A loess stratum genesis determination method and system under multi-source geological information constraint

Abstract

The application discloses a loess stratum genesis determination method and system under multi-source geological information constraints, relates to the technical field of stratum genesis determination, and comprises the following steps: acquiring multi-source remote sensing data of a target loess distribution area; constructing a dynamic model of dust transport path; collecting loess samples of the target area, determining constant element composition and trace element composition, and establishing a geochemical fingerprint library; performing fingerprint feature matching of the loess samples of the target area, calculating the contribution proportion of each source; and generating a stratum genesis determination result. The application solves the technical problem that the conventional static analysis method cannot realize dynamic determination of mixed genesis of loess stratum and accurate source tracing, and cannot determine the genesis of loess stratum accurately, achieves accurate source tracing and dynamic genesis determination of controversial mixed genesis of loess stratum, effectively solves the limitation of the conventional static analysis, and improves the technical effect of the accuracy of the determination of the genesis of loess stratum.

Description

Technical Field

[0001] This invention relates to the field of stratigraphic genesis determination technology, specifically to a method and system for determining the genesis of loess strata under the constraint of multi-source geological information. Background Technology

[0002] Loess, as an important record of paleoclimate and paleoenvironmental changes, has always been a core topic in Quaternary geological research, with its depositional processes and provenance tracing being key aspects. Current studies primarily employ traditional stratigraphic correlation, grain size analysis, and macro-element testing to analyze loess depositional characteristics. However, these methods fall short in retrieving loess transport pathways, accurately identifying provenance areas, and dynamically simulating depositional processes. This hinders a refined inversion of loess depositional processes and precise provenance tracing, limiting a deeper understanding of the geological evolution of loess distribution areas. Furthermore, existing techniques have limitations in characterizing the spatiotemporal evolution of loess deposition and elucidating depositional dynamics, resulting in insufficient accuracy and precision in determining the genesis of loess deposits.

[0003] Existing technologies, particularly traditional static analysis methods, struggle to dynamically determine the origin of loess strata and accurately trace their provenance, leading to insufficient accuracy in determining the origin of loess strata. Summary of the Invention

[0004] This application provides a method and system for determining the genesis of loess strata under the constraint of multi-source geological information. It is used to address the technical problem that traditional static analysis methods in the prior art are unable to achieve dynamic determination of the mixed genesis of loess strata and accurate traceability of material sources, resulting in insufficient accuracy in determining the genesis of loess strata.

[0005] In view of the above problems, this application provides a method and system for determining the genesis of loess strata under the constraint of multi-source geological information.

[0006] The first aspect of this application provides a method for determining the genesis of loess strata under the constraint of multi-source geological information, the method comprising:

[0007] Multi-source remote sensing data of the target loess distribution area is acquired. Based on the multi-source remote sensing data, surface mineral composition, surface roughness, and micro-topographic elevation features are extracted. Based on the surface mineral composition, surface roughness, and micro-topographic elevation features, and combined with wind speed and direction parameters from meteorological data, a dynamic model of dust transport path is constructed. Loess samples from the target area are collected, and the major and trace element compositions of the target loess distribution area are determined through elemental analysis. Simultaneously, a geochemical fingerprint database of known source areas is established, which includes elemental ratios and rare earth element distribution patterns. Based on the major and trace element compositions, the source area fingerprint features of the loess samples from the target area are matched using the geochemical fingerprint database to calculate the contribution ratio of each source. Based on the source contribution ratio and combined with the dynamic model of dust transport path, stratigraphic genesis is estimated, generating stratigraphic genesis determination results. The stratigraphic genesis determination results include the spatial distribution and temporal series variation trend of material sources.

[0008] A second aspect of this application provides a system for determining the genetic origin of loess strata under the constraint of multi-source geological information, the system comprising:

[0009] The remote sensing data acquisition module is used to acquire multi-source remote sensing data of the target loess distribution area, and extract surface mineral composition, surface roughness, and micro-topographic elevation features based on the multi-source remote sensing data. The path dynamic model construction module is used to construct a dynamic model of dust transport paths based on the surface mineral composition, surface roughness, and micro-topographic elevation features, combined with wind speed and direction parameters from meteorological data. The geochemical fingerprint database establishment module is used to collect loess samples from the target area, determine the major and trace element composition of the target loess distribution area through elemental analysis, and simultaneously establish known source areas. The system includes a geochemical fingerprint database containing elemental ratios and rare earth element distribution patterns; a source area fingerprint feature matching module, used to perform source area fingerprint feature matching of loess samples in the target area based on the major and trace element compositions and the geochemical fingerprint database, and calculate the contribution ratio of each source; and a genetic determination result generation module, used to perform stratigraphic genetic estimation based on the source contribution ratio and the dynamic model of wind and dust transport path, and generate stratigraphic genetic determination results, which include the spatial distribution and time series variation trend of material sources.

[0010] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0011] This method involves acquiring multi-source remote sensing data of the target loess distribution area; constructing a dynamic model of dust transport paths by combining wind speed and direction parameters from meteorological data; collecting loess samples from the target area and determining the major and trace element composition of the target loess distribution area through elemental analysis, while simultaneously establishing a geochemical fingerprint database of known source areas; matching the source area fingerprint characteristics of the loess samples in the target area using the geochemical fingerprint database based on the major and trace element compositions, and calculating the contribution ratio of each source; and then, based on the source contribution ratio and the dynamic model of dust transport paths, inferring stratigraphic genesis and generating stratigraphic gene determination results. This method achieves accurate source tracing and dynamic gene determination for the mixed genesis of disputed loess strata, effectively overcoming the limitations of traditional static analysis and improving the accuracy of loess strata gene determination. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A schematic flowchart of a method for determining the genesis of loess strata under the constraint of multi-source geological information provided in this application embodiment;

[0014] Figure 2 This is a schematic diagram of a loess stratum genesis determination system under multi-source geological information constraints, provided in an embodiment of this application.

[0015] Figure labeling: Remote sensing data acquisition module 10, path dynamic model construction module 20, geochemical fingerprint database establishment module 30, source area fingerprint feature matching module 40, and genetic determination result generation module 50. Detailed Implementation

[0016] This application provides a method and system for determining the genesis of loess strata under the constraint of multi-source geological information. This method addresses the technical problem that traditional static analysis methods in the prior art are unable to achieve dynamic determination of the mixed genesis of loess strata and accurate traceability of material sources, resulting in insufficient accuracy in determining the genesis of loess strata.

[0017] 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 a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0018] Example 1, as Figure 1 As shown, this application provides a method for determining the genesis of loess strata under the constraint of multi-source geological information, the method comprising:

[0019] Step S100: Obtain multi-source remote sensing data of the target loess distribution area, and extract surface mineral composition, surface roughness and micro-topographic elevation features based on the multi-source remote sensing data.

[0020] Specifically, multi-source remote sensing data of the target loess distribution area is obtained through remote sensing detection methods. This multi-source remote sensing data includes hyperspectral remote sensing data, synthetic aperture radar (SAR) data, and lidar (LiDAR) data. The surface mineral composition of the target area is extracted by relying on the spectral recognition characteristics of hyperspectral remote sensing data, the surface roughness of the target area is analyzed by utilizing the topographic detection advantages of SAR data, and the micro-topographic elevation characteristics of the target area are obtained by leveraging the high-precision ranging capabilities of lidar data, thus completing the accurate extraction of multi-dimensional surface features.

[0021] Step S200: Based on the surface mineral composition, surface roughness, and micro-topographic elevation characteristics, and combined with the wind speed and wind direction parameters in the meteorological data, construct a dynamic model of the dust transport path.

[0022] Specifically, the study first interprets the surface mineral assemblages based on the extracted surface mineral composition, identifying characteristic mineral endmembers representing different source areas. Simultaneously, relying on surface roughness and micro-topographic elevation characteristics, it quantitatively assesses the surface dynamic parameters of the target loess distribution area and its surrounding potential source areas. Next, it extracts core parameters of wind speed and direction from meteorological data of the target loess distribution area, couples them with characteristic mineral endmembers and surface dynamic parameters, and uses a particle trajectory model to simulate the entire process of dust particles being lifted, transported, and settled from the potential source area to the target area. Simulation data under different wind field scenarios are then integrated. Subsequently, cluster analysis is performed on the particle trajectories of different wind field scenarios and different dust-initiating points in the simulation data to identify frequently occurring dominant transport paths. The dust flux intensity corresponding to each dominant transport path, as well as the corresponding combination of wind speed, wind direction, and surface dynamic parameters, are statistically analyzed. All relevant data are integrated to generate a multi-scenario path-flux relationship matrix. Finally, based on this matrix, a dynamic model of dust transport paths reflecting the multi-temporal and spatial scale dust migration patterns is constructed.

[0023] Step S300: Collect loess samples from the target area, determine the major and trace element composition of the target loess distribution area through elemental analysis, and establish a geochemical fingerprint database of known source areas, which includes element ratios and rare earth distribution patterns.

[0024] Specifically, field sampling was conducted in the target loess distribution area to collect sufficient loess samples. The major and trace element compositions of the loess samples from the target area were accurately determined using professional elemental analysis methods. Simultaneously, potential source areas surrounding the target loess distribution area were identified as known source areas. Bedrock, weathering crust, and loose surface sediments were collected as source end-member samples. Elemental analysis was performed on these samples to obtain their major and trace element compositions. Characteristic element ratios and rare earth element distribution patterns for each known source area were calculated, and a standardized fingerprint database was constructed. Cluster analysis was then performed on the standardized fingerprint database to screen for index combinations that showed significant differences between geological units and exhibited stable geochemical behavior. Finally, these index combinations and their corresponding source area attribute information were integrated to construct a systematic geochemical fingerprint database of known source areas, including element ratios and rare earth element distribution patterns.

[0025] Step S400: Based on the aforementioned constant element composition and trace element composition, the source area fingerprint characteristics of the loess samples in the target area are matched using the aforementioned geochemical fingerprint database, and the contribution ratio of each source is calculated.

[0026] Specifically, based on the major and trace element compositions determined from loess samples in the target area, the target characteristic element ratios and target rare earth element distribution parameters consistent with those defined in the geochemical fingerprint database are first calculated. Then, the aforementioned target characteristic parameters are comprehensively compared with the endmember features of each known source region pre-stored in the geochemical fingerprint database to generate accurate source region fingerprint feature matching results. Subsequently, potential contributing endmembers shown in the matching results from the geochemical fingerprint database are selected as model input endmembers. A Bayesian endmember mixture model is adopted, with the geochemical composition observations of the loess samples in the target area set as the model response variable and the geochemical composition of the input endmembers set as the model prediction variable. The model is run and the posterior probability distribution of each input endmember with respect to the response variable is estimated through iterative sampling. The optimal estimate is extracted from this distribution to complete the endmember decomposition of the geochemical composition of the target sample. Finally, the source contribution ratio of each known source region in the loess samples of the target area is quantitatively calculated.

[0027] Step S500: Based on the contribution ratio of the material source, and combined with the dynamic model of the dust transport path, the stratigraphic genesis is calculated to generate a stratigraphic genesis determination result, which includes the spatial distribution and time series variation trend of the material source.

[0028] Specifically, the source contribution ratios of different target sampling points and stratigraphic units are first spatially integrated and interpolated to generate a spatial distribution map of material sources that reflects the spatial differentiation characteristics of contribution intensity in each source area. Then, from the dynamic model of wind-dust transport paths, simulated transport paths and flux information corresponding to the formation period of the target area's strata are extracted. The spatial distribution map of material sources is then spatially overlaid with this simulated information for consistency analysis, thereby tracing and verifying possible transport channels for loess material migrating from potential source areas to the target area. Subsequently, using the source contribution pattern presented in the spatial distribution map of material sources as spatial boundary conditions and stratigraphic chronology constraints as temporal boundary conditions, wind-dust transport is driven... A dynamic model of transport paths is used to simulate the transport and deposition of dust along verified possible transport channels under historical climate scenarios. Simultaneously, different spatial units in the spatial distribution map of material sources are assigned corresponding stratigraphic chronological constraints. Under these constraints, the spatial distribution maps of material sources from different periods are arranged in a time series, and principal component analysis is performed on the arranged maps to extract the main patterns of the evolution of the source contribution structure over time. Finally, based on these main patterns, time-varying curves of the contribution proportions of key source areas are plotted, quantifying the rate of change and evolutionary turning points, completing the overall stratigraphic genetic estimation, and ultimately generating a stratigraphic genetic determination result that includes the spatial distribution of material sources and the time-series trends of source contribution changes.

[0029] In one possible implementation, step S200 further includes:

[0030] Step S210: Based on the surface mineral composition, interpret the surface mineral assemblage and identify characteristic mineral endmembers representing different source regions.

[0031] Step S220: Based on the surface roughness and micro-topographic elevation characteristics, quantitatively evaluate the surface dynamic parameters of the target loess distribution area and its surrounding potential source area.

[0032] Step S230: Extract wind speed and wind direction parameters from meteorological data of the target loess distribution area, couple the characteristic mineral end-member with the surface dynamic parameters, use a particle trajectory model to simulate the lifting, transport and settling process of wind-blown dust particles, and integrate simulation process data under different wind field scenarios.

[0033] Step S240: Based on the simulation process data, analyze the multi-temporal and spatial scale trajectory and flux intensity of windblown dust migrating from the potential source area to the target loess distribution area, and construct a dynamic model of windblown dust transport path.

[0034] Specifically, based on the spectral and quantitative data of surface mineral composition in the target loess distribution area extracted from multi-source remote sensing data, mineral spectral interpretation technology combined with X-ray diffraction mineral quantitative analysis method is used to systematically interpret and label the surface mineral assemblage in the region. At the same time, combined with regional geological background data, mineral types with provenance indication significance, stable geochemical behavior and significant differences in different potential provenance intervals are screened out through the mineral characteristic comparison analysis method of provenance areas. Finally, characteristic mineral end-members that can characterize each potential provenance area are accurately identified, and the provenance marker minerals are determined.

[0035] Based on the extracted raw data of surface roughness and micro-topographic elevation characteristics of the target loess distribution area and its surrounding potential source areas, surface roughness is first quantified using numerical quantification methods. The root mean square of profile undulation is calculated using the roughness profile method, and the fractal dimension method is used to solve for the fractal characteristic values ​​of the surface, transforming surface roughness into a quantifiable numerical index. Next, topographic parameters are quantitatively analyzed for micro-topographic elevation characteristics. GIS spatial analysis tools are used to extract quantitative topographic parameters such as slope, aspect, topographic relief, and elevation variation coefficient, comprehensively characterizing the spatial shape of the micro-topography. The study investigated the characteristics of surface and elevation changes. Subsequently, combining wind erosion dynamics theory, a correlation calculation model was established for quantitative indicators of surface roughness, quantitative parameters of micro-topography, and surface dynamic parameters. Through model deduction, core surface dynamic parameters such as critical wind speed for sand erosion in the target area and surrounding potential source areas, near-surface wind field disturbance coefficient, dynamic sand transport flux, and wind-blown dust particle transport resistance were calculated. At the same time, each parameter was normalized and graded for quantitative evaluation, ultimately forming a standardized and quantitative set of surface dynamic parameters for the target loess distribution area and its surrounding potential source areas.

[0036] Core meteorological parameters such as wind speed and direction at different time scales and spatial locations were extracted from long-term and real-time meteorological monitoring data of the target loess distribution area and standardized. These wind speed and direction parameters were then coupled with characteristic mineral end-members representing different source areas and surface dynamic parameters of the target area and surrounding potential source areas obtained through quantitative evaluation to construct a comprehensive input parameter system for the particle trajectory model. Subsequently, the coupled parameter system was imported into the particle trajectory model, using characteristic mineral end-members as the basis for identifying dust particles. Combined with the regional wind erosion and sand-raising conditions and transport resistance characteristics defined by surface dynamic parameters, the model accurately simulated the complete dynamic process of dust particles under different wind fields, from their initial lifting and initiation in the potential source area, their long-distance transport and migration with airflow, to their deposition and accumulation in the target loess distribution area. At the same time, multi-scenario simulation experiments were conducted under wind field scenarios with different wind speed levels, wind direction types, and wind field durations. The simulation process data under each scenario were collected, classified, and stored in all dimensions. Finally, the simulation data of the entire process of dust particle lifting, transport, and deposition under all wind field scenarios were integrated and consolidated to form a standardized multi-scenario simulation process dataset.

[0037] Based on integrated simulation data of dust particles under different wind field scenarios, this study employs a combination of spatiotemporal series analysis and cluster analysis to perform multi-dimensional analysis of the trajectory data of dust particles migrating from various potential source areas to the target loess distribution area. First, the migration stages are divided into short-term, medium-term, and long-term phases according to time scale, and the migration ranges are divided into near-source, intermediate-source, and distant-source phases according to spatial scale. This identifies the set of frequently occurring dominant dust transport paths at different spatiotemporal scales. Then, the flux density calculation method is used to quantitatively statistically analyze the dust particle transport flux intensity corresponding to each path, and the flux density of each path is recorded simultaneously. The model combines wind speed, wind direction, and surface dynamic parameters that match the intensity of dust transport. Then, it integrates the spatiotemporal trajectory characteristics, quantitative flux intensity, and supporting parameter combinations of all advantageous transport paths to construct a multi-scenario path-flux relationship matrix. This matrix couples the spatiotemporal laws of dust transport with dynamic parameters. Finally, based on this matrix, a dynamic model of dust transport path is built that can accurately reflect the dynamic changes in the migration trajectory and flux intensity of dust materials at multiple spatiotemporal scales under different wind field conditions. This model can realize the dynamic deduction and parameterized expression of the entire process of dust rising from the potential source area, transporting to the target area, and settling.

[0038] In one possible implementation, step S240 further includes:

[0039] Step S241: Perform cluster analysis on the particle trajectories of different wind field scenarios and different dust generation points in the simulation process data to identify the set of frequently occurring advantageous transport paths.

[0040] Step S242: Calculate the dust flux intensity corresponding to each dominant transport path, and record the corresponding wind speed, wind direction and surface dynamic parameters combination.

[0041] Step S243: Integrate the dust flux intensity, wind speed, wind direction and surface dynamic parameters corresponding to all advantageous transport paths to generate a multi-scenario path-flux relationship matrix.

[0042] Step S244: Based on the multi-scenario path-flux relationship matrix, construct a dynamic model of dust transport path.

[0043] Specifically, for the integrated full data of the dust particle trajectory simulation process covering different wind field scenarios such as wind field intensity and wind direction type, as well as dust generation points in different potential source areas, a spatial trajectory clustering algorithm is used to conduct systematic clustering analysis on all particle migration trajectories. Using trajectory spatial overlap, migration direction similarity, and spatiotemporal occurrence frequency as core clustering indicators, particle trajectories in different scenarios and different dust generation points are classified, merged, and feature extracted. Low-frequency and occasional migration trajectories are eliminated, and high-frequency and representative dust material migration trajectories under various spatiotemporal conditions are accurately screened and identified. These are then integrated to form a set of advantageous transport paths, clarifying the core channels for dust migration from potential source areas to target loess distribution areas.

[0044] The flux density integral method was used to statistically analyze the dust flux intensity of each identified dominant transport path. Based on the simulation data of the particle trajectory model, the number and mass of dust particles passing through the path per unit time and per unit cross-section were calculated to quantitatively obtain the dust flux intensity value of each path. At the same time, through simulation data tracing and matching, the specific wind speed, wind direction and azimuth parameters set during the simulation of each dominant transport path, as well as the combination of surface dynamic parameters of the target loess distribution area and the surrounding potential source area corresponding to the path, were accurately retrieved from the wind field scenario parameter database and the surface dynamic parameter database. Finally, through data association mapping technology, the identification information of each dominant transport path was bound one by one with the corresponding dust flux intensity, wind speed and wind direction parameters, and surface dynamic parameter combinations to construct a structured path-parameter association dataset, completing the statistics and recording of all data.

[0045] The data on all the dominant dust transport paths that have been statistically analyzed are systematically sorted out. Using different wind field scenarios as the row dimension and each dominant transport path as the column dimension, the dust flux intensity corresponding to each path is quantified, and the matching wind speed, wind direction and surface dynamic parameters are used as the core metadata of the matrix. The data is structured and systematically filled in according to the correlation logic of wind field scenario-transport path-parameter group. At the same time, various parameters are standardized and normalized to ensure data dimension uniformity and comparability. Finally, a multi-scenario path-flux relationship matrix is ​​generated that can clearly reflect the correspondence between each dominant transport path and its dust flux intensity and dynamic driving parameters under different wind field scenarios, realizing the intensive, correlated storage and presentation of dust transport-related data.

[0046] Using a multi-scenario path-flux relationship matrix as the core training and validation dataset, the following steps are taken: First, the input feature parameters such as wind speed, wind direction, surface roughness, and topographic relief within the matrix are standardized and preprocessed. The spatial coordinate sequence of the dominant transport path, i.e., latitude / longitude / planar coordinates and flux intensity values, are used as the model's output labels. Second, a hybrid machine learning algorithm combining a Long Short-Term Memory (LSTM) network and an attention mechanism is introduced to construct the core architecture of the model. The LSTM layer is used to capture the sequential features of wind field and surface conditions changing over time / space, while the attention mechanism accurately assigns weights to the contributions of different wind field-surface parameter combinations in the matrix to transport paths and fluxes. The model parameters are iteratively optimized through backpropagation. The model is trained using 70% of the sample data in the matrix and validated using 30% of the sample data until the spatial error of the model's path prediction is ≤ The relative error of flux prediction is ≤8%; then, a dynamic response module is built for the model. The preprocessed real-time input wind field parameters, i.e., wind speed and wind direction, and surface condition parameters, i.e., surface dynamic parameters, are fed into the trained LSTM-Attention model. The model automatically parses the feature vectors corresponding to the input parameters by calling the learned matrix parameter association rules, and dynamically outputs the spatial trajectory sequence of the advantageous transport path of dust particles from the potential source area to the target loess distribution area, as well as the predicted value of dust flux intensity for each path; finally, an iterative update mechanism is embedded in the model to supplement the multi-scenario path-flux relationship matrix with new wind field-surface condition-path-flux measured data, and the LSTM-Attention model is fine-tuned and trained regularly to ensure that the model can always accurately respond to changes in input parameters, and finally complete the construction of a dynamic model of dust transport path with dynamic prediction capabilities.

[0047] In one possible implementation, step S300 further includes:

[0048] Step S310: Locate the potential source area surrounding the target loess distribution area as the known source area, and collect bedrock, weathering crust and loose surface sediment samples as source end-member samples.

[0049] Step S320: Perform elemental analysis on the source end-member sample to obtain its major element composition and trace element composition, calculate the element ratio and rare earth element distribution pattern of each source region, and construct a standardized fingerprint database.

[0050] Step S330: Perform cluster analysis on the standardized fingerprint database to screen out index combinations that have significant differences among geological units and stable geochemical behavior.

[0051] Step S340: Integrate the index combination and its corresponding source region attribute information to construct a systematic geochemical fingerprint database.

[0052] Specifically, taking the target loess distribution area as the core, and comprehensively considering factors such as topography, water system distribution and wind field characteristics, the surrounding geological units with potential source contributions, such as bedrock mountains, weathering crust development zones and river terraces, are identified as known source areas. Subsequently, in accordance with the principle of systematic sampling, bedrock samples of different lithologies, weathering crust samples of different depths, and loose aeolian and alluvial sediment samples are collected evenly in each known source area, and uniformly used as source end-member samples to ensure the representativeness and coverage of the samples.

[0053] The contents of major elements such as Si, Al, Fe, Ca, Mg, K, and Na in the samples were determined using inductively coupled plasma optical emission spectrometry (ICP-OES). Trace elements such as Sr, Ba, Zr, Ti, V, Cr, Ni, and Pb, as well as rare earth elements such as La, Ce, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu, were precisely analyzed using inductively coupled plasma mass spectrometry (ICP-MS). The test data were then standardized, including outlier removal, instrument error correction, calculation of rare earth element distribution patterns using chondrite-normalized methods, and calculation of characteristic element ratios for different source regions, such as La / Yb, Zr / Hf, and Ti / Zr. Finally, the elemental contents, elemental ratios, and rare earth element distribution pattern data of all source endmembers were integrated to construct a standardized fingerprint database. This enables quantitative characterization and comparative analysis of the geochemical characteristics of different source regions, providing data support for source tracing.

[0054] Using a standardized fingerprint database as the analysis object, the K-means clustering algorithm was employed to perform systematic clustering analysis on geochemical indicators of source end-member samples from various geological units. Major element content, trace element content, characteristic element ratios, and rare earth element distribution pattern parameters were used as clustering variables. The Euclidean distance between samples was calculated to determine the clustering threshold, grouping samples with similar geochemical characteristics into one category, thus achieving precise clustering of samples from different source areas. Simultaneously, the ANOVA method was used to test the significance of the clustering results, screening out geochemical indicators with significant differences between different cluster groups (P < 0.05). Furthermore, by analyzing the geochemical behavior of each indicator throughout the weathering, transportation, and deposition processes, unstable indicators susceptible to migration and differentiation due to later diagenesis and environmental disturbances were eliminated. Finally, a combination of indicators with stable geochemical behavior and strong source differentiation ability was selected.

[0055] The selected index combinations with significant provenance distinguishing capabilities and stable geochemical behavior are correlated one-to-one with the known provenance area attribute information corresponding to each index, including the geological unit type, spatial location, lithological characteristics, and sample number of the provenance area, to establish a structured data association table. Subsequently, database construction technology is used to standardize and classify the correlated index data, provenance attribute information, and corresponding geochemical characteristic parameters, such as element content, element ratios, and rare earth distribution pattern parameters, clarifying the provenance indication significance and discrimination threshold of each index combination. At the same time, a data retrieval and query module is set up to enable rapid retrieval by provenance area, index type, characteristic parameters, and other dimensions. Finally, a systematic geochemical fingerprint database covering core geochemical indicators, provenance correspondences, and discrimination criteria is constructed.

[0056] In one possible implementation, step S400 further includes:

[0057] Step S410: Based on the constant element composition and trace element composition, calculate the target characteristic element ratio and target rare earth distribution parameters that are consistent with the definition in the geochemical fingerprint database.

[0058] Step S420: Compare the target feature element ratio and target rare earth distribution parameters with the end-member features of each known source region pre-stored in the geochemical fingerprint database to generate source region fingerprint feature matching results.

[0059] Step S430: Based on the fingerprint feature matching results of the source region, the geochemical composition of the target sample is decomposed into endmembers using a geochemical hybrid model to quantify the source contribution ratio of each known source region in the loess sample of the target region.

[0060] Specifically, high-precision elemental content testing is conducted on the target loess samples to accurately obtain their major and trace element composition data. Subsequently, the target characteristic element ratios, such as Ti / Zr, Zr / Hf, La / Nd, etc., and the target rare earth element distribution parameters, including total rare earth element ΣREE, light rare earth to heavy rare earth ratio LREE / HREE, europium anomaly δEu, cerium anomaly δCe, lanthanum / ytterbium ratio (La / Yb)N and gadolinium / ytterbium ratio (Gd / Yb)N after chondrite standardization, are calculated. During the calculation process, the data undergoes preprocessing such as outlier removal and standardization correction to ensure that the calculated characteristic element ratios and rare earth element distribution parameters are completely consistent with the corresponding parameters of the end-member samples in the geochemical fingerprint database in terms of format, dimensions, and calculation methods. This forms sample geochemical fingerprint characteristic parameters that can be directly used for source region matching and comparison.

[0061] The calculated target feature element ratios, such as Ti / Zr and La / Nd, and target rare earth element distribution parameters, such as ΣLREE / ΣHREE and δEu, are organized into standardized data vectors. These vectors are then matched one by one with the feature vectors of endmembers of known source regions in the geochemical fingerprint database. By calculating quantitative indicators such as Euclidean distance or cosine similarity, the degree of matching between the sample data and the features of each source endmember is analyzed, generating a matching result table containing matching degree scores and deviation values. Finally, the source regions are ranked according to the matching degree to clarify the contribution probability of each potential source region, forming fingerprint feature matching results that can be used for source identification and tracing analysis.

[0062] Using the fingerprint feature matching results of the source region as input, representative endmembers in loess sediments, such as bedrock weathering endmembers and aeolian dust transport endmembers, are selected as endmember components in the geochemical mixing model. Then, numerical algorithms such as least squares are used to fit the linear relationship between the geochemical composition of the target sample, such as major elements and rare earth elements, and each endmember component, to solve the proportion coefficient of each endmember in the total composition of the sample. At the same time, non-negativity constraints and summation constraints of the endmember contribution ratio are set to ensure that the calculation results are consistent with geological significance. Finally, the quantitative results of the source contribution ratio of each known source region to the target loess sample are output, completing the quantitative analysis of source contribution.

[0063] In one possible implementation, step S430 further includes:

[0064] Step S431: Select known source region data that are shown as potential contributing endmembers in the source region fingerprint feature matching results from the geochemical fingerprint database as model input endmembers.

[0065] Step S432: Define the observed geochemical composition of the loess sample in the target area as the model response variable, and use the geochemical composition of the model input endmember as the model prediction variable.

[0066] Step S433: Run the Bayesian endmember mixture model and estimate the posterior probability distribution of each model input endmember with respect to the response variable through iterative sampling.

[0067] Step S434: Extract the optimal estimate from the posterior probability distribution to complete the endmember decomposition of the geochemical composition of the target sample.

[0068] Specifically, data entries marked as potential contributor endmembers based on fingerprint feature matching results from the geochemical fingerprint database are retrieved and screened, and key information such as source region attributes and geochemical composition characteristics are extracted. Subsequently, the extracted data undergoes consistency verification and outlier removal to ensure data quality. Finally, the screened and verified valid data are integrated as input endmembers for the model, ensuring that all data input into the model are source endmember metadata that have potential contributions to the geochemical composition of the target sample.

[0069] The measured geochemical composition data of loess samples from the target area, including the effective observations after cleaning and correction of element content, rare earth distribution parameters, and characteristic ratios, are defined as model response variables to characterize the true geochemical features of the samples. Subsequently, the geochemical composition data of the screened and validated model input endmembers, namely the endmembers of each potential source area, are used as model prediction variables to establish a quantitative correlation between them and the response variables of the target samples.

[0070] To run a Bayesian endmember mixture model, the core structure of the model must first be defined. This model consists of a prior distribution, a likelihood function, and a posterior distribution. The prior distribution sets the initial probability distribution of the contribution ratio of each input endmember based on geological prior knowledge, and must satisfy the non-negativity constraint and the constraint that the summation is 1. The likelihood function is used to characterize the observed geochemical composition of the target sample, i.e., the fitting relationship between the response variable and the geochemical composition of each input endmember, i.e., the predictor variable, and quantifies the degree of deviation between the observed value and the model prediction value. The posterior distribution is derived by combining Bayes' theorem with the prior distribution and the likelihood function, and characterizes the probability distribution features of the contribution ratio of each input endmember. After the model is built, the Markov chain Monte Carlo iterative sampling algorithm is used for computation. First, key parameters such as the number of sampling iterations and the number of combustion phase steps are set. During the combustion phase, the sampling trajectory is gradually adjusted to eliminate the influence of the initial value, so that the sampling process reaches a stable convergence state. After the combustion phase ends, iterative sampling continues. Each sampling generates a new input endmember contribution ratio sample based on the results of the previous round. At the same time, the fitting error corresponding to the sample is calculated through the likelihood function to continuously optimize the sampling accuracy. After sufficient iterative sampling and completion of convergence verification, based on a large number of effective sampling samples, the posterior probability distribution of each model input endmember with respect to the response variable is finally estimated. This distribution can intuitively reflect the possible value range and corresponding probability density of the contribution ratio of each input endmember.

[0071] The optimal estimate of the contribution ratio of each endmember is extracted from the posterior probability distribution. Usually, the mean or median of the posterior distribution is selected as the optimal estimate. At the same time, the 95% confidence interval is calculated to quantify the estimation uncertainty. Based on the optimal estimate, the geochemical observations of the target sample are decomposed into each end, i.e., a linear combination of the source regions. The endmember decomposition is completed by minimizing the residual between the observations and the model fitting values. Finally, the contribution ratio of each source endmember to the geochemical composition of the target sample is quantified, realizing the quantitative analysis of the source contribution.

[0072] In one possible implementation, step S500 further includes:

[0073] Step S510: Spatially integrate and interpolate the contribution ratios of the material sources for different target sampling points or different stratigraphic units to generate a spatial distribution map of material sources that reflects the spatial differentiation of contribution intensity in each source region.

[0074] Step S520: Extract simulated transport path and flux information under wind conditions corresponding to the formation period from the dynamic model of the dust transport path.

[0075] Step S530: Perform spatial overlay and consistency analysis on the spatial distribution map of the material source and the simulated transport path and flux information to trace and verify the possible transport channels of loess materials.

[0076] Step S540: Based on the possible transport channels, extrapolate the spatiotemporal process of loess deposition and generate the time series variation trend that reveals the contribution of the source over time.

[0077] Specifically, using different target sampling points or stratigraphic units as spatial nodes, the source contribution ratio data corresponding to each node is extracted. Based on the Geographic Information System (GIS) platform, the Kriging interpolation method is used to perform spatial interpolation calculations on the discrete source contribution ratio data, generating a continuous spatial distribution surface of source contribution. Subsequently, the calculation results are visualized using mapping software to generate a spatial distribution map of material sources that reflects the spatial differentiation of contribution intensity in each source area, intuitively presenting the spatial variation and distribution characteristics of source contribution.

[0078] Based on a dynamic model of dust transport paths, we screened and matched climate and wind conditions corresponding to the formation period of the strata, and identified key dynamic factors such as wind direction and wind speed that dominated the period. Then, we called the model simulation module to extract the transport path, diffusion range and spatial distribution trajectory of dust particles under the wind conditions. At the same time, we quantified the dust flux values ​​in different regions, including flux magnitude and flux density, to accurately depict the spatial trajectory and material transport intensity characteristics of dust transport. Finally, we obtained transport path and flux information that matched the formation period of the strata.

[0079] The spatial distribution map of material sources, vector data of simulated transport paths, and raster data of flux information were uniformly imported into the GIS platform to complete coordinate registration and format standardization. Then, using spatial overlay analysis tools, the simulated transport paths were overlaid with the spatial distribution areas of material source contributions, and the characteristic parameters of the overlapping areas were extracted. Next, the spatial matching degree between high-value areas of material source contributions and transport paths and high-value areas of flux was calculated to quantitatively analyze the consistency between path direction and material source distribution. Then, the rationality of transport channels was verified by consistency verification indicators, and the key paths of material transport were traced by combining flux distribution characteristics. Finally, based on the spatial analysis results, the transport channels of loess materials from the source area to the deposition area were identified, completing the tracing and verification of material transport paths.

[0080] Based on the identified transport channels, combined with the stratigraphic sequence of loess deposits and quantitative data on provenance contributions at different periods, time units were divided according to depositional stages. Parameters such as the provenance contribution ratio, transport path characteristics, and deposition rate of each unit were extracted to construct a time series dataset. The changing trend of the provenance contribution ratio over time was analyzed using methods such as moving average and linear regression to identify the stage-specific fluctuations and long-term evolution patterns of the contribution intensity. The driving factors of provenance contribution changes were analyzed in conjunction with paleoclimate and paleogeographic background, ultimately generating a time-series trend map of provenance contribution changes over time, clearly presenting the dynamic evolution characteristics of provenance contribution during loess deposition.

[0081] In one possible implementation, step S540 further includes:

[0082] Step S541: The source contribution pattern shown in the spatial distribution map of the material source is used as the spatial boundary condition, and the stratigraphic chronology constraint is used as the temporal boundary condition.

[0083] Step S542: Based on the spatial and temporal boundary conditions, drive the dynamic model of the dust transport path to simulate the transport and deposition process of dust along the possible transport channels under historical climate scenarios.

[0084] Specifically, the spatial distribution map of the material source shows the contribution pattern of the material source as the spatial boundary condition, which clarifies the spatial distribution range of the material source area, the contribution intensity gradient, and the spatial constraint boundary of the sedimentary area; the stratigraphic chronology constraint is used as the temporal boundary condition to define the time stage, sedimentation sequence, and key time nodes corresponding to the aeolian transport process, thus constructing an analytical framework that combines spatial pattern and temporal context.

[0085] Based on the defined spatial and temporal boundary conditions, a dynamic model of dust transport paths is driven to conduct simulation calculations. First, spatial parameters such as the spatial distribution of the source area, the contribution intensity gradient, and the constraint boundary of the sedimentary area are input. Combined with temporal parameters such as key time nodes and climate stages determined by stratigraphy, a spatiotemporal framework for model simulation is constructed. Then, historical climate scenario data, including environmental parameters such as paleoprecipitation and paleowind, are called to drive the model to simulate the entire process of dust material initiation, migration, and deposition along potential transport channels. During the simulation, the sedimentation simulation results are iteratively fitted with measured data such as the actual loess layer thickness, sediment grain size distribution, and source contribution ratio in the target area. Through comparative analysis, the dynamic parameters and transport efficiency coefficients of the model are continuously optimized until the simulation results and measured data reach the optimal matching degree. Finally, the complete spatiotemporal process of dust material from release, transport, to final deposition in historical periods is reconstructed, clearly depicting the path evolution, flux changes, and sedimentary distribution characteristics of dust transport.

[0086] In one possible implementation, step S540 further includes:

[0087] Step S543: Assign geochronological constraints to the corresponding strata of different spatial units in the spatial distribution map of the material source.

[0088] Step S544: Under the chronological constraints, arrange the spatial distribution maps of material sources from different periods into a time series.

[0089] Step S545: Perform principal component analysis on the spatial distribution map of the material sources arranged in time series to extract the main patterns of the evolution of the source contribution structures over time.

[0090] Step S546: Based on the main pattern, plot the time variation curve of the contribution ratio of the key source region, and quantify the rate of change and inflection point of the ratio to form a time series trend.

[0091] Specifically, the chronological constraints of the corresponding strata are matched to different spatial units in the spatial distribution map of material sources. That is, by comparing strata and dating data, such as optically stimulated luminescence, carbon-14, and paleomagnetism, the formation time interval, chronological order, or time node of the corresponding strata of each spatial unit is determined, the time coordinate of each spatial unit is clarified, and the relationship between the material source contribution and the time dimension is established.

[0092] Based on the established chronological constraints of each spatial unit, and according to the chronological order determined by stratigraphic formation time and dating results such as optically stimulated luminescence (OSL), carbon-14 dating, and paleomagnetic dating, spatial distribution maps of material sources in different periods are arranged systematically. First, the chronological nodes and time periods corresponding to each distribution map are extracted, clarifying their geological period, sedimentary stage, and chronological relationship. Then, they are arranged sequentially from early to late along the timeline, constructing a coherent temporal sequence framework. During this process, it is necessary to ensure that the distribution maps of adjacent time periods are naturally connected in terms of spatial range and source contribution characteristics, without logical breaks or overlaps. Finally, a set of spatial distribution sequences of material sources arranged according to temporal evolution is formed, clearly reflecting the spatial evolution and temporal characteristics of source contributions in different periods.

[0093] Principal component analysis was conducted on the spatial distribution map of material sources arranged in time series. First, the spatial distribution data of material source contributions in each period were organized into a standardized matrix. Using the type of material source area and the intensity of contribution as variables, the principal components that dominate the spatiotemporal changes of material source contributions were extracted by dimensionality reduction to clarify the explanatory power of each component for the variation of material source patterns. Then, typical patterns of material source contribution evolution over time were identified, including stable contribution type, fluctuating increase and decrease type, and stage transition type, and the duration and turning points of different patterns were quantified. Finally, the spatiotemporal evolution law of material source contribution structure was extracted to provide a quantitative basis for tracing the changes in transport channels and material supply.

[0094] Based on the main evolution patterns of provenance contribution structure extracted by principal component analysis, key provenance regions with high contributions to the overall change were first screened out, and the contribution ratios of these provenance regions in each period were extracted. A continuous time-varying curve was plotted with time as the horizontal axis and provenance contribution ratio as the vertical axis. Then, linear regression, sliding window fitting and other methods were used to quantify the rate of change of the contribution ratio of each provenance region. The time nodes when the contribution ratio changed significantly were identified by the mutation point detection algorithm. Combined with stratigraphic and geochronological constraints, the geological events or climate stages corresponding to the turning points were identified. Finally, a clear and quantitative trend of the time series change of provenance contribution was formed.

[0095] Example 2, based on the same inventive concept as the method for determining the genesis of loess strata under multi-source geological information constraints in the aforementioned examples, such as... Figure 2 As shown, this application provides a system for determining the genesis of loess strata under the constraint of multi-source geological information. The system and method embodiments in this application are based on the same inventive concept. The system includes:

[0096] The remote sensing data acquisition module 10 is used to acquire multi-source remote sensing data of the target loess distribution area, and extract surface mineral composition, surface roughness and micro-topographic elevation features based on the multi-source remote sensing data.

[0097] The path dynamic model construction module 20 is used to construct a dynamic model of the dust transport path based on the surface mineral composition, surface roughness and micro-topographic elevation characteristics, combined with wind speed and wind direction parameters in meteorological data.

[0098] The geochemical fingerprint database establishment module 30 is used to collect loess samples from the target area, determine the major and trace element composition of the target loess distribution area through elemental analysis, and establish a geochemical fingerprint database of known source areas. The geochemical fingerprint database includes element ratios and rare earth distribution patterns.

[0099] The source area fingerprint feature matching module 40 is used to perform source area fingerprint feature matching of loess samples in the target area based on the constant element composition and trace element composition and using the geochemical fingerprint database, and to calculate the contribution ratio of each source.

[0100] The causal determination result generation module 50 is used to perform stratigraphic origin estimation based on the contribution ratio of the material source and the dynamic model of the dust transport path, and generate stratigraphic origin determination results. The stratigraphic origin determination results include the spatial distribution and time series variation trend of the material source.

[0101] Furthermore, the system is also used to implement the following functions:

[0102] Based on the surface mineral composition, the surface mineral assemblage is interpreted to identify characteristic mineral endmembers representing different source areas. Based on the surface roughness and micro-topographic elevation characteristics, the surface dynamic parameters of the target loess distribution area and its surrounding potential source areas are quantitatively evaluated. Wind speed and direction parameters are extracted from meteorological data of the target loess distribution area, and the characteristic mineral endmembers are coupled with the surface dynamic parameters. A particle trajectory model is used to simulate the uplift, transport, and settling process of windblown dust particles, and simulation process data under different wind field scenarios are integrated. Based on the simulation process data, the multi-temporal and spatial scale trajectories and flux intensity of windblown dust migrating from potential source areas to the target loess distribution area are analyzed to construct a dynamic model of windblown dust transport paths.

[0103] Furthermore, the system is also used to implement the following functions:

[0104] Cluster analysis is performed on particle trajectories under different wind field scenarios and different dust-generating points in the simulation process data to identify the set of frequently occurring dominant transport paths; the dust flux intensity corresponding to each dominant transport path is statistically analyzed, and the corresponding wind speed, wind direction, and surface dynamic parameter combinations are recorded; the dust flux intensity, wind speed, wind direction, and surface dynamic parameter combinations corresponding to all dominant transport paths are integrated to generate a multi-scenario path-flux relationship matrix; based on the multi-scenario path-flux relationship matrix, a dynamic model of dust transport paths is constructed.

[0105] Furthermore, the system is also used to implement the following functions:

[0106] Potential source areas surrounding the target loess distribution area are identified as known source areas. Bedrock, weathering crust, and loose surface sediment samples are collected as source end-member samples. Elemental analysis is performed on these source end-member samples to obtain their major and trace element compositions. The elemental ratios and rare earth element distribution patterns of each source area are calculated to construct a standardized fingerprint database. Cluster analysis is performed on the standardized fingerprint database to screen out index combinations that exhibit significant differences among geological units and stable geochemical behavior. The index combinations and their corresponding source area attribute information are integrated to construct a systematic geochemical fingerprint database.

[0107] Furthermore, the system is also used to implement the following functions:

[0108] Based on the constant element composition and trace element composition, the target characteristic element ratios and target rare earth element distribution parameters consistent with those defined in the geochemical fingerprint database are calculated. The target characteristic element ratios and target rare earth element distribution parameters are compared with the endmember features of each known source region pre-stored in the geochemical fingerprint database to generate source region fingerprint feature matching results. Based on the source region fingerprint feature matching results, a geochemical mixture model is used to perform endmember decomposition of the geochemical composition of the target sample, and to quantify the source contribution ratio of each known source region in the loess sample of the target area.

[0109] Furthermore, the system is also used to implement the following functions:

[0110] From the geochemical fingerprint database, known source region data that appear as potential contributing endmembers in the fingerprint feature matching results of the source region are selected as model input endmembers; the observed geochemical composition of the loess sample in the target area is defined as the model response variable, and the geochemical composition of the model input endmembers is used as the model prediction variable; a Bayesian endmember mixture model is run, and the posterior probability distribution of each model input endmember with respect to the response variable is estimated through iterative sampling; the optimal estimate is extracted from the posterior probability distribution to complete the endmember decomposition of the geochemical composition of the target sample.

[0111] Furthermore, the system is also used to implement the following functions:

[0112] The contribution ratios of material sources for different target sampling points or different stratigraphic units are spatially integrated and interpolated to generate a spatial distribution map of material sources reflecting the spatial differentiation of contribution intensity in each source area. Simulated transport paths and flux information under wind conditions corresponding to the stratigraphic formation period are extracted from the dynamic model of wind-dust transport paths. The spatial distribution map of material sources is spatially overlaid and analyzed for consistency with the simulated transport paths and flux information to trace and verify possible transport channels for loess materials. Based on these possible transport channels, the spatiotemporal process of loess deposition is deduced, generating a time series trend revealing the change of material source contribution over time.

[0113] Furthermore, the system is also used to implement the following functions:

[0114] Using the source contribution pattern shown in the spatial distribution map of the material source as the spatial boundary condition and stratigraphic chronology constraints as the temporal boundary condition, the dynamic model of the dust transport path is driven based on the spatial and temporal boundary conditions to simulate the transport and deposition process of dust along the possible transport channels under historical climate scenarios.

[0115] Furthermore, the system is also used to implement the following functions:

[0116] Different spatial units in the spatial distribution map of material sources are assigned chronological constraints to their corresponding strata. Under these constraints, the spatial distribution maps of material sources from different periods are arranged in a time series. Principal component analysis is performed on the time series of the spatial distribution maps of material sources to extract the main patterns of the evolution of the source contribution structure over time. Based on these main patterns, time-varying curves of the contribution ratios of key source areas are plotted, and the rate of change and inflection points are quantified to form a time series trend.

[0117] It should be noted that the order of the embodiments described above is for descriptive purposes only and does not represent the superiority or inferiority of the embodiments. Specific embodiments of this specification have been described above. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0118] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0119] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A method for determining the genesis of loess strata under the constraint of multi-source geological information, characterized in that, The method includes: Acquire multi-source remote sensing data of the target loess distribution area, and extract surface mineral composition, surface roughness and micro-topographic elevation features based on the multi-source remote sensing data; Based on the surface mineral composition, surface roughness, and micro-topographic elevation characteristics, and combined with wind speed and wind direction parameters in meteorological data, a dynamic model of dust transport path is constructed. Loess samples were collected from the target area, and the major and trace element compositions of the target loess distribution area were determined by elemental analysis. At the same time, a geochemical fingerprint database of known source areas was established, which includes element ratios and rare earth distribution patterns. Based on the aforementioned major and minor element compositions, the source region fingerprint characteristics of loess samples from the target area are matched using the aforementioned geochemical fingerprint database, and the contribution ratio of each source is calculated. Based on the contribution ratio of the material source, and combined with the dynamic model of the dust transport path, the stratigraphic genesis is estimated, and a stratigraphic genesis determination result is generated. The stratigraphic genesis determination result includes the spatial distribution and time series variation trend of the material source. The dynamic model of dust transport path is constructed based on the surface mineral composition, surface roughness, and micro-topographic elevation characteristics, combined with wind speed and direction parameters from meteorological data, including: Based on the surface mineral composition, the surface mineral assemblage is interpreted to identify characteristic mineral endmembers representing different source regions; Based on the surface roughness and micro-topographic elevation characteristics, the surface dynamic parameters of the target loess distribution area and its surrounding potential source area are quantitatively evaluated. Wind speed and direction parameters are extracted from meteorological data of the target loess distribution area. The characteristic mineral end-members are coupled with the surface dynamic parameters. A particle trajectory model is used to simulate the lifting, transport and settling process of wind-blown dust particles, and simulation process data under different wind field scenarios are integrated. Based on the simulation process data, the multi-temporal and spatial scale trajectory and flux intensity of windblown dust from potential source areas to target loess distribution areas are analyzed, and a dynamic model of windblown dust transport path is constructed. The establishment of a geochemical fingerprint database for known source regions includes: The potential source areas surrounding the target loess distribution area are identified as known source areas, and bedrock, weathering crust, and loose surface sediment samples are collected as source end-member samples. Elemental analysis was performed on the source end-member samples to obtain their major element composition and trace element composition, and the element ratios and rare earth element distribution patterns of each source region were calculated to construct a standardized fingerprint database. Cluster analysis was performed on the standardized fingerprint database to screen out the combination of indicators that showed significant differences among different geological units and had stable geochemical behavior. By integrating the aforementioned index combinations and their corresponding source region attribute information, a systematic geochemical fingerprint database is constructed.

2. The method for determining the genesis of loess strata under the constraint of multi-source geological information as described in claim 1, characterized in that, Based on the simulation data, the multi-temporal and spatial scale trajectories and flux intensity of windblown dust migrating from potential source areas to target loess distribution areas are analyzed, and a dynamic model of windblown dust transport paths is constructed, including: Cluster analysis was performed on the particle trajectories of different wind field scenarios and different dust generation points in the simulation process data to identify the set of frequently occurring dominant transport paths; The dust flux intensity corresponding to each dominant transport path was statistically analyzed, and the corresponding wind speed, wind direction, and surface dynamic parameters were recorded. By integrating the dust flux intensity, wind speed, wind direction and surface dynamic parameters corresponding to all advantageous transport paths, a multi-scenario path-flux relationship matrix is ​​generated. Based on the multi-scenario path-flux relationship matrix, a dynamic model of dust transport path is constructed.

3. The method for determining the genesis of loess strata under the constraint of multi-source geological information as described in claim 1, characterized in that, Based on the aforementioned major and minor element compositions, and utilizing the geochemical fingerprint database, the provenance fingerprint characteristics of loess samples from the target area are matched to calculate the contribution ratio of each provenance, including: Based on the constant element composition and trace element composition, calculate the target characteristic element ratio and target rare earth distribution parameters that are consistent with the definition in the geochemical fingerprint database; The target feature element ratio and target rare earth distribution parameters are compared with the end-member features of each known source region pre-stored in the geochemical fingerprint database to generate source region fingerprint feature matching results. Based on the fingerprint feature matching results of the source region, a geochemical hybrid model is used to perform endmember decomposition of the geochemical composition of the target sample, and to quantify the source contribution ratio of each known source region in the loess sample of the target region.

4. The method for determining the genesis of loess strata under the constraint of multi-source geological information as described in claim 3, characterized in that, The geochemical mixture model is a Bayesian endmember mixture model. Based on the fingerprint feature matching results of the source region, the geochemical mixture model is used to perform endmember decomposition of the geochemical composition of the target sample, including: From the geochemical fingerprint database, known source region data that are displayed as potential contributing endmembers in the source region fingerprint feature matching results are selected as model input endmembers; The observed geochemical composition of the loess samples in the target area is defined as the model response variable, and the geochemical composition of the model input endmember is used as the model prediction variable. Run the Bayesian endmember mixture model and estimate the posterior probability distribution of each model input endmember with respect to the response variable through iterative sampling; The optimal estimate is extracted from the posterior probability distribution to complete the endmember decomposition of the geochemical composition of the target sample.

5. The method for determining the genesis of loess strata under the constraint of multi-source geological information as described in claim 1, characterized in that, Based on the contribution ratio of the material source, and combined with the dynamic model of the dust transport path, stratigraphic genesis is estimated to generate stratigraphic genesis determination results. These results include the spatial distribution and temporal series variation trends of the material source, including: The contribution ratios of the material sources for different target sampling points or different stratigraphic units are spatially integrated and interpolated to generate a spatial distribution map of material sources that reflects the spatial differentiation of contribution intensity in each source region. From the dynamic model of dust transport path, simulated transport path and flux information under wind conditions corresponding to the formation period of the strata are extracted; The spatial distribution map of the material source is spatially overlaid with the simulated transport path and flux information for consistency analysis to trace and verify possible transport channels of loess material. Based on the possible transport channels, the spatiotemporal process of loess deposition is deduced, and the time series variation trend revealing the contribution of the source material changes over time is generated.

6. The method for determining the genesis of loess strata under the constraint of multi-source geological information as described in claim 5, characterized in that, Based on the possible transport channels, the spatiotemporal process of loess deposition is deduced, including: The spatial distribution map of the material sources shows the source contribution pattern as the spatial boundary condition, and the stratigraphic chronology constraint is used as the temporal boundary condition. Based on the aforementioned spatial and temporal boundary conditions, the dynamic model of the dust transport path is driven to simulate the transport and deposition process of dust along the possible transport channels under historical climate scenarios.

7. The method for determining the genesis of loess strata under the constraint of multi-source geological information as described in claim 5, characterized in that, Generating the time series trend revealing the change of source contribution over time includes: Assign geochronological constraints to the corresponding strata of different spatial units in the spatial distribution map of the material source; Under the aforementioned chronological constraints, the spatial distribution maps of material sources from different periods are arranged in a time series. Principal component analysis was performed on the spatial distribution map of the source of the substances arranged in time series to extract the main patterns of the evolution of the source contribution structure over time. Based on the main pattern, the time-varying curves of the contribution ratio of key source regions are plotted, and the rate of change and inflection points of the ratio are quantified to form a time series trend.

8. A loess stratigraphic genesis determination system under multi-source geological information constraints, characterized in that, The system is used to implement the method for determining the genesis of loess strata under the constraint of multi-source geological information as described in any one of claims 1-7, and the system includes: The remote sensing data acquisition module is used to acquire multi-source remote sensing data of the target loess distribution area, and extract surface mineral composition, surface roughness and micro-topographic elevation features based on the multi-source remote sensing data; The path dynamic model construction module is used to construct a dynamic model of the dust transport path based on the surface mineral composition, surface roughness and micro-topography elevation characteristics, combined with wind speed and wind direction parameters in meteorological data. The geochemical fingerprint database establishment module is used to collect loess samples from the target area, determine the major and trace element composition of the target loess distribution area through elemental analysis, and establish a geochemical fingerprint database of known source areas. The geochemical fingerprint database includes element ratios and rare earth distribution patterns. The source area fingerprint feature matching module is used to perform source area fingerprint feature matching of loess samples in the target area based on the constant element composition and trace element composition and using the geochemical fingerprint database, and to calculate the contribution ratio of each source. The gene formation determination result generation module is used to perform stratigraphic gene formation estimation based on the contribution ratio of the material source and the dynamic model of the wind and dust transport path, and generate stratigraphic gene formation determination results. The stratigraphic gene formation determination results include the spatial distribution and time series variation trend of the material source. The system is also used for: Based on the surface mineral composition, the surface mineral assemblage is interpreted to identify characteristic mineral endmembers representing different source regions; Based on the surface roughness and micro-topographic elevation characteristics, the surface dynamic parameters of the target loess distribution area and its surrounding potential source area are quantitatively evaluated. Wind speed and direction parameters are extracted from meteorological data of the target loess distribution area. The characteristic mineral end-members are coupled with the surface dynamic parameters. A particle trajectory model is used to simulate the lifting, transport and settling process of wind-blown dust particles, and simulation process data under different wind field scenarios are integrated. Based on the simulation process data, the multi-temporal and spatial scale trajectory and flux intensity of windblown dust from potential source areas to target loess distribution areas are analyzed, and a dynamic model of windblown dust transport path is constructed. The potential source areas surrounding the target loess distribution area are identified as known source areas, and bedrock, weathering crust, and loose surface sediment samples are collected as source end-member samples. Elemental analysis was performed on the source end-member samples to obtain their major element composition and trace element composition, and the element ratios and rare earth element distribution patterns of each source region were calculated to construct a standardized fingerprint database. Cluster analysis was performed on the standardized fingerprint database to screen out the combination of indicators that showed significant differences among different geological units and had stable geochemical behavior. By integrating the aforementioned index combinations and their corresponding source region attribute information, a systematic geochemical fingerprint database is constructed.

Images