Research On GB-InSAR Complex Atmospheric Phase Correction Method Based On PS Point Classification

Facing the problems of geological disasters such as landslides which occur more frequently year by year,it seriously affects national economic development and people’s livelihood security.GB-In SAR has been widely used in deformation monitoring field in recent years due to its advantages of flexible layout,all-weather,and measuring accuracy up to sub-millimeter.However,changes in atmospheric conditions such as temperature,air pressure and humidity will causes refraction and a delay in the propagation path of the electromagnetic wave,which will lead to atmospheric phase errors and affect the accuracy of deformation measurement results.Therefore,this paper studies the complex atmospheric phase correction problem in GB-In SAR,aiming at its shortcomings on the basis of existing solutions.The corresponding atmospheric correction model and processing method are proposed.The main work of this paper is as follows:(1)Aiming at the problem of complex atmospheric phase fluctuations under complex environmental and meteorological conditions,this paper studied the main components of the differential interference phase of PS point in the scene: The time series data characteristics of deformation phase component,noise phase component and atmospheric phase component can be calculated and reflected by the three characteristic calculation methods of interference phase accumulation,interference phase standard deviation and least square fitting,and the differences of time dimension such as creep,time accumulation and volatility of different phase components of PS point can be calculated and reflected.(2)With the enlargement of the radar monitoring angle,the atmospheric phase will show complex spatiotemporal fluctuations.In order to solve the correction problem of complex atmospheric phase in the observation scence.In this paper,the traditional atmospheric correction method based on radar image is improved,and Transformer network model is introduced to match the change rules of each phase category and extracted phase features by using the timing multi-head attention mechanism,and a complex atmospheric phase correction based on machine learning PS point classification is proposed.In order to obtain accurate deformation monitoring results,the space dimension interpolation method is used for atmospheric correction of PS points which are classified without deformation and less affected by noise.Experimental results show that the improved method can effectively correct the complex atmospheric phase of GB-In SAR.(3)The PS point screening threshold mentioned above is set by human experience.The selection of PS point threshold will directly affect the accuracy and density of PS points,thus affecting the classification of PS points and the effect of atmospheric correction.Therefore,this paper makes improvements from the perspective of PS point threshold selection,and selects PS points through threshold optimization to conduct atmospheric phase correction based on PS classification.Compared with the previous complex atmospheric phase correction experiment for PS point classification with artificially set thresholds,it is proved that the classification and atmospheric correction method of PS points selected by the optimal thresholds in this paper can further improve the effect of atmospheric correction,thus improving the monitoring accuracy of GBIn SAR deformation.