Comparison of GOCI and Himawari-8 aerosol optical depth for deriving full-coverage hourly PM2.5 across the Yangtze River Delta

The aerosol optical depth (AOD) data from the Geostationary Ocean Color Imager (GOCI) and the Himawari-8 are valuable for deriving hourly ambient PM2.5 concentrations for assessing acute human exposure in East Asia. This study aims to comparatively evaluate the performance of these two AOD datasets for estimating the hourly PM2.5 on a 1-km grid by using the nonparametric approach with two random-forest submodels. The full-coverage AOD dataset was generated with the first submodel, followed by the PM2.5 estimation using the second submodel. For the Yangtze River Delta (YRD) in 2017, the validation R2 of filling AOD gaps in the GOCI and Himawari-8 was 0.992 and 0.978, respectively. Estimating the hourly PM2.5 concentrations by using the GOCI and Himawari-8 had similar performance, with the cross-validation R2 of 0.860 and 0.862, respectively. Because the PM2.5 predictions based on these two AOD datasets were almost identical, they were fused with the inverse-variance-weighting method to analyze the spatiotemporal patterns of PM2.5. The annual average hourly PM2.5 across YRD was the highest around 08:00 (45.9 μg/m3) and the lowest around 16:00 (39.0 μg/m3). The cumulative acute exposure assessment shows that approximately 21% of the YRD population was exposed to ambient PM2.5>250 μg/m3 for more than 10 h during 2017. This study demonstrates that the GOCI and Himawari-8 datasets are equally adequate to estimate 24-h full-coverage PM2.5 concentrations for air quality management and human health risk assessments.