The peripheral circulation of typhoon forms sustained ozone episodes. However, how it impacts the day-to-day ozone pollution levels during the episodes has not been clearly studied, which is crucial for better prediction of the daily ozone variation. In this study, the analysis of ground observation, wind profile data, and model simulation is integrated. By analysing the wind profile radar observations, we found a weak wind deepening (WWD; vertical depth of the weak winds increased), more correlated with the ground-level ozone variation than surface weak wind. Long-term statistical analyses showed that the WWD is a common weather phenomenon in the peripheral subsidence region of typhoons and is generally accompanied by ozone pollution episodes. The Weather Research and Forecasting model coupled with Chemistry (WRF-Chem) with process analysis simulation showed that the peripheral subsidence chemical formation (CHEM) and vertical mixing (VMIX) effects are two major contributors to the enhancement of ozone levels to form the episode, while the advection (ADV) showed negative values. However, the day-to-day variation of the daytime ozone levels during the episode is not determined by the daily variation of daytime CHEM and VMIX but is dominated by the ADV terms. Therefore, the ozone and its precursors accumulation, including the enhancement during the night-time, contribute to the daytime ozone increase in the following day. A detail day-to-day process analysis showed that in additional to decrease of negative ADV values (e.g. the weakened advection outflow or dispersion) on the ground, the integrated effect of the daily variation of the accumulative CHEM and ADV above the ground throughout the planetary boundary layer (PBL) together determined the overall day-to-day daytime ozone variation on the ground through the VMIX process. The results indicate that the peripheral characteristics of approaching typhoon not only form the ozone episode by the enhanced photochemical reactions, but also could increase the day-to-day daytime ozone levels via pollution accumulation throughout the PBL due to the WWD up to 3–5 km. These results illustrate the important role of the WWD in the lower troposphere for the formation of sustained ozone episodes due to the peripheral circulation of the typhoon, which helps to better predict the daily changes of daytime ozone levels.
The Pearl River Delta (PRD), located in the coastal region of southern China and often affected by typhoon systems, has experienced major economic development and urbanization accompanied by a large increase in air pollution and decrease in visibility (Wang et al., 1998, 2001; Lai and Sequeira, 2001). Ozone pollution is the most significant air pollution challenge in this region and has been the "primary pollutant" since 2014 (Ministry of Ecology and Environment of China, 2016). Ozone is harmful to human health and has adverse effects on vegetation and crops, among others (Aunan et al., 2000; Felzer et al., 2007; Feng et al., 2015). Ozone concentrations are determined by the photochemical reactions of its precursors and local meteorological conditions. However, ozone pollution episodes are mainly triggered by weather conditions rather than by sudden increases from emission sources (Ziomas et al., 1995; Giorgi and Meleux, 2007; Lin et al., 2019; Li et al., 2018).
There are many studies (Gao et al., 2018) that have classified the weather patterns affecting regional pollution events into several types, such as cold fronts, cold high-pressure systems moving towards the sea, uniform pressure fields, Western Pacific subtropical high (WPSH), tropical cyclone (TC) peripheries, and weak cold high-pressure ridges. Using observational data, several studies have reported the impacts of TC activity on meteorological factors that are favourable for air pollution over the PRD region (Feng et al., 2007; Chen et al., 2008; Wu et al., 2013). TCs are typical weather systems responsible for both high ozone and PM
Previous studies in the PRD and other coastal regions of China have illustrated the significant impact of TCs on forming ozone (TC-ozone) episodes (Zhang et al., 2012, 2013; Li et al., 2013, 2014; Jiang et al., 2015; Huang et al., 2015; Shu et al., 2016, 2019; Tan et al., 2018; Chen et al., 2018; Han et al., 2020). TC-ozone episodes generally occur when there are weather conditions such as high temperatures, radiation flux, low relative humidity, and weak wind (Cheng et al., 2016; Liu et al., 2017). Observation-based studies have reported that the TC-ozone episodes are associated with weak wind; however the mechanism underlying the effect of weak wind on ozone in TC-ozone episodes remains to be fully elucidated. In addition, previous process analysis based on numerical modelling simulations have shown that the chemical (CHEM) and vertical mixing (VMIX) effects are two major contributors to ozone episodes, whereas advective transport (ADV) is generally a consumptive process (Shu et al., 2016; Wang et al., 2010). The inconsistencies between observational and simulated results of wind contributions to ozone episodes are poorly understood, which may be attributed to the limited data on the influence of weak wind on ozone concentration enhancement.
In addition, for the air quality forecast and prevention, it is important to understand the mechanism underlying the day-to-day variation of the daytime ozone levels, since the ozone levels peak during the daytime due to photochemical effects; ozone is converted to
Thus, the objective of this study is to understand the impact processes of typhoon circulation characteristics on the day-to-day variation of daytime ozone concentration in TC-ozone episode. The analysis of ground observation, wind profile data, and Weather Research and Forecasting model coupled with Chemistry (WRF-Chem) simulation with process analysis is integrated. Detailed data and model description are provided in Sect. 2, followed by the results and discussion in Sect. 3. The main conclusions are summarized in Sect. 4.
In this study, hourly surface ozone concentrations from 2016 over mainland China were obtained from the Ministry of Environmental Protection of China. The 3D wind profiler data, automatic weather station data, cloud data, and solar radiation measurements were provided by the China Meteorological Administration and were used for the meteorological analyses of Typhoon Nepartak. The Final (FNL) Operational Global Analysis data used to describe the circulation of Typhoon Nepartak have a horizontal resolution of
The observations of a typical ozone episode that occurred in the PRD region during 7–10 July 2016 (local standard time, LST) before Typhoon Nepartak made landfall were collected and analysed. Typhoon Nepartak intensified into a supertyphoon at 20:00 LT on 5 July, then gradually moved northwest due to the forcing of the WPSH over its northeastern side (Fig. S2 in the Supplement). At 05:50 on 8 July, the typhoon made landfall in Taitung County, Taiwan, with a maximum wind speed of 60
WRF-Chem is a widely used and fully coupled online 3D Eulerian chemical transport model (https://ruc.noaa.gov/wrf/wrf-chem/, last access: 8 March 2022) that considers both chemical and physical processes (Zhang et al., 2010; Forkel et al., 2012); version 3.9.1.1 was applied in this study. Detailed descriptions of the meteorological and chemical aspects of the WRF-Chem model have been previously reported by Grell et al. (2005) and Skamarock et al. (2008). For the simulation, two nested domains (Fig. S1 in the Supplement) were set up with horizontal resolutions of 27 and 9 km and grids of
There were 39 vertical layers that extended from the surface up to a pressure maximum of 50 hPa, 12 of which were located in the lowest 2 km to fully describe the vertical structure of the planetary boundary layer (PBL). Carbon Bond Mechanism Z (CBM-Z), which includes 133 chemical reactions for 53 species and extends the model framework to function for a longer time period and at a larger spatial scale than its predecessor, was used as the gas-phase chemical mechanism (Zaveri and Peters, 1999). The corresponding aerosol chemical mechanism was the Model for Simulating Aerosol Interactions and Chemistry (MOSAIC) with eight bins (Zaveri et al., 2008), which is extremely efficient and does not compromise accuracy of the aerosol model calculations. Other major model configuration settings are listed in Table 1.
Table 1 Major model configuration options used in the simulations.
ITEM Selection Longwave radiation RRTMG Shortwave radiation RRTMG Microphysics scheme Lin scheme Boundary layer scheme Yonsei University (YSU) scheme Land surface option Noah land surface model Photolysis scheme Fast-J photolysis Dry deposition Wesely scheme
The ozone pollution level and the meteorological conditions of the typhoon Nepartak case were first analysed. As shown in Fig. 1, Guangdong Province experienced severe ozone pollution during the period 7–10 July; 28 % (7 July) to 57 % (10 July) of the air quality stations in Guangdong Province exceeded the national air quality standard level-II for ozone (200
Graph: Figure 1 The horizontal distribution of surface ozone concentration over PRD at 16:00 from (a) 6 July 2016 to (f) 11 July 2016. The yellow and blue triangles in (a) denote the positions of wind profiler stations 59 486 and 59 284. The black box denoted D0 indicates the area where the severe ozone pollution event occurred.
Graph: Figure 2 (a–d) The 1000 hPa wind vectors of NCEP-FNL data from 14:00 (7 July) to 14:00 (10 July) with red triangle and typhoon symbols representing the locations of the PRD centre and Nepartak, respectively. (e–h) Vertical cross sections of vertical velocity along the four straight lines linking PRD and the centres of Typhoon Nepartak in (a–d) from 14:00 on 7 July to 14:00 on 10 July 2016. The four dashed blue boxes denote the longitude range of PRD in (e–h).
The weather over the PRD region was characterized as clear sky, strong solar radiation (Fig. 3a), low relative humidity (Fig. 3b), and high temperatures (Fig. 3c), when the subsiding branches of vertical typhoon circulation were located over the PRD during 7 and 8 July (Fig. 2e and f). The variations in these surface meteorological variables exhibited favourable conditions for increasing ozone concentrations (Cheng et al., 2016; Liu et al., 2017). However, the height of the PBL increased significantly on 8 and 9 July (Fig. 3c), and the atmosphere was under unstable conditions, which was indicated by the comparison between the adiabatic lapse rate (blue) and the environmental lapse rate (red) (Fig. 3d–f). This instability is also shown by the large values of convective available potential energy (CAPE; Fig. 3d–f), which is another criterion used to determine the stability of atmosphere. When the CAPE is
Graph: Figure 3 Time series of diurnal mean (a) cloud cover, radiation at 59 287 observation station, (b) relative humidity, 24 h precipitation, and averaged (c) planetary boundary layer height (PBLH) and temperature anomaly of region D0 from 1 to 15 July. The SkewT/LogP at 14:00 on 7 (d), 8 (e), and 9 July (f). The solid thick red, blue, and yellow lines in (d, e, and f) denote the temperature sounding, the parcel path from the surface upward, and the dew point sounding, respectively.
The evolution of the vertical profile of horizontal winds at representative station 59 284 is shown in Fig. 4a. Before 5 July, the wind speed increased with the vertical atmospheric layers. There were relatively larger wind speeds above the PBL and weaker wind speeds below
Graph: Figure 4 The profile evolution of horizontal wind speed from 3 to 13 July. The solid black lines are the surface ozone concentrations at (a) 59 284 and (b) 59 486 wind profile radar stations.
By analysing the wind profile data (Fig. 4), we observed that the vertical depth of the horizontal weak wind generally increased from the surface up to the lower troposphere (
Graph: Figure 5 Correlation coefficient between the evolution of average wind speed and the evolution of ground ozone concentration in different altitude ranges of each wind profile radar station.
Long-term statistical analysis showed no stable atmospheric stratification and a decrease in the height of the boundary layer in this ozone pollution episode. The analysis of wind profile radar data and the correlation coefficients between the surface ozone concentrations and the average wind speeds between the surface and the altitude of each vertical layer (up to 6 km) indicated that in this episode of ozone pollution, WWD might have played an important role in the increase of ozone pollution at the surface. Guangdong Province is located on the western coast of the Pacific Ocean and is frequently affected by typhoons. To investigate whether the relationship between WWD and ground-level
The PRD regional background ozone concentration is generally less than 80–100
Table 2 The statistical results of the peripheral weak wind of 38 tropical cyclones for seven radar stations in Guangdong Province and ozone concentration from 2014 to 2018.
Radar station number 59 294 (87 %) (97 %) 59 486 (97 %) (94 %) 59 476 (97 %) (93 %) 59 285 (92 %) (100 %) 59 287 (92 %) (100 %) 59 284 (96 %) (100 %) 59 290 (93 %) (78 %) Avg. 93 % (87 %–97 %) 94 % (78 %–100 %)
The above correlation coefficients and statistical analysis indicate that WWD may be a common weather phenomenon in the periphery of typhoon and could impact the ground-level ozone concentration. In the subsequent section, the influence of WWD on ground-level ozone pollution and the impact of typhoon peripheral circulation on sustained ozone enhancement during Typhoon Nepartak are discussed based on the WRF-Chem numerical simulation.
To investigate the impact of typhoon periphery and WWD on formation of the sustained ozone episode, the numerical model with the process analysis was applied, prior to which the model performance was validated using the available observations. Figure S6a–d in the Supplement present the measured and simulated data for temperatures, wind speeds, wind directions, and ozone concentrations at Guangzhou from 00:00 on 3 July to 07:00 on 15 July 2016. With regards to the meteorological variables, there was good agreement between the measured and modelled results, especially the shifting wind features, implying that the model successfully captured the synoptic features. However, ozone concentrations (Fig. S6d) overestimated low values and underestimated high values. However, the simulated results and observed data agreed reasonably with each other and captured the ozone episode in the region.
Statistical metrics including the index of agreement (IOA), mean bias (MB), root mean square error (RMSE), and normalized mean bias (NMB) were used to further assess the model performance (Table 3). The IOA of the wind direction was determined according to Kwok et al. (2010), while the IOA values for the other variables were calculated as per Lu et al. (1997). Our simulation of the time series of ozone concentrations and meteorological variables was reasonable. All the meteorological parameters were close to the corresponding simulation results in the PRD region (Wang et al., 2006; Li et al., 2007; Hu et al., 2016). IOAs for temperature and wind speed (0.89 and 0.66, respectively) fulfilled the criteria (as presented in the brackets of Table 3). The model performed well at capturing the wind directions, with a small MB of 7.72
Table 3 Statistical comparison between the observed and simulated variables. The benchmarks are based on Emery et al. (2001) and EPA (Doll, 1991).
Variable IOA MB RMSE NMB (%) Temp () 0.89 () 1.90 2.68 Wspd. () 0.66 () 1.45 () 37.81 Wdir. () 0.77 7.72 () 85.88 4.24 Ozone (ppbv) 0.84 9.53 37.15 4.83 ()
Values that did not fulfil the criteria are indicated in italics.
Moreover, ozone concentrations are well simulated, with an IOA of 0.84 and an NMB of 4.83. Time series comparisons of ozone concentrations and meteorological factors at Shenzhen, Zhongshan, and Zhuhai are presented in Fig. S6a1–d1, a2–d2, and a3–d3. The overall results suggest that the model could reproduce ozone concentrations and capture the transport features in southern China.
Variations in ozone concentration are directly caused by physical and chemical processes (Zhu et al., 2015); the fact that peripheral circulation of a typhoon affects ozone concentration can be discussed using an process analysis. The following processes were considered in this analysis: (
Figure 6a shows the profile evolution of the average ozone concentrations in region D0 (black box D0 in Fig. 1) from 08:00 on 5 July to 20:00 on 10 July. The ozone concentrations gradually increased from 6–9 July throughout the PBL, with an increase in PBL height of up to
Graph: Figure 6 The profile evolution of averaged (a) ozone concentration and (b–e) CHEM, VMIX, CONV, and ADV of region D0 from 08:00, 5 July, to 20:00, 10 July. The black lines denote the planetary boundary layer height (PBLH).
It can be seen from Fig. 6b–e that during the period from 08:00 to 20:00 on 5–10 July, the contributions of CONV in PBL were zero; CHEM on the ground showed strong negative contributions, and VMIX on the ground showed strong positive contributions; ADV in PBL showed weak negative contributions during 6 and 7 July, and the negative contributions of ADV in PBL were strengthened on 8 and 9 July. Therefore, the contributions of ground VMIX and CHEM played a major role in the change of the PBL ozone concentrations, which is consistent with previous studies in the PRD region (Wang et al., 2010). The enhanced ozone above ground due to the CHEM effect contributed to the ground ozone enhancement through the increased VMIX effect. At the same time, changes in the strength of ADV contributions in PBL might also have a certain impact on the changes in the ozone concentrations on the ground.
In order to investigate the cause of the continued day-to-day increase of the daytime ozone concentration during the sustained ozone episode, the numerical relationship between the daytime (we used 08:00 to 20:00 in this study) average ozone concentration difference of 2 adjacent days and the various physical and chemical processes must be quantified. Based on the numerical process analysis, the difference between the daytime average ozone concentrations on two adjacent days (DDOC) can be further expressed by accumulative contribution between the periods, which can be expressed by three continuous contribution terms:
where
Graph: Figure 7 (a) Daytime and night-time ozone concentrations and (b) SUM and TOTAL_SUM on the ground within region D0 during 08:00 on 4 July to 20:00 on 10 July.
Notably, the ozone chemistry between the daytime and night-time is different. The SUM value during daytime is always positive, while the SUM of the night-time is always negative. In terms of the daily daytime variation, the separated three terms of TOTAL_SUM reveal that the daily variation of daytime ozone level is not only determined by the daytime chemistry, but is also influenced by the night-time ozone variation between the 2 adjacent days. For example, the night-time consumption or accumulation of ozone (as well as precursors) could contribute to the daytime ozone increase of the following day; therefore, in diagnostic forecasting of daily air quality, an increase in daytime ozone level can be expected if the concentration of ozone precursors id enhanced in the previous night but the meteorological condition remains unchanged between the 2 adjacent days.
Further, DDOC or TOTAL_SUM of 2 adjacent days can be decomposed into contributions of the different processes (CHEM, VMIX, CONV, ADV). We denote the four accumulative terms as TOTAL_SUM_CHEM, TOTAL_SUM_VMIX, TOTAL_SUM_CONV and TOTAL_SUM_ADV accordingly (see Eq. 5 in the Supplement for details). The details of the budget of the TOTAL_SUM_CHEM, TOTAL_SUM_VMIX, and TOTAL_SUM_CONV during the episode between 2 adjacent days are presented in Table 4. Each column shows an accumulative contribution of different process from 08:00 to 20:00 of the next day. The results show that both the VMIX and ADV enhancement contributed to the daily increase of daytime ozone concentration from 6 to 9 July on the ground. More specifically, during the episode (columns highlighted by brown colour), the TOTAL_SUM_VMIX contributions are always positive on the ground and reach maximum from 6 to 7 July, while the TOTAL_SUM_CHEM contributions are negative, which should be the result of the surface NO-titration effect. The TOTAL_SUM_CONV contributions are relatively ignorable, while the TOTAL_SUM_ADV contributions significantly increased from negative value to positive value during the episode period. Since the CHEM and VMIX are significantly associated with each other, the combined contribution of CHEM, VMIX, and CONV to the TOTAL_SUM is shown by the TOTAL_SUM_CVC in the Table 4. The CHEM+VMIX+CONV contribution to daily daytime ozone variation changed to negative values during the episode period, which did not determine the trend of the DDOC. By comparing the accumulative effect of individual process to the combined effect of the four processes (TOTAL_SUMs), the variation of DDOC (which increased from 5 to 9 July and decreased on 10 July) was determined by the integrated effect of four processes but mainly dominated by the TOTAL_SUM_ADV (sudden change from negative values to large positive values during the episode).
Table 4 The decomposed accumulative CHEM, VMIX, CONV, and ADV effects of the TOTAL_SUM on the ground.
Period (ppbv) 4_08-5_20 5_08_6_20 6_08-7_20 7_08-8_20 8_08-9_20 9_08-10_20 TOTAL_SUM_CHEM TOTAL_SUM_VMIX 118.85 113.40 88.91 70.38 105.23 TOTAL_SUM_CONV 33.70 13.50 0.81 12.13 TOTAL_SUM_ADV 10.97 15.06 14.01 6.91 TOTAL_SUM_CVC 14.39 13.089 TOTAL_SUMs 0.4242 9.7734 6.957 8.1045 6.5583
The highlighted column indicates the non-attainment (national air quality standard level-II) ozone period. TOTAL_SUM_CAC is the sum of the TOTAL_SUM_(CHEM+VMIX+CONV).
The VMIX effect links the ground ozone variation to the ozone variation in the upper PBL level, which is dependent on the vertical gradient of the concentration and the turbulence exchange coefficients (Gao et al., 2020). To understand the connection and why the VMIX contribution to the surface ozone reaches a maximum (131.0915 ppb) from 6 to 7 July, the vertical profiles of accumulative CHEM, ADV, CONV, and CAC (CHEM+ADV+CONV) to the TOTAL_SUM during the time period from 08:00 to 20:00 on 5–7 July are shown in Fig. 8. (For example, the accumulative CHEM effect from 08:00 to 20:00 on 6 July is denoted as the sum of CHEM 06_08-20.)
Graph: Figure 8 The vertical profiles of accumulative (a) CHEM, (b) ADV, (c) CONV, and (d) CVC (CHEM+ADV+CONV) during the periods from 08:00 to 20:00 on 5–7 July.
The gradient of vertical profile of accumulative CHEM contribution on 6 July was significantly larger than that of vertical profiles of accumulative CHEM contribution on 5 and 7 July (Fig. 8a). The CHEM increase in PBL is due to the impact of the periphery of the typhoon, which would produce a field of meteorological conditions conducive to photochemical reactions. These meteorological conditions also increased the absolute contribution and gradient of accumulative ADV contribution compared to that of 5 July (Fig. 8b). Therefore, the vertical profile gradient of the sum of CVC 06_08-20 was the largest, which contributed to the enhancement of VMIX contribution to the ozone on the ground. In short, both the daytime CHEM and ADV enhancement above the ground throughout the PBL have contributed to the increase in VMIX contribution to the ground-level ozone. The CHEM enhancement above the ground throughout the PBL is due to the increase in photochemical formations of precursors, while the ADV enhancement above the ground throughout the PBL is attributed to the WWD (weak wind deepening) effect in the whole lower troposphere during the episode.
In summary, under the influence of the peripheral subsidence of typhoon, the weak subsidence associated with typhoon periphery brings clear sky and warmer air, which is conducive to the ozone photolysis formation (CHEM) above the ground in planetary boundary layer (PBL) and compensates for the ozone through the positive VMIX effects on the ground. Therefore, the chemical formation (CHEM) and vertical mixing (VMIX) effects are two major contributors to forming TC-ozone episodes, while the ADV and CONV show negative values. However, the day-to-day daytime ozone levels do not associate with daily variation of daytime CHEM and VMIX but are dominated by the daily variation of ADV (e.g. weakened advection outflow or dispersion). The daily enhanced ADV during the episode on the ground and throughout the PBL is attributable to the WWD, which is a common phenomenon induced by the peripheral circulation of the typhoon system. In addition, both the enhanced CHEM and ADV above the ground contribute to the daily daytime ozone enhancement on the ground via the VMIX process during the episode.
In this study, the analysis of ground observation, wind profile data, and model simulation was integrated. By analysing the wind profile radar observations, we found that not only surface weak winds but also WWD generally appeared in the periphery of the typhoon. The statistics of wind fields and ground-level ozone at seven wind profile radar stations in PRD during the 38 typhoons in the Northwestern Pacific Ocean from 2014–2018 showed that the number of WWD occurrences accounted for 93 % (87 %–97 %) of the available number of radar stations for the seven radar stations in average. The number of ozone pollution occurrences accounted for 94 % of the number of WWD occurrences in average. The statistical results show that WWD is a common weather phenomenon in the periphery of typhoons associated with periphery subsidence of typhoon system and is often accompanied by significant increases in ozone concentrations.
The WRF-Chem model was used to simulate the daily daytime ozone variation in a sustained ozone pollution process in PRD during Typhoon Nepartak in 2016. Validation results showed that the model could reasonably reproduce the observed temperature, wind speed, wind direction, and ozone. Process analysis results showed that under the impact of the peripheral subsidence of typhoon, the chemical formation (CHEM) and vertical mixing (VMIX) effects are two major contributors to the enhancement of ozone levels to form an episode, while the ADV and CONV always show negative or small values. However, the day-to-day variation of the daytime ozone levels is not determined by the daily variation of daytime CHEM but is dominated by the daily variation of ADV terms on the ground (e.g. the weakened advection outflow or dispersion). So, the ozone and its precursors' accumulation, including the enhancement during the night-time, contribute to the daytime ozone increase in the following day. Via a detailed day-to-day analysis, we found that the decrease of negative ADV values during the event not only occurred on the ground, but also throughout the PBL. The daily enhanced VMIX contribution to the ground-level daytime ozone during the episode is associated with the enhanced CHEM and ADV in the upper PBL. Results show that in addition to the weakened advection outflow or dispersion on the ground, the integrated effect of the day-to-day variation of the accumulative CHEM above the ground and accumulative ADV contribution throughout the PBL together determined the overall day-to-day daytime ozone variation through the VMIX process on the ground.
This study reveals that the peripheral characteristics of the approaching typhoon not only form the ozone episode by enhanced photochemical reactions, but also change the day-to-day ozone levels by pollution accumulation throughout the PBL due to the weak wind deepening up to 3–5 km. This result explains the continued increase in daytime ozone, although the photochemical contribution began to decrease during the event. It also reveals the important role of WWD in the lower troposphere for the formation of sustained ozone episodes due to the peripheral circulation of the typhoon, which helps to better predict the daily changes of daytime ozone levels.
The supplement related to this article is available online at: https://doi.org/10.5194/acp-22-3861-2022-supplement.
We would like to acknowledge the National Centers for Environmental Prediction (NCEP) for the Final Operational Global Analysis data, the Ministry of Environmental Protection, China, for the hourly ambient surface O
By Ying Li; Xiangjun Zhao; Xuejiao Deng and Jinhui Gao
Reported by Author; Author; Author; Author