Modeling the Effects of Climate Change on Surface Ozone during Summer in the Yangtze River Delta Region, China

Future climate change can impact ozone concentrations by changing regional meteorological factors related to ozone (O3) pollution. To better understand the variations of meteorological factors and their effects on O3 formation processes under future climate conditions, we model the present and the future meteorology and air quality in summer over the Yangtze River Delta (YRD) region by using the Weather Research and Forecasting Model with Chemistry module (WRF/Chem), which is driven by the outputs of Community Climate System Model version 4 (CCSM4). The simulations predict that solar radiation, 2-m air temperature, and wind speed increase in the daytime over most of the YRD region. Absolute humidity and precipitation increase in the north and decrease in the south, while the planetary boundary layer height (PBLH) has an opposite change pattern displaying a decrease in the north and an increase in the south. The southerly wind will be strengthened in the daytime. At night, the change patterns of the meteorological factors are similar to the daytime but with small variations. Meanwhile, O3 and its precursors all increase in the north and decrease in the south. The increases of NOx, volatile organic compounds (VOC), and CO are related with the decreases of PBLH and the input effect of stronger southerly wind, while the decreases are attributed to the output effect of the stronger southerly wind. During the daytime, the increase of surface O3 in the north is dominated by the chemical processes related with the increases of solar radiation, air temperature, and O3 precursors. The decrease of surface O3 in the south is mainly caused by the transport process changing with the strengthened southerly wind. At night, the surface O3 changing the amplitude is less than the daytime. The less O3 variations at night can be attributed to an O3 titration reaction with NO, the changes in NOx concentrations, and the increases of nocturnal PBLH. With the aid of H2O2/HNO3, O3 formation in the YRD region is found to be easily affected by NOx in the future. The findings can help to understand the changing trend of O3 in the YRD region and can propose reasonable pollution control policies.


Introduction
At present, climate change and ambient air quality deterioration are serious issues of the atmospheric environment. In the past, they were separately studied by researchers and policymakers. As the research on the relationship between weather, climate, and air pollution deepened, the interaction between climate change and air quality has drawn more concern [1][2][3][4]). Previous studies focused on the impacts of greenhouse gases and aerosols on climate system [5,6], and there are relatively As natural emissions and their effects in the YRD under future climate conditions have been studied [8], the impacts of climate change on individual processes of O 3 formation are discussed in this paper. The remainder of this paper is organized as follows. The model, its configurations, and the input data are introduced in Section 2. The validation of model performance, the change in future meteorology, the impact of climate change on O 3 , its precursors, and O 3 -NO X -VOC sensitivity are discussed in Section 3. Finally, a brief summary is given in Section 4.

Model Description and Configuration
In this study, WRF-Chem version 3.5 (NOAA, Colorado, USA) was applied to investigate the impacts of climate change on air quality over the YRD. WRF-Chem is a new generation of air quality modeling system, which is composed of the meteorological component (WRF) and the air quality component (Chem). The two components are fully coupled, and they use the same coordinates and physical parameterizations [34]. WRF-Chem has been widely adopted in simulating the air quality of Chinese city clusters and investigating its formation mechanism [28][29][30][35][36][37]. Also, it is of a relatively good capability in simulating climate change and its effects on air quality [7,8].
As shown in Figure 1, three nested domains were used. The model domains were centered at (32. As natural emissions and their effects in the YRD under future climate conditions have been studied [8], the impacts of climate change on individual processes of O3 formation are discussed in this paper. The remainder of this paper is organized as follows. The model, its configurations, and the input data are introduced in Section 2. The validation of model performance, the change in future meteorology, the impact of climate change on O3, its precursors, and O3-NOX-VOC sensitivity are discussed in Section 3. Finally, a brief summary is given in Section 4.

Model Description and Configuration
In this study, WRF-Chem version 3.5 (NOAA, Colorado, USA) was applied to investigate the impacts of climate change on air quality over the YRD. WRF-Chem is a new generation of air quality modeling system, which is composed of the meteorological component (WRF) and the air quality component (Chem). The two components are fully coupled, and they use the same coordinates and physical parameterizations [34]. WRF-Chem has been widely adopted in simulating the air quality of Chinese city clusters and investigating its formation mechanism [28][29][30][35][36][37]. Also, it is of a relatively good capability in simulating climate change and its effects on air quality [7,8].
As shown in Figure 1, three nested domains were used. The model domains were centered at (32.  The major physical options selected in WRF-Chem simulations are shown in Table 1. The Purdue Lin microphysics scheme [38] makes progress that the addition of a snow field to the cloud model significantly modifies the microphysical processes; it is more consummate. In the RRTM (Rapid Radiative Transfer Model) longwave radiation scheme, the speed of execution compares favorably with those of other rapid radiation models, and the model can be used in general circulation models [39]. The Goddard shortwave radiation scheme results in considerable improvement in reproducing the model's thermal structure, such as the zonal mean air temperature, its latitudinal gradient, and vertically integrated temperature [40]. In the Kain-Fritsh cumulus parameterization scheme, the specific formation of the modifications can produce desired effects in numerical weather prediction The major physical options selected in WRF-Chem simulations are shown in Table 1. The Purdue Lin microphysics scheme [38] makes progress that the addition of a snow field to the cloud model significantly modifies the microphysical processes; it is more consummate. In the RRTM (Rapid Radiative Transfer Model) longwave radiation scheme, the speed of execution compares favorably with those of other rapid radiation models, and the model can be used in general circulation models [39]. The Goddard shortwave radiation scheme results in considerable improvement in reproducing the model's thermal structure, such as the zonal mean air temperature, its latitudinal gradient, and vertically integrated temperature [40]. In the Kain-Fritsh cumulus parameterization scheme, the specific formation of the modifications can produce desired effects in numerical weather prediction and render the scheme closer to observations and cloud-resolving modeling studies [41]. The NOAH/LSM (NCAR, Boulder, CO, USA) scheme can not only provide reasonable diurnal variations of surface heat flues but also correct seasonal evolutions of soil moisture in the context of a long-term data assimilation system [42]. The MYJ (Mellor-Yamada-Janjic) PBL scheme makes progress on the basis of the MY PBL scheme, and the new scheme for calculating the MY level 2.5 master length scale is rectified [43]. The chemical mechanism used to simulate gas concentration is the Carbon-Bond Mechanism version Z (CBMZ). The CBMZ is developed from a new lumped-structure mechanism, largely based on the widely used Carbon Bond Mechanism (CBM-IV). Major improvements include a revised inorganic chemistry, an explicit treatment of the lesser reactive paraffins, methane, and ethane [44]. The adopted option for aerosol is the four-bin sectional Model for Simulating Aerosol Interactions and Chemistry (MOSAIC). MOSAIC is found to be in excellent agreement with a benchmark version of the model using a rigorous solver for integrating the stiff ordinary differential equations (ODEs). Therefore, MOSAIC is a good choice in air quality and regional/global aerosol models [45]. The initial and boundary chemical conditions are derived from the modeling results of the global chemistry transport model MOZART-4.

Simulation Cases
To evaluate how climate change influences the O 3 concentration in the YRD, two simulation cases are specially designed and conducted. One uses the present meteorology in 2014, 2015, and 2016 and the present emission (referred to as PREMET hereafter). The other uses the future meteorology in 2050, 2051, and 2052 and the present emission (referred to as FUTMET hereafter). In the YRD, the observation results show that a high O 3 concentration usually appears in late spring and early summer and that O 3 is more sensitive to environment in summer [17,18,29]. Thus, the PREMET simulations are conducted from 00 UTC July 1st to 18 UTC July 31st in 2014, 2015, and 2016. Also, the FUTMET simulations are conducted from 00 UTC July 1st to 18 UTC July 31st in 2050, 2051, and 2052. An initial 48-hour model integration period is used for the spin-up of the simulations. The difference between the modeling results from FUTMET and PREMET can demonstrate the effect of climate change on air quality (EF climate ), which can be calculated by the following equation: where t represents one modeling time step; N represents the total modeled time; V FUTMET,t and V PREMET,t are the hourly modeling outputs of variable V (meteorological factors or air pollutants) from FUTMET and PREMET, respectively.

Present and Future Climate Data
The present and future climate data are used to drive WRF as the initial meteorological fields and boundary conditions. These data are obtained from the National Center for Atmospheric Research (NCAR) Community Climate System Model version 4 (CCSM4) outputs with a horizontal resolution of T85 (about 1.41 • ). The CCSM4 RCP4.5 outputs for 2050, 2051, and 2052 used in this study are based on the Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report. RCP4.5 is a midline future climate scenario for CO 2 emission and economic growth, which can ensure our results to be moderate. For the present year, the CCSM4 outputs in 2014, 2015, and 2016 are used. The data are also compared with the National Centers for Environmental Prediction (NCEP) Final Analysis (FNL) data with the spatial resolution of 1 • , and the difference is ignorable.

Anthropogenic Emissions
For the PREMET case, the anthropogenic emissions in China are from the Multi-resolution Emission Inventory (MEIC) developed by Tsinghua University based on a technology-based emission model. The MEIC inventory contains monthly anthropogenic emissions of SO 2 , NO X , CO, NH 3 , PM 2.5 , PM 10 , BC, OC, and VOCs in five sectors (agriculture, industry, power plants, residential, and transportation) [46]. Those for the areas out of China are mainly from the inventory developed by the NASA INTEX-B (NASA, Washington, DC, USA), including the emissions of SO 2 , NO x , CO, PM 10 , PM 2.5 , BC, OC, and VOCs from the power, residential, industry, and transportation sectors with a resolution of 0.5 • [47]. Furthermore, these data have been specially modified for simulations in the YRD following the work of Xie et al. [8,29].
For future anthropogenic emissions, some previous investigations estimated them by some hypothetical growth factors [7,26]. However, the Chinese government has formulated a series of strict emission reduction policies to protect air quality, implying that the hypothetical increase of emissions is inapplicable. On the other hand, the main purpose of this study is to demonstrate the changes of meteorological factors under future climate and their effects on individual processes of O 3 formation. Therefore, it is reasonable to assume that the future anthropogenic emissions will remain at the current level. In the FUTMET case, the emissions are set to be the same as PREMET.

Process Analysis Method
WRF-Chem version 3.5 contains a simple process analysis function, which can present the contributions of individual atmospheric processes, including chemical reaction (CHEM), vertical mixing coupled with dry deposition (VMIX), and advection transportation with horizontal and vertical components (ADVT). These variables can show the relative significance of each process and provide a particular interpretation of air pollution. In this study, by comparing the values of CHEM, VMIX, and ADVT from FUTMET with those from PREMET, we can figure out how the changes in meteorological factors influence the individual atmospheric processes of O 3 chemistry. Moreover, the O 3 change (accumulation or consuming) in different processes is called Chem_O 3 in the CHEM process, Vmix_O 3 in the VMIX process, and Advt_O 3 in the ADVT process. The unit of Chem_O 3 , Vmix_O 3 , and Advt_O 3 is ppb/h, implying the changes in one hour in this process.

Model Performance Evaluation
The mean bias (MB), root mean square error (RMSE), and correlation coefficient (CORR) between thet observations and the simulation results from PREMET are used to verify the performance of WRF-Chem. In statistics, they are usually defined as and Their hourly observational data of 2-m temperature (T 2 ), 10-m wind speed (WS 10 ), and 2-m relative humidity (RH 2 ) in July of 2014, 2015, and 2016 are obtained from the observation database in the Wyoming Weather Web. Their hourly air pollutant concentration records can be acquired from the air quality real-time publishing platform. The observational methods and the quality assurance/quality control (QA/QC) procedures for these data strictly follow the Chinese national standard. Also, the manual inspection of invalid and lacking data is performed during data processing [17,18,[28][29][30]. Table 2 shows the statistical comparisons between the observations and the modeling results from PREMET. OBS and SIM are represented the observation and simulation. For T 2 , the model slightly overvalues T 2 at all sites, which might result from the uncertainty of urban canopy and surface parameters [28,29,36,37]. However, the overestimation is acceptable because the MB values of T 2 are only 0.5-1.7 • C, the RMSE of T 2 is 2.8-3.9 • C, and the CORR between observations and simulations are over 0.8 at all sites (statistically significant at the 95% confidence level). Moreover, the lowest value 0.5 • C for MB and the highest value 0.9 for CORR show that the best simulation of T 2 is at NJ. For RH 2 , though the modeling results are slightly underestimated, the CORR values are over 0.8 at most sites. Therefore, the simulations for RH 2 are relatively acceptable. The lowest value 0.7 for CORR is at SH, which may be attributed to the uncertainty of the land-use data that cannot well describe waters around SH [29]. For WS 10 , the modeling values are overestimated at HZ and SH but underestimated at NJ. The CORR (close to 0.5) for WS 10 is the lowest in three meteorological factors. The wind components were not simulated very well, just as in previous studies, which may be caused by urban canopy parameters [28,29,37]. Compared with previous studies, the biases in this study are acceptable. With respect to O 3 , the CORRs are more than 0.6 at most sites. The lowest value for MB (−1.2 ppb) and the highest CORR value (close to 0.7) indicate that the simulation at SH is good. However, the modeling results overestimate O 3 concentrations at NJ and HZ, and the values of MB are around 8.0. These biases might be related to the modeled stronger solar radiation reaching the surface, which can cause positive biases in T 2 and thereby produces more O 3 [29,37]. At the same time, the high bias about simulated wind might result in the O 3 concentration uncertainty. In addition, there are also some uncertainties in the emission, which might cause O 3 concentration bias.

Model Evaluation for WRF-Chem
All in all, the performance of WRF-Chem in modeling climate and air quality is acceptable over the YRD region in this study. Some biases are still found in simulations, but the difference between PREMET and FUTMET can still demonstrate how meteorological factors could impact O 3 , as all other settings are the same in both simulations.

Regional Meteorology Changes
The regional meteorological changes caused by global climate change are shown in Figures 2 and 3. The values in these figures present the differences between FUTMET and PREMET. Figure 2 shows the changes during the daytime (from 7:00 to 18:00 LST (Local standard time)). For solar radiation (Figure 2a

Regional Meteorology Changes
The regional meteorological changes caused by global climate change are shown in Figure 2 and Figure 3. The values in these figures present the differences between FUTMET and PREMET. Figure  2 shows the changes during the daytime (from 7:00 to 18:00 LST (Local standard time)). For solar radiation (Figure 2a Figure 3 shows the changes at night (from 19:00 to 6:00 LST). For T2 (Figure 3a), it is predicted to increase about 1.9 °C over the entire YRD, with the higher increases over 5.0 °C in the southwest. The increments are slightly higher than those in the daytime, which may be related to more clouds at night. More clouds may strengthen the insulation effect. For absolute humidity (Figure 3b), the regional mean changing value (9.6 μg/m 3 ) is smaller than that in the daytime, related to a weaker evaporation at night. For PBLH (Figure 3c), higher increasing values occur in the southwest of the YRD, with a maximum of 268.6 m. The pattern is similar to Figure 3a, implying the effect of air temperature on PBLH. For wind field (Figure 3d), due to the rising of T2, the wind speed increases as well. The regional average increase is 0.8 m/s, and the maximum increase is 3.9 m/s. For the wind direction, the southerly wind dominates the region at night just as the changes in the daytime.  Figure 3 shows the changes at night (from 19:00 to 6:00 LST). For T 2 (Figure 3a), it is predicted to increase about 1.9 • C over the entire YRD, with the higher increases over 5.0 • C in the southwest. The increments are slightly higher than those in the daytime, which may be related to more clouds at night. More clouds may strengthen the insulation effect. For absolute humidity (Figure 3b), the regional mean changing value (9.6 µg/m 3 ) is smaller than that in the daytime, related to a weaker evaporation at night. For PBLH (Figure 3c), higher increasing values occur in the southwest of the YRD, with a maximum of 268.6 m. The pattern is similar to Figure 3a, implying the effect of air temperature on PBLH. For wind field (Figure 3d), due to the rising of T 2 , the wind speed increases as well. The regional average increase is 0.8 m/s, and the maximum increase is 3.9 m/s. For the wind direction, the southerly wind dominates the region at night just as the changes in the daytime. The estimated change trends and intensities of the above meteorological factors are in agreement with previous findings [7,8,26]. During the daytime, the increase of solar radiation and T2 can enhance the O3 photochemical reactions over the entire YRD. Although the increase of PBLH in the north can dilute O3 concentration in lower atmosphere, the decrease in the south can increase the O3 pollution level. However, the strengthened southerly wind can transport more O3 from the south to north, which may increase O3 in the north and decrease O3 in the south. The change trend of O3 in the daytime might be different from that at night. The increase of T2 can enhance the titration reaction between NO and O3. The increase of PBLH in the north can bring surface O3 to the higher layers. The southerly wind may increase O3 in the north but decrease it in the south. Consequently, it is hard to tell the change of O3 just by the changes in meteorological factors either in the daytime or at night. Further analyses are needed. Figure 4 shows the differences of ozone precursors at the surface between FUTMET and PREMET. For NOx, the change pattern during the daytime (Figure 4a) is similar to that at night (Figure 4b). The increases appear in the north of Shanghai and Hangzhou. In the daytime, the  The estimated change trends and intensities of the above meteorological factors are in agreement with previous findings [7,8,26]. During the daytime, the increase of solar radiation and T 2 can enhance the O 3 photochemical reactions over the entire YRD. Although the increase of PBLH in the north can dilute O 3 concentration in lower atmosphere, the decrease in the south can increase the O 3 pollution level. However, the strengthened southerly wind can transport more O 3 from the south to north, which may increase O 3 in the north and decrease O 3 in the south. The change trend of O 3 in the daytime might be different from that at night. The increase of T 2 can enhance the titration reaction between NO and O 3 . The increase of PBLH in the north can bring surface O 3 to the higher layers. The southerly wind may increase O 3 in the north but decrease it in the south. Consequently, it is hard to tell the change of O 3 just by the changes in meteorological factors either in the daytime or at night. Further analyses are needed. Figure 4 shows the differences of ozone precursors at the surface between FUTMET and PREMET. For NO x , the change pattern during the daytime (Figure 4a) is similar to that at night (Figure 4b).

Changes in Ozone Precursors
The increases appear in the north of Shanghai and Hangzhou. In the daytime, the maximum increase is 6.5 ppb in Shanghai. At night, the maximum increase is 11.9 ppb in Hangzhou. These increases of NO x match the decreases of PBLH, implying the accumulation of air pollutants under worse diffusion conditions. Also, the stronger southerly wind in the daytime can lead to increases of concentration in the north. However, more high values appear in Hangzhou at night than in the daytime, caused by an enhancement of the westerly wind along Nanjing-Shanghai line, which blocks the northward transportation of NO x . The change patterns of VOCs and CO are similar to that of NO x . The higher increments also appear in the northern YRD. The maximum changes for VOCs are 3.7 ppb (Figure 4c) in the day and 6.1 ppb at night (Figure 4d). Those values for CO are 134.3 ppb (Figure 4e) and 154.1 ppb (Figure 4f), respectively [48]. The differences of change patterns between day and night, as well as the main causes, are similar with those of NO x . maximum increase is 6.5 ppb in Shanghai. At night, the maximum increase is 11.9 ppb in Hangzhou. These increases of NOx match the decreases of PBLH, implying the accumulation of air pollutants under worse diffusion conditions. Also, the stronger southerly wind in the daytime can lead to increases of concentration in the north. However, more high values appear in Hangzhou at night than in the daytime, caused by an enhancement of the westerly wind along Nanjing-Shanghai line, which blocks the northward transportation of NOx. The change patterns of VOCs and CO are similar to that of NOx. The higher increments also appear in the northern YRD. The maximum changes for VOCs are 3.7 ppb (Figure 4c) in the day and 6.1 ppb at night (Figure 4d). Those values for CO are 134.3 ppb (Figure 4e) and 154.1 ppb (Figure 4f), respectively [48]. The differences of change patterns between day and night, as well as the main causes, are similar with those of NOx.   To understand the impacts of the changes caused by climate change in individual atmospheric process, the monthly mean differences between FUTMET and PREMET for the CHEM, VMIX, and ADVT processes of NOx in the YRD are investigated. The contributions of VMIX and ADVT to NOx variations are the largest, while CHEM has little effect. Thus, only the changes in the VMIX and ADVT processes of NOx (respectively referred to as Vmix_NOx, and Advt_NOx hereafter) at the surface are presented in Figure 5. As shown in Figure 5a   To understand the impacts of the changes caused by climate change in individual atmospheric process, the monthly mean differences between FUTMET and PREMET for the CHEM, VMIX, and ADVT processes of NO x in the YRD are investigated. The contributions of VMIX and ADVT to NO x variations are the largest, while CHEM has little effect. Thus, only the changes in the VMIX and ADVT processes of NOx (respectively referred to as Vmix_NOx, and Advt_NOx hereafter) at the surface are presented in Figure 5. As shown in Figure 5a  To understand the impacts of the changes caused by climate change in individual atmospheric process, the monthly mean differences between FUTMET and PREMET for the CHEM, VMIX, and ADVT processes of NOx in the YRD are investigated. The contributions of VMIX and ADVT to NOx variations are the largest, while CHEM has little effect. Thus, only the changes in the VMIX and ADVT processes of NOx (respectively referred to as Vmix_NOx, and Advt_NOx hereafter) at the surface are presented in Figure 5. As shown in Figure 5a Figure 6 shows the differences of surface ozone between FUTMET and PREMET in the YRD. During the daytime (Figure 7a), O3 increases in the north but decreases in the south. The high increase values with a maximum of 18.5 ppb are along the coastal areas of Jiangsu Province, while the high reductions (over −15.1 ppb) are mainly in the inland areas of Jiangsu, Zhejiang, and Anhui Provinces. At night (Figure 7b), the change pattern of O3 is similar to that in the day, but the increase and the decrease values are lower. The regional averaged change value is just −0.2 ppb, the maximum increase is 17.5 ppb, and the maximum decrease is about −12.0 ppb. Less O3 variations at night may be attributed to nocturnal O3 chemical reactions. Moreover, the above O3 change patterns are similar to those of NOx, VOCs, and CO, implying that O3 changes are tightly related to O3 precursors and are highly affected by chemical processes.   Figure 6 shows the differences of surface ozone between FUTMET and PREMET in the YRD. During the daytime (Figure 7a Figure 5a,c shows the changes during the daytime (from 7:00 to 18:00 LST). Figure 5b,d present those at night (from 19:00 to 6:00 LST). Vmix_NOx is the contribution of vertical mixing and dry deposition to NOx. Advt_NOx is the contribution of horizontal and vertical advection. Figure 6 shows the differences of surface ozone between FUTMET and PREMET in the YRD. During the daytime (Figure 7a), O3 increases in the north but decreases in the south. The high increase values with a maximum of 18.5 ppb are along the coastal areas of Jiangsu Province, while the high reductions (over −15.1 ppb) are mainly in the inland areas of Jiangsu, Zhejiang, and Anhui Provinces. At night (Figure 7b), the change pattern of O3 is similar to that in the day, but the increase and the decrease values are lower. The regional averaged change value is just −0.2 ppb, the maximum increase is 17.5 ppb, and the maximum decrease is about −12.0 ppb. Less O3 variations at night may be attributed to nocturnal O3 chemical reactions. Moreover, the above O3 change patterns are similar to those of NOx, VOCs, and CO, implying that O3 changes are tightly related to O3 precursors and are highly affected by chemical processes.  between -4.0 to 4.0 ppb (Figure 7c). Though PBLH would decrease in the north (Figure 2e, may increase O3 at surface), Chem_O3 can produce more O3 in lower atmospheres (Figure 8a) and more O3 would be transported to the upper atmospheric layer or deposition to the ground (Figure 8b). Thus, Vmix_O3 in the northern YRD mainly decreases. In the south, however, there is little change in Vmix_O3 (0-1.0 ppb/h). The high values close to some cities would be caused by the local atmospheric circulation related to urban heat island [28,29].   (Figure 2b). The higher increments over 3.0 ppb/h are mainly located along the coastal areas of Jiangsu Province, which is also the area with great increase of O 3 precursor concentration (Figure 4a,c,e). Over the southern YRD, the decreases of absolute humidity and O 3 precursor concentration should result in reductions of Chem_O 3 in extensive areas, with the maximum decrease about −2.2 ppb/h to the south of Hangzhou. Figure 7b illustrates that the changes in transport process of O 3 (Advt_O 3 ) are between −3.0 to 4.0 ppb/h, which is closely related to the variation of horizontal wind speed and wind direction [7]. Advt_O 3 rises up over 0-4.0 ppb/h in most areas of the northern YRD, with the higher increments over 2.1 ppb around Shanghai. The decreases of Advt_O 3 are mainly located in the southern YRD. The change pattern of Advt_O 3 further proves that the strengthened southerly wind in the YRD (Figure 2f (Figure 8b). Thus, Vmix_O 3 in the northern YRD mainly decreases. In the south, however, there is little change in Vmix_O 3 (0-1.0 ppb/h). The high values close to some cities would be caused by the local atmospheric circulation related to urban heat island [28,29].  Figure 7. The spatial distribution of monthly averaged differences of (a) Chem_O3, (b) Advt_O3, and (c) Vmix_O3 between FUTMET and PREMET during the daytime (from 7:00 to 18:00 LST). Chem_O3 represents the O3 chemical production process. Vmix_O3 means the contribution of vertical mixing and dry deposition to O3. Advt_O3 means the contribution of horizontal and vertical advections. At night, the changes in three main processes are smaller. The changes in Chem_O3 vary from −1.0 to 2.0 ppb (Figure 9a). The decreases of Chem_O3 mainly appear in the north, which should be attributed to the enhancement of an O3 titration reaction with NO at night due to temperature increase ( Figure 3a) and the sufficient NOx in the north (Figure 4b). In a similar way, the increases in the south are related with the decrease of NOx there (Figure 4b), which would weaken the O3 consumption reaction. Figure 10b shows that the changes in Advt_O3 are between −3.0 and 2.0 ppb in most areas. O3 accumulation along the Nanjing-Shanghai line is due to the strengthening of the easterly wind. The changes in Vmix_O3 range between −3.0 and 2.0 ppb (Figure 9c). The decreases of Vmix_O3 are related to the increases of nocturnal PBLH over the YRD (Figure 3c). The maximum increases of Vmix_O3 are in the south of the YRD. The increases are caused by great increases of wind speed that can weaken the dry deposition in this area (Figure 3d).  Figure 7. The spatial distribution of monthly averaged differences of (a) Chem_O3, (b) Advt_O3, and (c) Vmix_O3 between FUTMET and PREMET during the daytime (from 7:00 to 18:00 LST). Chem_O3 represents the O3 chemical production process. Vmix_O3 means the contribution of vertical mixing and dry deposition to O3. Advt_O3 means the contribution of horizontal and vertical advections. At night, the changes in three main processes are smaller. The changes in Chem_O3 vary from −1.0 to 2.0 ppb (Figure 9a). The decreases of Chem_O3 mainly appear in the north, which should be attributed to the enhancement of an O3 titration reaction with NO at night due to temperature increase ( Figure 3a) and the sufficient NOx in the north (Figure 4b). In a similar way, the increases in the south are related with the decrease of NOx there (Figure 4b), which would weaken the O3 consumption reaction. Figure 10b shows that the changes in Advt_O3 are between −3.0 and 2.0 ppb in most areas. O3 accumulation along the Nanjing-Shanghai line is due to the strengthening of the easterly wind. The changes in Vmix_O3 range between −3.0 and 2.0 ppb (Figure 9c). The decreases of Vmix_O3 are related to the increases of nocturnal PBLH over the YRD (Figure 3c). The maximum increases of Vmix_O3 are in the south of the YRD. The increases are caused by great increases of wind speed that can weaken the dry deposition in this area (Figure 3d).  . The spatial distribution of monthly averaged differences of (a) Chem_O3, (b) Advt_O3, and (c) Vmix_O3 between FUTMET and PREMET at night (from 19:00 to 06:00 LST). Chem_O3 represents the O3 chemical production process. Vmix_O3 means the contribution of vertical mixing and dry deposition to O3. Advt_O3 means the contribution of horizontal and vertical advections.

Impact of Climate Change on Regional Ozone Control Policy
The ratio of certain secondary photochemical products, such as H2O2/NOZ and H2O2/HNO3, can be used as the efficient indicators to distinguish NOX-or VOC-sensitive regimes of O3 chemistry . In this study, H2O2/HNO3 is adopted to investigate the O3-NOX-VOC sensitivity in the YRD under different climate conditions. According to the previous study [8], the transition value of H2O2/HNO3 in the present summer is quantified to be 0.3-0.5, while it will change to 0.4-0.8 under the future conditions. Figure 10 shows the mean values of H2O2/HNO3 at the first model layer over the YRD during afternoon (13:00-16:00 LST) from FUTMET and PREMET. For PREMET, low values (<0.3) of H2O2/HNO3 are mainly located in Shanghai, Nanjing, Hangzhou, and most land areas in the northern YRD (Figure 10a). The results mean that the O3 chemistry in the typical cities and the northern YRD is VOC-sensitive and that VOCs should be preferentially reduced in these areas at present. However, in the future, more and more areas in YRD are covered by high values of H2O2/HNO3 (Figure 10b). The values in the vast majority of grids are over 0.8, which means the O3 chemistry is NOx-sensitive in most parts of the YRD. Lower values (<0.4) only appear around Shanghai. The values in the cities of southern Jiangsu and northeastern Zhejiang are also close to 0.4. The change in the distribution pattern of H2O2/HNO3 implies that O3 chemistry in the future of YRD tends to be insensitive to VOCs and is more easily affected by NOx, which is in accordance with the findings of Xie et al [8] (c) Figure 9. The spatial distribution of monthly averaged differences of (a) Chem_O 3 , (b) Advt_O 3 , and (c) Vmix_O 3 between FUTMET and PREMET at night (from 19:00 to 06:00 LST). Chem_O 3 represents the O 3 chemical production process. Vmix_O 3 means the contribution of vertical mixing and dry deposition to O 3 . Advt_O 3 means the contribution of horizontal and vertical advections.

Conclusions
The effects of climate change on surface ozone in summer over the YRD region is studied by using WRF-Chem, with a special emphasis on the changes in meteorological factors and their impacts on individual atmospheric processes of O3 formation.
The simulations predict that solar radiation and a 2-m air temperature increase in the daytime in most of the YRD region with average increments of 33.

Impact of Climate Change on Regional Ozone Control Policy
The ratio of certain secondary photochemical products, such as H 2 O 2 /NO Z and H 2 O 2 /HNO 3 , can be used as the efficient indicators to distinguish NO X -or VOC-sensitive regimes of O 3 chemistry . In this study, H 2 O 2 /HNO 3 is adopted to investigate the O 3 -NO X -VOC sensitivity in the YRD under different climate conditions. According to the previous study [8], the transition value of H 2 O 2 /HNO 3 in the present summer is quantified to be 0.3-0.5, while it will change to 0.4-0.8 under the future conditions. Figure 10 shows the mean values of H 2 O 2 /HNO 3 at the first model layer over the YRD during afternoon (13:00-16:00 LST) from FUTMET and PREMET. For PREMET, low values (<0.3) of H 2 O 2 /HNO 3 are mainly located in Shanghai, Nanjing, Hangzhou, and most land areas in the northern YRD (Figure 10a). The results mean that the O 3 chemistry in the typical cities and the northern YRD is VOC-sensitive and that VOCs should be preferentially reduced in these areas at present. However, in the future, more and more areas in YRD are covered by high values of H 2 O 2 /HNO 3 (Figure 10b). The values in the vast majority of grids are over 0.8, which means the O 3 chemistry is NO x -sensitive in most parts of the YRD. Lower values (<0.4) only appear around Shanghai. The values in the cities of southern Jiangsu and northeastern Zhejiang are also close to 0.4. The change in the distribution pattern of H 2 O 2 /HNO 3 implies that O 3 chemistry in the future of YRD tends to be insensitive to VOCs and is more easily affected by NO x , which is in accordance with the findings of Xie et al. [8]

Conclusions
The effects of climate change on surface ozone in summer over the YRD region is studied by using WRF-Chem, with a special emphasis on the changes in meteorological factors and their impacts on individual atmospheric processes of O 3 formation.
The simulations predict that solar radiation and a 2-m air temperature increase in the daytime in most of the YRD region with average increments of 33.2 W/m 2 and 1. For the effects of climate change on air pollutants, the change patterns of O 3 precursors (NO x , VOC, and CO) are similar, with an increase in the north and a decrease in the south. The maximum increments of NO x , VOC, and CO in the day can reach 6.5, 3.7, and 134.3 ppb, respectively. Those at night are 11.9, 6.1, and 154.1 ppb, respectively. According to a process analysis, their increases are related with the decreases of PBLH and the input effect of stronger southerly wind, while the decreases are attributed to the output effect of the stronger southerly wind. Surface O 3 variations will increase in the north and decrease in the south during the daytime. According to process analysis, the increase of surface O 3 in the north is dominated by gas phase chemical process related to the increases of solar radiation, air temperature, and O 3 precursors. The decrease in the south is mainly caused by the transport process changing with the strengthened southerly wind. During the nighttime under the future climate, the surface O 3 changes amplitude is less than the daytime, with the same change pattern as that in the day. The less O 3 variations at night can be attributed to an O 3 titration reaction with NO, the changes in NO x concentrations, and the increases of nocturnal PBLH. Moreover, climate change can affect the chemical relationship between O 3 and its precursors. With the aid of H 2 O 2 /HNO 3 , O 3 -NO x -VOC sensitivity over the YRD region is found to be easily affected by NO x in the future.
This study provides us a scope of understanding how the future climate affects the surface ozone in the YRD region. However, only the IPCC RCP4.5 scenario from CCSM4 is considered. To obtain a comprehensive understanding, the future meteorological inputs should be provided by more global climate models based on more future climate scenarios, including Socio-Economic Pathways (SSPs) scenarios. Moreover, the manmade emissions may decrease in the future, and the land-use types may continue to change for decades. These future changes should be taken into consideration in the follow-up studies as well.