Study On The Cause Of The Spatial And Temporal Characteristics And Simulation Of Air Quality In North China

Air pollution is one of the major environmental problems in the world.With the frequent occurrence of "haze" in North China in recent years,the air quality has become a hot issue of concern and research of government and society.Therefore,it is of great practical significance to study the spatial and temporal distribution and simulation of air quality in North China.This paper mainly discusses the simulation of air quality surface,the spatial and temporal distribution of air quality,the influence factors of air quality and the prediction of air quality.Firstly,based on the monitoring data of air quantity index(AQI)in 71 major cities of North China from 2014 to 2016,this paper assesses the deterministic interpolation method and geostatistical method by cross-validation and use Kriging method to simulate the surface of air quality according to the theory of spatial variation.Secondly,a method which combining the point with surface,the spatial interpolation data and MODIS aerosol data,were used to analyze the spatial and temporal characteristics of air quality and discuss the time scale change by used Morlet wavelet.Then,the influence of precipitation,relative humidity,wind direction and wind speed of meteorological elements on AQI was emphatically discussed,and the spatial correlation analysis of AQI and socio-economic factors was carried out by ESDA-GWR model.Finally,taking Beijing as an example,considering the influence of meteorological factors and pollution sources,combined with BP neural network and genetic algorithm,an air quality prediction model based on GA-BP was designed which used for dynamic prediction of AQI.The main results of this paper are as follows:(1)Due to geostatistical method considers the spatial autocorrelation of AQI data,simulation effect is better than the deterministic interpolation method.The simulation results of kriging model considering spatial anisotropy are better than whole directional stable semivariable function model.The semi-variable functions of kriging have different application effects due to different time periods and the cross-validation parameters of gaussian function are better in the whole which have the better simulation results.(2)In terms of space characteristics,air pollution in North China is the highest and much higher than elsewhere.The air pollution is concentrated in Beijing,Tianjin,south of Hebei,north of Henan and west of Shandong and the air quality is relatively good in western shanxi,northern hebei and shandong peninsula.The aerosol data of MODIS and AQI converge in spatial distribution,and the differences are mainly found in plateau mountain and gulf coast.In terms of time characteristics,the AQI in winter is significantly higher than that in other seasons.The AQI in spring is dominated by high values,AQI in autumn and winter is dominated by low values and AQI in summer presents a balanced distribution feature.AOD and AQI are relatively consistent in the spatial variation of the season,but there is a large difference in the intensity becasuse the high value of AOD appears in summer which contrary to AQI.According to Morlet wavelet analysis,air quality in North China has a different time-scale structure and the main period is about 280 days.(3)Meteorological factors such as precipitation,relative humidity,wind direction and wind speed and AQI is mainly inverse correlation.When daily precipitation reaches more than 10 mm,the scouring effect of pollutants is obvious.At the interannual scale,the wind speed was significantly negatively correlated with AQI and the pollutants involved in AQI evaluation.At the same time,the wind direction has an influence on the spatial distribution of air quality,the pollutants are transmitted dynamically under the influence of prevailing wind direction,which can cause the space movement of air pollution.According to the analysis of the ESDA-GWR,car ownership,population aggolomeration and industrial production of social and economic factors are positively correlated with air pollution,while improve the GDP and forest coverage rate is conducive to curb pollution.(4)The BP neural network was constructed by using daily meteorological factors,pollutant factors and AQI,the fitting coefficient of neural network after training was 0.75.The genetic algorithm was used to optimize the BP neural network,and the fitting coefficient of GA-BP neural network reached 0.85,and the forecast accuracy increased from 60.01% to 75.87%,which reached the requirements of AQI prediction.