A multi-frequency filtering time-series InSAR deformation signal separation method, system, device and medium

Abstract

The application discloses a multi-frequency filtering time sequence InSAR deformation signal separation method, system, device and medium, and relates to the technical field of surface deformation monitoring. The method comprises the following steps: obtaining a radar line-of-sight deformation time sequence based on a time sequence InSAR technology and converting the radar line-of-sight deformation time sequence into a vertical direction; determining high and low frequency cutoff frequencies according to the time scale of geological and hydrological processes, and respectively extracting a high-frequency component reflecting short-term rapid fluctuation and a trend component reflecting long-term slow deformation through high-pass and low-pass filtering; and obtaining a low-frequency component reflecting seasonal changes by subtracting the high-frequency component and the trend component from an original sequence. The application avoids the uncertainty caused by complex statistical assumptions and model parameterization, realizes intuitive decoupling of multi-driving source deformation signals in the time-frequency domain, contributes to accurate quantification of deformation of different causes, deepens the understanding of the multi-scale driving mechanism of surface deformation, and provides a more direct and reliable technical tool.

Description

Technical Field

[0001] This invention relates to the field of surface deformation monitoring technology, and in particular to a method, system, device and medium for separating multi-frequency filtered time-series InSAR deformation signals. Background Technology

[0002] Land subsidence is a geological phenomenon caused by natural or anthropogenic factors, characterized by the slow sinking of the Earth's surface. It poses a serious threat to urban safety and sustainable development. Traditional monitoring methods struggle to achieve large-scale dynamic monitoring, while InSAR technology, although capable of acquiring long-term deformation data, often contains coupled superpositions of various components, including trends, seasonality, and short-term fluctuations, which traditional methods struggle to effectively separate and interpret. Existing signal decomposition methods largely rely on statistical assumptions and prior knowledge, limiting their ability to clearly define physical meaning and adapt to multiple scales.

[0003] With the widespread application of InSAR technology in Earth observation, the large-scale, high-precision, long-term series surface deformation data it acquires has become an indispensable tool for studying geological disaster processes such as land subsidence and landslides. However, InSAR time-series deformation signals are essentially a coupled superposition of deformations caused by multiple driving sources (such as long-term geological compaction, seasonal hydrological loads, and short-term human activities), exhibiting complex multi-scale mixed characteristics. How to accurately separate and quantitatively interpret the deformation components from different physical sources from this mixed signal is a core challenge currently facing the in-depth understanding of deformation causation mechanisms and the improvement of monitoring and early warning capabilities.

[0004] Currently, the analysis methods for time-series InSAR deformation signals can be mainly divided into the following categories: First, dimensionality reduction and decomposition methods based on mathematical statistics, such as Principal Component Analysis (PCA) and Independent Component Analysis (ICA). These methods separate signals through statistical assumptions, but the physical meaning of each separated component is often unclear, and they are highly dependent on prior knowledge and parameter settings. Second, methods based on time-frequency analysis, such as wavelet transform, can provide time-frequency localization information, but they still have limitations in terms of how to define the decomposition scale based on a clear physical process and how to intuitively associate specific frequency bands with specific driving sources. Third, simulation and inversion methods based on physical models, whose effectiveness is heavily dependent on the accuracy and completeness of model parameters, and whose applicability in complex geological environments at the regional scale is often challenged. Overall, existing methods struggle to achieve adaptive multi-scale signal separation guided by clear physical meaning when processing multi-source coupled deformation signals, and in particular, they lack a technical means to directly and clearly separate long-term trends, seasonal fluctuations, and short-term event-driven deformations systematically. Summary of the Invention

[0005] The purpose of this invention is to provide a method, system, device and medium for separating multi-frequency filtered time-series InSAR deformation signals, aiming to solve or improve at least one of the above-mentioned technical problems.

[0006] To achieve the above objectives, the present invention provides the following solution: A multi-frequency filtered time-series InSAR deformation signal separation method includes: Long-term synthetic aperture radar images covering the target area are acquired, and the images are processed using time-series InSAR technology. The cumulative deformation time series along the radar line of sight at each observation point is extracted and converted into a vertical cumulative deformation time series by combining the radar incident angle. For the cumulative deformation time series of each observation point, the low-frequency cutoff frequency and high-frequency cutoff frequency are determined according to the typical time scales of long-term geological processes and short-term hydrological activities. A high-pass filter is designed based on the high-frequency cutoff frequency to filter the original deformation time series and extract the high-frequency components that reflect short-term rapid fluctuations. A low-pass filter is designed based on the low-frequency cutoff frequency to filter the original deformation time series and extract the trend component reflecting long-term slow deformation. Subtracting the high-frequency component and the trend component from the original deformation time series yields the low-frequency component that reflects seasonal changes.

[0007] Optionally, the deformation period corresponding to the long-term geological process is ≥3 years, and the deformation period corresponding to the short-term hydrological activity is ≤4 months; the low-frequency cutoff frequency and the high-frequency cutoff frequency are set by normalization based on the sampling frequency of the InSAR deformation time series and the Nyquist frequency, wherein the low-frequency cutoff frequency corresponds to the frequency band boundary with a period ≥3 years, and the high-frequency cutoff frequency corresponds to the frequency band boundary with a period ≤4 months.

[0008] Optionally, both the low-pass filter and the high-pass filter are Butterworth filters with a filter order of 6. Before filtering, the two ends of the original deformation time series are extended using the mirror symmetry method, and the number of extension points is at least three times the filter order.

[0009] Optionally, both the low-pass filter and the high-pass filter employ zero-phase bidirectional filtering technology to process the deformation time series of each observation point. Specifically, the original deformation time series is subjected to forward causal filtering, time reversal, reverse filtering, and reversal operation in sequence, so that the filtered signal has no phase delay with the original signal on the time axis.

[0010] Optionally, it also includes: evaluating the reconstruction error of the trend component, high-frequency component and low-frequency component obtained by decomposition. The specific process is: summing the three components and comparing them with the original deformation time series, calculating the root mean square error to verify the accuracy of the decomposition process.

[0011] This invention provides a multi-frequency filtered time-series InSAR deformation signal separation system, comprising: The deformation signal extraction module is used to acquire long-term synthetic aperture radar images covering the target area. It uses time-series InSAR technology to process the images, extracts the cumulative deformation time series along the radar line of sight at each observation point, and converts it into a vertical cumulative deformation time series by combining the radar incident angle. The cutoff frequency calculation module is used to determine the low-frequency and high-frequency cutoff frequencies for the cumulative deformation time series of each observation point, based on the typical time scales of long-term geological processes and short-term hydrological activities. The high-frequency component calculation module is used to design a high-pass filter based on the high-frequency cutoff frequency, filter the original deformation time series, and extract the high-frequency components that reflect short-term rapid fluctuations. The trend component calculation module is used to design a low-pass filter based on the low-frequency cutoff frequency, filter the original deformation time series, and extract the trend component reflecting long-term slow deformation. The low-frequency component calculation module is used to subtract the high-frequency component and the trend component from the original deformation time series to obtain the low-frequency component that reflects seasonal changes.

[0012] The present invention provides an electronic device, including a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to perform the multi-frequency filtering time-series InSAR deformation signal separation method according to the above.

[0013] The present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the multi-frequency filtered timing InSAR deformation signal separation method as described above.

[0014] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects: This invention discloses a method, system, device, and medium for separating deformation signals using multi-frequency filtered temporal InSAR. The method includes acquiring long-term synthetic aperture radar (SAR) images covering a target area; processing the images using temporal InSAR technology to extract the cumulative deformation time series along the radar line of sight at each observation point, and converting it into a vertical cumulative deformation time series based on the radar incident angle; determining low-frequency and high-frequency cutoff frequencies for the cumulative deformation time series at each observation point based on typical time scales of long-term geological processes and short-term hydrological activities; designing a high-pass filter based on the high-frequency cutoff frequency to filter the original deformation time series and extract high-frequency components reflecting short-term rapid fluctuations; designing a low-pass filter based on the low-frequency cutoff frequency to filter the original deformation time series and extract trend components reflecting long-term slow deformation; and subtracting the high-frequency components and trend components from the original deformation time series to obtain low-frequency components reflecting seasonal changes. This invention avoids the uncertainties brought about by complex statistical assumptions and model parameterization, and realizes intuitive decoupling of deformation signals from multiple driving sources in the time-frequency domain. It provides a more direct and reliable technical tool for accurately quantifying the deformation contribution of different causes and deepening the understanding of the multi-scale driving mechanism of surface deformation. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating the multi-frequency filtering time-series InSAR deformation signal separation method in this embodiment; Figure 2 This is a comparison diagram of the experimental results before and after the decomposition of the settlement point and rebound point in this embodiment. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] The purpose of this invention is to provide a method, system, device and medium for separating multi-frequency filtered time-series InSAR deformation signals, aiming to solve or improve at least one of the above-mentioned technical problems.

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] As a first aspect, such as Figures 1-2 As shown, this invention provides a multi-frequency filtered time-series InSAR deformation signal separation method, comprising: S1. Acquire long-term synthetic aperture radar images covering the target area, process the images using time-series InSAR technology, extract the cumulative deformation time series along the radar line of sight at each observation point, and convert it into a vertical cumulative deformation time series by combining the radar incident angle.

[0021] S2. For the cumulative deformation time series at each observation point, determine the low-frequency cutoff frequency and high-frequency cutoff frequency based on the typical time scales of long-term geological processes and short-term hydrological activities. The deformation period corresponding to the long-term geological process is ≥3 years, and the deformation period corresponding to the short-term hydrological activity is ≤4 months. The low-frequency cutoff frequency and high-frequency cutoff frequency are set based on the normalization of the InSAR deformation time series sampling frequency and the Nyquist frequency. The low-frequency cutoff frequency corresponds to the frequency band boundary with a period ≥3 years, and the high-frequency cutoff frequency corresponds to the frequency band boundary with a period ≤4 months.

[0022] S3. Based on the high-frequency cutoff frequency, design a high-pass filter to filter the original deformation time series and extract the high-frequency components that reflect short-term rapid fluctuations.

[0023] S4. Design a low-pass filter based on the low-frequency cutoff frequency, filter the original deformation time series, and extract the trend component reflecting long-term slow deformation.

[0024] S5. Subtract the high-frequency component and the trend component from the original deformation time series to obtain the low-frequency component that reflects seasonal changes.

[0025] In addition, this method also includes: evaluating the reconstruction error of the trend component, high-frequency component and low-frequency component obtained by decomposition. The specific process is: summing the three components and comparing them with the original deformation time series, calculating the root mean square error to verify the accuracy of the decomposition process.

[0026] In one specific implementation, in steps S3 and S4, both the low-pass filter and the high-pass filter are Butterworth filters with an order of 6. Before filtering, the original deformation time series is extended at both ends using a mirror symmetry method, with the number of extension points being at least three times the filter order. Both the low-pass filter and the high-pass filter employ zero-phase bidirectional filtering technology to process the deformation time series at each observation point. Specifically, the original deformation time series is subjected to forward causal filtering, time reversal, reverse filtering, and a second reversal operation to ensure that the filtered signal has no phase delay with the original signal on the time axis.

[0027] Based on the above technical solution, the following embodiments are provided.

[0028] In this embodiment, a physical mechanism-driven multi-frequency filtering time-series InSAR deformation signal separation method is proposed to physically and clearly separate long-term trends, seasonal fluctuations, and short-term high-frequency deformation components from cumulative deformation signals coupled and superimposed from multiple driving sources. This method first determines the key parameters of the filter based on the physical timescale of the deformation process, then achieves distortion-free signal decomposition using zero-phase filtering technology, ultimately obtaining three types of deformation components with clear geological and hydrological significance to support accurate analysis of deformation causes. Specifically, this method first determines the filter type and cutoff frequency, then uses zero-phase bidirectional filtering to process the time series of each deformation point, extracting the trend and high-frequency components, and obtaining the low-frequency components through residual calculation. Finally, the reconstruction error of the decomposition results is evaluated to verify its accuracy. The specific process is as follows: Figure 1 As shown, it includes the following steps: S1. Acquire long-term synthetic aperture radar (SAR) images covering the study area, process them using time-series InSAR technology, extract the cumulative deformation time series along the radar line of sight at each observation point, and combine it with the radar incident angle to convert it into a vertical cumulative deformation time series.

[0029] S2. For the cumulative deformation time series at each point, determine the low-frequency cutoff frequency f based on the typical time scales of long-term geological processes (such as compaction and over-extraction) and short-term hydrological activities (such as sudden pumping changes and extreme precipitation). trend (Corresponding period ≥ 3 years) and high-frequency cutoff frequency f high (Corresponding cycle ≤ 4 months).

[0030] S3, based on a determined cutoff frequency f high Design a high-pass filter to filter the original deformation time series and extract the high-frequency components that reflect short-term rapid fluctuations.

[0031] S4. Based on a determined cutoff frequency f trendDesign a low-pass filter to filter the original deformation time series and extract the trend component that reflects long-term slow deformation.

[0032] S5. Subtract the high-frequency component and trend component obtained above from the original deformation time series, and the resulting residual is the low-frequency component that reflects the seasonal changes.

[0033] As for the more specific implementation steps, the calculation process for each step is as follows: 1. Determine the cutoff frequency based on InSAR deformation data and physical processes: The core of this method is the design of a set of filters with a clear physical orientation. Considering the characteristics of surface deformation signals, the Butterworth filter, which has the flattest amplitude-frequency response within the passband, is selected to avoid introducing amplitude distortion in the target frequency band. The two key parameters of the filter—the low-frequency cutoff frequency and the high-pass cutoff frequency—need to be determined based on the sampling characteristics of the InSAR data and the time scale of the surface deformation process.

[0034] Taking Sentinel-1 satellite imagery (148 episodes) covering the study area from 2018 to 2022 as an example, with a revisit period of 12 days, the sampling frequency of the InSAR deformation time series is: The corresponding Nyquist frequency is: .

[0035] Trend component cutoff frequency: take The trend component mainly reflects long-term geological processes (such as settlement caused by compaction and over-extraction), and its cycle is usually greater than 3 years, i.e., frequency. After normalization, it is approximately 0.022. .

[0036] High-frequency component cutoff frequency: take High-frequency components mainly correspond to short-term hydrological activities (such as sudden pumping events, extreme precipitation, etc.), and their cycles are generally less than 4 months, i.e.: After normalization, it is approximately 0.20 .

[0037] The filter order n is chosen to be 6 to achieve a balance between transition band steepness and computational stability. Based on the aforementioned cutoff frequencies, Butterworth low-pass filters (transfer function H) are designed respectively. trend (f), cutoff frequency f trend ) and high-pass filter (transfer function H high (f), cutoff frequency f high ).

[0038] 2. Perform boundary extension on the time-series signal of each InSAR deformation point: For N deformation points within the study area, let the original cumulative deformation time series of the i-th point be... , where M is the length of the time series. The mirror symmetry method is used to analyze the original sequence d. i Extending 3n points at each end (n is the filter order) makes its timing sequence become .

[0039] 3. Perform zero-phase bidirectional filtering on the time-series signal for each InSAR deformation point: To prevent the phase delay introduced by filtering from causing a time-axis shift in the deformation signal, a zero-phase bidirectional filter is used to process the timing data at each deformation point. Its mathematical expression is: in, This represents traditional causal filtering. This means reversing the time series, i.e., filtering the signal twice, once in the forward direction and once in the reverse direction, so that the system's net phase response is zero and the equivalent amplitude-frequency response is... The transition zone is steeper.

[0040] 4. Multi-scale component decomposition and extraction: Based on the above filter, the following steps are performed: (1) Trend component (T) i Extraction: Extract the zero-phase bidirectional low-pass filter H trend Applied to sequences The specific process is as follows: This component primarily reflects long-term, slow deformation with a period greater than 3 years.

[0041] (2) High-frequency components (H) i Extraction: Extract the zero-phase bidirectional low-pass filter H high Applied to sequences The specific process is as follows: This component primarily reflects short-term, rapid fluctuations with a cycle of less than 4 months.

[0042] (3) Low-frequency component (L) i Calculation: The residuals of the original signal minus the trend component and high-frequency component are used as the low-frequency component. A 5-point moving average is then used to smooth the front and rear boundaries at 3n points each to eliminate possible spurious fluctuations. The specific process is as follows: This component naturally occupies f trend with f highThe transition bands between these frequencies mainly reflect seasonal or interannual periodic deformations.

[0043] (4) Boundary truncation: The trend component, high frequency component and low frequency component obtained from the above decomposition are truncated to the middle M lengths as the final result.

[0044] 5. Reconstruction error assessment: Calculate the root mean square error (RMSE) of the reconstructed signal after decomposition to ensure that the decomposition process does not introduce significant distortion: In summary, the present invention has the following advantages: (1) The physical mechanism-driven multi-frequency filtering time-series InSAR deformation signal separation method proposed in this invention is the first to physically and adaptively separate deformation signals generated by long-term geological processes, seasonal hydrological loads and short-term human activities. This method overcomes the shortcomings of traditional statistical decomposition methods (such as PCA and ICA) with ambiguous physical meaning of components or time-frequency analysis methods (such as wavelet transform) with unclear correspondence between decomposition scale and driving source. It can provide three types of components that directly correspond to different driving mechanisms for InSAR deformation, which greatly improves the accuracy and reliability of quantitative attribution of deformation causes.

[0045] (2) This method innovatively determines the filter cutoff frequency based on the inherent time scale of the surface deformation process (long-term ≥3 years, short-term ≤4 months), giving the decomposition process a clear physical basis and strong interpretability. Compared with parameterized models that rely on prior knowledge or statistical assumptions, the parameter settings of this method are objective and transparent, significantly reducing the reliance on subjective experience and improving the repeatability and universality of the method, making it particularly suitable for large-scale, multi-scenario surface deformation analysis.

[0046] (3) The present invention employs zero-phase bidirectional filtering technology, which effectively avoids the signal phase delay problem caused by traditional filtering and ensures that the decomposed components are strictly aligned with the original signal on the time axis. This characteristic is crucial for accurately analyzing the temporal causal relationship between deformation and potential driving factors (such as precipitation events and pumping activities), and provides a distortion-free and high-fidelity signal foundation for studying the "drive-response" mechanism.

[0047] As a second aspect, the present invention provides a multi-frequency filtered time-series InSAR deformation signal separation system, comprising: The deformation signal extraction module is used to acquire long-term synthetic aperture radar images covering the target area. It uses time-series InSAR technology to process the images, extracts the cumulative deformation time series along the radar line of sight at each observation point, and converts it into a vertical cumulative deformation time series by combining the radar incident angle. The cutoff frequency calculation module is used to determine the low-frequency and high-frequency cutoff frequencies for the cumulative deformation time series of each observation point, based on the typical time scales of long-term geological processes and short-term hydrological activities. The high-frequency component calculation module is used to design a high-pass filter based on the high-frequency cutoff frequency, filter the original deformation time series, and extract the high-frequency components that reflect short-term rapid fluctuations. The trend component calculation module is used to design a low-pass filter based on the low-frequency cutoff frequency, filter the original deformation time series, and extract the trend component reflecting long-term slow deformation. The low-frequency component calculation module is used to subtract the high-frequency component and the trend component from the original deformation time series to obtain the low-frequency component that reflects seasonal changes.

[0048] As a third aspect, the present invention provides an electronic device including a memory and a processor, the memory being used to store a computer program, and the processor running the computer program to cause the electronic device to perform the multi-frequency filtering time-series InSAR deformation signal separation method according to the above description.

[0049] As a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the multi-frequency filtered timing InSAR deformation signal separation method as described above.

[0050] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0051] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for separating multi-frequency filtered time-series InSAR deformation signals, characterized in that, include: Long-term synthetic aperture radar images covering the target area are acquired, and the images are processed using time-series InSAR technology. The cumulative deformation time series along the radar line of sight at each observation point is extracted and converted into a vertical cumulative deformation time series by combining the radar incident angle. For the cumulative deformation time series of each observation point, the low-frequency cutoff frequency and high-frequency cutoff frequency are determined according to the typical time scales of long-term geological processes and short-term hydrological activities, respectively; the deformation period corresponding to the long-term geological processes is ≥3 years, and the deformation period corresponding to the short-term hydrological activities is ≤4 months. A high-pass filter is designed based on the high-frequency cutoff frequency to filter the original deformation time series and extract the high-frequency components that reflect short-term rapid fluctuations. A low-pass filter is designed based on the low-frequency cutoff frequency to filter the original deformation time series and extract the trend component reflecting long-term slow deformation. Subtracting the high-frequency component and the trend component from the original deformation time series yields the low-frequency component that reflects seasonal changes.

2. The multi-frequency filtered time-series InSAR deformation signal separation method according to claim 1, characterized in that, The low-frequency cutoff frequency and the high-frequency cutoff frequency are set by normalizing the sampling frequency and Nyquist frequency of the InSAR deformation time series, wherein the low-frequency cutoff frequency corresponds to the frequency band boundary with a period of ≥3 years, and the high-frequency cutoff frequency corresponds to the frequency band boundary with a period of ≤4 months.

3. The multi-frequency filtered time-series InSAR deformation signal separation method according to claim 1, characterized in that, Both the low-pass filter and the high-pass filter are Butterworth filters with a filter order of 6. Before filtering, the two ends of the original deformation time series are extended using the mirror symmetry method, and the number of extension points is at least three times the filter order.

4. The multi-frequency filtered time-series InSAR deformation signal separation method according to claim 1, characterized in that, Both the low-pass filter and the high-pass filter employ zero-phase bidirectional filtering technology to process the deformation time series of each observation point. Specifically, the original deformation time series is subjected to forward causal filtering, time reversal, reverse filtering, and reversal operation in sequence, so that the filtered signal has no phase delay with the original signal on the time axis.

5. The multi-frequency filtered time-series InSAR deformation signal separation method according to claim 1, characterized in that, It also includes: assessing the reconstruction error of the trend component, high-frequency component and low-frequency component obtained by decomposition. The specific process is: summing the three components and comparing them with the original deformation time series, calculating the root mean square error to verify the accuracy of the decomposition process.

6. A multi-frequency filtered time-series InSAR deformation signal separation system, characterized in that, include: The deformation signal extraction module is used to acquire long-term synthetic aperture radar images covering the target area. It uses time-series InSAR technology to process the images, extracts the cumulative deformation time series along the radar line of sight at each observation point, and converts it into a vertical cumulative deformation time series by combining the radar incident angle. The cutoff frequency calculation module is used to determine the low-frequency cutoff frequency and high-frequency cutoff frequency respectively for the cumulative deformation time series of each observation point, based on the typical time scales of long-term geological processes and short-term hydrological activities; the deformation period corresponding to the long-term geological process is ≥3 years, and the deformation period corresponding to the short-term hydrological activity is ≤4 months. The high-frequency component calculation module is used to design a high-pass filter based on the high-frequency cutoff frequency, filter the original deformation time series, and extract the high-frequency components that reflect short-term rapid fluctuations. The trend component calculation module is used to design a low-pass filter based on the low-frequency cutoff frequency, filter the original deformation time series, and extract the trend component reflecting long-term slow deformation. The low-frequency component calculation module is used to subtract the high-frequency component and the trend component from the original deformation time series to obtain the low-frequency component that reflects seasonal changes.

7. An electronic device, characterized in that, The device includes a memory and a processor, the memory being used to store a computer program, and the processor running the computer program to cause the electronic device to perform the multi-frequency filtered timing InSAR deformation signal separation method according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the multi-frequency filtered time-series InSAR deformation signal separation method as described in any one of claims 1-5.

Images