Study On The Boundary Layer Ozone Concentration Forecast And Meteorological Missing Data Imputation

With industrial and urban development, urban photochemical smog has become a serious environmental pollution problem in recent decades. Photochemical pollution is a new type of pollutant in the fossil fuels times. The main component of photochemical smog is ozone, which not only cause irreparable harm to the human body, and will damage the plant, ecosystems, buildings and rubber products.As more attention is attracted to the environment problem and the development of scientific research, great progress has been made on the study of atmospheric boundary layer ozone. Multivariate statistical analysis has been widely used in the ozone concentration forecast. Due to the non-Gaussian feature of ozone concentration distribution, generalized linear model (GLM) is used to forecast the boundary layer ozone concentration in this thesis.Missing data is a ubiquitous problem in the boundary layer ozone concentration forecasting. There will be a negative impact on the forecast result if the missing data was not be imputed by the appropriate way. In this thesis, the characteristics of meteorological missing data are analyzed, and the mass transfer of air pollutants in the atmosphere is incorporated to modify the missing data imputation methods.The main works and results include:(1) The sources of ozone and the main impact factors on the ozone concentration are described. The boundary layer ozone concentrations study history is briefly introduced. In addition, the characteristics of meteorological missing data are discussed in detail, and a new method (Wind Information modified Inverse Distance Weighting Method, WIIDWM) is proposed in this thesis based on the inverse distance weighting method. This method has a good performance when the high correlation variables are missing at the same time.(2) Because of the non-Gaussian feature of ozone concentration distribution, ordinary multiple linear regression (LM) has poor performance on the extreme points. According to this characteristic, the generalized linear model is applied to forecast boundary layer ozone concentration. The ozone concentration forecast result is better than that from the LM method.