A permafrost area insar deformation monitoring method based on spatial ICA separation

Abstract

The application provides a permafrost area InSAR deformation monitoring method based on spatial ICA separation, and specifically is: firstly, using time series InSAR technology to obtain the unwrapped differential interferometric phase map of the monitored permafrost area; secondly, using spatial ICA to separate the phase of the permafrost area; then, using different models for the separated permafrost area and seasonal frozen soil area respectively; further, solving the permafrost deformation parameters based on the parameter estimation method of Jacobi iteration; finally, adding the low-pass deformation of the model fitting and the filtered high-pass deformation to obtain the deformation monitoring result of the permafrost area. The application can improve the modeling accuracy and deformation estimation accuracy of the InSAR technology by extracting the phase signal related to the permafrost area deformation in advance for modeling, and provides a reference for the comprehensive environmental management and disaster assessment of the permafrost area.

Description

Technical Field

[0001] This invention relates to permafrost surface deformation monitoring, and more particularly, to an InSAR deformation monitoring method for permafrost regions based on spatial ICA separation. Background Technology

[0002] Traditional methods for monitoring surface subsidence in permafrost regions, such as total stations / prism methods, leveling, and global navigation satellite systems, can meet the accuracy requirements for permafrost monitoring. However, these methods are costly, have low spatiotemporal resolution, and are insufficient for observing overall surface subsidence in large areas of permafrost. Furthermore, the need for frequent on-site surveys by monitoring personnel severely limits their application. Differential Interferometric Synthetic Aperture Radar (D-InSAR), as an emerging space-based Earth observation technology, is primarily used to monitor centimeter-level or smaller surface deformations along the radar line of sight. With its wide monitoring range, high spatial resolution, and non-contact measurement capabilities, it significantly compensates for the shortcomings of traditional measurement methods, providing a new opportunity for monitoring subsidence in large-area permafrost regions. However, spatiotemporal loss of correlation and atmospheric delay limit the monitoring accuracy of D-InSAR technology. To overcome the shortcomings of D-InSAR technology, time-series InSAR technology has been proposed by scholars. This technology is mainly represented by permanent scatterer (PS), small baseline subset (SBAS), and temporally coherent point inSAR (TCP-InSAR). By performing phase analysis on multiple SAR images in a time series, and utilizing high-quality points with stable scattering characteristics, it can extract surface deformation, making it more promising for monitoring surface deformation in permafrost regions.

[0003] Permafrost subsidence is a complex, nonlinear, and distinctly periodic process, and its surface deformation is easily affected by external climate conditions. Traditional InSAR methods for estimating permafrost deformation mostly rely on assumed models to perform equal-weighted modeling of each deformation component in the unwrapped interferometric phase, without reasonably extracting the relevant phase components of permafrost deformation for different physical causes, resulting in low estimation progress.

[0004] Therefore, it is of great significance to design a method with high estimation accuracy that can be applied to settlement monitoring in permafrost regions. Summary of the Invention

[0005] This invention provides an InSAR deformation monitoring method for permafrost regions based on spatial ICA separation, addressing the shortcomings of existing InSAR permafrost deformation monitoring methods, such as low deformation estimation accuracy and difficulties in physical interpretation due to the failure to properly extract deformation-related components for different permafrost types. Specifically, the method involves: first, using time-series InSAR technology to obtain the unwrapped differential interferometric phase map of the monitored permafrost region; second, using spatial ICA to perform phase separation of the permafrost region; next, modeling the separated perennial and seasonal permafrost regions using different models; further, solving for the permafrost deformation parameters based on the Jacobi iteration parameter estimation method; and finally, adding the low-pass deformation fitted by the model to the filtered high-pass deformation to obtain the deformation monitoring results for the permafrost region. This invention uses temporal and spatial filtering methods to reasonably remove errors such as atmospheric and phase noise, effectively extracting boundary and temporal deformation phase signals for different types of permafrost, improving the accuracy of deformation modeling and estimation, and facilitating the scientific interpretation of deformation, thus providing a reference for comprehensive environmental management and disaster assessment in permafrost regions. The specific scheme is as follows:

[0006] A method for InSAR deformation monitoring in permafrost regions based on spatial ICA separation includes the following steps:

[0007] Step 1: Generation of temporal InSAR interferometric phase based on high coherence points, specifically: using time-series InSAR technology to obtain the differential interferometric phase map after unwrapping in the monitored permafrost region;

[0008] Step 2: Spatial ICA phase separation in the permafrost region, specifically: using spatial ICA to perform spatial phase separation in the permafrost region;

[0009] Step 3: InSAR permafrost deformation time series modeling, specifically: different models are used to model the separated permafrost region and seasonal permafrost region respectively;

[0010] Step 4: Estimation of frozen soil deformation parameters, specifically: using the Jacobi iteration parameter estimation method to solve for the frozen soil deformation parameters;

[0011] Step 5: Generating time-series deformation in the permafrost region. Specifically, the low-pass deformation fitted by the model is added to the filtered high-pass deformation to obtain the deformation monitoring results of the permafrost region.

[0012] Preferably, step one includes the following steps:

[0013] Step 1.1: Perform interferometric combination and baseline estimation on N+1 SAR image data to generate a connectivity map, then perform multi-look, image registration, and resampling to generate... Aspect-width interferogram and coherence coefficient diagram;

[0014] Step 1.2: Perform flattening, terrain removal, and filtering on the interferogram from Step 1.1 to generate a differential interferogram;

[0015] Step 1.3: Perform phase unwrapping on the differential interferogram from Step 1.2 to obtain the unwrapped differential interferogram;

[0016] Step 1.4: Using the coherence map generated in Step 1.1 and the unwrapped differential interferogram obtained in Step 1.3, high coherence points are extracted to obtain M differential interferometric phases with high coherence point information. Then, based on SVD decomposition, a temporal phase matrix with the first image as a reference is generated, which is the differential interferometric phase map of the monitored permafrost area after unwrapping.

[0017] Preferably, step two includes:

[0018] Construct the relationship between ICA and InSAR interferometric phases, where the ICA model is expressed as: X = A·S1);

[0019] Where: X is the mixed signal, i.e., the temporal phase matrix with the first image as a reference; the temporal phase X contains N columns of phase, where N is the number of images minus one, and the temporal phase is represented as matrix X in expression 2). N :

[0020] In the formula The differential phase of the Nth image relative to the first image

[0021] A is the mixing matrix, where each row represents the spatial response of the independent components; S is the independent component matrix, which is the independent components restored by ICA, where each row represents an independent component.

[0022] If the independent component matrix S is considered as a linear combination of mutually independent components, then the phase of all highly coherent points over the entire time series is represented by each independent component:

[0023]

[0024] In the formula In InSAR data, these are distinct independent components. Including independent signal deformation correlation components Terrain-related components orbital correlation components Atmospheric correlation components and noise-related components Deformation-related components It also includes linear deformation components related to long-term surface displacement. Environment-related periodic deformation components As shown in expression 4):

[0025]

[0026] Based on expression 3), expression 1) is represented as expression 5):

[0027]

[0028] In the formula: X M×N For time-series phase, M is the number of highly coherent points; A M×i S is the mixing matrix; i is the number of separate components; S i×N For all independent components obtained by ICA separation, each component composition;

[0029] For InSAR signals, the deformation phase and the noise-related phase are considered to be independent of each other in time and space. Based on the characteristics of the InSAR interferometric phase components, FastICA is used to decompose the InSAR interferometric phase and extract the required signal.

[0030] Preferably, step three specifically involves:

[0031] For permafrost regions, a multi-rate linear model is used to model and fit deformation:

[0032]

[0033] In the formula: For the deformation of the m-th highly coherent point in the permafrost region, v n The deformation rate between adjacent images, (t) n -t n-1 ) represents the time interval between neighboring images, n∈1,2,…,N;

[0034] For seasonally frozen soil regions, a frozen soil deformation model that takes into account environmental factors is used to fit the deformation:

[0035]

[0036] In the formula: For seasonally frozen soil regions m Deformation of a highly coherent point; T(t) is the temperature parameter, P(t) is the precipitation parameter, calculated from temperature and precipitation data; a1 and a2 are the coefficients of T(t) and P(t), respectively.

[0037] Preferably, in step four, for the multi-rate linear model, an equation can be established for each highly coherent point on the interferogram. Based on the time matrix B of the interferometric pair, the rate is solved using the Jacobi iteration parameter estimation algorithm. v It includes the following steps:

[0038] Step 4.1.1: Perform LDU decomposition on the time matrix B:

[0039] B = D + L + U (8);

[0040] In the formula: D is a diagonal matrix, which can be obtained by directly inverting it. -1 L is a lower triangular matrix; U is an upper triangular matrix;

[0041] Therefore, expression 6) can be written as:

[0042] Step 4.1.2: Based on the results obtained in Step 4.1.1, further determine the iteration format as expression 9):

[0043]

[0044] Step 4.1.3: According to expression 8), let -D -1 (L+U)=B0, Then, rewrite expression 9) and assign initial values ​​for iteration as follows:

[0045]

[0046] The termination condition for step 4.1.4 and expression 10) is |v (n+1) -v n |<ε, where the value of ε is set according to the actual computational requirements, and the obtained value of v is... (n+1) This is the optimal solution for unknown parameters.

[0047] Preferably, in step four, for the deformation model that takes environmental factors into account, given the time matrix B of the interferometric pair and the environmental factors including temperature and precipitation, the parameter T(t) is calculated. k P(t) k :

[0048] T(t) k =Tem g -Tem h 11);

[0049] P(t) k =Pre g -Pre h 12);

[0050] In the formula: T(t) k Let Tem be the temperature parameter of the k-th interference pair. g 、Tem h These represent the temperature data from the master and slave images of the interferometric pair, respectively; P(t) k For the precipitation parameters of the k-th interferometric pair, Pre g Pre hThese are precipitation data from the master-slave images of the interferometric pair, respectively.

[0051] B, T(t) k P(t) k Rewritten as a parameter matrix C:

[0052]

[0053] The unknown parameters are represented as: Y = [v a1 a2]; the parameter estimation algorithm using Jacobi iteration is used to solve for Y, including the following steps:

[0054] Step 4.2.1: Perform LDU decomposition on the coefficient matrix C:

[0055] C = D + L + U (14);

[0056] Expression 7) is written as:

[0057] Step 4.2.2: Based on the results of Step 4.2.1, determine the iteration format as follows:

[0058]

[0059] Step 4.2.3: According to expression 14), let -D -1 (L+U)=C0, Then rewrite expression 15) and assign initial values ​​for iteration:

[0060]

[0061] The termination condition for step 4.2.4 and expression 16) is |Y (n+1) -Y n |<ε, where the value of ε is set according to the actual calculation requirements, and the calculated Y is... (n+1) This is the optimal solution for unknown parameters.

[0062] Preferably, step five specifically involves:

[0063] Step 5.1: Fit the deformation of the multi-rate linear model obtained in Step 3. And the deformation of model fitting that takes into account environmental factors. Deformation phase converted to a multi-rate linear model fit Deformation phase of the model fitting to the model that takes environmental factors into account Calculate the residual phase

[0064]

[0065] In the formula: The original total interference phase;

[0066] Step 5.2: Apply mean filtering to the residual phase in expression 17). Spatial low-pass filtering, triangular filtering, and time high-pass filtering are performed to obtain the high-pass deformation phase, which is then converted into a high-pass deformation S. high ;

[0067] Step 5.3: Add the low-pass and high-pass deformations together to obtain the total deformation S of the region:

[0068] S = S low +S high 18);

[0069] Among them, low-pass deformation S low - Deformation fitted by multi-rate linear models and permafrost deformation models that take environmental factors into account:

[0070] Step 5.4: Geographically encode the total deformation of the coherent point line of sight to generate the vertical temporal deformation of the permafrost area, thus obtaining the deformation monitoring results of the permafrost area.

[0071] In addition to the objectives, features, and advantages described above, the present invention has other objectives, features, and advantages. The invention will now be described in further detail with reference to the figures. Attached Figure Description

[0072] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0073] Figure 1 A flowchart of an InSAR deformation monitoring method for permafrost regions based on spatial ICA separation;

[0074] Figure 2 The total deformation results on days 12, 144, 288, 456, 648, and 840 are shown for the simulated original mixed signal.

[0075] Figure 3 The simulation experiment shows the spatial response and time series based on ICA separation. In the left figure, from top to bottom (a1) and (b1), the spatial response and time series of linear deformation are respectively. In the right figure, from top to bottom (a2) and (b2), the spatial response and time series of periodic deformation are respectively. The horizontal axis represents the simulation time, and the vertical axis represents the deformation level of the component.

[0076] Figure 4 The mixed signals after ICA separation illustrate the overall deformation trends on days 12, 144, 288, 456, 648, and 840, respectively.

[0077] Figure 5 The graphs (a), (b), (c), (d), (e), and (f) show the comparison between calculated values ​​and simulated real values ​​under the high coherence point method in this study. They respectively illustrate the comparisons on day 12, day 144, day 288, day 456, day 648, and day 840. Detailed Implementation

[0078] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings. However, the present invention can be implemented in many different ways as defined and covered by the claims.

[0079] Example:

[0080] A method for monitoring deformation in permafrost regions based on spatial ICA separation, such as Figure 1 This includes the following steps:

[0081] Step 1: Generation of temporal InSAR interferometric phase based on high coherence points. Specifically, this involves using time-series InSAR technology to obtain the differential interferometric phase map of the unwrapped permafrost region, including the following steps:

[0082] Step 1.1: Perform interferometric combination and baseline estimation on N+1 SAR image data to generate a connectivity map, followed by multi-look, image registration, and resampling to generate... Aspect-width interferogram and coherence coefficient diagram;

[0083] Step 1.2: Perform flattening, terrain removal, and filtering on the interferogram from Step 1.1 to generate a differential interferogram;

[0084] Step 1.3: Perform phase unwrapping on the differential interferogram from Step 1.2 to obtain the unwrapped differential interferogram;

[0085] Step 1.4: Using the coherence coefficient map generated in Step 1.1 and the unwrapped differential interferogram obtained in Step 1.3, high coherence points are extracted (e.g., coherence reaches 0.75 or above), resulting in M ​​differential interferometric phases with high coherence point information. Subsequently, a temporal phase matrix with the first image as a reference is generated based on SVD decomposition, which is the differential interferometric phase map of the unwrapped frozen soil area.

[0086] Step 2: Spatial ICA phase separation in the permafrost region. Specifically, spatial ICA is used to perform phase separation in the permafrost region, including the following steps:

[0087] Construct the relationship between ICA and InSAR interferometric phases; the ICA model is expressed as:

[0088] X = A·S1);

[0089] in:

[0090] (1) X is the mixed signal, which is the time-series phase obtained after unwrapping and SVD decomposition in InSAR data, i.e. the differential interferometric phase map after unwrapping in the monitored permafrost area; A is the mixed matrix, each row of which represents the spatial response of the independent component; S is the independent component matrix, i.e. the independent component restored by ICA, each row of which represents an independent component.

[0091] (2) The temporal phase X contains N columns of phases, where N is the number of images minus one. The temporal phase is represented as matrix X. N :

[0092]

[0093] In the formula: The difference phase between the Nth image and the first image;

[0094] (3) If the independent component matrix S is considered as a linear combination of mutually independent components, then the phase of all highly coherent points over the entire time series is represented by each independent component:

[0095]

[0096] In the formula: In InSAR data, these are different independent components. Including independent signal deformation correlation components Terrain-related components orbital correlation components Atmospheric correlation components and noise-related components Deformation-related components It also includes linear deformation components related to long-term surface displacement. Environment-related periodic deformation components As shown in expression 4):

[0097] Based on expression 3), expression 1) is represented as expression 5):

[0098]

[0099] In the formula: X M×N For the timing phase in step 2.2, M is the number of highly coherent points; A M×i S is the mixing matrix; i is the number of separate components; S i×N For all independent components obtained by ICA separation, each component composition;

[0100] For InSAR signals, the deformation phase and the noise-related phase are considered to be independent of each other in time and space. Based on the characteristics of the InSAR interferometric phase components, FastICA is used to decompose the InSAR interferometric phase and extract the required signal.

[0101] Step 3: InSAR permafrost deformation time series modeling. Specifically, different models are used to model the separated permafrost regions and seasonal permafrost regions, including the following steps:

[0102] (1) For the linear deformation components related to long-term surface displacement obtained by spatial ICA phase separation, and considering the relatively stable and unidirectional deformation characteristics of permafrost regions, a multi-rate linear model (also known as a multi-rate model) is used to model and fit the deformation:

[0103]

[0104] In the formula: For the deformation of the m-th highly coherent point in the permafrost region, v n The deformation rate between adjacent images, (t) n -t n-1 ) represents the time interval between neighboring images, n∈1,2,…,N;

[0105] (2) For the periodic deformation components related to the external environment obtained by spatial ICA phase separation, based on the characteristics that seasonal frozen soil deformation changes significantly with temperature, exhibits obvious periodicity, and has a one-year cycle, a frozen soil deformation model that takes into account environmental factors is used to fit the deformation:

[0106]

[0107] In the formula: denoted as the deformation of the m-th high coherence point in the seasonally frozen soil region; T(t) is the temperature parameter, and P(t) is the precipitation parameter, calculated from temperature and precipitation data; a1 and a2 are the coefficients of T(t) and P(t), respectively.

[0108] Step 4: Estimation of frozen soil deformation parameters. Specifically, the Jacobi iteration parameter estimation method is used to solve for the frozen soil deformation parameters, as follows:

[0109] On the one hand, for the multi-rate linear model, an equation can be established for each highly coherent point on the interferogram. Based on the time matrix B of the interferometric pair, the rate can be solved using the Jacobi iteration parameter estimation algorithm. v It includes the following steps:

[0110] Step 4.1.1: Perform LDU decomposition on the time matrix B:

[0111] B = D + L + U (8);

[0112] In the formula: D is a diagonal matrix, which can be obtained by directly inverting it. -1 L is a lower triangular matrix; U is an upper triangular matrix;

[0113] Therefore, expression 6) can be written as:

[0114] Step 4.1.2: Based on the results obtained in Step 4.1.1, further determine the iteration format as expression 9):

[0115]

[0116] Step 4.1.3: According to expression 8), let -D -1 (L+U)=B0, Then, rewrite expression 9) and assign initial values ​​for iteration as follows:

[0117] ( Figure 1 The iterative formula is a general formula, where X is equivalent to v in this context.

[0118] The termination condition for step 4.1.4 and expression 10) is |v (n+1) -v n |<ε, where the value of ε is set according to the actual computational requirements, and the obtained value of v is... (n+1) The optimal solution for unknown parameters;

[0119] On the other hand, for the deformation model that takes into account environmental factors, given the time matrix B of the interferometric pair and the environmental factors including temperature and precipitation, the parameter T(t) is calculated. k P(t) k :

[0120] T(t) k =Tem g -Tem h 11);

[0121] P(t) k =Pre g -Pre h 12);

[0122] In the formula: T(t) k Let Tem be the temperature parameter of the k-th interference pair. g 、Tem h These represent the temperature data from the master and slave images of the interferometric pair, respectively; P(t) k For the precipitation parameters of the k-th interferometric pair, Pre g Pre h These are precipitation data from the master-slave images of the interferometric pair, respectively.

[0123] B, T(t) k P(t) k Rewritten as a parameter matrix C:

[0124]

[0125] The unknown parameters are represented as: Y = [v a1 a2]; the parameter estimation algorithm using Jacobi iteration is used to solve for Y, including the following steps:

[0126] Step 4.2.1: Perform LDU decomposition on the coefficient matrix C:

[0127] C = D + L + U (14);

[0128] Expression 7) is written as:

[0129] Step 4.2.2: Based on the results of Step 4.2.1, determine the iteration format as follows:

[0130]

[0131] Step 4.2.3: According to expression 14), let -D -1 (L+U)=C0, Then rewrite expression 15) and assign initial values ​​for iteration:

[0132] ( Figure 1 The iterative formula is a general formula, where X is equivalent to Y in this context.

[0133] The termination condition for step 4.2.4 and expression 16) is |Y (n+1) -Y n |<ε, where the value of ε is set according to the actual calculation requirements, and the calculated Y is... (n+1) This is the optimal solution for unknown parameters.

[0134] Step 5: Generating time-series deformation in the permafrost region. Specifically, this involves adding the low-pass deformation fitted by the model to the filtered high-pass deformation to obtain the deformation monitoring results for the permafrost region. This includes the following steps:

[0135] Step 5.1: Fit the deformation of the multi-rate linear model obtained in Step 3. And the deformation of model fitting that takes into account environmental factors. Deformation phase converted to a multi-rate linear model fit Deformation phase of the model fitting to the model that takes environmental factors into account Calculate the residual phase

[0136]

[0137] In the formula: The original total interference phase;

[0138] Step 5.2: Apply mean filtering to the residual phase in expression 17). Spatial low-pass filtering, triangular filtering, and time high-pass filtering are performed to obtain the high-pass deformation phase, which is then converted into a high-pass deformation S. high ;

[0139] Step 5.3: Add the low-pass and high-pass deformations together to obtain the total deformation S of the region:

[0140] S = S low +S high 18);

[0141] Among them, low-pass deformation S low Deformation fitted by multi-rate linear models and permafrost deformation models that take environmental factors into account:

[0142] Step 5.4: Geographically encode the total deformation of the coherent point line of sight to generate the vertical temporal deformation of the permafrost area, thus obtaining the deformation monitoring results of the permafrost area.

[0143] Applying the technical solution of this embodiment, a simulation experiment was set up for comparison with this embodiment. The satellite parameters used in the simulation experiment were all set according to the wide-swath interferometric mode generated by the C-band Sentinel-1A satellite and the parameters of the ascending SAR image. According to the characteristics and deformation trend of the test area, the deformation rate v of linear deformation was simulated by a two-dimensional Gaussian function model, with a range of [-30,0] mm / a, and the time t was the actual experimental satellite data collected; the known parameters T(t) and P(t) of periodic deformation were the actual experimental temperature and precipitation data collected, respectively, and the unknown parameters a1 and a2 were set to 0.01 and 0.02 according to the actual experimental area data.

[0144] Figure 2 This figure shows the mixed signal of two independent deformation sequences obtained from the simulation of real parameters in the experiment. The upper left corner of the figure represents the periodic deformation signal, and the lower right corner represents the linear deformation signal. The total deformation results on days 12, 144, 288, 456, 648, and 840 are illustrated respectively.

[0145] Figure 3 To separate the simulated mixed signal based on the spatial ICA method, after normalization, the spatial response value is defined between [-1, 1], resulting in the time series of two independent components and their corresponding spatial responses. Figure 3The left graph, from top to bottom, shows the spatial response and time series of linear deformation, while the right graph, from top to bottom, shows the spatial response and time series of periodic deformation. The horizontal axis represents the simulation time, and the vertical axis represents the deformation order of the components. Combining the simulated linear and periodic components demonstrates that ICA can reconstruct the original signal relatively well.

[0146] Figure 4 To model the independent components after ICA separation separately and then sum them to obtain the mixed signal, the overall deformation trend on days 12, 144, 288, 456, 648, and 840 was obtained. Combined with... Figure 2 The analog mixed signal in the data shows that ICA can very accurately separate the independent components in the original signal and exhibits good consistency in both time and space.

[0147] Figure 5 The figure shows a comparison between the simulated values ​​obtained from the simulation experiment of 500 randomly selected high coherence points and the actual values ​​calculated by the method in this paper. Statistical analysis of the data from day 12, day 144, day 288, day 456, day 648, and day 840 shows that the overall difference is small. The proportions of high coherence points with differences between [-4, 4] mm are 100%, 97.4%, 95.8%, 100%, and 96.6%, respectively, which meets the accuracy requirements.

[0148] This invention introduces blind source signal separation (ICA) technology into InSAR data processing, separating the InSAR differential interferometric phase to obtain the independent components affecting permafrost deformation, and then modeling them accordingly. This improves upon the limitations of traditional InSAR permafrost deformation monitoring models, which rely on a single, purely empirical mathematical model. It significantly enhances the modeling accuracy of InSAR technology and its accuracy in long-term permafrost deformation monitoring, and also makes the interpretation of deformation estimation results more reasonable. This provides a reference for permafrost deformation monitoring, helping to prevent permafrost degradation and protect the ecological environment.

[0149] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for monitoring deformation in permafrost regions using InSAR based on spatial ICA separation, characterized in that, Includes the following steps: Step 1: Generation of temporal InSAR interferometric phase based on high coherence points, specifically: using time-series InSAR technology to obtain the differential interferometric phase map after unwrapping in the monitored permafrost region; Step 2: Spatial ICA phase separation in the permafrost region, specifically: using spatial ICA to perform spatial phase separation in the permafrost region; Step 3: InSAR permafrost deformation time series modeling, specifically: different models are used to model the separated permafrost region and seasonal permafrost region respectively; Step 4: Estimation of frozen soil deformation parameters, specifically: using the Jacobi iteration parameter estimation method to solve for the frozen soil deformation parameters; Step 5: Generating time-series deformation in the permafrost region. Specifically, the low-pass deformation fitted by the model is added to the filtered high-pass deformation to obtain the deformation monitoring results of the permafrost region.

2. The InSAR deformation monitoring method for permafrost regions according to claim 1, characterized in that, Step one includes the following steps: Step 1.1, for Interferometric combination and baseline estimation are performed on SAR image data to generate a connectivity map, followed by multi-look processing, image registration, and resampling to generate... Aspect interferogram and coherence coefficient diagram, wherein: ; Step 1.2: Perform flattening, terrain removal, and filtering on the interferogram from Step 1.1 to generate a differential interferogram; Step 1.3: Perform phase unwrapping on the differential interferogram from Step 1.2 to obtain the unwrapped differential interferogram; Step 1.4: Using the coherence map generated in Step 1.1 and the unwrapped differential interferogram obtained in Step 1.3, high coherence points are extracted to obtain... A differential interferometric phase with highly coherent point information is generated, and then a temporal phase matrix with the first image as a reference is generated based on SVD decomposition, which is the differential interferometric phase map after unwrapping of the monitored permafrost area.

3. The InSAR deformation monitoring method for permafrost regions according to claim 2, characterized in that, Step two includes: Construct the relationship between ICA and InSAR interferometric phases, where the ICA model is expressed as: 1）； in: The signal is a mixed signal, specifically a temporal phase matrix referenced to the first image; temporal phase. Includes Column phase, where The number of images is reduced by one, and the temporal phase is represented by the matrix in expression 2). : 2); In the formula For the first The differential phase of the first scene image relative to the first scene image; This is a mixing matrix, where each row represents the spatial response of the independent components; It is an independent component matrix, that is, the independent components restored by ICA, with each row representing an independent component; Independent component matrix If we consider all highly coherent points as linear combinations of mutually independent components, then the phase of all highly coherent points over the entire time series is represented by each independent component: 3）； In the formula In InSAR data, these are distinct independent components. Including independent signal deformation correlation components Topographical components orbital correlation components Atmospheric related components and noise-related components Deformation-related components It also includes linear deformation components related to long-term surface displacement. Environment-related periodic deformation components As shown in expression 4): 4）； Based on expression 3), expression 1) is represented as expression 5): 5）； In the formula: The temporal phase is referenced to the first image. The number of highly coherent points; It is a mixed matrix; The number of components to be separated; For all independent components obtained by ICA separation, each component composition; For InSAR signals, the deformation phase and the noise-related phase are considered to be independent of each other in time and space. Based on the characteristics of the InSAR interferometric phase components, FastICA is used to decompose the InSAR interferometric phase and extract the required signal.

4. The InSAR deformation monitoring method for permafrost regions according to claim 3, characterized in that, Step three specifically involves: For permafrost regions, a multi-rate linear model is used to model and fit deformation: 6）； In the formula: For the first permafrost region Deformation of a highly coherent point This represents the deformation rate between adjacent images. The time interval between adjacent images ; For seasonally frozen soil regions, a frozen soil deformation model that takes into account environmental factors is used to fit the deformation: 7）； In the formula: For seasonally frozen soil regions Deformation of a highly coherent point; For temperature parameters, These are precipitation parameters, calculated from temperature and precipitation data. They are respectively coefficient.

5. The InSAR deformation monitoring method for permafrost regions according to claim 4, characterized in that, In step four, for the multi-rate linear model, an equation can be established for each highly coherent point on the interferogram, based on the time matrix of the interferometric pair. The rate is solved using a parameter estimation algorithm based on Jacobi iteration. It includes the following steps: Step 4.1.1, for the time matrix conduct break down: 8）； In the formula: Since it's a diagonal matrix, we can directly find its inverse. ; It is a lower triangular matrix; It is an upper triangular matrix; Therefore, expression 6) can be written as: ; Step 4.1.2: Based on the results obtained in Step 4.1.1, further determine the iteration format as expression 9). 9）； Step 4.1.3, according to expression 8), let , Then, rewrite expression 9) and assign initial values ​​for iteration as follows: 10）； The termination condition for step 4.1.4 and expression 10 is: , The size is set according to the actual calculation requirements, and the result obtained at this time This is the optimal solution for unknown parameters.

6. The InSAR deformation monitoring method for permafrost regions according to claim 4, characterized in that, In step four, for the deformation model that takes environmental factors into account, the time matrix of the interference pair is known. And environmental factors including temperature and precipitation, calculation parameters , : 11）； 12）； In the formula: For the first Temperature parameters of each interference pair These are the temperature data from the master and slave images of the interferometric pair, respectively. For the first Precipitation parameters of an interferometric pair, These are precipitation data from the master-slave images of the interferometric pair, respectively. Will Write as a parameter matrix : 13）； The unknown parameter is then represented as: ; Solving using the parameter estimation algorithm of Jacobi iteration This includes the following steps: Step 4.2.1, for the coefficient matrix conduct break down: 14）； Expression 7) is written as: ; Step 4.2.2: Based on the results of Step 4.2.1, determine the iteration format as follows: 15）； Step 4.2.3, according to expression 14), let , Then, rewrite expression 15) and assign initial values ​​for iteration: 16）； The termination condition for step 4.2.4 and expression 16 is: , The size is set according to the actual calculation requirements, and the result obtained at this time This is the optimal solution for unknown parameters.

7. The InSAR deformation monitoring method for permafrost regions according to claim 4, characterized in that, Step five specifically involves: Step 5.1: Fit the deformation of the multi-rate linear model obtained in Step 3. And the deformation of model fitting that takes into account environmental factors. Deformation phase converted to a multi-rate linear model fit Deformation phase of the model fitting to the model that takes environmental factors into account Calculate the residual phase : 17）； In the formula: The original total interference phase; Step 5.2: Apply mean filtering to the residual phase in expression 17). Spatial low-pass filtering, triangular filtering, and time high-pass filtering are performed to obtain the high-pass deformation phase, which is then converted into a high-pass deformation phase. ; Step 5.3: Add the low-pass and high-pass deformations to obtain the total deformation of the region. : 18）； Low-pass deformation Deformation fitted by multi-rate linear models and permafrost deformation models that take environmental factors into account: ; Step 5.4: Geographically encode the total deformation of the coherent point line of sight to generate the vertical temporal deformation of the permafrost area, thus obtaining the deformation monitoring results of the permafrost area.

Images