A method for identifying and dating submarine landslides based on big data analysis

Abstract

The present invention relates to the field of marine geological exploration technology, and specifically to a method for identifying and dating submarine landslides based on big data analysis. The method comprises the following steps: S1, collecting multi-source data such as multi-beam bathymetry, sidescan sonar, and shallow stratigraphic profiles, constructing a dynamic weight allocation model, and fusing these data to generate a three-dimensional terrain model; S2, extracting slope abrupt change zones and acoustic chaos zones based on the terrain model as initial boundaries, verifying the validity of these boundaries by combining shear wave velocity difference ratios, introducing turbidity correction coefficients, and outputting the final landslide region; S3, extracting profile sequence features and constructing a feature vector group. This is then combined with benthic foraminifera oxygen isotope data from core samples to establish a mapping table, and using a matching algorithm to output the age interval of the sedimentary unit. The present invention achieves high-precision identification of submarine landslides and intelligent dating of sedimentary units through multi-source data fusion, intelligent landslide boundary correction, and sequence-fossil combination matching.

Description

Technical Field

[0001] The present invention relates to the field of marine geological exploration technology, and in particular to a method for identifying and dating submarine landslides based on big data analysis. Background Art

[0002] Submarine landslides, as an important type of geological disaster, are widely distributed in deepwater areas such as continental slopes and basin margins. They pose a threat to the safety of submarine communication cables, oil and gas pipelines, deep-sea resource development and other projects, and may also induce local tsunamis. In order to understand their formation mechanism, accumulation characteristics and activity history, identifying landslide bodies and their age characteristics has become an important task in marine geological surveys and disaster assessments. In recent years, with the development of marine observation technology, data such as multi-beam bathymetry, side-scan sonar and shallow stratum profiles can be obtained with high precision, providing a rich data foundation for landslide body identification.

[0003] However, existing submarine landslide identification mostly relies on a single data source or manual interpretation methods, resulting in highly subjective landslide boundary identification, failure to eliminate the impact of sediment disturbance, and difficulty in accurately defining the age of deposits. On the one hand, multi-source data fusion lacks a weight adaptive mechanism, resulting in uneven contributions of different data types in the fusion process; on the other hand, sedimentary disturbances such as turbidity current re-deposition often interfere with boundary judgment, and the lack of structured sequence-age mapping methods results in inaccurate inference of landslide evolution time. Summary of the Invention

[0004] The present invention provides a method for identifying and dating submarine landslides based on big data analysis, which enables age analysis of each sedimentary unit within the landslide body and significantly improves the objectivity of identification and the accuracy of time definition.

[0005] The submarine landslide identification and age determination method based on big data analysis includes the following steps:

[0006] S1 acquires multi-source data, including seafloor multi-beam bathymetry data, side-scan sonar data, and shallow subsurface profile data. It then spatially registers and fuses these data using a dynamic weight allocation model to generate a three-dimensional seafloor terrain model. The dynamic weight allocation model automatically adjusts the weight coefficient based on the data confidence and spatial resolution.

[0007] S2, based on the three-dimensional seabed terrain model, the slope mutation area and the acoustic reflection chaotic area are extracted as the initial landslide boundary. The validity of the boundary is verified by combining the sediment shear wave velocity data. The turbidity current modification correction coefficient is introduced to modify the boundary. The area after eliminating the interference of turbidity current sedimentation is marked as the final landslide area.

[0008] S3: Extract the sedimentary sequence reflection characteristics from the shallow stratigraphic profile data within the final landslide area, construct a sequence feature vector group, and simultaneously obtain the benthic foraminifera oxygen isotope data of normal sedimentary core samples. Through detailed comparison with the existing standard stratigraphic oxygen isotope curve, establish a benthic foraminifera-age mapping table. Use the sequence matching algorithm to temporally and spatially associate the discontinuous sequence feature vectors with the mapping table, and output the age range of each sedimentary unit.

[0009] Optionally, the S1 includes:

[0010] S11, collects multi-beam bathymetric data, side-scan sonar data, and shallow subsurface profile data of the target sea area, corresponding to seabed topography, bottom reflection images, and sedimentary structure data, and pre-processes the collected multi-source data, including format unification, coordinate conversion, noise filtering, and outlier removal;

[0011] S12: Based on the spatial resolution and source confidence of multi-source data, a dynamic weight allocation model is constructed. Initial weights are set for different data sources, and weights are dynamically adjusted during the spatial position matching process. The coverage, signal-to-noise ratio, and detection accuracy of data within the same area are comprehensively considered to implement a regionally adaptive weight allocation strategy.

[0012] S13, based on the dynamic weight allocation results, performs spatial registration and fusion of multi-source data, and uses an interpolation reconstruction algorithm to map the spatially registered and fused multi-source data into a unified 3D geographic reference frame to generate a 3D seabed terrain model.

[0013] Optionally, the S11 includes:

[0014] The S111 uses a multi-beam bathymetric system, side-scan sonar system, and shallow subsurface profiler carried on board the survey vessel to collect seabed topography, underwater reflection images, and sedimentary structure data of the target sea area. Specifically, the following data are collected:

[0015] The multi-beam bathymetric system obtains a discrete set of water depth points and outputs A three-dimensional coordinate set of , where 、 is the two-dimensional plane coordinate of the multi-beam measurement point i, is the sounding value of the multi-beam sounding system at point i, 3D point sets generated for multibeam bathymetry systems;

[0016] The side scan sonar system obtains the reflection intensity image along the track and outputs it as a grayscale matrix ;

[0017] The shallow layer profiler obtains underground reflector information and outputs it as a depth-distance profile matrix , where d is the depth;

[0018] S112, unify the format of the collected multi-source data, build a standard data framework, convert it into a unified reference coordinate system (such as WGS84 or UTM), and convert the longitude and latitude of the center point into a projected coordinate system;

[0019] S113, removing noise points in the multi-beam bathymetric data and side-scan sonar data using a local median filtering method;

[0020] S114, using the Z score method to identify and eliminate outliers in the sounding values ​​after noise removal;

[0021] S115, storing the preprocessed multi-source data as a structured data set, specifically including:

[0022] The final output 3D point cloud model ;

[0023] The final output side scan sonar image ;

[0024] Final output profile reflection data ;

[0025] in, 、 are the horizontal coordinates and vertical projection coordinates in the unified projection coordinate system, is the effective water depth value.

[0026] Optionally, the S12 includes:

[0027] S121, based on the spatial resolution of multi-source data Source confidence , calculate the initial weight ;

[0028] S122, divide the target seabed area into small grid cells , calculate the coverage ratio of each data source in each cell , signal-to-noise ratio and detection stability , forming a comprehensive quality index within the region ;

[0029] S123, according to each small grid unit The data sources in and , calculate the final fusion weight after dynamic adjustment in the region ;

[0030] S124, all grid cells of Organized by spatial position, the weight field matrix for the 3D terrain fusion process is generated.

[0031] Optionally, the S13 includes:

[0032] S131, convert multi-beam bathymetric data, side-scan sonar data, and shallow subsurface profile data into a unified 3D geographic coordinate system (such as the UTM projection system), and use rigid transformation to achieve multi-source data coordinate registration;

[0033] S132, based on the final dynamic weight , perform weighted fusion of each data source in a unified coordinate grid and calculate the fused bathymetric value;

[0034] S133 uses the weighted inverse distance interpolation (IDW) algorithm to interpolate and reconstruct the discrete sampling points of the fused bathymetric values ​​to generate a three-dimensional seabed terrain model .

[0035] Optionally, the S2 includes:

[0036] S21, based on a 3D seabed terrain model , calculate the slope value at each location , and identify the slope mutation threshold (Set to ) is the continuous area of ​​slope mutation;

[0037] S22, using the local texture variance of side-scan sonar images to detect chaotic reflection areas, the high-variance areas were identified as sedimentary disturbance zones, reflecting the acoustic discontinuity of the landslide accumulation area;

[0038] S23, the spatial overlap between the identified slope mutation zone and the chaotic reflection zone is used as the initial landslide boundary area, and the sediment shear wave velocity is extracted inside and outside the initial landslide boundary area. , calculate the shear wave velocity difference rate on both sides of the boundary , when the shear wave velocity change rate Exceeding the shear wave velocity change rate judgment threshold (set to 20%), it is considered that there is a stratigraphic difference at the boundary, the geological rationality of the boundary is verified, and the boundary that meets the requirements is retained. The boundary subset of the output is the shear wave velocity verification boundary ;

[0039] S24, Verify Boundary Based on Output Shear Wave Velocity , introducing the turbidity transformation index , define the boundary correction factor And update the landslide boundary, and finally keep only The boundary points are output as the final landslide area, where Keep the lower limit for the boundary.

[0040] Optionally, the S22 includes:

[0041] S221, in side scan sonar image On the other hand, a fixed-size two-dimensional sliding window ( ), with each center pixel Extract the grayscale value set of its neighborhood for the center;

[0042] S222, calculate the local variance of the gray value for each sliding window , generate texture variance map;

[0043] S223, set the chaotic reflection area determination threshold (set to 50), the local texture variance exceeds the judgment threshold The position is marked as chaotic area, and a binary mask is generated. , expressed as:

[0044] .

[0045] Optionally, the S3 includes:

[0046] S31, extract shallow stratum profile data in the final landslide area, identify the reflection interface and extract its sequence characteristics, and construct a sequence feature vector group ;

[0047] S32, extract typical locations in a core sample of normal sedimentation Oxygen isotope data of benthic foraminifera at , through detailed comparison with existing standard formation oxygen isotope curves, a mapping table is constructed;

[0048] S33, according to the hierarchical feature vector group and benthic foraminifera oxygen isotope data Constructing similarity index based on structural similarity , select the best matching item, when and exceeds the threshold (set to 0.85), the age of the corresponding sedimentary unit x is assigned.

[0049] Optionally, the S31 includes:

[0050] S311, output profile reflection data Perform amplitude normalization and sliding window filtering to obtain a smooth reflection profile sequence ;

[0051] S312, calculate the local derivative response of the reflection profile to identify the sequence reflection interface and define the local extreme point sequence, which is expressed as:

[0052] ;

[0053] in, is the set of reflection interface positions, is the position of the i-th depth sampling point, is the derivative variable with respect to the depth dimension d, is the threshold for determining the reflection intensity change rate;

[0054] S313, based on the reflection interface position set , for each pair of adjacent reflection boundaries 、 Calculate the average amplitude of the profile segment between , peak spacing , texture roughness (variance) , Intra-layer reflection continuity coefficient (winding degree), forming each position The corresponding hierarchical feature vector group , where n is the number of profile reflection interfaces within the landslide area.

[0055] Optionally, the S32 includes:

[0056] S321, from m typical sampling points in the normal sediment layer Extract core samples and record their spatial coordinates and depth;

[0057] S322: Classify and identify benthic foraminifera in typical core samples, select the number of test species, and generate a benthic foraminifera combination vector. ;

[0058] S323, based on the regional geological database or stratigraphic standard atlas, obtain the geological age interval corresponding to each benthic foraminifera data to form a combination-age mapping relationship ,in, is the age interval of the corresponding combination;

[0059] S324, place the landslide horizon within the sedimentary age framework established by the normal sedimentary samples, with the time of the landslide event being the deposition time of the overlying samples (the underlying samples will be eroded by the occurrence of the landslide event);

[0060] For each fossil combination vector , sorted by species importance and indicativeness, and coded as benthic foraminifera oxygen isotope data , forming a lookup table mapping.

[0061] Beneficial effects of the present invention:

[0062] The present invention introduces a dynamic weight fusion mechanism for multi-source ocean exploration data. The system can fully utilize the spatial coverage and information complementarity of heterogeneous data such as multi-beam bathymetry, side-scan sonar and shallow stratum profiles, and realize adaptive weight allocation in different areas according to data confidence and spatial resolution, thereby effectively improving the three-dimensional restoration accuracy and local anomaly recognition capability in the process of constructing the seabed terrain model, and enhancing the ability to express landslide landform characteristics.

[0063] The present invention uses a synergistic mechanism of slope mutation detection, acoustic chaos identification, and shear wave velocity difference verification to accurately identify potential landslide boundaries and eliminate turbidity sediment interference zones. It further combines the turbidity transformation index driven by sonar energy to perform boundary correction, effectively improving the accuracy of landslide body identification and the geological rationality of the boundaries, and providing a stable spatial reference for subsequent sediment structure analysis.

[0064] The present invention establishes a matching mapping relationship between sedimentary sequence characteristics and microfossil assemblages within the landslide body, uses sparse core samples to establish an assembly-age mapping table, and introduces a similarity-driven sequence matching mechanism to achieve the age definition of non-continuous sedimentary units. Without relying on high-density core layout, it improves the ability to reconstruct the spatiotemporal evolution of deep-sea sedimentary events and the research efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0066] Figure 1 A flow chart of a definition method according to an embodiment of the present invention;

[0067] Figure 2 Schematic diagram of the dynamic weight fusion and terrain modeling process of multi-source seabed data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0068] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. Those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0069] like Figure 1-Figure 2 As shown in FIG, the method for identifying and dating submarine landslides based on big data analysis includes the following steps:

[0070] S1 acquires multi-source data, including seafloor multi-beam bathymetry data, side-scan sonar data, and shallow subsurface profile data. It then spatially registers and fuses these data using a dynamic weight allocation model to generate a three-dimensional seafloor terrain model. The dynamic weight allocation model automatically adjusts the weight coefficient based on the data confidence and spatial resolution.

[0071] S2, based on the three-dimensional seabed terrain model, the slope mutation area and the acoustic reflection chaotic area are extracted as the initial landslide boundary. The validity of the boundary is verified by combining the sediment shear wave velocity data. The turbidity current modification correction coefficient is introduced to modify the boundary. The area after eliminating the interference of turbidity current sedimentation is marked as the final landslide area.

[0072] S3: Extract the sedimentary sequence reflection characteristics from the shallow stratigraphic profile data within the final landslide area, construct a sequence feature vector group, and simultaneously obtain the benthic foraminifera oxygen isotope data of normal sedimentary core samples. Through detailed comparison with the existing standard stratigraphic oxygen isotope curve, establish a benthic foraminifera-age mapping table. Use the sequence matching algorithm to temporally and spatially associate the discontinuous sequence feature vectors with the mapping table, and output the age range of each sedimentary unit.

[0073] S1 includes:

[0074] S11, collects multi-beam bathymetric data, side-scan sonar data, and shallow subsurface profile data of the target sea area, corresponding to seabed topography, bottom reflection images, and sedimentary structure data, and pre-processes the collected multi-source data, including format unification, coordinate conversion, noise filtering, and outlier removal;

[0075] S12: Based on the spatial resolution and source confidence of multi-source data, a dynamic weight allocation model is constructed. Initial weights are set for different data sources, and weights are dynamically adjusted during the spatial position matching process. The coverage, signal-to-noise ratio, and detection accuracy of data within the same area are comprehensively considered to implement a regionally adaptive weight allocation strategy.

[0076] S13, based on the dynamic weight allocation results, performs spatial registration and fusion of multi-source data, and uses an interpolation reconstruction algorithm to map the spatially registered and fused multi-source data into a unified 3D geographic reference frame to generate a 3D seabed terrain model.

[0077] The S11 includes:

[0078] The S111 uses a multi-beam bathymetric system, side-scan sonar system, and shallow subsurface profiler carried on board the survey vessel to collect seabed topography, underwater reflection images, and sedimentary structure data of the target sea area. Specifically, the following data are collected:

[0079] The multi-beam bathymetric system obtains a discrete set of water depth points and outputs A three-dimensional coordinate set of , where 、 is the two-dimensional plane coordinate of the multi-beam measurement point i, is the sounding value of the multi-beam sounding system at point i, 3D point sets generated for multibeam bathymetry systems;

[0080] The side scan sonar system obtains the reflection intensity image along the track and outputs it as a grayscale matrix ;

[0081] The shallow layer profiler obtains underground reflector information and outputs it as a depth-distance profile matrix , where d is the depth;

[0082] S112: Unify the format of the collected multi-source data, build a standard data framework, and convert it into a unified reference coordinate system (such as WGS84 or UTM). The longitude and latitude of the center point are converted into a projected coordinate system, expressed as:

[0083] ;

[0084] ;

[0085] in, is the projection function, is the longitude and latitude, 、 To unify the horizontal and vertical coordinates in the projection coordinate system, is the standardized water depth value;

[0086] S113, the noise points in the multi-beam bathymetric data and side-scan sonar data are removed using the local median filtering method, which is expressed as:

[0087] ;

[0088] in, is the filtered depth value of point i, is the neighborhood point set around point i, is the median operation;

[0089] ;

[0090] in, for point The sonar grayscale value after filtering, For The image window neighborhood centered at is the horizontal offset of the image, is the offset in the vertical direction of the image;

[0091] S114, the Z-score method is used to identify and eliminate outliers in the sounding values ​​after noise removal, which is expressed as:

[0092] ;

[0093] in, is the depth value after filtering, 、 are the mean and standard deviation of the depth values ​​in the area, is the Z score, when When it is determined as an outlier and removed, is the abnormality identification threshold;

[0094] S115, storing the preprocessed multi-source data as a structured data set, specifically including:

[0095] The final output 3D point cloud model ;

[0096] The final output side scan sonar image ;

[0097] Final output profile reflection data ;

[0098] in, 、 are the horizontal coordinates and vertical projection coordinates in the unified projection coordinate system, is the effective water depth value.

[0099] S12 includes:

[0100] S121, based on the spatial resolution of multi-source data Source confidence , calculate the initial weight , expressed as:

[0101] ;

[0102] in, is the average spatial resolution of multi-source data, is the source confidence of multi-source data, 、 It is an adjustable parameter to control the influence of the two weights;

[0103] The source confidence of multi-source data is expressed as:

[0104] ;

[0105] in, is the equipment grade coefficient, is the data collection environment coefficient, is the data processing chain credibility coefficient, 、 、 is the weighting coefficient;

[0106] ;

[0107] in, is the actual measurement standard deviation of the equipment used, is the maximum measurement error within the acceptable range;

[0108] ;

[0109] in, For data sources The ratio of the number of effective sampling points to the total number of sampling points, is the noise ratio, which represents the ratio of the noise standard deviation to the signal mean, is the standard deviation of the background noise in the measurement, is the signal mean;

[0110] ;

[0111] in, is the missing rate, which indicates the proportion of missing data points, is the anomaly ratio, which indicates the proportion of outliers identified in valid data. For missing points, is the number of data points determined to be abnormal by rules (such as Z score);

[0112] S122, divide the target seabed area into small grid cells , calculate the coverage ratio of each data source in each cell , signal-to-noise ratio and detection stability , forming a comprehensive quality index within the region , expressed as:

[0113] ;

[0114] ;

[0115] in, For the unit The number of grid points containing valid observations, For unit The total number of all theoretical grid points to be fused within;

[0116] ;

[0117] in, is the mean value of the signal value of the data source in this unit, is the noise standard deviation of the data source in this unit, To prevent small positive numbers with denominators of 0;

[0118] ;

[0119] in, are consecutive sample values ​​arranged in order, To detect the degree of trajectory jump;

[0120] S123, according to each small grid unit The data sources in and , calculate the final fusion weight after dynamic adjustment in the region , expressed as:

[0121] ;

[0122] in, is the normalization factor, ensuring that the sum of all data source weights is 1;

[0123] S124, all grid cells of Organized by spatial position, the weight field matrix for the 3D terrain fusion process is generated.

[0124] S13 includes:

[0125] S131, convert the multi-beam bathymetric data, side-scan sonar data, and shallow subsurface profile data into the same 3D geographic coordinate system (such as the UTM projection system), and use rigid transformation to achieve multi-source data coordinate registration, which is expressed as:

[0126] ;

[0127] in, is the collection coordinate of the i-th point in the original data source, is the registration coordinate in the unified coordinate system after transformation, is the rotation matrix corresponding to the data source, is the translation vector;

[0128] S132, based on the final dynamic weight , perform weighted fusion of each data source in a unified coordinate grid, and calculate the fused bathymetric value, which is expressed as:

[0129] ;

[0130] in, is the fused depth value, For multi-source data in location The observed value of is the normalized fusion weight at that position in the corresponding weight field;

[0131] S133 uses the weighted inverse distance interpolation (IDW) algorithm to interpolate and reconstruct the discrete sampling points of the fused bathymetric values ​​to generate a three-dimensional seabed terrain model , expressed as:

[0132] ;

[0133] in, is the predicted depth value at the target interpolation point, is the depth value of the adjacent i-th sampling point, is the Euclidean distance between the target point and the i-th sampling point, p is the interpolation power index, N is the number of neighboring points involved in the interpolation, is the distance weighting coefficient;

[0134] ;

[0135] in, 、 are the horizontal coordinate and vertical coordinate of the j-th output point in the unified geographic reference system, For the interpolation algorithm at point The estimated depth or elevation value output at , M is the total number of interpolation output points.

[0136] S2 includes:

[0137] S21, based on a 3D seabed terrain model , calculate the slope value at each location , and identify the slope mutation threshold (Set to ) is taken as the slope mutation area, which is expressed as:

[0138] ;

[0139] in, 、 is the terrain aspect change rate calculated based on neighborhood differences;

[0140] S22, using the local texture variance of side-scan sonar images to detect chaotic reflection areas, the high-variance areas were identified as sedimentary disturbance zones, reflecting the acoustic discontinuity of the landslide accumulation area;

[0141] S23, the spatial overlap between the identified slope mutation zone and the chaotic reflection zone is used as the initial landslide boundary area, and the sediment shear wave velocity is extracted inside and outside the initial landslide boundary area. , calculate the shear wave velocity difference rate on both sides of the boundary , when the shear wave velocity change rate Exceeding the shear wave velocity change rate judgment threshold (set to 20%), it is considered that there is a stratigraphic difference at the boundary, the geological rationality of the boundary is verified, and the boundary that meets the requirements is retained. The boundary subset of the output is the shear wave velocity verification boundary , expressed as:

[0142] ;

[0143] in, is the average shear wave velocity in the initial landslide area, is the average shear wave velocity in the neighborhood outside the boundary;

[0144] S24, Verify Boundary Based on Output Shear Wave Velocity , introducing the turbidity transformation index , define the boundary correction factor And update the landslide boundary, and finally keep only The boundary points are output as the final landslide area, where =0.6 is the lower limit of the boundary, expressed as:

[0145] ;

[0146] ;

[0147] in, is the sonar chaos echo energy, is the total sonar energy.

[0148] S22 includes:

[0149] S221, in side scan sonar image On the other hand, a fixed-size two-dimensional sliding window ( ), with each center pixel Extract the grayscale value set of its neighborhood for the center, expressed as:

[0150] ;

[0151] in, For A grayscale window centered on is the side length of the sliding window, r is the radius, 、 is the offset coordinate relative to the center point;

[0152] S222, calculate the local variance of the gray value for each sliding window , generate the texture variance map, expressed as:

[0153] ;

[0154] ;

[0155] in, is the local texture variance at the center point, is the grayscale mean of the center point neighborhood, N is the side length of the sliding window, is the pixel position within the sliding window;

[0156] S223, set the chaotic reflection area determination threshold (set to 50), the local texture variance exceeds the judgment threshold The position is marked as chaotic area, and a binary mask is generated. , expressed as:

[0157] .

[0158] S3 includes:

[0159] S31, extract shallow stratum profile data in the final landslide area, identify the reflection interface and extract its sequence characteristics, and construct a sequence feature vector group ;

[0160] S32, extract typical locations in a core sample of normal sedimentation Oxygen isotope data of benthic foraminifera at , through detailed comparison with existing standard formation oxygen isotope curves, a mapping table is constructed;

[0161] S33, according to the hierarchical feature vector group and benthic foraminifera oxygen isotope data The similarity index is constructed based on the structural similarity of , select the best matching item, when and exceeds the threshold (set to 0.85), the age assigned to the corresponding sedimentary unit x is expressed as:

[0162] ;

[0163] ;

[0164] in, is the oxygen isotope data of the jth benthic foraminifera, is the era corresponding to position x, The age value that best matches the benthic foraminifera oxygen isotope combination.

[0165] S31 includes:

[0166] S311, output profile reflection data Perform amplitude normalization and sliding window filtering to obtain a smooth reflection profile sequence , expressed as:

[0167] ;

[0168] in, 、 are the mean and standard deviation of the original profile reflection data, It is a sliding mean operation;

[0169] S312, calculate the local derivative response of the reflection profile to identify the sequence reflection interface and define the local extreme point sequence, which is expressed as:

[0170] ;

[0171] in, is the set of reflection interface positions, is the position of the i-th depth sampling point, is the derivative variable with respect to the depth dimension d, =0.2 is the threshold for judging the reflection intensity change rate;

[0172] S313, based on the reflection interface position set , for each pair of adjacent reflection boundaries 、 Calculate the average amplitude of the profile segment between , peak spacing , texture roughness (variance) , Intra-layer reflection continuity coefficient (winding degree), forming each position The corresponding hierarchical feature vector group , where n is the number of profile reflection interfaces within the landslide area, specifically including:

[0173] ;

[0174] in, is the number of reflection points contained in the i-th sequence unit, For the location Depth The cross-sectional reflection intensity at ;

[0175] ;

[0176] in, 、 is the position of the adjacent reflection interface;

[0177] ;

[0178] in, is the average value of the reflection amplitude of the i-th layer, is the position of the j-th depth sampling point;

[0179] .

[0180] S32 includes:

[0181] S321, from m typical sampling points in the normal sediment layer Extract core samples and record their spatial coordinates and depth;

[0182] S322: Classify and identify benthic foraminifera in typical core samples, select the number of test species, and generate a benthic foraminifera combination vector. , expressed as:

[0183] ;

[0184] in, is the occurrence frequency of the kth foraminifera in the jth sample, , The total number of species identified for the target;

[0185] S323, based on the regional geological database or stratigraphic standard atlas, obtain the geological age interval corresponding to each benthic foraminifera data to form a combination-age mapping relationship ,in, is the age range of the corresponding combination, 、 are the lower limit and upper limit of the age range corresponding to the combination respectively;

[0186] S324, place the landslide horizon within the sedimentary age framework established by the normal sedimentary samples, with the time of the landslide event being the deposition time of the overlying samples (the underlying samples will be eroded by the occurrence of the landslide event);

[0187] For each fossil combination vector , sorted by species importance and indicativeness, and coded as benthic foraminifera oxygen isotope data , forming a lookup table mapping, expressed as:

[0188] ;

[0189] in, It is a combination-era mapping table used for subsequent matching.

[0190] The present invention encompasses any alternatives, modifications, equivalents, and solutions that fall within the spirit and scope of the present invention. To provide a thorough understanding of the present invention, specific details are described in detail below in connection with the preferred embodiments of the present invention, but those skilled in the art will be able to fully understand the present invention without these detailed descriptions. Furthermore, to avoid unnecessary confusion regarding the essence of the present invention, well-known methods, processes, procedures, components, and circuits have not been described in detail.

[0191] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for identifying and dating submarine landslides based on big data analysis, characterized by: The following steps are involved: S1 acquires multi-source data, including seafloor multi-beam bathymetry data, side-scan sonar data, and shallow subsurface profile data. It then spatially registers and fuses these data using a dynamic weight allocation model to generate a three-dimensional seafloor terrain model. The dynamic weight allocation model automatically adjusts the weight coefficient based on the data confidence and spatial resolution. S2, based on the three-dimensional seabed terrain model, extracts the slope mutation area and the acoustic reflection chaotic area as the initial landslide boundary. Combined with the sediment shear wave velocity data to verify the validity of the boundary, the turbidity current modification correction coefficient is introduced to modify the boundary. The area after eliminating the turbidity current sediment interference is marked as the final landslide area. Specifically, S21, based on a 3D seabed terrain model , calculate the slope value at each location , and identify the slope mutation threshold The continuous area is regarded as the slope mutation area; S22, using the local texture variance of side-scan sonar images to detect chaotic reflection areas, the high-variance areas were identified as sedimentary disturbance zones, reflecting the acoustic discontinuity of the landslide accumulation area; S23, the spatial overlap between the identified slope mutation zone and the chaotic reflection zone is used as the initial landslide boundary area, and the sediment shear wave velocity is extracted inside and outside the initial landslide boundary area. , calculate the shear wave velocity difference rate on both sides of the boundary , when the shear wave velocity change rate Exceeding the shear wave velocity change rate judgment threshold When the boundary is considered to have stratigraphic differences, the geological rationality of the boundary is verified and the boundary is retained to meet the requirements. The boundary subset of the output is the shear wave velocity verification boundary ; S24, Verify Boundary Based on Output Shear Wave Velocity , introducing the turbidity transformation index , define the boundary correction factor And update the landslide boundary, and finally keep only The boundary points are output as the final landslide area, where Keep a lower bound for the boundary; ； ； in, is the sonar chaos echo energy, is the total sonar energy; S3: Extract the sedimentary sequence reflection characteristics from the shallow stratigraphic profile data within the final landslide area, construct a sequence feature vector group, and simultaneously obtain the benthic foraminifera oxygen isotope data of normal sedimentary core samples. Through detailed comparison with the existing standard stratigraphic oxygen isotope curve, establish a benthic foraminifera-age mapping table. Use the sequence matching algorithm to temporally and spatially associate the discontinuous sequence feature vectors with the mapping table, and output the age range of each sedimentary unit.

2. The method for identifying and dating submarine landslides based on big data analysis according to claim 1, characterized in that: Said S1 comprises: S11, collects multi-beam bathymetric data, side-scan sonar data, and shallow subsurface profile data of the target sea area, corresponding to seabed topography, bottom reflection images, and sedimentary structure data, and pre-processes the collected multi-source data, including format unification, coordinate conversion, noise filtering, and outlier removal; S12: Based on the spatial resolution and source confidence of multi-source data, a dynamic weight allocation model is constructed. Initial weights are set for different data sources, and weights are dynamically adjusted during the spatial position matching process. The coverage, signal-to-noise ratio, and detection accuracy of data within the same area are comprehensively considered to implement a regionally adaptive weight allocation strategy. S13, based on the dynamic weight allocation results, performs spatial registration and fusion of multi-source data, and uses an interpolation reconstruction algorithm to map the spatially registered and fused multi-source data into a unified 3D geographic reference frame to generate a 3D seabed terrain model.

3. The method for identifying and dating submarine landslides based on big data analysis according to claim 2, characterized in that: The S11 includes: The S111 uses a multi-beam bathymetric system, side-scan sonar system, and shallow subsurface profiler carried on board the survey vessel to collect seabed topography, underwater reflection images, and sedimentary structure data of the target sea area. Specifically, the following data are collected: The multi-beam bathymetric system obtains a discrete set of water depth points and outputs A three-dimensional coordinate set of , where 、 is the two-dimensional plane coordinate of the multi-beam measurement point i, is the sounding value of the multi-beam sounding system at point i, 3D point sets generated for multibeam bathymetry systems; The side scan sonar system obtains the reflection intensity image along the track and outputs it as a grayscale matrix ; The shallow layer profiler obtains underground reflector information and outputs it as a depth-distance profile matrix , where d is the depth; S112, performing format unification operations on the collected multi-source data, constructing a standard data framework, converting it into a unified reference coordinate system, and converting the longitude and latitude of the center point into a projected coordinate system; S113, removing noise points in the multi-beam bathymetric data and side-scan sonar data using a local median filtering method; S114, using the Z score method to identify and eliminate outliers in the sounding values ​​after noise removal; S115, storing the preprocessed multi-source data as a structured data set, specifically including: The final output 3D point cloud model ; The final output side scan sonar image ; Final output profile reflection data ; in, 、 are the horizontal coordinates and vertical projection coordinates in the unified projection coordinate system, is the effective water depth value.

4. The method for identifying and dating submarine landslides based on big data analysis according to claim 3 is characterized in that: The S12 includes: S121, based on the spatial resolution of multi-source data Source confidence , calculate the initial weight ; S122, divide the target seabed area into small grid cells , calculate the coverage ratio of each data source in each cell , signal-to-noise ratio and detection stability , forming a comprehensive quality index within the region ; S123, according to each small grid unit The data sources in and , calculate the final fusion weight after dynamic adjustment in the region ; S124, all grid cells of Organized by spatial position, the weight field matrix for the 3D terrain fusion process is generated.

5. The method for identifying and dating submarine landslides based on big data analysis according to claim 4 is characterized in that: The S13 includes: S131, converts multi-beam bathymetric data, side-scan sonar data, and shallow subsurface profile data into a unified 3D geographic coordinate system, and uses rigid transformation to achieve multi-source data coordinate registration; S132, based on the final dynamic weight , perform weighted fusion of each data source in a unified coordinate grid and calculate the fused bathymetric value; S133 uses the weighted inverse distance interpolation algorithm to interpolate and reconstruct the discrete sampling points of the fused bathymetric values ​​to generate a three-dimensional seabed terrain model. .

6. The method for identifying and dating submarine landslides based on big data analysis according to claim 5, characterized in that: The S22 includes: S221, in side scan sonar image A fixed-size two-dimensional sliding window is selected, with each center pixel Extract the grayscale value set of its neighborhood for the center; S222, calculate the local variance of the gray value for each sliding window , generate texture variance map; S223, set the chaotic reflection area determination threshold , the local texture variance exceeds the judgment threshold The position is marked as chaotic area, and a binary mask is generated. , expressed as: 。 7. The method for identifying and dating submarine landslides based on big data analysis according to claim 6, characterized in that: The S3 includes: S31, extract shallow stratum profile data in the final landslide area, identify the reflection interface and extract its sequence characteristics, and construct a sequence feature vector group ; S32, extract typical locations in a core sample of normal sedimentation Oxygen isotope data of benthic foraminifera at , through detailed comparison with existing standard formation oxygen isotope curves, a mapping table is constructed; S33, according to the hierarchical feature vector group and benthic foraminifera oxygen isotope data The similarity index is constructed based on the structural similarity of , select the best matching item, when and exceeds the threshold , assign an age to the corresponding sedimentary unit x.

8. The method for identifying and dating submarine landslides based on big data analysis according to claim 7 is characterized in that: The S31 includes: S311, output profile reflection data Perform amplitude normalization and sliding window filtering to obtain a smooth reflection profile sequence ; S312, calculate the local derivative response of the reflection profile to identify the sequence reflection interface and define the local extreme point sequence, which is expressed as: ； in, is the set of reflection interface positions, is the position of the i-th depth sampling point, is the derivative variable with respect to the depth dimension d, is the threshold for determining the reflection intensity change rate; S313, based on the reflection interface position set , for each pair of adjacent reflection boundaries 、 Calculate the average amplitude of the profile segment between , peak spacing , texture roughness , Intra-layer reflection continuity coefficient , forming each position The corresponding hierarchical feature vector group , where n is the number of profile reflection interfaces within the landslide area.

9. The method for identifying and dating submarine landslides based on big data analysis according to claim 8, characterized in that: The S32 includes: S321, from m typical sampling points in the normal sediment layer Extract core samples and record their spatial coordinates and depth; S322: Classify and identify benthic foraminifera in typical core samples, select the number of test species, and generate a benthic foraminifera combination vector. ; S323, based on the regional geological database or stratigraphic standard atlas, obtain the geological age interval corresponding to each benthic foraminifera data to form a combination-age mapping relationship ,in, is the age interval of the corresponding combination; S324, placing the landslide body horizon into the sedimentary age framework established by the normal sedimentary samples, with the time of the landslide event being the deposition time of the overlying samples; For each fossil combination vector , sorted by species importance and indicativeness, and coded as benthic foraminifera oxygen isotope data , forming a lookup table mapping.

Images