Impacts of uncertainties in emissions on aerosol data assimilation and short-term PM2.5 predictions over Northeast Asia

To improve PM2.5. predictions in Northeast Asia, we estimated a new background error covariance matrix (BEC) for aerosol data assimilation using surface PM(2.5.)observations. In contrast to the conventional method of BEC estimation that uses perturbations in meteorological data, this method additionally considered the perturbations using two different emission inventories. By taking the emission uncertainty into account, we found that the standard deviations in the BEC were significantly increased. The standard deviations became around three times larger than those in the conventional method at the surface. The impacts of the new BEC were then tested for the prediction of surface PM2.5. over Northeast Asia using the Community Multiscale Air Quality (CMAQ) model initialized by three-dimensional variational method (3D-VAR). The surface PM(2.5.)data measured at 154 sites in South Korea and 1535 sites in China were assimilated every 6 h during the campaign period of the Korea-United States Air Quality Study (KORUS-AQ) (1 May-14 June 2016). The data assimilation with the new BEC showed better agreement with the surface PM(2.5.)observations than with the BEC from the conventional method. Our method was also more consistent with the observations in 24-h PM(2.5.)predictions than the conventional method (specifically, with a ~44% reduction of negative biases). We concluded that increased standard deviations, together with updated horizontal and vertical length scales in the new BEC, improved the data assimilation and short-term predictions of the surface PM2.5. This paper also suggests several research efforts to further improve the BEC for better short-term PM2.5 predictions in Northeast Asia.

Automated summary

This study improves the predictions of PM2.5 in Northeast Asia by estimating a new background error covariance matrix for aerosol data assimilation using surface PM2.5 observations. The method takes into account emission uncertainties and shows better agreement with surface PM2.5 observations compared to the conventional method.