A MTInSAR landslide monitoring method, device and storage medium

Abstract

The present application relates to the field of geological disaster prevention, and discloses a MTInSAR landslide monitoring method, device and storage medium, including: obtaining SAR image data of the landslide area to be monitored, and preprocessing the image data; establishing a PS monitoring network and a DS monitoring network according to the processed image data; using the PS monitoring network and the DS monitoring network to obtain the MTInSAR observation data of the landslide area to be monitored, including LOS deformation observation values ​​and slope elevation information; using independent component analysis and periodic analysis methods to identify the atmospheric delay component in the LOS deformation observation value; separating the atmospheric delay component from the LOS deformation observation value to obtain the atmospherically corrected LOS deformation data; using terrain and satellite orbit parameters, simulating the LOS deformation data as deformation along the slope direction. In this way, it does not rely on external weather models or meteorological data, can effectively complete the atmospheric automatic correction with high temporal and spatial resolution, improve the accuracy of MTInSAR slope monitoring, and contribute to landslide early warning and disaster prevention.

Description

Technical Field

[0001] The present invention relates to the field of geological disaster prevention, and in particular to a MTInSAR landslide monitoring method, equipment and storage medium. Background Art

[0002] Landslide is one of the major geological disasters in my country, especially in hilly cities such as Chongqing and Hong Kong. Frequent landslides and debris flows seriously affect the surrounding densely packed buildings, public infrastructure and the personal safety of residents. Regular slope stability monitoring is extremely important for preventing and controlling landslide disasters, but traditional ground measurement methods are expensive and lack monitoring efficiency and spatial detail reflection capabilities. As a newly developed remote sensing technology, Synthetic Aperture Radar Interferometry (InSAR) technology overcomes the limitations of ground instruments and can provide low-cost, high temporal and spatial resolution (1 meter in space, two weeks in time), high-precision (millimeter level) all-weather ground deformation monitoring methods.

[0003] In recent years, the Multi-Temporal InSAR (MTInSAR) technology has been developed and expanded, and has been maturely applied in the fields of geological disaster prevention, such as urban settlement, mining area deformation monitoring, and landslide monitoring. However, for coastal mountainous areas with multiple vegetation coverage and obvious atmospheric disturbances (mainly affected by atmospheric water vapor changes) (such as the Guangdong-Hong Kong-Macao Greater Bay Area in my country), the traditional MTInSAR processing method still has technical defects and deficiencies, and cannot accurately measure the deformation of the slope, which brings difficulties to the landslide stability analysis. On the one hand, the multiple vegetation coverage makes the radar decoherence on the slope surface serious, which directly affects the extraction of effective interference phase information; on the other hand, the atmospheric disturbance around the slope makes the MTInSAR measurement value have measurement errors caused by the atmospheric delay of the radar signal, which directly affects the accurate estimation of the slope deformation. The MTInSAR atmospheric disturbance correction can be achieved by using an external weather model, but the spatial resolution of the external weather model is very low, and the vertical atmospheric stratification of the weather model is not fine enough, which is not suitable for accurate atmospheric correction of the slope.

[0004] Therefore, how to solve the problem of phase decoherence and the influence of atmospheric disturbance in MTInSAR landslide monitoring is a technical problem that needs to be solved urgently by technical personnel in this field. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a MTInSAR landslide monitoring method, device and storage medium, which can improve the phase quality of multi-vegetation coverage areas, effectively complete the automatic atmospheric correction with high temporal and spatial resolution, and thus improve the MTInSAR slope monitoring accuracy. The specific scheme is as follows:

[0006] A MTInSAR landslide monitoring method, comprising:

[0007] Acquiring SAR image data of the landslide area to be monitored, and preprocessing the SAR image data;

[0008] Establishing a PS monitoring network and a DS monitoring network according to the processed SAR image data;

[0009] The PS monitoring network and the DS monitoring network are used to obtain the MTInSAR observation data of the landslide area to be monitored; the MTInSAR observation data includes LOS deformation observation values ​​and slope elevation information;

[0010] Independent component analysis and periodic analysis methods are used to identify the atmospheric delay components in the LOS deformation observations;

[0011] Separating the atmospheric delay component from the LOS deformation observation value to obtain LOS deformation data after atmospheric correction;

[0012] The LOS deformation data is simulated as deformation along the slope direction by using terrain data and satellite orbit parameters.

[0013] Preferably, in the MTInSAR landslide monitoring method provided in the embodiment of the present invention, an independent component analysis and a periodicity analysis method are used to identify the atmospheric delay component in the LOS deformation observation value, including:

[0014] Performing independent component analysis on the LOS deformation observations to obtain the spatial response and time series of each independent component;

[0015] Performing spatial correlation analysis and periodic spectrum analysis on the slope elevation information and the spatial responses and time series of each independent component obtained, to obtain the spatial correlation factors of each independent component and the periodicity of its time series;

[0016] The independent component that satisfies both the maximum spatial correlation factor and the most obvious time series periodicity is determined to be the atmospheric delay component in the LOS deformation observation value.

[0017] Preferably, in the MTInSAR landslide monitoring method provided in the embodiment of the present invention, independent component analysis is performed on the LOS deformation observation value to obtain the spatial response and time series of each independent component, including:

[0018] Generate a data matrix according to the LOS deformation observation value;

[0019] Determine the linear separation matrix corresponding to the data matrix by an iterative method;

[0020] According to the linear separation matrix and the data matrix, the spatial response and time series of each independent component are obtained.

[0021] Preferably, in the MTInSAR landslide monitoring method provided in the embodiment of the present invention, spatial correlation analysis and periodic spectrum analysis are performed on the slope elevation information and the spatial response and time series of each independent component obtained to obtain the spatial correlation factor of each independent component and the periodicity of its time series, including:

[0022] Calculate the spatial correlation factor of each independent component according to the slope elevation information and the mean of the elevation information, the obtained spatial response of each independent component and the mean of the spatial response;

[0023] The periodic spectrum of the time series of each independent component is analyzed by Fourier transform to obtain the periodicity of the time series of each independent component.

[0024] Preferably, in the above-mentioned MTInSAR landslide monitoring method provided by the embodiment of the present invention, before using independent component analysis and periodicity analysis methods to identify the atmospheric delay component in the LOS deformation observation value, it also includes:

[0025] The principal component analysis is performed on the LOS deformation observations, and only a set number of main components constituting the LOS deformation observation data are retained. The set number of main components should explain more than 85% of the data variance to achieve data dimensionality reduction.

[0026] Preferably, in the MTInSAR landslide monitoring method provided in the embodiment of the present invention, establishing a PS monitoring network includes:

[0027] Selecting PS candidate points according to the amplitude deviation index of the processed SAR image data;

[0028] All the selected PS candidate points are connected using the DeLorean triangulation network to build a PS monitoring network;

[0029] The deformation parameters and elevation parameters of the arc segments of the PS monitoring network are solved by using a robust estimation method;

[0030] According to the deformation parameters and elevation parameters of the arc segments of the PS monitoring network, the PS monitoring network is adjusted as a whole, and a PS candidate point is selected as a deformation monitoring reference point;

[0031] The elevation information and deformation time series of each PS candidate point relative to the deformation monitoring reference point are obtained.

[0032] Preferably, in the above-mentioned MTInSAR landslide monitoring method provided by the embodiment of the present invention, establishing a DS monitoring network includes:

[0033] Clustering homogeneous points using the terrain element information of the landslide area to be monitored;

[0034] Reconstructing the interference phase time series using the coherence matrix of the homogeneous points;

[0035] Based on the interferogram after phase reconstruction, and according to the threshold of the number of homogeneous points or the threshold of phase coherence, DS candidate points are selected;

[0036] Connect each DS candidate point with its nearest PS candidate point to form a DS monitoring network;

[0037] A robust estimation method is used to solve the deformation parameters and elevation parameters of the arc segments of the DS monitoring network;

[0038] The deformation parameters of the DS monitoring network arc segment are added to the deformation parameters of the connected PS candidate points to obtain the deformation value and elevation value of each DS candidate point.

[0039] Preferably, in the MTInSAR landslide monitoring method provided in the embodiment of the present invention, clustering homogeneous points using terrain element information includes:

[0040] The slope and slope angle values ​​of the landslide area to be monitored are calculated using a digital elevation model (DEM) having a spatial resolution superior to or equivalent to that of SAR images;

[0041] In a filtering window, pixels with similar slope and slope angle are selected as homogeneous points of the center pixel of the filtering window.

[0042] An embodiment of the present invention further provides an MTInSAR landslide monitoring device, comprising a processor and a memory, wherein the processor implements the above-mentioned MTInSAR landslide monitoring method provided in the embodiment of the present invention when executing the computer program stored in the memory.

[0043] An embodiment of the present invention further provides a computer-readable storage medium for storing a computer program, wherein when the computer program is executed by a processor, the MTInSAR landslide monitoring method as provided in the embodiment of the present invention is implemented.

[0044] It can be seen from the above technical scheme that the MTInSAR landslide monitoring method provided by the present invention includes: obtaining SAR image data of the landslide area to be monitored, and preprocessing the SAR image data; establishing a PS monitoring network and a DS monitoring network according to the processed SAR image data; using the PS monitoring network and the DS monitoring network to obtain MTInSAR observation data of the landslide area to be monitored; the MTInSAR observation data includes LOS deformation observation values ​​and slope elevation information; using independent component analysis and periodic analysis methods to identify the atmospheric delay component in the LOS deformation observation values; separating the atmospheric delay component from the LOS deformation observation values ​​to obtain LOS deformation data after atmospheric correction; using terrain data and satellite orbit parameters, simulating the LOS deformation data as deformation along the slope direction.

[0045] The above-mentioned MTInSAR landslide monitoring method provided by the present invention obtains LOS deformation observation values ​​by establishing PS monitoring network and DS monitoring network, which can increase the number of MTInSAR measurement points in areas with multiple vegetation coverage, and removes the atmospheric delay components in the LOS deformation observation values ​​based on independent component analysis and periodic analysis methods, and finally simulates the LOS deformation to the deformation along the slope direction. The whole process does not rely on external weather models or meteorological data, can effectively complete automatic atmospheric correction with high temporal and spatial resolution, improve the MTInSAR slope monitoring accuracy, and contribute to the implementation of landslide early warning and disaster prevention measures.

[0046] In addition, the present invention also provides corresponding equipment and computer-readable storage medium for the MTInSAR landslide monitoring method, which further makes the above method more practical. The equipment and computer-readable storage medium have corresponding advantages. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0048] Figure 1 A flow chart of the MTInSAR landslide monitoring method provided by an embodiment of the present invention;

[0049] Figure 2 A specific flow chart of the MTInSAR landslide monitoring method provided by an embodiment of the present invention;

[0050] Figure 3 and Figure 4They are respectively a set of spatial correlation factors of independent components of LOS deformation data and time series periodicity analysis results provided by embodiments of the present invention;

[0051] Figure 5 and Figure 6 They are respectively another set of spatial correlation factors and time series periodicity analysis results of independent components of LOS deformation data provided by an embodiment of the present invention;

[0052] Figure 7 A schematic diagram of the LOS deformation sensitivity factor of the landslide area to be monitored calculated from satellite orbit parameters and elevation model DEM provided in an embodiment of the present invention;

[0053] Figure 8a A schematic diagram of the relative change in time of the tropospheric delay in the landslide area to be monitored provided by an embodiment of the present invention;

[0054] Figure 8b A schematic diagram of the relative change of the process delay in space in the landslide area to be monitored provided by an embodiment of the present invention;

[0055] Figure 9a A three-dimensional display diagram of the slope deformation monitoring results of the landslide area to be monitored from December 2015 to May 2016 provided in an embodiment of the present invention;

[0056] Figure 9b A three-dimensional display diagram of the slope deformation monitoring results of the landslide area to be monitored from May 2016 to November 2016 provided in an embodiment of the present invention;

[0057] Fig.9c A schematic diagram showing the comparison between the average displacement of all MTInSAR measurement points on a slope and the annual accumulated rainfall provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] The present invention provides a MTInSAR landslide monitoring method, such as Figure 1 As shown, the following steps are included:

[0060] S101, obtaining SAR (Synthetic Aperture Radar) image data of the landslide area to be monitored, and preprocessing the SAR image data;

[0061] In specific implementation, Figure 2 As shown in the figure, the preprocessing of SAR image data may include: importing and registering SAR image data; using the interferometric module to perform interferometric processing, deflating phase processing and differential processing on the registered SAR image data. After registration, in order to ensure the spatial details of MTInSAR monitoring, the main image can be selected and the interferogram can be generated based on the full resolution.

[0062] S102, establishing a PS monitoring network and a DS monitoring network according to the processed SAR image data;

[0063] It should be noted that step S102 integrates permanent scatterers (PS) and distributed scatterers (DS) to establish a double-layer monitoring network of PS and DS. The DS monitoring network is based on the interference pattern after phase spatiotemporal filtering (to improve phase coherence). Since the surface coverage in the landslide area is mostly vegetation, bare soil, etc., this type of surface is mostly manifested as a distributed scatterer DS with medium / low coherence during MTInSAR analysis, or even completely lost coherence. The deformation signal of this type of area cannot be effectively restored by using only the PS solution method. Therefore, the DS network monitoring in the present invention is intended to restore useful deformation signals in medium / low coherence areas.

[0064] S103, using the PS monitoring network and the DS monitoring network to obtain MTInSAR observation data of the landslide area to be monitored; the MTInSAR observation data includes LOS deformation observation values ​​and slope elevation information;

[0065] It should be noted that the output information obtained by the PS and DS monitoring networks includes the LOS (Line-of-sight, satellite line of sight) deformation observation values ​​of the landslide area and the slope elevation information.

[0066] S104, using independent component analysis and periodicity analysis methods to identify atmospheric delay components in LOS deformation observations;

[0067] It should be noted that due to the influence of the tropospheric atmosphere, MTInSAR often contains atmospheric delay errors when monitoring slope deformation, and the atmospheric influence varies with the ground elevation, so it is more obvious in mountainous areas with undulating elevations. Tropospheric atmospheric delay is spatially related to elevation and is mostly seasonal in time, which is one of the main errors affecting the accuracy of landslide deformation monitoring. In the present invention, the atmospheric delay component in the LOS deformation observation value is automatically identified using independent component analysis and periodicity analysis methods.

[0068] S105, separating the atmospheric delay component from the LOS deformation observation value to obtain the LOS deformation data after atmospheric correction;

[0069] S106, using terrain data and satellite orbit parameters, simulating the LOS deformation data as deformation along the slope direction.

[0070] In the above-mentioned MTInSAR landslide monitoring method provided in an embodiment of the present invention, the LOS deformation observation values ​​are obtained by establishing a PS monitoring network and a DS monitoring network, which can increase the number of MTInSAR measurement points in areas with multiple vegetation coverage, and remove the atmospheric delay components in the LOS deformation observation values ​​based on independent component analysis and periodic analysis methods. Finally, the LOS deformation is simulated to the deformation along the slope direction. The whole process does not rely on external weather models or meteorological data, and can effectively complete automatic atmospheric correction with high temporal and spatial resolution, improve the MTInSAR slope monitoring accuracy, and contribute to the implementation of landslide early warning and disaster prevention measures.

[0071] Furthermore, in the specific implementation, in the above MTInSAR landslide monitoring method provided in the embodiment of the present invention, if Figure 2 As shown, step S102 establishes a PS monitoring network, which may specifically include: first, selecting PS candidate points according to the amplitude dispersion index (ADI) of the intensity information of the processed SAR image data; then, connecting all the selected PS candidate points using the DeLorean triangulation network to construct a PS monitoring network; using a robust estimation method (Beamforming) to solve the deformation parameters and elevation parameters of the arc segments of the PS monitoring network; after all arc segment parameters are solved, the PS monitoring network is overall adjusted according to the deformation parameters and elevation parameters of the arc segments of the PS monitoring network, and a PS candidate point in the landslide area to be monitored (generally a PS point located on a stable surface) is selected as a deformation monitoring reference point; and the elevation information and deformation time series of each PS candidate point relative to the deformation monitoring reference point are obtained.

[0072] In practical applications, PS candidate points are mostly phase-stable points (such as buildings, roads, and exposed rocks around the landslide area).

[0073] Furthermore, in the specific implementation, in the above MTInSAR landslide monitoring method provided in the embodiment of the present invention, if Figure 2As shown, step S102 establishes a DS monitoring network, which may specifically include: first, clustering homogeneous points using the terrain element information of the landslide area to be monitored; then, reconstructing the interference phase time series using the coherence matrix of the homogeneous points; selecting DS candidate points based on the interference pattern after phase reconstruction and according to the homogeneous point quantity threshold or phase coherence threshold; then, connecting each DS candidate point with its nearest PS candidate point to form a DS monitoring network; using a robust estimation method to solve the deformation parameters and elevation parameters of the DS monitoring network arc segment; finally, adding the deformation parameters of the DS monitoring network arc segment to the deformation parameters of the connected PS candidate point to obtain the deformation value and elevation value of each DS candidate point.

[0074] like Figure 2 As shown in the figure, DS processing first requires spatiotemporal homogeneity filtering of the interference pattern (homogeneous filtering includes two processes: homogeneous pixel clustering and phase reconstruction). Due to the poor quality of SAR images in landslide areas, the present invention uses terrain element information to cluster homogeneous points. After the homogeneous point clustering is completed, the coherence matrix of the homogeneous points is used to reconstruct the interference phase time series, and the phase reconstruction is achieved by the coherence weighted Phase-Linking method. The phase quality of the interference pattern after homogeneous filtering will be improved, which is helpful for the deformation parameter estimation of the subsequent DS arc segment. The selection of DS candidate points can be completed by setting a threshold value for the number of homogeneous points or a phase coherence threshold. The expansion of the DS monitoring network is established on the PS reference network, that is, the DS candidate points are connected to their nearest PS reference points to form a DS arc segment, and the deformation and elevation parameter solution of the DS arc segment is also based on a robust estimation method. Finally, the deformation parameters obtained by the DS arc segment plus the deformation parameters of the connected PS reference point are the deformation values ​​of the DS point. At this point, the DS extended network is completed.

[0075] The present invention introduces interferometric phase spatiotemporal filtering based on terrain data, which can further improve the phase quality of multi-vegetation coverage areas and reduce the impact of phase decoherence.

[0076] In specific implementation, in the above-mentioned MTInSAR landslide monitoring method provided in an embodiment of the present invention, the above-mentioned steps use terrain element information to cluster homogeneous points, which can specifically include: first, using a digital elevation model (DEM) that is better than or equivalent to the spatial resolution of the SAR image to calculate the slope and slope angle values ​​of the landslide area to be monitored; then, within a filtering window, select pixel points with similar slopes and slope angles as homogeneous points of the central pixel of the filtering window.

[0077] Furthermore, in the specific implementation, in the above MTInSAR landslide monitoring method provided in the embodiment of the present invention, if Figure 2As shown, step S104 uses independent component analysis and periodicity analysis methods to identify the atmospheric delay components in the LOS deformation observations, which may specifically include: first, performing independent component analysis on the LOS deformation observations to obtain the spatial response and time series of each independent component; then, performing spatial correlation analysis and periodic spectrum analysis on the slope elevation information and the obtained spatial responses and time series of each independent component to obtain the spatial correlation factors of each independent component and the periodicity of its time series; finally, the independent component that satisfies both the maximum spatial correlation factor and the most obvious time series periodicity is determined to be the atmospheric delay component in the LOS deformation observations.

[0078] It is understandable that the slope LOS deformation obtained by MTInSAR includes signal sources such as real deformation, atmospheric disturbance, systematic error, and random error. Among them, atmospheric disturbance is the error source with the greatest impact on deformation estimation in LOS measurement. In traditional atmospheric correction methods, external weather models are often used to obtain atmospheric delay signals in the landslide area to be monitored. On the one hand, the spatial resolution of external weather models is very limited, and most of them are interpolated from sparse ground meteorological stations or GPS (Global Positioning System) networks; on the other hand, the atmospheric stratification details of external weather models in the vertical direction are not high. In the vertical direction, most atmospheric delay products are interpolated from DEM according to a certain atmosphere-elevation relationship. These are not conducive to accurate atmospheric delay correction during MTInSAR landslide monitoring, especially when the slope deformation is not large, the deformation signal is more likely to be submerged in the atmospheric delay, which brings difficulties to the interpretation of slope deformation. Therefore, in the present invention, independent component analysis, spatial correlation analysis and periodic spectrum analysis are used to obtain the spatial correlation factors of each independent component and its time series periodicity, and to find the independent component that satisfies both the maximum spatial correlation factor and the most obvious time series periodicity as the atmospheric delay component in the LOS deformation observation value, so as to obtain an atmospheric delay signal with higher spatial resolution.

[0079] In specific implementation, in the above-mentioned MTInSAR landslide monitoring method provided by an embodiment of the present invention, the above-mentioned steps perform independent component analysis on the LOS deformation observation values ​​to obtain the spatial response and time series of each independent component, which may specifically include: first, generating a data matrix based on the LOS deformation observation values; then, using an iterative method to determine the linear separation matrix corresponding to the data matrix; finally, obtaining the spatial response and time series of each independent component based on the linear separation matrix and the data matrix.

[0080] Specifically, the LOS deformation observation data is regarded as a blind source separation (BBS) containing various signals. Therefore, the extraction and correction of atmospheric delay is a blind source signal separation problem.n×m is the data matrix generated by all LOS observations, where n represents the number of PS and DS obtained by monitoring (spatial dimension), and m is the number of SAR images (temporal dimension). The independent component signal separation can be expressed as:

[0081] LOS n×m =S n×k ·T k×m (1)

[0082] Among them, S n×k =[S1, S2, ..., S k ] is an unknown matrix, representing the spatial response of the separated k independent components (signal sources); specifically, S n×k The matrix consists of elements s ij (1≤i≤n, 1≤j≤k), s ij is the contribution of the jth independent component (signal source) at spatial position i (the spatial position is the position of the PS and DS points); T k×m It is also an unknown matrix, representing the specific performance of k independent components in time. For example, the time series of the jth independent component is [t j1 , ..., t jm ]. Since two unknown matrix variables cannot be obtained simultaneously when solving equation (1), when implementing independent component analysis, we first need to determine a linear separation (projection) matrix W. k×n , W k×n The original LOS n×m is converted into a set of independent variables, namely

[0083] Y k×m =W k×n ·LOS n×m (2)

[0084] Among them, Y k×m Includes the independent variable y j =[y j1 , ..., y jm ], 1≤j≤k, and Y k×m In fact, it represents T in formula (1) k×m .W k×n =[w1, w2, ..., w k ] T The solution can be achieved by iterative method, which is specifically implemented as follows: for each vector w i , i = 1, 2, ..., k, the iteration starts with a random vector And follow the steps below to find w i :

[0085] first step:

[0086] Step 2:

[0087] Among them, E represents the mathematical expectation, and X is the LOS after whitening. n×m matrix, G is the contrast function, G′ and G″ represent the first and second order derivatives of the contrast function respectively, and T is the transpose. The contrast function G is generally set as:

[0088]

[0089]

[0090]

[0091] Among them, 1≤a1≤2 and α2≈1.

[0092] Step 3:

[0093] Among them, w j , j = 1, ..., i-1 represents w obtained by the previous iteration;

[0094] Step 4:

[0095] Step 5: If w i Convergence, return to the first step, and i = i + 1,, i ≤ k, if not converged, return to the second step. The maximum number of iterations and convergence conditions can be set according to actual conditions. i After that, W k×n =[w1, w2, …, w k ] T Once W k×n Obtain, S in formula (1) n×k (W k×n Therefore, the spatial response (i.e., S in equation (1)) and time series (i.e., T in equation (1)) of each independent component (signal source) are also obtained accordingly.

[0096] Further, in the specific implementation, in the above-mentioned MTInSAR landslide monitoring method provided by the embodiment of the present invention, before executing step S104 to identify the atmospheric delay component in the LOS deformation observation value by using the independent component analysis and periodic analysis method, it can also include: performing principal component analysis on the LOS deformation observation value, and only retaining the set number of main components constituting the LOS deformation data to be corrected by the atmosphere. That is to say, before performing the independent component analysis, the LOS deformation observation data can be first subjected to principal component analysis. The purpose of the principal component analysis is to select the most important signal sources (i.e., the components with the largest data variance) in the LOS deformation observation data. Preferably, only the first few main components can be retained, and these main components explain 85% or more of the data variance, so as to achieve the purpose of data dimensionality reduction. The principal component analysis dimensionality reduction can help determine how many independent components are most appropriate to be retained in the final independent component analysis, i.e., determine the k value in formula (1).

[0097] In addition, in the specific implementation, in the above-mentioned MTInSAR landslide monitoring method provided by the embodiment of the present invention, the above-mentioned steps perform spatial correlation analysis and periodic spectrum analysis on the slope elevation information and the obtained spatial response and time series of each independent component to obtain the spatial correlation factor of each independent component and its time series periodicity, which may specifically include: calculating the spatial correlation factor of each independent component based on the slope elevation information and the mean of the elevation information, the obtained spatial response of each independent component and the mean of the spatial response; performing periodic spectrum analysis on the time series of each independent component using Fourier transform to obtain the time series periodicity of each independent component.

[0098] Specifically, since the atmospheric delay signal has a strong correlation with the terrain elevation, the present invention proposes to use the spatial correlation factor ρ to calculate the correlation degree between each independent component and the terrain elevation, that is:

[0099]

[0100] Where ρ(j) is the spatial correlation factor of the jth independent component (signal source) (j=1, .., k); S j is the spatial response of the jth independent component, i (i = 1…, n) represents different PS and DS pixels; H is the terrain elevation information output by MTInSAR; and Represent their respective means. The higher the value of ρ∈[0,1], the higher the correlation between the signal source and the terrain elevation. In addition, the atmospheric delay signal is affected by seasonal atmospheric changes and presents a periodic pattern in time. Therefore, the present invention uses Fourier transform to perform spectral analysis on the time series of each independent component to solve its periodic characteristics, that is:

[0101]

[0102] Where P(j) is the power spectrum of the jth (j=1, .., k) independent component time series; F is the Fourier transform; () * represents complex conjugate; L is T j The length of the vector, where T j It has been uniformly time interpolated before power spectrum analysis. The periodicity of each independent component can be obtained through spectrum analysis. Generally, the atmospheric delay signal has the same periodicity as the local water pressure, temperature, etc. Figures 3 to 6 The spatial correlation factors of the independent components of the two sets of LOS data and the time series periodicity analysis results are shown, among which the atmospheric delay signal shows the highest spatial elevation correlation and obvious age periodicity.

[0103] The above steps calculate the spatial correlation factor of each independent component (signal source) and its time series periodicity. Based on the characteristics of the atmospheric delay signal, the present invention determines that the independent component that satisfies both the maximum correlation factor and the most obvious periodicity (the periodicity is consistent with the local water vapor pressure, temperature, etc., generally the age periodicity) is the atmospheric delay term contained in the LOS deformation observation value. After determining the atmospheric delay term, it is necessary to remove this atmospheric delay from the original LOS deformation observation value to obtain the LOS deformation data after atmospheric correction. For example, it is determined that the tth (1≤t≤k) independent component among the k independent components is the atmospheric delay signal, then the atmospheric signal separation can be performed by LOS n×m -S t ·T t Implementation, where S t is S in formula (1) n×k The tth column, T t is T in formula (1) k×m At this point, the atmospheric delay correction of the LOS deformation observations obtained by MTInSAR monitoring is completed and does not depend on external weather models or meteorological data.

[0104] It should be noted that the slope deformation obtained by MTInSAR is in the satellite line of sight (LOS), which is not the traditional three-dimensional (north, east, sky) coordinate deformation, nor the deformation along the slope direction that is more concerned in engineering geology. Figure 2 As shown, the LOS deformation observation value after the atmospheric delay correction in step S105 can be simulated to the deformation value along the maximum slope gradient direction. The deformation simulation is completed by the following projection formula:

[0105]

[0106] Among them, S LOS is the LOS deformation sensitivity factor, which means that one unit of slope sliding (positive along the slope downward and negative along the slope upward) corresponds to SLOS The LOS displacement of the unit (a positive LOS displacement indicates movement toward the satellite, and a negative LOS displacement indicates movement away from the satellite); θ and are the satellite incidence angle and heading angle respectively; α and σ are the ground slope angle and aspect angle respectively. The slope and aspect information are obtained from the ground elevation model (DEM). Figure 7 The calculated S of the landslide area to be monitored is shown. LOS . Divide the LOS deformation observation after removing the atmospheric delay in step 5 by S LOS The deformation along the slope direction can be obtained, and the MTInSAR slope deformation monitoring is now complete. The slope deformation data can be used to verify the ground monitoring data and conduct further slope stability analysis.

[0107] The following takes the case where the landslide area to be monitored is a slope at Lantau Island in Hong Kong as an example to illustrate the MTInSAR landslide monitoring method provided by the embodiment of the present invention:

[0108] Specifically, 63 high-resolution CosmoSkyMed strip mode SAR images were used for single-view interferometry processing, with a time span from October 2013 to July 2019. The monitoring method of the present invention successfully obtained detailed slope deformation information of the landslide area to be monitored, and successfully separated the atmospheric disturbance from the original MTInSAR measurement value. The atmospheric delay information obtained from the experimental results was verified with the external atmospheric product European Centre for Medium-Range Weather Forecasts (ECMWF) ERA5, and the slope deformation information after atmospheric correction was also verified with ground leveling data. The results show that after the atmospheric delay is corrected by the present invention, the monitoring accuracy of the slope deformation in the landslide area to be monitored is improved by about 42%, and the deformation time series accuracy reaches the millimeter level. Figure 8a and Figure 8b The relative temporal and spatial changes of the tropospheric delay in the landslide area to be monitored obtained by the present invention are shown, wherein Figure 8a The relative change of tropospheric delay in time in the landslide area to be monitored is shown. Figure 8b The relative spatial variation of the tropospheric delay (normalized) in the landslide area to be monitored is shown. Figure 9a and Figure 9b It shows the slope deformation information of the landslide area to be monitored in different periods obtained by the present invention, and the positive value represents sliding down the slope; wherein, Figure 9a This is a three-dimensional display of the slope deformation monitoring results of the landslide area to be monitored from December 2015 to May 2016. Figure 9b This is a three-dimensional display of the slope deformation monitoring results of the landslide area to be monitored from May 2016 to November 2016. Fig.9cThe comparison between the average displacement of all MTInSAR measurement points on the slope and the annual accumulated rainfall is shown. The gray shading represents the standard deviation of all measurement points, and positive values ​​represent sliding down the slope.

[0109] Based on the same inventive concept, an embodiment of the present invention further provides a MTInSAR landslide monitoring device. Since the principle of solving the problem by the device is similar to that of the aforementioned MTInSAR landslide monitoring method, the implementation of the device can refer to the implementation of the MTInSAR landslide monitoring method, and the repeated parts will not be repeated.

[0110] In a specific implementation, the MTInSAR landslide monitoring device provided by the embodiment of the present invention specifically includes:

[0111] An image processing module is used to obtain SAR image data of the landslide area to be monitored and pre-process the SAR image data;

[0112] The network establishment module is used to establish the PS monitoring network and the DS monitoring network according to the processed SAR image data;

[0113] MTInSAR observation data acquisition module, used to obtain MTInSAR observation data of the landslide area to be monitored using the PS monitoring network and the DS monitoring network; MTInSAR observation data includes LOS deformation observation values ​​and slope elevation information;

[0114] The atmospheric delay component identification module is used to identify the atmospheric delay components in the LOS deformation observations using independent component analysis and periodic analysis methods;

[0115] The atmospheric delay component separation module is used to separate the atmospheric delay component from the LOS deformation observation value to obtain the LOS deformation observation data after atmospheric correction;

[0116] The deformation simulation module is used to simulate the LOS deformation data into deformation along the slope direction by using terrain data and satellite orbit parameters.

[0117] In the above-mentioned MTInSAR landslide monitoring device provided in the embodiment of the present invention, the number of MTInSAR monitoring points in multiple vegetation coverage areas can be increased through the interaction of the above-mentioned five modules, without relying on external weather models or meteorological data, and effectively completing high temporal and spatial resolution atmospheric automatic correction, thereby improving the MTInSAR slope monitoring accuracy and facilitating the implementation of landslide warning and disaster prevention measures.

[0118] For more specific working processes of the above modules, please refer to the corresponding contents disclosed in the above embodiments, which will not be repeated here.

[0119] Correspondingly, an embodiment of the present invention further discloses an MTInSAR landslide monitoring device, including a processor and a memory; wherein the processor implements the MTInSAR landslide monitoring method disclosed in the above embodiment when executing the computer program stored in the memory.

[0120] For more specific processes of the above method, please refer to the corresponding contents disclosed in the aforementioned embodiments, which will not be repeated here.

[0121] Furthermore, the present invention also discloses a computer-readable storage medium for storing a computer program; when the computer program is executed by a processor, the MTInSAR landslide monitoring method disclosed above is implemented.

[0122] For more specific processes of the above method, please refer to the corresponding contents disclosed in the aforementioned embodiments, which will not be repeated here.

[0123] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device and storage medium disclosed in the embodiment, since they correspond to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0124] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0125] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0126] In summary, an MTInSAR landslide monitoring method, device and storage medium provided by an embodiment of the present invention include: obtaining SAR image data of a landslide area to be monitored, and preprocessing the SAR image data; establishing a PS monitoring network and a DS monitoring network according to the processed SAR image data; using the PS monitoring network and the DS monitoring network to obtain MTInSAR observation data of the landslide area to be monitored; the MTInSAR observation data include LOS deformation observation values ​​and elevation information; using independent component analysis and periodic analysis methods to identify the atmospheric delay component in the LOS deformation observation values; separating the atmospheric delay component from the LOS deformation observation values ​​to obtain LOS deformation data after atmospheric correction; using terrain data and satellite orbit parameters to simulate the LOS deformation data as deformation along the slope direction. The above MTInSAR landslide monitoring method can obtain LOS deformation observation values ​​through the established PS monitoring network and DS monitoring network, which can increase the number of MTInSAR measurement points in multi-vegetation coverage areas, and remove the atmospheric delay components in the LOS deformation observation values ​​based on independent component analysis and periodic analysis methods, and finally simulate the LOS deformation to the deformation along the slope direction. The whole process does not rely on external weather models or meteorological data, and can effectively complete the atmospheric automatic correction with high temporal and spatial resolution, improve the MTInSAR slope monitoring accuracy, and help implement measures such as landslide early warning and disaster prevention. In addition, the present invention also provides corresponding equipment and computer-readable storage media for the MTInSAR landslide monitoring method, which further makes the above method more practical, and the equipment and computer-readable storage medium have corresponding advantages.

[0127] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0128] The MTInSAR landslide monitoring method, device and storage medium provided by the present invention are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A MTInSAR landslide monitoring method, characterized in that: include: Acquiring SAR image data of the landslide area to be monitored, and preprocessing the SAR image data; Establishing a PS monitoring network and a DS monitoring network according to the processed SAR image data; The PS monitoring network and the DS monitoring network are used to obtain the MTInSAR observation data of the landslide area to be monitored; the MTInSAR observation data includes LOS deformation observation values ​​and slope elevation information; Generate a data matrix according to the LOS deformation observation value; Determine the linear separation matrix corresponding to the data matrix by an iterative method; According to the linear separation matrix and the data matrix, the spatial response and time series of each independent component are obtained; Calculate the spatial correlation factor of each independent component according to the slope elevation information and the mean of the elevation information, the obtained spatial response of each independent component and the mean of the spatial response; The periodic spectrum of the time series of each independent component is analyzed by Fourier transform to obtain the periodicity of the time series of each independent component; The independent component that satisfies both the maximum spatial correlation factor and the most obvious time series periodicity is determined as the atmospheric delay component in the LOS deformation observation value; Separating the atmospheric delay component from the LOS deformation observation value to obtain LOS deformation data after atmospheric correction; The LOS deformation data is simulated as deformation along the slope direction by using terrain data and satellite orbit parameters.

2. The MTInSAR landslide monitoring method according to claim 1, characterized in that: Before using independent component analysis and periodic analysis methods to identify the atmospheric delay component in the LOS deformation observation value, the method further includes: A principal component analysis is performed on the LOS deformation observations, and only a set number of main components constituting the LOS deformation observation data are retained. The set number of main components should explain more than 85% of the data variance to achieve data dimensionality reduction.

3. The MTInSAR landslide monitoring method according to claim 1, characterized in that: Establish a PS monitoring network, including: Selecting PS candidate points according to the amplitude deviation index of the processed SAR image data; All the selected PS candidate points are connected using the DeLorean triangulation network to build a PS monitoring network; The deformation parameters and elevation parameters of the arc segments of the PS monitoring network are solved by using a robust estimation method; According to the deformation parameters and elevation parameters of the arc segments of the PS monitoring network, the PS monitoring network is adjusted as a whole, and a PS candidate point is selected as a deformation monitoring reference point; The elevation information and deformation time series of each PS candidate point relative to the deformation monitoring reference point are obtained.

4. The MTInSAR landslide monitoring method according to claim 3, characterized in that: Establish a DS monitoring network, including: Clustering homogeneous points using the terrain element information of the landslide area to be monitored; Reconstructing the interference phase time series using the coherence matrix of the homogeneous points; Based on the interferogram after phase reconstruction, and according to the threshold of the number of homogeneous points or the threshold of phase coherence, DS candidate points are selected; Connect each DS candidate point with its nearest PS candidate point to form a DS monitoring network; A robust estimation method is used to solve the deformation parameters and elevation parameters of the arc segments of the DS monitoring network; The deformation parameters of the DS monitoring network arc segment are added to the deformation parameters of the connected PS candidate points to obtain the deformation value and elevation value of each DS candidate point.

5. The MTInSAR landslide monitoring method according to claim 4, characterized in that: Use terrain feature information to cluster homogeneous points, including: The slope and slope angle values ​​of the landslide area to be monitored are calculated using a digital elevation model (DEM) having a spatial resolution superior to or equivalent to that of SAR images; In a filtering window, pixels with similar slope and slope angle are selected as homogeneous points of the center pixel of the filtering window.

6. A MTInSAR landslide monitoring device, characterized in that: It comprises a processor and a memory, wherein when the processor executes the computer program stored in the memory, the MTInSAR landslide monitoring method as claimed in any one of claims 1 to 5 is implemented.

7. A computer-readable storage medium, characterized in that: Used to store a computer program, wherein when the computer program is executed by a processor, the MTInSAR landslide monitoring method according to any one of claims 1 to 5 is implemented.

Images