SAR closed phase-based abnormal shallow water identification method for mountainous small watershed slope

Abstract

The application discloses a mountain small watershed slope shallow water anomaly recognition method based on SAR closed phase, which comprises the following steps: acquiring SLC image sequence and carrying out interference processing to obtain interference phase and coherence coefficient; selecting three combinations satisfying closed relation to construct closed phase time sequence stack, and forming water-sensitive observation mainly in the form of decorrelation phase by using closed operation; constructing closed constraint relation matrix to obtain multi-time-phase decorrelation phase sequence; inverting soil moisture variation through the decorrelation phase; aiming at the ordering ambiguity of the closed phase inversion, using the rule that the coherence coefficient changes with the moisture difference to eliminate ambiguity and constraint; constructing an abnormal index, introducing threshold, persistence and spatial consistency criteria to recognize water anomalies, and outputting abnormal results. The application realizes non-contact, continuous and interpretable recognition of mountain small watershed slope shallow water anomalies, and provides data support for slope flow anomaly enhancement and geological disaster precursor identification.

Description

Technical Field

[0001] This invention relates to the field of remote sensing monitoring and geological disaster precursor identification technology, and in particular to a method for identifying shallow water anomalies on slopes in small mountain watersheds based on SAR closed phase construction, decorrelation phase decomposition and soil moisture inversion. Background Technology

[0002] In mountainous areas with small watersheds and active slope flow, the shallow water content of slopes often exhibits anomalous characteristics such as rapid humidification, persistent high humidity, or spatial accumulation under the influence of hydrological processes such as heavy rainfall, snowmelt, or continuous wetting. These anomalies are closely related to the enhancement of slope flow processes and the gestation of disasters such as shallow landslides and slope-type debris flows. Therefore, obtaining the spatiotemporal variations of shallow slope moisture and identifying anomalies is of great significance for disaster precursor assessment and risk management. Existing moisture monitoring methods based on single-pair interferometric phases or single-temporal indicators are often affected by factors such as deformation terms, atmospheric delay terms, topographic residual terms, and unwrapping errors, resulting in aliasing of moisture change signals and insufficient temporal stability, making it difficult to reliably identify shallow slope moisture anomalies under complex mountainous terrain and various land cover conditions.

[0003] Therefore, it is necessary to propose a technical solution that can utilize closed phases to offset the influence of non-moisture factors and generate interpretable anomaly identification results at the small watershed slope scale. Summary of the Invention

[0004] The purpose of this section is to outline some aspects of embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0005] In view of the problems existing in the prior art, the present invention is proposed.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: A method for identifying shallow water anomalies on slopes in small mountain watersheds based on SAR closed phase, the method comprising the following steps: Step 1: Acquire the SLC image sequence of the same track synthetic aperture radar covering the target area, and perform fine registration and common geometric correction processing to obtain the SLC image set after registration and correction; Step 2: Perform interferometric processing on the SLC images in the registered and corrected SLC image set, and generate multi-temporal interferometric phase, coherence coefficient and geocoding products by constructing an interferometric network that satisfies the constraints of temporal and spatial baselines; Step 3: Select three-image combinations that meet the closed-phase calculation conditions from the interference network to form a closed-phase triplet set; Step 4: For each triplet in the set of closed-phase triplets, calculate the closed-phase value at the pixel level and form a closed-phase time sequence stack to obtain the set of closed-phase observations at the pixel level; Step 5: Construct a solution model between the decorrelation phase unknowns and the closed-phase observations, and use convex optimization to obtain the multi-temporal decorrelation phase sequence; Step 6: Invert the decorrelation phase sequence into a soil moisture content sequence; Step 7: The soil moisture content sequence is clipped and aggregated according to the slope flow mask to form pixel-level and slope flow unit-level soil moisture change products, and spatiotemporal consistency constraints and filtering are performed; Step 8: Construct moisture anomaly indicators and identify anomalies, outputting the anomaly occurrence time, anomaly intensity, duration, and spatial connectivity; Step 9: Output the results of shallow water anomaly identification on slopes in small watersheds in mountainous areas, including anomaly index raster, vector boundaries of anomaly areas, and time-series curves and statistical summaries for disaster precursor analysis.

[0007] Furthermore, the calculation of the closed phase value in step 4 is as follows:

[0008] in, Denotes the closed phase of any triplet, wherein the triplet is composed of the first... Jing, Di Jinghedi Scene image composition, the first Jing, Di Jinghedi Scene images can form interference pairs. Indicates by the first Jing and Di The interference phase of the interference pair composed of scenes Indicates by the first Jing and Di The interference phase of the interference pair composed of scenes Indicates by the first Jing and Di The interference phase of the interference pair composed of scenes.

[0009] Furthermore, the solution model in step 5 is as follows: By solving:

[0010] Obtain the decorrelation phase sequence of all interference pairs in the interferometric network. in, Indicates closed-phase observation. This represents the relation matrix, which encodes the topological relationship between each closed triplet and its corresponding three interference pairs. It is to remove the relevant phase unknowns. This represents a set of constraints that contains decorrelated phases. The constraints.

[0011] Further, step 6 specifically includes: performing forward modeling of the unknown soil moisture content sequence based on the decorrelation phase-wavenumber difference mapping relationship, the wavenumber-dielectric constant relationship, and the dielectric constant-moisture content model, to obtain the decorrelation phase simulation values ​​of each interference pair in the interferometric network. The decorrelation phase simulation value is soil moisture content. The function, and Indicates the image sequence number in any interference pair; Using the decorrelation phase simulation values ​​of each interference pair in the interferometric network Determine the closed-phase simulation value for each triplet. ,in, Indicates the image sequence number that makes up the triplet; Based on the closed-phase observations of all triplets Simulated values ​​of closed phase The inversion objective function is constructed by fitting the closed-phase residuals. :

[0012] in, For phase untangling operator, Weights set based on coherence coefficient or observation reliability; The soil moisture content sequence is obtained by minimizing the inversion objective function.

[0013] Furthermore, the mapping relationship between the decorrelated phase and the wavenumber difference is as follows:

[0014] in, Used to characterize increased electromagnetic wave attenuation or decreased penetration depth in soil. Indicates the first Jing and Di The electromagnetic wave number corresponding to the scene and wavenumber difference, Indicates the first Jing and Di The decorrelation phase in the interference phase of the interference pair composed of scenes; The wavenumber-dielectric constant relationship is:

[0015] in, This represents the wavenumber of electromagnetic waves propagating in the soil. Represents angular frequency. Permeability, The dielectric constant is related to soil moisture content; The dielectric constant-water content model is as follows:

[0016] in, , Soil moisture content, The coefficient is determined by soil texture parameters.

[0017] Furthermore, step 6 also includes using an approximate relationship of closed phase-moisture difference as a constraint or initialization, the approximate relationship being:

[0018] in, The proportional parameter is calibrated using measured data. , and This indicates the soil moisture content corresponding to the three images that make up the ternary set.

[0019] Furthermore, by minimizing the inversion objective function, the obtained soil moisture content sequence has multiple equivalent solutions, and the multiple equivalent solutions are cyclic permutation relationships. The method also includes calculating a stability index using the coherence coefficient, selecting the wettest reference image based on the stability index, and determining the final soil moisture content sequence from the multiple equivalent solutions based on the wettest reference image.

[0020] Furthermore, the method also includes: Each scene image Each candidate wettest baseline image is selected as an equivalent solution. Calculate ranking index for each candidate wettest baseline image. and obtained One of the possible values, where, , The coherence coefficient, To select the wettest image The sorting index of the water content corresponding to a certain image at the starting point in the corresponding equivalent solution; The obtained Perform linear fitting, i.e. ,in, and This indicates that the fitted parameters have been obtained; Select the one that satisfies Minimum As the final, wettest baseline image; The equivalent solution corresponding to the final wettest baseline image is used as the final soil moisture content sequence.

[0021] Furthermore, the moisture anomaly index in step 8 is defined in a robust standardized form as follows:

[0022] in, For slope units, Indicates in At any given moment, the slope unit Soil moisture content, For A sliding time window centered on the center. For any point in time within the sliding time window, This represents the function for finding the median. This represents the median absolute deviation. Prevent extremely small positive numbers with a denominator of zero; and identify anomalies using thresholds and persistence rules:

[0023] in, This is the abnormal threshold. For a continuous threshold, Indicates the length of the sliding window. Indicates a length of Within the sliding window, satisfy The number of times.

[0024] Furthermore, the anomaly index raster stores the anomaly index in units of pixels, and the vector attribute field of the anomaly region includes the anomaly start time and the anomaly end time; the time series curve includes the soil moisture content sequence at the scale of the anomaly region or slope unit, and the statistical summary includes at least the number of anomaly regions and the cumulative area of ​​anomalies.

[0025] The beneficial effects of this invention are: 1. By introducing a moisture-sensitive observation chain that combines closed-phase construction with closed-constraint decomposition, this invention selects closed triples and forms a closed-phase time series stack within a multi-temporal InSAR interferometric network. Utilizing the cancellation characteristics of closed phase on deformation, atmospheric delay, and topographic residual terms, the observation is transformed from "interferometric phase susceptible to multi-source error aliasing" to "moisture-sensitive constraint dominated by decorrelation phase". Furthermore, multi-temporal decorrelation phase sequences are obtained through linear programming / convex optimization, thereby significantly improving the stability and consistency of shallow moisture time series inversion and anomaly identification under complex mountainous terrain and complex cover conditions. 2. Based on this, the present invention establishes a physical inversion link of "decorrelated phase - electromagnetic propagation parameters - dielectric constant - water content", and quantitatively inverts the decorrelated phase sequence into the time series of shallow soil water content (or relative water content) on the slope by combining an empirical dielectric model. Pixel-level and unit-level water change products are generated at the slope flow mask and slope unit scales. Furthermore, robust standardized anomaly indicators based on median and MAD are adopted. The threshold-persistence threshold joint criterion suppresses isolated spikes and noise false alarms, enabling continuous, interpretable identification and event-based expression of shallow slope water anomalies. This can provide refined support for the assessment of enhanced slope flow anomalies and precursors of shallow landslides / slope debris flows under extreme rainfall conditions. 3. The ingenious aspects of this invention are as follows: On the one hand, by constructing a cyclic triple product of closed phases and using a closed cancellation mechanism, the constraints of deformation, atmospheric and topographic errors on moisture identification are structurally reduced, and closed phase observations are uniformly stacked into a closed constraint equation set, enabling the decorrelation phase sequence to be robustly solved within a global framework; on the other hand, to address the problem of equivalent solutions for moisture content ranking caused by the closed phase sign, the statistical law of coherence coefficient and moisture difference is introduced to select the most likely "wettest / driest" reference image to complete the ambiguity resolution, making the inversion results unique and traceable; finally, the anomaly identification results are organized into anomaly index grids, anomaly region vector boundaries, and time-series curves and statistical summaries for disaster precursor analysis, realizing an integrated process from physical inversion to anomaly product output, thereby obtaining highly reliable shallow moisture anomaly identification results in a non-contact manner even under conditions of fluctuating observation quality and complex mountain scenes. Attached Figure Description

[0026] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of 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, wherein: Figure 1This is a schematic diagram of the overall process of a method for identifying shallow water anomalies on slopes in small mountain watersheds based on SAR closed phase proposed in this invention. Figure 2 This is a flowchart of a preferred embodiment of a method for identifying shallow water anomalies on slopes in small mountain watersheds based on SAR closed phase proposed in this invention. Detailed Implementation

[0027] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0028] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0029] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0030] Reference Figure 1 As an embodiment of the present invention, a method for identifying shallow water anomalies on slopes of small watersheds in mountainous areas based on SAR closed phase is provided. This method includes the following steps: Step 1: Acquire a sequence of Synthetic Aperture Radar (SAR) SLC imagery covering the target area, and perform fine registration and common geometric correction to obtain a set of registered and corrected SLC images. Fine registration and common geometric correction ensure that all SLC images in the sequence are spatially aligned and eliminate any geometric distortions that may exist in the imagery, meeting the geometric consistency requirements for subsequent closed-phase calculations.

[0031] Step 2: Perform interferometric processing on the SLC images in the registered and corrected SLC image set, and generate multi-temporal interferometric phase, coherence coefficient and geocoding products by constructing an interferometric network that satisfies the constraints of temporal and spatial baselines.

[0032] Time baseline constraint refers to setting an appropriate time baseline threshold (e.g., 5 days, 7 days, 10 days, etc.) to filter out influence pairs with time differences within the threshold, reduce the influence of changes in surface scattering characteristics over time, and ensure the coherence of the interference pairs.

[0033] Spatial baseline constraint refers to setting an appropriate spatial baseline threshold (e.g., ±50m, ±100m) to filter out image pairs whose spatial baselines are within the threshold, avoiding coherence loss due to excessively large baselines and ensuring the effectiveness of the interferometric phase.

[0034] In this step, an interferometric network is formed by selecting interferometric pairs (i.e., effective interferometric pairs) that simultaneously satisfy both temporal and spatial baseline constraints (i.e., dual baseline constraints). The dual baseline constraints ensure the phase quality and temporal sensitivity of the acquired effective interferometric pairs, providing reliable data for subsequent moisture inversion.

[0035] By interferometric processing of SLC images, the interferometric phase and coherence coefficient for each interferometric pair can be calculated pixel-by-pixel. The interferometric phase ranges from -π to π, exhibiting phase entanglement (the true phase difference exceeding π is folded into this interval), requiring subsequent phase unwrapping to restore the true phase. The interferometric phase contains effective signals (characterizing soil moisture changes and surface deformation) and noise signals (characterizing atmospheric delay, orbital errors, topographic relief, etc.). The core of subsequent phase closure techniques is to remove noise from the interferometric phase.

[0036] After obtaining the interferometric phase and coherence coefficient, they can be geocoded. This involves mapping the interferometric phase, coherence coefficient, and other data, which were originally stored in the radar coordinate system (slant range-azimuth distance), to a unified geographic coordinate system (such as latitude and longitude coordinate system or UTM projection coordinate system). The resulting dataset has clear spatial location information, which is the interferometric phase coding product and the coherence coefficient coding product. This provides a foundation for subsequent hydrological analysis and disaster early warning.

[0037] Step 3: Select three-image combinations that meet the closed-phase calculation conditions from the interference network to form a closed-phase triplet set.

[0038] Step 2 yielded a large number of interferometric pairs satisfying the dual baseline constraint. By traversing these effective interferometric pairs, a combination of three images satisfying the "pairwise interferometric pair" condition was sought. For example, if an image... , , And interference - , - , - If all are in an interference network, then ( , , ) is a closed-phase triplet.

[0039] Step 4: For each triplet in the closed-phase triplet set, calculate the pixel-level closed-phase value and form a closed-phase time sequence stack to obtain the pixel-level closed-phase observation set.

[0040] The closed-phase time series stack is a set of closed-phase values ​​for each pixel on the time axis, reflecting the phase changes of each pixel at multiple time points. The closed-phase observation set is the closed-phase value calculated from all closed triplets at each time step. The time series stack stacks these closed-phase observation sets in chronological order to form a complete time series data set, which is used for subsequent estimation and inversion of decorrelation phase.

[0041] For example, for any triple ( , , ), respectively corresponding to the first in the SLC image sequence Jing, Di Jinghedi If the scene is such that the closed phase of the triple is... It can be represented as:

[0042] in, Denotes the closed phase of any triplet, wherein the triplet is composed of the first... Jing, Di Jinghedi Scene image composition, the first Jing, Di Jinghedi Scene images can form interference pairs. Indicates by the first Jing and Di The interference phase of the interference pair composed of scenes Indicates by the first Jing and Di The interference phase of the interference pair composed of scenes Indicates by the first Jing and Di The interference phase of the interference pair composed of scenes Indicates the first The phase of the scene, Indicates the first The phase of the scene, Indicates the first The phase of the scene.

[0043] A closed phase can also be equivalently represented as the argument of a cyclic triplet product, i.e.:

[0044] Among them, in the above formula For the first Jing and Di Complex interference quantities formed by the scene, among which, and Pick , and .

[0045] Furthermore, the closed phase is used to characterize the self-consistency of the closed loop formed by the three interference edges in the phase domain: when non-water-related terms (such as deformation, atmospheric, and topographic terms) in the loop have cancelling or significant attenuation characteristics under closed-loop operation, the closed phase is more sensitive to decorrelation phase and noise / unwrapping residual, thus serving as a key observation for subsequent decorrelation phase estimation; whereby This is the argument operator, used to extract the phase angle of a complex number.

[0046] Specifically, for any given interference phase, the interference phase can be decomposed into the following form:

[0047] in, For deformation phase, For atmospheric delayed phase, For terrain phase, This is for the decorrelated phase (this term can characterize changes in soil moisture and surface deformation). Zero-mean noise phase, For untangling integers.

[0048] The phase decomposition is used to clarify the contribution sources of different physical processes in the interferometric phase, so that the closed phase can weaken or cancel the components that are geometrically related to the observation and can be approximately canceled within the loop after summing the triplet loop (for example, after summing, the deformation phase, atmospheric delay phase, and topographic phase of the three interferometric pairs can be weakened or canceled), thereby concentrating the observation constraints on the decorrelation phase and noise / unwrapping terms.

[0049] Therefore, for any triplet, its closed phase can be simplified to an expression that relates only to the decorrelation phase and the noise / unwrapping term:

[0050] in, That is, the decorrelation phase of the corresponding interference pair. This is the equivalent unwrapping integer term after the closure operation, used to absorb the integer multiples caused by the inconsistency between phase wrapping and unwrapping. Residual.

[0051] Step 5: Construct a solution model between the decorrelation phase unknowns and the closed-phase observations, and use convex optimization to obtain the multi-temporal decorrelation phase sequence.

[0052] The solution model is a linear model, i.e. , First, define the sequential interferometric logarithm and the triplet closure phase number generated from the M-scene image as follows: and And construct a closed-phase stacked vector Decorrelated phase unknown vector Relationship matrix , Represent the set of real numbers, and solve using convex optimization methods:

[0053] This yields the decorrelation phase sequence of all interference pairs in the interferometric network. This decorrelation phase sequence serves as the phase input for subsequent soil moisture inversion. express The L2 norm.

[0054] in, This indicates closed-phase observation (i.e., using...) (The actual result obtained) The known relation matrix encodes the topological relationship between each closed triplet and its corresponding three interference pairs, so that each closed phase equation can be expressed as a linear combination of the decorrelation phase terms of several interference pairs. It is to eliminate the related phase unknowns (the quantities that need to be solved). This represents a set of constraints that contains decorrelated phases. The constraints. Constraint set. It is preferred to apply feasible region constraints (e.g., amplitude range constraints, prior smoothing or sparsity constraints, and reliability constraints based on coherence coefficients) to improve the stability and repeatability of the solution under noisy, missing edge or poor network conditions.

[0055] Step 6: Invert the decorrelation phase sequence into a soil moisture content sequence.

[0056] Step 6 specifically includes: performing forward modeling of the unknown soil moisture content sequence based on the decorrelation phase-wavenumber difference mapping relationship, the wavenumber-dielectric constant relationship, and the dielectric constant-moisture content model, to obtain the decorrelation phase simulation values ​​of each interference pair in the interferometric network. The decorrelation phase simulation value is soil moisture content. The function, and Indicates the image sequence number in any interference pair; The mapping relationship between the decorrelation phase and the wavenumber difference is as follows:

[0057] in, Used to characterize increased electromagnetic wave attenuation or decreased penetration depth in soil. Indicates the first Jing and Di The electromagnetic wave number corresponding to the scene and wavenumber difference, (i.e., the above-mentioned) ) indicates the first Jing and Di The decorrelation phase in the interference phase of the interference pair composed of scenes; where, the first Jing and Di The electromagnetic wave number corresponding to the scene and The wavenumber can be determined by the wavenumber-dielectric constant relationship.

[0058] The wavenumber-dielectric constant relationship is:

[0059] Right now: ,in, This represents the wavenumber of electromagnetic waves propagating in the soil. For radar wavelength, Angle of incidence Represents angular frequency. Permeability, The dielectric constant is related to soil moisture content; changes in soil moisture state cause changes in dielectric properties. This allows us to determine the correspondence between the wave number of any scene and the dielectric constant of that scene.

[0060] The dielectric constant-water content model uses an empirical dielectric model to express the complex dielectric constant, i.e.:

[0061] in, , Soil moisture content, (Right now , , , , , The coefficients in this model are calibrated by soil texture parameters. It represents the imaginary unit.

[0062] Based on the above mapping relationships and models, the water content of each image can be simulated using forward modeling to obtain the decorrelation phase simulation values ​​of each interferometric pair in the interferometric network. .

[0063] Next, the decorrelation phase simulation values ​​of each interference pair in the interferometric network are used. Determine the closed-phase simulation value for each triplet. ,in, This indicates the image sequence number that makes up the triplet. = + + ; Then, based on the closed-phase observations of all triples Simulated values ​​of closed phase The inversion objective function is constructed by fitting the closed-phase residuals. :

[0064] in, For phase untangling operator, Weights set based on coherence coefficient or observation reliability; The soil moisture content sequence is obtained by minimizing the inversion objective function.

[0065] In addition, in step 6, the approximate relationship of closed phase-water difference is used as a constraint or initialization, and the approximate relationship is as follows:

[0066] in, The proportional parameter is calibrated using measured data. , and This indicates the soil moisture content corresponding to the three images that make up the ternary set.

[0067] The approximation relationship is used to establish a priori connection between closed-phase observations and moisture content differences, so that when the moisture content is directly inverted from the decorrelation phase, it is sensitive to initial values ​​or has multiple solutions, and the closed phase can be used to constrain the relative ordering and difference magnitude of moisture content.

[0068] To address the ambiguity in moisture content ranking caused by closed phase symbols, the most probable "wettest / driest" reference image is selected using the coherence coefficient. The ambiguity satisfies the following conditions: when ranking solely based on closed phase symbols, the closed phase symbols remain unchanged for any cyclic permutation, and... Scene image exists There are 10 equivalent sorting solutions.

[0069] For example, given three images T1, T2, and T3, their corresponding water content order could be T1>T2>T3, T2>T3>T1, or T3>T1>T2. These three orders are equivalent under a closed-phase constraint (i.e., the calculated closed phase). (These are equivalent), and it is impossible to distinguish the water content corresponding to these three images solely by the value of the closed phase. Therefore, for the soil water content sequence obtained by minimizing the inversion objective function, there are multiple equivalent solutions, and these multiple equivalent values ​​are cyclic permutation relationships. The coherence coefficient... Reflects the first Jinghedi The correlation between landscape images is high; the smaller the difference in moisture content, the higher the coherence coefficient. Therefore, the coherence coefficient can be used to calculate a stability index. Based on the stability index, the wettest baseline image (or the driest baseline influence) can be selected, and the final soil moisture content sequence can be determined from multiple equivalent solutions based on the wettest baseline image (or the driest baseline influence). The following explanation uses the wettest baseline influence as an example.

[0070] Specifically, each image in the SLC image sequence can be... Each candidate wettest baseline image is selected as an equivalent solution. Calculate ranking index for each candidate wettest baseline image. and obtained One of the possible values, where, , The coherence coefficient, To select the wettest image Let be the sorting index of the water content corresponding to a certain image at the starting point in the corresponding equivalent solution. Here, we use... The image with the largest difference in moisture content among all interferometric pairs is selected and sorted. The image with the largest difference represents the image pair with the largest difference in moisture content under the current wettest baseline assumption. The coherence coefficient of this image pair should theoretically be the lowest among all interferometric pairs, and it is the most representative image to verify the stability of the assumption.

[0071] Ranking indices are calculated for each candidate wettest baseline image. You can then obtain multiple The set is composed of, and then the set is... Perform linear fitting, i.e. ,in, and Indicates the fitted parameters; The smaller the value, the weaker the coherence degradation under the wettest assumption, and the more consistent the ranking is with the observation stability. Therefore, it is determined as the most likely wettest baseline image, so as to provide a unique and stable reference for subsequent water content inversion.

[0072] Therefore, choose to satisfy Minimum Using the final wettest baseline image and the equivalent solution corresponding to the final wettest baseline image as the final soil moisture content sequence, this is a unique and stable moisture content time series.

[0073] Step 7: The soil moisture content sequence is clipped and aggregated according to the slope flow mask to form pixel-level and slope flow unit-level soil moisture change products, and spatiotemporal consistency constraints and filtering are performed.

[0074] Specifically, a predefined slope flow mask is used to filter (i.e., "crop") pixel-level moisture data, retaining only valid pixels within the mask area, thus avoiding interference from invalid regions in subsequent analysis. By preserving the moisture time series of each pixel, the resolution is high, reflecting moisture changes at the microscale of the slope. Furthermore, statistically summarizing the moisture data of all pixels belonging to the same slope flow unit (e.g., taking the average) yields the average moisture time series for that unit, facilitating watershed-scale hydrological analysis.

[0075] Spatiotemporal consistency constraints examine the rationality of data from both temporal and spatial dimensions. Temporally, it ensures that moisture changes are continuous, preventing unexplained abrupt changes (e.g., very wet one moment, suddenly dry the next, without supporting precipitation or evaporation events). Spatially, it ensures that moisture changes in adjacent pixels conform to topographical patterns (e.g., the top of a slope is drier than the bottom). Smoothing filters (such as moving averages) can further remove random noise from the data, making moisture changes smoother and highlighting the true spatiotemporal trends. This process generates high-quality hydrological analysis products, ensuring their reliability and usability.

[0076] Step 8: Construct moisture anomaly indicators and identify anomalies, outputting the anomaly occurrence time, anomaly intensity (i.e., the calculated moisture anomaly indicator value), duration, and spatial connectivity.

[0077] The moisture anomaly index is defined in a robust standardized form as follows:

[0078] in, For slope units, Indicates in At any given moment, the slope unit Soil moisture content, For A sliding time window centered on the time (i.e., a sliding window). For any point in time within this sliding time window, This represents the function for finding the median. This represents the median absolute deviation. To prevent extremely small positive numbers with a denominator of zero; Anomalies are identified using thresholds and persistence rules:

[0079] in, This is the abnormal threshold. For a continuous threshold, Indicates the length of the sliding window. Indicates a length of Within the sliding window, satisfy The number of times.

[0080] The robust standardization effectively suppresses the influence of extreme values ​​and non-Gaussian noise on anomaly detection by using the median within the sliding time window as the background level and MAD (median absolute deviation) as the scaling factor, thus improving the anomaly index. It is comparable across different slope units; the threshold The persistence threshold is used to determine whether an abnormal amplitude has reached a significant level. It is used to exclude isolated spikes that occur only in a single period or at a few moments, so that anomaly identification is more consistent with the physical characteristics of continuous response of shallow slope water processes, and outputs the occurrence time, duration and corresponding intensity description of the anomaly.

[0081] Step 9: Output the results of shallow water anomaly identification on slopes in small watersheds in mountainous areas, including anomaly index raster, vector boundaries of anomaly areas, and time-series curves and statistical summaries for disaster precursor analysis.

[0082] The anomaly index raster stores anomaly indices at the pixel level, and the vector attribute fields of the anomaly region include the start and end times of the anomaly. The time-series curve includes soil moisture content sequences at the scale of the anomaly region or slope unit, and the statistical summary includes at least the number of anomaly regions and the cumulative area of ​​the anomaly. The anomaly index raster is output periodically over time to facilitate spatial comparison and temporal tracking of the anomaly evolution process. The vector boundary of the anomaly region is expressed by boundarying the spatial set of anomaly pixels or units. The time-series curve is used to display the comparative characteristics of moisture changes in key slope units before and after the anomaly, and the statistical summary is used to form a searchable and comparable anomaly list at the small watershed scale, providing structured output for disaster precursor analysis, risk assessment, and subsequent manual verification.

[0083] Reference Figure 2 This application also provides a preferred embodiment of a method for identifying shallow water anomalies on slopes in small mountain watersheds based on SAR closed phase.

[0084] Step 1: Acquisition and common geometric correction of SLC image sequences on the same track: Time series data of co-track SAR single-look complex imagery (SLC) were acquired within the slope flow activity area of ​​a small watershed in the target mountainous region. Reference images were selected to perform fine registration of the entire sequence to achieve pixel-level consistent alignment. Common geometric correction and georeferenced consistency processing were also completed to provide a unified geometric basis for interferometric processing, closed phase construction, and subsequent inversion.

[0085] Step Two: Interference Processing and Interference Network Construction The SLC sequence is subjected to interferometry to generate multi-temporal interferometric phase and coherence coefficient products. Valid interferometric pairs are selected based on time and spatial baseline thresholds and an interferometric network is constructed to ensure sufficient connectivity and closed loop structure. The phase, coherence coefficient and geocoding products are then output.

[0086] Step 3: Closed Triple Screening In the interference network, three-scene combinations that satisfy the closure relationship are selected to form a set of closed-phase triplets, such that each triplet corresponds to a closed loop composed of three effective interference edges, which is used for subsequent closed-phase observation and constraint construction.

[0087] Step 4: Closed-phase calculation and timing stack construction: The closed phase is calculated for each closed triplet, and the closed phases of all triplets are stacked at the pixel scale of the study area to form a closed phase time series stack, resulting in a pixel-level closed phase observation set to provide closed observation input for decorrelation phase estimation.

[0088] Step 5: Phase decomposition modeling and construction of closed constraint equations: An interferometric phase decomposition model is established, clarifying that the interferometric phase is composed of the superposition of deformation, atmospheric, topographic, decorrelation, noise, and unwrapped integer terms. By utilizing the cancellation property of the closed phase on the deformation, atmospheric, and topographic terms, the closed phase observation is transformed into a set of constraint equations concerning the decorrelation phase and the noise / unwrapped residual terms, thereby highlighting the phase information related to the changes in medium scattering.

[0089] Step Six: Convex optimization to solve for the decorrelation phase sequence: Based on the set of constraint equations, a solution model is constructed between the decorrelation phase unknowns and closed observations. Linear programming or convex optimization is used to solve the problem to obtain a multi-temporal interferometric pair decorrelation phase sequence that satisfies the constraints and minimizes the residuals, providing a moisture-sensitive phase input for soil moisture inversion.

[0090] Step 7: Soil Moisture Temporal Inversion and Ambiguity Resolution: A forward modeling relationship between decorrelation phase and electromagnetic propagation parameters is established, and combined with the dielectric constant-moisture content model, the decorrelation phase sequence is inverted into a soil moisture time series. To address the ambiguity in moisture content ranking caused by closed phase symbols, the most likely "wettest / driest" reference image is selected using the coherence coefficient to determine the reference and direction of the moisture time series, thereby improving the uniqueness, stability and physical consistency of the inversion results.

[0091] Step 8: Moisture-based product trimming, aggregation, and consistency filtering: The soil moisture time series is spatially clipped according to the slope flow mask to form a pixel-level soil moisture change product; and the pixel-level moisture time series is aggregated according to the slope flow unit to form a slope flow unit-level soil moisture change product; at the same time, spatiotemporal consistency constraints and filtering are applied to the moisture time series to suppress isolated noise and enhance the continuity of the time series.

[0092] Step Nine: Construction of Anomaly Indicators and Anomaly Identification Based on the cell-level or slope flow unit-level soil moisture time series obtained in step eight, moisture anomaly indicators are constructed. Anomaly identification is performed by combining anomaly thresholds and persistence rules. Temporal attributes such as anomaly occurrence time, anomaly intensity, and duration are output, and spatial connectivity is combined to characterize the spatial organization structure of the anomaly.

[0093] Step 10: Output and Summary of Results The system outputs the results of shallow water anomaly identification on slopes in small watersheds in mountainous areas, including anomaly index raster, vector boundaries of anomaly regions, and time-series curves and statistical summaries for disaster precursor analysis. The statistical summaries include at least the number of anomaly regions and the cumulative area of ​​anomalies. It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications and substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for identifying shallow water anomalies on slopes of small watersheds in mountainous areas based on SAR closed phase, characterized in that, The method includes the following steps: Step 1: Acquire the SLC image sequence of the same track synthetic aperture radar covering the target area, and perform fine registration and common geometric correction processing to obtain the SLC image set after registration and correction; Step 2: Perform interferometric processing on the SLC images in the registered and corrected SLC image set, and generate multi-temporal interferometric phase, coherence coefficient and geocoding products by constructing an interferometric network that satisfies the constraints of temporal and spatial baselines; Step 3: Select three-image combinations that meet the closed-phase calculation conditions from the interferometric network to form a closed-phase triplet set; Step 4: For each triplet in the set of closed-phase triplets, calculate the closed-phase value at the pixel level and form a closed-phase time sequence stack to obtain the set of closed-phase observations at the pixel level; Step 5: Construct a solution model between the decorrelation phase unknowns and the closed-phase observations, and use convex optimization to obtain the multi-temporal decorrelation phase sequence; Step 6: Invert the decorrelation phase sequence into a soil moisture content sequence; Step 7: The soil moisture content sequence is clipped and aggregated according to the slope flow mask to form pixel-level and slope flow unit-level soil moisture change products, and spatiotemporal consistency constraints and filtering are performed; Step 8: Construct moisture anomaly indicators and identify anomalies, outputting the anomaly occurrence time, anomaly intensity, duration, and spatial connectivity; Step 9: Output the results of shallow water anomaly identification on slopes in small watersheds in mountainous areas, including anomaly index raster, vector boundaries of anomaly areas, and time-series curves and statistical summaries for disaster precursor analysis.

2. The method according to claim 1, characterized in that: The calculation of the closed phase value in step 4 is as follows: ， in, Denotes the closed phase of any triplet, wherein the triplet is composed of the first... Jing, Di Jinghedi Scene image composition, the first Jing, Di Jinghedi Scene images can form interference pairs. Indicates by the first Jing and Di The interference phase of the interference pair composed of scenes, Indicates by the first Jing and Di The interference phase of the interference pair composed of scenes Indicates by the first Jing and Di The interference phase of the interference pair composed of scenes.

3. The method according to claim 1, characterized in that: The solution model in step 5 is: By solving: ， Obtain the decorrelation phase sequence of all interference pairs in the interferometric network. in, Indicates closed-phase observation. This represents the relation matrix, which encodes the topological relationship between each closed triplet and its corresponding three interference pairs. It is to remove the relevant phase unknowns. This represents a set of constraints that contains decorrelated phases. The constraints.

4. The method according to claim 1, characterized in that: Step 6 specifically includes: performing forward modeling of the unknown soil moisture content sequence based on the decorrelation phase-wavenumber difference mapping relationship, the wavenumber-dielectric constant relationship, and the dielectric constant-moisture content model, to obtain the decorrelation phase simulation values ​​of each interference pair in the interferometric network. The decorrelation phase simulation value is soil moisture content. The function, and Indicates the image sequence number in any interference pair; Using the decorrelation phase simulation values ​​of each interference pair in the interferometric network Determine the closed-phase simulation value for each triplet. ,in, Indicates the image sequence number that makes up the triplet; Based on the closed-phase observations of all triplets Simulated values ​​of closed phase The inversion objective function is constructed by fitting the closed-phase residuals. : ， in, For phase untangling operator, Weights set based on coherence coefficient or observation reliability; The soil moisture content sequence is obtained by minimizing the inversion objective function.

5. The method according to claim 4, characterized in that: The mapping relationship between decorrelated phase and wavenumber difference is as follows: ， in, Used to characterize increased electromagnetic wave attenuation or decreased penetration depth in soil. Indicates the first Jing and Di The electromagnetic wave number corresponding to the scene and wavenumber difference, Indicates the first Jing and Di The decorrelation phase in the interference phase of the interference pair composed of scenes; The wavenumber-dielectric constant relationship is: ， in, This represents the wavenumber of electromagnetic waves propagating in the soil. Represents angular frequency. Permeability, The dielectric constant is related to soil moisture content; The dielectric constant-water content model is as follows: ， in, , Soil moisture content, The coefficient is determined by soil texture parameters.

6. The method according to claim 4, characterized in that: Step 6 further includes using an approximate relationship of closed phase-moisture difference as a constraint or initialization, wherein the approximate relationship is: ， in, The proportional parameter is calibrated using measured data. , and This indicates the soil moisture content corresponding to the three images that make up the ternary set.

7. The method according to claim 4, characterized in that: By minimizing the inversion objective function, the obtained soil moisture content sequence has multiple equivalent solutions. The multiple equivalent solutions are cyclic permutation relationships. The method also includes calculating a stability index using the coherence coefficient, selecting the wettest reference image based on the stability index, and determining the final soil moisture content sequence from the multiple equivalent solutions based on the wettest reference image.

8. The method according to claim 7, characterized in that: The method specifically includes: Each scene image Each candidate wettest baseline image is selected as an equivalent solution. Calculate ranking index for each candidate wettest baseline image. and obtained One of the possible values, where, , The coherence coefficient, To select the wettest image The sorting index of the water content corresponding to a certain image at the starting point in the corresponding equivalent solution; The obtained Perform linear fitting, i.e. ,in, and This indicates that the fitted parameters have been obtained; Select the one that satisfies Minimum As the final, wettest baseline image; The equivalent solution corresponding to the final wettest baseline image is used as the final soil moisture content sequence.

9. The method according to claim 1, characterized in that: In step 8, the moisture anomaly index is defined using a robust standardized form as follows: ， in, For slope units, Indicates in At any given moment, the slope unit Soil moisture content, For A sliding time window centered on the center. For any point in time within the time window, This represents the function for finding the median. This represents the median absolute deviation. Prevent extremely small positive numbers with a denominator of zero; and identify anomalies using thresholds and persistence rules: ， in, This is the abnormal threshold. For a continuous threshold, Indicates the length of the sliding window. Indicates a length of Within the sliding window, satisfy The number of times.

10. The method according to claim 1, characterized in that: The anomaly index raster stores the anomaly index in units of pixels, and the vector attribute field of the anomaly region includes the start time and end time of the anomaly; the time series curve includes the soil moisture content sequence at the scale of the anomaly region or slope unit, and the statistical summary includes at least the number of anomaly regions and the cumulative area of ​​anomalies.

Images