Impact of 3DVAR assimilation of surface PM2.5 observations on PM2.5 forecasts over China during wintertime

Publication date: August 2018
Source:Atmospheric Environment, Volume 187
Author(s): Shuzhuang Feng, Fei Jiang, Ziqiang Jiang, Hengmao Wang, Zhe Cai, Lin Zhang
Data assimilation is one of the effective ways to improve model predictions. In this study, surface fine particulate matter (PM2.5) observations during 16 December 2015 to 15 January 2016 over China are assimilated in a regional air quality forecasting system using the three-dimensional variational (3DVAR) method. Two parallel experiments with and without data assimilation (DA) are conducted. The results show that 3DVAR can significantly reduce the uncertainties of the initial PM2.5 fields and improve the subsequent PM2.5 forecasts at a certain extent. The influences of DA on both analysis and forecast fields are different in different areas. Overall, the root-mean-square error of analysis field could be reduced by at least 50%, and the correlation coefficient could be improved to more than 0.9. Less improvement appears in the North China Plain. For forecast field, similar with previous studies, the DA is effective only within a certain forecast time. On average, the benefits of DA could last more than 48 h over China. Much longer benefits (>24 h) are found in Sichuan basin, Xinjiang, southern China and part of northern China. In the first 24 h, there are more than half of Chinese cities with their daily mean PM2.5 hit rates increasing greater than 10%. The duration of DA benefits are mainly affected by weather condition and emission intensity. The areas with longer DA benefits generally have more stable weather condition and/or weaker emission intensity. The absence of heterogeneous reactions in chemical transport models may also has negative effects on the durations. In addition, we found that the assimilated observation information could transport along with the air masses, and the downwind areas generally have better DA benefits, indicating that when doing air quality forecasting using nested domains, we should conduct the DA in the largest domain rather than the innermost one.



https://ift.tt/2kMdErb