A three-dimensional variational data assimilation system for a size-resolved aerosol model: Implementation and application for particulate matter and gaseous pollutant forecasts across China

A three-dimensional variational (3DVAR) data assimilation (DA) system is presented here based on a size-resolved sectional aerosol model, the Model for Simulating Aerosol Interactions and Chemistry (MOSAIC) within the Weather Research and Forecasting model coupled to Chemistry (WRF-Chem) model. The use of this approach means that both gaseous pollutants such as SO2, NO2, CO, and O-3 as well as particulate matter (PM2.5, PM10) observational data can be assimilated simultaneously. Two one-month parallel simulation experiments were conducted, one with the assimilation of surface hourly concentration observations of the above six pollutants released by the China National Environmental Monitoring Centre (CNEMC) and one without assimilation in order to verify the impact of assimilation on initial chemical fields and subsequent forecasts. Results show that, in the first place, use of the DA system can provide a more accurate model initial field. The root-mean-square error of PM2.5, PM10, SO2, NO2, CO, and O-3 mass concentrations in analysis field fell by 29.27 mu g m(-3) (53.5%), 34.5 mu g m(-3) (50.9%), 30.36 mu g m(-3) (64.2%), 8.91 mu g m(-3) (39.5%), 0.46 mg m(-3) (47.4%), and 15.11 mu g m(-3) (51.0%), respectively, compared to a background field without assimilation. At the same time, mean fraction error was reduced by 42.6%, 53.1%, 45.2%, 43.1%, 69.9%, and 48.8%, respectively, while the correlation coefficient increased by 0.51, 0.55, 0.48, 0.38, 0.47, 0.65, respectively. Secondly, the results of this analysis reveal variable benefits from assimilation on different pollutants. DA significantly improves PM2.5, PM10, and CO forecasts leading to positive effects that last more than 48 h. The positive effects of DA on SO2 and O-3 forecasts last up to 8 h but that remains relatively poor for NO2 forecasts. Thirdly, the influence of assimilation varies in different areas. It is possible that the positive effects of DA on PM2.5 and PM10 forecasts can last more than 48 h across most regions of China. Indeed, DA significantly improves SO2 forecasts within 48 h over north China, and much longer CO assimilation benefits (48 h) are found in most regions apart from north and east China and across the Sichuan Basin. DA is able to improve O-3 forecasts within 48 h across China with the exception of southwest and northwest regions and the O-3 DA benefits in southern China are more evident, while from a spatial distribution perspective, NO2 DA benefits remain relatively poor.