Investigating tropospheric and surface ozone sensitivity from present day to future
Tropospheric ozone (O3) is an important reactive gas in the atmosphere influencing human health, ecosystems and climate. Since the mid-20th century, scientists started to explore the mechanism of tropospheric O3 formation after severe O3 air pollution in Los Angeles. They found that O3 is a photochemical pollutant as its formation involves energy from sunlight, as well as precursors nitrogen oxide (NOx), volatile organic compounds (VOCs) and carbon monoxide (CO). Nowadays, highly O3 polluted episodes can still occur in areas where emissions have been controlled strictly due to the non-linear chemical reactions of O3 formation. Therefore, it is important to implement suitable emission control strategies to mitigate O3 pollution, and to understand the impacts of emissions and climate on O3 changes in the future. Firstly, a chemistry scheme with more reactive VOC species is developed based on the Strat-Trop chemistry scheme in the United Kingdom Earth System Model, UKESM1. This permits a more realistic and photochemically active environment for O3 simulation in areas with high reactive VOC emissions. The effectiveness of emission controls in reducing surface O3 concentrations in the industrial regions of China in summer, 2016, is investi gated. The concentrations of surface O3 in those regions generally can be simulated accurately, and the diurnal variation of O3 can also be captured well by the model. O3 production in most regions is VOC-limited, suggesting that surface O3 concentrations will increase as NOx emissions decrease. In the VOC-limited regions, more than 70 % reductions in NOx emissions alone are required to reduce surface O3 concentrations. Reductions in 20 % VOC emissions alone lead to 11 % decreases in surface O3 concentrations, and are effective in offsetting increased O3 levels that would otherwise occur through decreased NOx emissions alone. Subsequently, the evolution of tropospheric O3 from the present day (2004- 2014) to the future (2045-2055) under the shared socio-economic pathways (SSPs) is investigated to demonstrate the impacts of different climate and emissions on O3 changes. In the context of climate change, changes in the tropospheric O3 burden in the future can be largely explained by changes in O3 precursor emissions. However, surface O3 changes vary substantially by season in high-emission regions due to different seasonal O3 sensitivity. VOC-limited areas are more extensive in winter (7 %) than in summer (3 %) across the globe. Reductions in NOx emissions are the key to transform O3 production from a VOC- to NOx-limited chemical environment, but will lead to increased O3 concentrations in high-emission regions, and hence emission controls on VOC and methane (CH4) are also necessary. Lastly, a deep learning model is developed to demonstrate the feasibility of correcting surface O3 biases in UKESM1, to identify key processes causing them, and to correct projections of future surface O3. Temperature and related geographic variables latitude and month show the strongest relationship with O3 biases. This indicates that O3 biases are sensitive to temperature and suggests weakness in representation of temperature-sensitive physical or chemical processes. Photolysis rates are also shown to be important for O3 biases likely due to uncertainties in cloud cover and insolation simulations. Chemical species such as the hydroxyl radical, nitric acid and peroxyacyl nitrates show a clear relationship to O3 biases, associated with uncertainties in emissions, chemical production and destruction, and deposition. Corrected seasonal O3 changes are generally smaller than those simulated with UKESM1 in high-emission regions. This demonstrates that O3 sensitivity to future emissions and climate in UKESM1 may be stronger than that in the real atmosphere. Given the uncertainty in simulating future ozone, we show that deep learning approaches can provide improved assessment of the impacts of climate and emission changes on future air quality, along with valuable information to guide future model development. The work presented here offers a valuable assessment of emission control strategies to resolve current O3 air pollution problems in China, and also quantifies the changes in the tropospheric O3 burden and global surface O3 sensitivity in the future under different emission and climate scenarios. Deep learning guides possible directions to improve model performance in surface O3 simulations for a global chemistry-climate model, and provides more accurate projections of O3 pollution in the future.