Inversion And Spatiotemporal Distribution Of Near-surface Ozonein The Beijing-Tianjin-Tangshan Region

With the development of urbanization and industrialization,the urban population of our country is increasing and the traffic scale is expanding continuously,which results in more and more serious urban air pollution and more prominent urban pollution problems.In recent years,ozone pollution caused by photochemical smog has occurred frequently in eastern cities of China,and ozone has gradually become the primary pollutant affecting urban air quality.In order to explore the characteristics and spatiotemporal variation trend of near-surface ozone pollution in the Beijing-Tianjin-Tangshan region（Beijing,Tianjin,Tangshan）,this study took the Beijing-Tianjin-Tangshan region（Beijing,Tianjin,Tangshan）as the research object from 2016 to 2019,combined with the ozone concentration of ground monitoring stations in the study area and the total ozone column data from satellite remote sensing,BP neural network,extreme learning machine（ELM）and support vector machine（S VM）were used to predict and evaluate near-surface ozone in the Beijing-Tianjin-Tangshan area.Then,based on the prediction results of the model,the annual,monthly,daily and hourly variations of near-surface ozone concentration and the regional distribution of ozone in the Beijing-Tianjin-Tangshan area were analyzed from the perspectives of time and space respectively.Finally,the interaction between meteorological factors and near-surface ozone concentration was systematically analyzed by combining the influencing factors such as temperature,sunshine duration,wind speed,air pressure and relative humidity in meteorological station data.Studies have shown that:（1）In the near-surface ozone prediction model established,BP model constantly adjusts the connection weight through network iteration to minimize the output error,and its accuracy and reliability are better than ELM and SVM.The determination coefficient R² of BP model training and prediction is about 0.9,but the training process of the model is relatively long,and the application effect is relatively poor in winter.（2）From 2016 to 2019,the annual average concentration of ozone in the Beijing-Tianjin-Tangshan region is the highest in Beijing and the fastest increase rate in Tianjin;The monthly variation pattern of near-surface ozone concentration in Beijing,Tianjin and Tangshan is the same,showing typical seasonal variation characteristics and a "bimodal" distribution;The ozone concentration was the highest in summer and the lowest in winter;The variation trend of hourly ozone concentration is similar,generally showing a "unimodal" variation,with the lowest ozone concentration in the morning and the highest at night.（3）The ozone concentration in Beijing was significantly higher than that in Tianjin and Tangshan.During the period from 2016 to 2019,the ozone concentration in the Beijing-Tianjin-Tangshan study area showed an obvious increasing trend,with Tianjin growing the fastest.（4）The effects of temperature,sunshine duration and air pressure on the ozone concentration are most significant.The higher the temperature,sunshine duration and air pressure are,the higher the ozone concentration will be;Higher relative humidity in summer inhibits the production of ozone;There is little correlation between ozone and wind speed;High concentration of PM_2.5 can effectively reduce the content of ozone in the atmosphere;The production of ozone is accompanied by NO₂ depletion The inversion of near-surface ozone concentration based on BP neural network can realize the accurate assessment of surface air quality scientifically and effectively,and provide technical support and scientific basis for the prediction,prevention and treatment of urban air pollution.