Short-Term Ozone Forecasts and Modeling of Long-Term Climate Change Impacts on Ozone Pollution in the Marmara Region

Abstract

Air pollution monitoring and data analysis have been the main components of an air quality management system. Thanks to advances in atmospheric and computer sciences over recent decades, air quality simulation models have emerged as powerful new tools for air quality management. The sophisticated structure of these models not only improves our understanding of the complex nature of the atmosphere but also enables us to make air quality forecasts for the near and far future. These capabilities cover a major gap in air quality management and turn the models into an essential part of the policy-making process. In this study, the deterministic atmospheric models and deep learning algorithms were employed to simulate the air quality of the Marmara region for the mid-21st century and near-future, respectively. The study consists of three parts: (1) investigating the effect of climate change on the summertime ozone concentration in the Marmara region of Turkey; (2) simulating the effect of climate change on the concentration of biogenic emissions; and (3) improving the performance of deep learning models by imposing the temporal characteristics of the daily ozone cycle on the model. The difference between past (2012) and future (2053) ozone concentrations was used to show how climate change impacts ozone concentration. The past and future (under the SSP2-4.5 and SSP5-8.5 scenarios) ozone forecasting were conducted using the WRF-CMAQ modeling system. The global bias-corrected CMIP6 data were used to give the meteorological initial and boundary conditions, and the anthropogenic and biogenic emissions were provided by the EMEP inventory and the MEGAN model, respectively. The CMIP6 data were downscaled using three nested domains with a spatial resolution of 36 km, 12 km, and 4km. Climate and air quality simulations' results show a significant (P< 0.05) increase in daily mean temperature and daily mean ozone concentration under future climate scenarios. The average rates of increase in ozone concentration in the Marmara domain were 13.6% and 16.02%, under the SSP2-4.5 and SSP5-8.5 scenarios, respectively. To answer the second question, i.e. how climate change impacts the biogenic emissions concentration, the biogenic emission simulations were performed by the MEGAN model using climate inputs from the past period and future scenarios. The results show that future climate scenarios cause a significant increase in biogenic emission concentration. This increase is about 28.2% and 38.46% for the average isoprene according to the SSP2-4.5 and SSP5-8.5 scenarios, respectively. Moreover, the rate of increase in the average terpenes concentration is 15.38% and 21.79% under the SSP2-4.5 and SSP5-8.5 scenarios, respectively. The final section of the thesis is dedicated to the improvement of deep learning models' performance in the prediction of hourly ozone concentration by imposing temporal characteristics of the diurnal ozone cycle on models. The results show that the proposed method significantly increased the performance of deep models. According to the best of our knowledge, the proposed approach has not been addressed in the literature. This is also the first study of the impact of climate change on tropospheric ozone and biogenic emission concentrations in the Marmara region. The results provide valuable details on how the meteorological parameters and emissions interact to form tropospheric ozone, depending on regional characteristics.