InSAR troposphere delay correction method and system based on DEM basin segmentation

Abstract

The present application relates to a DEM basin segmentation-based InSAR troposphere delay correction method and system, which comprises the following steps: processing DEM data, precise orbit data and time-series SAR images to obtain deformation rates without removing the influence of troposphere delay; performing sub-block segmentation on the research area in vertical and horizontal dimensions; modeling the troposphere delay of each sub-block area to obtain the troposphere delay model of each sub-block area; iteratively optimizing the troposphere delay model of each sub-block area to obtain the iteratively optimized troposphere delay phase of each sub-block area; and calculating the deformation rate without the influence of troposphere delay. The present application significantly improves the troposphere delay correction accuracy in high mountain valley areas with obvious terrain undulations and complex environments, provides a more applicable adaptive atmospheric delay correction for InSAR geological disaster monitoring, and improves the accuracy and application scope of InSAR deformation monitoring.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of ground deformation monitoring, in particular to an InSAR troposphere delay correction method and system based on DEM watershed segmentation. BACKGROUND

[0002] Interferometric Synthetic Aperture Radar (InSAR) technology has become the main tool for ground deformation monitoring at large spatial scales due to its high precision, high resolution and high efficiency. There are geological hazards such as mudslides and landslides in high mountain valley regions, and traditional ground monitoring methods are often limited by the difficulty of reaching the measurement area and low measurement efficiency, making it difficult to monitor high mountain valley regions with high precision and large scale. InSAR technology complements the shortcomings of traditional measurement strategies and has important value for deformation monitoring in high mountain valley regions. However, the accuracy of InSAR deformation measurement is significantly affected by atmospheric delay phase, mainly troposphere delay. These delays are caused by changes in water vapor content in the troposphere, which is usually related to topography and can cause significant errors in InSAR observations. At the same time, there are steep elevation gradients and dynamic atmospheric conditions in high mountain valley regions, driven by unique meteorological phenomena such as katabatic wind effect and inversion layer, the spatial heterogeneity of water vapor leads to horizontal and vertical variability of the atmosphere, making the correction of troposphere delay in this area further complicated and a major challenge.

[0003] Current methods for correcting troposphere delay mainly include two types: using external data for auxiliary correction and using the internal characteristics of time series InSAR data for correction. External auxiliary data sets include various high-resolution imaging spectrometers (such as MODIS and MERIS), numerical weather models (such as WRF and ERA), global satellite navigation systems (GPS), etc. However, the effectiveness of these methods depends largely on the amount, quality and resolution of the data used. In high mountain valley regions, it is not realistic to obtain high-resolution data sets, so it is usually not available to correct the troposphere delay in this area by external data assistance. The other method focuses on the internal characteristics of InSAR phase, using filtering and modeling to process time series InSAR data to correct troposphere delay. However, the general standard for selecting the filter window is usually based on empirical selection of prior information; common modeling methods (such as linear model and power law model) are not sufficient to describe the specific situation of troposphere delay in high mountain valley regions. Considering the frequent occurrence of geological disasters in high mountain valley regions and the difficulty of deformation monitoring, there is an urgent need for a suitable solution to correct the troposphere delay in this area. SUMMARY

[0004] The present application provides an InSAR troposphere delay correction method and system based on DEM watershed segmentation to solve at least one of the above technical problems.

[0005] The technical scheme for solving the above technical problem is as follows: an InSAR tropospheric delay correction method based on DEM basin segmentation, comprising:

[0006] S1, obtaining DEM data, precise orbit data and time-series SAR images within a preset time range of a research area, processing the time-series SAR images, the DEM data and the precise orbit data to obtain time-series InSAR interference images and deformation rates without removing the influence of tropospheric delay;

[0007] S2, performing sub-block segmentation in two dimensions of vertical and horizontal directions on the research area according to the time-series InSAR interference images and the basin information and DEM data of the research area to obtain a plurality of sub-block regions;

[0008] S3, modeling the tropospheric delay of each sub-block region based on the time-series InSAR interference images to obtain the tropospheric delay model of each sub-block region in the time-series InSAR interference images;

[0009] S4, using a weighted least squares iteration method to iteratively optimize the tropospheric delay model of each sub-block region in the time-series InSAR interference images based on the deformation rate without removing the influence of tropospheric delay to obtain the iterated tropospheric delay phase of each sub-block region in the time-series InSAR interference images;

[0010] S5, calculating the deformation rate without removing the influence of tropospheric delay according to the iterated tropospheric delay phase of each sub-block region in the time-series InSAR interference images.

[0011] Based on the above InSAR tropospheric delay correction method based on DEM basin segmentation, the application further provides an InSAR tropospheric delay correction system based on DEM basin segmentation.

[0012] The InSAR tropospheric delay correction system based on DEM basin segmentation comprises:

[0013] An interference processing module is configured to obtain DEM data, precise orbit data and time-series SAR images within a preset time range of a research area, process the time-series SAR images, the DEM data and the precise orbit data to obtain time-series InSAR interference images and deformation rates without removing the influence of tropospheric delay;

[0014] A region division module is configured to perform sub-block segmentation in two dimensions of vertical and horizontal directions on the research area according to the time-series InSAR interference images and the basin information and DEM data of the research area to obtain a plurality of sub-block regions;

[0015] a modeling module configured to model tropospheric delay for each sub-block region based on the time-series InSAR interferogram, to obtain a tropospheric delay model for each sub-block region in the time-series InSAR interferogram;

[0016] an iterative optimization module configured to iteratively optimize the tropospheric delay model for each sub-block region in the time-series InSAR interferogram based on the deformation rate without the influence of tropospheric delay by using a weighted least squares iterative method, to obtain an iterated tropospheric delay phase for each sub-block region in the time-series InSAR interferogram;

[0017] a deformation calculation module configured to calculate the deformation rate without the influence of tropospheric delay according to the iterated tropospheric delay phase for each sub-block region in the time-series InSAR interferogram.

[0018] The present application has the advantages that: the DEM watershed segmentation-based InSAR tropospheric delay correction method and system considers the complex topography and climate conditions in high mountain valley areas, corrects the tropospheric delay in combination with vertical and horizontal atmospheric changes, divides the research area into sub-regions in the horizontal dimension by using watershed boundaries, and makes the sub-regions naturally aligned with the topography and atmospheric model; in the vertical dimension, stratification caused by atmospheric inversion layers is taken into account to improve the correction accuracy of the tropospheric delay; by integrating the horizontal and vertical spatial features, the present application significantly improves the tropospheric delay correction accuracy in high mountain valley areas with obvious topography and complex environment, provides a more applicable adaptive atmospheric delay correction for InSAR geological disaster monitoring, and improves the accuracy and application breadth of InSAR deformation monitoring. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 a flowchart of the DEM watershed segmentation-based InSAR tropospheric delay correction method of the present application;

[0020] Figure 2 a baseline diagram for the time-series SAR image interferometric combination according to the SBAS-InSAR grouping principle in the embodiment;

[0021] Figure 3 a schematic diagram of the weighted least squares iterative method;

[0022] Figure 4 a diagram showing the DEM watershed segmentation and specific analysis;

[0023] Figure 5 a diagram showing the comparison between the results of the DEM watershed segmentation-based InSAR tropospheric delay correction method of the present application and the traditional correction method;

[0024] Figure 6A comparison diagram of residual phase after correction by the InSAR tropospheric delay correction method based on DEM basin segmentation of the present application and a traditional correction method;

[0025] Figure 7 A diagram of residual phase to height ratio after correction by the InSAR tropospheric delay correction method based on DEM basin segmentation of the present application and each correction method;

[0026] Figure 8 A diagram of annual average deformation rate change after atmospheric correction by the InSAR tropospheric delay correction method based on DEM basin segmentation of the present application and a traditional correction method;

[0027] Figure 9 A diagram of statistical analysis result of time displacement sequence by the InSAR tropospheric delay correction method based on DEM basin segmentation of the present application and a traditional method in a specific area;

[0028] Figure 10 A structural block diagram of the InSAR tropospheric delay correction system based on DEM basin segmentation of the present application. DETAILED DESCRIPTION

[0029] The principles and characteristics of the present application are described below in combination with the drawings, and the examples are only used to explain the present application and are not used to limit the scope of the present application.

[0030] As shown in Figure 1 , the InSAR tropospheric delay correction method based on DEM basin segmentation includes:

[0031] S1, obtaining DEM data, precise orbit data and time-series SAR images in a preset time range of a research area, processing the time-series SAR images, the DEM data and the precise orbit data to obtain time-series InSAR interference images and deformation rates without removing the influence of tropospheric delay;

[0032] S2, according to the time-series InSAR interference images and the basin information and DEM data of the research area, performing sub-block segmentation in two dimensions of vertical and horizontal directions on the research area to obtain a plurality of sub-block regions;

[0033] S3, based on the time-series InSAR interference images, respectively modeling the tropospheric delay of each sub-block region to obtain a tropospheric delay model of each sub-block region in the time-series InSAR interference images;

[0034] S4, based on the deformation rate without the influence of the tropospheric delay, using a weighted least squares iterative method to iteratively optimize the tropospheric delay model of each sub-block region in the time-series InSAR interferogram, and obtain the iterated tropospheric delay phase of each sub-block region in the time-series InSAR interferogram;

[0035] S5, according to the iterated tropospheric delay phase of each sub-block region in the time-series InSAR interferogram, calculate the deformation rate without the influence of the tropospheric delay.

[0036] The following will be described in detail:

[0037] The S1 includes the following S11-S16:

[0038] S11, obtaining DEM data, precise orbit data and time-series SAR images in a preset time range of a research area.

[0039] Specifically, the time-series SAR images can be directly downloaded through the Alaska Satellite Facility Center (https: / / search.asf.alaska.edu / # / ), and the time interval of the time-series SAR images used in the present example is 12 days. The obtained time-series SAR images should ensure that they have appropriate time baseline and spatial baseline to reduce the error influence in the interference process. The DEM data can be obtained from the COP-90 DEM data of the TanDEM mission of the German Aerospace Center, which can be directly downloaded from the OpenTopography website (https: / / portal.opentopography.org / ). The precise orbit data can be directly obtained from the Alaska Satellite Facility Center (https: / / s1qc.asf.alaska.edu / aux_poeorb / ), and the precise orbit data is usually generated 21 days after imaging.

[0040] S12, according to the group pairing principle of SBAS-InSAR, the time-series SAR images are interfered to obtain initial time-series InSAR interferogram.

[0041] Specifically, the specific combination of the time-series SAR images interfered according to the group pairing principle of SBAS-InSAR is as shown in Figure 2 The time-series SAR images include a plurality of SAR images arranged in time, and two SAR images satisfying a preset condition are combined as an image pair and interfered, wherein the number of image pair combinations formed in the time-series SAR images satisfies the following inequality:

[0042]

[0043] wherein K is a constant, K+1 represents the total number of SAR images in the time-series SAR images, and M represents the number of combinations of interferometric pairs, i.e. the total number of initial InSAR interferometric images in the initial time-series InSAR interferometric images.

[0044] S13, removing redundant terrain phases and orbit errors in the initial time-series InSAR interferometric images by using the DEM data and the precise orbit data to obtain intermediate time-series InSAR interferometric images.

[0045] Specifically, in the process of removing the redundant terrain phases and orbit errors in the initial time-series InSAR interferometric images by using the DEM data and the precise orbit data: the DEM data is used for terrain phase removal, and at the same time, the elevation of the study area can be calculated; and the precise orbit data is collected to remove the orbit errors.

[0046] S14, selecting and filtering coherent points of the intermediate time-series InSAR interferometric images to remove noise and turbulent delay to obtain time-series InSAR interferometric images.

[0047] Specifically, in the process of selecting and filtering coherent points of the intermediate time-series InSAR interferometric images, high-coherence points are selected as the research objects, and the phase data of the coherent points represent the phase calculation data of the overall region; filtering can be performed by using Goldstein filtering, mean filtering and other filtering methods, which can remove most of the noise and part of the turbulent delay, so as to realize the elimination of low-coherence regions.

[0048] S15, performing phase unwrapping on the time-series InSAR interferometric images to obtain unwrapped interferometric phases without removing the influence of the tropospheric delay.

[0049] Specifically, in the process of performing phase unwrapping on the time-series InSAR interferometric images to obtain unwrapped interferometric phases without removing the influence of the tropospheric delay, the unwrapped interferometric phases only contain the deformation phases in the time series and the tropospheric delay phases mainly in the vertical layering.

[0050] S16, performing singular value decomposition processing on the unwrapped interferometric phases without removing the influence of the tropospheric delay to obtain deformation rates without removing the influence of the tropospheric delay.

[0051] Specifically, the singular value decomposition processing is performed on the unwrapped interferometric phases, wherein singular value decomposition is a method of decomposing any matrix into the product of three matrices, which is suitable for solving the rank deficiency or overdetermined problem in a linear equation set. In SBAS-InSAR, singular value decomposition is used to process the phase equation set composed of multiple interferometric pairs, integrate redundant information and extract principal components. The specific formula for calculating the deformation rate based on singular value decomposition is as follows:

[0052]

[0053] In the formula, is a deformation rate, and are a space-time model and a deformation parameter, respectively.

[0054] In summary, after obtaining the DEM data, precise orbit data and time-series SAR images, the S1 performs image registration, interferometric phase calculation, terrain effect removal, orbit error removal, phase filtering and phase unwrapping, etc. Specifically, the terrain phase, orbit error, incoherent noise and most turbulent delay phases are removed by using the Goldstein filtering, mean filtering and eigenvalue decomposition of the coherence matrix, etc. The deformation phase without removing the tropospheric delay phase is obtained, and the deformation rate without removing the influence of the tropospheric delay is calculated.

[0055] The S2 includes the following S21-S22:

[0056] S21, considering the influence of the inversion layer on the vertical atmospheric water vapor distribution, according to the DEM data corresponding to the coherent points in the time-series InSAR interferogram, combining the appearance time and prior height of the inversion layer at the location of the research area, the research area is vertically divided into blocks;

[0057] S22, based on the segmentation strategy of the basin boundary, according to the basin information of the research area, the vertically divided research area is horizontally divided into blocks to obtain a plurality of sub-block regions.

[0058] Specifically, the atmospheric changes in the vertical and horizontal dimensions are considered in the segmentation process of the research area: the influence of the inversion layer on the vertical atmospheric water vapor distribution is considered in the vertical dimension, and the experimental area is divided in the vertical direction combined with the prior inversion layer height data; in the horizontal dimension, the specific terrain of the experimental area is taken as a reference, and the DEM basin-based window segmentation strategy is used for block processing according to the basin information, which is beneficial to the subsequent fine correction of the tropospheric delay.

[0059] In addition, the basin information is obtained by geographical analysis of the prior information of the topographic features of the research area.

[0060] The S3 is to model the tropospheric delay by using an independent linear model. Assuming that there are K+1 SAR images in the same sub-block region, the phase component of the qth pixel on the kth SAR image is which can be summarized as:

[0061]

[0062] where r k(q) is the reflectivity phase of the scatterer corresponding to the qth pixel on the kth SAR image, a k (q) is the reflectivity phase of the scatterer corresponding to the qth pixel on the kth SAR image, a k (q) is the tropospheric delay phase of the qth pixel on the kth SAR image, and λ is the wavelength. One of the K+1 SAR images is selected as the master image, and the interference phase of the qth pixel on the kth SAR image is k (q) is the signal of the acquisition geometry φ rk (q) is the signal of the terrain motion φ dk (q) is the signal of the decorrelation φ σk (q) is the signal of the tropospheric delay φ ak (q) is composed of:

[0063] φ k (q) = φ rk (q) + φ dk (q) + φ σk (q) + φ ak (q);

[0064] The tropospheric delay is affected by the water vapor data. Generally, the water vapor content decreases exponentially with the decrease of the altitude. Assuming that K+1 SAR images generate M InSAR interference images, the tropospheric delay phase of the mth InSAR interference image can be obtained by using a linear model on the local area scale of the mth InSAR interference image

[0065]

[0066] K m is the ratio of the tropospheric delay to the altitude in the mth InSAR interference image, is the intercept of the tropospheric delay phase in the mth InSAR interference image, and h is the height.

[0067] In each divided sub-block, fine modeling (re-expression) is performed according to the above linear model of the layered delay, and the tropospheric delay model of each sub-block region in the time-series InSAR interference image is obtained, which is:

[0068]

[0069] wherein, represents the tropospheric delay phase of the ith sub-block region in the mth InSAR interference image of the time-series InSAR interference image; K m,ithe tropospheric delay phase to height ratio of the i-th sub-block region in the m-th InSAR interferogram; h i the height of the i-th sub-block region; the tropospheric delay phase intercept of the i-th sub-block region, which has the physical meaning of the atmospheric delay reference value at the reference height in the selected study area, and the reference height is generally the sea level height.

[0070] The S4 comprises the following S41-S42:

[0071] S41, phase unwrapping is performed on the time-series InSAR interferogram to obtain unwrapped interferometric phases without removing the influence of the tropospheric delay, and the unwrapped interferometric phases without removing the influence of the tropospheric delay in each sub-block region of the time-series InSAR interferogram are taken as the initial tropospheric delay phases of each sub-block region of the time-series InSAR interferogram; based on the tropospheric delay model of each sub-block region of the time-series InSAR interferogram, the initial tropospheric delay phases and the height of each sub-block region of the time-series InSAR interferogram are least squares fitted, and the initial tropospheric delay phase to height ratio and the initial tropospheric delay phase intercept of each sub-block region of the time-series InSAR interferogram are obtained.

[0072] S42, the initial tropospheric delay phase, the initial tropospheric delay phase to height ratio, the initial tropospheric delay phase intercept, and the deformation rate without removing the influence of the tropospheric delay of each sub-block region of the time-series InSAR interferogram are substituted into the weighted least squares iterative algorithm model for iterative correction to obtain the iterative tropospheric delay phase of each sub-block region of the time-series InSAR interferogram.

[0073] Specifically, the weighted least squares iterative algorithm model is:

[0074]

[0075] wherein, the tropospheric delay phase of the i-th sub-block region in the m-th InSAR interferogram of the time-series InSAR interferogram at the j+1-th iteration, the tropospheric delay phase of the i-th sub-block region in the m-th InSAR interferogram at the j-th iteration, and the initial value of the iteration number j is 0; when j=0, the initial tropospheric delay phase of the i-th sub-block region of the time-series InSAR interferogram; W m,i,j the weight index of the i-th sub-block region in the m-th InSAR interferogram at the j-th iteration, and when j=0, W m,i,0calculated from the deformation rate without removing the effect of tropospheric delay; h i represents the elevation of the ith sub-block region; K m,i,j represents the ratio of the tropospheric delay phase to the elevation of the ith sub-block region in the mth InSAR interferogram in the jth iteration, represents the intercept of the tropospheric delay phase of the ith sub-block region in the mth InSAR interferogram in the jth iteration, K m,i,j and is obtained by least square fitting of the tropospheric delay phase of the ith sub-block region in the mth InSAR interferogram in the jth iteration and the elevation of the ith sub-block region, i.e., the tropospheric delay phase of the ith sub-block region in the mth InSAR interferogram in the jth iteration and the elevation of the ith sub-block region h i are substituted into and least square fitting is performed, so that K m,i,j and are obtained. m,i,0 and are the initial ratio of the tropospheric delay phase to the elevation and the initial intercept of the tropospheric delay phase of the ith sub-block region in the mth InSAR interferogram, respectively.

[0076] Specifically, the calculation formula of the weight index is:

[0077]

[0078] wherein, v m,i,j represents the deformation rate of the ith sub-block region in the mth InSAR interferogram in the jth iteration, v m,i,j is calculated from the tropospheric delay phase of the ith sub-block region in the mth InSAR interferogram in the jth iteration; represents the threshold of the deformation rate of the ith sub-block region in the mth InSAR interferogram in the jth iteration, and σ m,i,j represents the standard deviation of the deformation rate of the ith sub-block region in the mth InSAR interferogram in the jth iteration.

[0079] The iteration stopping condition of the weighted least square iteration algorithm model is:

[0080]

[0081] wherein, ||·|| represents the Euclidean norm, ε represents a priori parameter, ε is set to 0.002 rad in the embodiment, n m,i represents the number of coherent points of the ith sub-block region in the mth InSAR interferogram.

[0082] In summary, the principle of the weighted least squares iterative algorithm in S4 is as shown in the following formula (1) : Figure 3

[0083] S5 includes the following S51-S53:

[0084] S51, the tropospheric delay phase of each sub-block region in the time series InSAR interferogram is iterated and integrated to obtain the optimized time series tropospheric delay phase of the study area;

[0085] S52, the time series InSAR interferogram is phase unwrapped to obtain the unwrapped interferometric phase without removing the influence of tropospheric delay, and the unwrapped interferometric phase without removing the influence of tropospheric delay is subtracted from the optimized time series tropospheric delay phase to obtain the deformation phase removing the influence of tropospheric delay;

[0086] S53, the deformation phase removing the influence of tropospheric delay is singular value decomposed to obtain the deformation rate removing the influence of tropospheric delay.

[0087] After the S5 is completed, the processed experimental results can be compared with the results of the traditional tropospheric delay correction model to evaluate the applicability and performance in the high mountain valley area. It should be noted that the traditional tropospheric correction methods for result evaluation include: global linear model (LM), regular window segmentation linear model (RSLM), external data set assisted correction based on ERA-5 and GACOS;

[0088] The corrected experimental results include: corrected interferogram, corrected interferometric phase and corrected deformation rate;

[0089] Specifically, the correction effect of each strategy can be intuitively compared by the phase standard deviation corrected by each method; further, the performance can be distinguished by the removal of the linear relationship between the phase and the elevation before and after correction.

[0090] In order to increase the readability of the present application, the following will further explain and describe the InSAR tropospheric delay correction method based on DEM watershed segmentation provided by the present application through a high mountain valley tropospheric delay correction example that has occurred, obviously, the example is only as an example:

[0091] I. Overview of the study area

[0092] ​This study area is the high-altitude valley of the Deqin-Lancang River basin, roughly located between 98° and 99° east longitude and 28° and 29° north latitude. The region is characterized by mountains on both sides and valleys in the middle, with the Lancang River flowing through it. The terrain is highly varied, with elevations ranging from 1564m to 6644m, and prone to frequent geological disasters (such as landslides and debris flows), making it one of the most geologically hazardous areas in Yunnan Province. Furthermore, the Salween River, the Lancang-Mekong River, and the Jinsha River have carved steep valleys in this region, creating a complex geographical landscape. The towering mountains act as natural barriers, blocking the inflow of warm, humid air from the Indian Ocean. Simultaneously, the deep valleys provide efficient channels for water vapor transport. This unique combination results in highly variable and complex water vapor distribution within the valleys, leading to significant tropospheric delays in InSAR interferometric images and affecting InSAR deformation monitoring.

[0093] II. Data Acquisition

[0094] Considering the terrain orientation of the high mountains and valleys in the Deqin-Lancang River basin, Sentinel-1 time-series SAR imagery with down-orbiting was selected to complete the example, as shown in Table 1. Additionally, COP-90 DEM was collected for geocoding and removal of phase errors caused by topography in the interferometric phase.

[0095] Table 1

[0096]

[0097] III. Data Preprocessing

[0098] Twenty-seven Sentinel-1 time-series SAR images collected in 2021 were used. A total of 75 InSAR interferometric images were generated using GAMMA software, with spatial and temporal baselines limited to within 300 meters and 80 days, respectively. The contribution of terrain to phase was removed using COP-90DEM. Subsequently, the InSAR interferometric images were processed using the SBAS-InSAR algorithm in StaMPS software, identifying 5,404,200 coherent points with coherence values ​​exceeding 0.65, and calculating the deformation rate of these coherent points without removing tropospheric delay.

[0099] IV. Sub-block segmentation

[0100] In this embodiment, to account for geographical variability and better capture topographically related tropospheric delay characteristics, the study area is divided into 26 sub-blocks based on watersheds in the horizontal dimension, such as... Figure 4 (a) is shown by the red line. The size of each sub-block is further controlled within 10–15 km, which reduces the influence of deformation signals on elevation phase fitting while mitigating turbulence-related atmospheric effects. For a typical sub-block (Figure 4 The elevation phase scatter plot of the interferogram obtained between 20210225 and 20210414 is shown in Figure A). Figure 4 As shown in (b), the relationship between elevation and phase in this sub-block deviates significantly from a simple linear trend. Meteorological studies indicate that the inversion layer near the Tibetan Plateau is generally around 3000m high, resulting in a clear distribution of water vapor above and below the inversion layer. Therefore, in the vertical dimension, the elevation and phase data are segmented using a cutoff height of 3000m for linear fitting. Figure 4 The red and green lines in (b) represent the best linear fit for these two height ranges. Figure 4 (c) shows the topographic-related atmospheric delay estimates for all 26 sub-regions in the same InSAR interferometric image.

[0101] V. Iterative Correction

[0102] Based on the refined tropospheric delay model constructed in each sub-block region, the parameters in the tropospheric delay model are optimized by iterative weighted least squares method, thereby solving for the optimal tropospheric delay phase value.

[0103] VI. Deformation Solution

[0104] By removing the tropospheric delay phase from the interferometric phase and calculating the corrected deformation rate using the singular value decomposition method, InSAR tropospheric delay correction based on DEM watershed segmentation is achieved, and the corrected deformation rate is used in performance evaluation.

[0105] VII. Evaluation of Correction Results

[0106] The corrected experimental results were compared with those of the traditional tropospheric delay correction model to evaluate its applicability and performance in high mountain and valley regions.

[0107] The corrected experimental results include: the corrected interferogram, the corrected interference phase, and the corrected deformation rate.

[0108] It should be noted that traditional tropospheric correction methods used for result evaluation include: global linear model (LM), regular window segmentation linear model (RSLM), and external dataset-assisted correction based on ERA-5 and GACOS.

[0109] Specifically, a comparison of the interferograms after correction using the InSAR tropospheric delay correction method based on DEM watershed segmentation and the traditional correction method is as follows: Figure 5 As shown, it is clear that among the comparisons of the interferograms corrected by various methods and the original interferograms without removing tropospheric delay, the method of the present invention has the best performance in correcting the tropospheric delay caused by the interference process. It removes the influence of tropospheric delay on the interferogram to the greatest extent and achieves the most effective tropospheric delay correction.

[0110] Further, the comparison of residual phase after correction by the InSAR tropospheric delay correction method based on DEM watershed segmentation of the present application and the traditional correction method is shown in FIG. 3. Figure 6 As shown in FIG. 3, the standard deviation of residual phase after correction by each tropospheric delay method is shown, which proves the superior performance of the method of the present application in the correction of tropospheric delay in the high mountain valley region.

[0111] The ratio of residual phase to elevation after correction by the InSAR tropospheric delay correction method based on DEM watershed segmentation of the present application and the traditional correction method is shown in FIG. 4. Figure 7 As can be seen from FIG. 4, the ratio of residual phase to elevation after correction by the method of the present application is the smallest, which reflects the correction ability of the correction method of the present application for the tropospheric delay coupled with the terrain, and proves the superiority of the method of the present application in the correction of tropospheric delay in the high mountain valley region with complex terrain and large height fluctuation.

[0112] The annual average deformation rate in the line-of-sight (LOS) direction before and after atmospheric correction by the InSAR tropospheric delay correction method based on DEM watershed segmentation of the present application and the traditional correction method is shown in FIG. 5. Figure 8 As shown in FIG. 5, in the southern valley area of the working region, the atmospheric delay seriously distorts the measured deformation. In the absence of atmospheric correction, the deformation result of this area is significantly affected by the atmospheric effect, and the result is unreliable. In contrast, the method of the present application effectively divides the study area into multiple corresponding sub-block regions based on watershed terrain, isolates and corrects the layered atmospheric delay, and has the best correction effect.

[0113] Figure 9 The statistical analysis result graph of time displacement sequence of the InSAR tropospheric delay correction method based on DEM watershed segmentation of the present application in a specific area compared with the traditional method; in order to evaluate the effectiveness of these tropospheric delay correction methods, the stable area and the deformation area in the study area are selected for statistical analysis, and the specific conditions are shown in FIG. 6. Figure 9 As shown in FIG. 6, P1 is the Zhenggang landslide area, located about 1 km downstream of the dam site of Gushui Hydropower Station on the right bank, and has obvious displacement signal, and P2 is the stable area, and no deformation is observed. The line-of-sight (LOS) time displacement of representative points P1 and P2 before and after atmospheric correction is analyzed. In these two areas, due to the tropospheric delay, the time displacement sequence before atmospheric correction shows obvious fluctuations. These fluctuations mask the deformation signals of the stable area and the deformation area. After atmospheric correction, the fluctuations are reduced to different degrees, revealing the true ground deformation signal. For the deformation area (P1) shown in (a) of FIG. 7, Figure 9 compared with the RSLM method, the method of the present application more effectively reduces the time variation, and accurately reconstructs the true ground deformation. Similarly, for the stable area (P2), as shown in FIG. 8, Figure 9As shown in (b), the method of the application has the best performance in effectively restoring the actual displacement and minimizing the time variation.

[0114] Based on the InSAR tropospheric delay correction method based on DEM watershed segmentation, the application further provides an InSAR tropospheric delay correction system based on DEM watershed segmentation.

[0115] As shown in Figure 10 The InSAR tropospheric delay correction system based on DEM watershed segmentation comprises:

[0116] An interference processing module is configured to obtain DEM data of a research area, precise orbit data, and time-series SAR images within a preset time range, and process the time-series SAR images, the DEM data, and the precise orbit data to obtain time-series InSAR interference images and deformation rates without removing the influence of tropospheric delay.

[0117] A region division module is configured to divide the research area into sub-blocks in two dimensions of vertical and horizontal directions according to the time-series InSAR interference images and watershed information and DEM data of the research area to obtain a plurality of sub-block regions.

[0118] A modeling module is configured to model the tropospheric delay of each sub-block region based on the time-series InSAR interference images to obtain a tropospheric delay model of each sub-block region in the time-series InSAR interference images.

[0119] An iterative optimization module is configured to iteratively optimize the tropospheric delay model of each sub-block region in the time-series InSAR interference images based on the deformation rates without removing the influence of tropospheric delay by using a weighted least square iterative method to obtain an iterated tropospheric delay phase of each sub-block region in the time-series InSAR interference images.

[0120] A deformation calculation module is configured to calculate the deformation rates without removing the influence of tropospheric delay according to the iterated tropospheric delay phase of each sub-block region in the time-series InSAR interference images.

[0121] The specific functions of each module of the InSAR tropospheric delay correction system based on DEM watershed segmentation are described in the InSAR tropospheric delay correction method based on DEM watershed segmentation, and will not be repeated here.

[0122] The application is based on an InSAR troposphere delay correction method and system based on DEM basin segmentation, considers the complex topography and climate conditions in the high mountain valley area, and corrects the troposphere delay in combination with vertical and horizontal atmospheric changes; in the horizontal dimension, the research area is divided into sub-areas by using the basin boundary, and is naturally aligned with the terrain and atmospheric model; in the vertical dimension, stratification caused by atmospheric inversion layer is included to improve the correction accuracy of the troposphere delay; by integrating the above horizontal and vertical spatial features, the application significantly improves the troposphere delay correction accuracy in the high mountain valley area with obvious terrain undulations and complex environment, provides a more applicable adaptive atmospheric delay correction for InSAR geological disaster monitoring, and improves the accuracy and application breadth of InSAR deformation monitoring.

[0123] The above description is only the preferred embodiment of the application, and is not intended to limit the application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the application shall be included in the protection scope of the application.

Claims

1. An InSAR tropospheric delay correction method based on DEM watershed segmentation, characterized in that, include: S1, acquire DEM data, precise orbit data and time-series SAR images within a preset time range for the study area, process the time-series SAR images, DEM data and precise orbit data to obtain time-series InSAR interferometric images and deformation rates without removing tropospheric delay effects; S2, based on the time-series InSAR interferometric image and the watershed information and DEM data of the study area, the study area is divided into sub-blocks in both vertical and horizontal dimensions to obtain multiple sub-block regions; S3. Based on the time-series InSAR interferometric image, tropospheric delay modeling is performed on each sub-block region to obtain the tropospheric delay model of each sub-block region in the time-series InSAR interferometric image. S4. Based on the deformation rate without removing the influence of tropospheric delay, the weighted least squares iterative method is used to iteratively optimize the tropospheric delay model of each sub-block region in the time-series InSAR interferometric image to obtain the iterative tropospheric delay phase of each sub-block region in the time-series InSAR interferometric image. S5. Based on the iterative tropospheric delay phase of each sub-block region in the time-series InSAR interferometric image, calculate the deformation rate after removing the influence of tropospheric delay.

2. The InSAR tropospheric delay correction method based on DEM watershed segmentation according to claim 1, characterized in that, Specifically, S1 is: S11, acquire DEM data, precise orbit data, and time-series SAR images of the study area within a preset time range; S12, the time-series SAR images are interferometric according to the pairing principle of SBAS-InSAR to obtain an initial time-series InSAR interferometric image; S13, using the DEM data and the precise orbit data, remove the redundant terrain phase and orbit errors in the initial time-series InSAR interferometric image to obtain the intermediate time-series InSAR interferometric image; S14, perform coherence point selection and filtering on the intermediate time-series InSAR interferometric image to remove noise and turbulence delay, and obtain a time-series InSAR interferometric image; S15, perform phase unwrapping on the time-series InSAR interferometric image to obtain the unwrapped interferometric phase without removing the tropospheric delay effect; S16. Singular value decomposition is performed on the unwrapped interference phase without removing the tropospheric delay effect to obtain the deformation rate without removing the tropospheric delay effect.

3. The InSAR tropospheric delay correction method based on DEM watershed segmentation according to claim 1, characterized in that, Specifically, S2 is: S21. Considering the influence of the inversion layer on the vertical distribution of atmospheric water vapor, the study area is vertically divided into blocks based on the DEM data corresponding to the coherent points in the time-series InSAR interferometric image, combined with the occurrence time and prior height of the inversion layer at the location of the study area. S22, Based on the watershed boundary segmentation strategy, the vertically segmented study area is horizontally segmented according to the watershed information of the study area to obtain multiple sub-block areas.

4. The InSAR tropospheric delay correction method based on DEM watershed segmentation according to claim 1, characterized in that, In S3, the tropospheric delay model for each sub-block region in the time-series InSAR interferometric image is as follows: in, K represents the tropospheric delay phase of the i-th sub-block region in the m-th InSAR interferometric image of the time series; m,i h represents the ratio of tropospheric delay phase to elevation in the i-th sub-block region of the m-th InSAR interferometric image; i Indicates the elevation of the i-th sub-block region; This represents the tropospheric delay phase intercept of the i-th sub-block region in the m-th InSAR interferometric image.

5. The InSAR tropospheric delay correction method based on DEM watershed segmentation according to claim 1, characterized in that, Specifically, S4 is: The temporal InSAR interferometric image is unwrapped to obtain the unwrapped interferometric phase without removing the tropospheric delay effect. The unwrapped interferometric phase without removing the tropospheric delay effect in each sub-region of the temporal InSAR interferometric image is used as the initial tropospheric delay phase of each sub-region of the temporal InSAR interferometric image. Based on the tropospheric delay model of each sub-region of the temporal InSAR interferometric image, the initial tropospheric delay phase and elevation of each sub-region of the temporal InSAR interferometric image are fitted by least squares to obtain the ratio of the initial tropospheric delay phase to the elevation and the initial tropospheric delay phase intercept of each sub-region of the temporal InSAR interferometric image. The initial tropospheric delay phase, the ratio of the initial tropospheric delay phase to the elevation, the intercept of the initial tropospheric delay phase, and the deformation rate without removing the influence of tropospheric delay in each sub-block region of the time-series InSAR interferometric image are substituted into the weighted least squares iterative algorithm model for iterative correction to obtain the iterative tropospheric delay phase of each sub-block region in the time-series InSAR interferometric image.

6. The InSAR tropospheric delay correction method based on DEM watershed segmentation according to claim 5, characterized in that, The weighted least squares iterative algorithm model is as follows: in, This represents the tropospheric delay phase of the i-th sub-block region in the m-th InSAR interferometric image of the time series in the (j+1)-th iteration. This represents the tropospheric delay phase of the i-th sub-block region in the m-th InSAR interferometric image during the j-th iteration. The initial value of the iteration number j is 0; when j = 0, W represents the initial tropospheric delay phase of the i-th sub-block region in the temporal InSAR interferometric image; m,i,j Let W represent the weight exponent of the i-th sub-block region in the m-th InSAR interferometric image during the j-th iteration, and when j = 0, W m,i,0 Calculated from the deformation rate without removing the effects of tropospheric delay; h i K represents the elevation of the i-th sub-block region; m,i,j This represents the ratio of the tropospheric delay phase to the elevation of the i-th sub-block region in the m-th InSAR interferometric image during the j-th iteration. K represents the tropospheric delay phase intercept of the i-th sub-block region in the m-th InSAR interferometric image during the j-th iteration. m,i,j and The tropospheric delay phase and elevation of the i-th sub-block region in the m-th InSAR interferometric image during the j-th iteration are obtained through least-squares fitting, and when j = 0, K m,i,0 and These are the ratio of the initial tropospheric delay phase to the elevation and the initial tropospheric delay phase intercept, respectively, for the i-th sub-block region in the m-th InSAR interferometric image.

7. The InSAR tropospheric delay correction method based on DEM watershed segmentation according to claim 6, characterized in that, The formula for calculating the weighting index is as follows: Among them, v m,i,j v represents the deformation rate of the i-th sub-block region in the m-th InSAR interferometric image during the j-th iteration. m,i,j It is calculated from the tropospheric delay phase of the i-th sub-block region in the m-th InSAR interferometric image during the j-th iteration; Let represent the deformation rate threshold of the i-th sub-block region in the m-th InSAR interferometric image during the j-th iteration, and σ m,i,j It represents the standard deviation of the deformation rate of the i-th sub-block region in the m-th InSAR interferometric image during the j-th iteration.

8. The InSAR tropospheric delay correction method based on DEM watershed segmentation according to claim 6, characterized in that, The iteration stopping condition of the weighted least squares iterative algorithm model is: Where ||·|| represents the Euclidean norm, ε represents the prior parameter, and n m,i This represents the number of coherent points in the i-th sub-block region of the m-th InSAR interferometric image.

9. The InSAR tropospheric delay correction method based on DEM watershed segmentation according to claim 1, characterized in that, Specifically, S5 is: By integrating the iterative tropospheric delay phases of each sub-block region in the temporal InSAR interferometric image, the optimized temporal tropospheric delay phase of the study area is obtained. Phase unwrapping is performed on the time-series InSAR interferometric image to obtain the unwrapped interferometric phase without removing the tropospheric delay effect. The optimized time-series tropospheric delay phase is subtracted from the unwrapped interferometric phase without removing the tropospheric delay effect to obtain the deformation phase with the tropospheric delay effect removed. Singular value decomposition is performed on the deformation phase after removing the tropospheric delay effect to obtain the deformation rate after removing the tropospheric delay effect.

10. An InSAR tropospheric delay correction system based on DEM watershed segmentation, characterized in that, include: An interferometric processing module is used to acquire DEM data, precise orbit data, and time-series SAR images within a preset time range of the study area, and to process the time-series SAR images, DEM data, and precise orbit data to obtain a time-series InSAR interferometric image and a deformation rate without removing the effects of tropospheric delay. The region division module is used to divide the study area into sub-blocks in both vertical and horizontal dimensions based on the watershed information and DEM data of the study area, thereby obtaining multiple sub-block regions. The modeling module is used to perform tropospheric delay modeling on each sub-block region based on the time-series InSAR interferometric image, so as to obtain the tropospheric delay model of each sub-block region in the time-series InSAR interferometric image. The iterative optimization module is used to iteratively optimize the tropospheric delay model of each sub-block region in the time-series InSAR interferometric image based on the deformation rate without removing the influence of tropospheric delay, using a weighted least squares iterative method, to obtain the iterative tropospheric delay phase of each sub-block region in the time-series InSAR interferometric image. The deformation calculation module is used to calculate the deformation rate after removing the influence of tropospheric delay based on the iterative tropospheric delay phase of each sub-block region in the time-series InSAR interferometric image.

Images