Spatial-temporal Distribution Characteristics Of PM_2.5 Pollution Exposure Based On Multi-source Satellite Remote Sensing Of Population In The Yangtze River Delta

To assess the health risk of PM_2.5,it is necessary to accurately estimate the actual exposure level of people in PM_2.5.However,the sparse spatial distribution of PM_2.5 and the dense spatial distribution of population are often inconsistent.Therefore,it is necessary to develop a high spatial resolution and refined assessment of the exposure level of polluted population.The Yangtze River Delta region is located in the eastern part of China.It has developed industry,prosperous economy and dense population.The large number of source emissions combine with adverse meteorological conditions lead to serious pollution in the region.At present,PM_2.5pollution exposure research in the Yangtze River Delta region is mainly focused on the total population,public health data sampling points in administrative divisions or other similar population data combined with PM_2.5 observation data at stations.The spatial distribution characteristics and seasonal differences of PM_2.5 population exposure risk are still unknown.So in this study,we construct statistical model and random forest model to estimate the spatial distribution of population in the Yangtze River Delta.By comparing the accuracy of the two models,the high-precision random forest model will be selected to simulate the population of the Yangtze River Delta in 2010-2016.The estimated population spatial distribution in the Yangtze River Delta was combined with the PM_2.5 data retrieved from MODIS satellite remote sensing.On this basis,we assess the real exposure level of population to PM_2.5 and the exposure risk to PM_2.5 pollution in the Yangtze River Delta from three aspects:PM_2.5 concentration,the population exposure intensity,and the population weighted concentration.Furthermore,the seasonal and temporal spatial changes of PM_2.5 air pollution exposure level in high-resolution population are analyzed.Taking lung cancer as an example,the risk of PM_2.5 death toll increase under the current exposure is analyzed.The results show that:1)By comparing the street census population with the predicted population data.We can see the R² of statistical model method was 0.53,RMSE was 32848.13,while the R² of random forest model was 0.64,RMSE was 27874.50.By comparing the prediction accuracy of the two models,it can be found that R² of the random forest model is larger than that of the statistical model,while RMSE is smaller.Therefore,the method of combining machine learning and geographic information system?GIS?technology to construct the population random forest model has more advantages in spatial population.2)The proportion of over-standard population exceeding 35?g/m³ in the four seasons of the Yangtze River Delta in 2013 was close to 100%,and the high value concentrated in some areas.The spatial distribution of PM_2.5 exposed population is not continuous,showing obvious urban-rural differences in the whole Yangtze River Delta.The highest PM_2.5 pollution exposure risk is in winter and the lowest in summer.From the comparison results of PM_2.5 average concentration and population weighted average concentration in four provinces,it can be seen that except for the small difference between Shanghai PM_2.5 average concentration and population weighted PM_2.5 concentration,which can be ignored.There are certain differences in the other three provinces,among which Anhui Province has the largest difference,indicating that there is a spatial feature of population concentration in the high concentration area of PM_2.5.3)From 2010 to 2016,the population in the Yangtze River Delta has a higher health risk of PM_2.5 exposure,and more people are distributed in the heavily polluted areas.Generally speaking,PM_2.5 exposure is more serious in Shanghai,Jiangsu Province and the central and southern half of Anhui Province,and in some coastal cities in Zhejiang Province.PM_2.5exposure in urban areas is driven by population density,while PM_2.5 exposure in rural areas is driven by environmental pollution.There is a significant difference in the number of lung cancer deaths caused by PM_2.5 between urban and rural areas.