| With the development of industry and urbanization and the increase the emission of air pollutants,air quality is decreasing.The hazy which occurs frequently in recent years influences people's travel,lives and health.As a main component of hazy,PM2.5(the particular matter with aerodynamic diameter less than or equal to 2.5 ?m)arises the wide concern of public and government.Many environmental epidemiologic studies found that toxic substances attached on PM2.5 will come to lungs' alveoli with breath and cause asthma,respiratory disease,cardiovascular disease and lung cancer.With the closer cooperation with Pearl River Delta(PRD)and the increasing population and traffic,the air pollution problem becomes more obvious with the hazy days increase.Therefore,the spatial and temporal analysis and monitoring of PM2.5 concentrations are necessary for atmosphere protection.This paper takes Hong Kong as a case study.Based on the analysis of PM2.5 spatial and temporal variations and influencing factors,a PM 2.5 concentrations estimation model is built up using MODerate resolution Imaging Spectroradiometer(MODIS)data retrieved AOD products with higher resolution and accuracy.In order to promote the performance of model,the meteorological data(temperature,pressure,relative humidity,wind speed,and wind direction)and land use data are incorporated into model.The main outcomes and creative points are concluded as followings:(1)The PM2.5 concentrations spatial and temporal variations of Hong Kong in recent five years are analyzed to fill the research gap about PM2.5 variations after 2011.The researches about PM2.5 spatial and temporal variations of Hong Kong mainly concentrates before 2011.To fill the gap,this research used the hourly PM 2.5 measurements of 14 air quality monitoring station in Hong Kong during 2011-2015 and spatial interpolation technique to analyze the spatial and temporal variability characteristics of PM2.5.The analysis mainly includes three aspects: annual,seasonal and diurnal variability.According to the analysis results,the annual average value of PM2.5 is highest in 2011 and has a big decrease at 2012,then begins to increase in 2013 and keep decrease in 2014 and 2015.PM2.5 concentrtions are influenced by Asia monsoon circulation.PM2.5 content is highest in winter(36-50 ?g/m3)and lowest(12-30 ?g/m3)in summer.In terms of diurnal variation,PM2.5 content increase and reach the peak(about 28-54 ?g/m3)during rush hours and derease after rush hour.PM2.5 spatial distributions are related with land use,population,and traffic density.The urban area with high population and traffic has high PM2.5 concentrations while the area with high vegetation cover and low population has low PM2.5 concentrations.(2)The impact of meteorological factor on PM2.5 concentrations are analyzed and the reasons are explored.The meteorological data in 2013,including temperature,pressure,relative humidity(RH),wind speed,wind direction and rainfall,were collected from Hong Kong observatory for the study on influence of meteorological factors on variation of PM2.5.According to the correlation analysis results and monthly tendency,the PM2.5 concentrations have a positive relationship with pressure with correlation coefficients 0.507 while have negative relationships with temperature,RH,rainfall,wind speed with correlation coefficients-0.512,-0.237,-0.524,-0.284,respectively.The correlation coefficients are different in different month and seasons.The north wind in winter can increase the PM2.5 by bring pollutant from pollutant from mainland China while the east south wind in summer decreases the PM2.5 by bring clear air from south ocean to Hong Kong.Therefore,the meteorological factors influence PM2.5 concentration by influencing its aggregation,diffusion and propagation.This analysis result will provide reference for the variables selection in model establishment.(3)The Simplified Aerosol Retrieval Algorithm is revised and used for retrieving MODIS AOD data by development in resolution,accuracy and correlation with PM2.5.The procedure of retrieving MODIS operational AOD products at 10 km and 3 km resolution are compared.However,AOD product at rough resolution might introduce some error and cannot satisfy air pollution application in Hong Kong.Therefore,a AOD product with 500 m spatial resolution was produced using enhanced Simplified Aerosol Retrieval Algorithm(SARA).The two parameters of single scattering albedo and asymmetric factor required in SARA are acquired from AERONET AOD measurements.Therefore,SARA cannot be applied in those areas without AERONET stations.To make up this limitation,single scattering albedo and asymmetric factor are calculated from Optical Properties of Aerosols and Clouds(OPAC)software package.The enhanced SARA was used to retrieve AOD products with 500 m resolution in Hong Kong from 2012 to 2014.The retrieved AOD data were validated with AERONET AOD,which was found that the R2 is 0.938 and the RMSE is 0.072.Compared with 3 km AOD(R2=0.854,RMSE = 0.118)and 10 km AOD(R2=0.827,RMSE = 0.126),the retrieval accuracy is obviously improved.The correlation analysis with PM2.5 concentrations shows that the improved SARA retrieved AOD has higher R2 of 0.659 than 3 km AOD(0.515)and 10 km AOD(0.205).(4)Based on the correlations between AOD and PM2.5 concentrations,a PM2.5 spatial and temporal variations estimation model using geographically and temporally weighted regression is proposed.Most recent PM2.5 estimation models are global regression models of which coefficients are constants in space and time.The prediction accuracy usually is not high in most cases.In order to capture the spatial and temporal heterogeneity of PM2.5,a geographically and temporally weighted regression(GTWR)model based on AOD data was developed.The meteorolocal factors(temperature,pressure,RH,wind speed,wind direction and rainfall)and land use factors(NDVI and the road length in 500 m buffer)were added into the model to improve prediction accuracy.The PM2.5 concentration estimation from GTWR model were validated with ground PM2.5 measurements and compared with the estimation from geographically weighted regression(GWR),temporally weighted regression(TWR)and ordinary least squares(OLS).Compared with OLS,GWR,and TWR models,the R2 value of GTWR was improved from 0.598,0.655,0.778 to 0.792.The RMSE of GTWR was improved from 15.885,14.723,11.817 ?g/m3 to 11.463 ?g/m3.Therefore,the GTWR model may provide a more accurate way to predict PM2.5 spatio-temporal variation.In addition,the GTWR model were applied for retrieval PM2.5 concentration distribution in 2013.The results show the retrieved PM2.5 is generally consistant with ground measurments and shows more detailed spatial distribution of PM2.5.The research on the analysis and prediction of PM2.5 concentration using remote sensing AOD data can compensate the limitation of ground-based monitoring and provide more detailed information of PM2.5 spatial and temporal variations.This research can give data support for afterwards management of particulate matter pollution and relative relative epidemiologic research and provide reference for governments to make contingency plan for air pollution and urban development strategies.This is significant for atmosphere environment protection and sustainable development. |
PDF Full Text Request