Elevated 3D structures of PM 2.5 and impact of complex terrain-forcing circulations on heavy haze pollution over Sichuan Basin, China

. Deep basins create a uniquely favorable condition for the formation of air pollution, and the Sichuan Basin (SCB) in Southwest China is such a basin featuring frequent heavy pollution. A wintertime heavy haze pollution event in SCB was studied with conventional and intensive observation data and the WRF-chem model to explore the three-dimensional distribution of PM 2.5 for understanding the impact of regional pollutant emissions, basin circulations associated with plateaus, and downwind transport to 20 the adjacent areas. It was found that the vertical structure of PM 2.5 over SCB was characterized with a remarkable hollow sandwiched by high PM 2.5 layers at heights of 1.5–3 km and the highly polluted near-surface layer. The southwesterlies passed over the Tibetan Plateau (TP) and Yunan-Guizhou Plateau (YGP) resulted in a lee vortex over SCB, which helped form and maintain heavy PM 2.5 pollution. The basin PM 2.5 was lifted into the free troposphere and transported outside of SCB. At the bottom of SCB, 25 high PM 2.5 concentrations were mostly located in the northwest and southern regions. Due to the blocking effect of the plateau terrain on the northeasterly winds, PM 2.5 gradually increased from northeast to southwest in the basin. In the lower free troposphere, the high PM 2.5 centers were distributed over the northwestern and southwestern SCB areas, as well as the central SCB region. For this event, the regional emissions from SCB contributed 75.4–94.6 % to the surface PM 2.5 concentrations in SCB. The SCB 30 emission export was the major source of the PM 2.5 over the eastern regions of TP and the northern regions We also assessed the contribution of local emissions to the heavy PM 2.5 pollution within SCB, and the impact of external transport of the SCB PM 2.5 to the surrounding areas in Southwest China. conclusions squared error verification metrics a reasonable good model performance with a to the (Emery Hanna, 2004), RH a bit underestimated and wind speed was slightly overestimated. Furthermore, the statistical verification of the simulated the multiple ground observations, meteorological sounding data and Micro Pulse Lidar retrievals as well as conducting modeling experiments with the WRF-Chem model, this study investigated the three-dimensional structures and the development mechanisms of the PM 2.5 for a 310 wintertime heavy haze pollution episode over SCB, an isolated deep basin in Southwest China. The roles of the basin pollutant emission and the unique basin circulations were evaluated for their contributions to the 3D distribution of PM 2.5 over SCB and to the neighboring YGP, TP, and DMB regions. The vertical structure of the PM 2.5 in the lower troposphere over SCB was characterized with unique hollows located between a high PM 2.5 layer at heights of 1.5 – 3 km and the high PM 2.5 surface layer. It is 315 regions and affects the atmospheric environment changes in Southwest China. This work exposed the unique and important three-dimensional structures of PM 2.5 and investigated their formation mechanisms and downwind outflow transport over SCB. The deep basin terrain along with the TP and YGP forcing effect creates very complex PM 2.5 pollution conditions over the SCB region, 340 which is dramatically different from those over relatively flat regions. To generalize our findings, further work with more case studies and regional climatic analyses with long-term observation data and numerical modeling with data assimilation and refined physical and chemical schemes are desired. Furthermore, as pointed out in this study, the PM 2.5 emission sources in SCB greatly influence the regional environmental changes over Southwest China. Thus, the regional transport modeling of air

surface PM2.5 concentrations were shown in Table 4 with the normalized mean bias (NMB), the normalized mean error (NME), the mean fractional bias (MFB)  The vertical structure of the atmospheric boundary layer directly affects the vertical diffusion of atmospheric pollutants. Therefore, we compared the vertical profiles of the model simulation with the intensive sounding observations in terms of variation range and average profiles during the heavy haze episode. The potential temperature, wind speed and relative humidity of the simulation were validated 180 for both daytime and nighttime in Fig. 4. The simulated vertical profiles of meteorological variables were generally acceptable in the lower troposphere (Fig. 4). It should be pointed out that the significant underestimation of RH above 1km, where the observed RH reached nearly 100%, was caused by the clouds due to the abundant moisture at night, that the model failed to reproduce.
The MPL-4B lidar, located at site 15 ( Fig. 1) in the western edge of SCB to the east of TP, 185 continuously detected aerosol extinction ratios in the troposphere. The vertical distribution of PM2.5 mass concentrations were derived from the extinction ratio (Ansmann et al., 2012;Córdoba-Jabonero et al., 2016). The height-time cross-section of derived and simulated PM2.5 mass concentrations from 7:00 a.m. to 2:00 p.m. on 5 January 2017 were presented in Fig. 5. It can be seen that a good agreement between the lidar observation and the WRF-Chem simulation was achieved. One of the significant features is that 190 besides the occurrence of near-surface high PM2.5, which is typical for most heavy haze pollution over the areas with relatively flat terrain, a layer of high PM2.5 concentrations was developed between 1 and 2 km above ground level (Fig. 5a), leaving a hollow layer between the two heavy polluted layers. The upper high PM2.5 layer was built with uplifting and then overturning of the air flows associated with the blocking effect of the TP terrain, which will be addressed in the next section. urban agglomeration (Fig. 1). From the formation to the maintenance and the dissipation periods, the 200 prevailing northeasterly winds strengthened gradually over SCB (Fig. 6). The high plateaus and mountains, especially YGP and TP to the west of SCB blocked the upcoming northeasterly winds. The spatial distribution of surface PM2.5 concentrations (Fig. 6) clearly reflects the combined effect of the urban anthropogenic air pollutant emissions and the PM2.5 accumulations by the flow convergence forced by the TP and the YGP blocking to prevailing winds. During the formation and maintenance stage, the 205 surface winds were weak (1.4-1.7 m s -1 ) over SCB, which was insufficient to dispel the air pollutants, but to continuously accumulate PM2.5 locally from light to heavy pollution conditions (Fig. 6a, Fig. 6b).
In the maintenance period, heavy air pollution blanketed a large area in SCB with excessive PM2.5 concentrations (mostly >150.0 μg m -3 ). By the dissipation period, the northeasterly winds intensified and removed PM2.5 from SCB (Fig. 6c).

Vertical structures of PM2.5 concentrations
The high terrain of YGP and TP blocked the northeastern airflows over SCB by lifting the airflow along with air pollutants, altering the vertical PM2.5 distribution. Therefore, it was of great interest to analyze the vertical distribution and the transport structures of PM2.5 over SCB and the surrounding regions.

215
To examine the vertical structures of PM2.5 concentrations over SCB, we selected the urban site 1 (104.02° E; 30.67° N) in Chengdu (cf. In the current case, the lee vortex circulation, working together with the basin near-surface flows, drove https://doi.org/10.5194/acp-2020-1161 Preprint. Discussion started: 23 November 2020 c Author(s) 2020. CC BY 4.0 License. atmospheric environment over SCB. During the formation period of heavy air pollution event, the PM2.5 particles in the free troposphere were concentrated in the northwestern SCB (Fig. 9a); In the maintenance period, the high PM2.5 centers were developed in the northwestern SCB edge, and PM2.5 concentrations 260 increased obviously in the southwestern and central SCB regions (Fig. 9b), reflecting the strong vertical diffusion of PM2.5 in the lower troposphere during the heavy air pollution (Figs. 7c and 8c). Driven by strong northeasterly winds in the dissipation period (Fig. 6c), the high PM2.5 concentrations in the lower free troposphere were centered in the narrow southwestern and southern SCB areas (Fig. 9c), where the PM2.5 from the polluted SCB region were transported out at the gap between the eastern TP and northern 265 YGP edge.

Contribution of local emission and outflow transport
Local emission and regional transport of air pollutants are two key factors affecting air quality. Haze   respectively. However, it was interesting to point out that the averaged contribution rates of regional air pollutant emissions to surface PM2.5 concentrations in SCB were actually dropped down from 90.7 % in the formation period, 85.6 % in the maintenance period to 83.3 % in the dissipation period (Fig. 10). We think the exchanges of PM2.5 between the polluted air over SCB and the cleaner environment air over the surrounding plateaus and mountains in Southwest China play a role in this process. (Figs. 7 and 8).

285
To assess the impact of the PM2.5 transport from SCB on the air quality over the surrounding areas https://doi.org/10.5194/acp-2020-1161 Preprint. Discussion started: 23 November 2020 c Author(s) 2020. CC BY 4.0 License.
in Southwest China, we calculated the PM2.5 contribution amounts and rates of SCB's regional air pollutant emissions to the adjoining regions in the plateaus and mountains based on the differences of the PM2.5 concentrations between Emi-real and Emi-Non (Table 5). The near-surface prevailing northeasterly winds of SCB brought PM2.5 from SCB to the eastern TP edge, the northern YGP edge and 290 the DBM region (Fig. 6), resulting in importing concentrations of surface PM2.5 averaged respectively with 18.0, 31.3 and 10.4 μg m -3 during the heavy haze pollution (Table 5). TP and YGP, as clean regions in China (Song et al., 2017;Zhan et al., 2018), were remarkably polluted by the PM2.5 transport from SCB. During the dissipation period of the heavy air pollution episode, the eastern TP edge and northern YGP regions gained peak imports of PM2.5 at 22.9 and 41.9 μg m -3 (Table 5). Thus in this case, the 295 downwind adjoining TP and YGP regions is the main receptor area of the SCB emissions.
Finally, the PM2.5 contribution rates, i.e., the percentages between the PM2.5 concentrations transported from the basin to those in the adjacent regions of plateaus and mountains were calculated for different periods of the heavy PM2.5 pollution over SCB. The surface PM2.5 in the eastern TP edge were mostly originated from the source region of SCB, with the dominant contribution rates respectively of 300 63.6, 67.4 and 72.7 % in the formation, maintenance and dispersion periods. The PM2.5 import from the SCB pollutant emissions also contributed the majority of surface PM2.5 concentrations in the northern YGP, with contribution rates of 58.3, 52.8 and 70.5 % during three different periods with an overall contribution rate of 58.5 % averaged for the whole SCB heavy air pollution period. In contrast, the DBM region was less influenced by the SCB's emission sources with a contribution rate of 31.0 % averaged 305 during the heavy air pollution event.

Conclusions
By using the multiple ground observations, meteorological sounding data and Micro Pulse Lidar retrievals as well as conducting modeling experiments with the WRF-Chem model, this study investigated the three-dimensional structures and the development mechanisms of the PM2.5 for a 310 wintertime heavy haze pollution episode over SCB, an isolated deep basin in Southwest China. The roles of the basin pollutant emission and the unique basin circulations were evaluated for their contributions to the 3D distribution of PM2.5 over SCB and to the neighboring YGP, TP, and DMB regions.
The vertical structure of the PM2.5 in the lower troposphere over SCB was characterized with unique hollows located between a high PM2.5 layer at heights of 1.5-3 km and the high PM2.5 surface layer. It is Table 1. Names of 18 observation sites with the corresponding site number (Fig. 1b)