Intense atmospheric pollution modifies weather: A case of mixed biomass burning

Abstract. The influence of air pollutants, especially aerosols, on regional and global climate has been widely investigated, but only a very limited number of studies report their impacts on everyday weather. In this work, we present for the first time direct (observational) evidence of a clear effect of how a mixed atmospheric pollution changes the weather with a substantial modification in the air temperature and rainfall. By using comprehensive measurements in Nanjing, China, we found that mixed agricultural burning plumes with fossil fuel combustion pollution resulted in a decrease in the solar radiation intensity by more than 70%, a decrease in the sensible heat by more than 85%, a temperature drop by almost 10 K, and a change in rainfall during both daytime and nighttime. Our results show clear air pollution–weather interactions, and quantify how air pollution affects weather via air pollution–boundary layer dynamics and aerosol–radiation–cloud feedbacks. This study highlights cross-disciplinary needs to investigate the environmental, weather and climate impacts of the mixed biomass burning and fossil fuel combustion sources in East China.

1. My main concern is that the authors claimed that "the ERA-5 data are generated from an ECMWF IFS spectral model and do not yet assimilate the impact of aerosols on meteorology", but "the MERRA-2 data include the impact of dust on meteorology". Actually, both the two data are reanalysis, which means that they have assimilated tremendous atmospheric observations, including temperature measurements. Here is the detailed information on the data assimilation system for ERA-5 and MERRA2. The fact is that since 1997, ECMWF operations have applied 4D-var assimilation system. https://www.ecmwf.int/en/elibrary/20196-ifs-documentation-cy47r3-part-ii-dataassimilation.
https://journals.ametsoc.org/view/journals/clim/30/14/jcli-d-16-0758.1.xml These data assimilation systems do constrain the forecast by using surface observations, balloon data, aircraft reports, buoy observations, radar and satellite observations. Once the temperature and other meteorological fields are assimilated, the impact of aerosols on meteorology is certainly included in the reanalysis data. Investigations on relevant literature are highly suggested, based on which I also suggest the authors to reconsider the method or the datasets used in this work.
Response: Thanks for your comments. We agree with your opinion. Yes, as you mentioned above, the reanalysis data assimilation systems do constrain the forecast by using surface observations, balloon data, aircraft reports, buoy observations, radar and satellite observations. Generally, once the temperature and other meteorological fields are assimilated, the impact of aerosols on meteorology is certainly included in the reanalysis data.
In our study, we focused on the Tarim Basin (TB) region, which covers an area of 5.3×10 5 km 2 and contains the Taklimakan Desert (TD). And the TD is one of the major dust sources in Asia (Gong et al., 2003;Wang et al., 2005). However, at present, there are only 30 ground observation stations and 6 radiosonde stations in TB participated in global sharing (National Meteorological Information Center http://data.cma.cn/). To some degree, the scarce observational data could limit the quality of assimilation in both ERA-5 and MERRA-2 reanalysis data.
According to your suggestion, we have investigated the relevant literature. ERA-5 data indeed have assimilated multiple measurements through a four-dimensional variational data assimilation system in 12-hourly analysis cycles (Thepaut et al., 1996).
However, ERA-5 data are generated from a spectral model (ECMWF Integrated Forecast System) and have not considered the impact of aerosols on meteorology yet (Simmons, 2006). From the perspective of assimilation, reanalysis filed error includes model error and observation error (https://www.ecmwf.int/node/19997). Previous studies indicated that there are more than 100 dusty days per year in the TB (Zhou et al., 2020), meanwhile, these dust aerosols can suspend at attitude of 3-5 km for a long time, which have obvious positive radiation forcing, and the short-wave heating rate is greater than 6K/d . Therefore, if the effects of aerosol are not considered in the reanalysis model, the model error will be underestimated, which could somehow reflect the error in reanalysis field induced by dust aerosols.
Compared with ERA-5, MERRA-2 data include the assimilation of aerosol observations, thereby it provides a multidecadal reanalysis in which aerosol and meteorological observations are jointly assimilated within a global data assimilation system (Gelaro et al., 2017). More importantly, the aerosols and their interactions with weather and climate have been considered in MREEA-2 (Randles et al., 2017).
Therefore, there is an obvious difference between ERA-5 and MERRA-2. Figure R1 shows a comparison of MERRA-2 data with radiosonde observations in the TB region.
A good agreement is found between MERRA-2 data and radiosonde observations. Hence, we used the MERRA-2 reanalysis data to supplement the observation data. The OMR method is proposed by Ding et al. (2021). This method is based on the assumption that the difference between observations and reanalysis models reflects the impact of un-resolved processes (Huang and Ding, 2021;Huang et al., 2018;Kalnay and Cai, 2003;Wang et al., 2013;Zhao et al., 2014). It means that these real biases, which result from missing physical or chemical processes in the model, have been misinterpreted as observational errors and discarded during the data assimilation procedure for the reanalysis data (Huang and Ding, 2021;Huang et al., 2018). As illustrated above, ERA-5 data were generated from a spectral model (ECMWF Integrated Forecast System) and has not considered the impact of aerosols on meteorology yet (Simmons, 2006). Therefore, investigation of the difference between observation (radiosonde observation and MERRA-2 reanalysis data) and ERA-5 reanalysis data can provide an opportunity to study the heating effect of dust aerosols, especially in a region with dust aerosol pollution (Huang and Ding, 2021;Ding et al., 2013). This method has been tested by previous work and proved to be well suited to identify the effects from aerosol impacts on the air temperature (Huang and Ding, 2021;Ding et al., 2013;Huang et al., 2018;Kalnay and Cai, 2003;Wang et al., 2013;Zhao et al., 2014). We have supplemented the reanalysis data and method in the revised version (Lines 101-103; Lines 139-143).

References:
Ding K., Huang, X., and Ding, A., et al., Aerosol-boundary-layer- Figure 4 is too small to be clearly identified and needs to be improved.