Research And Application Of Parallel Algorithm Of PM_2.5 Retrieval Based On Active And Passive Remote Sensing

Atmospheric environmental conditions have a greater impact on various industries and personallives.The well-known particulate PM_2.5 is one of the major atmospheric pollutants and an important indicator of national air quality monitoring.The PM_2.5 concentration data released by most current information platforms is mainly derived from the real-time monitoring of ground-based remote sensing sites.Modern remote sensing technology has been booming since the middle of the last century,promoting the development of satellite remote sensing technology as a more effective way to monitor the atmospheric environment,with better temporal-spatial resolution coverage.The most widely used data of satellite remote sensing in the field of air pollution monitoring is Aerosol Optical Depth?AOD?.Active laser radar can obtain high-precision aerosol vertical distribution information,but it can only observe the atmosphere near the sub-satellite track,and limited spatial coverage.Passive instruments have abundant spectral information and wide coverage,but it is difficult to analyze the vertical distribution of aerosols and other pollutants.The key to the estimation of PM_2.5 in satellite remote sensing products is to establish the relationship between PM_2.5 and AOD.The main factor affecting the correlation is the vertical distribution of aerosol information.While researchers constantly improve the accuracy of remote sensing,various sectors of the society have also put forward urgent demands for the real-time nature of satellite remote sensing,especially in areas such as disaster warning,environmental detection,and military investigation,and even require real or near-real-time processing.At the same time,how to make atmospheric remote sensing technology better integrated into the public life service and achieve close integration with“Internet Plus”in such areas as public health and travel planning.In order to help solve the above problems,this study will use the combination of active and passive remote sensing to invert PM_2.5,improve the accuracy of inversion,and use GPU parallel acceleration to improve the efficiency of PM_2.5inversion algorithm.Finally,using satellite remote sensing to invert PM_2.5.5 data,based on Android platform,design a PM_2.5 monitoring system to achieve monitoring of user movement trajectory PM_2.5 exposure statistics.The main research content of this paper is as follows:?1?Retrieval of PM_2.5 using a combination of active and passive remote sensing.Firstly,based onthe vertical profile of aerosol extinction and the aerosol elevation retrieved from active and passive load observations,the vertical distribution of regional aerosols simulated by the atmospheric model was corrected.and a new method to calibrate aerosol extinction coefficient profiles using MODIS column concentrations was proposed.The extinction coefficient correlation between different layers of CALIPSO and GEOS-Chem was increased from 0.69 to 0.91.On this basis,aerosol extinction was obtained by AOD vertical correction,and then based on a statistical correlation model.The aerosol extinction information was converted to PM_2.5concentrations at the regional scale,and we finally retrieved PM_2.5 data released by the China Environmental Monitoring Center with a R reached 0.87.?2?Based on GPU to achieve parallel PM_2.5 retrieval algorithm.Firstly,parallel analysis of PM_2.5inversion algorithm is performed.Based on the characteristics of CUDA?Compute Unified DeviceArchitecture?parallel computing architecture,the CPU and GPU task allocation,GPU thread mapping anddata fetching are rationally designed,Finally,a single Kernel function,asynchronous data transmission,and GPU storage of process data are formed,and the algorithm is efficiently parallelized.Experiments show that CUDA parallelization accelerates.The inversion algorithm can achieve up to 7 times more acceleration.?3?Based on Android platform to design a PM_2.5 monitoring system.Using satellite remotesensing retrieved PM_2.5 data as a data source,integrating Baidu map SDK service to record daily trajectories of users,realizing monitoring and statistics of the user's motion trajectory PM_2.5 concentrations,and using Jpush SDK service for real-time reminding,making atmospheric remote sensing technology better integrated into the public life service and achieve close integration with“Internet Plus”in such areas as public health and travel planning.