Slope deformation monitoring method based on spaceborne synthetic aperture radar

Abstract

The invention relates to the technical field of geological monitoring, and discloses a side slope deformation monitoring method based on spaceborne synthetic aperture radar, which comprises the following steps: S1, data acquisition: acquiring multi-source spaceborne synthetic aperture radar (SAR) data of a side slope area, the multi-source spaceborne SAR data comprising C-band SAR data and L-band SAR data; s2, self-adaptive preprocessing: self-adaptive preprocessing is performed on the multi-source spaceborne SAR data according to a ground surface coverage scene of a slope area, and the ground surface coverage scene comprises a low vegetation area, a medium-high vegetation area, a flying dust interference area and a high and steep terrain area. According to the slope deformation monitoring method based on the spaceborne synthetic aperture radar, a self-adaptive wave band fusion strategy is adopted for different vegetation coverage scenes, a time sequence coherence enhancement algorithm is combined, the vegetation area monitoring precision is remarkably improved, and noise signals are effectively eliminated and interference errors are reduced through flying dust and construction interference dynamic identification and a multi-dimensional filtering algorithm.

Description

Technical Field

[0001] This invention relates to the field of geological monitoring technology, specifically to a method for monitoring slope deformation based on spaceborne synthetic aperture radar. Background Technology

[0002] Stability monitoring of steep mine slopes (such as spoil heaps and mining slopes) is a crucial aspect of ensuring safe mine production. Spaceborne synthetic aperture radar (SAR) technology, with its advantages of all-weather, all-time, and wide-area coverage, has become an important means of slope deformation monitoring. However, steep mine slopes present complex scenarios such as vegetation cover, dust interference, and steep terrain, leading to insufficient accuracy in existing spaceborne SAR monitoring methods. In vegetated areas, SAR signals are easily affected by vegetation scattering, resulting in reduced coherence. In particular, the monitoring error of C-band SAR in medium-to-high vegetation areas can reach 5-10 mm. In areas affected by dust and construction interference, high-frequency noise is generated by mining blasting, transportation dust, and mechanical vibration, which mixes with the actual deformation signal of the slope and is difficult to separate effectively using traditional filtering methods. In steep terrain areas, radar shadows and overlapping phenomena lead to data loss, and one-dimensional line-of-sight deformation calculations cannot reflect the three-dimensional motion state of the slope, which can easily cause misjudgment of risks. Therefore, there is an urgent need for a spaceborne SAR slope deformation monitoring method that can specifically address the above problems and improve the monitoring accuracy and reliability in complex scenarios. Summary of the Invention

[0003] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a slope deformation monitoring method based on spaceborne synthetic aperture radar, which solves the aforementioned problems.

[0004] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: a slope deformation monitoring method based on spaceborne synthetic aperture radar, comprising the following steps: S1, Data Acquisition: Acquire multi-source spaceborne synthetic aperture radar (SAR) data for the slope area, including C-band SAR data and L-band SAR data; S2, Adaptive preprocessing: For the surface cover scenario in the slope area, adaptive preprocessing is performed on the multi-source spaceborne SAR data. The surface cover scenario includes low vegetation area, medium and high vegetation area, dust interference area and steep terrain area. S3, Interference Filtering: Interference filtering is performed on the preprocessed SAR data to remove noise signals caused by dust, construction machinery, and vegetation growth. S4, Three-dimensional deformation calculation: A three-dimensional deformation model is constructed based on the filtered SAR data to calculate the horizontal and vertical displacements of the slope. S5, Precision Calibration and Result Output: The solution results of the three-dimensional deformation model are calibrated by a ground truth verification network to output high-precision slope deformation monitoring results.

[0005] Preferably, in step S2, the preprocessing for low vegetation areas (coverage < 30%) adopts a weight allocation algorithm that combines the high resolution characteristics of the C-band with the penetration characteristics of the L-band. For medium-to-high vegetation areas (coverage of 30%-70%), L-band SAR data is used as the main source, and the band weights are dynamically adjusted in combination with vegetation height data.

[0006] Preferably, in step S3, interference filtering marks high-incidence areas by using a dust spatiotemporal distribution model, marks interference areas by combining mechanical operation trajectories, and uses frequency domain filtering and amplitude deviation thresholding to remove noise, thereby reducing dust interference error from ±8mm to within ±2mm.

[0007] Preferably, in step S4, the three-dimensional deformation model constructs a high-precision terrain foundation by fusing multi-level DEMs (topographic map + UAV DEM + GNSS elevation), and uses a 3+2 satellite network to acquire multi-baseline data, calculates the three-dimensional displacement components, and increases the data coverage of high and steep terrain areas to 98%.

[0008] Preferably, in step S5, the ground truth verification network achieves accuracy calibration through three types of verification points, ensuring that the monitoring error in the vegetated area is ≤3mm, the error in the steep area is ≤4mm, and the coherence retention rate is ≥0.6%.

[0009] (III) Beneficial Effects

[0010] Compared with existing technologies, this invention provides a slope deformation monitoring method based on spaceborne synthetic aperture radar, which has the following advantages: 1. This slope deformation monitoring method based on spaceborne synthetic aperture radar adopts an adaptive band fusion strategy for different vegetation cover scenarios and combines it with a temporal coherence enhancement algorithm to significantly improve the monitoring accuracy of vegetated areas. Through dynamic identification of dust and construction interference and multi-dimensional filtering algorithms, noise signals are effectively eliminated and interference errors are reduced.

[0011] 2. This slope deformation monitoring method based on spaceborne synthetic aperture radar, based on multi-level DEM fusion and multi-baseline networking technology, realizes accurate three-dimensional deformation calculation in steep terrain areas, fills data gaps, constructs a ground truth verification network, realizes dynamic calibration, ensures the reliability of monitoring results, and provides accurate data support for mine slope safety management.

[0012] 3. This slope deformation monitoring method based on spaceborne synthetic aperture radar adopts multi-satellite collaborative observation (Sentinel 1 6-day revisit + ALOS-214-day cycle) and automated processing link (GPU cluster to achieve data to results output within 6 hours). It can carry out intensive monitoring of key periods 1 hour after blasting and 48 hours after heavy rain, and capture the "accelerated deformation signal" before landslide (such as the sudden increase in deformation rate from 5 mm / day to 30 mm / day), so as to buy 3-6 hours of buffer time for emergency response.

[0013] 4. This slope deformation monitoring method based on spaceborne synthetic aperture radar, through an adaptive fusion strategy of C / L bands (60% C-band ratio in low vegetation areas to retain resolution, and 80% L-band ratio in medium and high vegetation areas to enhance penetration), combined with a temporal coherence enhancement algorithm (interpolation of neighboring high coherence pixels + phenological factor correction), reduces the deformation monitoring error in vegetation areas from 5-10mm in traditional methods to within 3mm, and improves the coherence retention rate by more than 30%. For example, on a mine slope with 50% vegetation coverage, it can stably capture minute deformations at the 0.5mm level, solving the "signal submersion" problem caused by vegetation scattering.

[0014] 5. This slope deformation monitoring method based on spaceborne synthetic aperture radar features an innovative spatiotemporal dynamic interference identification mechanism (combining blasting plans, machinery trajectories, and meteorological data to mark high-incidence areas) and a multi-dimensional filtering algorithm (frequency domain elimination of noise >0.1Hz + amplitude deviation threshold smoothing). This reduces dust interference error from ±8mm to ±2mm, and achieves a false deformation identification rate of over 95% caused by construction machinery vibration. An application case in an open-pit mine shows that this method reduces the number of invalid warnings by 60%, avoiding production interruptions caused by misjudgments. Detailed Implementation

[0015] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0016] A slope deformation monitoring method based on spaceborne synthetic aperture radar includes the following steps: S1: Data Acquisition Multi-source spaceborne SAR data for this slope area was acquired through customized observation missions by satellite operators. C-band data: Sentinel-1 satellite imagery, 5m resolution, 6-day revisit period, VV+VH polarization; L-band data: ALOS-2 satellite imagery, 3m resolution, 14-day revisit period, polarization HH+HV; By employing multi-satellite collaborative observation (Sentinel 1 revisiting every 6 days + ALOS-214-day cycle) and automated processing links (GPU clusters enabling data to be processed and output within 6 hours), it is possible to conduct intensive monitoring of critical periods such as 1 hour after blasting and 48 hours after heavy rain, capturing "accelerated deformation signals" before landslides (such as the sudden increase in deformation rate from 5 mm / day to 30 mm / day), thus providing 3-6 hours of buffer time for emergency response.

[0017] S2: Adaptive preprocessing Low vegetation area (coverage <30%): Sentinel-1 and ALOS-2 data were fused using a weighted allocation algorithm, with C-band accounting for 60% and L-band accounting for 40%, preserving high resolution while enhancing penetration. Medium to high vegetation areas (coverage 30%-70%): Based primarily on ALOS-2 data, using dual-channel penetration mode, and combined with vegetation height data collected by UAV lidar (accuracy ±0.2m), a dynamic mapping model between vegetation coverage and band weights was established. When vegetation height > 1m, the L-band weight was increased to 80%. For steep terrain areas: For areas with a slope greater than 60°, a 5m high-precision DEM is generated by fusing 1:5000 topographic maps (error ±1m), UAV oblique photogrammetry DEM (accuracy ±0.5m), and GNSS measured elevation points (50m / point). For radar shadow areas, data is repaired by combining 3D laser scanning (point cloud density 100 points / m²) with SAR deformation trend extrapolation. Temporal coherence enhancement: Calculate pixel-level coherence coefficients for multiple image periods, and interpolate nearby high-coherence pixels within a 50m radius for areas with coherence coefficients <0.3. Introduce vegetation phenology factors (based on the local rainy season vegetation growth cycle from April to September) to eliminate seasonal pseudo-deformation. In step S2, the preprocessing of low vegetation areas (coverage <30%) adopts a weight allocation algorithm, which combines the high resolution characteristics of the C-band and the penetration characteristics of the L-band. For medium-to-high vegetation areas (coverage of 30%-70%), L-band SAR data is used as the main source, and the band weights are dynamically adjusted in combination with vegetation height data. By employing an adaptive fusion strategy of C / L bands (60% C band ratio in low vegetation areas to retain resolution, and 80% L band ratio in medium and high vegetation areas to enhance penetration), combined with a temporal coherence enhancement algorithm (interpolation of neighboring high coherence pixels + phenological factor correction), the deformation monitoring error in vegetation areas has been reduced from 5-10 mm in traditional methods to within 3 mm, and the coherence retention rate has been improved by more than 30%. For example, on a mine slope with 50% vegetation coverage, it can stably capture minute deformations at the 0.5 mm level, solving the "signal overload" problem caused by vegetation scattering.

[0018] S3: Interference Filtering Dust interference identification: Combining the mine blasting plan (3 times a week, with known coordinates of each blasting point), transportation route map, and wind speed and direction data from the weather station (sampling rate 10 minutes / time), the area within 500m of the blasting point and 200m on both sides of the transportation road are marked as high-incidence areas for dust. The high-incidence period is within 1 hour after blasting and when the wind force is >3. Mechanical interference marking: Obtain the GPS tracks of excavators and trucks (positioning accuracy ±3m) through the mine production scheduling system, and mark the densely populated areas of machinery (more than 3 machines per 100m²) on the SAR image. Noise filtering: Frequency domain filtering is performed on the marked area data (high-frequency noise >0.1Hz is removed), the standard deviation of the pixel amplitude time series is calculated, and adaptive smoothing is performed on amplitude abrupt change points with standard deviation >0.2 (window size 5×5 pixels). In step S3, interference filtering marks high-incidence areas by using a dust spatiotemporal distribution model and marks interference areas by combining mechanical operation trajectories. Noise is eliminated by using frequency domain filtering and amplitude deviation thresholding, reducing dust interference error from ±8mm to within ±2mm. An innovative spatiotemporal dynamic interference identification mechanism (combining blasting plans, machinery trajectories, and meteorological data to mark high-incidence areas) and a multi-dimensional filtering algorithm (frequency domain elimination of noise >0.1Hz + amplitude deviation threshold smoothing) have reduced dust interference error from ±8mm to ±2mm and achieved a false deformation identification rate of over 95% caused by construction machinery vibration. A case study of its application in an open-pit mine shows that this method has reduced the number of invalid warnings by 60%, avoiding production interruptions caused by misjudgments.

[0019] S4: Three-dimensional deformation calculation Multi-baseline data acquisition: An observation equation set was constructed using ALOS-2 images from three different orbits (incident angles of 30°, 45°, and 60°) and two Sentinel-1 images (side view ±10°). Three-dimensional displacement calculation: The horizontal eastward, northward and vertical displacement components are solved by the least squares method. The calculation weight of areas where the horizontal component accounts for more than 60% (such as the edge of the slope) is increased by 20% to ensure that the horizontal deformation error is ≤5mm. In step S4, the three-dimensional deformation model constructs a high-precision terrain foundation through multi-level DEM fusion (topographic map + UAV DEM + GNSS elevation), and uses a 3+2 satellite network to acquire multi-baseline data, calculates the three-dimensional displacement components, and improves the data coverage of high and steep terrain areas to 98%. Multi-level DEM fusion technology (topographic map + UAV DEM + GNSS elevation) constructs a 5m resolution high-precision terrain model. Combined with 3D laser supplementation of shaded areas and deformation trend extrapolation, the effective monitoring coverage of steep slopes (slope > 60°) is increased from 70% to 98% using traditional methods. Simultaneously, by solving 3D displacement through a 3+2 multi-baseline network, the horizontal slip monitoring accuracy reaches ±5mm, solving the "perception blind spot" of one-dimensional LOS monitoring for lateral slippage, and successfully providing an early warning of slope collapse risk caused by horizontal displacement at a mine spoil heap. The calculated horizontal eastward, northward, and vertical displacement components can be directly correlated with the mechanical stability of the mine slope. For example, the ratio of horizontal displacement to vertical settlement (>1.5 indicates slip risk) helps engineers determine the slope instability mode (whether it is a push-type landslide or a traction-type landslide), improving the targeting of reinforcement schemes by 40%. Based on this, a mine optimized the layout of anti-slide piles, saving 2 million yuan in project costs.

[0020] S5: Precision Calibration and Result Output Ground truth deployment: 10 GNSS monitoring stations were deployed in high vegetation areas, 5 fixed inclinometers were deployed in dusty areas, and 8 wire displacement meters were deployed in steep areas. Calibration mechanism: The SAR monitoring values ​​are compared with the ground true values ​​every month. When the deviation is >3mm, the filtering threshold (such as adjusting the amplitude deviation threshold to 0.18) and DEM fusion weights are automatically optimized. Output results: Generate daily deformation rate map (accuracy ±2mm), 3D deformation cloud map and time series curves of key areas, and push them to the mine safety management and control platform simultaneously; In step S5, the ground truth verification network achieves accuracy calibration through three types of verification points, ensuring that the monitoring error in the vegetated area is ≤3mm, the error in the steep area is ≤4mm, and the coherence retention rate is ≥0.6%.

[0021] Using the method in this embodiment, the effective coverage rate of monitoring high and steep slopes in the mine reaches 98%, with errors of ≤3mm in vegetated areas, ≤2mm in dusty areas, and ≤4mm in high and steep areas. The accuracy is improved by more than 60% compared with traditional methods, meeting the needs of mine safety management and control. The ground-based truth verification network (GNSS + inclinometer + guy wire displacement meter) achieves monthly dynamic calibration. When the deviation between the SAR monitoring value and the truth value is >3mm, the algorithm parameters are automatically optimized to ensure long-term monitoring accuracy stability (annual drift <2mm). Compared with traditional manual inspection (each person covers an average of 0.5km² per day), this method can cover an area of ​​70km² with a single scene image, reducing monitoring costs by 60% and avoiding the safety hazards of personnel entering high-risk areas.

[0022] Working principle: The wavelength of electromagnetic waves in spaceborne SAR determines its surface penetration capability and spatial resolution: C-band (wavelength 5.6cm) has high spatial resolution (1-5m) but weak vegetation penetration, suitable for capturing subtle deformations in bare land or low vegetation areas; L-band (wavelength 23.5cm) has strong penetration (can penetrate vegetation layers below 50cm) but lower resolution (3-10m), suitable for medium to high vegetation areas. This invention uses C / L band data fusion to utilize the complementary characteristics of the two bands—C-band is used to preserve details in low vegetation areas, while L-band is used to penetrate vegetation in medium to high vegetation areas, forming a data source with a balance of "resolution-penetration", solving the problem of insufficient adaptability of a single band in complex vegetation scenarios; Coherence Enhancement Mechanism: The accuracy of SAR interferometry depends on the coherence (phase stability) of multi-period images. Vegetation movement and surface changes can lead to a decrease in coherence. This invention calculates the pixel-level coherence coefficient (γ) and uses high-coherence pixels within a 50m radius for low-coherence areas (γ < 0.3) to repair phase information using spatial correlation. At the same time, vegetation phenological factors (such as the rainy season growth index) are introduced, and periodic vegetation disturbances (such as pseudo-deformation caused by the seasonal withering of herbaceous plants) are eliminated through time series decomposition, thereby improving coherence by more than 30%. The principle of high and steep terrain correction: High and steep slopes (slope > 60°) will cause radar signal "shadow areas" (unable to be illuminated) and "overlapping areas" (signals arrive ahead). Traditional DEM correction errors can reach 1-2m. This invention adopts multi-level DEM fusion: 1:5000 topographic map provides macro-topographic framework, UAV oblique photogrammetry DEM (point cloud density 100 points / m²) supplements slope details, GNSS measured points (50m / point) control absolute elevation, generating a 5m resolution high-precision DEM. Combined with radar incident angle model (θ=30°-60°), the shadow and overlap effects are eliminated through geometric correction, and the data coverage is increased from 70% of the traditional method to 98%. Dust and mechanical interference identification: Dust particles (diameter 1-100μm) scatter SAR signals randomly, manifesting as high-frequency noise (frequency > 0.1Hz); instantaneous displacement of construction machinery vibration can cause abrupt amplitude changes (standard deviation > 0.2). This invention marks the interference area (500m around the blasting point, dense machinery area) through a spatiotemporal model, removes high-frequency components in the frequency domain using a Butterworth filter, and smooths the amplitude change points in the spatial domain using a 5×5 pixel window, achieving separation of noise and real deformation signals, reducing the interference error from ±8mm to ±2mm. Dynamic threshold adaptation mechanism: Different mines have different dust intensity and mechanical density. This invention automatically adjusts the amplitude deviation threshold through ground truth feedback (such as relaxing the threshold to 0.25 in areas with frequent blasting) to avoid loss of deformation signal due to excessive filtering. Spaceborne SAR can only measure one-dimensional deformation along the radar line of sight, which cannot fully reflect the horizontal sliding and vertical settlement of slopes. This method achieves three-dimensional monitoring through the following means: Multi-directional monitoring: Using 5 sets of satellite data from different angles (3 sets of L-band observations from 30°, 45°, and 60° angles, and 2 sets of C-band observations from ±10° angles to the left and right), slope deformation is recorded from multiple directions; Establish a simple system of equations: Each set of data corresponds to an equation, reflecting the relationship between deformation and three-dimensional displacement in that direction. Group 1: L-band (incident angle 30°, observation along the track northward) The LOS deformation (corresponding to the satellite observation of the slope from due north at a 30° angle) Relationship with three-dimensional displacement:

[0023] Coefficient Explanation: Eastward coefficient = 0 (observed along the track in a northward direction; eastward displacement is not projected to LOS); Northward coefficient = cos30° ≈ 0.866; Vertical coefficient = sin30° ≈ 0.5 The deformation in this direction is mainly contributed by the "northward displacement + vertical displacement"; Group 2: L-band (incident angle 45°, observation along the track northward) The LOS deformation (corresponding to the satellite observation of the slope from due north at a 45° angle) Relationship with three-dimensional displacement:

[0024] Coefficient Explanation: Eastward coefficient = 0 (same as the observation logic along the track); Northward coefficient = cos45°≈0.707; Vertical coefficient = sin45°≈0.707 In this direction of deformation, the northward displacement and the vertical displacement contribute equally to the deformation. Group 3: L-band (incident angle 60°, observation along the track northward) The LOS deformation (corresponding to the satellite observation of the slope from due north at a 60° angle) Relationship with three-dimensional displacement:

[0025] Coefficient Explanation: Eastward coefficient = 0; Northward coefficient = cos60° ≈ 0.5; Vertical coefficient = sin60° ≈ 0.866 Deformation in this direction is mainly dominated by "vertical displacement" (with a larger vertical coefficient). Group 4: C-band (left-side view, incident angle 35°) The LOS deformation is observed from a westward (left-side view) angle of 35° on the slope. Relationship with three-dimensional displacement: Coefficient Explanation: Eastward coefficient = cos35° ≈ 0.819 (left-side view observation, direct projection of eastward displacement); Northward coefficient = 0 (side-view direction perpendicular to north); Vertical coefficient = sin35° ≈ 0.574 The deformation in this direction is mainly contributed by the "eastward displacement + vertical displacement". Group 5: C-band (right-side view, incident angle 35°) The LOS deformation is observed from the east (right side view) at a 35° angle by the corresponding satellite. Relationship with three-dimensional displacement:

[0026] Coefficient Explanation: Eastward coefficient = -cos35° ≈ -0.819, the right-side view and left-side view directions are opposite, and the eastward displacement projection sign is reversed; Northward coefficient = 0; Vertical coefficient = sin35° ≈ 0.574 In this direction of deformation, the contribution of eastward displacement is opposite to that of the left-side view, which is used to distinguish the positive and negative directions of eastward displacement.

[0027] in, The line-of-sight (LOS) deformation is represented by the radar observations from 5 satellites, with positive values ​​for closer satellites and negative values ​​for farther satellites. in, The displacement is horizontal eastward, with eastward being positive and westward being negative; in, Horizontal displacement in the northward direction is positive, and displacement in the southward direction is negative; Where u_z is the vertical displacement, upward (lifting) is positive and downward (settling) is negative; The actual observations from 5 groups of satellites Substituting into the system of equations, the solution can be obtained by using the least squares method and a programming approach. , The three-dimensional displacement values ​​are u_z; Solving the three-dimensional displacement: By solving the above system of equations using mathematical methods (least squares method), the horizontal eastward displacement can be directly calculated. ), horizontal north ( Displacement values ​​in the three directions: ), perpendicular (u_z), and vertical (u_z). For high and steep slopes: increase the calculation weight for areas with obvious horizontal sliding (horizontal displacement accounts for more than 60%) to ensure the monitoring accuracy of such high-risk areas (error ≤ 5mm). The ground truth verification network achieves multi-dimensional calibration through three types of equipment: GNSS monitoring stations (±1mm) provide absolute displacement benchmarks, fixed inclinometers (±0.01°) verify slope dip angle changes, and string line displacement meters (0.1mm resolution) capture minute slippage in steep areas. SAR monitoring values ​​are compared with true values ​​monthly, and when the deviation is >3mm, the algorithm parameters are automatically optimized—such as increasing L-band weight in vegetated areas and reducing the filter threshold in dusty areas—forming a "monitoring-verification-optimization" closed loop to ensure long-term monitoring accuracy stability.

[0028] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for monitoring slope deformation based on spaceborne synthetic aperture radar, characterized in that, Includes the following steps: S1, Data Acquisition: Acquire multi-source spaceborne synthetic aperture radar (SAR) data for the slope area, including C-band SAR data and L-band SAR data; S2, Adaptive preprocessing: For the surface cover scenario in the slope area, adaptive preprocessing is performed on the multi-source spaceborne SAR data. The surface cover scenario includes low vegetation area, medium and high vegetation area, dust interference area and steep terrain area. S3, Interference Filtering: Interference filtering is performed on the preprocessed SAR data to remove noise signals caused by dust, construction machinery, and vegetation growth. S4, Three-dimensional deformation calculation: A three-dimensional deformation model is constructed based on the filtered SAR data to calculate the horizontal and vertical displacements of the slope. S5, Precision Calibration and Result Output: The solution results of the three-dimensional deformation model are calibrated by a ground truth verification network to output high-precision slope deformation monitoring results.

2. The slope deformation monitoring method based on spaceborne synthetic aperture radar according to claim 1, characterized in that: The preprocessing for low-vegetation areas described in step S2 includes: fusing C-band SAR data and L-band SAR data using a weighted allocation algorithm, wherein C-band SAR data accounts for 60% and L-band SAR data accounts for 40%.

3. The slope deformation monitoring method based on spaceborne synthetic aperture radar according to claim 1, characterized in that: The pretreatment for medium-to-high vegetation zones described in step S2 includes: Using L-band SAR data as the main data source, a dynamic mapping model between vegetation coverage and band weights was established by enabling dual-channel penetration mode and combining vegetation height data collected by UAV lidar.

4. The slope deformation monitoring method based on spaceborne synthetic aperture radar according to claim 1, characterized in that: The interference filtering in step S3 includes: Establish a spatiotemporal distribution model for dust, and mark high-incidence areas and times of dust by combining mine blasting plans, transportation routes and wind speed and direction data; The GPS tracks of construction machinery are obtained through the mine production scheduling system, and areas with high machinery density are marked. Frequency domain filtering is used to remove high-frequency noise with a frequency greater than 0.1 Hz. An adaptive smoothing process is performed on amplitude abrupt change points using the amplitude deviation threshold method. The amplitude abrupt change points are pixels with a time series standard deviation of amplitude values ​​greater than 0.

2.

5. The slope deformation monitoring method based on spaceborne synthetic aperture radar according to claim 1, characterized in that: The construction of the three-dimensional deformation model in step S4 includes: A multi-level DEM fusion model is constructed. The multi-level DEM fusion model is based on a 1:5000 topographic map and integrates UAV oblique photogrammetry DEM and GNSS measured elevation points to generate a high-precision DEM with a resolution of 5m. Multi-baseline SAR data is acquired using a 3+2 satellite network configuration, which includes 3 L-band SAR images from different orbits and 2 C-band SAR images from different side views. The horizontal eastward, northward, and vertical displacement components are solved using the least squares method, with separate calculation weights set for regions where the horizontal component accounts for more than 60%.

6. The slope deformation monitoring method based on spaceborne synthetic aperture radar according to claim 1, characterized in that: The ground truth verification network mentioned in step S5 includes: GNSS monitoring stations deployed in areas with high vegetation cover have a sampling rate of 1Hz and an accuracy of ±1mm. A fixed inclinometer is deployed in the dust interference area, wherein the range of the fixed inclinometer is ±30° and the accuracy is ±0.01°. A wire displacement meter is deployed in a high and steep terrain area, and the resolution of the wire displacement meter is 0.1 mm.

7. The slope deformation monitoring method based on spaceborne synthetic aperture radar according to claim 1, characterized in that: Step S2 also includes preprocessing for steep terrain areas: For radar shadow areas with a slope greater than 60°, a combination of three-dimensional laser scanning and SAR deformation trend extrapolation was used to fill in the data gaps.

8. The slope deformation monitoring method based on spaceborne synthetic aperture radar according to claim 1, characterized in that: The adaptive preprocessing in step S2 further includes an improved temporal coherence enhancement process: Calculate the pixel-level coherence coefficient of multi-phase SAR images; For low coherence regions with a coherence coefficient < 0.3, a neighboring high coherence pixel interpolation method is used, and the search radius of the neighboring high coherence pixels is 50m; Introduce vegetation phenological factors to eliminate pseudo-deformation caused by seasonal vegetation growth.

9. The slope deformation monitoring method based on spaceborne synthetic aperture radar according to claim 1, characterized in that: The calibration described in step S5 includes: comparing the SAR monitoring values ​​with the ground true values ​​every month, and automatically triggering algorithm parameter optimization when the deviation is >3mm.

10. A slope deformation monitoring method based on spaceborne synthetic aperture radar according to claim 1, characterized in that: The slopes referred to are steep slopes in mines, including slopes at mine spoil heaps and slopes at mining operations.

Images