A Large-Scale Sand Dune Landform Fine Classification Method Based on DEM

Abstract

This invention discloses a large-scale, fine-grained classification method for dune landforms based on DEM (Diagram of Earth Image Processing). The method includes: establishing dune landform calculation units with low flow thresholds using inverse topography of a desert region's DEM and flow data generated by a priority flooding algorithm; extracting elevation values ​​corresponding to the boundary nodes of the dune landform calculation units to construct the base topography of the desert region; calculating the difference between the original DEM and the base topography to obtain the local relative elevation difference of the dune region; reclassifying the dune region according to a set threshold to obtain an initial classification result for the dune landforms; eliminating fragmentation in the initial classification result according to a set area threshold, and smoothing the boundaries according to a tolerance threshold to obtain the final, fine-grained classification result. This invention enables automatic, rapid, continuous, and complete classification of various dune landform types at different scales and morphologies, thus providing an important data supplement for global geomorphological classification and mapping, and providing data support for related research.

Description

Technical Field

[0001] This invention belongs to the field of geomorphological classification, specifically to a method for fine classification of large-scale dune landforms based on DEM. Background Technology

[0002] Geomorphological classification has always been an important fundamental theoretical issue in geomorphological research, a major way for geomorphology to serve society, and a cutting-edge research topic in the current field of digital geomorphology.

[0003] Sand dunes, as unique landforms formed by the erosion, transportation, and deposition of loose sand on the Earth's surface, have extremely high research value in global landform classification. For example, by analyzing the morphology and material composition of sand dunes in the Namib Desert of Africa, information on climate and surface material changes along the west coast of Africa over tens of millions of years can be obtained, which is of great value to disciplines such as atmospheric science, earth science, and archaeology.

[0004] Traditional classification and extraction of aeolian landforms are primarily qualitative, relying heavily on small-scale manual measurements or remote sensing imagery for manual interpretation of dune types. While these methods offer high accuracy within a certain range, they are highly subjective and limited, and cannot quickly and automatically classify large-scale dune landform types. In contrast, Digital Elevation Models (DEMs), as a digital representation of landform and elevation information, have been widely applied in digital topography and geomorphology analysis, providing support for global landform classification. Currently, mainstream DEM data covers the vast majority of desert regions on Earth, serving as a data source for large-scale dune landform classification.

[0005] Existing DEM-based dune landform classification methods are mostly limited to specific dune types in small areas and cannot be extended to regions with different dune landform types. These methods generally use landform factors such as slope and relief for classification. However, these factors are calculated based on pixels, making them highly susceptible to the size of the set convolution kernel, and they do not consider that landforms are composed of individual landform object units. The results of these methods contain a large number of fragmented patches, significant boundary effects, and low refinement, making the final classification results unsuitable for large-scale application. Summary of the Invention

[0006] Purpose of the Invention: To overcome the limitations of existing technologies, which rely on topographic factors such as slope and undulation to classify sand dune landforms, which vary in shape and size due to differences in climate, surface material composition, and topography, and which are restricted by convolution kernel size and calculation methods, thus failing to meet the needs of global landform classification, this invention provides a large-scale, fine-grained classification method for sand dune landforms based on DEM. This method can automatically, quickly, continuously, and completely classify various types of sand dune landforms at different scales, thus providing an important data supplement for global landform classification and mapping, and providing data support for related research such as desert morphology and texture and desert dynamics.

[0007] Technical solution: To achieve the above objectives, this invention provides a method for fine classification of large-scale sand dune landforms based on DEM, comprising the following steps:

[0008] S1: Using the inverse topography of the desert region's DEM and combined with the flow data generated by the priority flooding algorithm, establish a dune landform calculation unit with a small flow threshold.

[0009] S2: Extract the elevation values ​​corresponding to the boundary nodes of the dune landform calculation unit, and use TIN to construct the base topography of the desert region;

[0010] S3: Calculate the difference between the original DEM and the base topography to obtain the local relative elevation difference in the dune area;

[0011] Based on the obtained local relative elevation differences, the dune area is reclassified according to the set threshold to obtain the initial classification results of the dune landform.

[0012] S4: Based on the set area threshold, eliminate the fragments in the initial classification results and smooth the boundaries according to the tolerance threshold to obtain the final refined classification results.

[0013] Further, step S1 specifically includes:

[0014] A1: Invert the original DEM to obtain inverse terrain DEM data;

[0015] A2: Using the priority flooding algorithm, the water flow direction data of the inverse topography DEM is obtained;

[0016] A3: Calculate the cumulative runoff based on the runoff direction data, set the flow threshold, calculate the watershed boundary of the dune area, and use it as the calculation unit for dune landforms.

[0017] Further, step S2 specifically includes:

[0018] B1: Based on the original DEM data, extract the elevation values ​​corresponding to the boundary nodes of the dune landform calculation unit obtained in step S1.

[0019] B2: Set the nodes of the dune landform calculation unit as the elements for constructing the irregular triangular mesh surface, and use the elevation values ​​in the DEM at the corresponding locations as the elevation source for constructing the irregular triangular mesh, thus constructing the irregular triangular mesh surface of the dune area.

[0020] B3: Convert the generated irregular triangular network into raster data format to construct a continuous and complete base topographic elevation data source for the dune area.

[0021] Further, step S3 specifically includes:

[0022] C1: Subtract the base elevation data of the dune area processed in step B3 from the original DEM to obtain the local relative elevation difference data of the dune area;

[0023] C2: Using the obtained local relative elevation difference data, perform binary classification based on the threshold to divide the sand dune and inter-dune plain areas.

[0024] Further, step S4 specifically includes:

[0025] D1: Convert the dunes and inter-dune plains obtained in step C2 into polygon features of the vector layer, without simplifying the polygons;

[0026] D2: Eliminate unnecessary small patches based on the set area threshold;

[0027] D3: Using a set tolerance threshold, smooth the boundaries of the dune classification to obtain the final, refined classification result.

[0028] Furthermore, in step A3, for a square area with a longitude of 10° multiplied by a latitude of 10°, the smaller flow threshold range is 6000 to 12000.

[0029] Furthermore, the threshold range for binary classification in step C2 is 2 to 10.

[0030] The GIS software used in this invention includes ArcGIS and ArcGIS Pro.

[0031] Furthermore, in step B2, the sampling points for constructing the base topography of the dune landform are each node of its computational unit surface.

[0032] Furthermore, the area threshold for eliminating redundant small patches in step D2 is 0.01 to 0.03 square kilometers.

[0033] Furthermore, the tolerance for smoothing the boundary in step D3 is 120–180 meters.

[0034] Beneficial Effects: Compared with existing technologies, this invention primarily targets large-scale dune landform areas, enabling the acquisition of classification results for most desert regions globally using existing DEM data, filling a gap in large-scale research on dune landform classification. This invention utilizes DEM inverse topography, based on a priority flooding algorithm, to construct continuous dune landform computational units. Then, it fits the local base topography of the dune landform, calculates the relative elevation difference of the dune landform area, and performs binary classification based on thresholds to obtain dunes and inter-dune regions. Finally, it smooths boundaries and eliminates fragmentation to obtain the final classification result. The classification unit of this invention is transformed from the previous pixel scale to the landform unit object scale, breaking through the limitation of the neighborhood window size threshold of the original method. This reduces the "myopia" effect in the dune landform classification method, enabling it to be applied to the classification of more diverse and larger-scale dune landforms. Furthermore, the original method is highly susceptible to noise in DEM data, while the dune landform classification method based on landform units proposed in this invention significantly reduces the impact of noise on the overall results. In addition, this invention incorporates cartographic synthesis steps such as fragmentation removal and boundary smoothing to correct the classification results. Therefore, this invention addresses the multi-form, large-scale dune landforms by automating dune landform classification, ensuring the continuity and precision of the overall results, and achieving high accuracy in the final classification. This provides data support for research on dune morphology and dune formation, and can play an important role in research on desert texture and desert dynamics. Attached Figure Description

[0035] Figure 1 This is a flowchart of a large-scale, fine-classification method for sand dune landforms based on DEM.

[0036] Figure 2 This is a schematic diagram of the DEM data of the area to be classified in the embodiment;

[0037] Figure 3 This is a schematic diagram of the dune landform classification calculation unit generated in the embodiment;

[0038] Figure 4 This is a schematic diagram of the base topography of the dune landform area constructed using TIN in the embodiment;

[0039] Figure 5 This is a schematic diagram of the local relative elevation difference data calculated in the embodiment;

[0040] Figure 6 This is a map showing the initial sand dune landform region classification results obtained from the second classification in the example;

[0041] Figure 7 This is a map showing the final classification results of the dune landform area obtained by the present invention in the embodiments;

[0042] Figure 8The image shows the final classification result of the dune landform area obtained using the existing method in the example. Detailed Implementation

[0043] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. After reading this invention, any modifications of the invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0044] This invention provides a method for fine classification of large-scale sand dune landforms based on DEM, such as... Figure 1 As shown, it includes the following steps:

[0045] S1: Using the inverse topography of the desert region's DEM and combined with the flow data generated by the priority flooding algorithm, establish a dune landform calculation unit with a small flow threshold.

[0046] S2: Extract the elevation values ​​corresponding to the boundary nodes of the dune landform calculation unit, and use TIN to construct the base topography of the desert region;

[0047] S3: Calculate the difference between the original DEM and the base topography to obtain the local relative elevation difference in the dune area;

[0048] Based on the obtained local relative elevation differences, the dune area is reclassified according to the set threshold to obtain the initial classification results of the dune landform.

[0049] S4: Based on the set area threshold, eliminate the fragments in the initial classification results and smooth the boundaries according to the tolerance threshold to obtain the final refined classification results.

[0050] This embodiment applies the method provided by the present invention in a specific example. The experimental area was selected from a region in the Sahara Desert of Africa (e.g., [location missing]). Figure 2 (As shown). The Sahara Desert, formed approximately 2.5 million years ago, is the world's largest desert region, boasting the largest scale and extent of sand dunes globally. The experimental sample area exhibits diverse dune landforms of varying shapes and sizes, including crescent-shaped dunes, pyramidal dunes, and composite transverse dunes. Furthermore, the experimental sample area covers nearly 1.2 million square kilometers, fulfilling the objective of this invention for large-scale dune landform classification. Therefore, this dune landform region is highly suitable as a case study demonstration area to verify the effectiveness of the proposed method.

[0051] The specific process is as follows:

[0052] Step 1: Establish a dune landform calculation unit with a low flow threshold:

[0053] (11) Select the DEM of the sand dune area and use ArcGIS's Raster Calculator tool to obtain the inverse terrain DEM, specifically set to 9999 and subtract it from the original DEM;

[0054] (12) Use the priority flooding algorithm, with the input being the reverse terrain DEM, and the calculation result being the water flow direction of the reverse terrain;

[0055] (13) Use the [Flow] tool to calculate the cumulative water flow of the anti-topography. The input data is the water flow direction of the anti-topography, and the result is the anti-topography flow data, that is, the ridgeline of the landform area.

[0056] (14) Using the [Raster Calculator] tool, the flow data of this terrain area is divided into binary categories based on the threshold of 12000 for flow binary classification;

[0057] (15) Use the [River Link] tool to assign unique values ​​to each part of the runoff accumulation in this geomorphic area, mark the outlets and confluences of the runoff network, and input the data as the data after the previous binary partitioning;

[0058] (16) Use the [Catchment Area] tool to obtain the watershed boundary line with anti-topography, which is the target result of step (1) - the sand dune landform calculation unit with low flow threshold, such as Figure 3 As shown, the input data is the data obtained in step (15).

[0059] Step 2: Construct a continuous and complete base topographic elevation data source for the dune region:

[0060] (21) Use the [Raster to Polygon] tool to vectorize the dune landform calculation unit finally generated in step (1) without simplifying the inflection points;

[0061] (22) Use the [Line to Point] tool to extract the inflection points of the dune landform calculation unit;

[0062] (23) Using the [Value Extraction to Point] tool, assign the values ​​on the original DEM to the point data extracted in step (22), with the field name being dem;

[0063] (24) Using the Create TIN tool, convert the data processed in step (23) into an irregular triangular network structure. The input features are the results of step (22), where the elevation field is dem, the feature type is multipoint, and the triangulation options that conform to Delaunay are used.

[0064] (25) Using the [TIN to Raster] tool, the TIN processed in step (24) is converted to raster format. The method is linear, that is, the pixel value is calculated by applying linear interpolation to the TIN triangle. The resolution of the result is set to 30m. The result is the local base topographic data of the dune landform area, such as Figure 4 As shown.

[0065] Step 3: Calculate the local relative elevation differences in the dune landform area:

[0066] (31) Use the [Resampling] tool to resample the final result of step (2) to the same resolution and range as the original DEM;

[0067] (32) Using the [Raster Calculator] tool, subtract the result obtained in step (31) from the original DEM to obtain a schematic diagram of the relative elevation difference in a local area of ​​the dune landform, as shown in the figure below. Figure 5 As shown;

[0068] (33) Using the [Reclassification] tool, a binary classification operation is performed on the dune landform types according to the set threshold to obtain the initial dune landform classification results. Regions above the threshold are set to 0, representing dune regions, and regions below the threshold are set to 1, representing inter-dune plain regions, as detailed below the threshold. Figure 6 As shown.

[0069] Step 4: Generate refined dune landform classification results:

[0070] (41) Use the [Raster to Polygon] tool to convert the binary classification results generated in step (41) into vector format, and select not to simplify the polygons;

[0071] (42) Using the "Select Features by Attributes" tool, select the small patches that need to be eliminated based on the set area threshold of 0.01 square kilometers;

[0072] (43) Use the [Eliminate] tool to eliminate the small patches in the result of step (42), and check the "Eliminate surfaces through boundaries" option to reduce the amount of noise in the original result;

[0073] (44) Using the "Smooth Shared Edges" tool, with 120m as the smoothing tolerance, the polynomial approximate smoothing method with an exponential kernel is selected to smooth the boundaries in the result of step (44), making the result more refined, and obtaining the final refined dune landform classification result, as shown in the following figure. Figure 7 As shown.

[0074] To demonstrate the effectiveness of the method of the present invention, existing methods are used to obtain the following results: Figure 8 The dune landform classification results shown are obtained through... Figure 7 and Figure 8 The comparison shows that the method of the present invention has better continuity and more accurate classification results.

Claims

1. A method for fine classification of large-scale dune landforms based on DEM, characterized in that, The method comprises the following steps: S1: using the inverse topography of the DEM of the desert area, combining the flow data generated by the priority flooding algorithm, and establishing a sand dune landform calculation unit with a small flow threshold value; S2: extracting the elevation values corresponding to the boundary nodes of the sand dune landform calculation unit, and constructing the base terrain of the desert area by using a TIN; S3: calculating the difference between the original DEM and the base terrain to obtain the local relative height difference of the sand dune area; According to the obtained local relative height difference, the sand dune area is reclassified according to the set threshold value to obtain the initial classification result of the sand dune landform; S4: according to the set area threshold value, eliminating the small patches in the initial classification result, and smoothing the boundary according to the tolerance threshold value to obtain the final fine classification result; Step S1 is specifically: A1: taking the inverse of the original DEM to obtain inverse topography DEM data; A2: using the priority flooding algorithm to obtain the catchment flow direction data of the inverse topography DEM; A3: based on the catchment flow direction data, calculating the catchment accumulation, setting the flow threshold value, calculating the watershed boundary of the sand dune area, and taking it as the calculation unit of the sand dune landform; Step S2 is specifically: B1: based on the original DEM data, extracting the elevation values corresponding to the boundary nodes of the sand dune landform calculation unit obtained in step S1; B2: setting the nodes of the sand dune landform calculation unit as the elements for constructing the irregular triangular net surface, and taking the elevation values in the DEM corresponding to the positions as the elevation sources for constructing the irregular triangular net, to construct the irregular triangular net surface of the sand dune area; B3: converting the generated irregular triangular net into a grid data format to construct a continuous and complete base terrain elevation data source of the sand dune area.

2. The method according to claim 1, wherein, The step S3 is specifically: C1: subtracting the sand dune area base terrain elevation data processed in step B3 from the original DEM to obtain the local relative height difference data of the sand dune area; C2: using the obtained local relative height difference data to perform two classification according to the threshold value to divide the sand dune and inter-dune plain areas.

3. The method according to claim 2, wherein, The step S4 is specifically: D1: converting the sand dune and inter-dune plain results obtained in step C2 into face elements of a vector layer without simplifying the face; D2: eliminating the redundant small patches with a set area threshold value; D3: smoothing the boundary of the sand dune classification with a set tolerance threshold value to obtain the final fine classification result.

4. The method according to claim 1, wherein, The flow threshold value in step A3 is 6000-12000.

5. The method according to claim 2, wherein, The threshold value for two classification in step C2 is 2-10.

6. The method according to claim 1, wherein, In step B2, the sampling points for constructing the base terrain of the sand dune landform are the nodes of the calculation unit surface.

7. The method according to claim 3, wherein, In step D2, the area threshold value for eliminating redundant small patches is 0.01-0.03 square kilometers.

Images