Spatial And Temporal Distribution Of PM_2.5 In Shenzhen And Its Relationship With Land Use

PM_2.5 is the primary pollutant in China’s urban atmosphere.Its physicochemical properties have important effects on air quality,visibility,Earth’s radiation balance,and human health.Shenzhen is located on the east coast of the Pearl River Delta in China.With the advancement of urbanization and industrialization,the environmental pressure in the Pearl River Delta region has increased,and atmospheric pollution events have frequently occurred.The related research on PM_2.5 has become a hot spot.The research on the spatial and temporal distribution of PM_2.5 is limited by the monitoring data conditions.the analysis time is short and the regional representation is limited.The research on the relationship between PM_2.5 and land use has less attention to the spatial structure,the spatial scale is too coarse and the land use category is too comprehensively,it is difficult to fully reflect the environmental effects of urban land use.This paper uses Shenzhen as the research area,and uses the continuous five-year PM_2.5 surface monitoring data to analyze the characteristics of spatio-temporal changes of PM_2.5 from multiple perspectives.Using domestic satellites and other multi-source data,combined with remote sensing classification and remote sensing inversion techniques,to analyze the relationship between the spatial distribution of PM_2.5 pollution and the spatial distribution of land use,and explore the impact of land use resulting from varying degrees of human activities on the PM_2.5pollution caused.Providing references for airpollution control and urban land planning.and lay the foundation for the comprehensive application of domestic data.The main work includes:（1）Using HJ-CCD,SPOT5,DEM,and other multi-source data for urban land use classification.Through multi-level segmentation,feature extraction and feature optimization,combined with prior knowledge,an object-oriented decision tree classification rule set was constructed.The urban typical landscape residential land,industrial land,traffic land,forest land,plantation,lawn,reservoir/pit pond,river and other types are distinguished,the classification accuracy of the rule set reaches89.6%.Then,by manually interpreting the mixed category in the decision tree classification results,the land use classification results of Shenzhen are obtained.（2）According to the hourly monitoring data of 45 national air quality monitoring points PM_2.5in Shenzhen and its surrounding areas from 2013 to 2017,the variation pattern and pollution status of PM_2.5 in Shenzhen were analyzed from two aspects of time and space.The time perspective is measured by years,quarters,months,and days.The spatial angle is based on the analysis of Shenzhen’s regional characteristics and the differences between Shenzhen and surrounding areas.（3）Using HJ-CCD data,urban aerosol optical thickness was inferred based on the 6S model,improved dark-pixel method,and using the PM_2.5 monitoring data to construct a fitting model of AOD and PM_2.5.The one-dimensional quadratic function model was the optimal model.The fitting correlation reached 0.72,the average relative error was less than 20%,and the inversion accuracy was better.The feasibility of using AOD to calculate the quantitative inversion of PM_2.5 in the coastal area of South China is proved based on the domestic environmental data,and the spatial distribution data of Shenzhen PM_2.5 is obtained.（4）Based on the inversion results of PM_2.5 and its corresponding land use data,the spatial relationship and quantitative relationship between land use and PM_2.5concentration were performed through statistical analysis of spatial data,spatial autocorrelation analysis,and geo-detectors.The study also analyzed the contribution of Shenzhen urben landscapes pattern to the spatial differentiation of PM_2.5.Based on the above research work,the following conclusions are drawn:The air quality of Shenzhen is good overall.The annual average values of PM_2.5 concentration during 2013-2017 were 37.75?g/m3,33.36?g/m3,29.76?g/m3,27.07?g/m3,and 28.71?g/m3.Respectively,showing a decreasing trend overall.However,due to the high temperature and low precipitation in 2017,the PM_2.5 mass concentration have rebounded.From the daily,monthly,and seasonal changes of PM_2.5 concentration in Shenzhen,the―Bimodal characteristic‖of PM_2.5 changes in 2013 and 2014 were significant.The intensity of the"first peak"in the double-peak phenomenon caused by traffic morning and evening peaks has been gradually weakened since 2015,and the difference between peak and valley values has decreased.The monthly variation of PM_2.5concentration in Shenzhen shows a―U‖distribution,with peaks appearing mostly in December and January,and valleys mostly appearing from June to August.The seasonal characteristics of PM_2.5concentration affected by climate were significant,showing in winter>autumn>spring>summer.The spatial distribution difference of PM_2.5 concentration is very significant in Shenzhen and its surrounding areas.Whether in the inter-regional perspective or in the Shenzhen region,the spatial distribution pattern is characterized by high elevations from west to east and high inland and low coastal height.Regardless of the statistical units at any time,the characteristics of PM_2.5 in Shenzhen are consistent with the surrounding cities,and the overall air quality level is in a moderate gradient,second only to Hong Kong and Huizhou.The monthly and quarterly spatial changes fully reflect the monsoon climate characteristics.It’s dry and rainy in winter.The northern monsoon transmits pollutants from Guangzhou and Dongguan to the south through the air.The pollution is aggravated,and in the summer,it is dominated by the southern monsoon.Warm and humid airflow brings precipitation,and the overall PM_2.5 mass concentration is generally reduced.The spatial distribution of PM_2.5 concentration in Shenzhen has an agglomeration feature,and spatial agglomeration corresponds to the spatial differentiation of land use.The distribution of―high-level‖correlation,―high-low‖correlation,and distribution location of construction land and agro-forestry land match.The bivariate spatial autocorrelation index between PM_2.5 concentration and NDVI is-0.43,which has a negative spatial correlation.The degree of vegetation coverage has a greater influence on the spatial agglomeration of PM_2.5,and the effect of vegetation on the reduction of PM_2.5.To achieve a certain area has a significant effect,strengthening the construction and protection of vegetation accumulation areas is of great significance for the control of PM_2.5.The impact of human activities on PM_2.5 concentrations was very significant.Through analysis of the differences in PM_2.5 concentrations over land use types in Shenzhen,it was found that the order of average PM_2.5 concentrations from high to low was industrial land>residential land>traffic land>rivers>other land>reservoirs/pits pond>plantation>lawn>forest land,from high-concentration PM_2.5 areas to low-concentration areas,the proportion of agricultural and forestry land use has gradually increased,and the proportion of construction land has gradually decreased.agricultural and forestry land,especially forest lands The absorption of PM_2.5 is most prominent.Construction land,especially industrial land,has a significant impact on PM_2.5 pollution.Pollution prevention and control can be started from cleaner production.The spatial structure of urban landscape is closely related to the spatial differentiation of PM_2.5concentration.At Class level,the proportion of residential land,industrial land,traffic land,and other lands use areas is positively correlated with the concentration of PM_2.5.The proportion of forest land,plantation,lawn area is negatively correlated with the concentration of PM_2.5.In the category of waters,the proportion of reservoirs/pits pond is negatively correlated with the concentration of PM_2.5,and the correlation between the concentration of rivers and PM_2.5 does not pass the significance test and the relationship is not clear.In the Landscape level,CONTAG and LPI are negatively correlated with PM_2.5 concentration,and SHDI is positively correlated with PM_2.5concentration,indicating that the higher degree of fragmentation of the urban landscape is,the more prominent the pollution is.the correlation between PAFRAC and the PM_2.5 concentration has not passed the significance test,and the shape index has no significant effect on the PM_2.5 concentration.The contribution of different urban landscape features to the spatial dissimilarity of PM_2.5differs greatly.The proportion of forestland area in the type level contributes most to the spatial disparity of PM_2.5 concentration,and vegetation characteristic is the main reason for the spatial differentiation of PM_2.5 concentrations in Shenzhen.The contribution of CONTAG in the landscape level is the largest,indicating that the spatial distribution of PM_2.5 concentration is greatly affected by the aggregation state of the landscape.The composite contribution of type level and landscape level（P:0.007 to 0.416）was higher than that of any two independent factors（P:0.001 to 0.366）,reflecting the complexity of the spatial variation of PM_2.5 concentration.The combined effect of the area ratio of residential land and forest land types has the largest contribution to the spatial distribution of PM_2.5 concentration in Shenzhen,which is basically consistent with the spatial distribution response of PM_2.5 concentration,which is consistent with perceptual knowledge.