields of six conventional air pollutants in China, namely PM_2.5, PM₁₀, SO₂, NO₂, CO and O₃, at high spatial (15km×15km) and temporal (1 hour) resolutions. This paper aims to document this dataset by providing the detailed descriptions of the assimilation system and presenting the first validation results for the reanalysis fields of air pollutants in China. A twenty-fold cross validation (CV) method was used to assess the quality of CAQRA. The CV results show that the CAQRA has excellent performances in reproducing the magnitude and variability of the air pollutants in China with the biases (normalized mean bias) of the reanalysis data about -2.6 (-4.9%) μg/m³ for PM_2.5, -6.8 (-7.6%) μg/m³ for PM₁₀, -2.0 (-8.5%) μg/m³ for SO₂, -2.3 (-6.9%) μg/m³ for NO₂, -0.06 (-6.1%) mg/m³ for CO and -2.3 (-4.0%) μg/m³ for O₃. The interannual changes of the air quality in China were also well represented by the CAQRA in terms of the six air pollutants. Comparisons with previous datasets of daily PM_2.5, SO₂ and NO₂ concentrations indicate that the CAQRA is more accurate with smaller RMSE values. We also compared our reanalysis dataset to the CAMSRA (The Copernicus Atmosphere Monitoring Service reanalysis) produced by ECMWF (European Centre for Medium-Range Weather Forecasts), which suggests that the CAQRA has higher accuracy in representing the surface air pollutants in China due to the assimilation of surface observations. This reanalysis dataset can provide us comprehensive pictures of the air quality in China from 2013 to 2018 with a complete spatial and temporal coverage, which can be used in the assessment of health impacts of air pollution, validation of model simulations and providing training data for the statistical or AI (Artificial Intelligence) based forecast.