CN117148340B - A method and system for detecting active geological hazards based on multi-source earth observation - Google Patents

Abstract

The invention discloses an active geological disaster detection method and system based on multisource earth observation, which relate to the technical field of geological disaster detection and comprise the following steps: acquiring a satellite remote sensing image; processing the satellite remote sensing image by combining ENHANCED INSAR STACKING technology, SAR POT, OPOT and DEM differential technology to obtain millimeter-to-meter deformation information; processing millimeter-to-meter deformation information by using MCD and DBSACN methods, and automatically circling the position and boundary of the active geological disaster; the average gradient within the range of the active geological disaster is utilized to divide the active geological disaster into an active landslide and ground subsidence. The invention obtains millimeter-to-meter deformation information, improves the detection capability, automatically circles the position and the range of the active geological disaster by using the MCD and DBSACN method, and improves the efficiency of marking the position and the range of the active geological disaster.

Description

A method and system for detecting active geological hazards based on multi-source earth observation

Technical Field

The present invention relates to the technical field of geological disaster detection, and in particular to a method and system for detecting active geological disasters based on multi-source earth observation.

Background Art

Geological disasters are geological phenomena that commonly occur on the earth's surface. The occurrence of geological disasters often causes casualties and economic losses. In order to reduce the risk of geological disasters, it is crucial to identify the list of active geological disasters.

For areas of hundreds of thousands of square kilometers, only InSAR technology and POT technology based on satellite remote sensing images can obtain wide-area surface deformation in a low-cost and high-efficiency manner, especially in dangerous areas where field investigations are challenging or impractical. They have become indispensable tools for detecting geological disasters.

Most previous studies have used a single technical method to obtain large-scale surface deformation, but in fact, geological disasters at different stages have deformation signals of different gradients, and geological disasters with a moving speed of meters per year are often ignored. Especially in large areas of hundreds of thousands of square kilometers, the location and scope of active geological disasters after obtaining deformation information currently rely on manual interpretation, which is time-consuming and labor-intensive.

Summary of the invention

The purpose of the present invention is to address the deficiencies of the above-mentioned prior art and to provide a method for detecting active geological disasters based on multi-source earth observation, so as to solve the problems in the prior art that geological disasters with a moving speed of meters and deformation per year are often ignored, and after obtaining deformation information, the location and scope of active geological disasters are determined by relying on manual interpretation, which is time-consuming and labor-intensive.

The present invention specifically provides the following technical solution: a method for detecting active geological disasters based on multi-source earth observation, comprising the following steps:

Acquire several satellite remote sensing images;

Optimizing and processing the satellite remote sensing image to obtain millimeter to meter level deformation information;

The satellite remote sensing images are processed by combining Enhanced InSAR Stacking technology, SAR POT, OPOT and DEM difference technology to obtain millimeter to meter level deformation information;

The MCD method is used to statistically identify moving pixels in millimeter to meter-level deformation information;

The moving pixels are processed by using the DBSCAN method to delineate the location and boundary of active geological disasters and obtain the scope of active geological disasters;

The average slope within the active geological disaster range is collected, and the active geological disaster is classified into active landslide or ground subsidence using the average slope.

Preferably, the satellite remote sensing images include: multi-orbit multi-band synthetic aperture radar images, multi-period high spatial resolution satellite remote sensing images and multi-period high spatial resolution DEM images.

Preferably, the optimizing processing of the satellite remote sensing image to obtain millimeter to meter level deformation information comprises the following steps:

The satellite remote sensing images are processed by combining Enhanced InSAR Stacking technology, SAR POT, OPOT and DEM difference technology to obtain millimeter to meter level deformation information.

Preferably, the satellite remote sensing image is processed using Enhanced InSAR Stacking technology, comprising the following steps:

The strategy of InSAR Stacking and atmospheric correction algorithm is adopted to reduce the estimation bias caused by non-stationary signals in satellite remote sensing images;

Generate interferograms using satellite remote sensing images and auxiliary data with reduced estimation bias;

Suppress random noise in the interferogram through adaptive filtering;

The filtered interference graph is phase unwrapped using the minimum cost flow algorithm MCF;

The atmospheric delay in the interferogram after phase unwrapping is corrected, and the average velocity of each point in the interferogram after phase unwrapping is estimated using the least squares method to obtain the final surface deformation information.

Preferably, the satellite remote sensing image is processed using SAR POT and OPOT technology, comprising the following steps:

The SAR POT and OPOT track moving targets in multi-temporal images by calculating the correlation between synthetic aperture radar image blocks in the search window, capturing the offset of the entire pixel and sub-pixel level, and obtaining the final surface deformation information.

Preferably, the satellite remote sensing image is processed using DEM difference technology, comprising the following steps:

The DEM difference technology is to accurately align the DEM images of two phases of satellite remote sensing images and then perform difference to obtain the final surface deformation information.

Preferably, the millimeter to meter level deformation information includes: the annual deformation rate of Sentinel-1 in the LOS direction when ascending, the annual deformation rate of Sentinel-1 in the LOS direction when descending, the deformation amount in the Range direction of Sentinel-1 when ascending, the deformation amount in the Range direction of Sentinel-1 when descending, the east-west deformation obtained by Sentinel-2, and the north-south deformation obtained by Sentinel-2.

Preferably, the DBSCAN method is used to process the moving pixels and delineate the location and boundary of active geological disasters, including the following steps:

Clustering the moving pixels determined by the MCD using the DBSCAN method;

The regions with sufficient density after clustering are divided into clusters, and the density threshold parameter MinPts of the clusters is set to 3;

The clusters whose density threshold parameter MinPts is greater than or equal to 3 are defined as active geological hazards;

The cluster parameter Eps is set to a size of 2 pixels, and clusters with a distance less than 2 pixels are connected;

The convex hull algorithm is used to draw the boundaries of clusters that meet the conditions after setting parameters, and the final location and boundaries of active geological hazards are obtained.

Preferably, the method of using the average slope to divide active geological disasters into active landslides and ground subsidence comprises the following steps:

The threshold of the average slope is set to 5°, and active geological disasters with an average slope greater than the threshold are classified as active landslides, and active geological disasters with an average slope less than the threshold are classified as active geological disasters;

When the active landslide velocity V<100mm/yr, the active landslide is defined as a first-level landslide;

When the active landslide velocity V>100mm/yr, the active landslide is defined as a secondary landslide;

Wherein, the speed of the secondary landslide is greater than the speed of the primary landslide.

The present invention also provides an active geological disaster detection system based on multi-source earth observation, comprising:

Image acquisition module, used to acquire several satellite remote sensing images; optimize the satellite remote sensing images to obtain millimeter to meter level deformation information;

The information processing module uses the MCD method to statistically identify moving pixels in millimeter to meter deformation information;

The detection module processes the moving pixels using the DBSCAN method, delineates the location and boundary of the active geological disaster, and obtains the scope of the active geological disaster;

The qualitative module is used to collect the average slope within the scope of the active geological disaster, and use the average slope to classify the active geological disaster into active landslide or ground subsidence.

Compared with the prior art, the present invention has the following significant advantages:

The present invention obtains millimeter to meter level deformation information by processing different satellite remote sensing images, thereby improving the detection capability, and uses the MCD method to statistically identify moving pixels in the millimeter to meter level deformation information, and utilizes the DBSCAN method to process the moving pixels, thereby delineating the location and boundaries of active geological disasters, and automatically delineating the location and scope of active geological disasters, thereby solving the current situation of manual delineation, greatly improving the efficiency of delineating the location and scope of active geological disasters, and qualitatively identifying the type of geological disasters in combination with the average slope of the active geological disasters, thereby facilitating timely protection by the disaster prevention part in the later stage.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is an overall flow chart provided by the present invention;

FIG2 is a diagram of geographical location and satellite image coverage according to an embodiment of the present invention;

FIG3 is a diagram showing the applicability analysis of SAR images in an embodiment of the present invention;

FIG4 is a millimeter to meter level surface deformation map obtained in an embodiment of the present invention;

FIG5 is a diagram for verifying the accuracy of the deformation results in an embodiment of the present invention;

FIG6 is an enlarged view of two active geological disaster automatic detection areas in an embodiment of the present invention;

FIG7 is a distribution map of active geological disasters in the Hexi Corridor region according to an embodiment of the present invention;

FIG8 is an example diagram of automatic detection of active landslides in an embodiment of the present invention;

FIG. 9 is an example diagram of automatic detection of ground subsidence in an embodiment of the present invention.

DETAILED DESCRIPTION

The following is a clear and complete description of the technical solutions of the embodiments of the present invention in conjunction with the drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

For ease of understanding and explanation, as shown in FIG1 , the present invention provides a method for detecting active geological hazards based on multi-source earth observation, comprising the following steps:

Step S1: Acquire several satellite remote sensing images.

Among them, satellite remote sensing images include multi-orbit and multi-band synthetic aperture radar images, multi-period high spatial resolution satellite remote sensing images, and two or even more periods of high spatial resolution DEM images.

Step S2: Optimize the satellite remote sensing image to obtain millimeter to meter level deformation information.

The satellite remote sensing images are processed by combining Enhanced InSAR Stacking technology, SAR POT, OPOT and DEM difference technology to obtain deformation information at the millimeter to meter level.

Among them, InSAR Stacking technology is an effective InSAR method that can quickly estimate the deformation rate. However, when it comes to SAR images of wide areas, the spatiotemporal changes in water vapor will cause obvious atmospheric signals to appear in the interference pattern.

Step S21: Use Enhanced InSAR Stacking technology to process satellite remote sensing images.

Enhanced InSAR Stacking technology is: the strategy of InSAR Stacking and atmospheric correction algorithm (such as GACOS) can reduce the estimation bias caused by non-stationary signals in satellite remote sensing images.

Step S211: Use GAMMA software to generate an interferogram using Sentinel-1 satellite remote sensing images and auxiliary data (precise orbits, external DEM, etc.).

Step S212: Suppress random noise in the interference pattern through adaptive filtering.

Step S213: performing phase unwrapping on the filtered interference graph using the minimum cost flow (MCF) algorithm.

Step S214: Correct the atmospheric delay in the interferogram after phase unwrapping, increase the probability of residual noise randomness, and use the least squares method to estimate the average velocity of each point in the interferogram after correcting the phase unwrapping. This method can detect deformation information greater than 10 mm/yr to several hundred mm/yr.

Step S22: Use SAR POT and OPOT technology to process the satellite remote sensing image, that is, by calculating the correlation between the satellite remote sensing image blocks in the search window, the moving target in the multi-temporal image can be tracked, the offset of the entire pixel and sub-pixel level can be captured, and the final surface deformation information can be obtained.

The minimum detectable deformation depends on the spatial resolution of the satellite image, which is about 1/20 of the spatial resolution of the image. For example, when performing POT calculation on Sentinel-2 images, the minimum detectable deformation is 0.5 m and the maximum deformation can reach 1 m/hour.

Step S23: using DEM difference technology to process the satellite remote sensing image as follows: using DEM difference technology to accurately align the DEM images of two phases of satellite remote sensing images and then perform difference to obtain the final surface deformation information.

The minimum detectable deformation depends on the spatial resolution of the DEM, and deformation detection greater than 5 m/min can be achieved.

Among them, the millimeter to meter level deformation information includes: the annual deformation rate of Sentinel-1 in the LOS direction when ascending, the annual deformation rate of Sentinel-1 in the LOS direction when descending, the deformation amount in the Range direction of Sentinel-1 in ascending, the deformation amount in the Range direction of Sentinel-1 in descending, the east-west deformation obtained by Sentinel-2, and the north-south deformation obtained by Sentinel-2.

Step S3: Using the MCD method to statistically identify moving pixels in millimeter to meter level deformation information.

The specific process is to use the MCD module of LIBRA software to statistically identify moving pixels. Just input the deformation information and the moving pixels can be automatically detected adaptively.

Step S4: Use the DBSCAN method to process the moving pixels, delineate the location and boundaries of active geological disasters, and obtain the scope of active geological disasters.

Step S41: clustering the moving pixels determined by MCD using the DBSCAN method.

Step S42: After clustering, the regions with sufficient density are divided into clusters, and the density threshold parameter MinPts of the cluster is set to 3.

Step S43: define clusters with a density threshold parameter MinPts greater than or equal to 3 as active geological hazards.

Step S44: Set the cluster parameter Eps to a spatial resolution of 2 pixels to ensure that all moving pixels with a distance less than 2 pixels are connected.

Step S45: using the convex hull algorithm to draw the boundaries of the clusters after clustering (after setting parameters and meeting the conditions) to obtain the final active geological disaster locations and boundaries.

Step S5: Collect the average slope within the scope of active geological disasters, and use the average slope to divide the active geological disasters into active landslides and ground subsidence.

Step S51: setting the threshold of the average slope to 5°, classifying active geological disasters with an average slope greater than the threshold as active landslides, and classifying active geological disasters with an average slope less than the threshold as ground subsidence.

Step S52: When the active landslide velocity V<100 mm/yr, the active landslide is classified as a first-level landslide.

Step S53: When the active landslide velocity V>100 mm/yr, the active landslide is classified as a secondary landslide.

Among them, the speed of the secondary landslide is greater than that of the primary landslide.

Geological disaster prevention and mitigation. The final obtained millimeter to meter level active geological disaster catalog will be provided to the community, so that information can be shared and feedback can be circulated between the community, experts, families and other organizations. This disaster prevention and mitigation model centered on the community and assisted by experts, families and other organizations will effectively improve the disaster prevention and mitigation capabilities of geological disasters in any region.

The present invention also provides an active geological disaster detection system based on multi-source earth observation, comprising:

The image acquisition module is used to acquire a number of satellite remote sensing images; optimize the satellite remote sensing images to obtain millimeter to meter level deformation information.

The information processing module is used to statistically identify moving pixels in millimeter to meter level deformation information using the MCD method.

The detection module is used to process moving pixels using the DBSCAN method, delineate the location and boundaries of active geological disasters, and obtain the scope of active geological disasters.

The qualitative module is used to collect the average slope within the scope of active geological hazards, and use the average slope to classify active geological hazards into active landslides or ground subsidence.

Example

Take the Hexi Corridor region of China as an example:

Figure 2 shows the geographical location and satellite image coverage of this embodiment. Figure 2 (a) shows the geographical location of the Hexi Corridor, and Figure 2 (b) shows the land cover and satellite remote sensing image coverage of the Hexi Corridor in 2022.

Figure 3 is a suitability analysis of the SAR image of this embodiment, Figure 3 (a) is the spatial distribution of geometric distortion of the Sentinel-1 ascending orbit image; Figure 3 (b) is the spatial distribution of geometric distortion of the Sentinel-1 descending orbit image; Figure 3 (c) is the spatial distribution of geometric distortion of the combined Sentinel-1 ascending and descending orbit images; Figure 3 (d) is a schematic diagram of SAR satellite geometric distortion. It can be seen that the geometric distortion and visible area of the ascending orbit image are 6651 km² (3.5%) and 187,032 km² (96.5%), respectively. In contrast, the geometric distortion and visible area of the descending orbit image are 6617 km² (3.4%) and 187,065 km² (96.6%), respectively. Combining the ascending and descending orbit images, the total visible area is 192,429 km² (99.4%), and the geometric distortion area is reduced to 1253 km² (0.6%). Therefore, the combined use of multi-orbit images will greatly improve the ability to detect active geological disasters.

Figure 4 shows the millimeter to meter surface deformation obtained by this embodiment. Figure 4(a) shows the annual deformation rate in the LOS direction of Sentinel-1 orbit ascending; Figure 4(b) shows the annual deformation rate in the LOS direction of Sentinel-1 orbit descending; Figure 4(c) shows the deformation in the Range direction of Sentinel-1 orbit ascending; Figure 4(d) shows the deformation in the Range direction of Sentinel-1 orbit descending; Figure 4(e) shows the east-west deformation obtained by Sentinel-2; Figure 4(f) shows the north-south deformation obtained by Sentinel-2. It can be seen from the figure that the mean and standard deviation of the millimeter to meter deformation results obtained by this embodiment are within the acceptable range, indicating that the millimeter to meter deformation results obtained by this embodiment are reliable.

FIG5 is a verification of the accuracy of the deformation results; from the histogram of the surface deformation difference of the six deformation results in the overlapping area of different orbital images, it can be seen that the mean and standard deviation are small, and the histogram conforms to the Gaussian distribution, which further verifies that the accuracy of the surface deformation results obtained in this embodiment is reliable.

Figure 6 is an enlarged view of two active geological disaster automatic detection areas. It can be seen from the figure that the boundaries of active geological disasters automatically delineated by the MCD+DBSCAN method are completely consistent with the deformation, indicating that the method for automatically delineating the boundaries of active geological disasters proposed in this patent has good performance.

Figure 7 is a distribution map of active geological disasters in the Hexi Corridor region. Active landslides are mainly distributed in the Qilian Mountains, and ground subsidence is mainly distributed in Jiuquan City. Figure 8 is an example diagram of automatic detection of active landslides in an embodiment of the present invention; Figure 9 is an example diagram of automatic detection of ground subsidence in an embodiment of the present invention.

The above content is a further detailed description of the present invention in combination with a specific preferred embodiment. For those skilled in the art to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as belonging to the protection scope of the present invention.

Claims (5)

1. A method for detecting active geological disasters based on multi-source earth observation, characterized in that it comprises the following steps: Acquire a number of satellite remote sensing images; the plurality of satellite remote sensing images include: multi-orbit multi-band synthetic aperture radar images, multi-period high spatial resolution satellite remote sensing images and multi-period high spatial resolution DEM images; Optimizing and processing the satellite remote sensing image to obtain millimeter to meter level deformation information; The MCD method is used to statistically identify moving pixels in millimeter to meter-level deformation information; The moving pixels are processed by using the DBSCAN method to delineate the location and boundary of active geological disasters and obtain the scope of active geological disasters; Collecting the average slope within the scope of the active geological disaster, and using the average slope to classify the active geological disaster into active landslide or ground subsidence; The step of optimizing the satellite remote sensing image to obtain millimeter to meter level deformation information comprises the following steps: The satellite remote sensing images are processed by combining Enhanced InSAR Stacking technology, SAR POT, OPOT and DEM difference technology to obtain millimeter to meter level deformation information; The satellite remote sensing image is processed using the Enhanced InSAR Stacking technology, including the following steps: The strategy of InSAR Stacking and atmospheric correction algorithm is adopted to reduce the estimation bias caused by non-stationary signals in satellite remote sensing images; Generate interferograms using satellite remote sensing images and auxiliary data with reduced estimation bias; Suppress random noise in the interferogram through adaptive filtering; The filtered interference graph is phase unwrapped using the minimum cost flow algorithm MCF; Correct the atmospheric delay in the interferogram after phase unwrapping, and use the least squares method to estimate the average velocity of each point in the interferogram after correcting the phase unwrapping to obtain the final surface deformation information; The satellite remote sensing image is processed using SAR POT and OPOT technologies, including the following steps: The SAR POT and OPOT track moving targets in multi-temporal images by calculating the correlation between synthetic aperture radar image blocks in the search window, capturing the offset of the entire pixel and sub-pixel level, and obtaining the final surface deformation information; The satellite remote sensing image is processed using DEM difference technology, including the following steps: The DEM difference technology accurately aligns the DEM images of two phases of satellite remote sensing images and then performs difference to obtain the final surface deformation information. 2. A method for detecting active geological disasters based on multi-source earth observation as described in claim 1, characterized in that the millimeter to meter deformation information includes: the annual deformation rate in the LOS direction of Sentinel-1 ascending orbit, the annual deformation rate in the LOS direction of Sentinel-1 descending orbit, the deformation amount in the Range direction of Sentinel-1 ascending orbit, the deformation amount in the Range direction of Sentinel-1 descending orbit, the east-west deformation obtained by Sentinel-2, and the north-south deformation obtained by Sentinel-2. 3. The method for detecting active geological disasters based on multi-source earth observation according to claim 1 is characterized in that the DBSCAN method is used to process the motion pixels to delineate the location and boundary of the active geological disasters, comprising the following steps: Clustering the moving pixels determined by the MCD using the DBSCAN method; The regions with sufficient density after clustering are divided into clusters, and the density threshold parameter MinPts of the clusters is set to 3; The clusters whose density threshold parameter MinPts is greater than or equal to 3 are defined as active geological hazards; The cluster parameter Eps is set to a size of 2 pixels, and clusters with a distance less than 2 pixels are connected; The convex hull algorithm is used to draw the boundaries of clusters that meet the conditions after setting parameters, and the final location and boundaries of active geological hazards are obtained. 4. The method for detecting active geological disasters based on multi-source earth observation according to claim 1, characterized in that the method of using the average slope to divide active geological disasters into active landslides and ground subsidence comprises the following steps: The threshold of the average slope is set to 5°, and active geological disasters with an average slope greater than the threshold are classified as active landslides, and active geological disasters with an average slope less than the threshold are classified as ground subsidence; When the active landslide velocity V<100mm/yr, the active landslide is defined as a first-level landslide; When the active landslide velocity V>100mm/yr, the active landslide is defined as a secondary landslide; Wherein, the speed of the secondary landslide is greater than the speed of the primary landslide. 5. An active geological disaster detection system based on multi-source earth observation, characterized by comprising: An image acquisition module is used to acquire a number of satellite remote sensing images; the plurality of satellite remote sensing images include: multi-orbit multi-band synthetic aperture radar images, multi-period high spatial resolution satellite remote sensing images and multi-period high spatial resolution DEM images; Optimize the processing of satellite remote sensing images to obtain deformation information at the millimeter to meter level; An information processing module, used to statistically identify moving pixels in millimeter to meter level deformation information using the MCD method; A detection module is used to process the moving pixels using the DBSCAN method, delineate the location and boundary of the active geological disaster, and obtain the scope of the active geological disaster; A qualitative module, used to collect the average slope within the scope of the active geological disaster, and use the average slope to classify the active geological disaster into active landslide or ground subsidence; The step of optimizing the satellite remote sensing image to obtain millimeter to meter level deformation information comprises the following steps: The satellite remote sensing images are processed by combining Enhanced InSAR Stacking technology, SAR POT, OPOT and DEM difference technology to obtain millimeter to meter level deformation information; The satellite remote sensing image is processed using the Enhanced InSAR Stacking technology, including the following steps: The strategy of InSAR Stacking and atmospheric correction algorithm is adopted to reduce the estimation bias caused by non-stationary signals in satellite remote sensing images; Generate interferograms using satellite remote sensing images and auxiliary data with reduced estimation bias; Suppress random noise in the interferogram through adaptive filtering; The filtered interference graph is phase unwrapped using the minimum cost flow algorithm MCF; Correct the atmospheric delay in the interferogram after phase unwrapping, and use the least squares method to estimate the average velocity of each point in the interferogram after correcting the phase unwrapping to obtain the final surface deformation information; The satellite remote sensing image is processed using SAR POT and OPOT technologies, including the following steps: The SAR POT and OPOT track moving targets in multi-temporal images by calculating the correlation between synthetic aperture radar image blocks in the search window, capturing the offset of the entire pixel and sub-pixel level, and obtaining the final surface deformation information; The satellite remote sensing image is processed using DEM difference technology, including the following steps: The DEM difference technology accurately aligns the DEM images of two phases of satellite remote sensing images and then performs difference to obtain the final surface deformation information.