Method, device and medium for constructing integrated water and land digital elevation model

Abstract

The application discloses a method and device for constructing a water-land integrated digital elevation model and a medium. The method comprises the following steps: determining a first digital elevation model of a land area based on an application scenario identifier; determining a spatially aligned grid image set and an average water level value based on a shoal remote sensing image and a water level observation data set, and determining a water body type of each pixel according to a preset identification frequency threshold; extracting shoal three-dimensional water level contour data by combining a preset area threshold, the water body type and the average water level value, and generating a second digital elevation model of the shoal by spatial interpolation; determining a deposition state according to historical underwater topographic data of a deep water area and a real-time water depth of a check point, and constructing a third digital elevation model of the deep water area based on the deposition state; and generating a water-land integrated digital elevation model of a target area based on the first, second and third digital elevation models and topographic control point data. The application can improve the efficiency and precision of constructing a water-land integrated digital elevation model.

Description

Technical Field

[0001] This application relates to the field of remote sensing imagery and geographic information processing technology, specifically to a method, device, and medium for constructing an integrated land and water digital elevation model. Background Technology

[0002] Against the backdrop of the digital and intelligent transformation of water conservancy projects, the construction of digital twin watersheds is a crucial foundation for supporting the intelligent advancement of reservoir safety management and water resource optimization scheduling. Digital elevation models (DEMs), as the core foundation of watershed spatial data, play an irreplaceable role in watershed topographic analysis, hydrological process simulation, and water-related engineering planning. Currently, relying on a three-dimensional monitoring system encompassing air, space, ground, water, and engineering, a watershed topographic information acquisition system has been established, primarily based on aerospace remote sensing (such as publicly available DEM products like SRTM and AW3D30), supplemented by UAV oblique photography and lidar. However, when it comes to the high-precision construction of integrated land-water digital twin watersheds, existing technologies still face significant technical bottlenecks and efficiency limitations.

[0003] Firstly, in terms of land topography construction, conventional airborne lidar data processing typically employs automated filtering and global interpolation algorithms to build a DEM. However, this standardized approach often performs poorly when dealing with hydraulic engineering areas featuring significant artificial characteristics. Due to the lack of constraints on specific terrain structures, the automatic interpolation process is prone to over-smoothing key hydraulic structures with steep ridges or abrupt changes, such as dams, levees, and canals. This results in the final land model losing crucial terrain framework features, failing to meet the stringent structural accuracy requirements of hydraulic engineering design.

[0004] Secondly, acquiring underwater topographic information, especially monitoring shallow and nearshore areas, faces challenges in terms of operational efficiency and implementation. Current mainstream DEM products primarily cover land areas and cannot directly acquire underwater topographic data. Traditional underwater surveying methods (such as shipborne sonar) are limited by vessel draft and navigation safety, making effective operations difficult in shallow waters and mudflats where land and water meet. Even with small unmanned vessels or manual wading surveying, the coverage is extremely limited, and the process is time-consuming, labor-intensive, and inefficient. This reliance on high-cost, long-cycle manual or semi-automated surveying methods cannot meet the urgent need for rapid, comprehensive updates to large-scale underwater topography in digital twin watersheds.

[0005] Finally, in the comprehensive processing of multi-source data, land and underwater topographic data from different platforms (satellites, drones, unmanned vessels, etc.) and different time phases are often based on different vertical benchmarks and coordinate systems, and there are inherent systematic biases between different acquisition devices. Existing integrated stitching methods lack a systematic unification of elevation benchmarks and a rigorous overall adjustment mechanism, and only fuse data through simple boundary stitching. This makes the synthesized model prone to elevation breaks, logical inconsistencies, "step" effects, or ghosting phenomena at the land-water interface, severely damaging the natural continuity and overall accuracy of the topographic surface, and making it difficult to construct a complete and coordinated digital base for the entire watershed. Summary of the Invention

[0006] The purpose of this application is to provide a method, device, and medium for constructing an integrated land and water digital elevation model, so as to solve the problem of low efficiency in constructing integrated land and water digital elevation models in the prior art.

[0007] To achieve the above objectives, the first aspect of this application provides a method for constructing an integrated land and water digital elevation model, the method comprising:

[0008] The first digital elevation model of the land area in the target region is determined based on the application scenario identifier; Based on multiple remote sensing images and water level observation datasets of the shallow water area in the target region, determine multiple sets of spatially aligned raster images and the average water level value corresponding to each set of raster images; The water type at each pixel coordinate of each raster image set is determined based on a preset recognition frequency threshold. The three-dimensional water level contour data of the shoal area are determined based on the preset area threshold, water body type and average water level value; A second digital elevation model of the shoal area is generated based on three-dimensional water level contour data and spatial interpolation methods; The siltation status of the deep water area is determined based on historical underwater topographic data of the deep water area in the target region and real-time water depth data of multiple verification points. A third digital elevation model for determining deep water areas based on siltation conditions; A land-water integrated digital elevation model of the target area is generated based on the first digital elevation model, the second digital elevation model, the third digital elevation model, and the terrain control point data of the target area.

[0009] In this embodiment of the application, the step of determining the first digital elevation model of the land area in the target region based on the application scenario identifier includes: when the application scenario identifier is a high-precision terrain modeling scenario, determining the ground point cloud data based on terrain control point data, the original three-dimensional point cloud data of the land area, and the point cloud ground filtering algorithm; fusing the three-dimensional terrain profile lines of the land area into the ground point cloud data to obtain fused data; and performing spatial gridding calculation on the fused data through a spatial interpolation method to obtain the first digital elevation model.

[0010] In this embodiment of the application, the method further includes: when the application scenario is identified as a large-scale macroscopic analysis scenario, acquiring an aerospace remote sensing digital elevation model covering the target area; and determining a first digital elevation model based on the aerospace remote sensing digital elevation model.

[0011] In this embodiment of the application, the method further includes: when the application scenario is identified as a local supplementary survey and a special terrain scenario, acquiring high-precision discrete terrain points of the land area; the high-precision discrete terrain points are collected along the terrain feature lines and terrain changes of the land area based on real-time dynamic carrier phase differential technology; and constructing a first digital elevation model based on the high-precision discrete terrain points and spatial interpolation methods.

[0012] In this embodiment, the water level observation dataset stores water levels corresponding one-to-one with each remote sensing image. The steps of determining multiple spatially aligned raster image sets and the average water level value corresponding to each raster image set based on multiple remote sensing images of the shallow water area in the target region and the water level observation dataset include: determining multiple target water level groups based on the water levels corresponding one-to-one with each remote sensing image; constructing a water body identification feature layer corresponding to any remote sensing image based on the spectral features or backscattering intensity of that remote sensing image; performing binarization processing on the water body identification feature layer corresponding to each remote sensing image using an adaptive threshold segmentation algorithm to output a water body binary map of each remote sensing image; dividing the water body binary map of each remote sensing image into the corresponding target water level group according to the water level corresponding to each remote sensing image; spatially aligning all water body binary maps within each target water level group to obtain multiple raster image sets; and determining the average water level value corresponding to each raster image set.

[0013] In this embodiment of the application, the step of determining the water body type at each pixel coordinate of each group of raster images according to a preset recognition frequency threshold includes: determining the pixel value of each raster image in each group of raster images at each pixel coordinate; summing the pixel values ​​of each raster image in any group of raster images at the same pixel coordinate to obtain the total pixel value of the group of raster images at that pixel coordinate; calculating the ratio between the total pixel value and the total number of raster images in the group of raster images as the water body recognition frequency of the group of raster images at that pixel coordinate; and determining the water body type of the group of raster images at that pixel coordinate based on the water body recognition frequency and the preset recognition frequency threshold.

[0014] In this embodiment of the application, the step of determining the three-dimensional water level contour data of the shoal area based on a preset area threshold, water body type, and average water level value includes: generating a stable water body range raster corresponding to any set of raster images at each pixel coordinate based on the water body type of the set of raster images; performing morphological processing on the stable water body range raster using a preset area threshold, and performing boundary smoothing and vectorization processing on the morphologically processed raster to obtain the water surface boundary line corresponding to the set of raster images; assigning the average water level value corresponding to each set of raster images as an elevation attribute to the water surface boundary line corresponding to the set of raster images to obtain the three-dimensional water level contour data of the shoal area.

[0015] In this embodiment of the application, the step of determining the siltation state of the deep water area based on historical underwater topographic data and real-time water depth data of multiple verification points in the target area includes: acquiring historical underwater topographic data and real-time water depth data of multiple verification points; extracting historical elevation data of each verification point from the historical underwater topographic data based on the planar coordinates of each verification point; the historical underwater topographic data and the planar coordinate system of each verification point are the same; determining the elevation difference between the real-time water depth data of any verification point and the historical elevation data of that verification point; determining the arithmetic mean of the corresponding verification points based on the elevation difference of each verification point; determining the siltation state of the deep water area as slight siltation when the arithmetic mean is less than a preset change threshold; and determining the siltation state of the deep water area as significant siltation when the arithmetic mean is greater than or equal to the preset change threshold.

[0016] In this embodiment of the application, the step of determining the third digital elevation model of deep water based on the siltation state includes: generating a first underwater elevation discrete point based on historical underwater topographic data when the siltation state is slight; and generating a third digital elevation model based on the spatial interpolation method and the first underwater elevation discrete point.

[0017] In this embodiment of the application, the method further includes: acquiring high-density underwater topographic point cloud data of deep water when the siltation state is significant; performing sound velocity profile correction and tidal level correction on the high-density underwater topographic point cloud data, and removing abnormal noise points to obtain a second underwater elevation discrete point; and reconstructing the second underwater elevation discrete point into a grid using a spatial interpolation method to obtain a third digital elevation model.

[0018] In this embodiment of the application, the steps of generating an integrated land and water digital elevation model of the target area based on a first digital elevation model, a second digital elevation model, a third digital elevation model, and terrain control point data include: converting the plane coordinate systems of the first, second, and third digital elevation models into a unified coordinate system to obtain multiple digital elevation models with a unified coordinate system; matching the spatial position of each digital elevation model with the terrain control point data to calculate the elevation system deviation of the corresponding digital elevation model; correcting the elevation values ​​of each digital elevation model using a regional network adjustment method with the goal of eliminating the elevation system deviation of each digital elevation model to obtain corrected digital elevation models; and generating an integrated land and water digital elevation model of the target area based on the corrected digital elevation models.

[0019] A second aspect of this application provides a computer device, comprising: The memory is configured to store instructions; and The processor is configured to retrieve instructions from memory and to implement the methods described above when executing instructions.

[0020] A third aspect of this application provides a machine-readable storage medium storing instructions that cause a machine to perform the methods described above.

[0021] The above technical solutions first introduce a land model construction mechanism based on application scenario identification, enabling the system to adaptively match the optimal data acquisition strategy according to actual needs, avoiding resource waste and time delays caused by blindly adopting high-cost measurement methods in low-precision scenarios. Second, for the shallow "blind spots" that are difficult to access with traditional measurement methods, the spatiotemporal correlation between multi-temporal remote sensing images and water level data is utilized to generate three-dimensional water level contour lines through statistical frequency analysis. This fills the data gaps in the water-land transition zone in a low-cost and automated manner without the need for inefficient on-site wading measurements. More importantly, a "diagnosis first, decision later" mechanism based on checkpoints is established for deep water areas. By comparing real-time data from a small number of checkpoints with historical data, the siltation status is accurately determined. This allows for the direct reuse of historical data when the terrain is stable (slight siltation), and high-density field measurements are only triggered when the terrain changes significantly. This "condition-triggered" operational logic fundamentally avoids redundant full-coverage scanning of stable water areas. Finally, the above-mentioned sub-domain models are fused under a unified benchmark by combining terrain control points. In summary, through the synergistic optimization of "land scene adaptation", "shallow water remote sensing inversion" and "deep water on-demand mapping", the unnecessary field measurement workload is reduced to the greatest extent while ensuring the continuity and accuracy of the topographic representation of the entire region, thereby significantly improving the overall efficiency of constructing an integrated land and water digital elevation model.

[0022] Other features and advantages of the embodiments of this application will be described in detail in the following detailed description section. Attached Figure Description

[0023] The accompanying drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the following detailed description to explain the embodiments of this application, but do not constitute a limitation on the embodiments of this application. In the drawings: Figure 1 The flowchart illustrates a method for constructing an integrated land and water digital elevation model according to an embodiment of this application. Figure 2 This diagram schematically illustrates the spatial distribution and composition of an integrated land and water digital elevation model according to an embodiment of this application. Figure 3 A flowchart illustrating a method for determining a water body type according to an embodiment of this application is shown schematically. Figure 4 This illustration schematically shows the effect of generating three-dimensional water level contour data of a shoal area according to an embodiment of this application; Figure 5 The diagram illustrates a structural diagram of a computer device according to an embodiment of this application. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0025] It should be noted that if the embodiments of this application involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0026] Furthermore, if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0027] The acquisition, transmission, storage, use, and processing of data in this application comply with relevant laws and regulations. Furthermore, it should be noted that certain software, components, models, and other existing industry solutions may be mentioned in the embodiments of this application. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0028] It should be noted that all data involved in this application (including but not limited to data used for analysis, data stored, data displayed, etc.) are information and data that have been fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0029] Figure 1 A flowchart illustrating a method for constructing an integrated land and water digital elevation model according to an embodiment of this application is shown. Figure 1 As shown in the figure, this application provides a method for constructing an integrated land and water digital elevation model, which may include the following steps.

[0030] Step 101: Determine the first digital elevation model of the land area based on the terrain control point data of the target area, the original three-dimensional point cloud data of the land area, and the three-dimensional terrain profile.

[0031] In this embodiment, the target area refers to the geographic area to be mapped, encompassing a reservoir or lake and its surrounding land. Topographic control point data refers to spatial reference point data with high-precision three-dimensional coordinate information, serving as an absolute spatial benchmark for position calibration and accuracy control of other multi-source data. Land area refers to the land surface area within the target area excluding the water body coverage. Original three-dimensional point cloud data refers to a dense point cloud collection containing the three-dimensional spatial coordinate information of all land features within the target area, acquired through remote sensing equipment such as lidar. This data typically contains noise data from non-ground points such as vegetation and buildings. Three-dimensional topographic profile lines refer to three-dimensional vector line segments describing the outlines of hydraulic structures such as dams and levees, or key topographic features such as topographic faults. The first digital elevation model refers to a grid model or vector model specifically generated based on the processed land area data, used for digitally representing the undulating morphology of the land surface.

[0032] First, precise coordinate registration is performed on the original 3D point cloud data of the land area using terrain control point data. Then, a point cloud ground filtering algorithm is applied to iteratively process the registered data, effectively removing non-ground points such as vegetation and buildings, thereby extracting ground point cloud data that truly reflects the landform. Next, 3D terrain profiles that accurately describe terrain abrupt changes and structural features are fused into the ground point cloud data as feature constraints for terrain construction. Finally, a spline function interpolation algorithm with obstacles is used to perform spatial gridding calculations on the fused data, generating a continuous and smooth first digital elevation model.

[0033] By introducing high-precision terrain control point data to establish a spatial benchmark, using filtering algorithms to effectively filter out interference from land cover, and combining three-dimensional terrain profile lines to enforce constraints on key terrain features, this step significantly improves the ability and accuracy of the land terrain model to reproduce the real land surface. This effectively solves the problem of terrain distortion caused by vegetation obstruction or feature loss, laying a solid land data foundation for ultimately improving the accuracy of constructing an integrated land and water digital elevation model.

[0034] Step 102: Based on multiple remote sensing images and water level observation datasets of the shallow water area in the target region, determine multiple sets of spatially aligned raster images and the average water level value corresponding to each set of raster images.

[0035] In this embodiment, the shoal area refers to a water-land transition zone located between the perennial water level and the highest flood level, exhibiting periodic exposure or submersion characteristics due to water level fluctuations; its terrain is gentle and varied. Remote sensing imagery refers to multi-temporal surface observation data acquired at different time points by remote sensing platforms such as satellites and aircraft, recording surface spectral characteristics or radar backscattering intensity, encompassing various data sources such as optical images or synthetic aperture radar images. The water level observation dataset refers to a collection of data containing precise water level elevation values ​​at various time points, recorded by on-site monitoring at hydrological stations or water level gauges; this dataset corresponds to the remote sensing imagery in the time dimension. Multiple raster image sets refer to several image clusters formed by classifying and summarizing the extracted binary water body maps according to water level ranges. Each set contains multiple rasterized images of the corresponding water level within the same specific range, and these images have been geometrically corrected to achieve precise overlap in geographic space. The average water level value refers to the representative value obtained by statistically calculating the measured water level corresponding to all individual images contained in each raster image set, and is used to characterize the macroscopic hydraulic state reflected by the image set.

[0036] First, a spatiotemporal correlation is established between remote sensing images and water level observation datasets, matching the corresponding measured water level to each remote sensing image based on its imaging time. Then, a feature layer is constructed based on spectral features or backscattering intensity, and an adaptive threshold segmentation algorithm is used to accurately extract the water body extent from each remote sensing image, generating a binary water body map. According to preset water level intervals, these binary water body maps are divided into different target water level groups, and strict spatial registration and alignment are performed on the images within each group, thus constructing a multi-group hierarchical raster image set. Simultaneously, the arithmetic mean of the water levels corresponding to all images within each group is calculated and determined as the average water level value of that group of raster image sets.

[0037] By accurately matching and statistically grouping long-term remote sensing imagery with measured water level data, a direct mapping relationship was effectively established between planar image features and vertical water level elevation. This statistical classification processing based on multi-temporal data effectively mitigates random errors caused by environmental noise, wave disturbance, or differences in image quality in single-period images, ensuring the stability and representativeness of the shallow water body extraction. This provides highly reliable data support for subsequent accurate inversion of shallow water topography, thereby significantly improving the accuracy of constructing an integrated land-water digital elevation model.

[0038] Step 103: Determine the water body type at each pixel coordinate of each raster image set according to the preset recognition frequency threshold.

[0039] In this embodiment, the preset identification frequency threshold refers to a pre-set statistical limit value used to determine whether a spatial location belongs to the stable water body coverage area under specific water level conditions. It acts as a probability filter to distinguish between real water bodies and accidental noise interference. Pixel coordinates refer to the row and column indices or spatial coordinate parameters used to uniquely identify the planar position of each smallest data unit in a rasterized data matrix. Water body type refers to the classification definition of specific pixel location attributes based on time-series statistical results, mainly used to distinguish whether the location belongs to a stable area continuously covered by water or a non-water body area affected by environmental disturbances.

[0040] First, the pixel values ​​at the same pixel coordinates are extracted from each raster image in the group. These values ​​are then summed to obtain the total number of times that location was identified as a water body in multi-temporal observations. Next, the ratio of this total number of times to the total number of images in the group is calculated to determine the water body identification frequency at that pixel coordinate. Finally, the calculated water body identification frequency is compared with a preset identification frequency threshold. If the frequency is higher than the threshold, the water body type at that pixel coordinate is confirmed as a stable water body; otherwise, it is marked as a non-water body, thus completing the probabilistic screening of the water body range.

[0041] By employing a frequency-based statistical thresholding strategy, this step effectively isolates false water body signals caused by cloud cover, wave spray, temporary water accumulation, or image noise from complex multi-temporal observation data, accurately identifying the actual water area coverage with high confidence at a specific average water level. This statistically-based denoising process minimizes the interference of random errors on topographic inversion, ensuring that the subsequently extracted water level lines accurately reflect topographic contour features, thereby significantly improving the accuracy of constructing an integrated land-water digital elevation model.

[0042] Step 104: Determine the three-dimensional water level contour data of the shoal area based on the preset area threshold, water body type and average water level value.

[0043] In this embodiment, the preset area threshold refers to a minimum area limit set in advance to eliminate interference from minor noise in the image and ensure the continuity and integrity of the extracted water body morphology. In morphological operations, it serves as a filtering condition to remove isolated patches smaller than this area or fill internal voids smaller than this area. Three-dimensional water level contour data refers to a set of vector curves with precise elevation attributes. It depicts the geometric contour of the water-land boundary under specific water level conditions in three-dimensional geographic space, essentially transforming dynamic boundary lines at different water levels into topographically significant contour data.

[0044] First, based on the distribution of water body types at each pixel coordinate, a raster of stable water body ranges under the given water level conditions is identified and aggregated. Then, a preset area threshold is applied to perform morphological optimization on the raster image. This filters out small, fragmented patches lacking hydrogeographic significance and fills in small holes within the water body area caused by observational gaps or identification errors, thereby reconstructing a water body mask with clear edges and internal connectivity. Next, the optimized raster boundaries are smoothed and vectorized to extract continuous water surface boundary lines. The average water level value corresponding to this set of raster images is then assigned to these boundary lines as elevation dimension information, generating water level contour data with three-dimensional spatial coordinates.

[0045] By introducing area threshold constraints for refined morphological processing, the problems of fragmented water body boundaries and internal cavities caused by image noise or surface complexity were effectively corrected, ensuring that the extracted water level lines could realistically and continuously reflect the natural contour features of the shoal topography. Simultaneously, the statistically derived average water level values ​​were accurately mapped to spatial elevation attributes, achieving a crucial transformation from two-dimensional image boundaries to a three-dimensional terrain framework. This ensured the accuracy and reliability of the basic data for constructing the shoal topography, thereby significantly improving the accuracy of building an integrated land-water digital elevation model.

[0046] Step 105: Generate a second digital elevation model of the shoal area based on the three-dimensional water level contour data and spatial interpolation method.

[0047] In this embodiment, a second digital elevation model of the shoal area is generated based on three-dimensional water level contour data and spatial interpolation methods. The spatial interpolation method refers to a mathematical or statistical method that estimates unknown point data using known point data. In this step, it is specifically used to transform discrete three-dimensional water level contour data into a continuous raster surface. Algorithms such as the irregular triangular network (TIN) model, inverse distance weighting, Kriging interpolation, or spline function method can be used to ensure a smooth spatial transition in the generated model. In this embodiment, a second digital elevation model of the shoal area can be generated based on the TIN (Triangulated Irregular Network) model and three-dimensional water level contour data. The TIN model is a spatial interpolation and modeling technique that connects discrete spatial points or terrain feature lines into a series of non-overlapping adjacent triangular patches to construct a continuous surface. It can effectively maintain the fault lines and skeleton features of the terrain through constraints and is often used to simulate complex irregular terrain surfaces. The second digital elevation model refers to a digital model of shoal terrain generated based on water level inversion data, specifically used to fill data gaps between land and deep water areas. Three-dimensional water level contour data is used as rigid constraint lines or terrain skeletons for constructing triangulation networks (TINs). An irregular triangulation model is applied to rigorously triangulate the spatial areas covered by these contour lines. During network construction, the algorithm mandates that the edges of generated triangles must not cross the water level contour lines, which serve as constraints, but rather connect along the contour lines. Then, by rasterizing or linearly interpolating the generated triangulation network, a continuously changing shallow water terrain surface is constructed. Through the constraint capability of the TIN model on feature lines, discrete water level contour lines are transformed into a topologically rigorous continuous terrain surface, avoiding the step-like errors that may occur with conventional interpolation methods. This effectively solves the data loss problem caused by the ineffectiveness of conventional measurement methods in shallow water areas, achieving high-fidelity simulation of complex shallow water terrain and significantly improving the accuracy of constructing an integrated land-water digital elevation model. The construction of shallow water topography using high-density, high-precision three-dimensional water level contour data obtained by remote sensing inversion effectively fills the blind spots of traditional measurement methods in shallow water areas. It eliminates the need to rely on costly and inefficient shipborne sonar or manual measurement, greatly improving the efficiency of acquiring shallow water area topographic data, and thus enhancing the overall efficiency and completeness of constructing an integrated land and water digital elevation model.

[0048] Step 106: Determine the siltation status of the deep water area based on historical underwater topographic data of the deep water area in the target region and real-time water depth data of multiple verification points.

[0049] In this embodiment, the siltation state of the deep water area is determined based on historical underwater topographic data and real-time water depth data from multiple verification points within the target area. Historical underwater topographic data refers to as-built topographic maps, past surveying data, or historical digital elevation models acquired and archived in the deep water area over previous periods, serving as a benchmark for assessing current topographic evolution. Verification points are discrete detection locations selected within the deep water area based on spatial representativeness principles, used to infer the overall topographic change trend through local sampling data. Real-time water depth data refers to the current water depth or bottom elevation values ​​collected on-site at each verification point using depth sounding equipment, reflecting the current actual geomorphological characteristics of the deep water area. Siltation state refers to the level of deep water topographic change determined based on the degree of difference between the real-time measurements at the verification points and historical benchmark values. This state characterizes the significance of bottom sediment deposition or erosion and serves as a basis for subsequent topographic construction strategy selection. This step enables the rapid diagnosis of terrain changes in deep water areas using a small amount of discrete verification data. It avoids the need for blindly conducting full-coverage, high-cost unmanned surface surveys when the terrain is stable or undergoing minimal changes. This allows for the adaptive selection of the optimal data acquisition strategy based on the actual terrain conditions, significantly reducing the workload and cost of field measurements while ensuring data accuracy. It also effectively improves the overall efficiency of constructing an integrated land and water digital elevation model.

[0050] Step 107: Determine the third digital elevation model of deep water based on the siltation state.

[0051] In this embodiment, based on the degree of deep-water topographic evolution determined in the preceding steps, a differentiated topographic construction path is dynamically matched. This means that existing data is reused when topographic changes are insignificant, while high-precision field measurements are employed when topographic changes are significant, achieving a balance between cost and accuracy. The third digital elevation model refers to a digital surface model that reflects the underwater topographic morphology of deep water areas. It is generated from processed underwater elevation data through gridded reconstruction and is a key component of the final integrated land-water model. This step avoids the inefficient traditional approach of applying full-coverage field measurements to all deep-water areas. Instead, it flexibly allocates data resources based on actual siltation conditions. While ensuring the model accurately reflects the current topography, it significantly reduces unnecessary field surveying workload and data processing time, thereby significantly improving the overall efficiency of constructing the integrated land-water digital elevation model.

[0052] Step 108: Generate an integrated land and water digital elevation model of the target area based on the first digital elevation model, the second digital elevation model, the third digital elevation model, and the terrain control point data.

[0053] In this application embodiment, the integrated land-water digital elevation model refers to a comprehensive digital model generated by seamlessly stitching and fusing geospatial data of land, shallow water, and deep water areas, capable of continuously and uniformly representing the surface and underwater topographic relief of the entire target area. A unified coordinate system refers to a unique and standardized national standard plane coordinate system selected to eliminate spatial reference differences between multi-source data. Elevation system deviation refers to the systematic elevation numerical differences at the same geographical location caused by differences in acquisition equipment, processing algorithms, or vertical datum between digital elevation models from different sources. The regional network adjustment method is a mathematical adjustment technique based on the least squares principle, using control points to jointly solve multiple models to eliminate geometric deformation and systematic errors.

[0054] First, a spatial datum normalization operation is performed, transforming the plane coordinate systems of the first, second, and third digital elevation models to a unified coordinate system, achieving precise alignment of the data in the horizontal dimension. Then, under the unified plane datum, each digital elevation model is spatially matched with high-precision topographic control point data, accurately calculating the elevation system deviation of each model relative to the absolute datum. Next, with the optimization objective of eliminating these system deviations, a regional network adjustment method is applied to calculate correction parameters and perform overall correction on the elevation values ​​of each model, unifying all three to the same vertical datum. Finally, based on the unified and corrected plane and elevation datum data, the data is integrated and reconstructed to generate the final integrated land and water digital elevation model for the target area.

[0055] By rigorously unifying the plane coordinates and correcting the elevation benchmarks of the multi-source heterogeneous data before fusion, this step effectively eliminates systematic errors and misalignments caused by different measurement methods, ensuring seamless spatial connection and logical consistency of land, shallow water and deep water topography. It fundamentally solves the common problems of cracks, steps or ghosting in traditional splicing methods, thereby significantly improving the accuracy of constructing an integrated land and water digital elevation model.

[0056] The above technical solutions first introduce a land model construction mechanism based on application scenario identification, enabling the system to adaptively match the optimal data acquisition strategy according to actual needs, avoiding resource waste and time delays caused by blindly adopting high-cost measurement methods in low-precision scenarios. Second, for the shallow "blind spots" that are difficult to access with traditional measurement methods, the spatiotemporal correlation between multi-temporal remote sensing images and water level data is utilized to generate three-dimensional water level contour lines through statistical frequency analysis. This fills the data gaps in the water-land transition zone in a low-cost and automated manner without the need for inefficient on-site wading measurements. More importantly, a "diagnosis first, decision later" mechanism based on checkpoints is established for deep water areas. By comparing real-time data from a small number of checkpoints with historical data, the siltation status is accurately determined. This allows for the direct reuse of historical data when the terrain is stable (slight siltation), and high-density field measurements are only triggered when the terrain changes significantly. This "condition-triggered" operational logic fundamentally avoids redundant full-coverage scanning of stable water areas. Finally, the above-mentioned sub-domain models are fused under a unified benchmark by combining terrain control points. In summary, through the synergistic optimization of "land scene adaptation", "shallow water remote sensing inversion" and "deep water on-demand mapping", the unnecessary field measurement workload is reduced to the greatest extent while ensuring the continuity and accuracy of the topographic representation of the entire region, thereby significantly improving the overall efficiency of constructing an integrated land and water digital elevation model.

[0057] In this embodiment of the application, the step of determining the first digital elevation model of the land area in the target region based on the application scenario identifier includes: when the application scenario identifier is a high-precision terrain modeling scenario, determining the ground point cloud data based on terrain control point data, the original three-dimensional point cloud data of the land area, and the point cloud ground filtering algorithm; fusing the three-dimensional terrain profile lines of the land area into the ground point cloud data to obtain fused data; and performing spatial gridding calculation on the fused data through a spatial interpolation method to obtain the first digital elevation model.

[0058] In this embodiment, the application scenario identifier refers to a logical marker used in the system to distinguish different engineering task requirements or data accuracy levels. It serves as the basis for triggering different data processing paths, ensuring the on-demand allocation of computing resources and data acquisition costs. The original 3D point cloud data refers to an initial discrete set of points containing massive spatial 3D coordinates and intensity information, obtained through airborne lidar scanning or oblique photogrammetry technology. It completely records the spatial morphology of surface vegetation, buildings, and exposed surfaces. The point cloud ground filtering algorithm refers to a computer processing logic used to automatically identify and separate non-ground points (such as vegetation, vehicles, and buildings) from ground points in a mixed original point cloud. For example, progressive densification triangulation filtering or morphological filtering aims to extract ground point cloud data that only represents the undulations of the real surface. The 3D terrain profile line refers to a 3D vector line used to describe abrupt changes in terrain structure or skeletal features, such as ridge lines, valley lines, roadbed edges, or embankment edges. It contains accurate elevation information and participates in modeling as a mandatory constraint condition. The first digital elevation model refers to a digital terrain surface model that specifically covers the land portion of the target area. Firstly, by utilizing application scenario identifiers, rapid routing of modeling strategies was achieved, avoiding high-cost data processing in unnecessary scenarios. More importantly, when high-precision modeling was required, a combination of automated filtering and feature line constraints was adopted. This approach not only leveraged the efficiency advantage of algorithms in rapidly processing massive amounts of data but also effectively solved the problem of automatic filtering mistakenly deleting or smoothing key terrain features such as steep slopes and dam crests by fusing terrain profiles. This avoided the need for extensive and time-consuming manual refinement of the model in the future. While ensuring the accuracy of key structural features of the land model, the efficiency of constructing the first digital elevation model was significantly improved, thereby enhancing the efficiency of constructing the overall integrated land and water model.

[0059] In this embodiment of the application, the method further includes: when the application scenario is identified as a large-scale macroscopic analysis scenario, acquiring an aerospace remote sensing digital elevation model covering the target area; and determining a first digital elevation model based on the aerospace remote sensing digital elevation model.

[0060] In this embodiment, the large-scale macroscopic analysis scenario refers to an operational state where the spatial resolution requirements for terrain data are relatively relaxed, but the requirements for coverage and acquisition timeliness are high, such as watershed master planning, flood inundation trend assessment, or large-scale site selection analysis. The aerospace remote sensing digital elevation model refers to a standardized wide-area terrain elevation data product acquired by satellite-borne synthetic aperture radar or optical sensors and high-altitude aircraft, such as the Space Shuttle Radar Topographic Mapping Mission (SRTM) data, the global 30-meter resolution digital surface model AW3D30, or the Copernicus Digital Elevation Model. This step utilizes existing, widely covered, and easily accessible mature remote sensing data products to directly construct land terrain, avoiding the need for blindly conducting high-cost, long-cycle UAV aerial surveys or manual field measurements in scenarios requiring only macroscopic terrain trend analysis. This achieves on-demand matching of data accuracy and acquisition cost, greatly simplifies the data acquisition process, and significantly improves the efficiency of constructing the first digital elevation model.

[0061] In this embodiment of the application, the method further includes: when the application scenario is identified as a local supplementary survey and a special terrain scenario, acquiring high-precision discrete terrain points of the land area; the high-precision discrete terrain points are collected along the terrain feature lines and terrain changes of the land area based on real-time dynamic carrier phase differential technology; and constructing a first digital elevation model based on the high-precision discrete terrain points and spatial interpolation methods.

[0062] In this application embodiment, local supplementary surveys and special terrain scenarios refer to specific engineering environments where aerial photogrammetry cannot be effectively carried out due to severe obstruction by tall vegetation or limited airspace, or where only rapid resurveying of small flat areas is required. These represent blind spots or uneconomical work areas that are difficult to cover by conventional large-area automated operation methods. High-precision discrete terrain points refer to a set of three-dimensional spatial coordinate points with centimeter-level or higher positioning accuracy directly obtained through field measurement methods, which serve as the basic sampling data for reconstructing local surface morphology. Real-time dynamic carrier phase differential technology refers to a high-precision satellite positioning measurement method that uses carrier phase observations between a base station and a rover to perform real-time differential calculations, such as RTK technology, which can provide the precise three-dimensional position of the measurement point in a specified coordinate system in real time during field operations. Terrain feature lines refer to linear elements that cause changes in surface morphology or control the trend of terrain, such as ridge lines, valley lines, slope change lines, or the outlines of artificial structures. Terrain change points refer to key node locations where the surface relief or slope aspect changes significantly. Through this step, this application abandons the high-cost, long-cycle aerospace remote sensing full-coverage scanning method for complex obscured areas or scattered small areas, and instead adopts a flexible and mobile field point-marking method. By utilizing the high precision and real-time performance of RTK technology, it collects a small amount of precise data only for key feature locations that control the terrain morphology. While ensuring the accuracy of local terrain representation, it greatly reduces the workload of collecting and processing redundant data, avoids resource waste, and significantly improves the efficiency and flexibility of building the first digital elevation model in special scenarios.

[0063] In this embodiment, the point cloud ground filtering algorithm can employ a progressive triangulation densification point cloud ground filtering algorithm. This refers to a point cloud classification technique based on an iterative strategy of constructing irregular triangulation networks. It selects local low points as initial seed points to build a basic network and gradually densifies it. It uses angle and distance parameters during the iteration process to determine and remove non-ground noise points such as vegetation and buildings. Fusion data refers to a mixed dataset formed by superimposing and integrating vector profile lines describing the rigid structure of the terrain with discretely distributed ground point cloud data under a unified spatial reference. This dataset possesses both high-density point location information and clear terrain feature lines. Spline function interpolation algorithm is a numerical approximation method that uses piecewise polynomial functions to fit a smooth, continuous surface between discrete data points. It is often used to generate mathematical models that conform to natural surface morphological changes. Spatial gridding computation refers to the process of resampling spatially non-uniformly distributed discrete point data into regularly arranged raster matrix data through interpolation operations. First, the original point cloud is corrected using topographic control point data. A filtering algorithm is then applied to remove surface cover, resulting in a clean ground point cloud. Next, 3D topographic profile lines are fused into the point cloud as topographic fault features. Spline interpolation is then used to perform regularization operations on this fused data, which includes feature constraints, to generate the first digital elevation model. By employing a high-precision filtering algorithm to remove vegetation interference, combining topographic profile lines to impose strict constraints on key topographic structures, and using spline interpolation to maintain the smoothness and continuity of the topographic surface, the true undulating morphology of the land surface is effectively restored. This avoids elevation distortion caused by vegetation shading or missing feature lines, thus significantly improving the accuracy of constructing an integrated land-water digital elevation model.

[0064] In one embodiment, firstly, for application scenarios such as engineering design that have high requirements for terrain accuracy and timeliness, the original three-dimensional point cloud data of the land area collected by the lidar sensor carried by the UAV is acquired, and high-precision positioning is performed by combining the terrain control point data of the target area measured on the ground.

[0065] Secondly, the Progressive TIN Densification (PTD) point cloud ground filtering algorithm is applied to process the original three-dimensional point cloud data of the land area. By iteratively constructing a triangular network and determining the positional relationship between points and the network, non-ground points such as surface buildings and vegetation are removed, thereby determining the ground point cloud data.

[0066] Furthermore, in order to address the issue that the automatic filtering process may mistakenly remove point clouds of some hydraulic structures (such as dams and canals) that should be retained, the three-dimensional topographic profiles of the land area described by manual collection of key topographic features are fused into the ground point cloud data as a mandatory constraint on the topographic features, thus obtaining fused data.

[0067] Finally, the fused data is spatially gridded using a spline function interpolation algorithm with obstacles. Based on the principle of minimum curvature, this algorithm can generate a smooth digital elevation surface that conforms to the discontinuous characteristics of actual terrain, even under conditions of abrupt terrain changes or the presence of "obstacles" such as man-made structures. This results in a first digital elevation model that ensures the accuracy and integrity of the engineering terrain representation.

[0068] In this embodiment, the water level observation dataset stores water levels corresponding one-to-one with each remote sensing image. The steps of determining multiple spatially aligned raster image sets and the average water level value corresponding to each raster image set based on multiple remote sensing images of the shallow water area in the target region and the water level observation dataset include: determining multiple target water level groups based on the water levels corresponding one-to-one with each remote sensing image; constructing a water body identification feature layer corresponding to any remote sensing image based on the spectral features or backscattering intensity of that remote sensing image; performing binarization processing on the water body identification feature layer corresponding to each remote sensing image using an adaptive threshold segmentation algorithm to output a water body binary map of each remote sensing image; dividing the water body binary map of each remote sensing image into the corresponding target water level group according to the water level corresponding to each remote sensing image; spatially aligning all water body binary maps within each target water level group to obtain multiple raster image sets; and determining the average water level value corresponding to each raster image set.

[0069] In this embodiment, the target water level group refers to several consecutive water level intervals pre-defined based on the range of observed water level variations, used to summarize discretely distributed remote sensing images over time into a statistically significant set. The water body identification feature layer refers to intermediate process data specifically designed to enhance the contrast between water targets and background features, generated based on spectral band differences or radar backscattering intensity differences in the original remote sensing images, such as normalized difference water index images or radar intensity filtered images. The adaptive threshold segmentation algorithm is an image processing algorithm that can dynamically calculate the optimal segmentation threshold based on the image's histogram features or local statistical characteristics, automatically adapting to changes in lighting conditions or environmental noise in images from different time phases. The water body binary map refers to a binary raster image output after segmentation, whose pixel attributes only contain two logical states: water body and non-water body. The average water level value is the numerical value obtained by arithmetically averaging the measured water levels corresponding to all remote sensing images belonging to the same target water level group; this value serves as the unified elevation attribute of this group of raster images in the vertical direction.

[0070] First, a spatiotemporal mapping relationship between images and water levels is established. Enhanced recognition layers are generated based on spectral or scattering features. An adaptive algorithm is used to extract the water body extent and generate binary maps. Then, these binary maps are grouped into corresponding water level groups based on water level elevation, and spatial correction and alignment are performed on the data within each group. Finally, the average water level of each group is calculated. This water level-driven grouping strategy, adaptive feature extraction technology, and averaging of water levels within groups effectively solve the recognition instability problem caused by differences in imaging conditions in multi-source images. It also provides accurate vertical elevation references for each image set, ensuring the robustness of the water body extraction results and the accuracy of elevation attributes. This provides a high-quality data foundation for subsequent accurate inversion of shallow water topography, thus significantly improving the accuracy of constructing an integrated land-water digital elevation model.

[0071] In this embodiment of the application, the step of determining the water body type at each pixel coordinate of each group of raster images according to a preset recognition frequency threshold includes: determining the pixel value of each raster image in each group of raster images at each pixel coordinate; summing the pixel values ​​of each raster image in any group of raster images at the same pixel coordinate to obtain the total pixel value of the group of raster images at that pixel coordinate; calculating the ratio between the total pixel value and the total number of raster images in the group of raster images as the water body recognition frequency of the group of raster images at that pixel coordinate; and determining the water body type of the group of raster images at that pixel coordinate based on the water body recognition frequency and the preset recognition frequency threshold.

[0072] In this embodiment, pixel value refers to a numerical identifier in a binary raster image used to characterize whether a specific location is a water body. Typically, a specific integer value (e.g., 1 represents water, 0 represents non-water) is used to distinguish the target from the background. Total pixel value refers to the statistical result obtained by summing the pixel values ​​of all temporal data within the image set for the same spatial coordinates over time. Its physical meaning represents the cumulative frequency of water body characteristics at that location in the observation sequence. Water body identification frequency is a probabilistic index characterizing whether a specific location belongs to a water body coverage area under the current water level group conditions, obtained by normalizing the total pixel value. The pixel values ​​of each image within the group at each coordinate point are read one by one, and the total pixel value of each point is calculated through spatial superposition. Subsequently, the water body identification frequency is obtained by calculating the ratio of the total pixel value to the total number of images, and this is combined with a threshold to determine whether the point is a stable water body area. By using pixel-level statistics and frequency analysis based on multi-temporal big data, this step effectively separates random classification noise from stable water body signals, greatly reducing the impact of single-period image recognition errors on the overall results, ensuring high reliability of shallow water-land boundary extraction, and thus significantly improving the accuracy of constructing an integrated water-land digital elevation model.

[0073] In this embodiment of the application, the step of determining the three-dimensional water level contour data of the shoal area based on a preset area threshold, water body type, and average water level value includes: generating a stable water body range raster corresponding to any set of raster images at each pixel coordinate based on the water body type of the set of raster images; performing morphological processing on the stable water body range raster using a preset area threshold, and performing boundary smoothing and vectorization processing on the morphologically processed raster to obtain the water surface boundary line corresponding to the set of raster images; assigning the average water level value corresponding to each set of raster images as an elevation attribute to the water surface boundary line corresponding to the set of raster images to obtain the three-dimensional water level contour data of the shoal area.

[0074] In this embodiment, the stable water body range raster refers to a binary image matrix generated based on pixel-level water body type determination results, which intuitively represents the continuous water body coverage morphology under specific water level conditions. The preset area threshold is a minimum connected region area parameter pre-set to filter out broken noise patches in the image and fill in small voids within the water body; it serves as a quantitative constraint for morphological operations. Morphological processing refers to the technical means of optimizing and correcting the topological structure of the raster image using mathematical morphological algorithms such as erosion, dilation, opening, or closing operations, aiming to smooth edges and maintain regional connectivity. Boundary smoothing and vectorization processing refers to the data processing process of converting jagged discrete raster edges into mathematically continuous and smooth geometric curves and converting image coordinates into geospatial vector coordinates. The water surface boundary line refers to the two-dimensional vector line extracted after the above processing, capable of accurately depicting the water-land interface. Three-dimensional water level contour data refers to the set of spatial curves used to construct the terrain skeleton in three-dimensional space after assigning absolute vertical elevation attributes to the planar water surface boundary. First, stable water body raster ranges are extracted based on water body type. Morphological operations are then performed using a preset area threshold to remove noise and fill holes. Next, the processed raster boundaries are vectorized and smoothed to generate continuous water surface boundary lines. Finally, the corresponding average water level value is assigned as the elevation attribute to this boundary line. This data processing workflow, combining morphological optimization and vectorization techniques, effectively eliminates quantization errors and fragmented noise in raster data, transforming discrete image classification results into continuous, smooth, and clearly defined three-dimensional terrain feature lines. This ensures the geometric accuracy and realism of the shallow water terrain skeleton, significantly improving the accuracy and efficiency of constructing an integrated land-water digital elevation model.

[0075] In one embodiment, the data acquisition and preprocessing are first performed. Specifically, the Global Surface Water Data Set (GSWD) is used as the remote sensing image data source based on the Google Earth Engine platform. Long-term monthly synthetic water body distribution data (generated from Landsat satellite imagery) of the target area (e.g., from 1984 to 2021) is extracted and combined with water level observation datasets obtained from hydrological stations. Remote sensing images with similar measured water levels are divided into the same target water level group. Subsequently, all water body distribution data within each target water level group are spatially aligned to obtain multiple raster image sets, and the average water level value corresponding to each raster image set is calculated.

[0076] Secondly, the frequency of water body identification and type determination are performed. The frequency of water bodies identified is counted pixel-by-pixel within each group. This involves determining the pixel value at each pixel coordinate of each raster image in each group and summing the values. The ratio of the total pixel value to the total number of images is then calculated to obtain the water body identification frequency. This frequency is compared with a preset identification frequency threshold (e.g., 60%), and pixel coordinates with frequencies higher than this threshold are identified as stable water body types.

[0077] Next, three-dimensional water level contour data is extracted. A stable water body range raster is generated based on the determined stable water body type. Morphological processing is performed on this raster using a preset area threshold to fill internal holes and remove small patches within the water body range. The processed raster is then smoothed and vectorized to extract the water surface boundary lines corresponding to different water levels. The average water level value corresponding to each raster image set is assigned as an elevation attribute to the corresponding water surface boundary line, treating it as equivalent three-dimensional water level contour data for the shallow water area.

[0078] Finally, a second digital elevation model is generated. Based on the obtained series of three-dimensional water level contour data, an irregular triangular mesh model is used for terrain reconstruction to generate a second digital elevation model covering the shallow water area (e.g., areas with a water depth of approximately 7m or less). Validation results show that the model generated by this method has a high correlation coefficient with the depth sounder's measured data, and can meet the accuracy requirements for shallow water terrain reconstruction.

[0079] In this embodiment of the application, the step of determining the siltation state of the deep water area based on historical underwater topographic data and real-time water depth data of multiple verification points in the target area includes: acquiring historical underwater topographic data and real-time water depth data of multiple verification points; extracting historical elevation data of each verification point from the historical underwater topographic data based on the planar coordinates of each verification point; the historical underwater topographic data and the planar coordinate system of each verification point are the same; determining the elevation difference between the real-time water depth data of any verification point and the historical elevation data of that verification point; determining the arithmetic mean of the corresponding verification points based on the elevation difference of each verification point; determining the siltation state of the deep water area as slight siltation when the arithmetic mean is less than a preset change threshold; and determining the siltation state of the deep water area as significant siltation when the arithmetic mean is greater than or equal to the preset change threshold.

[0080] In this embodiment, historical elevation data refers to the underwater elevation values ​​at historical moments corresponding to the same geographical coordinates, retrieved from past surveying results based on the plane location index of the verification points; elevation difference refers to the vertical deviation between real-time measured values ​​and historical records, used to quantify the magnitude of topographic changes at a single sampling location; the arithmetic mean refers to the average index obtained by statistically calculating the elevation differences of all verification points, reflecting the average siltation or scouring trend of deep water at the overall level, effectively eliminating random errors in single-point measurements; the preset change threshold refers to the numerical limit set according to engineering accuracy requirements or historical siltation patterns, used to distinguish whether the topography has undergone substantial changes; slight siltation specifically refers to the calculated statistical index not exceeding the limit, indicating that the timeliness of historical data meets current application needs; significant siltation specifically refers to the statistical index exceeding the limit, indicating that the historical data is invalid and cannot accurately reflect the current topography. This step establishes a rapid diagnostic mechanism for overall terrain changes based on sparse sample statistical inference. It uses low-cost single-point verification data to replace expensive full-coverage pre-scanning, accurately identifying areas that do need to be re-surveyed. This avoids invalid repeated measurements of deep water areas that are in a stable state, thereby maximizing the optimization of surveying and mapping resource allocation while ensuring data reliability, and significantly improving the efficiency of constructing deep water areas and overall integrated land-water digital elevation models.

[0081] In this embodiment of the application, the step of determining the third digital elevation model of deep water based on the siltation state includes: generating a first underwater elevation discrete point based on historical underwater topographic data when the siltation state is slight; and generating a third digital elevation model based on the spatial interpolation method and the first underwater elevation discrete point.

[0082] In this embodiment, the first underwater elevation discrete point refers to a set of three-dimensional coordinate points directly parsed from historical underwater topographic data through data processing methods such as digital extraction, resampling, or format conversion, under the premise that the topography has not changed significantly. This set is considered to represent the current underwater surface morphology and serves as the direct input source for subsequent mathematical modeling. Through this step, after verifying that the underwater topography is in a stable state, existing surveying results are directly reused to replace entirely new field measurements. This transforms the originally time-consuming marine or surface surveying task into a rapid data processing task, thereby ensuring data validity while completely avoiding the time and economic costs of repeated measurements. This greatly improves the efficiency of constructing deep-water topographic models and overall integrated land-sea digital elevation models.

[0083] In this embodiment of the application, the method further includes: acquiring high-density underwater topographic point cloud data of deep water when the siltation state is significant; performing sound velocity profile correction and tidal level correction on the high-density underwater topographic point cloud data, and removing abnormal noise points to obtain a second underwater elevation discrete point; and reconstructing the second underwater elevation discrete point into a grid using a spatial interpolation method to obtain a third digital elevation model.

[0084] In this embodiment, the spatial interpolation method can be Kriging interpolation. Significant siltation refers to a physical change in the actual topography of the deep water area that exceeds a preset tolerance, as determined by previous steps, indicating that the historical data can no longer meet current accuracy requirements. High-density underwater topographic point cloud data refers to a massive set of three-dimensional coordinate points acquired by active scanning equipment such as multibeam echo sounders mounted on unmanned surface vessels, capable of comprehensively covering the seabed and reflecting its subtle undulations. Sound velocity profile correction refers to the data processing process of correcting the echo time or path of sonar measurements based on the changing patterns of sound wave propagation speed at different water depths to eliminate sound ray bending errors. Tide level correction refers to uniformly converting the measured water depth based on the instantaneous water surface to a preset vertical reference plane (such as the theoretical maximum). The calculation process of the low tide level; abnormal noise points refer to non-topographic pseudo-data generated by bubbles, fish schools, suspended matter, or multipath effects in the water; the second underwater elevation discrete points refer to the high-precision net water depth point set that truly represents the current underwater morphology after the above rigorous error correction and denoising processing; Kriging interpolation is a linear unbiased optimal estimation method based on variogram theory and structural analysis. It considers not only the known point values ​​but also the spatial positional relationship and correlation between points, and can restore the natural undulation characteristics of the underwater topography to the greatest extent; grid reconstruction specifically refers to the process of using this algorithm to convert irregularly distributed discrete points into a regularly arranged raster matrix. This step activates a high-precision, full-process measurement and automated processing mechanism when significant changes in the terrain are confirmed. It utilizes sound velocity and tidal level correction techniques to eliminate systematic errors and employs Kriging interpolation to effectively overcome the smoothing effect of a single interpolation method. This ensures that the current underwater terrain details can be accurately reproduced in the newly generated model, thereby avoiding engineering design deviations and subsequent rework caused by using outdated historical data. While ensuring the timeliness and accuracy of the data, it also guarantees the overall efficiency of building a high-quality integrated digital elevation model for land and water.

[0085] In this embodiment, historical underwater topographic data refers to raw mapping data collected in the deep waters of the target area using a single-beam or multi-beam sonar detection system, recording underwater reflection signals and initial depth values. Underwater elevation discrete points refer to a set of discontinuous, valid measurement points with precise three-dimensional spatial coordinates and elevation attributes, obtained after preprocessing steps such as sound velocity profile correction, tidal level observation correction, and outlier removal from the raw underwater topographic data. First, the collected raw underwater topographic data undergoes rigorous sound velocity and tidal level correction and denoising, transforming it into spatially accurate underwater elevation discrete points. Then, a semi-variogram model is constructed based on these discrete points, and Kriging interpolation is applied to weighted estimate and gridded reconstruction of the blank areas between discrete points, thereby generating a third digital elevation model. By employing the statistically advantageous Kriging interpolation method to process sparse or unevenly distributed underwater measurement points, the problem of local abrupt changes or smooth transitions that are easily generated by traditional geometric interpolation methods in the construction of complex underwater terrain is effectively overcome. It can maximize the use of spatial correlation of data to restore the natural continuity of underwater terrain, thereby significantly improving the accuracy of constructing an integrated land and water digital elevation model.

[0086] In this embodiment of the application, the steps of generating an integrated land and water digital elevation model of the target area based on a first digital elevation model, a second digital elevation model, a third digital elevation model, and terrain control point data include: converting the plane coordinate systems of the first, second, and third digital elevation models into a unified coordinate system to obtain multiple digital elevation models with a unified coordinate system; matching the spatial position of each digital elevation model with the terrain control point data to calculate the elevation system deviation of the corresponding digital elevation model; correcting the elevation values ​​of each digital elevation model using a regional network adjustment method with the goal of eliminating the elevation system deviation of each digital elevation model to obtain corrected digital elevation models; and generating an integrated land and water digital elevation model of the target area based on the corrected digital elevation models.

[0087] In this embodiment, the planar coordinate system refers to the local or independent two-dimensional projection reference system upon which each sub-domain digital elevation model is based during the initial construction or acquisition phase, and these systems may differ from each other. Spatial location matching refers to the computational steps of establishing a one-to-one geometric relationship between raster nodes in the digital elevation model and measured terrain control points through search and comparison algorithms within a unified geographic coordinate framework. The corrected digital elevation models refer to high-quality terrain data prepared for final fusion, which, after planar coordinate normalization and elevation system deviation correction, achieves benchmark unification in three-dimensional space and eliminates systematic errors. First, coordinate transformation is performed to unify the original planar coordinate systems of the three models to the national standard coordinate system. Next, the unified models are spatially matched with terrain control points to accurately calculate elevation system deviations. Subsequently, regional network adjustment methods are used to eliminate these deviations, generating corrected digital elevation models, and based on these, the final integrated land and water model is constructed. By employing this rigorous processing logic of "unifying the plane first and then correcting the elevation," spatial benchmark differences and systematic misalignments between multi-source data are eliminated, ensuring geometric continuity and logical consistency of different terrain regions at the fusion boundary. This fundamentally avoids the generation of splicing cracks or elevation steps, thereby significantly improving the accuracy of constructing an integrated land and water digital elevation model.

[0088] In one embodiment, firstly, a plane coordinate unification process is performed, converting the plane coordinate systems of the first digital elevation model of the land area, the second digital elevation model of the shallow water area, and the third digital elevation model of the deep water area into a unified coordinate system (e.g., the CGCS2000 coordinate system), resulting in multiple digital elevation models after the unified coordinate system.

[0089] Secondly, the elevation benchmark is calibrated. Each digital elevation model after unifying the coordinate system is spatially matched with high-precision terrain control point data, and the elevation system deviation of the corresponding digital elevation model is calculated. With the goal of eliminating the elevation system deviation of each digital elevation model, the elevation value of each digital elevation model is calibrated as a whole through the regional network adjustment method to obtain the calibrated digital elevation models, thereby ensuring the continuity and consistency of the terrain surface in the subsequent fusion process.

[0090] Finally, based on the corrected digital elevation models, feature constraint and reconstruction techniques are used to generate the final model. Specifically, vector contour lines are extracted from the corrected first, second, and third digital elevation models (DEMs) at a preset contour interval (e.g., 0.5m). Simultaneously, terrain fault lines are determined. The water surface line, acquired concurrently with the data from the first DEM, is defined as the terrain fault boundary between the first and second DEMs. The outer boundary of the measured data corresponding to the third DEM is defined as the fault segment connecting the second and third DEMs. Subsequently, terrain buffer zones are established along the terrain fault lines. For data gaps or anomalies within the buffer zones, spatial interpolation is performed using reliable surrounding terrain points to supplement key elevation points. Finally, by integrating all vector contour lines, terrain fault lines, and the densified terrain points, a constrained irregular triangular network (UTRI) is constructed. Linear interpolation is then performed based on this UTRI to generate a spatially continuous, terrain-coordinated, and preset grid resolution (e.g., 0.5m) integrated land-water digital elevation model of the target area.

[0091] The above technical solutions first introduce a land model construction mechanism based on application scenario identification, enabling the system to adaptively match the optimal data acquisition strategy according to actual needs, avoiding resource waste and time delays caused by blindly adopting high-cost measurement methods in low-precision scenarios. Second, for the shallow "blind spots" that are difficult to access with traditional measurement methods, the spatiotemporal correlation between multi-temporal remote sensing images and water level data is utilized to generate three-dimensional water level contour lines through statistical frequency analysis. This fills the data gaps in the water-land transition zone in a low-cost and automated manner without the need for inefficient on-site wading measurements. More importantly, a "diagnosis first, decision later" mechanism based on checkpoints is established for deep water areas. By comparing real-time data from a small number of checkpoints with historical data, the siltation status is accurately determined. This allows for the direct reuse of historical data when the terrain is stable (slight siltation), and high-density field measurements are only triggered when the terrain changes significantly. This "condition-triggered" operational logic fundamentally avoids redundant full-coverage scanning of stable water areas. Finally, the above-mentioned sub-domain models are fused under a unified benchmark by combining terrain control points. In summary, through the synergistic optimization of "land scene adaptation", "shallow water remote sensing inversion" and "deep water on-demand mapping", the unnecessary field measurement workload is reduced to the greatest extent while ensuring the continuity and accuracy of the topographic representation of the entire region, thereby significantly improving the overall efficiency of constructing an integrated land and water digital elevation model.

[0092] Figure 2 The illustration schematically shows the spatial distribution and composition of an integrated land and water digital elevation model according to an embodiment of this application. For example... Figure 2As shown in this embodiment, the target area is precisely divided into three continuous geographical units: the land area (red), the shoal area (yellow-green), and the deep water area (purple). The land area displays a refined first digital elevation model constructed through interpolation based on lidar point clouds, topographic control points, and three-dimensional topographic profiles, clearly presenting the topographic framework of the land surface and hydraulic structures. The shoal area displays a second digital elevation model extracted from long-term remote sensing imagery and water level data, utilizing water surface boundary lines at different water levels to form dense three-dimensional water level contour lines with clear elevation attributes, effectively filling blind spots in conventional measurements and achieving smooth terrain transitions. The deep water area displays a third digital elevation model generated based on sonar detection data and Kriging interpolation. After unified coordinate system transformation, elevation system deviation correction based on control points, and regional network adjustment, these three sub-models achieve seamless spatial splicing and logical fusion, thus forming a continuous and high-precision integrated land-water digital elevation model covering the entire area.

[0093] Figure 3 A flowchart illustrating a method for determining water body type according to an embodiment of this application is shown schematically. Figure 3 As shown in this embodiment, the multiple layers stacked on the left represent binary water bodies (i.e., raster image sets) of multiple remote sensing images belonging to the same target water level group, divided according to the water level observation dataset and strictly spatially aligned; the gridding calculation process in the middle schematically shows the steps to determine the water body type corresponding to this group of raster image sets, that is, by accumulating pixel values ​​pixel by pixel and calculating the ratio of it to the total number of images to obtain the water body identification frequency, and comparing this frequency with a preset identification frequency threshold (60% in the example in the figure), thereby determining the high-frequency areas that meet the threshold condition (shown in red in the figure) as stable water bodies; the merging result on the right corresponds to the stable water body range raster (raster image set) generated according to the above determination, which will serve as the basic data source for subsequent extraction of three-dimensional water level contour data of the shoal area by combining the average water level value.

[0094] Figure 4 The illustration shows a rendering of a generated three-dimensional water level contour data for a shallow water area according to an embodiment of this application. Figure 4As shown in the embodiments of this application, the multiple colored curves superimposed on the real surface texture are the water surface boundary lines generated based on long-term remote sensing images and synchronous water level observation data. Specifically, multiple sets of raster images corresponding to different average water level values ​​are processed to extract the range of stable water bodies under each water level and vectorize them. Then, the corresponding average water level values ​​of each set (such as the values ​​clearly marked as 120m, 110m, 100m, 90m, etc. in the figure) are assigned as elevation attributes to the corresponding boundary lines. These dense curves with accurate elevation information constitute the three-dimensional water level contour data of the shoal area. They are distributed in a hierarchical manner according to the terrain trend, effectively covering the shoal area that is difficult to reach by conventional measurement, and serve as the terrain skeleton for constructing an irregular triangular network to generate the final second digital elevation model of the shoal area.

[0095] Figure 5 A schematic diagram illustrating the structure of a computer device according to an embodiment of this application is provided. Figure 5 As shown in the illustration, this application provides a computer device that may include: Memory 510 is configured to store instructions; and Processor 520 is configured to retrieve instructions from memory 510 and to implement the methods described above when executing instructions.

[0096] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0097] This application also provides a machine-readable storage medium storing instructions that cause a machine to perform the above-described method.

[0098] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0099] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0100] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0101] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0102] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0103] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0104] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0105] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0106] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for constructing an integrated land and water digital elevation model, characterized in that, The method includes: The first digital elevation model of the land area in the target region is determined based on the application scenario identifier; Based on multiple remote sensing images and water level observation datasets of the shallow water area in the target region, multiple sets of spatially aligned raster images and the average water level value corresponding to each set of raster images are determined; the water level observation dataset stores the water level corresponding to each remote sensing image. The water type at each pixel coordinate of each raster image set is determined based on a preset recognition frequency threshold. Based on the water type at each pixel coordinate of any set of raster images, generate a stable water body range raster corresponding to that set of raster images; The stable water body range grid is morphologically processed by a preset area threshold, and the grid after morphological processing is smoothed and vectorized to obtain the water surface boundary line corresponding to the set of grid images. The average water level value corresponding to each set of raster images is used as the elevation attribute and assigned to the water surface boundary line corresponding to that set of raster images to obtain the three-dimensional water level contour data of the shoal area. A second digital elevation model of the shoal area is generated based on the three-dimensional water level contour data and spatial interpolation method; The siltation status of the deep water area is determined based on historical underwater topographic data of the deep water area in the target region and real-time water depth data of multiple verification points. A third digital elevation model for the deep water area is determined based on the siltation state; A land-water integrated digital elevation model of the target area is generated based on the first digital elevation model, the second digital elevation model, the third digital elevation model, and the terrain control point data of the target area. The step of determining multiple spatially aligned raster image sets and the average water level value corresponding to each raster image set based on multiple remote sensing images and water level observation datasets of the shallow water area in the target region includes: Multiple target water level groups are determined based on the one-to-one correspondence between the water levels in each remote sensing image; Construct a water body identification feature layer corresponding to any remote sensing image based on the spectral characteristics or backscattering intensity of the remote sensing image. The water body identification feature layer corresponding to each remote sensing image is binarized using an adaptive threshold segmentation algorithm, and a water body binary map of each remote sensing image is output. Based on the water level corresponding to each remote sensing image, the binary water body map of the remote sensing image is divided into the corresponding target water level group; Spatially align the binary images of all water bodies within each target water level group to obtain multiple sets of raster images; Determine the average water level value corresponding to each set of raster images.

2. The method according to claim 1, characterized in that, The steps for determining the first digital elevation model of land in the target area based on the application scenario identifier include: When the application scenario is identified as a high-precision terrain modeling scenario, the ground point cloud data is determined based on the terrain control point data, the original three-dimensional point cloud data of the land area, and the point cloud ground filtering algorithm. The three-dimensional terrain profile of the land area is fused into the ground point cloud data to obtain fused data; The first digital elevation model is obtained by performing spatial gridding calculation on the fused data using the spatial interpolation method.

3. The method according to claim 2, characterized in that, The method further includes: When the application scenario is identified as a large-scale macroscopic analysis scenario, an aerospace remote sensing digital elevation model covering the target area is obtained; The first digital elevation model is determined based on the aforementioned aerospace remote sensing digital elevation model.

4. The method according to claim 2, characterized in that, The method further includes: When the application scenario is identified as a local supplementary survey and a special terrain scenario, high-precision discrete terrain points of the land area are obtained; the high-precision discrete terrain points are collected along the terrain feature lines and terrain changes of the land area based on real-time dynamic carrier phase differential technology. The first digital elevation model is constructed based on the high-precision discrete terrain points and the spatial interpolation method.

5. The method according to any one of claims 1 to 4, characterized in that, The step of determining the water body type at each pixel coordinate of each raster image set according to a preset recognition frequency threshold includes: Determine the pixel value at each pixel coordinate for each raster image in each set of raster images; The pixel values ​​at the same pixel coordinates of each raster image in any set of raster images are summed to obtain the total pixel value of the set of raster images at that pixel coordinate. The ratio between the total pixel value and the total number of raster images in the set of raster images is calculated and used as the water body identification frequency of the set of raster images at the pixel coordinates. Based on the water body identification frequency and the preset identification frequency threshold, the water body type at the pixel coordinates of the raster image set is determined.

6. The method according to claim 5, characterized in that, The step of determining the siltation state of the deep water area based on historical underwater topographic data and real-time water depth data from multiple verification points in the target area includes: Acquire historical underwater topographic data of the deep water area and real-time water depth data of multiple verification points; Based on the planar coordinates of each verification point, historical elevation data of each verification point is extracted from the historical underwater topographic data; the historical underwater topographic data and the planar coordinate system of each verification point are the same. Determine the elevation difference between the real-time water depth data and the historical elevation data of any verification point; Determine the arithmetic mean of the corresponding verification points based on the elevation differences of each verification point; If the arithmetic mean is less than a preset change threshold, the siltation state of the deep water area is determined to be slight siltation; If the arithmetic mean is greater than or equal to the preset change threshold, the siltation state of the deep water area is determined to be significant siltation.

7. The method according to claim 6, characterized in that, The step of determining the third digital elevation model of the deep water area based on the siltation state includes: When the siltation state is described as slight siltation, a first underwater elevation discrete point is generated based on the historical underwater topographic data. The third digital elevation model is generated based on the spatial interpolation method and the first underwater elevation discrete points.

8. The method according to claim 6, characterized in that, The method further includes: When the siltation state is significant siltation, high-density underwater topographic point cloud data of the deep water area is acquired; The high-density underwater topographic point cloud data is subjected to sound velocity profile correction and tidal level correction, and abnormal noise points are removed to obtain the second underwater elevation discrete points. The third digital elevation model is obtained by reconstructing the second underwater elevation discrete points using the spatial interpolation method.

9. The method according to claim 8, characterized in that, The step of generating an integrated land and water digital elevation model of the target area based on the first digital elevation model, the second digital elevation model, the third digital elevation model, and the terrain control point data of the target area includes: The plane coordinate systems of the first digital elevation model, the second digital elevation model, and the third digital elevation model are respectively converted into a unified coordinate system to obtain multiple digital elevation models with a unified coordinate system. Each digital elevation model is spatially matched with the terrain control point data to calculate the elevation system deviation of the corresponding digital elevation model. With the goal of eliminating the elevation system bias of each digital elevation model, the elevation values ​​of each digital elevation model are corrected by the regional network adjustment method to obtain the corrected digital elevation models. A land-water integrated digital elevation model of the target area is generated based on the corrected digital elevation models.

10. A computer device, characterized in that, include: The memory is configured to store instructions; as well as A processor configured to retrieve the instructions from the memory and, when executing the instructions, to implement the method according to any one of claims 1 to 9.

11. A machine-readable storage medium, characterized in that, The machine-readable storage medium stores instructions for causing the machine to perform the method according to any one of claims 1 to 9.

Images