Heavy mineral placer prediction method, device and equipment based on paleo-shoreline reconstruction

Abstract

This invention discloses a method, apparatus, and equipment for predicting heavy mineral placer deposits based on ancient coastline reconstruction. The method includes: determining the relative height of the ancient sea level in a sample area; generating a target ancient coastline using a region growing algorithm; generating a paleogeographic elevation model; identifying multiple paleogeographic units based on topographic location indices; selecting exposed paleogeographic units; identifying minerals in the exposed paleogeographic units to obtain a heavy mineral abundance map; constructing a hydrodynamic-sedimentary coupling model to identify sediment accumulation zones in the sample area and comparing them with the heavy mineral abundance map to obtain heavy mineral enrichment zones; dividing the sample area into positive and negative sample areas and training a heavy mineral placer deposit prediction model; predicting the mineralization probability distribution map of the target area; and determining the heavy mineral target area based on the mineralization probability distribution map. This invention improves the prediction accuracy of heavy mineral mineralization probability, thereby improving the efficiency and accuracy of heavy mineral placer deposit location exploration and can be applied to the field of resource exploration technology.

Description

Technical Field

[0001] This invention relates to the field of resource exploration technology, and in particular to a method, apparatus and equipment for predicting heavy mineral placer deposits based on ancient coastline reconstruction. Background Technology

[0002] Existing methods for exploring heavy mineral placer deposits mainly rely on traditional techniques, but these methods have significant limitations:

[0003] 1) Geological survey and borehole sampling techniques: Based on extensive exploration data from mining areas, previous researchers analyzed and organized the topography, grain size, mineral composition, geochemistry, and material sources of sample areas to delineate heavy mineral placer mineralization zones and infer enrichment mechanisms. However, this method is costly, inefficient, and fails to capture the control effect of large-scale ancient coastline migration on enrichment areas. Furthermore, the history of ancient coastline migration reflects sea-level changes and tectonic activity, factors that determine the spatial distribution of placer resources; therefore, it cannot reflect the spatial heterogeneity at the regional scale.

[0004] 2) Simple Remote Sensing Techniques: Some studies use satellite imagery for coastline monitoring, but lack multispectral or hyperspectral data fusion, making it impossible to accurately identify sedimentary environments associated with ancient coastlines (such as beach ridges, tidal channels, or lagoons). Simple remote sensing techniques can only provide macroscopic topographic information and cannot distinguish mineral spectral characteristics, thus making it difficult to correlate placer enrichment mechanisms. For example, silicate minerals such as ilmenite have unique Reststrahlen features in the 9-10 micrometer range, which belongs to the thermal infrared band. Ordinary multispectral data has a large band width and cannot capture the fine spectral structure of minerals.

[0005] Furthermore, the enrichment of heavy minerals is the result of the combined effects of long-term geological history (such as sea-level fluctuations and changes in ancient coastlines) and short-term hydrodynamic processes. Existing technologies lack a systematic analytical framework that combines "paleoenvironmental reconstruction" with "modern process simulation," resulting in insufficient prediction accuracy for heavy mineral placer enrichment areas, which in turn affects the efficiency and accuracy of heavy mineral placer location exploration.

[0006] The above problems urgently need to be addressed. Summary of the Invention

[0007] The purpose of this invention is to at least partially solve one of the technical problems existing in the prior art.

[0008] Therefore, one objective of this invention is to provide a method for predicting heavy mineral placer deposits based on ancient coastline reconstruction. This method improves the prediction accuracy of heavy mineral mineralization probability, thereby improving the efficiency and accuracy of heavy mineral placer deposit location exploration.

[0009] Another objective of this invention is to provide a heavy mineral placer prediction device based on ancient coastline reconstruction.

[0010] To achieve the above-mentioned technical objectives, the technical solutions adopted in the embodiments of the present invention include:

[0011] On one hand, embodiments of the present invention provide a method for predicting heavy mineral placer deposits based on ancient coastline reconstruction, comprising the following steps:

[0012] Acquire optical remote sensing data, thermal infrared remote sensing data, reference endmember spectral data, topographic elevation data, hydrodynamic environmental parameters, and geological and borehole data for the sample area;

[0013] Determine the relative height of the paleosea level in the sample area, and generate the target paleoshoreline using a region growing algorithm based on the relative height of the paleosea level and the topographic elevation data.

[0014] Based on the geological and borehole data and the target ancient coastline, a paleogeographic elevation model is generated. Based on the topographic location index, multiple paleogeographic units are identified according to the paleogeographic elevation model. Based on the optical remote sensing data and the geological and borehole data, paleogeographic exposed units are selected from the paleogeographic units.

[0015] Mineral identification is performed on the paleogeographic exposed units based on the thermal infrared remote sensing data and the reference end-member spectral data to obtain a heavy mineral abundance map of the sample area.

[0016] A hydrodynamic-sedimentary coupling model is constructed based on the hydrodynamic environmental parameters and the target paleo-coastline. Sedimentary accumulation zones in the sample area are identified based on the hydrodynamic-sedimentary coupling model, and the sedimentary accumulation zones are compared with the heavy mineral abundance map to obtain heavy mineral enrichment zones.

[0017] Based on the heavy mineral enrichment area, each grid region of the sample area is divided into positive sample region and negative sample region, and a heavy mineral placer prediction model is trained based on the positive sample region and the negative sample region.

[0018] The mineralization probability distribution map of the target area is predicted based on the heavy mineral placer prediction model, and the heavy mineral target area of ​​the target area is determined based on the mineralization probability distribution map.

[0019] Furthermore, in one embodiment of the present invention, determining the relative height of the paleosea level in the sample area, and generating a target paleocoastline using a region growing algorithm based on the relative height of the paleosea level and the topographic elevation data, specifically includes:

[0020] Based on the original geomorphic body of the sample area, the top surface elevation and formation age of the paleosea level geological markers are measured, and the relative height of the paleosea level is calculated based on the top surface elevation and the formation age.

[0021] or,

[0022] The global mean sea level height corresponding to the time interval between the target mineralization period and the current time is determined based on the global sea level change curve, and the average vertical tectonic rate of the sample area is obtained. Then, the relative height of the paleosea level is calculated based on the global mean sea level height, the average vertical tectonic rate, and the time interval.

[0023] A digital elevation model of the sample area is constructed based on the topographic elevation data. Seed pixels with an elevation less than or equal to the relative height of the ancient sea level are determined based on the digital elevation model and added to the ancient sea area set.

[0024] Starting from the seed pixel, determine whether the altitude of the neighboring pixels of the seed pixel is less than or equal to the relative altitude of the ancient sea level. If so, add the neighboring pixel as a new seed pixel to the ancient sea area set.

[0025] The ancient sea area grid region is determined based on the ancient sea area set, and the target ancient coastline is determined based on the boundary line of the ancient sea area grid region.

[0026] Furthermore, in one embodiment of the present invention, the step of generating a paleogeographic elevation model based on the geological and borehole data and the target paleo-coastline, identifying multiple paleogeographic units based on the topographic location index according to the paleogeographic elevation model, and selecting paleogeographic outcrop units from the paleogeographic units based on the optical remote sensing data and the geological and borehole data, specifically includes:

[0027] Based on the geological and borehole data, multiple target paleogeographic surfaces were identified, and the paleosurface elevation points of the target paleogeographic surfaces were determined.

[0028] Spatial interpolation is performed on the paleosurface elevation points to obtain the paleosurface grid surface. The paleosurface grid surface is then adjusted according to the target paleo-coastline to obtain the paleo-geomorphic elevation model.

[0029] The topographic location index of multiple micro-geomorphic units is calculated based on the paleomorphic elevation model, and the micro-geomorphic units are divided into paleomorphic ridge units, paleomorphic slope units, and paleomorphic valley units based on the topographic location index, thus obtaining the paleomorphic units.

[0030] The elevation changes of each paleogeographic unit are calculated based on the topographic elevation data and the paleogeographic elevation model, and suspected exposed units are screened out based on the elevation changes.

[0031] Based on the optical remote sensing data, identify whether the suspected outcrop unit has sand or gravel features, and designate the suspected outcrop unit with sand or gravel features as a high-probability outcrop unit.

[0032] Based on the geological and borehole data, it is verified whether the high-probability outcrop unit is an outcrop area or a region with extremely shallow strata burial depth, and the high-probability outcrop unit belonging to the outcrop area or region with extremely shallow strata burial depth is taken as the paleogeographic outcrop unit.

[0033] Furthermore, in one embodiment of the present invention, the step of identifying minerals in the paleogeographic exposed units based on the thermal infrared remote sensing data and the reference endmember spectral data to obtain a heavy mineral abundance map of the sample area specifically includes:

[0034] The pixel spectrum of each pixel in the paleogeographic exposed unit is determined based on the thermal infrared remote sensing data, and the spectral similarity between the pixel spectrum and the reference end-member spectral data is calculated. Based on the spectral similarity, high-potential areas with the spectral characteristics of the target mineral are screened out.

[0035] The relative abundance of heavy minerals in each pixel of the high-potential area is estimated using the LSU algorithm, and the heavy mineral abundance map is generated based on the relative abundance.

[0036] Furthermore, in one embodiment of the present invention, the step of constructing a hydrodynamic-sedimentary coupling model based on the hydrodynamic environmental parameters and the target paleocoastline, identifying sediment accumulation zones in the sample area based on the hydrodynamic-sedimentary coupling model, and comparing the sediment accumulation zones with the heavy mineral abundance map to obtain heavy mineral enrichment zones specifically includes:

[0037] The lower boundary topography is determined based on the target paleo-coastline. Paleo-wind direction parameters, paleo-wave period, and paleo-wave height are determined based on regional paleoclimate studies. The flow velocity field and wave energy flux density of the sample area are modeled using a hydrodynamic model to obtain the hydrodynamic-sedimentation coupling model.

[0038] The volumetric transport rate of the mixed sediments in each sub-region is calculated based on the hydrodynamic-sedimentation coupling model, and the sediment accumulation zone is selected based on the volumetric transport rate.

[0039] The sediment accumulation area is compared with the heavy mineral abundance map, and the sediment accumulation area with heavy mineral abundance greater than a preset abundance threshold is designated as the heavy mineral enrichment area.

[0040] Furthermore, in one embodiment of the present invention, the step of dividing each grid region of the sample area into positive sample regions and negative sample regions according to the heavy mineral enrichment region, and training a heavy mineral placer prediction model based on the positive sample regions and the negative sample regions, specifically includes:

[0041] The grid regions within the heavy mineral enrichment area in the sample region are designated as the positive sample regions, and the grid regions outside the heavy mineral enrichment area in the sample region are designated as the negative sample regions.

[0042] The spatial location features are determined based on the shortest Euclidean distance from the positive / negative sample regions to the target paleo-coastline; the paleo-geomorphic features are determined based on the paleo-geomorphic type of the positive / negative sample regions; the heavy mineral features are determined based on the heavy mineral abundance of the positive / negative sample regions; and the hydrodynamic features are determined based on the velocity field and wave energy flux density of the positive / negative sample regions.

[0043] Based on the spatial location features, paleogeographic features, heavy mineral features, and hydrodynamic features, corresponding positive / negative training samples are constructed, and the sample labels of positive training samples are set to mineral-containing, while the sample labels of negative training samples are set to non-mineral-containing, thus obtaining the training dataset.

[0044] The heavy mineral placer prediction model is obtained by training the training dataset using a random forest, gradient boosting tree, or neural network model.

[0045] Furthermore, in one embodiment of the present invention, the step of predicting the mineralization probability distribution map of the target area based on the heavy mineral placer prediction model, and determining the heavy mineral target area of ​​the target area based on the mineralization probability distribution map, specifically includes:

[0046] The mineralization probability of each pixel in the target area is predicted based on the heavy mineral placer prediction model, and the mineralization probability distribution map is generated based on the mineralization probability.

[0047] The favorable depositional zones in the target area were determined based on hydrodynamic-sedimentation coupled simulation.

[0048] When the mineralization probability is greater than a preset first threshold and the corresponding pixel is located in the favorable deposition area, the corresponding pixel is classified as a first-level heavy mineral target area.

[0049] When the mineralization probability is greater than or equal to a preset second threshold and less than or equal to the first threshold, and the corresponding pixel is located in the favorable deposition area, the corresponding pixel is divided into a secondary heavy mineral target area.

[0050] When the mineralization probability is greater than or equal to the second threshold and the corresponding pixel is not located in the favorable deposition area, the corresponding pixel is classified as a level three heavy mineral target area.

[0051] On the other hand, embodiments of the present invention provide a heavy mineral placer deposit prediction device based on ancient coastline reconstruction, comprising:

[0052] The data acquisition module is used to acquire optical remote sensing data, thermal infrared remote sensing data, reference end-member spectral data, topographic elevation data, hydrodynamic environmental parameters, and geological and borehole data of the sample area.

[0053] The ancient coastline generation module is used to determine the ancient sea level relative height of the sample area, and generate the target ancient coastline based on the ancient sea level relative height and the topographic elevation data using a region growing algorithm.

[0054] The paleogeographic outcrop unit screening module is used to generate a paleogeographic elevation model based on the geological and borehole data and the target paleo-coastline, identify multiple paleogeographic units based on the topographic location index and the paleogeographic elevation model, and screen out paleogeographic outcrop units from the paleogeographic units based on the optical remote sensing data and the geological and borehole data.

[0055] The heavy mineral abundance identification module is used to identify minerals in the paleogeographic exposed units based on the thermal infrared remote sensing data and the reference end-member spectral data, and to obtain a heavy mineral abundance map of the sample area.

[0056] The heavy mineral enrichment area determination module is used to construct a hydrodynamic-sedimentary coupling model based on the hydrodynamic environmental parameters and the target paleo-coastline, identify sediment accumulation areas in the sample area based on the hydrodynamic-sedimentary coupling model, and compare the sediment accumulation areas with the heavy mineral abundance map to obtain heavy mineral enrichment areas.

[0057] The model training module is used to divide each grid region of the sample area into positive sample regions and negative sample regions according to the heavy mineral enrichment area, and to train a heavy mineral placer prediction model based on the positive sample regions and the negative sample regions.

[0058] The heavy mineral target area determination module is used to predict the mineralization probability distribution map of the target area based on the heavy mineral placer prediction model, and to determine the heavy mineral target area of ​​the target area based on the mineralization probability distribution map.

[0059] On the other hand, embodiments of the present invention provide an electronic device, including:

[0060] At least one processor;

[0061] At least one memory for storing at least one program;

[0062] When the at least one program is executed by the at least one processor, the at least one processor implements the above-described method for predicting heavy mineral placer deposits based on ancient coastline reconstruction.

[0063] On the other hand, embodiments of the present invention also provide a computer-readable storage medium storing a processor-executable computer program that, when executed by a processor, implements the above-described method for predicting heavy mineral placer deposits based on ancient coastline reconstruction.

[0064] On the other hand, embodiments of the present invention also provide a computer program product, including a computer program that, when executed by a processor, implements the above-described method for predicting heavy mineral placer deposits based on ancient coastline reconstruction.

[0065] The advantages and beneficial effects of the present invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention:

[0066] This invention acquires optical remote sensing data, thermal infrared remote sensing data, reference endmember spectral data, topographic elevation data, hydrodynamic environmental parameters, and geological and borehole data of a sample area to determine the relative height of the paleosea level in the sample area. Based on the relative height of the paleosea level and the topographic elevation data, a target paleo-coastline is generated using a region growing algorithm. A paleo-geomorphic elevation model is generated based on the geological and borehole data and the target paleo-coastline. Multiple paleo-geomorphic units are identified based on the topographic location index and the paleo-geomorphic elevation model. Paleo-geomorphic exposed units are selected from these units based on the optical remote sensing data and the geological and borehole data. Finally, the paleo-geomorphic features are analyzed based on the thermal infrared remote sensing data and the reference endmember spectral data. Mineral identification is performed on exposed units to obtain a heavy mineral abundance map of the sample area. A hydrodynamic-sedimentary coupling model is constructed based on hydrodynamic environmental parameters and the target paleo-coastline. Sedimentary accumulation zones in the sample area are identified using this model, and compared with the heavy mineral abundance map to identify heavy mineral enrichment zones. Based on these enrichment zones, each grid area of ​​the sample area is divided into positive and negative sample areas. A heavy mineral placer prediction model is trained based on these positive and negative sample areas. The model predicts the mineralization probability distribution map of the target area, and the target heavy mineral area is determined based on this map. This embodiment of the invention, based on multi-source data fusion and process mechanism coupling, integrates remote sensing, paleoenvironmental indicators, and a hydrodynamic model to accurately characterize the transport path and enrichment sites of heavy minerals, improving the prediction accuracy of heavy mineral mineralization probability and thus enhancing the efficiency and accuracy of heavy mineral placer exploration. Attached Figure Description

[0067] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the embodiments of the present invention are described below. It should be understood that the drawings described below are only for the convenience of clearly describing some embodiments of the technical solutions of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0068] Figure 1 A flowchart illustrating the steps of a method for predicting heavy mineral placer deposits based on ancient coastline reconstruction, provided in an embodiment of the present invention;

[0069] Figure 2 A structural block diagram of a heavy mineral placer prediction device based on ancient coastline reconstruction provided in an embodiment of the present invention;

[0070] Figure 3 This is a structural block diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0071] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of this invention; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this invention as detailed in the appended claims.

[0072] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein is for the purpose of describing embodiments of the invention only and is not intended to limit the invention.

[0073] The method for predicting heavy mineral placer deposits based on ancient coastline reconstruction provided in this invention can be applied to terminals, servers, or software running on either terminal or server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, or vehicle terminal, but is not limited to these. The server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network. The software can be an application that implements the method for predicting heavy mineral placer deposits based on ancient coastline reconstruction, but is not limited to the above forms.

[0074] This invention can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This invention can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This invention can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0075] It should be noted that in various specific embodiments of the present invention, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when embodiments of the present invention require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirection to a confirmation page. Only after obtaining the user's separate permission or consent is the necessary user-related data for the normal operation of the embodiments of the present invention acquired.

[0076] Placer minerals refer to minerals with a density greater than 2.9 and stable physicochemical properties found in loose sedimentary sand grains formed from the weathering and transportation of rocks or ores. Common types include native gold, native platinum, diamond, magnetite, ilmenite, zircon, and rutile, which can form placer deposits after enrichment. The placer minerals in this application mainly refer to sand bodies rich in zircon, ilmenite, rutile, monazite, etc., in coastal sedimentary environments, and are widely used in the metallurgical, ceramics, and aerospace industries.

[0077] Existing methods for heavy mineral placer exploration lack a systematic analytical framework that combines paleoenvironmental reconstruction with modern process simulation, resulting in insufficient accuracy in predicting heavy mineral placer enrichment areas, which in turn affects the efficiency and accuracy of heavy mineral placer location exploration.

[0078] To this end, the present invention provides a method for predicting heavy mineral placer deposits that can integrate multi-source spatial data, penetrate the bottleneck of mineral spectral identification, and couple ancient and modern geological processes. This method achieves a complete technical solution from "identification" to "prediction" of the enrichment mechanism of heavy mineral placer deposits (including zircon, ilmenite, rutile, monazite, etc.), and is applicable to the exploration and evaluation of heavy mineral placer resources in coastal zones.

[0079] Reference Figure 1 This invention provides a method for predicting heavy mineral placer deposits based on ancient coastline reconstruction, specifically including the following steps:

[0080] S101. Acquire optical remote sensing data, thermal infrared remote sensing data, reference endmember spectral data, topographic elevation data, hydrodynamic environmental parameters, and geological and borehole data of the sample area.

[0081] S102. Determine the relative height of the ancient sea level in the sample area, and generate the target ancient coastline using the region growing algorithm based on the relative height of the ancient sea level and the topographic elevation data.

[0082] S103. Generate a paleogeographic elevation model based on geological and borehole data and the target paleo-coastline. Based on the topographic location index, identify multiple paleogeographic units according to the paleogeographic elevation model. Select paleogeographic exposed units from the paleogeographic units based on optical remote sensing data and geological and borehole data.

[0083] S104. Based on thermal infrared remote sensing data and reference end-member spectral data, mineral identification is performed on paleogeographic exposed units to obtain a heavy mineral abundance map of the sample area.

[0084] S105. Construct a hydrodynamic-sedimentary coupling model based on hydrodynamic environmental parameters and the target paleo-coastline. Identify sediment accumulation zones in the sample area based on the hydrodynamic-sedimentary coupling model, and compare the sediment accumulation zones with the heavy mineral abundance map to obtain heavy mineral enrichment zones.

[0085] S106. Based on the heavy mineral enrichment area, the sample area is divided into positive sample area and negative sample area, and a heavy mineral placer prediction model is trained based on the positive sample area and negative sample area.

[0086] S107. Based on the heavy mineral placer prediction model, the mineralization probability distribution map of the target area is obtained, and the heavy mineral target area of ​​the target area is determined based on the mineralization probability distribution map.

[0087] The embodiments of this invention are based on multi-source data fusion and process mechanism coupling. By integrating remote sensing, paleoenvironmental indicators and hydrodynamic models, it can accurately characterize the transport path and enrichment site of heavy minerals, improve the prediction accuracy of heavy mineral mineralization probability, and thus improve the efficiency and accuracy of heavy mineral placer deposit location exploration.

[0088] The specific steps of the embodiments of the present invention will be described in detail below.

[0089] The first step is multi-source data acquisition and preprocessing. This step is the data foundation of the entire research, and its core objective is to systematically collect, organize, and standardize all multi-source spatial and attribute data required for subsequent analyses. High-quality, uniformly formatted, and coordinate-consistent data is a prerequisite for ensuring the reliable operation of the entire technical chain from paleoenvironmental reconstruction to mineral prediction. The standardized database output from this step will directly serve all subsequent stages.

[0090] 1. Data list, the data to be collected is as follows:

[0091] (1) Optical remote sensing data: High-resolution series (such as GF-1B / 1C), Landsat-8 / 9 OLI and other multispectral data are used. The high spatial resolution information provided by its visible-near infrared band is mainly used to assist geological interpretation, land use classification and identification of modern surface exposure features in the third step.

[0092] (2) Thermal infrared remote sensing data: L1T level radiance data and L2 level surface emissivity (LSE) products from the ASTER sensor on Terra satellite were used. Given that opaque minerals (such as ilmenite) have diagnostic spectral features in the thermal infrared band (such as absorption valleys at ~10.75 μm), ASTER TIR data is the core data source for the fourth step of spectral angle mapping (SAM) and linear spectral unmixing (LSU) to directly identify heavy minerals and invert their abundance.

[0093] (3) Laboratory end-member spectral data: Fourier transform infrared spectroscopy (FTIR) was used to measure heavy mineral samples collected in the field. The measurement results (reflectivity R) were converted to emissivity (ε=1-R) according to Kirchhoff's law and resampled to the ASTER TIR band as reference end-members for subsequent spectral matching.

[0094] (4) Digital Elevation Model (DEM): 30-meter resolution DEM data or higher precision DEM data from the domestic "Tianhui" series or ALOS WORLD 3D are used. This data forms the basis of modern topography and is used for the reconstruction of the ancient coastline in the second step and the elevation change analysis in the third step.

[0095] (5) Hydrodynamic environmental parameters: Water temperature (°C), salinity (‰), and flow velocity (m / s) were simultaneously measured by deploying CTD (temperature, salinity, depth) and ADCP (acoustic Doppler current profiler) on site. This mainly provides boundary conditions and verification data for the hydrodynamic-sedimentation coupling simulation in step 5.

[0096] (6) Geological and borehole data: Collect geological maps, mineral exploration reports, and borehole data (including stratigraphic lithology, sedimentary facies, target strata top burial depth, heavy mineral analysis results, etc.) of the sample area, as well as the spatial location and grade information of known mineral deposits. These data are the source of control points for the third step of paleo-DEM restoration and also the positive and negative sample labels for the fifth step of machine learning model training.

[0097] 2. Data preprocessing and dataset construction:

[0098] (1) Spectral resampling:

[0099] To enable high-resolution spectral features measured in the laboratory to match the spectral response functions of remote sensing satellite sensors, thus allowing laboratory spectra to be directly used as "reference endmembers" for satellite image classification and identification, spectral resampling is necessary. Resampling involves calculating a weighted average of the spectral response functions for each satellite band, converting the high-precision "laboratory-scale" spectrum into "satellite-scale" spectral values ​​comparable to the image data. The resulting endmember spectral library matrix will serve as the direct input for the fourth step of spectral analysis.

[0100] For a given spectral band k in thermal infrared remote sensing data, its resampled emissivity value It can be calculated using the following formula:

[0101]

[0102] In the formula: λ is the calculated emissivity value (dimensionless, between 0 and 1) for the k-th band in the thermal infrared remote sensing data after resampling; λ is the wavelength (μm). and Let be the lower and upper limits of the spectral range for the k-th band; The high-resolution emissivity spectrum (dimensionless) obtained from laboratory measurements is a continuous curve. The spectral response function for the k-th band refers to the ratio of the received radiance to the incident radiance at each wavelength, describing the sensor's response intensity at different wavelengths (dimensionless, between 0 and 1). The spectral response function is obtained by the sensor's manufacturing and launch institutions through laboratory calibration and calculation before satellite launch.

[0103] (2) DEM data preprocessing: data mosaicking, projection transformation, filling data holes, removing obvious outliers to ensure elevation accuracy.

[0104] (3) Remote sensing image preprocessing: Radiometric calibration and atmospheric correction are performed on optical images to obtain surface reflectance; for thermal infrared images (ASTER), the official preprocessed L2 level surface emissivity (LSE) product is preferred to ensure spectral fidelity to the greatest extent and avoid errors introduced by additional processing.

[0105] (4) Standardization of geological and borehole data: Establish a GIS database, unify the coordinate system, and standardize the input of borehole attributes (such as grade, thickness, and burial depth) to ensure that they can be used as reliable inputs for spatial interpolation and model training.

[0106] (5) Coordinate system integration with database: All vector and raster data are converted to the same projected coordinate system and integrated into a unified spatial database in the GIS platform to ensure that the data is traceable and reusable.

[0107] The second step is the reconstruction of the ancient coastline, a crucial bridge connecting modern topography and the paleoenvironment. The location of the ancient coastline directly controls the energy zones of waves and coastal currents, thus determining the initial locations of placer deposits. An accurate ancient coastline is a prerequisite for subsequent identification of paleogeographic units (such as sandbars and lagoons). This step utilizes the DEM data processed in the first step, constrained by sea level height, to directly reconstruct the paleogeographic pattern.

[0108] As a further optional implementation, the relative height of the paleosea level in the sample area is determined, and the target paleocoastline is generated using a region growing algorithm based on the relative height of the paleosea level and topographic elevation data. This specifically includes:

[0109] S1021. Measure the top elevation and formation age of paleosea-level geological markers based on the original geomorphic bodies of the sample area, and calculate the relative height of paleosea-level based on the top elevation and formation age;

[0110] or,

[0111] S1022. Determine the global mean sea level height corresponding to the time interval between the target mineralization period and the current time based on the global sea level change curve, and obtain the average vertical tectonic rate of the sample area. Then, calculate the relative height of the paleosea level based on the global mean sea level height, the average vertical tectonic rate, and the time interval.

[0112] S1023. Construct a digital elevation model of the sample area based on topographic elevation data, determine seed pixels whose elevation is less than or equal to the relative height of the ancient sea level based on the digital elevation model, and add the seed pixels to the ancient sea area set.

[0113] S1024. Starting from the seed cell, determine whether the altitude of the adjacent cells of the seed cell is less than or equal to the relative altitude of the ancient sea level. If so, add the adjacent cells as new seed cells to the ancient sea area set.

[0114] S1025. Determine the ancient sea area grid region based on the ancient sea area set, and determine the target ancient coastline based on the boundary line of the ancient sea area grid region.

[0115] Specifically, the primary key to reconstructing ancient coastlines is determining the height (H) of the ancient sea level relative to the modern sea level. palaeo (Unit: meter). It can be determined using the following two methods:

[0116] 1) Direct dating and elevation measurement based on native landforms: Locate and identify well-preserved geological markers formed at paleosea level, such as shell ridges and beach rocks. Use high-precision GPS to measure their top elevation H. sample Samples were collected and subjected to radiocarbon or optically stimulated luminescence dating to determine their formation age t.

[0117] If the sample region is structurally stable, H can be approximated as H palaeo ≈H sample If the Global Post-Ice Balance Adjustment (GIA) needs to be considered, then the formula is used: H palaeo =H sample -ΔGIA(t), where ΔGIA(t) can be obtained from global models, for example, by querying NASA's ICE-6G_C (VM5a) model online calculator.

[0118] 2) Global Sea-Level Curve and Regional Tectonic Correction: Consult published global sea-level change curves to obtain the global mean sea-level height SLIP(t) (in meters) corresponding to the time t (in years) since the target mineralization period. Simultaneously, obtain the mean vertical tectonic rate U (uplift is positive, subsidence is negative, in meters / year) for the sample area from regional geological research literature. The paleo-sea-level relative height H can be calculated using the following formula. palaeo :

[0119]

[0120] Obtain H palaeo Then, the ancient coastline can be reconstructed using a digital elevation model (DEM) in a GIS environment, as shown in the following formula:

[0121]

[0122] In the formula: Con is a conditional judgment function in GIS raster calculation. Its function is to judge each cell in the raster; if the condition is true, it outputs a value (e.g., 1); if it is false, it outputs another value (e.g., 0). DEM is the digital elevation model matrix of the sample area, where the value of each cell represents the elevation (m) of that point, derived from the preprocessed DEM data in the first step. == is a logical operator representing "equal to". Since DEM is continuous data, completely identical cells are extremely rare. In practice, Con(DEM<=H) can be used. palaeo This method is used to extract all areas with elevations below or equal to the ancient sea level (i.e., ancient sea masking), but this may include some inland depressions that are not associated with the ancient ocean.

[0123] Therefore, a boundary condition needs to be introduced—only those low-lying areas connected to the "paleoceanic body outside the sample area boundary" are identified as paleooceanic areas. This can be achieved by introducing a "seed point" and a "region growing" algorithm, as follows:

[0124] 1) Define seed points: First, manually set one or a group of "seed points" at the edge of the sample area, making sure the area is located on the ancient ocean side, and set its cell value to 1, while the values ​​of all other cells not selected as seeds are usually set to 0.

[0125] 2) Region Growing: This is an iterative process that starts from the seed point and checks its neighboring cells (using an 8-connected neighborhood rule). If the elevation of a neighboring cell satisfies DEM≤H... palaeo If a cell meets the condition, it is included in the "ancient sea area" set, and its value is assigned to 1. This process is repeated until no new cells can be added. This step identifies all cells connected to the seed point whose elevation satisfies DEM≤H. palaeo Cells that meet the conditions are selected, thus creating a connected region. Cells that do not meet the conditions remain unchanged (with a value of 0).

[0126] 3) Generating the final ancient coastline: The connected regions obtained after region growing represent the true ancient sea area. This is represented by a new raster layer. All cells with a value of 1 in the layer form a continuous, internally connected region, representing the ancient sea area. All cells with a value of 0 represent ancient land or inland depressions not connected to the ancient ocean outside the sample area. Connecting the outer boundaries of all cells with a value of 1 within the sample area yields the preliminary reconstructed ancient coastline.

[0127] By introducing seed points and region growing algorithms, inland depressions that are not connected to the ancient ocean can be excluded, making the reconstructed ancient coastline more geologically plausible in space.

[0128] After performing the above operations in ArcGIS, a binary map will be generated, in which the boundary formed by connecting pixels with a value of 1 is the preliminary restored ancient coastline.

[0129] Shorelines extracted directly from DEMs are often jagged. To obtain a smooth and continuous shoreline that conforms to the natural shape, spatial interpolation is required for the discrete elevation points extracted from the DEM.

[0130] First, extract the elevation value from the DEM that is equal to (or very close to) H. palaeo The coordinates of the center points of all pixels are determined. These points constitute a discrete approximation of the ancient coastline. An inverse distance weighting algorithm is used to interpolate these discrete points, generating a continuous line. The interpolation formula is:

[0131]

[0132] In the formula: H0 is the predicted elevation value (in meters) of any point on the smooth coastline, and the goal is to make H0 infinitely close to H palaeo H i d is the measured elevation value (in meters) of the i-th known point (a discrete point extracted from the DEM); i Let (x0, y0) be the point to be interpolated and (x0, y0) be the i-th known point. i ,y i The Euclidean distance (in meters) between the points is denoted by p. p is the distance decay coefficient, which is dimensionless and usually takes the value of 2. The larger the value of p, the greater the weight of the closer points, and the more the interpolation result will appear as a "bull's eye". The smaller the value of p, the more even the weight distribution and the smoother the result. N is the number of known points participating in the interpolation. When interpolating, a search radius is set, and all known points falling within this radius will participate in the calculation. The total number of these known points is N.

[0133] By using inverse distance weighted interpolation, a smooth and continuous ancient coastline can be obtained. This line better reflects the equilibrium profile formed by ocean waves and tides in the real world, providing a more accurate basis for subsequent paleomorphological identification and hydrodynamic simulation.

[0134] The third step is paleogeographic identification. This step is the core turning point of the entire technical chain. It follows the boundary of the metallogenic system defined in the second step (paleocoastline reconstruction), expanding the research from a one-dimensional "line" to a two-dimensional "surface." The aim is to identify the specific micro-geomorphic units (such as sandbars and coastal dikes) controlling placer enrichment within the paleocoastal zone. Simultaneously, it paves the way for the fourth step (mineral identification), providing precise target areas for spectral analysis, avoiding the inefficiency of "whole-area scanning," and achieving a transition from "morphological analysis" to "material identification." Its fundamental significance lies in establishing the spatial-genetic connection between "paleogeographic environment" and "metallogenesis," providing theoretical target areas for subsequent exploration.

[0135] As a further optional implementation, a paleogeomorphic elevation model is generated based on geological and borehole data and the target paleo-coastline. Multiple paleogeomorphic units are identified based on the paleogeomorphic elevation model using a topographic location index. Then, paleogeomorphic outcrop units are selected from these units based on optical remote sensing data and geological and borehole data. Specifically, this includes:

[0136] S1031. Identify multiple target paleogeomorphic surfaces based on geological and borehole data, and determine the paleosurface elevation points of the target paleogeomorphic surfaces.

[0137] S1032. Spatial interpolation of ancient surface elevation points yields the ancient surface grid surface. The ancient surface grid surface is adjusted according to the target ancient coastline to obtain the ancient geomorphological elevation model.

[0138] S1033. Calculate the topographic location index of multiple micro-geomorphic units based on the paleomorphic elevation model, and divide the micro-geomorphic units into paleomorphic ridge units, paleomorphic slope units, and paleomorphic valley units based on the topographic location index to obtain paleomorphic units.

[0139] S1034. Calculate the elevation changes of each paleogeomorphic unit based on topographic elevation data and paleogeomorphic elevation model, and screen out suspected exposed units based on the elevation changes.

[0140] S1035. Identify whether the suspected outcrop unit has sand or gravel features based on optical remote sensing data, and regard the suspected outcrop unit with sand or gravel features as a high probability outcrop unit.

[0141] S1036. Verify whether the high-probability outcrop unit is an outcrop area or a region with extremely shallow strata based on geological and borehole data, and designate the high-probability outcrop unit that belongs to the outcrop area or region with extremely shallow strata as the paleogeographic outcrop unit.

[0142] The specific process of paleomorphological identification is as follows:

[0143] 1. Paleo-DEM reconstruction, based on simple modeling of the "fundamental laws of stratigraphy," which state that sediments were initially deposited nearly horizontally, with younger sedimentary layers overlying older ones. It is particularly suitable for coastal plains with clearly defined layered sedimentary structures.

[0144] 1) Identify and track paleosurfaces. Identify target paleomorphic surfaces (such as the basement peat layer, paleosol layer, and the bottom plate of marine transgressive sand layer) from the borehole data. Extract the paleosurface elevation points from these boreholes.

[0145] 2) Spatial interpolation. Using these discrete, known paleosurface elevation points as control points, spatial interpolation algorithms (such as kriging or inverse distance weighting) are used to generate a continuous raster surface that reflects the paleosurface—this is the paleoDEM.

[0146] 3) Integrate ancient coastlines. Use the previously restored ancient coastlines as boundary constraints to adjust the interpolated ancient DEM, ensuring that the coastline position matches the 0-meter contour line of the DEM (or the sea level at that time).

[0147] 2. Automatic identification of paleogeographic units based on Topographic Position Index (TPI)

[0148] Based on the generated ancient DEM, the TPI algorithm is applied to identify micro-topographic units. TPI calculates the difference between the elevation of a point and the average elevation of its surrounding area to determine whether the point is located in a relatively convex or concave position. TPI is a powerful tool for identifying positive and negative terrain, and is very suitable for distinguishing ridges, valleys, and other landforms. The specific formula is as follows:

[0149]

[0150] In the formula: is the topographic location index (in meters) at pixel (i,j). Elevation (in meters) of the center pixel. It is the average elevation (in meters) of all pixels within a certain range around the central pixel.

[0151] In practice, the size of the neighborhood (also known as the "kernel") is a key adjustable parameter that determines the scale at which landforms are identified. For identifying large sandbars and seawalls, a large-scale kernel (e.g., 201x201 pixels) should be used to capture the macroscopic pattern. For identifying small beach ridges and gullies, a small-scale kernel (e.g., 21x21 pixels) is needed to reflect microscopic changes. Multiple kernel sizes (e.g., 201, 101, 51, 31, 21, and 11) can be experimented with and compared to find the most suitable size for the landform characteristics of the sample area.

[0152] After calculating the TPI grid, continuous TPI values ​​are classified by setting a threshold (T), thereby classifying the landform units:

[0153] 1) TPI > +T: Defined as a "ridge" unit. Corresponding to positive landforms such as ancient sandbars and coastal dikes, it is the most favorable enrichment area for placer deposits;

[0154] 2) -T≤TPI≤+T: Defined as a "slope" or flat land unit;

[0155] 3) TPI > -T: Defined as a "valley" unit. Corresponds to negative landforms such as ancient river channels and lagoons.

[0156] The final result is a paleogeographic classification map, which clearly marks different geomorphic units.

[0157] 3. Delineation of ancient geomorphic exposure units

[0158] Based on the identified paleogeomorphic units (such as ancient sandbars and coastal dikes), further screening and delineation are conducted on units that have been exposed on the surface or covered by extremely thin sediments due to later geological alteration (such as tectonic uplift and erosion). These units are the most likely parts of the paleogeomorphic bodies to be directly detected by modern remote sensing technology and are the "key evidence bodies" connecting deep prediction and surface exploration.

[0159] 1) Initial screening based on elevation changes quantifies the net sedimentation or erosion in the target area since the mineralization period. Elevation changes are calculated to obtain an elevation change map; the specific formula is: Elevation Change = Modern DEM - Ancient DEM. A threshold is set (e.g., -1 meter < Elevation Change < 1 meter; in tectonically active, strongly eroded bedrock coasts, the threshold can be relaxed, such as -2 meters to +2 meters; in gently sloping coasts dominated by sedimentation, the threshold should be narrowed, such as -0.5 meters to +0.5 meters), to initially delineate "suspected outcrop areas."

[0160] 2) Remote sensing image verification: Overlay high-resolution remote sensing images, and visually interpret or use image classification algorithms to identify "high-probability outcrops" with sand and gravel as the main components within the "suspected outcrop areas".

[0161] 3) Geological data verification: Overlay "high-probability outcrop areas" with geological maps and borehole data. Retain those areas marked as outcrops on geological maps, or areas where boreholes confirm that the target strata are buried at very shallow depths, to generate the final "distribution map of directly detectable units".

[0162] 4) Output Results: In the distribution map of directly detectable units in the sample area, clearly mark the directly exposed areas and the shallowly buried-quasi-exposed areas with eye-catching legends. Attribute information is added to each delineated unit, such as: exposure type (fully exposed / shallowly buried), main surface material (sand, gravel, etc.), area, and confidence level (assessed based on the sufficiency of evidence).

[0163] The fourth step is mineral identification and abundance inversion based on thermal infrared spectroscopy. This step directly converts remote sensing data into mineralogical information. Its core significance lies in breaking through the limitations of traditional geological mapping, achieving large-scale, rapid, and quantitative identification of heavy minerals, and providing direct mineralogical evidence for mineralization prediction. It complements the third step (paleomorphological identification) in terms of both "morphology" and "material": paleomorphological units delineate potential enrichment sites, while this step directly verifies whether these units are rich in target minerals. Its output (mineral distribution map and abundance map) is the most critical input data for the fifth step of comprehensive modeling and target area delineation.

[0164] As a further optional implementation, mineral identification is performed on paleogeographic exposed units based on thermal infrared remote sensing data and reference endmember spectral data to obtain a heavy mineral abundance map of the sample area, which specifically includes:

[0165] S1041. Determine the pixel spectrum of each pixel in the paleogeographic exposed unit based on thermal infrared remote sensing data, calculate the spectral similarity between the pixel spectrum and the reference end-member spectral data, and screen out high-potential areas with the spectral characteristics of the target mineral based on the spectral similarity.

[0166] S1042. Estimate the relative content of heavy minerals in each pixel of the high potential area using the LSU algorithm, and generate a heavy mineral abundance map based on the relative content.

[0167] The specific steps for mineral identification and abundance inversion based on thermal infrared spectroscopy are as follows:

[0168] 1. Laboratory Spectroscopic Measurement and Endmember Establishment

[0169] Heavy mineral samples collected in the field were analyzed in the laboratory using a Fourier transform infrared spectrometer (FTIR) to measure the reflectance spectrum R(λ) of the samples, with a wavelength range covering 8-14 μm.

[0170] According to Kirchhoff's laws, reflectivity can be converted to emissivity, a physical quantity directly detected by thermal infrared remote sensing. The conversion formula is as follows: (λ)=1 R(λ), where (λ) is the emissivity of the sample at wavelength λ (dimensionless, range 0-1).

[0171] High-resolution laboratory spectra and the spectral response function (SRF) of the ASTER TIR band were compared. band(λ) Perform convolution integrals to generate reference endmember spectra that match satellite data. band :

[0172]

[0173] 2. Image Preprocessing

[0174] We used the ASTER L2 Surface Emissivity (LSE) product from NASA's LP DAAC Data Center. This product has been atmospherically corrected using a temperature and emissivity separation algorithm, directly providing emissivity values ​​for each pixel in five TIR bands with an accuracy of approximately ±0.015. Using the corrected LSE product ensures the reliability of subsequent analysis results.

[0175] 3. Spectral Angle Mapping (SAM) Classification: This step aims to quickly screen out "high-potential areas" containing the spectral characteristics of the target mineral from large-area images and provide masks and endmember spectra for the next step of more refined abundance inversion.

[0176] (1) Construction of endmember spectral library: Endmember spectra are the basis for subsequent analysis. An endmember spectral library was constructed by combining image self-extraction and standard spectral library verification.

[0177] On the preprocessed ASTER L1T radiance image, combined with the "directly detectable cell distribution map" delineated in step 3.3, spectrally pure pixels are interactively selected using the Pixel Purity Index (PPI) and n-dimensional visualization tools. The spectra of these pixels are used as initial endmembers.

[0178] The PPI method is a statistical approach that automatically identifies the "purest" pixels from high-dimensional remote sensing imagery. While most pixels in an image are mixtures of different land features, a very small number are relatively pure (e.g., pure sand, pure vegetation, or water). The goal of PPI is to find these "extreme" pixels. In an N-dimensional spectral space (where N is the number of bands), thousands of random unit vectors are generated. These vectors can be viewed as randomly placed "pointers." The spectral data points of each pixel are projected onto these random vectors. For each projection, the pixels falling at the two extremes ("corners") of the vector are recorded. Each time a pixel is marked as an "endpoint," its PPI count increases by 1. After thousands of random projections, each pixel receives a final PPI count value. This value is its "purity index." A higher value indicates that the pixel has been at the edge of the data distribution more often, and its spectrum is purer. The final result is a PPI image, where brighter pixels represent purer pixels. By setting a threshold (such as PPI value > a certain value), these high-purity pixels can be extracted, and their spectral curves are the candidate endmember spectra.

[0179] The extracted endmember spectra were compared and analyzed with standard mineral spectra from the U.S. Geological Survey Spectral Library and the JHU Spectral Library to identify and confirm the corresponding land cover types (such as heavy mineral sand, quartz sand, vegetation, etc.).

[0180] The endmember library used in this invention includes: (1) target endmembers: the spectra of one or more heavy minerals; and (2) background endmembers: the spectra of necessary land features such as quartz sand, vegetation, and water bodies. For heavy minerals and quartz sand, we prepared 2-3 representative spectral variants for each to construct a multi-endmember model.

[0181] (2) SAM Classification: The SAM algorithm is used for pixel-level classification, which can quickly identify pixels containing target minerals, providing a mask for subsequent more refined analysis. The reference endmember spectra (selecting the most representative heavy mineral endmembers in the selected area) in the endmember library constructed in the previous step are matched with the spectra of each pixel in the ASTER LSE product. Both are considered as vectors in n-dimensional space (n is the number of bands; in ASTER TIR, n=5). The spectral similarity is measured by calculating the angle between them. The smaller the angle, the higher the similarity. The calculation formula is:

[0182]

[0183] In the formula: θ is the spectral angle, m i The emissivity value (dimensionless) of the reference endmember in band i is derived from the aforementioned endmember library. i The emissivity value (dimensionless) of the image pixel in band i is available from ASTER LSE products. N is the number of bands (5 in this case).

[0184] Through repeated experiments and verification with known sample points in the sample area, a threshold angle (e.g., θ ≤ 0.1 radians) is set. Pixels below this threshold are judged as "containing heavy minerals," assigned a value of 1, and can be displayed as white; conversely, those above this threshold are assigned a value of 0 and can be displayed as black. The output is a binary mask image used to limit the analysis range of subsequent linear spectral unmixing, ensuring that abundance inversion is only performed within high-potential areas.

[0185] 4. Linear Spectral Unmixing (LSU) Based on Multi-Terminal Component Model

[0186] Within the mask area generated by SAM, the LSU algorithm is applied to estimate the relative content of heavy minerals in each pixel.

[0187] The linear mixture model assumes that the spectral signal of a pixel is a linear combination of the spectra of several land cover classes within it. The multi-endmember model extends this by allowing a land cover class to be represented by multiple spectral endmembers, thus better capturing spectral variations within land cover classes. Its mathematical model is as follows:

[0188]

[0189] In the formula: r k f represents the emissivity value (dimensionless) of a pixel in the k-th band, from an ASTER LSE product.i f is the abundance of the i-th endmember in a pixel (dimensionless, and 0 ≤ f). i ≤1,∑f i =1). This is the target variable to be solved. i,k Let be the emissivity of the i-th endmember in the k-th band. It is derived from the constructed endmember spectral library and is dimensionless. k Let N be the model residual for the k-th band. N is the total number of endmembers.

[0190] Solve a set of f using optimization algorithms such as fully constrained least squares. i The value that makes the model predictions and actual observations r equal. k The difference between (e) k The minimum abundance is calculated by summing the abundance values ​​of multiple endmembers belonging to the same land cover class (such as "heavy minerals") in all pixels. This yields the total abundance of that land cover class, which is then used to create a heavy mineral abundance map.

[0191] The fifth step is target delineation through hydrodynamic-sedimentary coupling simulation and multi-source information fusion. This step is the final integration and output stage of the entire methodology. Its core significance lies in breaking through the traditional "static" feature superposition and constructing a "process-driven" prediction model. It not only answers "where is enriched," but also explains "why is it enriched here" from a dynamic mechanism perspective, significantly improving the scientific validity and reliability of the prediction results. It is logically closely related to the fourth step, where the heavy mineral abundance map output from the fourth step is the key validation data and modeling input for this step. The results of the hydrodynamic simulation need to be compared and calibrated with the actual mineral distribution, and mineral abundance itself is one of the most important predictive factors for machine learning models.

[0192] As a further optional implementation, a hydrodynamic-sedimentary coupling model is constructed based on hydrodynamic environmental parameters and the target paleo-coastline. The model is used to identify sediment accumulation zones in the sample area, and these zones are compared with heavy mineral abundance maps to obtain heavy mineral enrichment zones, which specifically include:

[0193] S1051. Determine the lower boundary topography based on the target paleo-coastline, determine the paleo-wind direction parameters, paleo-wave period and paleo-wave height based on regional paleoclimate research, and use the hydrodynamic model to model the velocity field and wave energy flux density of the sample area to obtain the hydrodynamic-sedimentation coupling model.

[0194] S1052. Calculate the volumetric transport rate of mixed sediments in each sub-region based on the hydrodynamic-sedimentation coupling model, and screen out sediment accumulation zones based on the volumetric transport rate.

[0195] S1053. Compare the sediment accumulation area with the heavy mineral abundance map, and designate the sediment accumulation area with heavy mineral abundance greater than the preset abundance threshold as the heavy mineral enrichment area.

[0196] The specific process of hydrodynamic-sedimentation coupled simulation is as follows:

[0197] (1) Model construction and parameter setting

[0198] The data includes ancient DEMs and boundary conditions. The ancient DEMs are derived from the ancient coastline reconstruction results in the second step and serve as the lower boundary topography of the model. The boundary conditions are based on regional paleoclimate studies, setting parameters such as paleowind direction, paleowave period, and height.

[0199] The model uses hydrodynamic models such as Delft3D, MIKE21, or ROMS to construct two-dimensional or three-dimensional model meshes.

[0200] The simulation output includes the velocity field (U, V) and wave energy flux density (P). The velocity field reflects the hydrodynamic transport capacity; the wave energy flux density directly characterizes the energy of sediment sorting and transport, with units of watts per meter (W / m), and the formula is:

[0201]

[0202] In the formula: ρ is the density of seawater, in kilograms per cubic meter, a physical constant; g is the acceleration due to gravity, in meters per second. 2 H is a physical constant. s The significant wave height, in meters, can be obtained from the simulation results of the hydrodynamic model; C g Wave group velocity is the speed at which wave energy propagates, measured in meters per second. It is the simulation result of the hydrodynamic model and is determined by water depth and wave period.

[0203] (2) Simulation of sediment transport and sorting

[0204] In this invention, ancient coastal sediments are considered as a mixed system composed of heavy mineral particles and light mineral particles such as quartz. The goal of the simulation is not to directly simulate the separation process of different minerals, but to predict the hydrodynamic and geomorphic locations most conducive to the final mineralization of heavy minerals by calculating the transport and deposition of the mixed sediments.

[0205] The volumetric transport rate of mixed sediments was calculated using the Soulsby-Van Rijn et al. integrated transport formula. This formula comprehensively considers sediment transport under the combined effects of water flow and waves, and its basic form is as follows:

[0206]

[0207] In the formula: S totalTotal sediment volume transport rate (TVR) is the core output variable of the model, measured in square meters per second (m² / s), representing the transport flux of sediment per unit width. U represents the depth-average current velocity above the seabed, measured in m / s, derived from simulation outputs of hydrodynamic models (such as Delft3D). rms This represents the near-bottom wave trajectory velocity caused by the wave, in m / s, derived from wave simulation output. (C) d U is the drag coefficient, dimensionless, an empirical parameter related to bed surface roughness, and an input parameter for model setting. cr The critical initiation velocity for sediments, expressed in m / s, is a key parameter in this invention. It is determined by the median grain size and density of the sediment particles. For mixed sediments, this parameter is the fundamental cause of hydraulic differentiation. Heavy mineral placer deposits, due to their much higher density than quartz sand, have a higher U... cr The value is significantly higher. A s This is a comprehensive coefficient, the calculation of which includes properties such as median grain size and density of sediments. It is defined by the formula itself and is an intermediate variable calculated based on sediment properties.

[0208] Heavy mineral placer deposits, due to their higher density, have a critical initiation velocity much higher than that of quartz sand. In simulation calculations, the median grain size of the mixed sediments is used to estimate a representative overall critical initiation velocity. This is based on the hydrodynamic conditions of the simulated region (determined by U0 and U1). rms The weakening of the characterization (results in) causes the current flow rate to be close to but lower than the U of heavy minerals. cr But still higher than quartz U cr At this time, heavy mineral particles will settle first, while quartz sand is more easily transported. Therefore, the model calculates S... total Areas with significantly reduced sediment concentration (i.e., net sediment accumulation areas) are theoretically the high-potential areas most conducive to the selective enrichment of heavy mineral placer deposits.

[0209] The simulated sediment accumulation zone was overlaid and compared with the heavy mineral abundance map obtained in step four. If the high abundance zone closely matched the simulated accumulation zone, the corresponding area was designated as a heavy mineral enrichment zone. This not only verified the accuracy of the simulation but also explained the formation of the enrichment zone from the perspective of sedimentary dynamics.

[0210] As a further optional implementation, the sample area is divided into positive and negative sample areas based on the heavy mineral enrichment zone, and a heavy mineral placer prediction model is trained based on the positive and negative sample areas, which specifically includes:

[0211] S1061. The grid regions located within the heavy mineral enrichment area in the sample region are taken as positive sample regions, and the grid regions located outside the heavy mineral enrichment area in the sample region are taken as negative sample regions.

[0212] S1062. Determine the corresponding spatial location features based on the shortest Euclidean distance from the positive / negative sample area to the target paleo-coastline, determine the corresponding paleo-geomorphic features based on the paleo-geomorphic type of the positive / negative sample area, determine the corresponding heavy mineral features based on the heavy mineral abundance of the positive / negative sample area, and determine the corresponding hydrodynamic features based on the velocity field and wave energy flux density of the positive / negative sample area.

[0213] S1063. Construct corresponding positive / negative training samples based on spatial location features, paleogeographic features, heavy mineral features, and hydrodynamic features. Set the sample label of the positive training samples to "mineralized" and the sample label of the negative training samples to "non-mineralized" to obtain the training dataset.

[0214] S1064. Based on the training dataset, a heavy mineral placer prediction model is obtained by training a random forest, gradient boosting tree, or neural network model.

[0215] Specifically, this embodiment of the invention achieves intelligent quantitative prediction of heavy mineral placer deposits based on multi-source information fusion and machine learning prediction, as detailed below:

[0216] (1) Predictors

[0217] The results of the previous steps are all converted into digital features (raster layers) that can be recognized by the machine learning model, including the following four aspects:

[0218] 1) Spatial location characteristics: Calculate the shortest Euclidean distance from each pixel within the sample area to the ancient coastline reconstructed in the second step;

[0219] 2) Paleomorphological features: Paleomorphological types identified by the TPI algorithm in the third step (e.g., encoding "ridge" units as 1 and others as 0);

[0220] 3) Heavy mineral characteristics: The heavy mineral abundance output by the LSU algorithm in the fourth step is the most direct indicator for mineral exploration.

[0221] 4) Hydrodynamic characteristics: The parameters such as flow velocity and wave energy flux density obtained from the simulation in step 5 are normalized.

[0222] (2) Machine learning model training and prediction

[0223] Algorithms such as Random Forest, Gradient Boosting, or neural network models are employed. These algorithms can handle high-dimensional features and have a strong ability to learn nonlinear relationships between features.

[0224] The sample data are labeled with positive and negative samples. Positive samples are the pixels containing known mineral deposits (from the first step database), labeled as 1, representing the presence of minerals. Negative samples are areas within the sample region that are clearly free of minerals (such as deep-sea areas or bedrock areas) or randomly selected non-mineral deposit pixels, labeled as 0.

[0225] During model training, the aforementioned predictive factor layer is used as input (X), and the sample label (mineralized / non-mineralized) is used as output (Y). The classification model is trained, and the model will autonomously learn the relationship between each factor and mineralization, ultimately obtaining a heavy mineral placer deposit prediction model. The model training process is not the focus of this invention and will not be elaborated here.

[0226] The trained model is used to predict the mineralization probability map of each pixel (value range 0-1).

[0227] This step allows for the quantitative integration of multi-source information, avoiding the subjectivity of manually setting weights. Machine learning models can capture complex mineralization patterns that are difficult for the human brain to intuitively perceive; for example, the combination of "moderate flow velocity and location on the flank of an ancient sandbar" may have the greatest mineralization potential. The generated mineralization probability map is the most direct and objective basis for delineating target areas.

[0228] As a further optional implementation, a mineralization probability distribution map of the target area is predicted based on a heavy mineral placer deposit prediction model, and the heavy mineral target area of ​​the target area is determined based on the mineralization probability distribution map, specifically including:

[0229] S1071. Predict the mineralization probability of each pixel in the target area based on the heavy mineral placer prediction model, and generate a mineralization probability distribution map based on the mineralization probability.

[0230] S1072. Determine the favorable depositional zones in the target area based on hydrodynamic-sedimentation coupled simulation;

[0231] S1073. When the mineralization probability is greater than the preset first threshold and the corresponding pixel is located in a favorable sedimentation area, the corresponding pixel is classified as a first-level heavy mineral target area.

[0232] S1074. When the mineralization probability is greater than or equal to the preset second threshold and less than or equal to the first threshold, and the corresponding pixel is located in a favorable sedimentation area, the corresponding pixel is classified as a secondary heavy mineral target area.

[0233] S1075. When the mineralization probability is greater than or equal to the second threshold and the corresponding pixel is not located in a favorable depositional area, the corresponding pixel is classified as a level 3 heavy mineral target area.

[0234] Specifically, for the target area, its corresponding spatial location features, paleogeomorphological features, heavy mineral features, and hydrodynamic features are obtained through steps similar to those described above. These features are then input into the heavy mineral placer prediction model to obtain the mineralization probability of each pixel in the target area. A mineralization probability distribution map is then generated based on these probabilities. Simultaneously, based on steps similar to those described above, a hydrodynamic-sedimentary coupling simulation is performed on the target area to determine the favorable sedimentary areas (corresponding to sediment accumulation areas). Overlaying the mineralization probability map with the favorable sedimentary areas obtained from the hydrodynamic simulation allows for the delineation and classification of mineralization target areas, which can be categorized into the following three types:

[0235] 1) Primary heavy mineral target areas: mineralization probability > 0.7 and located in favorable sedimentary areas. These target areas are "theoretically supported and shown by geophysical and geochemical exploration," and have the best prospecting prospects.

[0236] 2) Secondary heavy mineral target areas: These target areas have a mineralization probability between 0.4 and 0.7 and are located within favorable sedimentary zones. Further investigation is needed for these target areas.

[0237] 3) Level III target area: Areas with a mineralization probability ≥ 0.4 but located outside the favorable sedimentary area, with only one type of evidence, are designated as exploration areas.

[0238] In some alternative embodiments, cross-validation (such as 10-fold cross-validation) can be used to evaluate the generalization ability of the model. The AUC value is commonly used as an indicator, and the closer the value is to 1, the better the model performance.

[0239] In some optional embodiments, the heavy mineral target area can be field-verified, i.e., shallow drilling or trenching is deployed in the delineated high-priority target area to collect sediment samples for heavy mineral analysis, thereby verifying the reliability of the heavy mineral placer prediction model.

[0240] The method steps of the embodiments of the present invention have been described above. It can be understood that the embodiments of the present invention are based on multi-source data fusion and process mechanism coupling. By integrating remote sensing, paleoenvironmental indicators and hydrodynamic models, the accurate characterization of heavy minerals from transport paths to enrichment sites is achieved, which improves the prediction accuracy of heavy mineral mineralization probability, thereby improving the efficiency and accuracy of heavy mineral placer deposit location exploration.

[0241] Reference Figure 2 This invention provides a heavy mineral placer prediction device based on ancient coastline reconstruction, comprising:

[0242] The data acquisition module is used to acquire optical remote sensing data, thermal infrared remote sensing data, reference end-member spectral data, topographic elevation data, hydrodynamic environmental parameters, and geological and borehole data of the sample area.

[0243] The ancient coastline generation module is used to determine the relative height of the ancient sea level in the sample area, and to generate the target ancient coastline based on the relative height of the ancient sea level and the topographic elevation data through a region growing algorithm.

[0244] The paleogeographic outcrop unit screening module is used to generate a paleogeographic elevation model based on geological and borehole data and the target paleo-coastline. Based on the topographic location index, multiple paleogeographic units are identified according to the paleogeographic elevation model, and paleogeographic outcrop units are screened from the paleogeographic units based on optical remote sensing data and geological and borehole data.

[0245] The heavy mineral abundance identification module is used to identify minerals in paleogeographic exposed units based on thermal infrared remote sensing data and reference end-member spectral data, and to obtain a heavy mineral abundance map of the sample area.

[0246] The heavy mineral enrichment area identification module is used to construct a hydrodynamic-sedimentary coupling model based on hydrodynamic environmental parameters and the target paleo-coastline, identify sediment accumulation areas in the sample area based on the hydrodynamic-sedimentary coupling model, and compare the sediment accumulation areas with the heavy mineral abundance map to obtain the heavy mineral enrichment areas.

[0247] The model training module is used to divide each grid region of the sample area into positive sample regions and negative sample regions according to the heavy mineral enrichment area, and to train a heavy mineral placer prediction model based on the positive sample regions and negative sample regions.

[0248] The heavy mineral target area determination module is used to predict the mineralization probability distribution map of the target area based on the heavy mineral placer prediction model, and to determine the heavy mineral target area of ​​the target area based on the mineralization probability distribution map.

[0249] It is understood that the content of the above method embodiments is applicable to the present device embodiments. The specific functions implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0250] Reference Figure 3 This invention provides an electronic device, comprising:

[0251] At least one processor;

[0252] At least one memory for storing at least one program;

[0253] When the above-mentioned at least one program is executed by the above-mentioned at least one processor, the above-mentioned at least one processor implements the above-mentioned method for predicting heavy mineral placer deposits based on ancient coastline reconstruction.

[0254] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0255] This invention also provides a computer-readable storage medium storing a processor-executable computer program that, when executed by a processor, implements the above-described method for predicting heavy mineral placer deposits based on ancient coastline reconstruction.

[0256] This invention provides a computer-readable storage medium that can execute a method for predicting heavy mineral placer deposits based on ancient coastline reconstruction, as provided in the method embodiments of this invention. It can execute any combination of the implementation steps of the method embodiments and possesses the corresponding functions and beneficial effects of the method.

[0257] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method for predicting heavy mineral placer deposits based on ancient coastline reconstruction.

[0258] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0259] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0260] The embodiments described in this invention are for the purpose of more clearly illustrating the technical solutions of the embodiments of this invention, and do not constitute a limitation on the technical solutions provided by the embodiments of this invention. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this invention are also applicable to similar technical problems.

[0261] The terms "first," "second," "third," "fourth," etc. (if present) in the specification and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0262] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the aforementioned blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this invention are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is changed and sub-operations described as part of a larger operation are executed independently.

[0263] Furthermore, although the invention has been described in the context of functional modules, it should be understood that, unless otherwise stated, one or more of the aforementioned functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the invention. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional skill of an engineer. Therefore, those skilled in the art can implement the invention as set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of the invention, which is determined by the full scope of the appended claims and their equivalents.

[0264] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0265] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0266] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the aforementioned program can be printed, because the aforementioned program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0267] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0268] In the foregoing description of this specification, references to terms such as "one embodiment," "another embodiment," or "some embodiments" indicate that a specific feature, structure, material, or characteristic described in connection with an embodiment or example is included in at least one embodiment or example of the present invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0269] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

[0270] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of the present invention.

Claims

1. A method for predicting heavy mineral placer deposits based on ancient coastline reconstruction, characterized in that, Includes the following steps: Acquire optical remote sensing data, thermal infrared remote sensing data, reference endmember spectral data, topographic elevation data, hydrodynamic environmental parameters, and geological and borehole data for the sample area; Determine the relative height of the paleosea level in the sample area, and generate the target paleoshoreline using a region growing algorithm based on the relative height of the paleosea level and the topographic elevation data. Based on the geological and borehole data and the target ancient coastline, a paleogeographic elevation model is generated. Based on the topographic location index, multiple paleogeographic units are identified according to the paleogeographic elevation model. Based on the optical remote sensing data and the geological and borehole data, paleogeographic exposed units are selected from the paleogeographic units. Mineral identification is performed on the paleogeographic exposed units based on the thermal infrared remote sensing data and the reference end-member spectral data to obtain a heavy mineral abundance map of the sample area. A hydrodynamic-sedimentary coupling model is constructed based on the hydrodynamic environmental parameters and the target paleo-coastline. Sedimentary accumulation zones in the sample area are identified based on the hydrodynamic-sedimentary coupling model, and the sedimentary accumulation zones are compared with the heavy mineral abundance map to obtain heavy mineral enrichment zones. Based on the heavy mineral enrichment area, each grid region of the sample area is divided into positive sample region and negative sample region, and a heavy mineral placer prediction model is trained based on the positive sample region and the negative sample region. The mineralization probability distribution map of the target area is predicted based on the heavy mineral placer prediction model, and the heavy mineral target area of ​​the target area is determined based on the mineralization probability distribution map.

2. The method for predicting heavy mineral placer deposits based on ancient coastline reconstruction according to claim 1, characterized in that, The process of determining the relative height of the paleosea level in the sample area, and generating the target paleoshoreline using a region growing algorithm based on the relative height of the paleosea level and the topographic elevation data, specifically includes: Based on the original geomorphic body of the sample area, the top surface elevation and formation age of the paleosea level geological markers are measured, and the relative height of the paleosea level is calculated based on the top surface elevation and the formation age. or, The global mean sea level height corresponding to the time interval between the target mineralization period and the current time is determined based on the global sea level change curve, and the average vertical tectonic rate of the sample area is obtained. Then, the relative height of the paleosea level is calculated based on the global mean sea level height, the average vertical tectonic rate, and the time interval. A digital elevation model of the sample area is constructed based on the topographic elevation data. Seed pixels with an elevation less than or equal to the relative height of the ancient sea level are determined based on the digital elevation model and added to the ancient sea area set. Starting from the seed pixel, determine whether the altitude of the neighboring pixels of the seed pixel is less than or equal to the relative altitude of the ancient sea level. If so, add the neighboring pixel as a new seed pixel to the ancient sea area set. The ancient sea area grid region is determined based on the ancient sea area set, and the target ancient coastline is determined based on the boundary line of the ancient sea area grid region.

3. The method for predicting heavy mineral placer deposits based on ancient coastline reconstruction according to claim 1, characterized in that, The process of generating a paleogeographic elevation model based on the geological and borehole data and the target paleo-coastline, identifying multiple paleogeographic units based on the topographic location index and the paleogeographic elevation model, and selecting paleogeographic outcrop units from the paleogeographic units based on the optical remote sensing data and the geological and borehole data, specifically includes: Based on the geological and borehole data, multiple target paleogeographic surfaces were identified, and the paleosurface elevation points of the target paleogeographic surfaces were determined. Spatial interpolation is performed on the paleosurface elevation points to obtain the paleosurface grid surface. The paleosurface grid surface is then adjusted according to the target paleo-coastline to obtain the paleo-geomorphic elevation model. The topographic location index of multiple micro-geomorphic units is calculated based on the paleomorphic elevation model, and the micro-geomorphic units are divided into paleomorphic ridge units, paleomorphic slope units, and paleomorphic valley units based on the topographic location index, thus obtaining the paleomorphic units. The elevation changes of each paleogeographic unit are calculated based on the topographic elevation data and the paleogeographic elevation model, and suspected exposed units are screened out based on the elevation changes. Based on the optical remote sensing data, identify whether the suspected outcrop unit has sand or gravel features, and designate the suspected outcrop unit with sand or gravel features as a high-probability outcrop unit. Based on the geological and borehole data, it is verified whether the high-probability outcrop unit is an outcrop area or a region with extremely shallow strata burial depth, and the high-probability outcrop unit belonging to the outcrop area or region with extremely shallow strata burial depth is taken as the paleogeographic outcrop unit.

4. The method for predicting heavy mineral placer deposits based on ancient coastline reconstruction according to claim 1, characterized in that, The step of identifying minerals in the paleogeographic exposed units based on the thermal infrared remote sensing data and the reference endmember spectral data to obtain a heavy mineral abundance map of the sample area specifically includes: The pixel spectrum of each pixel in the paleogeographic exposed unit is determined based on the thermal infrared remote sensing data, and the spectral similarity between the pixel spectrum and the reference end-member spectral data is calculated. Based on the spectral similarity, high-potential areas with the spectral characteristics of the target mineral are screened out. The relative abundance of heavy minerals in each pixel of the high-potential area is estimated using the LSU algorithm, and the heavy mineral abundance map is generated based on the relative abundance.

5. The method for predicting heavy mineral placer deposits based on ancient coastline reconstruction according to claim 1, characterized in that, The process involves constructing a hydrodynamic-sedimentary coupling model based on the hydrodynamic environmental parameters and the target paleo-coastline, identifying sediment accumulation zones in the sample area using the hydrodynamic-sedimentary coupling model, and comparing these sediment accumulation zones with the heavy mineral abundance map to obtain heavy mineral enrichment zones. Specifically, this includes: The lower boundary topography is determined based on the target paleo-coastline. Paleo-wind parameters, paleo-wave periods, and paleo-wave heights are determined based on regional paleoclimate studies. The flow velocity field and wave energy flux density of the sample area are modeled using a hydrodynamic model to obtain the hydrodynamic-sedimentation coupling model. The volumetric transport rate of the mixed sediments in each sub-region is calculated based on the hydrodynamic-sedimentation coupling model, and the sediment accumulation zone is selected based on the volumetric transport rate. The sediment accumulation area is compared with the heavy mineral abundance map, and the sediment accumulation area with heavy mineral abundance greater than a preset abundance threshold is designated as the heavy mineral enrichment area.

6. The method for predicting heavy mineral placer deposits based on ancient coastline reconstruction according to claim 1, characterized in that, The step of dividing the sample region into positive and negative sample regions based on the heavy mineral enrichment area, and training a heavy mineral placer prediction model based on the positive and negative sample regions, specifically includes: The grid regions within the heavy mineral enrichment area in the sample region are designated as the positive sample regions, and the grid regions outside the heavy mineral enrichment area in the sample region are designated as the negative sample regions. The spatial location features are determined based on the shortest Euclidean distance from the positive / negative sample regions to the target paleo-coastline; the paleo-geomorphic features are determined based on the paleo-geomorphic type of the positive / negative sample regions; the heavy mineral features are determined based on the heavy mineral abundance of the positive / negative sample regions; and the hydrodynamic features are determined based on the velocity field and wave energy flux density of the positive / negative sample regions. Based on the spatial location features, paleogeographic features, heavy mineral features, and hydrodynamic features, corresponding positive / negative training samples are constructed, and the sample labels of positive training samples are set to mineral-containing, while the sample labels of negative training samples are set to non-mineral-containing, thus obtaining the training dataset. The heavy mineral placer prediction model is obtained by training a random forest, gradient boosting tree, or neural network model based on the training dataset.

7. A method for predicting heavy mineral placer deposits based on ancient coastline reconstruction according to any one of claims 1 to 6, characterized in that, The step of obtaining the mineralization probability distribution map of the target area based on the heavy mineral placer prediction model, and determining the heavy mineral target area of ​​the target area based on the mineralization probability distribution map, specifically includes: The mineralization probability of each pixel in the target area is predicted based on the heavy mineral placer prediction model, and the mineralization probability distribution map is generated based on the mineralization probability. The favorable depositional zones in the target area were determined based on hydrodynamic-sedimentation coupled simulation. When the mineralization probability is greater than a preset first threshold and the corresponding pixel is located in the favorable deposition area, the corresponding pixel is classified as a first-level heavy mineral target area. When the mineralization probability is greater than or equal to a preset second threshold and less than or equal to the first threshold, and the corresponding pixel is located in the favorable deposition area, the corresponding pixel is divided into a secondary heavy mineral target area. When the mineralization probability is greater than or equal to the second threshold and the corresponding pixel is not located in the favorable deposition area, the corresponding pixel is classified as a level three heavy mineral target area.

8. A heavy mineral placer deposit prediction device based on ancient coastline reconstruction, characterized in that, include: The data acquisition module is used to acquire optical remote sensing data, thermal infrared remote sensing data, reference end-member spectral data, topographic elevation data, hydrodynamic environmental parameters, and geological and borehole data of the sample area. The ancient coastline generation module is used to determine the ancient sea level relative height of the sample area, and generate the target ancient coastline based on the ancient sea level relative height and the topographic elevation data using a region growing algorithm. The paleogeographic outcrop unit screening module is used to generate a paleogeographic elevation model based on the geological and borehole data and the target paleo-coastline, identify multiple paleogeographic units based on the topographic location index and the paleogeographic elevation model, and screen out paleogeographic outcrop units from the paleogeographic units based on the optical remote sensing data and the geological and borehole data. The heavy mineral abundance identification module is used to identify minerals in the paleogeographic exposed units based on the thermal infrared remote sensing data and the reference end-member spectral data, and to obtain a heavy mineral abundance map of the sample area. The heavy mineral enrichment area determination module is used to construct a hydrodynamic-sedimentary coupling model based on the hydrodynamic environmental parameters and the target paleo-coastline, identify sediment accumulation areas in the sample area based on the hydrodynamic-sedimentary coupling model, and compare the sediment accumulation areas with the heavy mineral abundance map to obtain heavy mineral enrichment areas. The model training module is used to divide each grid region of the sample area into positive sample regions and negative sample regions according to the heavy mineral enrichment area, and to train a heavy mineral placer prediction model based on the positive sample regions and the negative sample regions. The heavy mineral target area determination module is used to predict the mineralization probability distribution map of the target area based on the heavy mineral placer prediction model, and to determine the heavy mineral target area of ​​the target area based on the mineralization probability distribution map.

9. An electronic device, characterized in that, include: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements a method for predicting heavy mineral placer deposits based on ancient coastline reconstruction as described in any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements a method for predicting heavy mineral placer deposits based on ancient coastline reconstruction as described in any one of claims 1 to 7.

Images