Deformation Monitoring Based On Optical And Radar Image Pixel Offset Tracking Technology

Commonly used satellite imagery methods for monitoring the surface deformation generally include two types: Interferometric Synthetic Aperture Radar(In SAR)technology and pixel offset tracking technology.Among them,the advantages of In SAR accuracy are outstanding,but when the ground surface is deformed by a large amount,it is susceptible to the influence of Spatiotemporal decoherence,and effective measurement values cannot be obtained.Pixel offset tracking technology is a supplementary method.Although it is inferior to In SAR in terms of accuracy,it is not affected by phase decoherence and has a larger measurable deformation range.It can acquire two-dimensional deformation(the east-west and east-west direction of the optical image).North-south,range and azimuth of radar images)to deal with various large-deformation geological disasters or geological movements.However,limited by cloud cover and time resolution,the number of available images from a single optical satellite is often insufficient to support long-term monitoring of surface deformation;In addition,as more and more remote sensing images are opened to the public,how to make full use of these data and apply them to surface deformation monitoring becomes more and more important.In response to the above problems,this article will explore the application potential and the possibility of combining satellite images of different platforms for three typical large-deformation geological movements(landslides,earthquakes,and glaciers)based on medium-resolution remote sensing images.The main research work of this paper is as follows:(1)A method of synthesizing panchromatic bands is proposed.Based on the integral area of the spectral response function,the red,green,and blue bands of Landsat-8(L8)are synthesized to obtain an analog band that is highly correlated with the actual panchromatic band.According to the similar characteristics of Landsat-8 and Sentinel-2(S2)spectral characteristics,this method was applied to S2 image and compared with the four 10 m spatial resolution bands of S2.The result proves that the simulated panchromatic band has the best effect in this experiment.(2)Aiming at the Baige landslide,which has been sliding for a long time and accumulated deformation of tens of meters,this paper attempts to combine the offset results of optical satellite images of different platforms.Using 13 scenes of L8 images and 8 scenes of S2 images,the displacement time series from 2014 to 2018 were inverted,and the displacement time and space changes of Baige landslide in Sichuan were analyzed.After checking the consistency of the displacement patterns,an attempt was made to fuse the two displacement maps for time series inversion.The result of the fusion increases the time density of the displacement sequence and reflects more details of the displacement change.The results show that the landslide has experienced four periods of speed change from 2014 to 2018,and the maximum displacement can reach 54 m.(3)For the Kaikoura earthquake in New Zealand with a displacement of several meters,an attempt was made to invert the three-dimensional deformation field by combining the offset results of the optical image and the radar image.Through the feature point-based pixel migration technology,the Sentinel-1(S1)is used to obtain the distance and azimuth displacement of the earthquake;the conventional optical image-based offset tracking technology is used to obtain the two-dimensional image of the Sentinel-2 Deformation field(east-west direction and north-south direction).According to the geometric model of the image obtained by the radar satellite and the two-dimensional displacement of S2 as a constraint,the three-dimensional deformation field of this earthquake is inverted.The results are basically consistent with GPS points,verifying the feasibility of this method.(4)Based on the time series inversion method,analyze and compare the performance of Sentinel-1 and Landsat-8 in the high-speed Patagonian glacier.By comparing the inverted displacement maps of two different satellites,it is found that L8 can detect glacier movement for a longer period of time,and the detection range can penetrate deep into the ice sheet.On the other hand,S1 has a large number of low-quality points on the edge of the glacier and its diversion,which are eliminated,resulting in sparse displacement points and distributed in blocks,which cannot fully reflect the displacement changes of the glacier.Therefore,L8 is a better choice for the displacement detection of high-speed glaciers.