Research On Inversion Model And Spatial-Temporal Change Evolution Of PM_2.5 In Typical Region Of China In Recent 20 Years

With the continuous advancement of the industrialization and urbanization,a series of environmental pollution problems have arisen and hindered the sustainable development of urbanization in China.It is necessary to study and reveal the spatial distribution and law of ecological environment pollution.In recent years,PM_2.5pollution has further intensified.Government departments and relevant scholars have continued to pay attention to this issue.At present,frequent and widespread haze pollution events occur in China.PM_2.5is the main component of haze pollution.PM_2.5 pollution affects air quality,regional and global climate and human health.As the rate of economic growth continues to accelerate,the contradiction between the pursuit of material life and the demand for a green and healthy living environment has become increasingly prominent.PM_2.5pollution has received widespread attention.In the past 20 years,the PM_2.5.5 pollution generated in the context of China’s rapid economic development is inconsistent with the new concept of"eco-priority green development"advocated by our country.It is necessary to control air pollution,improve regional air quality,adjust energy and industrial structure,promoting industrial green development and strengthening the construction of ecological civilization,and mastering the temporal and spatial evolution of PM_2.5 pollution is the basis for the above-mentioned work.In this study,four typical regions of Beijing-Tianjin-Hebei,the Yangtze River Delta,the northeastern three provinces of China and the Pearl River Delta are selected as the study area.The PM_2.5.5 observations of 308 monitoring stations from 2013 to 2016,the MODIS aerosol optical depth and five meteorological factors（border layer height,relative humidity,temperature,wind speed,wind direction）from 2000 to 2016 are the basic data sets.Support vector regression（SVR）and Back Propagation Artificial Neural Network（BPANN）methods were applied to form a combined simulation model.Based on the simulation model,the PM_2.5 concentration in the study area from2000 to 2012 was estimated.The spatial variation of PM_2.5 concentration over the past20 years was reconstructed using geospatial analysis techniques.The results showed that the R² of the combined model was 0.02⁰.15,0.16⁰.33 and 0.01⁰.07 higher than the BPANN,PLSR andε-SVR models,respectively,the RMSE was 0.90³.97and 3.54⁵.57 and 0.20⁰.63 respectively,the MAPE was 1.22%⁷.65%,8.64¹2.28and 0.15%¹.83%lower than the average of the three models.The analysis results of the characteristics of spatio-temporal changes showed that the concentration of PM_2.5in the northeastern three provinces of China and Beijing-Tianjin-Hebei increased at the rate of 0.104μg/m³ and 2.87μg/m³ from 2000 to 2010,and the rate of the Pearl River Delta decreased by 0.095μg/m³ per year.The trend of the gradual increase of PM_2.5 pollution in these three research areas was obvious,while the Yangtze River Delta showed minor fluctuations in the range of 47 to 52μg/m³,and the pollution scope did not show significant changes.In 2012,the concentrations of PM_2.5 in four research areas decreased,and the pollution scope narrowed as compared with previous years.However,the concentration of PM_2.5 increased slightly and then decreased from 2013 to 2016,the range of serious PM_2.5 pollution decreased（except for the Yangtze River Delta）.This is related to the Chinese government’s emphasis on ecological development and green development and has implemented a series of regional air pollution control measures.