Xing et al. (2011) developed the first response surface model (RSM) for China's nonlinear ozone response to nitrogen oxide (NOx) and volatile organic compound (VOC) emission changes, built on Community Multi-scale Air Quality (CMAQ) simulations of a single, meteorologically atypical month (July 2005), and derived quantitative emission-reduction portfolios for Beijing, Shanghai, and Guangzhou. The authors explicitly flagged that their ozone sensitivities could suffer from emission-inventory errors — with VOC inventories carrying far wider uncertainty than NOx — and that their findings were conditioned on the atypical meteorology of the single simulated case. Successor methods developed since (ERSM, ERSMv2.0, pf-RSM, Epf-RSM, machine-learning RSM, 2015–2022) resolved the original design/dimensionality limitation and re-applied response surface modeling to multiple seasons, years, and regions, but none quantified how inventory errors propagate into the predicted response surfaces or into the control portfolios derived from them. The open problem is therefore to quantify the sensitivity and uncertainty of emulator-based ozone-response predictions and their derived emission-reduction portfolios with respect to anthropogenic emission-inventory uncertainty — especially speciated VOC emissions — and to establish the robustness of such control requirements across meteorological conditions rather than a single case.

Background

Surface ozone (O3) is a secondary air pollutant produced photochemically from nitrogen oxides (NOx, the sum of NO and NO2) and volatile organic compounds (VOCs, reactive carbon-containing gases from solvents, vehicles, industry, and vegetation). Its response to precursor emission changes is strongly nonlinear: in so-called VOC-limited (NOx-saturated) regimes, common in polluted urban cores, reducing NOx alone can increase local ozone because less NO is available to titrate ozone away, whereas in NOx-limited regimes NOx reductions lower ozone. Because ozone attainment strategies commit billions of dollars to emission controls, regulators need to know not just the direction but the magnitude of the ozone response to any proposed combination of NOx and VOC reductions.

The tool of record for this question is a three-dimensional chemical transport model (CTM) such as the Community Multi-scale Air Quality (CMAQ) model, which simulates emissions, chemistry, and transport on a gridded domain. Brute-force evaluation of many candidate control strategies is computationally prohibitive, so Xing, Wang, Jang, Zhu, and Hao (2011) trained a response surface model (RSM) — a statistical meta-model (emulator) interpolating CTM outputs across perturbed emission inputs — on an ensemble of CMAQ runs for July 2005 (12 km grid over highly populated eastern China, nested within a 36 km national domain; six control variables: four NOx and two VOC source-category groups). From the resulting ozone isopleths (curves of constant ozone as a function of NOx and VOC emission levels), they quantified the VOC-limited character of the Beijing, Shanghai, and Guangzhou urban cores, the downwind benefits of NOx control, and concrete reduction portfolios attaining 80 ppb, for example roughly 80% power-plant NOx reduction combined with about 60% NOx and 60% VOC reduction from other categories.

The paper's own conclusions, however, state that these findings 'are limited to the modeling cases in this study due to the atypical meteorological conditions as well as the uncertainties from simulations and predictions', and that 'the ozone sensitivities may still suffer from uncertainties in the emission inventory. Therefore, it is important for future work to better understand the precursor emission inventory, especially for VOC emissions.' The inventory used carried reported 95% confidence intervals of about −10% to +36% for national NOx and −44% to +109% for VOC emissions, and VOC uncertainty is compounded by speciation — the allocation of total VOC mass to individual chemical species — which determines reactivity and hence ozone-forming potential. A speciated-VOC inventory error does not merely shift the baseline; because the response surface is curved and regime-dependent, it can move the location of the NOx-saturated/VOC-limited transition and thereby change the sign of the predicted effect of a NOx reduction, so the error could plausibly reorder or invalidate derived control portfolios. The paper also acknowledged experimental-design constraints (the number of control variables should not exceed about eight before prediction errors grow) and unrealized links to activity growth and control costs; those two methodological and integration aspects were substantially taken up by successor work after 2014.

Problem Statement

Quantify how uncertainty in China's anthropogenic emission inventories — in particular the magnitude and speciation of volatile organic compound (VOC) emissions, whose reported 95% confidence intervals are far wider than those of NOx — propagates into response-surface-model (emulator) predictions of the nonlinear ozone response to combined NOx and VOC emission reductions, and determine how robust the emission-reduction portfolios derived from those response surfaces (the control ratios required to attain a stated ozone target) are to such inventory errors, including how their validity depends on meteorological conditions rather than a single simulated month. This quantification comprises (i) characterizing the input inventory uncertainty at species and source-category resolution with stated provenance, (ii) propagating that uncertainty through the chemical-transport-model ensemble and emulator pipeline, (iii) reporting the resulting uncertainty in predicted ozone responses and in the derived control portfolios, identifying where inventory error is large enough to change the sign of the predicted response or to reorder portfolio choices, and (iv) assessing how these conclusions change across seasons, years, or meteorological regimes. The single overarching objective is a defensible error bar — or a demonstrated insensitivity — on emulator-based ozone-control design in China; the components above serve that objective jointly and are not independent questions.

References

  1. Xing, J., Wang, S. X., Jang, C., Zhu, Y., Hao, J. M.: Nonlinear response of ozone to precursor emission changes in China: a modeling study using response surface methodology, Atmospheric Chemistry and Physics, 11, 5027–5044, 2011. https://doi.org/10.5194/acp-11-5027-2011 (source paper)
  2. Zhao, B., Wang, S. X., Xing, J., Fu, K., Fu, J. S., Jang, C., Zhu, Y., Dong, X. Y., Gao, Y., Wu, W. J., Wang, J. D., Hao, J. M.: Assessing the nonlinear response of fine particles to precursor emissions: development and application of an extended response surface modeling technique v1.0, Geoscientific Model Development, 8, 115–128, 2015. https://doi.org/10.5194/gmd-8-115-2015
  3. Xing, J., Wang, S., Zhao, B., Wu, W., Ding, D., Jang, C., Zhu, Y., Chang, X., Wang, J., Zhang, F., Hao, J.: Quantifying nonlinear multiregional contributions to ozone and fine particles using an updated response surface modeling technique, Environmental Science & Technology, 51, 11788–11798, 2017. https://doi.org/10.1021/acs.est.7b01975
  4. Xing, J., Ding, D., Wang, S., Dong, Z., Kelly, J. T., Jang, C., Zhu, Y., Hao, J.: Development and application of observable response indicators for design of an effective ozone and fine-particle pollution control strategy in China, Atmospheric Chemistry and Physics, 19, 13627–13646, 2019. https://doi.org/10.5194/acp-19-13627-2019
  5. Jin, J., Zhu, Y., Jang, J., Wang, S., Xing, J., Chiang, P., Fan, S., Long, S.: Enhancement of the polynomial functions response surface model for real-time analyzing ozone sensitivity, Frontiers of Environmental Science & Engineering, 15, 31, 2020. https://doi.org/10.1007/s11783-020-1323-0
  6. Li, J., Dai, Y., Zhu, Y., Tang, X., Wang, S., Xing, J., Zhao, B., Fan, S., Long, S., Fang, T.: Improvements of response surface modeling with self-adaptive machine learning method for PM2.5 and O3 predictions, Journal of Environmental Management, 303, 114210, 2022. https://doi.org/10.1016/j.jenvman.2021.114210
  7. Wang, F., An, J., Li, Y., Tang, Y., Lin, J., Qu, Y., Chen, Y., Zhang, B., Zhai, J.: Impacts of uncertainty in AVOC emissions on the summer ROx budget and ozone production rate in the three most rapidly-developing economic growth regions of China, Advances in Atmospheric Sciences, 31, 1331–1342, 2014. https://doi.org/10.1007/s00376-014-3251-z
  8. Zhao, Y., Mao, P., Zhou, Y., Yang, Y., Zhang, J., Wang, S., Dong, Y., Xie, F., Yu, Y., Li, W.: Improved provincial emission inventory and speciation profiles of anthropogenic non-methane volatile organic compounds: a case study for Jiangsu, China, Atmospheric Chemistry and Physics, 17, 7733–7756, 2017. https://doi.org/10.5194/acp-17-7733-2017
  9. Shen, H., Sun, Z., Chen, Y., Russell, A. G., Hu, Y., Odman, M. T., Qian, Y., Archibald, A. T., Tao, S.: Novel method for ozone isopleth construction and diagnosis for the ozone control strategy of Chinese cities, Environmental Science & Technology, 55, 15625–15636, 2021. https://doi.org/10.1021/acs.est.1c01567
  10. Li, K., Jacob, D. J., Liao, H., Shen, L., Zhang, Q., Bates, K. H.: Anthropogenic drivers of 2013–2017 trends in summer surface ozone in China, Proceedings of the National Academy of Sciences, 116, 422–427, 2019. https://doi.org/10.1073/pnas.1812168116
  11. Li, K., Jacob, D. J., Shen, L., Lu, X., De Smedt, I., Liao, H.: Increases in surface ozone pollution in China from 2013 to 2019: anthropogenic and meteorological influences, Atmospheric Chemistry and Physics, 20, 11423–11433, 2020. https://doi.org/10.5194/acp-20-11423-2020
  12. Xu, D., Yuan, Z., Wang, M., Zhao, K., Liu, X., Duan, Y., Fu, Q., Wang, Q., Jing, S., Wang, H., Zhao, X.: Multi-factor reconciliation of discrepancies in ozone-precursor sensitivity retrieved from observation- and emission-based models, Environment International, 158, 106952, 2022. https://doi.org/10.1016/j.envint.2021.106952