Multi-source heterogeneous spatio-temporal big data mining method for beach investigation monitoring

Abstract

The application discloses a multi-source heterogeneous spatio-temporal big data mining method for beach investigation monitoring, and relates to the technical field of big data mining and geographic spatial information processing. The method extracts a beach measured elevation profile and a historical tide line to construct an orthogonal coordinate system of a shoreline, projects heterogeneous topographic elevation, wave flow time series and sediment parameters to grid nodes to establish a multi-source spatio-temporal characteristic grid, extracts elevation differences and average particle size gradients of adjacent nodes, calculates a sediment transport vector in combination with a wave flow sequence, constructs a beach erosion and deposition power spatio-temporal distribution matrix according to a magnetic susceptibility characteristic, correlates and mines a historical elevation evolution sequence with the matrix, calculates an erosion and deposition state transition probability, iteratively generates a future beach three-dimensional terrain evolution grid with the sediment transport vector as a mass exchange constraint, extracts an elevation undercut erosion block for reverse particle tracking, locates a sediment blockage source, and outputs a beach maintenance parameter set containing three-dimensional sand supplement coordinates, backfill volume and matching particle size.

Description

Technical Field

[0001] This invention relates to the fields of big data mining and geospatial information processing technology, specifically a method for mining multi-source heterogeneous spatiotemporal big data for beach surveys and monitoring. Background Technology

[0002] Beaches, as vital coastal natural resources, play a fundamental role in disaster prevention and mitigation as well as ecosystem maintenance. Affected by the combined effects of global climate change, extreme weather, and frequent human engineering activities, beaches in many areas exhibit a complex evolutionary pattern of alternating erosion and material coarsening. To achieve scientific restoration and sustainable management of coastlines, obtaining accurate dynamics of beach evolution, long-term, precise investigation and dynamic monitoring of the complex and ever-changing beach environment have become crucial aspects of natural resource geological environment management.

[0003] In conventional beach surveys and monitoring, existing technologies largely rely on periodic artificial topographic mapping, local trench sampling, and laboratory-level sediment grain size testing and geochemical analysis. However, actual beach evolution is coupled with high-frequency dynamic hydrological environments such as waves and tidal currents, and low-frequency static geological properties. Existing methods struggle to overcome spatial baseline biases caused by alternating wet and dry conditions and data acquisition frequency discrepancies when processing multi-source heterogeneous mapping and monitoring data. This results in a disconnect between topographic elevation, hydrodynamic timing, and sediment grain size parameters. Furthermore, existing beach evolution early warning models typically rely on two-dimensional empirical formulas or post-hoc statistical induction, which are insufficient for cross-modal physical evolution mapping between hydrodynamic forces and microscopic sand grain sorting, heavy mineral enrichment, and sedimentation. When faced with variable sea conditions and tidal inputs, these models are prone to three-dimensional spatial extrapolation biases regarding topographic erosion and sedimentation, making it difficult to accurately trace the dynamic blocking points of erosion and directly output quantitative replenishment and maintenance parameters suitable for real beach restoration projects. Therefore, it is still necessary to provide a multi-source heterogeneous spatiotemporal big data mining method for beach surveys and monitoring, so as to improve the accuracy of beach topographic evolution prediction and the scientific implementation of coastal restoration decisions. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a multi-source heterogeneous spatiotemporal big data mining method for beach surveys and monitoring to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a multi-source heterogeneous spatiotemporal big data mining method for beach survey and monitoring, comprising the following steps: S1, extracting measured elevation profiles and historical tide lines from the beach to construct an orthogonal coordinate system of the shoreline, projecting heterogeneous topographic elevation points, high-frequency wave-current time series, and sediment average grain size, sorting coefficient, and magnetic susceptibility values ​​onto the grid nodes corresponding to the orthogonal coordinate system of the shoreline, and establishing a multi-source spatiotemporal feature grid; S2, extracting the elevation difference and average grain size gradient of adjacent nodes in the multi-source spatiotemporal feature grid, calculating the sediment transport vector in combination with the wave-current time series, identifying high-density heavy mineral siltation points and low-density light mineral transport points based on the magnetic susceptibility values ​​in the grid nodes, and constructing a spatiotemporal distribution matrix of beach erosion and deposition dynamics; S3, extracting from the multi-source spatiotemporal feature grid... The historical elevation evolution sequence is correlated with the spatiotemporal distribution matrix of beach erosion and deposition dynamics. The probability of erosion and deposition state transition for each grid node under different wave and tidal input conditions is calculated. The sediment transport vector is used as a quantitative constraint for sediment volume exchange between adjacent nodes. The elevation incision and deposition of each node are iteratively updated under the future wave and tidal prediction sequence to generate a three-dimensional beach topographic evolution grid for future time. S4. Erosion blocks with elevation incision greater than a set threshold are extracted from the three-dimensional beach topographic evolution grid. The erosion blocks are traced in reverse based on the sediment transport vector to locate the source of sediment blockage that causes beach erosion. Based on the volume loss of the erosion block and the original sediment sorting coefficient, a set of beach maintenance parameters including three-dimensional replenishment coordinates, backfill volume and matching particle size is output.

[0006] Preferably, the process of constructing an orthogonal shoreline coordinate system by extracting measured elevation profiles of the beach and historical tide lines, and projecting heterogeneous topographic elevation points, high-frequency wave-current time series, and sediment average grain size, sorting coefficient, and magnetic susceptibility values ​​onto the corresponding grid nodes of the orthogonal shoreline coordinate system is as follows: Extracting historical high tide and low tide lines to determine the intertidal zone wet-dry transition boundary; using the measured elevation profile perpendicular to the coastline as a reference, establishing an orthogonal shoreline coordinate system that is horizontally perpendicular to the coastline and vertically parallel to the coastline; obtaining topographic elevation points, sediment average grain size, sorting coefficient, and magnetic susceptibility values ​​for different observation periods, and mapping them to discrete spatial nodes of the orthogonal shoreline coordinate system through spatial interpolation to generate a static geological attribute layer; extracting wave direction, wave height, and velocity parameters from the high-frequency wave-current time series, extracting tidal wet-dry transition timestamps based on the frequency of tidal level changes, and attaching the wave direction, wave height, and velocity parameters containing the wet-dry transition timestamps to the corresponding discrete spatial nodes to generate dynamic hydrological time series characteristics of the grid nodes.

[0007] Preferably, the specific process of establishing a multi-source spatiotemporal feature grid is as follows: synchronizing the timestamp sequences of static geological attribute layers and dynamic hydrological time-series features at the same discrete spatial node, and aligning the sampling frequencies of heterogeneous data; extracting the frequency-aligned topographic elevation points, average sediment grain size, sorting coefficient, and magnetic susceptibility values, and combining them to generate a multi-dimensional geological spatial feature vector; fusing the multi-dimensional geological spatial feature vector with the dynamic hydrological time-series features, and constructing a multi-source spatiotemporal feature grid containing three-dimensional spatial coordinates and one-dimensional time-series parameters in the shoreline orthogonal coordinate system.

[0008] Preferably, the specific process of extracting the elevation difference and average grain size gradient of adjacent nodes in the multi-source spatiotemporal feature grid and calculating the sediment transport vector in combination with the wave-current time series is as follows: In the multi-source spatiotemporal feature grid, extract the elevation difference and average grain size gradient of adjacent nodes along the direction perpendicular to the coastline, and calculate the sediment grain size variation gradient in the direction from the beach shoulder to the low tide line; extract the elevation difference and average grain size gradient of adjacent nodes along the direction parallel to the coastline, and extract the differences in the spatial distribution of sediments under the action of coastal currents; extract the maximum tidal velocity vector and the constant wave direction vector in the high-frequency wave-current time series, and calculate them by vector superposition with the sediment grain size variation gradient and the differences in the spatial distribution of sediments to obtain the direction and intensity of sediment spatial transport, which is recorded as the sediment transport vector.

[0009] Preferably, the specific process for identifying high-density heavy mineral siltation points and low-density light mineral transport points based on the magnetic susceptibility values ​​in the grid nodes and constructing the spatiotemporal distribution matrix of beach erosion and deposition dynamics is as follows: Extract the spatial distribution sequence of magnetic susceptibility values ​​from the multi-source spatiotemporal feature grids; identify grid nodes with magnetic susceptibility values ​​greater than the high-value envelope threshold and mark them as high-density heavy mineral siltation points; identify grid nodes with magnetic susceptibility values ​​less than the low-value envelope threshold and mark them as low-density light mineral transport points; map the high-density heavy mineral siltation points to the beach wave erosion incision area, and map the low-density light mineral transport points to the beach sediment deposition area; extract the sediment transport vector, combine the spatial coordinate parameters of the beach wave erosion incision area and the beach sediment deposition area, calculate the erosion and deposition dynamic parameters of different grid nodes on the beach under temporal evolution, and generate the spatiotemporal distribution matrix of beach erosion and deposition dynamics.

[0010] Preferably, the specific process of extracting historical elevation evolution sequences from multi-source spatiotemporal feature grids, associating them with the spatiotemporal distribution matrix of beach erosion and deposition dynamics, and calculating the erosion and deposition state transition probability of each grid node under different wave and tidal input conditions is as follows: extract elevation node data from multi-source spatiotemporal feature grids with different sampling periods, and reorganize them according to the time dimension to generate historical elevation evolution sequences containing the intertidal wet-dry alternation time states; spatially map the historical elevation evolution sequences with the spatiotemporal distribution matrix of beach erosion and deposition dynamics, extract the mapping relationship between wave and tidal dynamic evolution parameters and the elevation increase or decrease of grid nodes, and generate associated feature vectors; statistically analyze the historical erosion and deposition transition frequencies under different wave and tidal conditions in the associated feature vectors, and calculate the erosion and deposition state transition probability of each grid node evolving from the current state to the erosion downcutting state or the deposition uplift state.

[0011] Preferably, the specific process of generating a future-temporal three-dimensional beach topographic evolution grid by iteratively updating the elevation incision and sedimentation of each node under the future wave and tidal current prediction sequence, using the sediment transport vector as a quantitative constraint for sediment volume exchange between adjacent nodes, is as follows: Extract the flow direction and velocity parameters from the sediment transport vector, calculate the material exchange threshold between adjacent grid nodes, and use it as a quantitative constraint for coastal sediment volume exchange; import the future wave and tidal current prediction sequence, combine the scour and sedimentation state transition probability and the material conservation evolution boundary to perform spatiotemporal evolution calculation, and iteratively update the elevation incision and sedimentation of each node; superimpose the iteratively updated elevation incision and sedimentation onto the spatial coordinate system of the initial multi-source spatiotemporal feature grid to reconstruct the three-dimensional spatial morphology and generate a future-temporal three-dimensional beach topographic evolution grid.

[0012] Preferably, the specific process of locating the source of sediment blockage leading to beach erosion by comparing and extracting erosion blocks with elevation downcutting greater than a set threshold in the three-dimensional topographic evolution grid of the beach, and performing reverse particle tracking on the erosion blocks based on sediment transport vectors is as follows: Compare the node elevation changes of the initial multi-source spatiotemporal feature grid and the three-dimensional topographic evolution grid of the beach, extract continuous spatial nodes with elevation downcutting greater than a set threshold, and cluster them to generate erosion blocks; extract sediment transport vectors, establish a reverse particle tracking path along the opposite direction of coastal sediment transport, and calculate the upstream material supply path of the erosion blocks; calculate the coastal sediment transport dynamic attenuation parameter in the reverse particle tracking path, and extract the spatial coordinate nodes with attenuation parameters exceeding a preset attenuation threshold as the source of sediment blockage leading to beach erosion.

[0013] Preferably, the specific process of outputting a beach maintenance parameter set including three-dimensional replenishment coordinates, backfill volume, and matching grain size based on the volume loss of the erosion block and the original sediment sorting coefficient is as follows: Extract the three-dimensional topological boundary of the erosion block, calculate the envelope volume of the elevation incision difference within the three-dimensional topological boundary by integration, generate the backfill volume, and extract the center projection coordinates of the erosion block as the three-dimensional replenishment coordinates; extract the original sediment sorting coefficient and average grain size of the erosion block before erosion evolution, perform gradation mapping by combining the sediment spatial distribution gradient under the lateral sorting action of waves, and calculate the matching grain size required for backfill; and construct the beach maintenance parameter set by multi-dimensional parameter correlation of the three-dimensional replenishment coordinates, backfill volume, and matching grain size.

[0014] The technical effects and advantages of this invention are as follows: (1) In the technical solution described in this invention, in the initial stage of data processing, the measured elevation profile of the beach and the historical tide line are extracted to construct an orthogonal coordinate system of the shoreline. The heterogeneous topographic elevation points, high-frequency wave-current time series and sediment multidimensional numerical features are projected onto the corresponding grid nodes to establish a multi-source spatiotemporal feature grid. Subsequently, the elevation difference and average grain size gradient of adjacent nodes are extracted, and the sediment transport vector is calculated in combination with the wave-current time series. The heavy mineral siltation points and light mineral transport points are identified according to the magnetic susceptibility values, and then the spatiotemporal distribution matrix of beach erosion and deposition dynamics is constructed. As a result, the spatial misalignment deviation of multi-source data caused by the alternation of wet and dry conditions on the beach can be overcome. The macroscopic wave-current evolution and microscopic sand grain transport are coupled and physically quantified under a unified spatiotemporal reference, solving the problem of the separation between topographic elevation, hydrodynamic time series and bottom sediment parameters in the existing methods, and ensuring the spatial alignment accuracy and dynamic characterization accuracy of the bottom excavation data.

[0015] (2) In the technical solution described in this invention, the historical elevation evolution sequence and the spatiotemporal distribution matrix of beach erosion and deposition dynamics are extracted and correlated during the temporal evolution stage. The erosion and deposition state transition probability of each grid node is calculated, and the sediment transport vector is used as a quantitative constraint for sediment volume exchange. The three-dimensional topographic evolution grid of the beach in the future time state is generated by iteratively updating under the wave and current prediction sequence. Finally, the erosion blocks with extracted elevation incision are compared, and the source of sediment blockage is located by reverse particle tracking based on the sediment transport vector. The beach maintenance parameter set including three-dimensional replenishment coordinates, backfill volume and matching particle size is output by combining the volume loss and the original sediment sorting coefficient. Thus, it can overcome the limitation of traditional models in mapping cross-modal physical mechanisms, ensure that the three-dimensional topographic erosion and deposition simulation process follows the law of material conservation, improve the reliability of geomorphic deformation prediction under variable sea conditions, and directly open up the decision-making link from evolution early warning to erosion source tracing and quantitative replenishment maintenance, thereby enhancing the adaptability and on-site implementation of the coastal restoration plan.

[0016] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0017] Figure 1 This is a flowchart of the multi-source heterogeneous spatiotemporal big data mining method for beach survey and monitoring according to the present invention; Figure 2 This is a schematic diagram illustrating the bimodal clustering threshold division of sediment magnetic susceptibility in an embodiment of the present invention; Figure 3 This is a schematic diagram of the three-dimensional topographic evolution and sediment transport vector space grid of the beach in an embodiment of the present invention. Detailed Implementation

[0018] This application's embodiments address the issues of insufficient fusion of beach monitoring and survey data and the lack of scientific maintenance guidance in geomorphological evolution projection through a multi-source heterogeneous spatiotemporal big data mining method for beach surveys and monitoring.

[0019] The overall approach of the scheme in this application is as follows: First, an orthogonal coordinate system for the shoreline adapted to the alternating wet and dry characteristics of the intertidal zone is constructed to unify the spatiotemporal benchmark of heterogeneous geological mapping and high-frequency hydrological monitoring data. On this basis, the dynamic linkage mechanism between elevation gradient, micro-mineral sedimentary characteristics and macro-wave current sequence is explored in depth to quantify the distribution of sediment erosion and deposition on the real beach. Then, based on the historical evolution sequence and dynamic matrix, the node state transition trend is calculated. Under the physical constraints of sediment volume conservation and exchange, the three-dimensional topographic evolution grid driven by the future hydrological sequence is iteratively calculated. Finally, the deformation block is subjected to reverse dynamic backtracking to find the root causes of local erosion, and the refined beach backfilling and maintenance execution parameters that guide the actual restoration project are calculated accordingly.

[0020] Example 1; please refer to Figure 1This invention provides a technical solution: a multi-source heterogeneous spatiotemporal big data mining method for beach surveys and monitoring, comprising the following steps: S1, extracting measured elevation profiles and historical tide levels from the beach to construct an orthogonal coordinate system of the shoreline, projecting heterogeneous topographic elevation points, high-frequency wave-current time series, and sediment average grain size, sorting coefficient, and magnetic susceptibility values ​​onto the grid nodes corresponding to the orthogonal coordinate system of the shoreline, and establishing a multi-source spatiotemporal feature grid; S2, extracting the elevation difference and average grain size gradient of adjacent nodes in the multi-source spatiotemporal feature grid, calculating the sediment transport vector by combining the wave-current time series, identifying high-density heavy mineral siltation points and low-density light mineral transport points based on the magnetic susceptibility values ​​in the grid nodes, and constructing a spatiotemporal distribution matrix of beach erosion and deposition dynamics; S3, extracting historical data from the multi-source spatiotemporal feature grid. The elevation evolution sequence is correlated with the spatiotemporal distribution matrix of beach erosion and deposition dynamics. The probability of erosion and deposition state transition for each grid node under different wave and tidal input conditions is calculated. The sediment transport vector is used as a quantitative constraint for sediment volume exchange between adjacent nodes. The elevation incision and deposition of each node are iteratively updated under the future wave and tidal prediction sequence to generate a three-dimensional beach topographic evolution grid for the future time. S4. Erosion blocks with elevation incision greater than a set threshold are extracted from the three-dimensional beach topographic evolution grid. The erosion blocks are traced in reverse based on the sediment transport vector to locate the source of sediment blockage that causes beach erosion. Based on the volume loss of the erosion block and the original sediment sorting coefficient, a set of beach maintenance parameters including three-dimensional replenishment coordinates, backfill volume and matching particle size is output.

[0021] In this implementation plan, step S1 primarily addresses the issues of inconsistent sampling frequencies and spatial misattribution of beach topographic, hydrological, and geological mapping data. The established shoreline orthogonal coordinate system refers to a two-dimensional grid reference system established parallel and perpendicular to the coastline, which eliminates the interference of the intertidal wet-dry cycle on data mapping. Topographic elevation points are low-frequency, static spatial attribute data, while wave-current time series are high-frequency, dynamic hydrological observation data. Sediment average grain size, sorting coefficient, and magnetic susceptibility values ​​are geological sampling and measurement data characterizing the physical properties and mineral composition of sand grains. The sorting coefficient reflects the uniformity of grain size, and the magnetic susceptibility represents the enrichment of magnetic heavy minerals in the sediment. Projecting heterogeneous data with completely different sampling frequencies and data structures onto the same orthogonal grid node achieves alignment of macroscopic hydrodynamic parameters and microscopic geological parameters under the same spatiotemporal reference, forming the foundational data for subsequent correlation mining.

[0022] Step S2 is used to quantify the physical driving relationship between hydrodynamics and sediment transport on the beach surface in the actual beach environment. The average grain size gradient represents the continuous spatial variation trend of sand grain size between adjacent spatial nodes. The system uses elevation difference and wave-current time series to calculate the sediment transport vector characterizing the direction and intensity of spatial sediment transport. Since hydrodynamics has a lateral sorting effect on sand grains in the scour zone, low-density light particles are easily transported and lost by ocean currents, while high-density heavy minerals and other fine particles are easily concentrated and accumulated in the erosion area. Therefore, based on the distribution of magnetic susceptibility values ​​in the grid, it is possible to distinguish and mark erosion and degradation points where heavy minerals are enriched and deposited, as well as nodes where light minerals are transported and deposited. Through the fusion calculation of relevant parameters, a data matrix reflecting the dynamic strength and scour and deposition distribution of all areas of the beach is finally constructed, providing boundary constraints based on physical laws for subsequent topographic evolution inference.

[0023] Step S3 is used to extrapolate and generate predictions of 3D geographic environmental deformation in future timeframes. The probability of erosion-deposition transition refers to the mathematical probability parameter obtained through statistical mining of historical elevation change patterns, indicating whether a specific grid node will transition from its current state to an erosion-incision state or a deposition-uplift state. Based on the established probabilistic correlation prediction, the sediment transport vector calculated in step S2 is used as a quantitative constraint on material exchange between adjacent grid nodes, establishing a boundary for material conservation evolution between grids and preventing numerical abrupt changes in nodes that deviate from basic geophysical principles due to purely probabilistic mining. Under the boundary of material conservation evolution, future wave and tidal data are input, and the system iteratively calculates the elevation incision and deposition of each grid at each spatiotemporal step, obtaining the vertical erosion depth or uplift height of the beach, and outputting a dynamic evolution grid that includes 3D topographic changes.

[0024] Step S4 is used to complete the operational support transformation from topographic evolution data prediction to actual coastal zone restoration engineering schemes. Erosion blocks refer to continuous spatial patches in the predicted grid where the elevation incision depth exceeds a preset safety threshold. Reverse particle tracing refers to establishing a spatial tracing path in the opposite direction of coastal sediment transport calculated in step S2, tracing back to find coordinate nodes where upstream dynamic material replenishment exhibits abrupt attenuation, and locating the source of sediment blockage caused by artificial dikes or abnormal wave currents leading to beach erosion. After locating the source, the system calculates the required backfill volume based on the actual volume loss of the erosion block, and calculates the matching grain size that can resist on-site hydrodynamic scouring by referring to the original sediment sorting coefficient of the eroded area. Finally, it outputs a set of beach maintenance parameters containing specific three-dimensional replenishment coordinates, filling volume, and required sand grain specifications to guide the actual beach restoration construction operations.

[0025] Specifically, the process of constructing an orthogonal coordinate system for the shoreline by extracting measured elevation profiles of the beach and historical tide levels, and projecting heterogeneous topographic elevation points, high-frequency wave-current time series, and sediment average grain size, sorting coefficient, and magnetic susceptibility values ​​onto the corresponding grid nodes of the orthogonal coordinate system is as follows: The historical high tide and low tide lines are extracted to determine the intertidal zone wet-dry transition boundary. Using the measured elevation profile perpendicular to the coastline as a reference, an orthogonal coordinate system is established that is horizontally perpendicular to the coastline and vertically parallel to the coastline. Topographic elevation points, sediment average grain size, sorting coefficient, and magnetic susceptibility values ​​from different observation periods are obtained and mapped to discrete spatial nodes of the orthogonal coordinate system through spatial interpolation to generate a static geological attribute layer. Wave direction, wave height, and velocity parameters are extracted from the high-frequency wave-current time series. Tidal wet-dry transition timestamps are extracted based on the frequency of tidal level changes. Wave direction, wave height, and velocity parameters containing the wet-dry transition timestamps are then attached to the corresponding discrete spatial nodes to generate dynamic hydrological time series characteristics of the grid nodes.

[0026] In this implementation scheme, when establishing an orthogonal coordinate system for the shoreline, the measured vertical coastline elevation profile is used as the X-axis, the average fitted direction of historical high tide and low tide lines is used as the Y-axis, and the elevation direction is used as the Z-axis to establish a spatial discrete benchmark. For topographic elevation points, average sediment grain size, sorting coefficient, and magnetic susceptibility values ​​with low acquisition frequency, a spatial interpolation algorithm with shoreline anisotropy is used to map them to the corresponding grid nodes. Let the set of known geological sampling points be P, and for any discrete spatial node to be interpolated... Its static geological attribute values The interpolation process is represented as follows: ; ; In the formula, K represents the total number of known geological sampling points; This represents the measured static geological property value of the k-th known geological sampling point; This represents the spatial distribution weight of the k-th known geological sampling point to be interpolated. This represents the anisotropy correction distance between the node to be interpolated and the k-th known geological sampling point; Indicates the distance decay index; This represents the Cartesian coordinates of the k-th known geological sampling point; Indicates the spatial variation coefficient in the direction perpendicular to the coastline; This represents the spatial variation coefficient along the direction parallel to the coastline. The spatial variation coefficient is... and The determination method is as follows: extract the semivariance functions of historical beach geological exploration data in the directions perpendicular to and parallel to the coastline, respectively. Use the inverse of the range when the semivariance function reaches the sill value as the coefficient of variation for the corresponding direction, so that the attribute interpolation smoothness in the direction parallel to the coastline is higher than that in the direction perpendicular to the coastline. After completing the static geological attribute mapping, obtain the high-frequency wave-current time series. Based on the dynamic curve of tide level evolution over time given by the local tide gauge station, calculate the critical time stamps at which the elevation of the discrete spatial node is submerged and exposed. Assign the wave direction, wave height, and current velocity parameters between two adjacent critical time stamps to the node, forming a dynamic hydrological time series characteristic truncated by the wet-dry transition boundary.

[0027] Specifically, the process of establishing a multi-source spatiotemporal feature grid is as follows: Synchronize the timestamp sequences of static geological attribute layers and dynamic hydrological time-series features at the same discrete spatial nodes, and align the sampling frequencies of heterogeneous data; extract the frequency-aligned topographic elevation points, average sediment grain size, sorting coefficient, and magnetic susceptibility values, and combine them to generate a multi-dimensional geological spatial feature vector; fuse the multi-dimensional geological spatial feature vector with the dynamic hydrological time-series features, and construct a multi-source spatiotemporal feature grid containing three-dimensional spatial coordinates and one-dimensional time-series parameters in the shoreline orthogonal coordinate system.

[0028] In this implementation scheme, when synchronizing the timestamp sequences of static geological attribute layers and dynamic hydrological time-series features, a cubic spline interpolation algorithm is used to resample and align the truncated dynamic hydrological time-series features to the reference time mining frequency, addressing the sampling frequency differences of heterogeneous data. Subsequently, the frequency-aligned spatial data of each dimension are extracted, and normalization and nonlinear activation processing are performed when constructing the multidimensional feature vector of geological space. Let the time be... Discrete space nodes The corresponding topographic elevation points, average sediment grain size, sorting coefficient, and magnetic susceptibility values ​​constitute a feature subset. The set of wave direction, wave height, and flow velocity parameters after synchronization frequency constitutes a feature subset. When constructing a multi-source spatiotemporal feature grid, the nodes at time... Comprehensive spatiotemporal feature tensor The solution process is expressed as follows: ; In the formula, to Representing feature subsets respectively The topographic elevation points, average sediment grain size, sorting coefficient, and magnetic scalar characteristics are represented by numerical values ​​for magnetic scalars. to Representing feature subsets respectively Numerical scalars representing wave direction, wave height, and flow velocity characteristics; Represents a static attribute vector mapping matrix; Represents the dynamic hydrological vector mapping matrix; This represents the tensor cascade concatenation operator; Indicates time The activation factor for the submerged water depth at the lower tide level; Indicates time Dynamic tidal level elevation; Represents discrete space nodes The topographic elevation values ​​of the nodes; This represents the scale constant of the substrate hydrodynamic attenuation. Specifically, the scale constant of the substrate hydrodynamic attenuation... The determination method is as follows: extract the vertical relative water depth difference from the historical hydrodynamic scour records of the current beach environment when the bottom shear stress decays to half of the critical starting shear force, and calibrate it as the bottom hydrodynamic decay scale constant. Through the above spatial matrix transformation and nonlinear modulation of the tidal submerged water depth activation factor, a multi-source spatiotemporal feature grid with both three-dimensional spatial geometric coordinate position distribution and one-dimensional time hydrological evolution sequence characteristics is generated.

[0029] Specifically, the process of extracting the elevation difference and average grain size gradient of adjacent nodes in the multi-source spatiotemporal feature grid and calculating the sediment transport vector in combination with the wave-current time series is as follows: In the multi-source spatiotemporal feature grid, the elevation difference and average grain size gradient of adjacent nodes are extracted along the direction perpendicular to the coastline, and the sediment grain size variation gradient in the direction from the beach shoulder to the low tide line is calculated; the elevation difference and average grain size gradient of adjacent nodes are extracted along the direction parallel to the coastline, and the differences in the spatial distribution of sediments under the action of coastal currents are extracted; the maximum tidal velocity vector and the constant wave direction vector in the high-frequency wave-current time series are extracted, and they are vector superimposed with the sediment grain size variation gradient and the differences in the spatial distribution of sediments to obtain the direction and intensity of sediment spatial transport, which is denoted as the sediment transport vector.

[0030] In this implementation scheme, within a multi-source spatiotemporal feature grid, along the grid's principal axis perpendicular to the coastline, the absolute elevation difference between spatially adjacent nodes and the measured difference in average sediment grain size are extracted grid by grid. The actual topographic slope distance along the route is obtained by taking the square root of the sum of the squares of the elevation difference and the projected distance. The measured difference in average sediment grain size is divided by this actual topographic slope distance to calculate the sediment grain size variation gradient from the beach shoulder to the low tide line. Similarly, the same difference calculation process is performed along the grid's principal axis parallel to the coastline to extract the spatial distribution differences of sediments under the influence of coastal currents. After obtaining the above-mentioned spatial gradient characteristics of geological sediments in the horizontal and vertical directions, the maximum current velocity vector and the constant wave direction vector of a specific tidal cycle in the high-frequency wave current time series are retrieved, and these are superimposed with the grain size variation gradient and spatial distribution differences to calculate the vector superposition of physical kinetic energy levels, thus obtaining the direction and intensity of sediment spatial transport. Specifically, for any discrete spatial node... Its sediment transport vector The derivation and calculation process is expressed as follows: In the formula, This represents the density constant of nearshore seawater; The sediment response weighting coefficient represents the direction perpendicular to the coastline. This represents the gradient of sediment grain size variation at discrete spatial nodes in the direction from the beach shoulder to the low tide line; A unit reference vector representing the direction perpendicular to the coastline; The sediment response weighting coefficient represents the direction parallel to the coastline. This represents the spatial distribution differences of sediments at discrete spatial nodes under the influence of coastal currents. A unit reference vector representing the direction parallel to the coastline; Indicates the conversion rate of shear stress in the subbed; This represents the vector representing the maximum velocity of the tidal current. This represents the constant wave direction vector. Among them, the sediment response weighting coefficient... and The method for determining the sediment transport vector is as follows: Measured cross-sectional and longitudinal profile data of the historical erosion and deposition evolution of the target beach are extracted. The partial least squares regression coefficients between the actual change in beach elevation per unit time span and the purely hydrodynamic calculated change in elevation within the measured profiles are used as weighting coefficients for the corresponding directions. Through this vector superposition calculation, the strength of macroscopic wave current hydrodynamics is combined with the microscopic characteristics of sand grain sorting to generate a sediment transport vector characterizing the direction and intensity of spatial sediment transport.

[0031] Specifically, the process of identifying high-density heavy mineral siltation points and low-density light mineral transport points based on the magnetic susceptibility values ​​in the grid nodes and constructing the spatiotemporal distribution matrix of beach erosion and deposition dynamics is as follows: Extract the spatial distribution sequence of magnetic susceptibility values ​​from the multi-source spatiotemporal feature grids; identify grid nodes with magnetic susceptibility values ​​greater than the high-value envelope threshold and mark them as high-density heavy mineral siltation points; identify grid nodes with magnetic susceptibility values ​​less than the low-value envelope threshold and mark them as low-density light mineral transport points; map the high-density heavy mineral siltation points to the beach wave erosion incision area, and map the low-density light mineral transport points to the beach sediment deposition area; extract the sediment transport vector, combine the spatial coordinate parameters of the beach wave erosion incision area and the beach sediment deposition area, calculate the erosion and deposition dynamic parameters of different grid nodes on the beach under temporal evolution, and generate the spatiotemporal distribution matrix of beach erosion and deposition dynamics.

[0032] In this implementation scheme, the spatial distribution sequence of magnetic susceptibility values, representing the magnetic mineral content of sediments, is extracted from the grid nodes. A Gaussian mixture model is used to perform bimodal clustering decomposition on the spatial distribution sequence of magnetic susceptibility values. The result of subtracting one standard deviation from the mean of the Gaussian component representing the peak value of heavy minerals is defined as the high-value envelope threshold, and the result of adding one standard deviation to the mean of the Gaussian component representing the peak value of light minerals is defined as the low-value envelope threshold. The multi-source spatiotemporal feature grid is traversed, and nodes with magnetic susceptibility values ​​greater than the high-value envelope threshold are marked as high-density heavy mineral stagnation points. Based on the depositional pattern of wave-driven currents preferentially transporting low-density coarse particles and stagnating high-density fine particles, these nodes are spatially mapped to the beach wave erosion incision region. Similarly, nodes with magnetic susceptibility values ​​less than the low-value envelope threshold are marked as low-density light mineral transport points, and their spatial mapping is defined as the beach sediment deposition region. Based on this, the erosion and deposition dynamic parameters of different grid nodes on the beach are calculated under temporal evolution by combining the sediment transport vector with the spatial coordinate parameters of the divided erosion and deposition regions. Specifically, for any discrete space node Its erosion and deposition dynamic parameters The calculation process is expressed as follows: In the formula, This represents the region category mapping indicator function. When the discrete spatial node belongs to the beach wave erosion incision region, the value is a positive erosion coefficient, and when the discrete spatial node belongs to the beach sedimentation environment siltation region, the value is a negative siltation coefficient. This represents the orthogonal decomposition components of the sediment transport vector in the direction parallel to the coastline; This represents the orthogonal decomposition components of the sediment transport vector in the direction perpendicular to the coastline; This represents the measured magnetic susceptibility value of the discrete spatial node; This represents the background average magnetic susceptibility across the entire beach survey and monitoring area. Erosion and sedimentation dynamic parameters are calculated by iterating through all nodes. This involves generating a spatiotemporal distribution matrix of beach erosion and deposition dynamics with temporal evolution attributes and spatial dynamic intensity distribution within an orthogonal coordinate system along the shoreline. Combined with... Figure 2 As shown, Figure 2 The bar chart in the figure represents the frequency distribution of actual magnetic susceptibility sampling for each node in the multi-source spatiotemporal feature grid. The two smooth curves superimposed on this represent the low-density light mineral component (left peak) and the high-density heavy mineral component (right peak), respectively, generated by fitting a Gaussian mixture model. The two vertical dashed lines in the figure correspond to the calculated low-value envelope threshold and high-value envelope threshold, respectively. Through these two envelope threshold lines, the system performs physical hard truncation of nodes in the spatial grid: node regions located to the right of the high-value envelope threshold exhibit heavy mineral enrichment, mapped to the incised area of ​​beach wave erosion; node regions located to the left of the low-value envelope threshold exhibit light mineral accumulation, mapped to the siltation area of ​​a beach sedimentary environment.

[0033] Specifically, the process of extracting historical elevation evolution sequences from multi-source spatiotemporal feature grids, associating them with the spatiotemporal distribution matrix of beach erosion and deposition dynamics, and calculating the probability of erosion and deposition state transition for each grid node under different wave and tidal input conditions is as follows: Elevation node data from different sampling periods in the multi-source spatiotemporal feature grids are extracted and reorganized along the time dimension to generate historical elevation evolution sequences containing the intertidal wet-dry alternation timescale; the historical elevation evolution sequences are spatially mapped to the spatiotemporal distribution matrix of beach erosion and deposition dynamics, and the mapping relationship between wave and tidal dynamic evolution parameters and the elevation increase / decrease of grid nodes is extracted to generate associated feature vectors; the frequency of historical erosion and deposition transitions under different wave and tidal conditions in the associated feature vectors is statistically analyzed, and the probability of erosion and deposition state transition for each grid node in its current state to either erosion downcutting or deposition uplift is calculated.

[0034] In this implementation scheme, elevation node data from different sampling periods are extracted within a multi-source spatiotemporal feature grid. These data are then arranged and recombined chronologically to construct a historical elevation evolution sequence encompassing the intertidal wet-dry transition. This historical elevation evolution sequence is spatially mapped to corresponding nodes in the spatiotemporal distribution matrix of beach erosion and deposition dynamics, aligning the elevation changes at the same spatial coordinates with wave and tidal dynamic evolution parameters at the same time, generating a multi-dimensional correlation feature vector. This correlation feature vector is analyzed to statistically determine the frequency of historical elevation transitions between erosion and deposition under specific wave and tidal input conditions. For any grid node... The probability of its transition from the current state to the erosion-incision state, and the probability of its transition to the scouring-deposition state. The calculation process is expressed as follows: In the formula, The probability of a grid node transitioning to a scouring / deposition state as it evolves into an erosion / incision state is represented by ; A and B represent the two-dimensional discrete spatial coordinates of the grid node, respectively; f represents the historical observation time window index. This represents the total number of historical observation time windows; This represents the change in elevation within the f-th historical observation time window; This represents a status indicator function. When the change in elevation is negative, the output is 1; otherwise, the output is 0. This represents the spatiotemporal distribution parameters of beach erosion and deposition dynamics within the f-th historical observation time window; Indicates the dynamic response scaling factor; This represents the baseline probability calibration term. Among them, the dynamic response scaling factor... The method for determining the coefficient is as follows: calculate the Pearson correlation coefficient between the historical elevation change of global grid nodes and the corresponding scouring and silting dynamic parameters, and extract the absolute value of the Pearson correlation coefficient as the dynamic response scaling factor.

[0035] Specifically, using sediment transport vectors as quantitative constraints for sediment volume exchange between adjacent nodes, the process of iteratively updating the elevation incision and sedimentation of each node under future wave and tidal current prediction sequences to generate a future-temporal three-dimensional beach topographic evolution grid is as follows: Extract the flow direction and velocity parameters from the sediment transport vectors, calculate the material exchange threshold between adjacent grid nodes, and use it as a quantitative constraint for coastal sediment volume exchange; import the future wave and tidal current prediction sequences, combine the scouring and sedimentation state transition probability and the material conservation evolution boundary to perform spatiotemporal evolution calculations, and iteratively update the elevation incision and sedimentation of each node; superimpose the iteratively updated elevation incision and sedimentation onto the spatial coordinate system of the initial multi-source spatiotemporal feature grid to reconstruct the three-dimensional spatial morphology and generate a future-temporal three-dimensional beach topographic evolution grid.

[0036] In this implementation scheme, the flow direction and velocity parameters in the sediment transport vector are extracted, and their normal projection components between adjacent grid nodes are defined as the material flow direction. A future wave and tidal current prediction sequence is imported, and based on the topological connectivity of adjacent nodes and the probability of scour and deposition state transition, the spatiotemporal evolution calculation of the spatial material transfer volume is performed. For any grid node... The combined change in elevation shear and sedimentation at future time step g. The calculation process is expressed as follows: In the formula, This represents the combined change in elevation incision and sedimentation at prediction time step g; i and j represent the spatial horizontal and vertical coordinates of the evolving grid, respectively. This represents the probability of scouring and silting transition corresponding to the grid node; This represents the set of adjacent topological nodes of the grid node; o represents the adjacent node in the set of adjacent topological nodes; This represents the input sediment transport vector flowing from adjacent nodes into the current grid node; This represents the inward unit normal vector of the current mesh node; This represents the threshold for material exchange between adjacent grid nodes, i.e., the quantitative constraint on the volume exchange of sediment along the coast. This represents the output sediment transport vector flowing from the current grid node to adjacent nodes; This represents the outward unit normal vector of the current mesh node; This parameter represents the time step size for iterative solution. This represents the projected base area of ​​a single grid node in an orthogonal coordinate system. The material exchange threshold is also included. The method for determining the maximum stable scour depth limit is as follows: Extract the maximum stable scour depth limit recorded in the historical measured profile data of the beach area, multiply it by the porosity of the sand sediments and the projected bottom area, and calculate the maximum allowable sediment loss volume per grid, which is the material exchange threshold. After completing the iteration of all prediction time steps, the comprehensive change values ​​are superimposed on the elevation values ​​of the initial grid spatial coordinate system to complete the three-dimensional spatial morphology reconstruction and generate a three-dimensional topographic evolution grid of the beach for future timeframes.

[0037] Specifically, the process of comparing and extracting erosion blocks in the three-dimensional topographic evolution grid of the beach where the elevation downcut exceeds a set threshold, and locating the source of sediment blockage leading to beach erosion by performing reverse particle tracking based on sediment transport vectors is as follows: Compare the node elevation changes of the initial multi-source spatiotemporal feature grid and the three-dimensional topographic evolution grid of the beach, extract continuous spatial nodes with elevation downcut exceeding a set threshold, and cluster them to generate erosion blocks; extract sediment transport vectors, establish reverse particle tracking paths along the opposite direction of coastal sediment transport, and calculate the upstream material supply path of the erosion blocks; calculate the coastal sediment transport dynamic attenuation parameters in the reverse particle tracking path, and extract the spatial coordinate nodes whose attenuation parameters exceed a preset attenuation threshold as the source of sediment blockage leading to beach erosion.

[0038] In this implementation scheme, corresponding nodes are extracted from the initial multi-source spatiotemporal feature grid and the beach 3D topographic evolution grid, and node elevation difference calculation is performed. Continuous spatial nodes with elevation incision exceeding a set threshold are extracted and aggregated into erosion blocks using a density-based spatial clustering algorithm. Sediment transport vectors are extracted at the upstream boundary of the erosion blocks, and a reverse particle tracking path is iteratively generated along the opposite direction of this vector at a fixed grid resolution step size. During the evolution of the reverse particle tracking path, the dynamic attenuation characteristics of the path nodes are extracted to locate the source of blockage. For any node step w on the tracking path, its coastal sediment transport dynamic attenuation parameters are... The solution process is expressed as follows: In the formula, This represents the coastal sediment transport dynamic attenuation parameter at tracking step w; This represents the sediment transport vector at tracking step w+1; This represents the sediment transport vector at the tracking step w; This represents the regularization minimum constant term to prevent the denominator from being zero; Indicates the sensitivity coefficient to local terrain undulations; This represents the spatial elevation difference between the grid nodes corresponding to tracking step w+1 and tracking step w. The preset attenuation threshold is determined by extracting the distribution sequence of coastal sediment transport dynamics attenuation parameters from the entire beach grid under historical normal sea conditions, and taking the sum of the mean and three standard deviations of this distribution sequence as the preset attenuation threshold. This will be used during the reverse tracking process. The first spatial coordinate node exceeding the preset attenuation threshold is extracted and marked as the source of sediment blockage. Combined with... Figure 3 As shown, Figure 3 The 3D surface at the bottom represents the evolution of topographic elevation in an orthogonal coordinate system (X-axis parallel to the coastline, Y-axis perpendicular to the coastline). Areas where the surface color shows a concave trend are the erosion incision blocks extracted by the system. The 3D black arrows floating above these erosion incision blocks represent the extracted sediment transport vectors. This vector array not only indicates the spatial transport direction of sediment in 3D space, but its magnitude also represents the transport intensity. Based on this sediment transport vector array, the system traces particles in reverse order from the topological boundary of the erosion block, moving in the opposite direction of the arrows. Figure 3 The vector distribution in the data intuitively presents the dynamic decay process of the upstream material supply path, and finally locates the spatial coordinate node of the abrupt change in decay parameter as the source of sediment blockage.

[0039] Specifically, based on the volume loss of the erosion block and the original sediment sorting coefficient, the process of outputting a beach maintenance parameter set including three-dimensional replenishment coordinates, backfill volume, and matching grain size is as follows: Extract the three-dimensional topological boundary of the erosion block, integrate and calculate the elevation incision difference envelope volume within the three-dimensional topological boundary to generate the backfill volume, and extract the center projection coordinates of the erosion block as the three-dimensional replenishment coordinates; extract the original sediment sorting coefficient and average grain size of the erosion block before erosion evolution, combine the sediment spatial distribution gradient under the lateral sorting action of waves to perform gradation mapping, and calculate the matching grain size required for backfill; correlate the three-dimensional replenishment coordinates, backfill volume, and matching grain size using multi-dimensional parameters to construct the beach maintenance parameter set.

[0040] In this implementation plan, the three-dimensional topological boundary point set of the erosion block is extracted, a spatial closed surface is constructed, and the discrete surface volume integral operator is applied to the elevation evolution difference of each grid node within the closed surface to calculate the backfill volume required to restore the block. The center projection coordinates of the erosion block are extracted as the three-dimensional sand replenishment coordinates for the construction operation. Simultaneously, the sorting coefficient and average grain size attributes of the original sediments of the erosion block before topographic evolution are extracted, and combined with the hydrodynamic lateral sorting parameters of the wave scour zone, scour-resistant grain size mapping is performed. For the discrete node index corresponding to the erosion block... Its backfill volume Matching particle size The association calculation process is represented as follows: ; In the formula, Indicates the required backfill volume for the eroded area; This represents the set of grid nodes covered by the eroded block; and These represent the horizontal and vertical indices of the spatial grid within the grid node set, respectively. This represents the baseline elevation in the initial multi-source spatiotemporal feature grid; This represents the predicted future elevation in a 3D topographic evolution grid of a beach. Represents the planar projected area of ​​a single grid node; Indicates the matching particle size of the gradation mapping output; Indicates the average grain size of the original sediment; Indicates the original sediment sorting coefficient; This indicates the spatial distribution gradient of sediments at this node under the lateral sorting effect of waves; Represents the empirical gradient normalization constant; This indicates the erosion resistance armoring compensation item for the subgrade. Specifically, the erosion resistance armoring compensation item for the subgrade... The determination method is as follows: A sample set of average grain size differences between residual surface sediments in historical beach erosion areas and sediments in adjacent stable areas is statistically analyzed, and the 75th percentile of this sample set is extracted as the seabed erosion resistance armoring compensation term. After completing the gradation mapping, the extracted three-dimensional replenishment coordinates and backfill volume are used... Matching particle size Combined into a structured data dictionary, a set of beach maintenance parameters is generated.

[0041] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for mining multi-source heterogeneous spatiotemporal big data for beach surveys and monitoring, characterized in that, Includes the following steps: S1. Extract the measured elevation profile of the beach and the historical tide line to construct the shoreline orthogonal coordinate system. Project the heterogeneous topographic elevation points, high-frequency wave current time series and sediment average grain size, sorting coefficient and magnetic susceptibility values ​​onto the grid nodes corresponding to the shoreline orthogonal coordinate system to establish a multi-source spatiotemporal feature grid. S2. Extract the elevation difference and average particle size gradient of adjacent nodes in the multi-source spatiotemporal feature grid, calculate the sediment transport vector by combining the wave-current time series, identify high-density heavy mineral siltation points and low-density light mineral transport points based on the magnetic susceptibility values ​​in the grid nodes, and construct the spatiotemporal distribution matrix of beach erosion and deposition dynamics. S3. Extract the historical elevation evolution sequence from the multi-source spatiotemporal feature grid, correlate it with the spatiotemporal distribution matrix of beach scour and deposition dynamics, calculate the scour and deposition state transition probability of each grid node under different wave and tidal input conditions, use the sediment transport vector as the quantitative constraint for sediment volume exchange between adjacent nodes, iteratively update the elevation downcut and deposition of each node under the future wave and tidal prediction sequence, and generate a three-dimensional beach topographic evolution grid for future time. S4. Compare and extract erosion blocks in the three-dimensional topographic evolution grid of the beach where the elevation downcut exceeds the set threshold. Based on the sediment transport vector, perform reverse particle tracking on the erosion blocks to locate the source of sediment blockage that causes beach erosion. Based on the volume loss of the erosion blocks and the original sediment sorting coefficient, output a set of beach maintenance parameters including three-dimensional sand replenishment coordinates, backfill volume and matching particle size.

2. The multi-source heterogeneous spatiotemporal big data mining method for beach survey and monitoring according to claim 1, characterized in that: The specific process of extracting measured elevation profiles of the beach and historical tide lines to construct an orthogonal coordinate system for the shoreline, and projecting heterogeneous topographic elevation points, high-frequency wave current time series, and sediment average grain size, sorting coefficient, and magnetic susceptibility values ​​onto the corresponding grid nodes of the orthogonal coordinate system is as follows: Historical high tide and low tide lines are extracted to determine the wet and dry boundary of the intertidal zone. Based on the measured elevation profile perpendicular to the coastline, an orthogonal coordinate system of the coastline is established that is horizontally perpendicular to the coastline and longitudinally parallel to the coastline. Topographic elevation points, average sediment grain size, sorting coefficient, and magnetic susceptibility values ​​for different observation periods are obtained and mapped to discrete spatial nodes of the shoreline orthogonal coordinate system through spatial interpolation to generate a static geological attribute layer. Wave direction, wave height, and velocity parameters are extracted from the high-frequency wave-current time series. Tidal wet-dry alternation timestamps are extracted based on the frequency of tidal level changes. Wave direction, wave height, and velocity parameters containing wet-dry alternation timestamps are then attached to the corresponding discrete spatial nodes to generate dynamic hydrological time series features of the grid nodes.

3. The multi-source heterogeneous spatiotemporal big data mining method for beach survey and monitoring according to claim 2, characterized in that: The specific process of establishing a multi-source spatiotemporal feature grid is as follows: Synchronize the timestamp sequences of static geological attribute layers and dynamic hydrological temporal characteristics at the same discrete spatial node, and align the sampling frequencies of heterogeneous data. Extract frequency-aligned topographic elevation points, average sediment grain size, sorting coefficient, and magnetic susceptibility values, and combine them to generate a multi-dimensional feature vector of geological space. By integrating multi-dimensional feature vectors of geological space with dynamic hydrological time-series features, a multi-source spatiotemporal feature grid containing three-dimensional spatial coordinates and one-dimensional time-series parameters is constructed under the orthogonal coordinate system of the shoreline.

4. The multi-source heterogeneous spatiotemporal big data mining method for beach survey and monitoring according to claim 1, characterized in that: The specific process of extracting the elevation difference and average particle size gradient of adjacent nodes in a multi-source spatiotemporal feature grid, and combining it with wave-current time series to calculate sediment transport vectors is as follows: In the multi-source spatiotemporal feature grid, the elevation difference and average grain size gradient of adjacent nodes are extracted along the direction perpendicular to the coastline, and the sediment grain size variation gradient in the direction from the beach shoulder to the low tide line is calculated. The elevation difference and average grain size gradient of adjacent nodes were extracted along the parallel coastline to extract the spatial distribution differences of sediments under the action of coastal currents. The maximum tidal velocity vector and the constant wave direction vector are extracted from the high-frequency wave current time series. They are then superimposed with the sediment grain size variation gradient and the differences in sediment spatial distribution to calculate the direction and intensity of sediment spatial transport, which are denoted as sediment transport vectors.

5. The multi-source heterogeneous spatiotemporal big data mining method for beach survey and monitoring according to claim 1, characterized in that: The specific process of identifying high-density heavy mineral stagnation points and low-density light mineral transport points based on the magnetic susceptibility values ​​in the grid nodes, and constructing the spatiotemporal distribution matrix of beach erosion and deposition dynamics, is as follows: Extract the spatial distribution sequence of magnetic susceptibility values ​​from the multi-source spatiotemporal feature grid, identify grid nodes with magnetic susceptibility values ​​greater than the high-value envelope threshold and mark them as high-density heavy mineral stagnation points, and identify grid nodes with magnetic susceptibility values ​​less than the low-value envelope threshold and mark them as low-density light mineral transport points. High-density heavy mineral siltation points are mapped to beach wave erosion incision areas, and low-density light mineral transport points are mapped to beach sediment deposition areas. Extract sediment transport vectors, combine spatial coordinate parameters of the beach wave erosion incision area and the beach sediment deposition area, calculate the erosion and deposition dynamic parameters of different grid nodes on the beach under temporal evolution, and generate the spatiotemporal distribution matrix of beach erosion and deposition dynamics.

6. The multi-source heterogeneous spatiotemporal big data mining method for beach survey and monitoring according to claim 1, characterized in that: The specific process of extracting historical elevation evolution sequences from multi-source spatiotemporal feature grids, correlating them with the spatiotemporal distribution matrix of beach scour and deposition dynamics, and calculating the scour and deposition state transition probability of each grid node under different wave and tidal input conditions is as follows: Elevation node data with different sampling periods are extracted from multi-source spatiotemporal feature grids and reorganized according to the time dimension to generate a historical elevation evolution sequence containing the intertidal wet-dry alternation time. The historical elevation evolution sequence is spatially mapped to the spatiotemporal distribution matrix of beach erosion and deposition dynamics. The mapping relationship between wave and tide dynamic evolution parameters and the increase or decrease of grid node elevation is extracted to generate associated feature vectors. The historical scour-deposition transition frequency under different wave and tidal conditions is statistically correlated in the feature vector, and the scour-deposition state transition probability of each grid node evolving from the current state to the erosion downcutting state or the siltation uplift state is calculated.

7. The multi-source heterogeneous spatiotemporal big data mining method for beach survey and monitoring according to claim 1, characterized in that: Using sediment transport vectors as a quantitative constraint for sediment volume exchange between adjacent nodes, the specific process of iteratively updating the elevation downcut and sedimentation of each node under the future wave and tidal current prediction sequence to generate a future temporal 3D beach topographic evolution grid is as follows: Extract the flow direction and velocity parameters from the sediment transport vector, calculate the material exchange threshold between adjacent grid nodes, and use it as a quantitative constraint for the volume exchange of sediment along the coast. Import future wave and tidal current prediction sequences, combine the probability of scour and sedimentation state transition with the material conservation evolution boundary to perform spatiotemporal evolution calculations, and iteratively update the elevation downcut and sedimentation of each node; The updated elevation downcut and sedimentation are superimposed onto the spatial coordinate system of the initial multi-source spatiotemporal feature grid to reconstruct the three-dimensional spatial morphology and generate a three-dimensional beach topographic evolution grid for future timeframes.

8. The multi-source heterogeneous spatiotemporal big data mining method for beach survey and monitoring according to claim 1, characterized in that: The specific process of comparing and extracting erosion blocks with elevation downcutting exceeding a set threshold in the 3D topographic evolution grid of the beach, and locating the source of sediment blockage leading to beach erosion by performing inverse particle tracking based on sediment transport vectors is as follows: By comparing the node elevation changes of the initial multi-source spatiotemporal feature grid and the beach three-dimensional topographic evolution grid, continuous spatial nodes with elevation downcutting amount greater than a set threshold are extracted and clustered to generate erosion blocks. Extract sediment transport vectors, establish reverse particle tracking paths along the opposite direction of sediment transport along the coast, and calculate the upstream material supply paths of the erosion block; In the reverse particle tracking path, the attenuation parameters of the coastal sediment transport dynamics are calculated, and the spatial coordinate nodes whose attenuation parameters exceed the preset attenuation threshold are extracted as the source locations of sediment blockage that cause beach erosion.

9. The multi-source heterogeneous spatiotemporal big data mining method for beach survey and monitoring according to claim 1, characterized in that: The specific process of outputting a set of beach maintenance parameters, including three-dimensional replenishment coordinates, backfill volume, and matching grain size, based on the volume loss of the eroded block and the original sediment sorting coefficient is as follows: Extract the three-dimensional topological boundary of the eroded block, calculate the envelope volume of the elevation incision difference within the three-dimensional topological boundary by integration, generate the backfill volume, and extract the center projection coordinates of the eroded block as the three-dimensional sand replenishment coordinates. Extract the sorting coefficient and average grain size of the original sediments in the erosion block before erosion evolution, combine them with the spatial distribution gradient of sediments under the action of wave lateral sorting to perform gradation mapping, and calculate the matching grain size required for backfilling. By correlating the three-dimensional sand replenishment coordinates, backfill volume, and matching particle size with multi-dimensional parameters, a set of beach maintenance parameters is constructed.

Images