A method, device, equipment, medium and product for identifying paleofloods

Abstract

The application discloses a palaeoflood identification method, device, equipment, medium and product, relates to the palaeoflood field, and the method comprises the following steps: acquiring granularity distribution data and content ratio of each sediment sample; performing EMM processing on the granularity distribution data of all sediment samples to obtain an end member intensity matrix; the end member intensity matrix comprises intensity values of each potential end member in each sediment sample; constructing an end member matrix of the potential end member according to the intensity values of the potential end member in each sediment sample in the end member intensity matrix; obtaining a to-be-refined end member according to the correlation of each potential end member and a ratio matrix; performing EMM processing on the end member matrix of the to-be-refined end member by taking the ratio matrix as prior knowledge to obtain a high-fidelity flood end member matrix; performing processing on the high-fidelity flood end member matrix by adopting an iterative three-classification threshold method to obtain a flood identification threshold; and completing palaeoflood identification according to the threshold. The application improves the accuracy, reliability and repeatability of palaeoflood identification.

Description

Technical Field

[0001] This application relates to the field of paleofloods, and in particular to a method, apparatus, equipment, medium, and product for identifying paleofloods. Background Technology

[0002] Extreme flood events pose a serious threat to socio-economic and ecological environments. Due to the lack of long-term instrumental hydrological records, reconstructing high-precision paleoflood sequences is crucial for understanding flood patterns and assessing future risks.

[0003] Current technologies primarily rely on sediment grain size or geochemical indicators to identify paleofloods. However, these technologies suffer from the following key technical problems: 1. Grain size effect interference: Numerous studies have shown that in fluvial sedimentary environments, sediment grain size significantly influences its geochemical composition, a phenomenon known as the "grain size effect." Current technologies typically analyze grain size and geochemical indicators as two separate modules, or use only the latter for simple verification, failing to effectively address the inherent interference of grain size on geochemical signals, thus affecting the accuracy of flood identification.

[0004] 2. Subjectivity in flood identification: Currently known techniques mostly use static thresholds based on simple statistics such as "mean + standard deviation" to classify flood layers. The threshold setting of this method has a certain degree of subjectivity, and for complex sedimentary sequences, the reliability and repeatability of the identification results are poor.

[0005] Therefore, there is an urgent need for a paleoflood identification method that can avoid interference from granularity effects and subjectivity, in order to improve the accuracy, reliability and repeatability of paleoflood identification. Summary of the Invention

[0006] The purpose of this application is to provide a method, apparatus, equipment, medium, and product for paleoflood identification that can avoid interference from granularity effects and subjective paleoflood identification methods, thereby improving the accuracy, reliability, and repeatability of paleoflood identification.

[0007] To achieve the above objectives, this application provides the following solution: In the first aspect, this application provides a method for identifying ancient floods, including: continuously collecting multiple sediment samples from borehole cores in the target area.

[0008] Obtain the grain size distribution data and the content ratio of each sediment sample; the content ratio is the ratio of the content of the target flood-sensitive chemical element to the content of the target reference stationary chemical element.

[0009] EMM processing was performed on the grain size distribution data of all sediment samples to obtain the endmember intensity matrix; the endmember intensity matrix includes the intensity value of each potential endmember in each sediment sample.

[0010] For any potential endmember, the endmember matrix of the potential endmember is constructed based on the intensity values ​​of the potential endmember in each sediment sample as described in the endmember intensity matrix.

[0011] Based on the correlation coefficient values ​​of the endmember matrix and the ratio matrix of each potential endmember, endmembers to be refined are selected from each potential endmember; the ratio matrix includes the content ratio of all sediment samples.

[0012] Using the ratio matrix as prior knowledge, EMM processing was performed on the grain size distribution data of all sediment samples to obtain a high-fidelity flood end-member matrix. The high-fidelity flood end-member matrix includes the intensity values ​​of the end-members to be refined in each sediment sample.

[0013] An iterative three-class classification threshold method was used to process the high-fidelity flood end-member matrix to obtain the flood identification threshold.

[0014] The sediment samples corresponding to the target intensity values ​​in the high-fidelity flood end-member matrix are identified as paleoflood samples, thus completing the paleoflood identification; the target intensity value is the intensity value that is greater than the flood identification threshold.

[0015] Secondly, this application provides an ancient flood identification device, including: a sample collection module for continuously collecting multiple sediment samples from borehole cores in a target area.

[0016] The acquisition module is used to acquire the grain size distribution data and the content ratio of each sediment sample; the content ratio is the ratio of the content of the target flood-sensitive chemical element to the content of the target reference stationary chemical element.

[0017] The end-member determination module is used to perform EMM processing on the grain size distribution data of all sediment samples to obtain an end-member intensity matrix. The end-member intensity matrix includes the intensity value of each potential end-member in each sediment sample. For any potential end-member, the end-member matrix of the potential end-member is constructed based on the intensity value of the potential end-member in each sediment sample as described in the end-member intensity matrix. The end-members to be refined are screened from each potential end-member based on the correlation coefficient value between the end-member matrix and the ratio matrix of each potential end-member. The ratio matrix includes the content ratio of all sediment samples.

[0018] The high-fidelity flood end-member matrix determination module is used to perform EMM processing on the grain size distribution data of all sediment samples using the ratio matrix as prior knowledge to obtain the high-fidelity flood end-member matrix. The high-fidelity flood end-member matrix includes the intensity values ​​of the end-members to be refined in each sediment sample.

[0019] The flood identification threshold determination module is used to process the high-fidelity flood end-member matrix using an iterative three-classification threshold method to obtain the flood identification threshold.

[0020] The paleoflood identification module is used to identify sediment samples corresponding to target intensity values ​​in the high-fidelity flood end-member matrix as paleoflood samples, thus completing the paleoflood identification; the target intensity value is the intensity value that is greater than the flood identification threshold.

[0021] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the ancient flood identification method described above.

[0022] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the paleoflood identification method described above.

[0023] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the paleoflood identification method described above.

[0024] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides a method, device, equipment, medium, and product for paleoflood identification. A ratio matrix is ​​constructed based on the ratio of the content of the target flood-sensitive chemical element to the content of the target baseline immobile chemical element. Based on the grain size distribution data of all sediment samples, endmembers to be refined are obtained. Using the ratio matrix as prior knowledge, the endmember matrix of the endmember to be refined is subjected to EMM processing to obtain a high-fidelity flood endmember matrix. This corrects the interference of grain size effect on geochemical signals, realizes the constraint and refinement of element content on grain size modeling, and achieves deep fusion between data, thereby improving the accuracy of paleoflood identification.

[0025] This application replaces the traditional static threshold with an iterative three-class threshold method to identify flood events in an objective and adaptive manner, thereby improving the reliability and repeatability of the identification results. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This is a flowchart illustrating an ancient flood identification method provided in one embodiment of this application.

[0028] Figure 2 This is a schematic diagram of the functional modules of an ancient flood identification device provided in an embodiment of this application.

[0029] Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0030] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0031] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0032] In one exemplary embodiment, such as Figure 1 As shown, a method for identifying ancient floods is provided, including: Step 201: Continuously collecting multiple sediment samples from borehole cores in the target area.

[0033] Step 202: Obtain the grain size distribution data and the content ratio of each sediment sample; the content ratio is the ratio of the content of the target flood-sensitive chemical element to the content of the target reference stationary chemical element.

[0034] Step 203: Perform EMM processing on the grain size distribution data of all sediment samples to obtain the endmember intensity matrix; the endmember intensity matrix includes the intensity value of each potential endmember in each sediment sample.

[0035] Step 204: For any potential endmember, construct the endmember matrix of the potential endmember based on the intensity values ​​of the potential endmember in each sediment sample as described in the endmember intensity matrix.

[0036] Step 205: Based on the correlation coefficient values ​​of the endmember matrix and ratio matrix of each potential endmember, screen the endmembers to be refined from each potential endmember; the ratio matrix includes the content ratio of all sediment samples.

[0037] Step 206: Using the ratio matrix as prior knowledge, perform EMM processing on the grain size distribution data of all sediment samples to obtain a high-fidelity flood end-member matrix; the high-fidelity flood end-member matrix includes the intensity values ​​of the end-members to be refined in each sediment sample.

[0038] Step 207: The high-fidelity flood endmember matrix is ​​processed using an iterative three-class classification threshold method to obtain the flood identification threshold.

[0039] Step 208: Identify the sediment samples corresponding to the target intensity values ​​in the high-fidelity flood end-member matrix as paleoflood samples to complete paleoflood identification; the target intensity value is an intensity value greater than the flood identification threshold.

[0040] In practical applications, the grain size distribution data of each sediment sample is obtained as follows: a first preset number of sediment samples are selected as elemental samples, and a second preset number of sediment samples are selected as grain size samples; the union of the elemental samples and the grain size samples is all sediment samples.

[0041] The particle size distribution data of each sample was obtained, and linear interpolation was performed on the particle size distribution data of each sample to obtain the particle size distribution data of each sediment sample.

[0042] Flood-sensitive chemical elements are those chemically active in surface geochemical environments, easily affected by chemical weathering and leaching, or prone to adsorption on the surface of fine-grained clay minerals, thus exhibiting a significant response to changes in hydrodynamic conditions (such as grain sorting caused by floods). These elements typically include large ion lithophile elements. Since flood events are usually accompanied by changes in the input of suspended transported material (fine-grained clay), fluctuations in the content of these elements can effectively record hydrodynamic processes. In practical applications, flood-sensitive chemical elements can be selected from one or more of rubidium (Rb), strontium (Sr), potassium (K), calcium (Ca), and magnesium (Mg). Reference elements can be selected from one or more of zirconium (Zr), titanium (Ti), hafnium (Hf), niobium (Nb), yttrium (Y), and thorium (Th).

[0043] In practical applications, the content ratio of each sediment sample is obtained by linear interpolation of the content ratio of each element sample. The process of determining the content ratio of each sediment sample includes several methods.

[0044] The first method involves obtaining the content of the target flood-sensitive chemical element (e.g., Rb) and the content of the target reference stationary chemical element (e.g., Zr) in each element sample. For any element sample, the ratio of the two contents is calculated to obtain the content ratio of the element sample. The content ratio of each element sample is then linearly interpolated to obtain the content ratio of each sediment sample.

[0045] The second method involves obtaining the content of the target flood-sensitive chemical element (e.g., Rb) and multiple baseline stationary chemical elements (e.g., Zr and Ti) in each element sample. For any element sample, the ratio of Rb content to Zr and Ti content is calculated to obtain Rb / Zr and Rb / Ti. Linear interpolation is performed on the Rb / Zr ratios for all element samples to obtain the Rb / Zr ratio for each sediment sample. Then, linear interpolation is performed on the Rb / Ti ratios for all element samples to obtain the Rb / Ti ratio for each sediment sample. Correlation analysis is then performed on the Rb / Zr and Rb / Ti ratios for each sediment sample to determine the baseline stationary chemical element that best represents the flood signal. This baseline stationary chemical element is then used as the target baseline stationary chemical element, and the content ratios for each sediment sample are obtained.

[0046] The third method involves obtaining the contents of multiple flood-sensitive chemical elements (e.g., Rb and Sr) and the contents of a target baseline fixed chemical element (e.g., Zr) in each element sample. For any element sample, the ratios of Rb to Zr and Sr to Zr in the sample are calculated to obtain Rb / Zr and Sr / Zr. Linear interpolation is then performed on the Rb / Zr ratios for all element samples to obtain the Rb / Zr ratio for each sediment sample. Similarly, linear interpolation is performed on the Sr / Zr ratios for all element samples to obtain the Sr / Zr ratio for each sediment sample. Correlation analysis is then performed on the Rb / Zr and Sr / Zr ratios for each sediment sample to identify the flood-sensitive chemical element that best represents the flood signal. This element is then used as the target flood-sensitive chemical element, and the content ratios for each sediment sample are obtained.

[0047] The fourth method involves obtaining the contents of flood-sensitive chemical elements (e.g., Rb and Sr) and multiple baseline stationary chemical elements (e.g., Zr and Ti) in each element sample. For any element sample, the ratios of Rb content to Zr and Ti content, and the ratios of Sr content to Zr and Ti content, are calculated to obtain Rb / Zr, Rb / Ti, Sr / Zr, and Sr / Ti. Linear interpolation is then performed on the Rb / Zr corresponding to all element samples to obtain the Rb / Zr corresponding to each sediment sample. Similarly, linear interpolation is performed on the Rb / Ti corresponding to all element samples to obtain the Rb / Ti corresponding to each sediment sample. Finally, linear interpolation is performed on the Sr / Zr corresponding to all element samples to obtain the Sr / Ti corresponding to each sediment sample. Correlation analysis was performed on the Rb / Zr, Rb / Ti, Sr / Zr, and Sr / Ti ratios of each sediment sample to obtain the baseline immobilized chemical elements and flood-sensitive chemical elements that best represent the flood signal. These were then used as the target baseline immobilized chemical elements and target flood-sensitive chemical elements, and the content ratios of each sediment sample were obtained.

[0048] In another exemplary embodiment of this application, the process of selecting endmembers to be refined from each potential endmember based on the correlation coefficient value between the endmember matrix and the ratio matrix of each potential endmember specifically includes: calculating the correlation coefficient value between the endmember matrix and the ratio matrix of each potential endmember.

[0049] The potential endmembers corresponding to the maximum correlation coefficient values ​​are identified as endmembers to be refined.

[0050] In another exemplary embodiment of this application, the ratio matrix is ​​used as prior knowledge to perform EMM processing on the grain size distribution data of all sediment samples to obtain a high-fidelity flood endmember matrix. Specifically, at the current iteration number, the endmember intensity matrix and endmember feature matrix at the current iteration number are updated to obtain the endmember intensity matrix and endmember feature matrix at the next iteration number. The endmember intensity matrix and endmember feature matrix at the initial iteration number can be initialized or obtained using the matrix results obtained in step 203.

[0051] The objective function value for the current iteration number is calculated based on the grain size distribution data, ratio matrix, endmember strength matrix of the endmember to be refined in each sediment sample under the next iteration number, and endmember strength matrix and endmember feature matrix under the next iteration number.

[0052] Determine whether the iteration stopping condition has been met based on the objective function value at the current iteration number and the objective function value at the previous iteration number.

[0053] If the iteration stopping condition is met, the endmember intensity matrix for the next iteration is determined to be the high-fidelity flood endmember matrix.

[0054] If the iteration stopping condition is not met, the iteration count is updated and the next iteration begins.

[0055] In practical applications, EMM processing is performed on the grain size distribution data of all sediment samples. Specifically, matrix V is constructed based on the grain size distribution data of all sediment samples, and then EMM processing is performed on V.

[0056] In another exemplary embodiment of this application, the objective function J is: Where V represents the matrix composed of grain size distribution data of all sediment samples, W represents the endmember feature matrix at the next iteration number, and H represents the endmember intensity matrix at the next iteration number. This represents a vector composed of the intensity values ​​of the endmembers to be refined in each sediment sample within the endmember intensity matrix for the next iteration. Represents the ratio matrix, Represents the constraint strength coefficient at the current iteration number, || || 2 The L2 norm represents the square root of the sum of squares of all elements in the matrix. The W matrix is ​​the granularity distribution curve of the endmembers themselves, used for qualitative judgment. You need to look at the graph drawn from the W matrix to determine which endmember (e.g., EM4) represents the sediment brought by the flood (usually coarser or with a specific distribution).

[0057] In another exemplary embodiment of this application, the method further includes, before linear interpolation, dating all sediment samples to obtain the time corresponding to each sediment sample.

[0058] In another exemplary embodiment of this application, the paleoflood identification method further includes: calculating the paleoflood intensity index value of each paleoflood sample based on the intensity value of the end-member to be refined in each paleoflood sample in the high-fidelity flood end-member matrix and the content ratio of each paleoflood sample.

[0059] The paleoflood intensity index (PII) values ​​of each paleoflood sample are sorted according to the time corresponding to each paleoflood sample to obtain the PII time series; the time corresponding to each paleoflood sample is obtained by dating each paleoflood sample.

[0060] The values ​​of climate proxy indices for the climate factors to be analyzed at the corresponding time for each paleoflood sample are obtained, and the values ​​of the obtained climate proxy indices are sorted according to the time corresponding to each paleoflood sample to obtain the time series of climate proxy indices.

[0061] A cross-wavelet transform is performed between the PII time series and the climate proxy index time series to obtain a time-frequency plot.

[0062] The event synchronization intensity and delay time set are obtained based on the time corresponding to each paleoflood sample and the extreme points in the climate event sequence corresponding to the climate factors to be analyzed.

[0063] Attributing paleofloods to time-frequency diagrams, event synchronization intensity, and delay time sets helps determine whether they are driven by the climate factors to be analyzed.

[0064] In another exemplary embodiment of this application, the paleoflood intensity index value of each paleoflood sample is calculated based on the intensity value of the end-member to be refined in the high-fidelity flood end-member matrix in each paleoflood sample and the content ratio of each paleoflood sample. Specifically, for any paleoflood sample, the weighted sum of the intensity value of the end-member to be refined in the high-fidelity flood end-member matrix in the paleoflood sample and the content ratio of the paleoflood sample is calculated to obtain the paleoflood intensity index value of the paleoflood sample.

[0065] This application also provides a more specific embodiment to describe the above-mentioned ancient flood identification method in detail. The specific steps include: step (1): data acquisition and standardization preparation.

[0066] This step aims to ensure the initial quality and consistency of the input data, laying the foundation for subsequent high-fidelity analysis. The main components are data acquisition equipment and laboratory processing equipment.

[0067] Sample collection: Sediment samples are continuously collected from the borehole core at predetermined, uniform intervals (e.g., 4 cm). During the collection process, strict adherence to pollution-free operating procedures is maintained to avoid damaging the original sedimentary bedding.

[0068] Sample preparation: The collected samples were processed using laboratory equipment for subsequent grain size and geochemical analysis. To protect the natural aggregate structure of the sediments and ensure that the grain size data accurately reflects the hydrodynamic conditions at the time of deposition, the use of mild wet sieving and settling techniques was mandatory, and methods such as freeze-drying, which could damage the sample structure, were explicitly prohibited.

[0069] Data Acquisition: Chemical element content was collected from a portion of the samples using data acquisition equipment, while particle size distribution data was collected from another portion. The chemical elements included: active chemical elements (Rb, Sr, K), semi-active chemical elements, and immobile chemical elements (Zr, Ti, Hf, Nb).

[0070] Step (2): Integrated geochemistry-grain size pretreatment.

[0071] This step aims to solve the technical challenge of the "granularity effect" through computational methods, elevating geochemical data from an auxiliary verification tool to a core input. The execution entity is a computer processor.

[0072] Step 2.1: Grain size effect correction based on fixed element normalization. This is used to eliminate the interference of sediment grain size on geochemical measurement results and obtain geochemical signals that truly reflect the intensity of chemical weathering and changes in provenance.

[0073] Step 2.1.1: The computer receives the input geochemical data (chemical element content) and grain size distribution data. For any element sample, the ratio of the content of the selected target flood-sensitive chemical element (e.g., Rb) to the content of the target reference stationary chemical element (e.g., Zr) in the element sample is calculated to obtain a ratio matrix for subsequent analysis (including the Rb / Zr ratios for all element samples) to eliminate preliminary grain size background interference.

[0074] Step 2.1.2: Establish a unified time axis or depth axis and perform interpolation. First, date the sediment samples (carbon-14 dating) to establish a depth-time model and determine a unified time axis or depth axis. The computer receives the grain size distribution data of each grain size sample and the content ratios of each element sample calculated in Step 2.1.1. Linear interpolation is performed on the grain size distribution data and the content ratios of each element sample for each grain size sample, projecting them onto the unified time axis or depth axis to obtain an aligned data matrix. Each row of the data matrix represents the data corresponding to a time or depth point, including the time or depth value of that point, the interpolated grain size distribution vector (containing the content of each grain size component), and the interpolated content ratio value of that point. This processing ensures that each time point has corresponding grain size data and geochemical data, and that the two are strictly corresponding in space and time.

[0075] Step 2.2: Feedback-based, iterative end-member modeling (EMM). This step creates a collaborative analysis system that deeply integrates content ratios and granularity data, going beyond simple linear combinations of multiple indicators to improve the accuracy of end-member identification. The general approach is as follows: Step A: Perform a standard granularity analysis to obtain preliminary results. Step B: Using interpolated and aligned geochemical data, identify end-members in the preliminary results that are correlated with the ratio index (Rb / Zr). Step C: Transform these correlations into mathematical constraints. Step D: With these constraints, perform another standard granularity analysis. Step E: Check if the new results from Step D have converged. If not, adjust the constraint strength coefficient λ and continue with Steps C and D. λ controls the influence of the ratio matrix on the end-member decomposition results. A larger λ forces the decomposed flood of end-members to conform more closely to the ratio matrix curve. A smaller λ respects the natural distribution of the original granularity data.

[0076] Step 2.2.1: Initial EMM Analysis: The computer performs a preliminary EMM analysis (such as nonnegative matrix factorization) on the granular distribution data in the data matrix to identify potential endmembers (e.g., EM1 to EM5) and obtain a preliminary endmember intensity matrix and endmember feature matrix. A column of the endmember intensity matrix corresponds to the intensity value of a specific potential endmember across all samples.

[0077] Step 2.2.2: Geochemical Targeted Refining: Computer analysis is performed on the geochemical data related to each initial endmember, after correction in Step 2.1, to identify endmembers that not only differ in grain size characteristics but also have clearly indicative geochemical features. Specifically, the correlation coefficient between the endmember matrix and the ratio matrix of each potential endmember is calculated, and the potential endmember with the highest correlation is selected as the endmember to be refined.

[0078] Step 2.2.3: Constrained EMM reanalysis: The computer uses the ratio matrix as prior knowledge and constructs an objective function containing directional constraint terms based on the column index k of the end-member to be refined in the preliminary end-member strength matrix determined in Step 2.2.2. Then, it performs EMM analysis again on the grain size distribution data (matrix V) of all sediment samples.

[0079] Specifically, a penalty term is introduced during the iterative process of nonnegative matrix factorization. By minimizing the objective function, the intensity trends of the decomposed endmembers to be refined are optimized under the constraint of the ratio matrix P, while other endmembers are freely decomposed based on the original granularity data. This process forms a feedback loop for a geochemical data refining granularity model, which can be iteratively executed until the model converges.

[0080] Step 2.2.3 specifically includes: Step 2.2.3.1: Construct the objective function.

[0081] Define the objective function as: .

[0082] Among them, the first item ||V-WH|| 2 The second term represents the reconstruction error of the original granular data. The introduced penalty term is used to constrain... Fitting matrix P.

[0083] Step 2.2.3.2: Perform nonnegative matrix factorization with penalty terms.

[0084] Step 2.2.3.2.1: Initialization.

[0085] Randomly generate non-negative initial endmember feature matrix and initial endmember strength matrix. To speed up convergence, the endmember strength matrix and endmember feature matrix obtained in step 2.2.1 can also be used.

[0086] Step 2.2.3.2.2: Iterative update (core algorithm).

[0087] A multiplicative update rule is used to alternately update the elements in the initial endmember feature matrix and the initial endmember strength matrix. After each update, the value of the objective function J is calculated. It is then determined whether the rate of change of the objective function J between two adjacent iterations is less than a preset threshold (e.g., 10). -4 The algorithm checks whether the number of iterations has reached the preset maximum number of iterations (e.g., 1000). If either condition is met, the iteration is considered to have stopped (i.e., the objective function has converged), the iteration stops, and the final endmember intensity matrix is ​​output as the high-fidelity flood endmember matrix.

[0088] The multiplicative update rule is as follows: Update the W matrix: Based on the fitting error between V and the reconstructed data WH, adjust the elements in W to maintain non-negativity. Update the H matrix (introducing constraints): This is a key step in this application. When calculating the update gradient of H, not only the fitting error (V-WH) is considered, but also a penalty term is added. The derived gradient components, if the intensity values ​​of the endmembers to be refined in the current endmember intensity matrix... If the ratio matrix P deviates from the target ratio matrix, the algorithm will generate a reverse correction force during the update process. The value of P is pulled towards the value of P.

[0089] Step (3): Dynamic flood event detection and intensity quantification.

[0090] This step replaces the traditional subjective threshold method with an objective and complex algorithm, generating a completely new quantitative output. The execution entity is a computer processor.

[0091] Step 3.1: Flood identification is performed using the Iterative Triclass Thresholding (ITT) method. This method overcomes the subjectivity and limitations of the traditional "mean + standard deviation" thresholding method by employing an iterative process that perceives uncertainty, thereby improving the sensitivity and objectivity of detecting weak or complex flood signals.

[0092] Step 3.1.1: Initial segmentation: The computer selects the high-fidelity flood end-member matrix obtained in step (2), applies the Otsu algorithm to it, and finds an initial optimal segmentation threshold by maximizing the inter-class variance.

[0093] Step 3.1.2: Define the To-Be-Determined (TBD) region: Unlike traditional binary segmentation, this application divides the high-fidelity flood endmember matrix into a flood region and a background region based on the optimal segmentation threshold. The mean values ​​of the flood region and the background region (μ_flood and μ_background) are used to define the to-be-determined region. The intensity values ​​of the high-fidelity flood endmember matrix that fall within the interval [μ_background, μ_flood] are determined as the to-be-determined region.

[0094] Step 3.1.3: Iterative Refinement: Apply the Otsu algorithm again to the data points within the region to be defined for secondary segmentation. This process can be repeated a preset number of times, or until the threshold change (T) calculated in the nth iteration is obtained. n -T n-1 The process continues until the threshold value is less than a preset convergence criterion (e.g., 0.001), otherwise, it re-enters step 3.1.2. The threshold value at the last iteration number is determined as the flood identification threshold. n This represents the threshold at the nth iteration number. Paleoflood samples can be obtained based on the flood identification threshold and the intensity values ​​of the endmembers to be refined for each sediment sample.

[0095] Step 3.2: Construction of the Paleoflood Intensity Composite Index (PII). Based on the formula... The Paleo-flood Intensity Index (PII) value was calculated for each sediment sample identified as a paleoflood event by the ITT method, and the PII time series was constructed.

[0096] in, This represents the intensity value of the EM4 endmember to be refined corresponding to the paleoflood sample. This indicates the content ratio of ancient flood samples. and All are weighting coefficients, which can be optimized through principal component analysis or by training a machine learning model on known historical flood event data (the default value in this application is 0.5).

[0097] Step (4): Quantitative Attribution Analysis of Climate Drivers. This step elevates traditional qualitative visual comparisons to statistically robust quantitative analysis implemented by computers to reveal the climate-driven mechanisms behind flood events. The execution entity is a computer processor.

[0098] Step 4.1: Cross wavelet transform (XWT) analysis. Used to quantitatively analyze the correlation, common periodicity, and leading or lagging relationships between flood series and climate proxy indicator series in the time-frequency domain.

[0099] Step 4.1.1: Data preparation: The computer acquires the PII time series generated in step (3) and obtains one or more high-resolution climate proxy time series from public climate databases (e.g., the δ18O record of Donggedong stalagmites as a proxy for El Niño-Southern Oscillation (ENSO) activity).

[0100] Step 4.1.2: XWT Calculation: Perform cross wavelet transform between the PII time series and the climate proxy index time series to obtain a time-frequency plot. Combine the historical climate data in the climate proxy index time series with the PII values ​​in the PII time series obtained in Step 3.2, and perform XWT analysis on the data of the two, with the horizontal axis set to time, to determine the relationship between them.

[0101] Time-frequency plots can visually show when and at which periods two sequences have a significant common power spectrum (energy resonance), and use phase arrows to indicate their leading or lagging relationship (phase difference).

[0102] Step 4.2: Event Synchronization (ES) analysis. This is used to quantify the nonlinear coupling strength and synchronicity between flood events and extreme climate events from the perspective of discrete events, without parameters.

[0103] Step 4.2.1: Event Sequence Definition: Based on the results of the ITT algorithm, extract the corresponding times for each ancient flood sample to form a "flood event sequence". Simultaneously, extract the time points of extreme events (such as the strongest El Noo event) from the climate proxy index time series to form a "climate event sequence". In scientific data analysis, a threshold is first set, and all extreme events exceeding that threshold are extracted. Specifically, a single stalagmite may record climate change over the past 10,000 years. A characteristic of ENSO is that it does not occur only once every 10,000 years, but rather every few years. Here, the extreme point refers to the moment when oxygen isotopes reach their peak during each ENSO eruption. In a 10,000-year record, there may have been 500 intense ENSO events. Extracting these 500 time points forms the "climate event sequence".

[0104] Step 4.2.2: ES Calculation: The computer applies an event synchronization algorithm to the two event sequences described above. This algorithm determines whether the two events are "synchronized" by dynamically calculating the delay time. Its core formula is: ,in, Represents the first in the flood event sequence The time delay between time point m and the m-th time point in the climate event sequence. Indicates the first flood event in the sequence of flood events. At that moment, Let m represent the m-th time in the climate event sequence, and min{} represent the minimum value. If it is less than a set threshold (e.g., 0.1 seconds), it indicates that the [number]th [time]... The flood event at time m and the climate factor to be analyzed at time m are synchronous event pairs.

[0105] Step 4.2.3: A delay time set can be obtained by calculating according to the above formula. The event synchronization strength Q is obtained by dividing the number of synchronized event pairs in the delay time set by the total number of event pairs. This value can reveal the degree of coupling between two event sequences.

[0106] Step (5): Attribute the paleoflood to the time-frequency diagram, event synchronization intensity, and delay time set. Based on the above statistical indicators, the computer executes the following comprehensive logical judgment process to determine whether the paleoflood is driven by the climate factors to be analyzed.

[0107] Step 5.1: Time-Frequency Plot Feature Determination (Condition A). Determine whether there is a resonance energy region in the time-frequency plot that falls within a preset significance test region (e.g., a red noise standard spectrum region that has passed the 5% significance level test); if so, extract the periodic frequency band corresponding to the resonance energy region and determine whether the periodic frequency band is within the characteristic period range of the climate factor to be analyzed (e.g., 2-7 years); if it is, then the first driving condition is satisfied.

[0108] Step 5.2: Event synchronization strength determination (condition B). Determine whether the event synchronization strength Q is greater than the preset upper limit threshold of the confidence interval (preferably the upper limit of the 95% confidence interval calculated by Monte Carlo simulation); if so, it is determined that the second driving condition is met.

[0109] Step 5.3: Phase and Delay Time Determination (Condition C). Based on the phase relationship extracted from the time-frequency diagram, determine whether the anomalous event of the climate factor to be analyzed occurs before or simultaneously with the paleoflood event (e.g., in-phase or fixed leading relationship indicated by arrows pointing to the right or vertically), and determine whether the representative delay time in the delay time set is within a preset physically interpretable lag time range (e.g., a lag range of several months to one year set based on experience). The representative delay time includes the average or median of each delay time in the delay time set. If both conditions are met, the third driving condition is satisfied.

[0110] Step 5.4: Attribution Conclusion Generation. If the first, second, and third driving conditions are all met simultaneously, the paleoflood is determined to be driven by the climate factor to be analyzed; if any one of the first, second, and third driving conditions is not met, the paleoflood is determined not to be driven by the climate factor to be analyzed.

[0111] This application aims to achieve high-fidelity reconstruction, quantitative intensity assessment, and quantitative analysis of climate-driven mechanisms of paleoflood events by deeply integrating geochemical and granular data and combining them with advanced signal processing technology. This will fundamentally improve the fidelity and scientific rigor of paleoflood reconstruction and solve problems such as granular effect interference, strong subjectivity in flood identification, and qualitative analysis of climate correlation in current known technologies.

[0112] This application significantly improves the accuracy and reliability of paleoflood identification, fundamentally solving the technical challenge of "grain size effect" interference. Existing technologies typically treat grain size and geochemical data as independent modules for simple linear superposition or parallel verification. This method cannot truly eliminate the "grain size effect" interference caused by sediment grain size variations, leading to index changes that may only reflect hydrodynamic sorting rather than actual chemical weathering or provenance changes. This application breaks with the traditional "data parallel" paradigm, proposing a deeply integrated feedback endmember model (EMM) analysis protocol: First, normalization is performed using the ratio of flood-sensitive elements to baseline stationary elements (e.g., Rb / Zr) to initially offset the dilution effect caused by sediment grain size variations in signal intensity. Then, this ratio sequence is creatively transformed into mathematical prior knowledge and introduced into a non-negative matrix factorization algorithm. This is not a simple comparative verification, but rather utilizes the high sensitivity of geochemical data to hydrodynamic conditions as a penalty term in the objective function, directionally forcing (constraining) the grain size model to separate sedimentary components (endmembers) highly consistent with flood chemical characteristics. Through this constraint mechanism, this application essentially uses geochemical data to clean the granular data, forcing the model to extract high-fidelity flood endmembers driven solely by hydrodynamics, completely removing granular noise mixed in by non-hydrodynamic factors, achieving deep fusion between data, and greatly improving the accuracy of flood analysis.

[0113] This application enhances the objectivity and sensitivity of flood detection. Existing technologies use static thresholds based on the "mean + standard deviation" to identify floods. This method is highly subjective and insensitive to signals with low signal-to-noise ratios. This application replaces the traditional static threshold with an iterative three-classification threshold method, identifying flood events in an objective and adaptive manner. The identification process is transformed into an objective, data-driven, iterative algorithm capable of intelligently handling uncertain data. This allows the application to more sensitively capture weak or complex flood events, significantly improving the objectivity of the detection.

[0114] This application achieves a leap from qualitative to quantitative analysis. Existing technologies typically output binary data (flood or non-flood). This application, through step 3.2 of its technical solution, constructs a novel paleoflood intensity index (PII) that integrates multi-source information, thus achieving a leap from binary identification to quantitative assessment. The PII is a continuous, quantitative measure of flood intensity, with an information content far exceeding that of binary classification, providing more refined data support for flood risk assessment.

[0115] In exploring the relationship between floods and large-scale climate models such as ENSO, current techniques largely rely on qualitative visual comparisons of charts, lacking quantitative and statistically robust analytical methods. This fails to accurately reveal the deep-seated dynamic mechanisms, such as the coupling strength and phase relationship (e.g., leading or lagging), between the two. Furthermore, the primary reliance on qualitative visual comparisons to explore the association between floods and climate (e.g., ENSO) lacks statistical rigor and fails to uncover the underlying mechanisms. This application provides a novel mechanistic analysis capability. By introducing cross-wavelet transform and event synchronization algorithms, it achieves, for the first time, a quantitative attribution of the climate-flood relationship. It not only identifies correlations but also precisely quantifies the phase relationship (leading or lagging) and nonlinear coupling strength between the two at different time scales, thus providing a mechanistic understanding and predictive insight unattainable by current techniques.

[0116] This application employs signal processing techniques such as cross-wavelet transform and event synchronization to quantify the coherence, phase relationship, and nonlinear coupling strength between flood sequences and global climate driving factors (such as ENSO), thereby revealing their intrinsic driving mechanism.

[0117] Current known techniques only use mean, median, or sensitive grain size components to identify floods. This method has low accuracy because it ignores geochemical information that also reflects flood processes and is susceptible to interference from factors such as changes in provenance in complex sedimentary environments. This application significantly improves the ability to analyze hydrodynamic processes through step 2.2.3.

[0118] Traditional multi-indicator combination methods simply perform linear superposition or parallel verification of granularity indicators and geochemical indicators. This approach does not achieve deep fusion of the two types of data at the model level, nor does it address the granularity effect inherent in geochemical data itself. This application, through step 2.2, achieves a paradigm shift from "data parallelism" to "model fusion."

[0119] Traditional time series analysis methods for climate correlation analysis can employ conventional linear correlation analysis or spectral analysis. However, these methods typically assume that the signal is stationary, making it difficult to handle typical non-stationary time series such as paleoclimate and paleoflood. This application utilizes cross-wavelet transform and event synchronization to capture time-varying frequency and phase information, resulting in analytical capabilities far exceeding those of traditional methods.

[0120] Based on the same inventive concept, this application also provides an ancient flood identification device for implementing the ancient flood identification method described above. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations in one or more ancient flood identification device embodiments provided below can be found in the limitations of the ancient flood identification method described above, and will not be repeated here.

[0121] In one exemplary embodiment, such as Figure 2 As shown, an ancient flood identification device is provided, comprising: a sample collection module for continuously collecting multiple sediment samples from borehole cores in a target area.

[0122] The acquisition module is used to acquire the grain size distribution data and the content ratio of each sediment sample; the content ratio is the ratio of the content of the target flood-sensitive chemical element to the content of the target reference stationary chemical element.

[0123] The end-member determination module is used to perform EMM processing on the grain size distribution data of all sediment samples to obtain an end-member intensity matrix. The end-member intensity matrix includes the intensity value of each potential end-member in each sediment sample. For any potential end-member, the end-member matrix of the potential end-member is constructed based on the intensity value of the potential end-member in each sediment sample as described in the end-member intensity matrix. The end-members to be refined are screened from each potential end-member based on the correlation coefficient value between the end-member matrix and the ratio matrix of each potential end-member. The ratio matrix includes the content ratio of all sediment samples.

[0124] The high-fidelity flood end-member matrix determination module is used to perform EMM processing on the grain size distribution data of all sediment samples using the ratio matrix as prior knowledge to obtain the high-fidelity flood end-member matrix. The high-fidelity flood end-member matrix includes the intensity values ​​of the end-members to be refined in each sediment sample.

[0125] The flood identification threshold determination module is used to process the high-fidelity flood end-member matrix using an iterative three-classification threshold method to obtain the flood identification threshold.

[0126] The paleoflood identification module is used to identify sediment samples corresponding to target intensity values ​​in the high-fidelity flood end-member matrix as paleoflood samples, thus completing the paleoflood identification; the target intensity value is the intensity value that is greater than the flood identification threshold.

[0127] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 3As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores paleoflood identification data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a paleoflood identification method.

[0128] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0129] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described method embodiments.

[0130] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the above-described method embodiments.

[0131] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the above-described method embodiments.

[0132] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0133] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0134] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, etc., and are not limited to these.

[0135] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0136] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for identifying ancient floods, characterized in that, The paleoflood identification method includes: Multiple sediment samples were obtained by continuously collecting core samples from boreholes in the target area; Obtain the grain size distribution data and content ratio of each sediment sample; the content ratio is the ratio of the content of the target flood-sensitive chemical element to the content of the target reference stationary chemical element. Endmember model analysis was performed on the grain size distribution data of all sediment samples to obtain the endmember intensity matrix; the endmember intensity matrix includes the intensity value of each potential endmember in each sediment sample; For any potential endmember, the endmember matrix of the potential endmember is constructed based on the intensity values ​​of the potential endmember in each sediment sample as described in the endmember intensity matrix. Based on the correlation coefficient values ​​of the endmember matrix and ratio matrix of each potential endmember, endmembers to be refined are screened from each potential endmember; the ratio matrix includes the content ratio of all sediment samples. Using the ratio matrix as prior knowledge, endmember model analysis was performed on the grain size distribution data of all sediment samples to obtain a high-fidelity flood endmember matrix. The high-fidelity flood endmember matrix includes the intensity values ​​of the endmembers to be refined in each sediment sample. An iterative three-class thresholding method is used to process the high-fidelity flood end-member matrix to obtain the flood identification threshold; specifically: Step 3.1.1: Initial segmentation: Select the high-fidelity flood end-member matrix and apply the Otsu algorithm to it to find an initial optimal segmentation threshold by maximizing the inter-class variance; Step 3.1.2: Define the undetermined region: Divide the high-fidelity flood endmember matrix into a flood region and a background region according to the optimal segmentation threshold. Use the mean value μ_flood of the flood region and the mean value μ_background of the background region to define the undetermined region. The intensity value of the high-fidelity flood endmember matrix that is located in the interval [μ_background, μ_flood] is determined as the undetermined region. Step 3.1.3: Iterative Refinement: Apply the Otsu algorithm again to the data points in the area to be defined for secondary segmentation; this process can be repeated a preset number of times, or until the threshold change calculated in the nth iteration is less than a preset convergence criterion, otherwise return to step 3.1.2; determine the threshold under the last iteration number as the flood identification threshold; The sediment samples corresponding to the target intensity values ​​in the high-fidelity flood end-member matrix are identified as paleoflood samples, thus completing the paleoflood identification; the target intensity value is the intensity value that is greater than the flood identification threshold.

2. The paleoflood identification method according to claim 1, characterized in that, The paleoflood identification method also includes: The paleoflood intensity index value of each paleoflood sample is calculated based on the intensity value of the end-member to be refined in each paleoflood sample and the content ratio of each paleoflood sample in the high-fidelity flood end-member matrix. The paleoflood intensity index (PII) values ​​of each paleoflood sample were sorted according to the time corresponding to each paleoflood sample to obtain the PII time series; the time corresponding to each paleoflood sample was obtained by dating each paleoflood sample. The values ​​of climate proxy indices for the climate factors to be analyzed at the time corresponding to each paleoflood sample are obtained, and the values ​​of the obtained climate proxy indices are sorted according to the time corresponding to each paleoflood sample to obtain the time series of climate proxy indices. A cross-wavelet transform is performed between the PII time series and the climate proxy time series to obtain a time-frequency plot; The synchronization intensity and delay time set of events are obtained based on the time corresponding to each paleoflood sample and the climate event sequence; the climate event sequence is the time corresponding to the extreme point in the climate proxy index time series. Attributing paleofloods to time-frequency diagrams, event synchronization intensity, and delay time sets helps determine whether they are driven by the climate factors to be analyzed.

3. The paleoflood identification method according to claim 1, characterized in that, Based on the correlation coefficient values ​​between the endmember matrix and the ratio matrix of each potential endmember, endmembers to be refined are selected from each potential endmember, specifically including: Calculate the correlation coefficient between the endmember matrix and the ratio matrix for each potential endmember; The potential endmembers corresponding to the maximum correlation coefficient values ​​are identified as endmembers to be refined.

4. The paleoflood identification method according to claim 1, characterized in that, Using the ratio matrix as prior knowledge, endmember model analysis was performed on the grain size distribution data of all sediment samples to obtain a high-fidelity flood endmember matrix, specifically: At the current iteration number, the endmember strength matrix and endmember feature matrix at the current iteration number are updated to obtain the endmember strength matrix and endmember feature matrix at the next iteration number; The objective function value for the current iteration number is calculated based on the grain size distribution data, ratio matrix, endmember strength matrix of the endmember to be refined in each sediment sample under the next iteration number, and endmember strength matrix and endmember feature matrix under the next iteration number. Determine whether the iteration stopping condition has been met based on the objective function value at the current iteration number and the objective function value at the previous iteration number. If the iteration stopping condition is met, the endmember intensity matrix for the next iteration number is determined to be the high-fidelity flood endmember matrix. If the iteration stopping condition is not met, the iteration count is updated and the next iteration begins.

5. The paleoflood identification method according to claim 4, characterized in that, The objective function J is: Where V represents the matrix composed of grain size distribution data of all sediment samples, W represents the endmember feature matrix at the next iteration number, and H represents the endmember intensity matrix at the next iteration number. This represents a vector composed of the intensity values ​​of the endmembers to be refined in each sediment sample within the endmember intensity matrix for the next iteration. Represents the ratio matrix, Represents the constraint strength coefficient at the current iteration number, || || 2 This represents the 2-norm.

6. The paleoflood identification method according to claim 2, characterized in that, The paleoflood intensity index value of each paleoflood sample is calculated based on the intensity value of the end-member to be refined in each paleoflood sample and the content ratio of each paleoflood sample in the high-fidelity flood end-member matrix. Specifically, for any paleoflood sample, the weighted sum of the intensity value of the end-member to be refined in the paleoflood sample and the content ratio of the paleoflood sample is calculated to obtain the paleoflood intensity index value of the paleoflood sample.

7. A device for identifying ancient floods, characterized in that, The ancient flood identification device includes: The sample acquisition module is used to continuously collect multiple sediment samples from borehole cores in the target area; The acquisition module is used to acquire the grain size distribution data and the content ratio of each sediment sample; the content ratio is the ratio of the content of the target flood-sensitive chemical element to the content of the target reference stationary chemical element. The end-member determination module is used to perform end-member model analysis on the grain size distribution data of all sediment samples to obtain an end-member intensity matrix. The end-member intensity matrix includes the intensity value of each potential end-member in each sediment sample. For any potential end-member, an end-member matrix of the potential end-member is constructed based on the intensity value of the potential end-member in each sediment sample as described in the end-member intensity matrix. Based on the correlation coefficient value between the end-member matrix and the ratio matrix of each potential end-member, end-members to be refined are screened from each potential end-member. The ratio matrix includes the content ratio of all sediment samples. The high-fidelity flood end-member matrix determination module is used to perform end-member model analysis on the grain size distribution data of all sediment samples using the ratio matrix as prior knowledge, and obtain the high-fidelity flood end-member matrix. The high-fidelity flood end-member matrix includes the intensity values ​​of the end-members to be refined in each sediment sample. The flood identification threshold determination module is used to process the high-fidelity flood end-member matrix using an iterative three-class threshold method to obtain the flood identification threshold; specifically: Step 3.1.1: Initial segmentation: Select the high-fidelity flood end-member matrix and apply the Otsu algorithm to it to find an initial optimal segmentation threshold by maximizing the inter-class variance; Step 3.1.2: Define the undetermined region: Divide the high-fidelity flood endmember matrix into a flood region and a background region according to the optimal segmentation threshold. Use the mean value μ_flood of the flood region and the mean value μ_background of the background region to define the undetermined region. The intensity value of the high-fidelity flood endmember matrix that is located in the interval [μ_background, μ_flood] is determined as the undetermined region. Step 3.1.3: Iterative Refinement: Apply the Otsu algorithm again to the data points in the area to be defined for secondary segmentation; this process can be repeated a preset number of times, or until the threshold change calculated in the nth iteration is less than a preset convergence criterion, otherwise return to step 3.1.2; determine the threshold under the last iteration number as the flood identification threshold; The paleoflood identification module is used to identify sediment samples corresponding to target intensity values ​​in the high-fidelity flood end-member matrix as paleoflood samples, thus completing the paleoflood identification; the target intensity value is the intensity value that is greater than the flood identification threshold.

8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the ancient flood identification method according to any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the ancient flood identification method according to any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the ancient flood identification method according to any one of claims 1-6.

Images