Mitigating ozone damage to ecosystem productivity through sectoral and regional emission controls: a case study in the Yangtze River Delta, China

GEOS-Chem is a widely used global 3D chemical transport model for simulating concentrations of gas-phase pollutants and aerosols with a detailed HO_x –NO_x –VOC–O₃–halogen–aerosol chemistry mechanism (Kim et al 2015a, Lee et al 2017, Lu et al 2019, Porter and Heald 2019, Gong et al 2021). This model uses Fast-JX scheme to compute photolysis rates (Bian and Prather 2002). Biogenic VOC (BVOC) emissions are computed based on the Model of Emissions of Gases and Aerosols from Nature (MEGAN v2.1) (Guenther et al 2012). The dry deposition for gases is computed based on the resistance-in-series scheme (Wesely 1989). The global anthropogenic emission inventory is obtained from Community Emissions Data System (Hoesly et al 2018). Anthropogenic emissions in China are computed based on Multiresolution Emission Inventory (MEIC), which estimates SO₂, NO_x , CO, NMVOC, NH₃, PM₁₀, PM_2.5, BC, OC, and CO₂ emissions from five source sectors including agriculture (AGR), industry (IND), power (POW), residential (RES), and transportation (TRA) (Zheng et al 2018). Notably, agriculture source sector only includes NH₃ emissions. Total anthropogenic emissions from each sector in YRD are summarized in table S1 (available online at stacks.iop.org/ERL/17/065008/mmedia). In this study, nested GEOS-Chem version 12.0.0 with resolution of 0.5° $\times$0.625° is driven by the Modern-Era Retrospective analysis for Research and Applications, version 2 (MERRA2) meteorological fields to predict the changes of tropospheric O₃ under different anthropogenic emission controls.

The YIBs model is a process-based vegetation model designed to simulate global carbon cycle with dynamical prediction of leaf area index (LAI) and tree height (Yue and Unger 2015). Leaf-level photosynthesis is computed using Farquhar et al (1980) scheme. The leaf photosynthesis is further upscaled to canopy level by separating the responses of sunlit and shaded leaves to diffuse and direct radiation (Spitters 1986). Calculations of LAI and carbon allocation follow the Joint UK Land Environment Simulator (JULES) model (Clark et al 2011). The Moderate Resolution Imaging Spectroradiometer (MODIS) land types and cover fractions are aggregated into eight plant functional types (figure S1), including evergreen needleleaf forest, deciduous broadleaf forest, evergreen broadleaf forest, shrubland, C₃/C₄ grasses, and C₃/C₄ crops to calculate carbon uptake in the YRD. The YIBs model applies Kim et al (2015b) scheme to build phenology. For crops, the plant date and harvest date are fixed based on observations. For other plant functional types, leaf phenology is generally controlled by temperature, water availability, and photoperiod. The YIBs model has been comprehensively validated against site-level observations and satellite retrievals (Yue et al 2015, 2017, Yue and Unger 2018). Since 2020, the YIBs model has joined the multi-model ensemble project 'Trends in the land carbon cycle (TRENDY)' to estimate global carbon budget (Friedlingstein et al 2020). In this study, the YIBs model is driven by the same meteorological fields (MERRA2) as GEOS-Chem with a higher resolution of 0.25° $\times$0.3125°.

The semi-mechanistic O₃ damage scheme is coupled into YIBs to quantify the carbon losses by surface O₃ (Sitch et al 2007, Lei et al 2020, Unger et al 2020). The O₃-induced photosynthesis loss ($A^{\prime}$, $\mu {\text{mol }}{{\text{m}}^{ - 3}}$) is calculated by multiplying the O₃ damage factor ($\alpha$) to the original leaf photosynthesis ($A$, $\mu {\text{mol }}{{\text{m}}^{ - 3}}$):

$$\begin{equation}A{^{^{\prime}}} = A \times \alpha .\end{equation} \tag{ 1 }$$

The $\alpha$ is calculated based on excessive O₃ flux (${\text{E}}{{\text{F}}_{{{\text{O}}_{\text{3}}}}}$, $\mu {\text{mol }}{{\text{m}}^{ - 2}}{{\text{s}}^{ - 1}}$) and sensitivity coefficient ($\beta$):

$$\begin{equation}\alpha = {\text{E}}{{\text{F}}_{{{\text{O}}_{\text{3}}}}} \times \beta .\end{equation} \tag{ 2 }$$

The $\beta$ has two values for each vegetation type, representing low to high O₃-damaging sensitivities. The ${\text{E}}{{\text{F}}_{{{\text{O}}_{\text{3}}}}}$ is calculated based on stomatal O₃ flux (${F_{{{\text{O}}_{\text{3}}}}}$, $\mu {\text{mol }}{{\text{m}}^{ - 2}}{{\text{s}}^{ - 1}}$) and O₃ damaging threshold (${T_{{{\text{O}}_{\text{3}}}}}$, $\mu {\text{mol }}{{\text{m}}^{ - 2}}{{\text{s}}^{ - 1}}$):

$$\begin{equation}{\text{E}}{{\text{F}}_{{{\text{O}}_{\text{3}}}}} = {\text{max}}\left( {{F_{{{\text{O}}_{\text{3}}}}} - {T_{{{\text{O}}_{\text{3}}}}},0} \right).\end{equation} \tag{ 3 }$$

The ${T_{{{\text{O}}_{\text{3}}}}}$ represents the O₃ tolerance for each vegetation type. The ${F_{{{\text{O}}_{\text{3}}}}}$ is calculated based on surface O₃ concentrations ($\left[ {{{\text{O}}_{\text{3}}}} \right]$, $\mu {\text{mol }}{{\text{m}}^{ - 3}}$), aerodynamic resistance (${R_{\text{a}}}$, ${\text{s }}{{\text{m}}^{ - 1}}$), boundary layer resistance (${R_{\text{b}}}$, ${\text{s }}{{\text{m}}^{ - 1}}$), stomatal conductance (${R_{\text{s}}}$, ${\text{s }}{{\text{m}}^{ - 1}}$), and the ratio of leaf resistance O₃ to leaf resistance of water vapor ($k$):

$$\begin{equation}{F_{{{\text{O}}_3}}} = \left[ {{{\text{O}}_{\text{3}}}} \right]/\left( {{R_{\text{a}}} + {R_{\text{b}}} + k \times {R_{\text{s}}}} \right).\end{equation} \tag{ 4 }$$

The O₃-damaging gross primary productivity (GPP) is calculated by integrating $A^{\prime}$ over all canopy layers:

$$\begin{equation}{\text{GPP}}^{\prime} = \int\limits_0^{{\text{LAI}}} A{^{^{\prime}}}{\text{d}}L.\end{equation} \tag{ 5 }$$

This scheme has been well validated against hundreds of experimental data collected globally and in China (Yue et al 2017, Yue and Unger 2018).

In this study, we conduct ten GEOS-Chem runs and 20 YIBs runs to explore the changes of O₃-induced GPP reductions in the YRD by different anthropogenic emission control strategies (table 1). In the control run (CTRL), the GEOS-Chem model is forced with all sources of emissions. The SHUT run uses the same emissions as CTRL, except for shutting down anthropogenic emissions in China. In five sensitivity runs, the model is forced with all anthropogenic emissions but 50% emission reductions in YRD from a specific sector, including agriculture (AGR_RED50%), industry (IND_RED50%), power (POW_RED50%), residential source (RES_RED50%), or transportation (TRA_RED50%). We perform another three runs with GEOS-Chem to explore the impacts of collaborative emission reductions and regional transport on O₃ pollution, including 50% reductions of anthropogenic emissions outside YRD (YRD_{out_}RED50%), inside YRD (YRD_in_RED50%), or the whole anthropogenic emissions in China (CHN_RED50%). For above simulations, anthropogenic emissions of all pollution species are retained or perturbed with the same ratios (e.g. reduce by 50% or 100%). All runs are conducted from January to September in 2017 and the results of last 6 months are used to investigate the mitigation effects of anthropogenic emission controls on O₃-induced GPP losses during the growing season (April–September). The simulated hourly surface O₃ concentrations from each GEOS-Chem run are used to drive the YIBs model with either high or low sensitivities, making a total of 20 YIBs runs. The derived GPP losses with low and high O₃-damaging sensitivities are averaged for comparisons and one standard deviation of low and high O₃-damaging sensitivities represents uncertainties of O₃ damage on GPP.

Ground-based hourly O₃ concentrations in 2017 growing-season are collected from the observational network of the China Ministry of Ecology and Environment (MEE) (http://datacenter.mee.gov.cn/). With quality control, continuous measurements at 157 sites in YRD are selected to evaluate the simulated surface O₃ from GEOS-Chem. We use the daily maximum 8 h average (MDA8) O₃ as the metric in validation. Benchmark GPP are obtained from the Global LAnd Surface Satellite (GLASS) product generation system (Yuan et al 2010, Zhao et al 2013). The GLASS GPP agrees well with FLUXNET observations (Yu et al 2018) and has been widely used in the estimate of global and regional carbon budgets (Jiang et al 2021).

Simulated mean surface O₃ concentrations and GPP during the growing season in 2017 are validated using site-level measurements from MEE and benchmark GPP from GLASS (figure 1). We choose 2017 here as the regional MEIC inventory used in GEOS-Chem is only updated to 2017 (http://meicmodel.org/). Observed O₃ shows high values in the north and east of YRD (figure 1(a)), following the large regional emissions (figure S2). Compared to observations, the GEOS-Chem model well captures the spatiotemporal variation of surface O₃, with a high correlation coefficient of 0.6 (p < 0.01) and a low normalized mean bias of 4% for 28725 statistical samples (figures 1(b) and (c)). Similarly, the YIBs model reasonably depicts the spatial distribution of GPP with an increasing gradient from the north to the south of YRD (figures 1(d) and (e)). The correlation coefficient is 0.7 (p < 0.01) and the normalized mean bias is only 10% for 467 statistical samples between observations and simulations (figure 1(f)). The limited biases in both the simulated O₃ and GPP consolidate our strategies in quantifying the changes of O₃-induced carbon losses under different emission controls using the GEOS-Chem and YIBs models.

Zoom In Zoom Out Reset image size

Ecosystem productivity in YRD is threatened by the severe O₃ pollution (figure 2). In 2017, surface O₃ reduces regional GPP by 184.1 ± 44.5 Gg[C] d⁻¹ during the growing season in YRD (figure 2(a)), accounting for a fraction of 10% to the total GPP. Regional maximum reductions are distributed in Zhejiang province and southern Anhui province, although the high surface O₃ concentrations are predicted in the northern Anhui province and southern Jiangsu province. Such discrepancy is likely because that northern Anhui province and southern Jiangsu province is mainly covered by crops which have a higher O₃ damage critical threshold than other ecosystem types (Unger et al 2020). In the growing season, O₃-induced GPP reductions reach maximums in summer (June–August) due to large GPP and high surface O₃ concentrations (figure 2(c)). With sensitivity simulation SHUT, we quantify the effects of anthropogenic emissions on ecosystem productivity in YRD. It is found that shutting down all anthropogenic emissions in China decreases regional mean O₃ by 29.2 ppbv during the growing season in YRD (figure S3(a)). Consequently, O₃-induced GPP loss is alleviated by 144.9 ± 31.5 Gg[C] d⁻¹ in YRD (figure 2(b)). The largest benefit of GPP recovery occurs in July in response to the largest O₃ reduction by shutting down all anthropogenic emissions (figures 2(c) and S3(b)). Although anthropogenic emissions contribute only 45% to the total surface O₃, shutting down anthropogenic emissions helps mitigate 79% of total O₃-induced ΔGPP during the growing season in YRD. Therefore, exploring the effects of anthropogenic emission controls on O₃-induced ΔGPP is of great value for ecosystem health in YRD.

Zoom In Zoom Out Reset image size

Simulated surface O₃ in YRD is influenced by different anthropogenic emission control strategies (figure 3). For 50% anthropogenic emission reductions in industrial and transportation sectors, regional mean surface O₃ in YRD decreases by 0.6 and 0.4 ppb during the growing season, respectively (figures 3(b) and (e)). In contrast, regional mean surface O₃ increases by 1.0 ppb during the growing season with 50% reductions in agriculture emissions (figure 3(a)). For this specific sector, emissions are mainly in the form of ammonia, the reduction of which does not affect O₃ precursors but decrease ammonium and PM_2.5 concentrations (figure S4). Consequently, the reductions of aerosols reduce the uptake of HO₂ and results in more production of surface O₃ (Li et al 2019). As a comparison, reductions of anthropogenic emissions in other sectors such as power and residential have smaller impacts on surface O₃ with regional mean changes of 0.2 and 0.1 ppb, respectively (figures 3(c) and (d)).

Zoom In Zoom Out Reset image size

Compared to emission controls of a single sector, collaborative emission reductions have better effects in the mitigation of O₃ pollution. For 50% emission reductions in all sectors within YRD, regional mean surface O₃ shows a large reduction of −2.0 ppb during the growing season (figure 3(f)), which has better mitigation effects than the sum of independent emission reductions of five sectors (figures 3(a)–(e)). This is because that there are inhomogeneous emissions for O₃ precursors, such as NO_x and VOCs from different sectors. For example, power sector accounts for 20% of total NO_x emissions but only <1% of total VOCs emissions. Thus, anthropogenic emission control from power sector means only NO_x emissions reductions, which will increase surface O₃ in YRD. Regionally, the largest surface O₃ reduction of −7.0 ppb is found in central Zhejiang province. However, an increase of surface O₃ up to 6.0 ppb is predicted in the south of Jiangsu province when regional anthropogenic emissions are reduced by 50% (figure 3(f)). This is because southern Jiangsu is considered as a VOC-limited regime due to low HCHO/NO₂ ratios (Wang and Liao 2020, Li et al 2021, Xu et al 2021). The O₃ concentrations in the region would be elevated with the reduced NO_x emissions alone or the same proportional reduction of VOCs and NO_x emissions (Yang et al 2021). Such phenomenon that O₃ increases with the reduction of VOCs and NO_x emissions in YRD is also observed during the 2019 novel coronavirus pandemic, which provides a natural experiment to assess the effects of precursor changes on surface O₃ (Wang et al 2021, Zhu et al 2021).

Similar to local emission control, the 50% reductions of anthropogenic emissions outside YRD results in a reduction of −1.9 ppb in regional mean O₃ concentrations (figure 3(g)). Such change is most significant in the western YRD with O₃ reductions exceeding −5.0 ppb. The magnitude of O₃ reduction decreases gradually from the west to east, suggesting that regional transport of O₃ and its precursors to YRD is mainly from inland areas. As a comparison, reducing 50% anthropogenic emissions over China results in a considerable reduction of −4.7 ppb in regional mean surface O₃ over YRD (figure 3(h)), with a spatial pattern merging the O₃ changes by 50% reductions of emissions inside (figure 3(f)) and outside (figure 3(g)) YRD.

We further investigate the seasonal variations in O₃ changes by different anthropogenic emission controls during the growing season in YRD (figure 3(i)). The results show that O₃ reductions driven by regional anthropogenic emission controls, including YRD_{out_}RED50%, YRD_in_RED50%, and CHN_RED50% exhibit similar seasonal variations, increasing from April to July and then decreasing. However, there are inconsistent seasonal variations of O₃ changes driven by sectoral anthropogenic emission controls. The O₃ changes reach a maximum in July for 50% anthropogenic emission reductions in power, transportation, and residential sectors, but in June for 50% anthropogenic emission reductions in agriculture and industry.

Emission controls bring benefits to the recovery of ecosystem productivity during the growing season in YRD (figure 4). O₃-induced regional ΔGPP is mitigated by 3.1 ± 0.4 Gg[C] d⁻¹ with 50% emission reductions from industry sector (figure 4(b)) and 2.2 ± 0.2 Gg[C] d⁻¹ with 50% emission reductions from transportation sector (figure 4(e)). However, 50% emission reductions from agriculture, power, and residential sectors result in additional carbon losses of 5.2 ± 0.9, 0.1 ± 0.1, and 1.4 ± 0.3 Gg[C] d⁻¹, respectively (figures 4(a), (c) and (d)) due to enhanced regional surface O₃. The sum of regional ΔGPP is −1.5 ± 0.8 Gg[C] d⁻¹ from the individual emission controls at the five sectors during the growing season in YRD.

Zoom In Zoom Out Reset image size

We further explore the impacts of regional and sectoral collaborative emission controls on ecosystem productivity during the growing season in YRD. With the YRD_in_RED50% simulation, the O₃-induced regional ΔGPP is mitigated by 10.1 ± 1.6 Gg[C] d⁻¹ (figure 4(f)), which is much larger than sum of independent emission reduction in five sectors. Such difference illustrates that collaborative emission reductions in all sectors have larger benefits for the recovery of ecosystem productivity. Although there is smaller surface O₃ reduction from YRD_out_RED50% than YRD_in_RED50% simulations, a larger regional GPP recovery of 18.6 ± 3.5 Gg[C] d⁻¹ is achieved in the former experiment (figure 4(f)). This is because surface O₃ reductions from YRD_out_RED50% simulation is mainly located in southwestern YRD (figure 3(f)), where the largest GPP is predicted (figure 1(e)). The largest benefit of ecosystem productivity is simulated by reducing 50% anthropogenic emissions in China, which mitigates O₃-induced ΔGPP by 33.3 ± 6.1 Gg[C] d⁻¹ in YRD (figure 4(h)).

The seasonal variations in GPP recovery from sectoral and regional anthropogenic emission controls follow those in O₃ changes (figure 4(i)). For regional anthropogenic emission controls, the O₃-induced GPP losses are mitigated by 43.2 ± 7.2, 19.8 ± 3.1, and 72.2 ± 12.1 Gg[C] d⁻¹ in July for YRD_{out_}RED50%, YRD_in_RED50%, and CHN_RED50% simulations, respectively. These numbers account for 39%, 33%, and 36% of total GPP recovery during the growing season in YRD. For sectoral anthropogenic emission controls, the maximum GPP recovery of 6.3 ± 0.8 Gg[C] d⁻¹ occurs in June for IND_RED50% simulation, but 4.7 ± 0.5 Gg[C] d⁻¹ in July for TRA_RED50% simulation. These results indicate that summer months, especially July are the best period for GPP recovery from anthropogenic emission controls.