Impact of Taihu Lake on City Ozone in the Yangtze River Delta

With their rapid industrialization and economic development, the scale of Chinese cities is increasing. A consequence of this is that they emit large amounts of ozone (O₃) precursors [nitrogen oxides (NO_x), volatile organic compounds (VOCs) etc.] into the atmosphere, meaning atmospheric environmental problems are becoming increasingly serious and a direct threat to human health. O₃ is a secondary pollutant generated by NO_x and VOCs (Chameides and Walker, 1973; Oltmans, 1981; Logan, 1985). Currently, regional photochemical pollution is apparent in the Beijing-Tianjin-Hebei region, the Yangtze River Delta region, and the Pearl River Delta region, along with other economically developed areas. The phenomenon of high concentrations of O₃ has become the focus of many studies into urban air quality (e.g., Vingarzan, 2004; Qu et al., 2014; Han et al., 2015; Shang et al., 2015).

A number of studies have been published that have highlighted the regional transport and photochemistry of severe O₃ pollution. (Ling et al., 2011) found that reducing the different sources of VOCs can control O₃ formation in the Hong Kong area. The O₃ pollution process is closely related to tropical cyclones in the Pearl River Delta region during the summer season (Huang et al., 2005; Jiang et al., 2008). (Tie et al., 2013) used the WRF model coupled with Chemistry (WRF-Chem) to analyze the impacts of a megacity (Shanghai) on O₃ formation. (Li et al., 2012) used the CMAQ (Community Multiscale Air Quality) model to analyze the contributions of individual physical and chemical processes on O₃ concentrations over the Yangtze River Delta. A recent study by (Zhu et al., 2015) showed that Shanghai has a greater impact on a downstream city (Kunshan) regarding the formation of O₃. (Chung, 1977) found that the lake-breeze (lake effect) has a significant influence on concentrations of O₃. (Lyons and Cole, 1976) found that the lake effect changes the weather field, which results in a large quantity of ozone being transported downstream. (Levy et al., 2010) found that O₃ concentrations were high over the southern Great Lakes. During the day, updrafts transport O₃ to higher altitudes over the city, and downdrafts transport O₃ to the southern Great Lakes. Generally, O₃ pollution is becoming increasingly serious. In the Yangtze River Delta region, although studies on regional urban photochemical pollution have been extensive, research on the impacts of the Taihu lake effect on city O₃ has been limited.

Figure 1.  (a) Modeling domains used in the WRF-Chem model, with a horizontal spacing of 9 km and 3 km for the two nested domains. (b) Land-use categories in the Yangtze River Delta in the inner domain. SH, SZ, CZ, WX and HZ represent Shanghai, Suzhou, Changzhou, Wuxi, and Hangzhou respectively. d01: the first layer of the model, d02: the inner layer of the model.

Taihu Lake is the fifth largest freshwater lake in China and is a typical large shallow lake. Taihu Lake Basin is one of the most developed, most densely populated and most industrial areas in China. Currently, air pollution is serious in Taihu Lake Basin. Industrial production, residential emissions and traffic generate large quantities of emissions (Huang et al., 2011). Due to the difference between the specific heat capacity of Taihu Lake and the land, the lake effect changes the meteorological field in the surrounding area, and then changes the concentrations of O₃. Therefore, it is necessary to study the impact of Taihu Lake on city O₃, which we did in the present study via two simulation experiments. The specific purpose was to study the temporal and spatial distribution characteristics, the physical and chemical mechanisms, and the interaction of O₃ between cities and Taihu. The ultimate aim was to provide scientific support for the control of O₃ pollution in the Yangtze River Delta region.

The meteorological data (wind direction) for this study were from the automatic weather stations in Suzhou, Hangzhou and Shanghai; and the concentrations of O₃ and nitrogen dioxide (NO₂) were from the atmospheric pollution monitoring stations of Suzhou, Hangzhou and Shanghai.

The impact of the Taihu Lake effect was examined using the WRF-Chem model (Grell et al., 2005; Peckam et al., 2013). This model was developed by NCAR, PNNL (Pacific Northwest National Laboratory), NOAA, and other institutions. The modeling system includes two components: a dynamic module and a chemical module. The dynamic module can calculate many different dynamic parameters and microphysical variables. The chemical module includes complete transmission (advection and diffusion), dry/wet deposition and chemical processes. A detailed description of the chemical component of the model is available in (Grell et al., 2005). The biggest advantage of the model is that the dynamic module and the chemical transmission module are completely coupled in time and space.

Figure 2.  Comparison of simulated (Sim; gray lines with circles) and observed (Obs; black lines with squares) O₃ concentrations (top row), NO₂ concentrations (middle row), and meteorological condition (wind direction; bottom row; units: degrees), at Suzhou (left column), Hangzhou (middle column), and Shanghai (right column), during 18-21 May 2014.

Figure 1 shows the modeling domains used in the WRF-Chem model, with horizontal grid spacing (grid numbers) of 9 km (99× 99) and 3 km (78× 78) for the outer and inner domains, respectively. The model was used to simulate two nested domains, and the inner domain was the Yangtze River Delta region. The central latitude and longitude of the outer and inner domain were (32^°N, 119^°E) and (31.16^°N, 120.81^°E), respectively. The projection method used was Lambert projection. The meteorological input data were from NCEP, and the grid resolution was 1^°× 1^°. The gas-phase chemical mechanism of the WRF-Chem model is RADM2 (Regional Acid Deposition Model, version 2) (Stockwell et al., 1990). The photolysis rate of the photochemical reaction process is calculated by the online Fast-J method. The Purdue Lin scheme was used for the microphysics (Lin et al., 1983). The RRTM scheme was used for the parametrization of longwave radiation. The Noah land surface scheme was used for the parametrization of the land surface process (Chen and Dudhia, 2001). The emission source used in this paper was INTEX-B (Intercontinental Chemical Transport Experiment-Phase B), and the initial conditions and outermost boundary conditions were provided by MOZART-4 (Model for Ozone And Related Chemical Traces, version 4) (Emmons et al., 2010). The simulation time of the study was 16-21 May 2014. The first two days were the spin-up time of the model, and the main time period for the analysis occurred on 20 May.

We evaluated the impact of the effect of Taihu Lake on the weather conditions and O₃ in the surrounding cites of Taihu. Fine-resolution (30 s) MODIS 20-category land-use data were used to optimally represent the Taihu Lake area. The land-use types of the inner domain are shown in Fig. 1. Shanghai, Suzhou, Changzhou, Wuxi, and Hangzhou abbreviated as SH, SZ, CZ, WX and HZ respectively. Two simulation experiments were conducted with different land-use types: the first was a reference experiment (R-E), in which we used the MODIS land-use categories. Then, a sensitivity experiment, NO_Taihu (NO_TH), was conducted in which we replaced the water of Taihu Lake with cropland, but retained the land-use types in other regions.

To evaluate the performance of the model, we compared the simulation results with observational data. Figure 2 compares the simulated and observed O₃ concentrations, NO₂ concentrations and meteorological condition (wind direction) at Suzhou, Hangzhou and Shanghai during 18-21 May 2014. The locations of the three stations are shown in Fig. 3. The wind direction shifted from the northwest during the day to the south at night on 20 May. The observed wind direction was in good agreement with the model-estimated values in the three city sites. The calculated concentrations of O₃ and NO₂ were very similar to the measured values at all of the sites, but the model slightly underestimated the peak concentrations of O₃ in the three cities in the afternoon on 20 May.

Statistical methods (correlation coefficient and normalized mean error) were used to quantitatively describe the accuracy of the model results.

The calculation method of the correlation coefficient is \begin{equation} r=\dfrac{\sum_{i=1}^N(p_i-P)(o_i-O)}{\sqrt{\sum_{i=1}^N(p_i-P)^2}\sqrt{\sum_{i=1}^N(o_i-O)^2}} ,(1) \end{equation} and that of the mean normalized error (NME) is \begin{equation} {\rm NME}=\dfrac{1}{N}\sum_{i=1}^N\dfrac{(p_i-o_i)}{o_i} . (2)\end{equation}

In the above two equations, p_i is the simulated concentration, o_i is the observed concentration, P is the average simulated concentration, O is the average observed concentration, and N is the sample number.

Figure 3.  Distribution of the simulated surface average concentrations of O₃ (shading; units: ppbv) and the wind field (vectors; units: m s^-1) in the (a) afternoon (1200-1600 LST) and (b) evening (1800-2400 LST) on 20 May 2014.

According to the recommendations of the United States EPA (O'Nell et al., 2006), if the normalized mean error of O₃ is within 15%, the results of the O₃ simulation can be accepted. The statistical analysis results of the O₃ and wind direction simulation and observed values are shown in Table 1.

The correlation coefficients between the simulated and observed concentrations of O₃ at Suzhou, Hangzhou and Shanghai were 0.91, 0.88 and 0.82, respectively, showing good correlation between the two. The normalized mean error of the simulation results at Suzhou, Hangzhou and Shanghai was 7.63%, 7.95% and 8.56%, respectively. The normalized mean errors of O₃ were all within 15%, and so the results of the O₃ simulation were accepted. The statistical analysis showed that the WRF-Chem model simulation was able to reasonably reproduce the magnitude and variation in the O₃ concentrations.

A weather chart that was published by the Hong Kong Observatory shows that the period of 20-21 May 2014 was a special springtime situation (http://gb.weather.gov.hk/wxinfo/currwx/wxchtc.htm). At 0800 LST (LST=UTC+8) 20 May, the Yangtze River Delta was experiencing a high-pressure front and a low-pressure tail. The wind direction was northwest and, with the high-pressure transition, the weather became sunny. This condition was conducive to the formation of photochemical pollution in the Yangtze River Delta. On 21 May, the Yangtze River Delta was under the control of high pressure and the wind direction was southeast. The change in the weather system caused the change in the meteorological field.

Figure 3 shows the afternoon (1200-1600 LST) and evening (1800-2400 LST) average surface concentrations of O₃ and the 5-m wind fields from the R-E experiment. A northwest wind prevailed over the Yangtze River Delta during the afternoon (1200-1600 LST), as shown in Fig. 3a. However, the wind direction was affected by the high-pressure shift from the northwest to the south in the evening (1800-2400 LST), as shown in Fig. 3b. High concentrations of O₃ were distributed in Taihu, Suzhou, Wuxi, and in the downwind region in the afternoon (1200-1600 LST). The concentrations of O₃ reached 100 ppbv over most of the Yangtze River Delta region. However, most of the large cities had low concentrations of O₃ in the evening (1800-2400 LST). Vehicle exhaust produced large amounts of NO_x emissions in the urban area at night. A large amount of NO_x emissions accumulated in the urban area, and O₃ was consumed more rapidly. In contrast with the case of the city, the concentrations of O₃ in the Taihu Lake region were still high in the evening (1800-2400 LST). The distributional characteristics of O₃ in Taihu Lake were similar to the research results of (Lyons and Cole, 1976) and (Levy et al., 2010).

Figure 4.  The distribution of the vertical cross-section of the concentrations of O₃ (shading; units: ppbv) and the wind field (vectors; units: m s^-1) through Taihu, Suzhou and Shanghai at (a) 1500 LST and (b) 2000 LST 20 May 2014.

Figure 5.  The differences of in the (a) afternoon (1200-1600 LST) and (b) evening (2000-2400 LST) average surface air temperature (shading; units: ^°C) and the 2-m wind fields (vectors; units: m s^-1) between the R-E and NO_TH experiments on 20 May 2014.

Figure 4 shows the distribution of the vertical cross section of the concentrations of O₃ and the wind field through Taihu, Suzhou and Shanghai at 1500 and 2000 LST 20 May. At 1500 LST in the afternoon, it is clear that the concentrations of O₃ were still high in Suzhou and Taihu Lake at 1 km. Compared with Taihu, the concentrations of O₃ in Suzhou were significantly higher at high altitude. This phenomenon was caused by the large source of emissions and the urban heat island of Suzhou. Because of this urban heat island, the wind field showed a convergent structure and an upward flow in Suzhou. As shown in Fig. 4a, O₃ was transported to the downwind region under the influence of prevailing winds and the lake-breeze. At 2000 LST in the evening, it was clear that the concentrations of O₃ were still high in Taihu Lake at 500 m, and that the ozone residual layer was at high altitude in Suzhou, as shown in Fig. 4b. Under the influence of the lake-breeze, O₃ was transported from Taihu Lake to Suzhou and the downstream area.

Next, we investigate the effects of Taihu Lake on the meteorological fields and O₃ in the surrounding area on 20 May 2014. Figure 5 shows the differences in the afternoon (1200-1600 LST) and evening (2000-2400 LST) average surface air temperature and 2-m wind fields between the R-E and NO_TH experiments. The differences were calculated by subtracting the NO_TH result from that of the R-E result. Taihu Lake had a substantial impact on the local temperature and wind field structure. As shown in Fig. 5a, because the specific heat capacity of the water body was larger than that of the land, the surface temperature of Taihu Lake was reduced by 4^°C-8^°C under the impact of Taihu Lake in the afternoon (1200-1600 LST), but the influence of Taihu Lake on the urban heat island was not obvious. A huge temperature difference caused a strong lake-breeze effect; the wind speed reached 5 m s^-1 and the wind field showed a divergent structure. The Taihu Lake effect was weakened during the evening; the difference in air temperature was smaller (approximately 2^°C-4^°C) and the wind field changed under the influence of the prevailing winds, as shown in Fig. 5b. This phenomenon is referred to as the Taihu Lake effect.

Figure 6 shows the differences in the afternoon (1200-1600 LST) average boundary layer height and 2-m wind fields between the NO_TH and R-E experiments. The Taihu Lake effect made the boundary layer height of the Taihu Lake region lower. The atmosphere was more stable. The Taihu Lake effect was not conducive to the diffusion of O₃ in the Taihu region. In addition, it was difficult for such high concentrations of O₃ to dissolve in the water. Therefore, these high concentrations of ozone remained over Taihu Lake for the entire day.

Figure 7 shows the differences in the afternoon (1200-1600 LST) and evening (2000-2400 LST) average surface concentrations of O₃ and the 2-m wind fields between the R-E and NO_TH experiments. In the afternoon (1200-1600 LST), Suzhou was in the right area of Taihu Lake under the influence of the prevailing wind (northwest). Therefore, Taihu Lake had less effect on the concentrations of O₃ in Suzhou region, and the difference in the average concentrations of O₃ between the R-E and NO_TH experiments just reached 6 ppbv in Suzhou. However, Taihu Lake had a significant effect on the concentrations of O₃ in the region of Taihu Lake, and the difference in the average concentrations of O₃ between the R-E and NO_TH experiments reached more than 12 ppbv in the Taihu region, as shown in Fig. 7a. However, under the influence of the background wind (south) and the lake-breeze, the impact of the Taihu Lake effect on the concentrations of O₃ in the Suzhou region and the region of Taihu Lake was obvious in the evening (2000-2400 LST), and the difference in the average concentrations of O₃ between the R-E and NO_TH experiments reached more than 12 ppbv in most regions, as shown in Fig. 7b. This phenomenon is similar to previous research results (Chung, 1977), in which the author found that lake effects influenced the concentrations of photochemical air pollution and that high concentrations of O₃ were frequently associated with a lake-breeze.

Figure 6.  The differences in the afternoon (1200-1600 LST) average boundary layer height (shading; units: m) and the 2-m wind fields (vectors; units: m s^-1) between the NO_TH and R-E experiments on 20 May 2014.

Figure 7.  The differences in the (a) afternoon (1200-1600 LST) and (b) evening (2000-2400 LST) average surface concentrations of O₃ (shading; units: ppbv) and the 2-m wind fields (vectors; units: m s^-1) between the R-E and NO_TH experiments on 20 May 2014.

Figure 8.  The average contributions of individual physical and chemical processes to the buildup of O₃ over (a) Taihu and (b) Suzhou during daytime (0800-1600 LST) and nighttime (1800-2400 LST) on 20 May 2014. The abbreviations of the process quantities are chem (chemical processes), vmix (vertical mixing), conv (convection processes), advz (vertical advection), and advh (horizontal advection).

To study the influence of the Taihu Lake effect on O₃ formation, the contributions of the physical and chemical processes to O₃ formation were analyzed. The WRF-Chem model provided the process quantities of O₃ formation (chemical processes, vertical mixing, convection processes, vertical advection and horizontal advection). The chemical process quantity is the photochemical reaction of O₃. If the chemical process is positive, the O₃ formation rate is larger than the consumption rate in the O₃ photochemical reaction. The vertical mixing is connected with turbulence, and the vertical advection is connected with the vertical wind direction and the concentration of O₃. The horizontal advection is connected with the horizontal wind direction and the concentration of O₃. If the horizontal advection is positive, the O₃ is transported from the other regions to the local area. Figure 8 shows the contributions of the individual physical and chemical processes to the buildup of O₃ over Taihu Lake and Suzhou during the daytime (0800-1600 LST) and nighttime (1800-2400 LST) on 20 May. Due to the influence of the Taihu Lake effect, the larger changes are horizontal and vertical advections in the R-E experiment relative to the NO_TH experiment in Taihu Lake in the daytime (0800-1600 LST), as shown in Fig.8a. Research suggests that O₃ chemical production mainly occurs at high altitude in the Yangtze River Delta region (Li et al., 2012). During the daytime (0800-1600 LST), the obvious downward flows were caused by the strong divergent wind field structure in Taihu. The O₃ generated by the gas-phase chemical processes at high altitude were transported to the near-surface in Taihu. Therefore, the vertical advection was transformed from a small negative contribution (-13.4 ppbv) in the NO_TH experiment into a major positive contribution (25.9 ppbv) in the R-E experiment, to the O₃ formation in Taihu; and the horizontal advection was transformed from a small positive contribution (10.2 ppbv) in the NO_TH experiment into a major negative contribution (-34.5 ppbv) in the R-E experiment, to the generation of O₃ in Taihu. The chemical processes both constituted a small positive contribution (13.5 ppbv and 8.4 ppbv for R-E and NO_TH, respectively) to the generation of O₃ in Taihu. Vertical mixing both made a small negative contribution (-3.2 ppbv and -4.8 ppbv for R-E and NO_TH, respectively). During nighttime (1800-2400 LST), and because the Taihu Lake effect became weakened, the changes in the contributions of physical and chemical processes to the generation of O₃ were small in Taihu Lake, as shown in Fig. 8a. However, the influence of the Taihu Lake effect on O₃ was obvious. The difference in the average surface concentrations of O₃ between R-E and NO_TH reached more than 12 ppbv in most areas of Taihu Lake in the evening (2000-2400 LST).

In the afternoon (1200-1600 LST), Suzhou (the surrounding city of Taihu Lake) was in the upstream area of the prevailing wind (northwest). Therefore, the influence of the Taihu Lake effect on O₃ in Suzhou was not significant, and the difference in the average concentrations of O₃ between the R-E and NO_TH experiments just reached 6 ppbv in Suzhou. Because of the urban heat island, the wind field showed a convergent structure and an upward air current in Suzhou. The Taihu Lake effect was able to weaken the urban heat island effect in Suzhou. Therefore, the negative contribution of vertical advection to O₃ generation was smaller in the R-E experiment (-35.6 ppbv) than in the NO_TH experiment (-42.5 ppbv) in Suzhou in the daytime (0800-1600 LST). Because O₃ in Taihu Lake was transported to Suzhou under the influence of a lake-breeze, the positive contribution of horizontal advection to O₃ generation was bigger in the R-E experiment (37.4 ppbv) relative to the NO_TH experiment (29.4 ppbv) in Suzhou in the daytime (0800-1600 LST), as shown in Fig. 8b. High concentrations of surface NO_x caused a low contribution of the chemical process to the O₃ generation in Suzhou (3.5 ppbv and 3.4 ppbv for R-E and NO_TH, respectively), and the difference in the chemical processes in the two experiments was small. The vertical mixing made a small positive contribution to the O₃ generation in Suzhou (12.5 ppbv and 16.3 ppbv for R-E and NO_TH, respectively). In the evening (2000-2400 LST), Suzhou (the surrounding city of Taihu) was in the downstream area of the direction of the prevailing wind (south) and the lake-breeze of Taihu Lake. Therefore, O₃ in the Taihu Lake region was clearly transported to Suzhou. Therefore, the influence of the Taihu Lake effect on O₃ in Suzhou was significant, and the difference in the average surface concentrations of O₃ between the R-E and NO_TH experiments reached more than 12 ppbv in Suzhou in the evening (2000-2400 LST). During nighttime (1800-2400 LST), although the Taihu Lake effect became weakened, the changes in the contributions of the vertical advection and horizontal advection to O₃ generation were similar to the changes in the daytime (0800-1600 LST), as shown in Fig. 8b. With a large amount of discharged NO_x and the photochemical reaction having stopped, the chemical process resulted in a large negative contribution to the O₃ generation in Suzhou. Similar to that during the daytime (0800-1600 LST), vertical mixing made a positive contribution to the O₃ generation in Suzhou. Compared with daytime, the positive contribution of vertical mixing became smaller. The results showed that the concentrations of O₃ in the Taihu Lake region were continuously high all day, which resulted in a large quantity of O₃ being transported to the downstream area. However, a large number of O₃ precursors from human emissions caused the high concentrations of O₃ in the Taihu Lake region. So, an aim in the region must be to reduce the emissions of pollutants into the atmosphere.

This study investigated the impact of Taihu Lake on the O₃ in the Yangtze River Delta region using observations and WRF-Chem model simulations. The WRF-Chem model simulation was able to reasonably reproduce the magnitude and variation in O₃ and NO₂ concentrations, as well as the meteorological condition (wind direction) at Suzhou, Hangzhou and Shanghai, during 18-21 May 2014.

Comparisons between the control and sensitivity experiments revealed the Taihu Lake effect and the impact of Taihu Lake on the O₃ in the Yangtze River Delta region. The difference in the average surface concentrations of O₃ between the R-E and NO_TH experiments reached more than 12 ppbv in most areas of Taihu Lake, throughout the day, on 20 May. During daytime (0800-1600 LST), the influence of the Taihu Lake effect on O₃ in the Suzhou region was not significant. However, the influence of the Taihu Lake effect on O₃ in the Suzhou region was obvious at nighttime (1800-2400 LST).

It was also determined that Taihu Lake had a large impact on the meteorological fields, and that the local temperature and wind field structure changed. As a result, the larger changes in the physical and chemical processes were in the form of horizontal and vertical advections under the influence of the Taihu Lake effect in Taihu Lake and Suzhou. For example, the vertical advection was transformed from a small negative contribution (NO_TH) into a major positive contribution (R-E), to the O₃ generation in Taihu Lake, during the daytime (0800-1600 LST).