Decoupling between PM2.5 concentrations and aerosol optical depth at ground stations in China

Abstract

Surface PM2.5 concentrations and aerosol optical depth (AOD) are two air pollution metrics tightly connected. Many studies have used AOD to derive PM2.5 concentrations without investigating their inconsistencies. Here, we explored the associations between surface PM2.5 and AOD using ground-level data from 19 stations in China during 2017–2019. Unexpectedly, we found low correlation coefficients of 0.03–0.60 between daily PM2.5 and AOD for most sites. Such decoupling between PM2.5 and AOD is further compared to simultaneous meteorological factors such as air temperature, specific humidity, sea level pressure, and wind speed. We found that specific humidity dominates the correlations with normalized PM2.5-AOD differences at 14 out of 19 sites. On average, specific humidity increases from 2.83 g kg−1 for the cases with low AOD but high PM2.5–11.89 g kg−1 for those with high AOD but low PM2.5, indicating that hygroscopic growth of aerosols may play an important role in decoupling the associations between PM2.5 and AOD. Random forest (RF) models using AOD as the only input yield a low R of 0.49 between the predicted and observed PM2.5 concentrations. The inclusion of specific humidity in the RF model increases the R to 0.74, close to the R of 0.81 with three additional meteorological factors. Our study revealed a strong decoupling between PM2.5 and AOD and suggested including specific humidity as a key parameter in the retrieval of long-term PM2.5 using AOD data in China.