Geostationary Satellite Based Aerosol Optical Depth Retrieval And Analysis Of Its Temporal-Spatial Distribution

Aerosol is the gas particle in the air,which is a key factor affecting the ecological environment,climate change and human health.It is also the core content of the environmental monitoring by the environmental protection department and the meteorological forecasting department.Monitoring Aerosol by Ground-based site is a time-consuming and expensive method and cannot meet the requirements for monitoring large areas.Therefore how to use wide coverage and high time resolution remote sensing satellite data to monitor aerosol is a scientific issue and practical technology that needs to be improved.At present,most of the methods for retrieving aerosol from satellite remote sensing data are the traditional method which base on atmospheric Radiative transfer model This method is time-consuming,laborious,and complicated.There are too many assumptions and the atmospheric environment factors about this method,and the result cannot be obtained quickly.The traditional method of Aerosol optical thickness inversion based on remote sensing satellite data is deep blue algorithm,dark target method and empirical statistical method,etc.But there is less attention to focus on the application of deep learning method.In this study,we use the aerosol optical thickness data set from AERONET and remete sensing data set from advanced Himawari-8 Imager onboard Himawari-8 satellite to establish a deep learning algorithm to retrieve aerosol optical depth in Hubei Province.Then,we use the aerosol optical thickness data set from Whuan ground site to validate the feasibility of deep learning algorithm.We compare the AOD from deep learning algorithm and the AOD from Japan Aerospace Exploration Agency(JAXA)official base on the AOD fromNational Aeronautics and Space Administration(NASA)official to evaluate the accuracy of the AOD model.We retrieve AOD by deep learning algorithm in 2015-2018,and research on the distribution of time and space and reveal the intraday,monthly,quarterly and annual AOD in Hubei Province.We introduce the concept of background aerosol,comprehensively analyze the influencing factors of the temporal and spatial changes of AOD base on the meteorological site data and air quality site data in Hubei Province in the day.Finally,this research provides detailed temporal and spatial variation laws and scientific theoretical foundations.The main conclusions are as follows:(1)This use the good nonlinear generalization ability of deep learning algorithm to establish inversion model of AOD based on the nonlinear idea of atmospheric radiation transmission process in Hubei Province.In the ten-fold cross-validation result of the model,the R ratio is 0.98,the RMSE is 0.06,the MRE is 26%,and the WithEE is 92%.The results show that the model performs well in the data set.(2)To verify and analyze the AOD from deep learning algorithm,this paper compare it to the AOD from the JAXA official base on the AOD from MODIS official.The result show that,MRE is 32.76%,RMSE is 0.19,R is 0.81,and WithEE is 52.58%.The results of comparative analysis indicate that the AOD from deep learning algorithm is more refined and of higher quality than JAXA's official AOD products.(3)To analyze the spatial and temporal variation of AOD in the average annual,seasonal,monthly and intraday days of Hubei Province from 2015 to 2018 base on the AOD retrieved by deep learning algorithm.The results show that the average AOD is decreasing year by year in Hubei province from 2015 to 2018,means that the environmental management is effective.The average AOD is small in spring and summer,and more in autumn and winter.The average monthly AOD is polluted in month 11,12,and 1.Seriously,the average daily AOD from 8:00 to 17:00 in the day is first decreased and then increased.(4)This paper study the factors affecting AOD in Hubei Province based on longtime meteorological monitoring data and air quality monitoring data.The results show that rainfall in the past hour and NO2 are the main factors affecting daytime BAOD..In addition,through the study of AOD during the daytime interval,it is found that AOD is mainly affected by wind direction and wind speed,while BAOD is mainly affected by air pressure,temperature,water vapor pressure,rainfall in the past hour,SO2 and NO2.The correlation coefficients are 0.73,-0.66,-0.73,-0.71,0.73 and 0.71,respectively.The analysis results indicate that the transport aerosol is the main source of aerosol in Hubei province,while the background aerosol is mainly related to climate status and industrial production in Hubei Province.