Research On Rapid Construction Method Of Remote Sensing Inversion Parameters For AOD

National environmental protection agency require real-time and wide-range atmospheric monitoring data for air quality assessment,but existing atmospheric remote sensing inversion methods for scientific research need to be further studied in related fields.Aerosol Optical Depth(AOD)is an index of atmospheric pollution.Accelerating AOD's rapid inversion can effectively grasp the spatio-temporal distribution of atmospheric information in real time.This has been used to study multiple haze intrusions in most areas of China.Significance.At present,China has already launched high-resolution satellites with high spatial resolution of sub-meters,which also imposes higher requirements on the computational capabilities of remote sensing inversion algorithms.This paper focuses on the construction of remote sensing inversion parameters used in the AOD remote sensing inversion of satellite-to-ground coverage simulation system under the auspices of the Chinese academy of sciences institute of remote sensing and digital earth,the goal is to solve the problem of long formation time of inversion parameters.Aiming at this problem,this paper proposes a task scheduling and scheduling algorithm that is suitable for the rapid construction of remote sensing inversion parameters in a heterogeneous GPU cluster environment.The task scheduling strategy includes the task scheduling model for the rapid construction of remote sensing inversion parameters.Task scheduler structure design.The task scheduling algorithm includes dynamic decomposition algorithm and dynamic allocation algorithm of the task.The construction of AOD remote sensing inversion parameters requires a large number of input parameters.These large amounts of input parameters are calculated,and finally the required AOD remote sensing inversion parameters are generated.This generation process requires a lot of calculation time.The scheduling strategy and scheduling algorithm proposed in this paper can group a large number of input parameters reasonably,decompose it into different independently executed parameter combinations,and then assign these parameter combinations to different node machines to perform calculations to generate AOD remote sensing inversion.parameter.This paper verifies the scheduling strategy and scheduling algorithm in a heterogeneous GPU cluster environment and obtains the experimental results.The experimental results are compared and analyzed to verify the feasibility of the scheduling strategy and scheduling algorithm.The main research contents of this article are:(1)The research status of AOD remote sensing inversion at home and abroad,the construction principles of remote sensing inversion parameters,and heterogeneous GPU cluster architecture are studied.(2)A scheduling strategy for the construction of AOD remote sensing inversion parameters was proposed.The strategy includes the design of the scheduling model for the construction of remote sensing inversion parameters and the structure design of the task scheduler.(3)A scheduling algorithm for constructing AOD remote sensing inversion parameters is proposed.The algorithm is designed based on the proposed task scheduling strategy,including dynamic decomposition algorithm and dynamic allocation algorithm.(4)Experimental verification of the scheduling strategy and scheduling algorithm proposed in this paper,and comparative analysis of the test results,thus illustrating the feasibility of the strategy and algorithm.The scheduling strategy and scheduling algorithm proposed in this paper provide a more stable construction environment for the rapid construction of remote sensing inversion parameters,which ensures the accuracy of the construction results of remote sensing inversion parameters,and can effectively shorten the construction time of remote sensing inversion parameters.The construction of AOD remote sensing inversion parameters is fast and accurate to meet the needs of the relevant departments for real-time monitoring of atmospheric pollution,which provides a certain method and theoretical basis for the future research and practice of rapid remote sensing inversion.