Köppen Climate Identification and Classification Method Based on Geomorphological Features

Abstract

This invention discloses a Köppen climate identification and classification method based on geomorphic features. It acquires multi-source geomorphic feature data and utilizes bilinear resampling and Gaussian smoothing techniques to achieve spatial normalization and terrain denoising preprocessing. Seventeen feature parameters are extracted to construct a geomorphic feature dataset. A terrain roughness index is constructed using the local standard deviation of elevation to divide complex and flat terrain regions. Spatial iterative clustering with variable-density seed points is performed to generate geographically homogeneous superpixel units. A training set is constructed through visual interpretation of sample points. A stacked ensemble learning framework is used to generate base layer predicted features and train a meta-classifier layer to complete the fusion result feature set. Global superpixel classification is performed for backfilling, and the results are topologically merged and extracted based on spatial adjacency principles to generate maps. This method achieves detailed identification and classification of Köppen climate at the hundred-meter level, providing a robust, convenient, and economically feasible solution for accurate local climate identification in complex geographical environments.

Description

Technical Field

[0001] This invention relates to a Köppen climate identification and classification method based on topographic features, belonging to the field of climate identification technology. Background Technology

[0002] The Köppen climate classification system is currently the most widely used empirical climate classification system in geography, meteorology, and ecology research. It systematically uses two key climate factors, temperature and precipitation, to define different climate zones.

[0003] Because the Köppen classification is a typical empirical classification method, it relies solely on two surface statistics—temperature and precipitation—while ignoring aspects, slope, and local circulation caused by topography. This makes it difficult to effectively represent the significant microclimate differences controlled by landforms in complex terrain regions, resulting in low accuracy in mountainous and basin areas with dramatic topographic changes. Furthermore, existing and improved climate zoning schemes in China do not adequately map the interaction mechanism between topography and climate, and they all rely on measured temperature and precipitation data from climate stations, resulting in low spatial resolution. In addition, collecting meteorological data in remote areas using traditional measurement methods requires a significant amount of manpower and resources.

[0004] A stable physical mapping relationship exists between climate and landforms: landforms alter the solar radiation budget and near-surface energy balance through elevation, slope, aspect, and topographic shading; they regulate water vapor transport and precipitation distribution, shaping climate patterns; and soil organic carbon storage reflects the intensity of rainfall and heat. This mapping relationship is stable on multi-year scales, enabling landform elements not only to explain the spatial distribution patterns of Köppen climate types but also to serve as high-resolution spatial constraints when meteorological data is insufficient or spatial resolution is low, for characterizing climate type boundaries and identifying transition zone details. By constructing a machine learning mapping framework between landform factors and climate types, high-precision identification and zoning of Köppen climate types can be achieved at a scale of hundreds of meters.

[0005] Therefore, there is a need to propose a method that can identify and classify Köppen climate based on known public data on geomorphological features. Summary of the Invention

[0006] The technical problem to be solved by this invention is to provide a Köppen climate identification and classification method based on geomorphological features for classifying Köppen climate regions, so as to meet the requirements of convenience and economic feasibility.

[0007] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0008] The Köppen climate identification and classification method based on geomorphological features includes the following steps:

[0009] Step 1: Obtain multi-source heterogeneous data of the study area, including DEM data, soil physicochemical property data and geomorphological type map. Resample the above multi-source heterogeneous data to a uniform spatial resolution and project the original coordinate system to a uniform projection coordinate system.

[0010] Step 2: Based on Step 1, import the resampled and projected multi-source heterogeneous data into a water body filter mask, identify and remove interference from non-land surface on topographic gradient calculation, and use a smoothing operator to perform convolution summation on the masked DEM data.

[0011] Step 3: Extract 13 feature parameters from the smoothed DEM data: elevation, slope, aspect, hill shadow, north azimuth, east azimuth, shape index, horizontal curvature, vertical curvature, mean curvature, Gaussian curvature, minimum curvature, and maximum curvature. Extract 2 feature parameters from the masked soil physicochemical property data: average density of soil organic carbon in 0-5 cm and average content of soil organic carbon in 0-5 cm. Extract 2 feature parameters from the masked landform type map: genetic type landform and secondary genetic type landform.

[0012] Step 4: Based on the smoothed DEM data, set a local window, calculate the local standard deviation of elevation in the study area to quantify the terrain roughness, divide the complex terrain area and the flat terrain area according to the terrain roughness threshold, and construct a variable density seed point model that generates high-density seed points in the complex terrain area with fine-grained step size and low-density seed points in the flat terrain area with coarse-grained step size.

[0013] Step 5: Based on the 17 feature parameters extracted in Step 3, construct a geomorphic feature dataset and input the variable density seed point model from Step 4 into the spatial iterative clustering algorithm model. Perform nonlinear iterative clustering with variable density seed point constraints in the spatial dimension to segment the study area and generate spatially homogeneous superpixel units and corresponding unit feature data.

[0014] Step 6: Based on historical Köppen climate distribution maps and normalized difference vegetation index data, spatial differences are visually identified, sample points are selected within the study area, and the corresponding Köppen climate label coding attribute data are labeled to construct a climate sample point dataset.

[0015] Step 7: Construct a Köppen climate identification and classification model based on stacked ensemble machine learning, including a base classifier layer and a meta classifier layer. Use the superpixel unit feature data generated in Step 5 as the independent variable, perform K-fold partitioning on the climate sample point dataset constructed in Step 6, use the generated training subset as the dependent variable, perform parallel classification in the base classifier layer, and generate the outer fold prediction feature set of the base classifier layer.

[0016] Step 8: Using the outward folding prediction feature set output in Step 7 as the independent variable and the climate sample point dataset constructed in Step 6 as the dependent variable, the optimal ensemble mapping is learned through the meta-classifier layer, and the fusion result of the meta-classifier layer is output.

[0017] Step 9: Use the fusion results output in Step 8 to predict the superpixel units in the study area, backfill the climate label codes generated by the classification into the corresponding superpixel unit space, and perform topological merging on superpixel units that are geographically adjacent and have completely consistent climate category codes according to the spatial adjacency principle to eliminate false boundaries within the same climate type and generate Köppen climate identification classification map.

[0018] Compared with the prior art, the present invention, employing the above technical solution, has the following technical effects:

[0019] 1. This invention fully considers the impact of near-surface thermal feedback on local climate, enabling detailed identification and classification of Köppen climate at the 100-meter level, improving the resolution of Köppen climate identification and classification maps, and without requiring the use of temperature and precipitation data from actual meteorological stations.

[0020] 2. The results of this invention can be widely applied to fields such as agricultural precision zoning, ecological monitoring and environmental risk assessment, providing a highly robust, convenient and economically feasible technical means for the accurate identification of local climate in complex geographical environments. Attached Figure Description

[0021] Figure 1 This is a flowchart of the Köppen climate identification and classification method based on landform features according to the present invention;

[0022] Figure 2 This is a schematic diagram of the geomorphic feature dataset provided by the present invention;

[0023] Figure 3 This is a model result of surface roughness and variable density seed points extracted from the sample area of ​​the example;

[0024] Figure 4 This is a diagram showing the nonlinear iterative clustering results of the sample region with and without variable density seed point constraints in the example.

[0025] Figure 5 It is a historical Köppen climate distribution map of the sample area from 2000 to 2020 at a resolution of 1 km.

[0026] Figure 6 This is a schematic diagram of a climate sample point dataset constructed by visual interpretation of historical Köppen climate distribution maps and Normalized Difference Vegetation Index (NDVI) data base maps in the example sample area.

[0027] Figure 7 This is a schematic diagram of an example sample area stacking integrated machine learning classification model;

[0028] Figure 8 This is the final 300m resolution Köppen climate identification and classification result image of the embodiment. Detailed Implementation

[0029] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0030] like Figure 1 As shown, this invention proposes a Köppen climate identification and classification method based on landform features, comprising the following steps:

[0031] Step 1: Obtain DEM (Digital Elevation Model) data, soil physicochemical property dataset, and geomorphological type map of the study area. Perform spatial scale resampling on the above multi-source heterogeneous data to a uniform spatial resolution, setting the spatial resolution size to 300 meters. Then, transform the original coordinate system to a uniform WGS-84 projected coordinate system.

[0032] The resampling method used is the bilinear resampling method, and its resampling formula is:

[0033] ,

[0034] In the formula, These are the attribute estimates of the target pixel points in the WGS-84 projected coordinate system after sampling. These represent the four pixel center points closest to the target pixel in the original coordinate system. These are the geographic attribute values ​​corresponding to the center points of the four pixels closest to the target pixel; and The weighting coefficients are calculated based on the distances between the target pixel and the original pixel before resampling of the multi-source heterogeneous data in the latitude and longitude directions.

[0035] In this embodiment, the resampled DEM data of the study area comes from the 30-meter DEM image of the SRTM satellite, the soil physicochemical property dataset comes from the 250-meter global soil property dataset provided by the International Soil Reference and Information Center (ISRIC), and the geomorphic type dataset comes from the China 1:1,000,000 digital geomorphic vector dataset provided by the National Earth System Science Data Center.

[0036] Step two involves importing the aforementioned multi-source heterogeneous data into a water body filtering mask to identify and remove non-land surfaces such as rivers and lakes to prevent them from interfering with the topographic gradient calculation. Subsequently, a Gaussian smoothing operator is used to perform convolution and summation operations on the masked DEM data to calculate a weighted average of the cell neighborhood.

[0037] The formula for the Gaussian smoothing operator is:

[0038] ,

[0039] In the formula, For the Gaussian smoothing operator, within the operator The weighting coefficient at the location, The standard deviation of the normal distribution. These represent the spatial coordinate distance offsets of DEM cell points within the operator relative to the central DEM cell point of the operator in the horizontal and vertical directions, respectively.

[0040] In this embodiment, the water filtering mask is set to the MOD44W.006 water mask extracted by the MODIS satellite; the standard deviation of the Gaussian smoothing operator is set to 2, and the convolution kernel radius is 3.

[0041] Step 3: Based on the smoothed DEM data of the study area, extract 13 feature parameters, including elevation, slope, aspect, hill shadow, north azimuth, east azimuth, shape index, and six types of curvature: horizontal, vertical, average, Gaussian, minimum, and maximum. Figure 2 As shown;

[0042] Based on the masked soil physicochemical property dataset, two feature parameters were extracted: average density of soil organic carbon in 0-5 cm and average content of soil organic carbon in 0-5 cm.

[0043] Based on the masked landform type dataset, two types of feature parameters are extracted: causal landform type and secondary causal landform type.

[0044] Step four: Based on the smoothed DEM data of the study area, a local window is set, and the local standard deviation of elevation in the study area is calculated to quantify the terrain roughness. A threshold is set according to the terrain roughness to divide the area into complex terrain regions and flat terrain regions. A variable-density seed point model is constructed, using a fine-grained step size to generate high-density seed points in complex terrain regions and a coarse-grained step size to generate low-density seed points in flat terrain regions.

[0045] In this embodiment, the local window size is defined as 3×3; the formula for terrain roughness is:

[0046] ,

[0047] In the formula, This represents the terrain roughness value. This represents the total number of pixels within the local window. For the first in the local window The elevation value of each pixel. This represents the average elevation value within the local window.

[0048] The terrain roughness threshold is set to 30 meters; areas with a roughness greater than or equal to 30 meters are classified as complex terrain, while those less than 30 meters are classified as flat terrain. The fine-grained step size is set to 12 pixels, and the coarse-grained step size is set to 36 pixels. The extracted surface roughness and variable density seed point model results are as follows: Figure 3 As shown.

[0049] Step 5: Construct a geomorphic feature dataset based on the 17 feature parameters extracted in Step 3, input it into the SNIC spatial iterative clustering algorithm model, perform nonlinear iterative clustering with variable density seed point constraints in the spatial dimension, segment the study area to generate spatially homogeneous superpixel units and corresponding unit feature data;

[0050] In this embodiment, the seed point is set to the variable-density seed point obtained in step four, with a compactness factor of 5 and a cell connectivity of 4. Nonlinear iterative clustering constrained by the variable-density seed point is performed in the spatial dimension, and the segmentation of the study area generates the following results: Figure 4 As shown, this paper presents two types of spatially homogeneous superpixel unit segmentation and corresponding unit feature data with and without variable density seed point constraints.

[0051] Step Six, Combining Figure 5 Historical Köppen climate distribution maps and Normalized Difference Vegetation Index (NDVI) data extracted from MODIS satellite imagery were used to visually identify spatial differences. Typical sample points were selected within the study area and their corresponding Köppen climate label coding attribute data were labeled. A climate sample point dataset was constructed through visual interpretation, and the dataset was randomly sampled and divided into training and test sets at a ratio of 7:3.

[0052] The historical Köppen climate distribution map was generated by interpolating monthly temperature and precipitation raster data from 2000 to 2020 within the CHELSA 1km resolution land surface climatology dataset; such as Figure 6 As shown in this embodiment, according to the four Köppen climate types in the study area, 200 temperate semi-arid climate points were selected and assigned the coded attribute value 1, 200 warm temperate monsoon climate points were assigned the coded attribute value 2, 200 cold temperate monsoon climate points were assigned the coded attribute value 3, and 200 subarctic monsoon climate points were assigned the coded attribute value 4.

[0053] Step 7: Construct a Köppen climate identification and classification model based on stacked ensemble machine learning, as shown in the model below. Figure 7 The diagram shows a two-level structure consisting of a base classifier layer and a meta-classifier layer: the base classifier layer contains three candidate machine learning classifiers: random forest, gradient boosting tree, and decision tree, which are trained under the same constraints; the meta-classifier layer uses a random forest as its machine learning classifier.

[0054] The superpixel unit feature data generated in step five is used as the independent variable; the training set divided in step six is ​​further divided into K folds, with K being 5 folds, and each fold iteration uses the same feature parameter set and consistent partitioning logic to ensure the statistical consistency and reproducibility of the outer fold prediction feature set. The generated training subset is used as the dependent variable and input into the basic classifier layer for parallel classification, and the output generates the outer fold prediction feature set of the basic classifier layer.

[0055] Step 8: Using the outward folding prediction feature set output in Step 7 as the independent variable and the training set generated in Step 6 as the dependent variable, the meta-classifier layer learns the optimal ensemble mapping and outputs the fusion result of the generated meta-classifier layer.

[0056] Step nine involves using the fusion results from step eight to predict superpixel units across the entire study area, and backfilling the climate label codes generated from the classification into the corresponding superpixel unit space. Based on the spatial adjacency principle, superpixel units that are geographically adjacent and have completely identical climate category codes are topologically merged to eliminate false boundaries within the same climate type, generating units such as… Figure 8 The image shown is a Köppen climate identification and classification map. The spatial adjacency principle refers to the criterion for determining whether there is a spatial connection, contact, or proximity relationship between climate type superpixel units. If there are superpixel units that are geographically adjacent and have completely identical climate category codes, then topological merging is performed.

[0057] For the final Köppen climate identification and classification map results, the accuracy was evaluated using the climate sample point dataset test set from step four, and the evaluation results were reflected by user accuracy, overall accuracy, and kappa coefficient, all rounded to four decimal places.

[0058] The quantitative evaluation results are shown in Table 1. The Köppen climate identification and classification map results extracted by the method provided by this invention have a producer accuracy of ≥0.8957, a user accuracy of ≥0.8957, an overall accuracy of 93.73%, and a kappa coefficient of 0.9223.

[0059] Table 1. Accuracy Evaluation of Köppen Climate Identification and Classification Results in the Examples

[0060]

[0061] Based on the same inventive concept, embodiments of this application provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the aforementioned Köppen climate identification and classification method based on topographic features.

[0062] Based on the same inventive concept, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the aforementioned Köppen climate identification and classification method based on topographic features.

[0063] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention 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.

[0064] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. 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 illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0065] 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.

[0066] 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.

[0067] The above embodiments are merely illustrative of the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solution based on the technical concept proposed in this invention shall fall within the scope of protection of this invention.

Claims

1. A Köppen climate identification and classification method based on geomorphological features, characterized in that, Includes the following steps: Step 1: Obtain multi-source heterogeneous data of the study area, including DEM data, soil physicochemical property data and geomorphological type map. Resample the above multi-source heterogeneous data to a uniform spatial resolution and project the original coordinate system to a uniform projection coordinate system. Step 2: Based on Step 1, import the resampled and projected multi-source heterogeneous data into a water body filter mask, identify and remove interference from non-land surface on topographic gradient calculation, and use a smoothing operator to perform convolution summation on the masked DEM data. Step 3: Extract 13 feature parameters from the smoothed DEM data: elevation, slope, aspect, hill shadow, north azimuth, east azimuth, shape index, horizontal curvature, vertical curvature, mean curvature, Gaussian curvature, minimum curvature, and maximum curvature; extract 2 feature parameters from the masked soil physicochemical property data: average density of soil organic carbon in 0-5 cm and average content of soil organic carbon in 0-5 cm; extract 2 feature parameters from the masked geomorphological type map: genetic type geomorphology and secondary genetic type geomorphology. Step 4: Based on the smoothed DEM data, set a local window, calculate the local standard deviation of elevation in the study area to quantify the terrain roughness, divide the complex terrain area and the flat terrain area according to the terrain roughness threshold, and construct a variable density seed point model that generates high-density seed points in the complex terrain area with fine-grained step size and low-density seed points in the flat terrain area with coarse-grained step size. Step 5: Based on the 17 feature parameters extracted in Step 3, construct a geomorphic feature dataset and input the variable density seed point model from Step 4 into the spatial iterative clustering algorithm model. Perform nonlinear iterative clustering with variable density seed point constraints in the spatial dimension to segment the study area and generate spatially homogeneous superpixel units and corresponding unit feature data. Step 6: Based on historical Köppen climate distribution maps and normalized difference vegetation index data, spatial differences are visually identified, sample points are selected within the study area, and the corresponding Köppen climate label coding attribute data are labeled to construct a climate sample point dataset. Step 7: Construct a Köppen climate identification and classification model based on stacked ensemble machine learning, including a base classifier layer and a meta classifier layer. Use the superpixel unit feature data generated in Step 5 as the independent variable, perform K-fold partitioning on the climate sample point dataset constructed in Step 6, use the generated training subset as the dependent variable, perform parallel classification in the base classifier layer, and generate the outer fold prediction feature set of the base classifier layer. Step 8: Using the outward folding prediction feature set output in Step 7 as the independent variable and the climate sample point dataset constructed in Step 6 as the dependent variable, the optimal ensemble mapping is learned through the meta-classifier layer, and the fusion result of the meta-classifier layer is output. Step 9: Use the fusion results output in Step 8 to predict the superpixel units in the study area, backfill the climate label codes generated by the classification into the corresponding superpixel unit space, and perform topological merging on superpixel units that are geographically adjacent and have completely consistent climate category codes according to the spatial adjacency principle to eliminate false boundaries within the same climate type and generate Köppen climate identification classification map.

2. The Köppen climate identification and classification method based on geomorphological features according to claim 1, characterized in that, In step 1, the resampling method used is the bilinear resampling method, and the formula is: ， In the formula, For the target pixel points in the resampled projected coordinate system The attribute estimates, These are the four pixel center points closest to the target pixel in the original coordinate system. These are the geographic attribute values ​​corresponding to the center points of the four pixels closest to the target pixel. The weighting coefficient is calculated based on the distances between the target pixel and the original pixel before resampling of the multi-source heterogeneous data in the latitude and longitude directions.

3. The Köppen climate identification and classification method based on geomorphological features according to claim 1, characterized in that, In step 2, a Gaussian smoothing operator is used to perform convolution and summation operations on the masked DEM data. The formula for the Gaussian smoothing operator is as follows: ， In the formula, For the Gaussian smoothing operator, within the operator The weighting coefficient at the location, The standard deviation of the normal distribution. These represent the spatial coordinate distance offsets of the DEM pixel points within the operator relative to the DEM pixel point at the operator center in the horizontal and vertical directions, respectively.

4. The Köppen climate identification and classification method based on geomorphological features according to claim 1, characterized in that, In step 4, the local window size is set to 3*3, and the formula for calculating the terrain roughness of the local window is: ， In the formula, This represents the terrain roughness value. This represents the total number of pixels within the local window. For the first in the local window The elevation value of each pixel. This represents the average elevation value within the local window. Set the terrain roughness threshold to 30 meters, and classify local windows with terrain roughness values ​​greater than or equal to 30 meters as complex terrain areas, and local windows with terrain roughness values ​​less than 30 meters as flat terrain areas.

5. The Köppen climate identification and classification method based on geomorphological features according to claim 1, characterized in that, In step 5, the spatial iterative clustering algorithm model is the SNIC superpixel segmentation algorithm, and the formula is: ， In the formula, For the first The center vector of each superpixel unit To be categorized into the first The number of pixels in a superpixel unit For the first Within the first superpixel unit 17 feature parameters per pixel.

6. The Köppen climate identification and classification method based on geomorphological features according to claim 1, characterized in that, In step 6, the historical Köppen climate distribution map is generated using monthly temperature data and precipitation raster data from the CHELSA 1km resolution land surface climatology dataset. The types include temperate semi-arid climate, warm temperate humid climate, cold temperate humid climate, and subarctic humid climate, totaling four climate types.

7. The Köppen climate identification and classification method based on geomorphological features according to claim 1, characterized in that, In step 7, the basic classifier layer and the meta-classifier layer constitute a two-level structure. The basic classifier layer includes three parallel candidate machine learning classifiers: random forest, gradient boosting tree, and decision tree. The three candidate machine learning classifiers are trained under the same conditions and constraints. The machine learning classifier used in the meta-classifier layer is random forest.

8. The Köppen climate identification and classification method based on geomorphological features according to claim 1, characterized in that, In step 9, the spatial adjacency principle refers to the criteria for determining whether there is a spatial connection, contact or proximity relationship between superpixel units. If there are superpixel units that are geographically adjacent and have completely identical climate category codes, then topological merging is performed.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the Köppen climate identification and classification method based on topographic features as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the Köppen climate identification and classification method based on topographic features as described in any one of claims 1 to 8.

Images