Impact of assimilating multi-source observations on meteorological and PM_2.5 forecast over Central China

Highlights

• Assimilating satellite radiances and conventional observations helps meteorological forecasts get better over Central China.

• Based on improved meteorological fields, WRF-Chem model improve PM_2.5 forecasts over the study area.

• Data assimilation also helps analyzing the formation of an air pollution episode over Central China.

Abstract

Data assimilation (DA) is a promising approach to improve meteorological and PM_2.5 forecasts, but to what extent and by what process the DA of meteorological fields helps improve PM_2.5 forecast still call for more discussion. By utilizing WRFDA and WRF-Chem models, we have assimilated AMSU-A, MHS radiances and conventional observations, and studied the influences of the meteorological DA on meteorological and PM_2.5 forecasts over Central China through a series of experiments. The results show that multi-source meteorological DA helps improve temperature and relative humidity forecasts in the lower atmosphere, and the improved meteorological fields further improve PM_2.5 forecast with a reduction of bias and RMSE by 7.4% and 4.1% over the study area, especially during PM_2.5 episode. This study also helps understand how DA improve the PM_2.5 forecasts over Central China.

Introduction

Fine particulate matter (PM_2.5, particles with an aerodynamic diameter ≤ 2.5 μm) is a major concern of the deteriorating air quality. With the rapid economic development and environmental deterioration, China, the largest developing country, is facing extremely severe air quality in the world (Kan et al., 2012; Zhang et al., 2017). Environmental epidemiologic studies show that the exposure to PM_2.5 produces harmful effect on human health and may induce respiratory diseases, lung problems, cardiac disease and additional health problems (Arden Pope III et al., 2002; Kappos et al., 2004). Accurate prediction of air-quality is essential to meet the requirements of protecting the health of the public.

PM_2.5 has a complex composition in micro scale, basically can be classified as black carbon, organic carbon and ionic components (ammonium, chloride, nitrate, sulfate, etc.) (Briggs and Long, 2016; Yang et al., 2005). Macroscopically, the formation of PM_2.5 pollution highly depends on meteorological conditions, such as relative humidity (RH), air temperature, mixing height and wind speed (Cai et al., 2017; Hien et al., 2002; Tai et al., 2010), which has been demonstrated by many studies from different places of the world. For example, Tiwari et al. (2013) has revealed a negative correlation between black carbon and wind speed over New Delhi, especially during the post-monsoon season and winter. Gui et al. (2019) shows that planetary boundary layer height (PBLH) and wind speed play a dominant role among the meteorological factors in PM_2.5 changes in Eastern China. Zhang et al. (2015) finds that high RH and temperature in winter highly contribute to the formation of secondary organic carbon in Wuhan. Therefore, more accurate meteorological factors predictions will make for better PM_2.5 forecasts.

However, the large uncertainty of the numerical weather model (NWP) is still a challenge, though remarkable progress has been done in the past decades. Data assimilation is considered as an efficient method for improving the accuracy of the initial condition (IC) of a meteorological forecast (Courtier et al., 1998; Rabier et al., 1996; Mao et al., 2017; Raju et al., 2015). In addition, satellite data has been widely used and demonstrated to be helpful to improve the IC and the accuracy of meteorological predictions of the NWP model due to its large spatial coverage and relatively low cost (Barker et al., 2012; Eyre et al., 1993), especially over some areas where conventional observations are sparse (McNally et al., 2006; Singh et al., 2008).”

DA has been widely used to optimize both the chemical (Meirink et al., 2008; Pagowski et al., 2010; Schwartz et al., 2014) and meteorological (Lee et al., 2017; Park et al., 2011) initial conditions as well as emissions separately or simultaneously. Most of these previous studies mainly focus on the impacts of meteorological data assimilation on the initial conditions and forecasts of temperature and moisture (Bao et al., 2015), wind (Tan et al., 2007), precipitation (Wang et al., 2014) and hurricane (Zou et al., 2013), but few have considered the direct impact on pollution forecasts.

To understand the influences of meteorological data assimilation on meteorological and PM_2.5 forecasts, we conducted a set of weather/chemistry forecast experiments in Central China, an area playing an important role in economy development and inhabiting a large population in China. Both satellite data (AMSU-A, MHS) and conventional data are used in this study. Section 2 describes the data, methodology and models. Section 3 presents the results of experiments. And the results are summarized in Section 4.