Intelligent Reconstruction Of Three-dimensional Marine Environmental Elements From Satellite Remote Sensing

Marine environment affects underwater sound propagation and navigation of submersibles,and is of great significance to activities such as target detection and positioning,underwater communication and countermeasures.Due to the limitations in time and space of various kinds of underwater parameter field measurement equipment,it is increasingly difficult to meet the needs of underwater acoustic detection and countermeasure.How to quickly and accurately obtain underwater environmental parameters has become a hot issue in the field of Marine research in recent years.Marine remote sensing observation data has continuity in time and wide region in space,and there is correlation with ocean water parameters.Reconstruction of underwater environment through satellite remote sensing data has become the research direction of many scholars.The observation satellite design for system error is inevitable,observations from the actual situation there is error,and the error analysis of Marine underwater environment real-time model predicted results will affect,understand this influence,to explore the real-time analysis under what kind of error model can predict the results of the optimal design for satellite to provide parameter basis.In view of the above problems,this paper firstly uses multiple linear regression method to evaluate the influence of satellite observation errors on the temperature reconstruction field.By using SLA of sea surface height anomaly data in the South China Sea and SSTA of sea surface temperature data,different errors are added into the control variable method to reconstruct the sea surface temperature field.The errors added by SLA and SSTA when the reconstruction effect is optimal are found,which provides reference data for the design of Marine remote sensing satellite.Secondly,the algorithm for constructing the reconstruction model of seawater temperature field is studied.After data cleaning of the original data,the reconstruction models of three seawater temperature fields are respectively constructed by SVR-EOF,lightgbm(Light Gradient Boosting Machine)and one-dimensional variational algorithm,and the measured data of Argo are used to compare the models at 50m,100m,A preliminary test of 500m has been carried out to verify the effectiveness of the model.Finally,root mean square error(RMSE)and determination coefficient(R~2)were used to evaluate the three models,and Argo raw data were used to test the performance of the three reconstruction models to explore the advantages of different reconstruction models.By analyzing the correlation between sea surface parameters and ocean temperature and the distribution of mesoscale vortexes,the relationship between sea surface elements and mesoscale vortexes and model reconstruction performance and error distribution was explored.