Simi-parametric Models Explore The Relationship Between Air Pollutants And Meteorological Factors

As ozone pollution and PM2.5 pollution have a great impact on human health and ecological environment,this paper studies the regression relationship between these two kinds of pollutants and meteorological factors in view of air ozone pollution and PM2.5 pollution in some cities.In view of the complex relationship between meteorological factors and pollutants,including linear relationship,nonlinear relationship and interactive relationship.Therefore,this paper uses partial linear model and semi variable coefficient panel data model with fixed effect to model these two kinds of problems,which can better describe the regression relationship between meteorological factors and pollutants.In ozone pollution modeling,meteorological factors include daily maximum temperature,solar radiation and wind speed,in which ozone content is linearly related to daily maximum temperature and wind speed,and is non-linear related to solar radiation.Therefore,this paper uses partial linear model to fit this regression relationship,and combines the nuclear method and self-help method to get the average estimation of ozone content and partial important quantile return Estimate.The results show that the average ozone content of Roosevelt Island is 70.58293 ppb,and its quantiles of 10%,25%,50%,75% and 90% are 42.98 ppb,58.64 ppb,71.84 ppb,85.49 ppb and 96.30 ppb,respectively.In PM2.5 pollution,this paper mainly considers the regression relationship between PM2.5 and five meteorological factors: temperature,cumulative wind speed,hourly precipitation,air pressure and dew point in Beijing,Shanghai,Chengdu,Shenyang and Guangzhou from April 23,2013 to December 31,2015,and considers the individual effects of different cities,as well as the linear and nonlinearrelationship between these factors In this paper,a semi variable coefficient panel data model with fixed effect is proposed to model the relationship between these variables.The numerical results show that the model can reasonably describe the relationship between meteorological factors and PM2.5.To sum up,when using statistical methods to analyze the actual data,we should roughly analyze the relationship between variables(linear,non-linear,interactive,etc.)according to the scatter diagram and primary regression relationship,and then use appropriate model modeling.According to the model,an appropriate estimation method is proposed to get the fitting regression relationship,draw the fitting curve,and compare the advantages and disadvantages with other methods.Finally,the predicted value can be obtained.