Isotope source tracing analysis system based on fourth century loess

Abstract

The invention discloses an isotope source tracing analysis system based on fourth-century loess, and relates to the technical field of data processing. Fourth-century loess sample data is acquired and preprocessed, and a standardized data set is constructed; establishing a three-dimensional feature tensor based on the standardized data set, performing dimensionality reduction, calculating information entropy, dynamically distributing adaptive weights of all modes, weighting to obtain a coupling feature vector, and outputting the coupling feature vector and a weight distribution result; receiving a coupling feature vector and a weight distribution result, taking a multi-isotope end member value as a constraint, and outputting a contribution proportion of each potential source region through a Bayesian mixed linear model and constraint optimization solution; and receiving the contribution ratio, combining the isotope fractionation effect and a distance attenuation model, matching an optimal object source migration path by constructing a path cost function, and outputting migration path parameters. The invention provides a material source tracing analysis system which adopts multi-source indexes and is high in accuracy and high in quantification capability.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and more specifically to a source tracing analysis system based on Quaternary loess isotopes. Background Technology

[0002] Quaternary loess paleosols are the core carriers of global change research. Their provenance tracing relies on techniques such as elemental geochemistry, isotopes, and zircon U-Pb age spectroscopy. Current mainstream methods use X-ray fluorescence spectroscopy to determine elemental content, isotope mass spectrometry to test isotope ratios, manual calculation of characteristic ratios and plotting of scatter plots, and comparison with sediment data from potential source areas to determine provenance.

[0003] However, existing technologies involve cumbersome data processing, requiring manual data standardization, ratio calculation, and graphical drawing, resulting in low efficiency and difficulty in adapting to large-scale sample analysis. Traditional methods rely on single or a few indicators, failing to fully utilize multi-dimensional data information and exhibiting low accuracy in identifying complex sources. They are also highly subjective, with source judgment relying on manual comparison of graphical clustering results, making them susceptible to human error. Furthermore, they lack quantitative capabilities, only qualitatively identifying source areas and failing to quantify the contribution ratio of different source areas, thus failing to meet the needs of refined geological research.

[0004] To address the aforementioned shortcomings, providing an analytical system that employs multiple indicators and boasts high accuracy and strong quantification capabilities is a pressing technical problem that needs to be solved in this field. Summary of the Invention

[0005] In view of this, the present invention provides a Quaternary loess isotope source tracing analysis system to solve the problems existing in the background art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A source tracing and analysis system based on Quaternary loess isotopes includes: The data acquisition and processing module acquires multi-isotope data, grain size data, and chemical composition data of Quaternary loess samples; it preprocesses the acquired data and constructs a standardized dataset. The multi-dimensional feature fusion module establishes a three-dimensional feature tensor based on the standardized dataset, reduces the dimensionality of the three-dimensional feature tensor, calculates the information entropy based on the principal component contribution rate obtained from the decomposition of the three-dimensional feature tensor, dynamically allocates adaptive weights for each modality, and finally obtains a coupled feature vector through weighted summation, and outputs the coupled feature vector and the weight allocation result. The source contribution quantification module receives the coupled feature vector and weight allocation result, uses the multi-isotope endmember value as a constraint, solves the problem through a Bayesian mixture linear model and constraint optimization, and outputs the contribution ratio of each potential source region. The migration path tracing module receives the contribution ratio, combines the isotope fractionation effect and distance decay model, matches the optimal source migration path by constructing a path cost function, and outputs the migration path parameters. The result verification module receives the migration path parameters and the contribution ratio, and verifies the reliability of the results by checking the consistency between the root mean square error and the contribution.

[0007] Preferably, in the data acquisition and processing module, the multi-isotope data includes... 87 Sr / 86 Sr、 143 Nd / 144 Nd, d 18 O、 d 13 C, Particle size data includes the volume fraction of sand, powder, and clay particles, and chemical composition data includes the mass fraction of SiO2, Al2O3, Fe2O3, and CaO.

[0008] Preferably, the three-dimensional feature tensor is tensor elements This is the product of the standardized values ​​of the three corresponding indicators; =4 corresponds to four multiple isotope indices. =3 corresponds to 3 granularity indices. =4 corresponds to 4 chemical component indicators.

[0009] Preferably, the dimensionality reduction of the three-dimensional feature tensor specifically includes: ; Where G is the core tensor, These are the factor matrices for the three types of indicators; Principal component contribution rate The calculation formula is: ; in, Factor matrix The feature value of the kth column, This represents the number of principal components in this mode.

[0010] Preferably, the weight calculation in the multi-dimensional feature fusion module adopts the information entropy method, and the formula is: ; in, Let i be the weight of the i-th type of indicator. i =1, 2, and 3 correspond to three categories of indicators: isotope, particle size, and chemical composition, respectively.

[0011] Preferably, the formula for calculating the coupled feature vector in the multi-dimensional feature fusion module is: ; in, For the core tensor G Along the first i The expansion matrix of the modes, This represents the dimensionality reduction feature of the mode after tensor decomposition.

[0012] Preferably, the core equation of the source contribution quantification module is: ; objective function ; in, The ratios of the four isotopes are as follows: For the first j The proportion of material contribution from each potential material source area meets the requirements. and , The penalty coefficient is... This is the initial value for the source contribution based on granularity endmembers. For the corresponding first eigenvector in the coupled feature vector k Characteristic components of isotopes, For the first j The first potential resource area k The end-member ratio of the isotopes denoted as the deviation between the observed value and the model prediction of the k-th isotope ratio.

[0013] Preferably, the isotope fractionation correction equation in the migration path tracing module is: ; The path cost function is ; in, d The distance from the source region, The fractionation coefficient is... This is the path weight coefficient. This is the paleowind speed derived from the granularity parameter inversion.

[0014] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a source tracing analysis system based on Quaternary loess isotopes. By integrating multi-source data of multiple isotopes, grain size and chemical composition, it achieves deep fusion of multi-dimensional information by constructing a three-dimensional feature tensor and adaptive weight allocation. It achieves accurate quantification of the source contribution ratio by relying on a Bayesian mixed linear model, completes migration path tracing by combining isotope fractionation and distance decay models, and ensures the reliability of the results through dual verification. It effectively solves the defects of traditional methods such as cumbersome data processing, reliance on a single indicator, strong subjectivity and inability to quantify source contribution. It significantly improves the efficiency, accuracy and objectivity of Quaternary loess source tracing and meets the needs of fine geological research. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the structure provided by the present invention; Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] This invention discloses an isotopic provenance tracing and analysis system based on Quaternary loess, such as... Figure 1 As shown, it includes: The data acquisition and processing module acquires multi-isotope data, grain size data, and chemical composition data of Quaternary loess samples; it preprocesses the acquired data and constructs a standardized dataset. The multi-dimensional feature fusion module establishes a three-dimensional feature tensor based on a standardized dataset, reduces the dimensionality of the three-dimensional feature tensor, calculates the information entropy based on the principal component contribution rate obtained from the decomposition of the three-dimensional feature tensor, dynamically allocates adaptive weights for each modality, and finally obtains the coupled feature vector through weighted summation, and outputs the coupled feature vector and the weight allocation result. The source contribution quantification module receives the coupled feature vector and weight allocation results, uses multi-isotope endmember values ​​as constraints, solves the problem through a Bayesian mixture linear model and constraint optimization, and outputs the contribution ratio of each potential source region. The migration path tracing module receives the contribution ratio, combines the isotope fractionation effect and distance decay model, matches the optimal source migration path by constructing a path cost function, and outputs the migration path parameters. The results verification module receives migration path parameters and contribution percentages, and verifies the reliability of the results by checking the consistency between root mean square error and contribution percentage.

[0019] In one specific embodiment, the multi-isotope data in the data acquisition and processing module includes... 87 Sr / 86 Sr、 143 Nd / 144 Nd, d 18 O、 d 13 C, Particle size data includes the volume fraction of sand, powder, and clay particles, and chemical composition data includes the mass fraction of SiO2, Al2O3, Fe2O3, and CaO.

[0020] In one specific embodiment, the three-dimensional feature tensor is: tensor elements This is the product of the standardized values ​​of the three corresponding indicators; =4 corresponds to four multiple isotope indices. =3 corresponds to 3 granularity indices. =4 corresponds to four chemical component indicators. In source tracing, the indicative power of a single indicator is limited, but the combination of isotope binding particle size and chemical components is unique. Tensor elements are multiplied by the three types of indicators, which can strengthen this synergistic correlation and preserve the inherent logic of multidimensional data.

[0021] In one specific embodiment, dimensionality reduction of the three-dimensional feature tensor specifically includes: ; Where G is the core tensor, These are the factor matrices for the three types of indicators. High-dimensional tensors (4×3×4) have data redundancy, and directly inputting them into the model can lead to computational complexity and overfitting. Tucker decomposition, through the core tensor G and factor matrices, compresses the data dimensionality without destroying the multidimensional coupling relationship between isotopes, granularity, and chemical composition, effectively extracting information and preventing indicator data redundancy.

[0022] Principal component contribution rate The calculation formula is: ; in, Factor matrix The feature value of the kth column, This represents the number of principal components in this mode.

[0023] In one specific embodiment, the weight calculation in the multi-dimensional feature fusion module adopts the information entropy method, and the formula is: ; in, Let i be the weight of the i-th type of indicator. i =1, 2, and 3 correspond to three categories of indicators: isotopes, particle size, and chemical composition, respectively. Information entropy measures the information value of an indicator. The more concentrated the principal component contribution rate, the smaller the entropy value, indicating that the indicator's indicative power is more explicit; conversely, the larger the entropy value, the more dispersed the indicator's information. By converting information value into weights using 1-entropy, high-indicative indicators receive higher weights, and low-value indicators receive lower weights. This avoids the subjectivity of fixed weights in traditional methods and achieves data-driven adaptive weighting.

[0024] In one specific embodiment, the formula for calculating the coupled feature vector in the multi-dimensional feature fusion module is as follows: ; in, For the core tensor G Along the first i The expansion matrix of the modes, The dimensionality reduction feature of this mode after tensor decomposition retains the core indicative information of the indicator; after weighted summation, the coupled feature vector F not only integrates the key information of isotopes, particle size and chemical composition, but also highlights the role of high-value indicators through weights, transforming three-dimensional high-dimensional data into one-dimensional feature fingerprints, which simplifies the subsequent model input and ensures the integrity of information.

[0025] Geologically, loess samples are mixtures of sediments from multiple source areas, and their isotopic ratios are necessarily the result of weighting the end-member values ​​of each source area according to their contribution. The introduction of coupled feature components is because these components integrate grain size and chemical composition information, allowing the weighted calculations to not only match isotopes but also indirectly match grain size and chemical composition characteristics, resulting in higher fitting accuracy. This aligns with the geological logic of multi-evidence chain fusion. The core equation of the source contribution quantification module is: ; objective function ; in, The ratios of the four isotopes are as follows: For the first j The proportion of material contribution from each potential material source area meets the requirements. and , The penalty coefficient is... This is the initial value for the source contribution based on granularity endmembers. For the corresponding first eigenvector in the coupled feature vector k Characteristic components of isotopes, For the first j The first potential resource area k The end-member ratio of the isotopes This represents the deviation between the observed value and the model prediction of the k-th isotope ratio. A coupled characteristic component is introduced. F k Essentially, this involves fitting the model to multiple dimensions. The larger the coupled feature component, the higher the indicative weight of the corresponding isotope index. During fitting, the endmember features corresponding to that isotope are matched more preferentially, ensuring that the fitting result simultaneously satisfies isotope matching and synergistic matching of particle size and chemical composition, thereby improving the accuracy of source contribution calculation. (First item) To minimize the sum of squared residuals and ensure a good fit between the model's predicted values ​​and the measured values ​​(core of the likelihood function); the second term The initial contribution value of granular endmembers is introduced as a penalty term (Bayesian prior information) to avoid physically unreasonable results in the solution; constraints and Based on the fact that the contribution of material sources is a probability proportion, we ensure that the results conform to geological logic.

[0026] In one specific embodiment, the isotope fractionation correction equation in the migration path tracing module is: During wind migration, loess particles undergo slight isotopic fractionation due to distance and environmental factors (the fractionation is more pronounced with increasing distance), and the fractionation rate follows an exponential decay law, with a fractionation coefficient... These are empirical geological parameters. After correction, they can avoid path misjudgment caused by directly comparing sample values ​​with source end-member values, making distance calculations more consistent with actual geological processes.

[0027] The path cost function is ; in, d The distance from the source region, The fractionation coefficient is... This is the path weight coefficient. Paleowind speeds derived from granularity parameters; Ensure that the isotope correction value corresponding to the path matches the sample value, the second item Considering the energy cost of the migration process, the lower the paleowind speed, the greater the difficulty and cost of migration over the same distance d. This helps to avoid selecting paths with high fitting but infeasible migration, and to make path tracing more consistent with the geological laws of wind transport.

[0028] Dating was performed on the loess profile to determine the formation age of different strata; stratum thickness was measured and deposition rate was calculated. ,in h For the formation thickness, Δ tTo establish the time span, a wind speed-sedimentation rate relationship was established, and obtained by fitting the data through modern aeolian sedimentary observations. , k For regional calibration coefficients; paleosedimentation rates S Substituting the values, we obtain the ancient wind speed.

[0029] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. The methods disclosed in the embodiments are described simply because they correspond to the methods disclosed in the embodiments; relevant parts can be found in the method section.

[0030] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A system for tracing and analyzing the isotopic provenance of Quaternary loess, characterized in that, include: The data acquisition and processing module acquires multi-isotope data, particle size data, and chemical composition data of Quaternary loess samples. The collected data is preprocessed to construct a standardized dataset; The multi-dimensional feature fusion module establishes a three-dimensional feature tensor based on the standardized dataset, reduces the dimensionality of the three-dimensional feature tensor, calculates the information entropy based on the principal component contribution rate obtained from the decomposition of the three-dimensional feature tensor, dynamically allocates adaptive weights for each modality, and finally obtains a coupled feature vector through weighted summation, and outputs the coupled feature vector and the weight allocation result. The source contribution quantification module receives the coupled feature vector and weight allocation result, uses the multi-isotope endmember value as a constraint, solves the problem through a Bayesian mixture linear model and constraint optimization, and outputs the contribution ratio of each potential source region. The migration path tracing module receives the contribution ratio, combines the isotope fractionation effect and distance decay model, matches the optimal source migration path by constructing a path cost function, and outputs the migration path parameters. The result verification module receives the migration path parameters and the contribution ratio, and verifies the reliability of the results by checking the consistency between the root mean square error and the contribution.

2. The Quaternary loess isotope provenance tracing and analysis system according to claim 1, characterized in that, The data acquisition and processing module includes multi-isotope data. 87 Sr / 86 Sr、 143 Nd / 144 Nd, δ 18 O、 δ 13 C, Particle size data includes the volume fraction of sand, powder, and clay particles, and chemical composition data includes the mass fraction of SiO2, Al2O3, Fe2O3, and CaO.

3. The Quaternary loess isotope provenance tracing and analysis system according to claim 2, characterized in that, The three-dimensional feature tensor is tensor elements This is the product of the standardized values ​​of the three corresponding indicators; =4 corresponds to four multiple isotope indices. =3 corresponds to 3 granularity indices. =4 corresponds to 4 chemical component indicators.

4. The Quaternary loess isotope provenance tracing and analysis system according to claim 3, characterized in that, The dimensionality reduction of the three-dimensional feature tensor specifically includes: ； Where G is the core tensor, These are the factor matrices for the three types of indicators; Principal component contribution rate The calculation formula is: ； in, Factor matrix The feature value of the kth column, This represents the number of principal components in this mode.

5. The Quaternary loess isotope provenance tracing and analysis system according to claim 4, characterized in that, The weight calculation in the multi-dimensional feature fusion module uses the information entropy method, and the formula is: ； in, Let i be the weight of the i-th type of indicator. i =1, 2, and 3 correspond to three categories of indicators: isotope, particle size, and chemical composition, respectively.

6. The Quaternary loess isotope provenance tracing and analysis system according to claim 5, characterized in that, The formula for calculating the coupled feature vector in the multi-dimensional feature fusion module is as follows: ； in, For the core tensor G Along the first i The expansion matrix of the modes, This represents the dimensionality reduction feature of the mode after tensor decomposition.

7. The Quaternary loess isotope provenance tracing and analysis system according to claim 1, characterized in that, The core equation of the source contribution quantification module is: ； objective function ; in, The ratios of the four isotopes are as follows: For the first j The proportion of material contribution from each potential material source area meets the requirements. and , The penalty coefficient is... This is the initial value for the source contribution based on granularity endmembers. For the corresponding first eigenvector in the coupled feature vector k Characteristic components of isotopes, For the first j The first potential resource area k The end-member ratio of the isotopes denoted as the deviation between the observed value and the model prediction of the k-th isotope ratio.

8. The Quaternary loess isotope provenance tracing and analysis system according to claim 7, characterized in that, The isotope fractionation correction equation in the migration path tracing module is: ; The path cost function is ; in, d The distance from the source region, The fractionation coefficient is... This is the path weight coefficient. This is the paleowind speed derived from the granularity parameter inversion.

Images