A method for dividing rural landscape ecological sensitive areas based on K-MEANS clustering algorithm

Abstract

The present invention discloses a method for dividing rural landscape ecological sensitive areas based on the K-means clustering algorithm, which belongs to the field of landscape architecture. The steps of the method include establishing an ecological sensitivity evaluation system suitable for rural ecological landscape; using the grid method to collect basic information of the study area, superimposing and merging the grids, dividing the basic ecological units and naming them; performing K-means clustering analysis on the data samples of the three-level single factor, selecting different k values for clustering, and assigning values to all the single factor results obtained in the previous step; using the entropy method to obtain the weight of each factor; within each ecological unit, performing K-means clustering analysis again on the basis of the statistical results of the three-level single factor elements, and obtaining the division results of the secondary elements of each unit; repeating the steps of K-means clustering analysis on the secondary elements in each ecological unit, and obtaining the sensitivity level of the primary elements of each unit; and unifying the base map for visual expression.

Description

Technical Field

[0001] The present invention relates to the field of landscape architecture, and in particular to a method for dividing rural landscape ecologically sensitive areas based on a K-MEANS clustering algorithm. Background Art

[0002] Currently, rural development is underway and progressing across the country. However, while rural areas are undergoing a transformation, many are adopting urban development models and approaches. This has led to a tendency towards urbanization or urbanization, a loss of their original rural character and spiritual qualities, a loss of natural features, and the potential for constructive damage to the rural ecological environment and landscape. The rural ecological landscape, a comprehensive reflection of regional geographical characteristics, resource characteristics, resource endowments, and natural features formed through human intervention, is the foundation of sustainable rural development. Therefore, maintaining, protecting, restoring, and enhancing the unique rural ecological landscape is a crucial task in rural revitalization. In particular, accurately identifying ecologically sensitive areas in rural landscapes and scientifically zoning and grading them is the first step in the comprehensive improvement and optimization of rural ecological landscapes.

[0003] Ecological sensitivity refers to the degree to which an ecosystem responds to changes in the natural environment and human interference. Specifically, it refers to the ability of ecological factors to adapt to external changes or pressures without sacrificing or degrading environmental quality. It reflects the probability of ecological and environmental problems occurring in a region. Ecologically sensitive areas are defined as ecological elements or entities that are critical to the overall regional ecosystem and have poor self-recovery capabilities under human intervention. Changes to these elements will impact the regional ecosystem and require control or protection. Ecological sensitivity assessments are generally conducted using ecological factor grid analysis and geographic information system analysis, and are widely used in various fields.

[0004] Currently, most studies on ecological sensitivity analysis are semi-quantitative, employing mathematical modeling, methods based on grey system theory, and vector projection principles to reveal the ecological and environmental sensitivity of the research subjects. In recent years, with the increasing maturity of computer technology and ecological theory, methods combining ArcGIS geographic information systems and RS remote sensing technology have become increasingly popular in ecological and environmental sensitivity research. However, these methods mostly focus on single-factor, large-scale analysis, leaving relatively few studies focused on the comprehensive analysis and evaluation of ecological sensitivity in scenic areas and rural areas. Summary of the Invention

[0005] In view of the shortcomings of the existing technology, the present invention proposes a method for dividing rural landscape ecological sensitive areas based on the K-means clustering algorithm.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A method for dividing rural landscape ecological sensitive areas based on K-MEANS clustering algorithm includes the following steps:

[0008] Step A: Establish a rural ecological landscape sensitivity evaluation index system that can effectively identify rural ecologically sensitive areas;

[0009] Step B, divide the research area into several basic landscape ecological units;

[0010] Step C: Collect the three-level indicator data required for the rural landscape ecological sensitivity analysis within the research scope, and sort out the three-level single factor data samples required for the classification of landscape ecological sensitive areas;

[0011] Step D, performing K-means cluster analysis on the data samples of the three-level single factors, including estimating the clustering trend, selecting different k values and determining the optimal k value through testing and measuring the clustering quality;

[0012] Step E: performing data standardization on the single-factor clustering results obtained in step C;

[0013] Step F, using the entropy method to calculate the weights of indicators at each level in the indicator grading system;

[0014] Step G: Taking the landscape ecological unit as the unit, the standardized single factor clustering results obtained in step D are combined with the weights determined in step E to calculate the division results of the secondary indicators of each unit;

[0015] In step H, the K-means cluster analysis steps are repeated for the secondary factors, and the sensitivity level of the primary factors in each unit is obtained by combining the weights;

[0016] In step I, the K-means cluster analysis steps are repeated for the primary elements, and the final ecologically sensitive zoning of each landscape ecological unit is obtained by combining the weights, and visualized using a unified base map. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The present invention will be further described below with reference to the accompanying drawings.

[0018] Figure 1 A schematic flow chart of the method of this application;

[0019] Figure 2 The data corresponding to the third-level elements in the embodiments of this application;

[0020] Figure 3 Coordinates the elevation data in the embodiments of this application;

[0021] Figure 4 Graph showing elevation clustering results under different k values in the embodiment of the present application;

[0022] Figure 5 This is a diagram of the elevation data clustering result in an embodiment of the present application;

[0023] Figure 6 This is a diagram showing the sensitivity of some ecological units in the embodiments of this application;

[0024] Figure 7 Schematic diagram of ecological sensitivity in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0026] The present invention discloses a method for dividing rural landscape ecological sensitive areas based on the K-means clustering algorithm. The K-means clustering algorithm is:

[0027] The K-means clustering algorithm, also known as the K-means algorithm, uses distance as the similarity indicator and the sum of squared errors from sample points to category centers as the evaluation criterion for clustering effectiveness. It uses continuous iteration to minimize the sum of squared error function of the overall classification.

[0028] The method for dividing the rural landscape ecological sensitive areas is as follows:

[0029] Step A: Establish a rural ecological landscape sensitivity evaluation index system that can effectively identify rural ecologically sensitive areas;

[0030] Step B, divide the research area into several basic landscape ecological units;

[0031] Step C: Select a specific village as the research object, collect the three-level indicator data required for the rural landscape ecological sensitivity analysis within the research scope, and preprocess each data, including data cleaning and data integration. After obtaining data that can be effectively processed, organize the three-level single factor data samples required for the classification of landscape ecological sensitive areas;

[0032] Step D, performing K-means cluster analysis on the data samples of the three-level single factors, including estimating the clustering trend, selecting different k values and determining the optimal k value through testing and measuring the clustering quality;

[0033] Step E: performing data standardization on the single-factor clustering results obtained in step C;

[0034] Step F, using the entropy method to calculate the weights of indicators at each level in the indicator grading system;

[0035] Step G: Taking the landscape ecological unit as the unit, the standardized single factor clustering results obtained in step D are combined with the weights determined in step E to calculate the division results of the secondary indicators of each unit;

[0036] In step H, the K-means cluster analysis steps are repeated for the secondary elements, and the sensitivity level of the primary elements in each unit is obtained by combining the weights;

[0037] In step I, the K-means cluster analysis steps are repeated for the primary elements, and the final ecologically sensitive zoning of each landscape ecological unit is obtained by combining the weights, and visualized using a unified base map.

[0038] As a further technical solution of the present invention, step A comprises the steps of:

[0039] The landscape ecological sensitivity evaluation factors in relevant literature were sorted out and summarized, and after reclassifying the characteristics of rural ecological landscapes, a rural ecological landscape sensitivity evaluation system that can effectively identify rural landscape ecological sensitive areas was established (Table 1). The index system includes three first-level indicators, A1 stability, A2 restoration and A3 landscape value; seven second-level indicators, B1 terrain, B2 land use, B3 water body, B4 habitat sensitivity, B5 soil vulnerability, B6 man-made landscape and B7 natural landscape; and further refined into 18 third-level indicators, namely C1 elevation, C2 slope, C3 land use type, C4 vegetation coverage, C5 vegetation type, C6 water body area, C7 water quality, C8 water body coastline, C9 water body buffer zone width, C10 biodiversity, C11 endangered species, C12 habitat continuity, C13 natural factors, C14 human factors, C15 village historical buildings, C16 ancient trees and famous trees, C17 special topography and landforms and C18 orientation of important landscape nodes, basically covering all aspects of the rural ecological landscape environment.

[0040] Table 1 Classification system of rural landscape ecological sensitivity indicators

[0041]

[0042] As a further technical solution of the present invention, step B comprises the steps of:

[0043] (1) For the selected rural research area, basic site information was collected, including land use type, hydrological climate, elevation, and vegetation coverage.

[0044] Land Use Type: According to the "Current Land Use Classification" (GB / T21010-2017), my country's urban and rural land use types are divided into 12 primary categories: cultivated land, gardens, forests, grasslands, commercial and service land, industrial and mining storage land, residential land, public administration and public service land, special land, transportation land, water areas and water conservancy facilities land, and other land. Site land use information is collected based on this.

[0045] Hydroclimate: This includes both hydrological and climatic factors. Hydrological factors, depending on whether or not there is surface runoff, determine data collection requirements and include water level, flow, hydrology, sediment content, water quality, and flow velocity. Climatic factors include temperature characteristics, precipitation characteristics, wind direction and force, and precipitation amount.

[0046] Elevation: Elevation data is collected while aspect and slope are recorded.

[0047] Vegetation coverage: refers to the percentage of the vertical projection of vegetation on the ground to the total area of the statistical area. The vegetation type is recorded when collecting data.

[0048] (2) Organize the collected basic information data and perform data processing, perform overlay analysis on each layer in ArcGIS, visualize the output results, and use color blocks of different grayscales to distinguish the classification results.

[0049] (3) Divide the landscape ecological units based on the analysis results and give systematic names to different units for easy subsequent use.

[0050] As a further technical solution of the present invention, the step C comprises the steps of:

[0051] (1) According to the three-level indicator factors listed in the rural landscape ecological sensitivity evaluation index system, data were collected for each unit with the landscape ecological unit as the unit. The data from multiple information sources and multiple platforms were organized into a common format, and the data were comprehensively arranged using Excel to establish a database.

[0052] (2) Data cleaning: Delete empty data, meaningless data, and duplicate data from the original data. The processing content includes all data that affect the sensitivity attributes of the rural ecological landscape. Each grid cell only counts the indicator factor information that appears in the cell, and deletes the factor information that does not appear. For example, if there is no water body within the grid range, the values of "C6 Water Body Area", "C7 Water Quality", "C8 Water Body Coastline", and "C9 Water Body Buffer Zone Width" are all empty and need to be deleted accordingly.

[0053] (3) Data numbering: sort the cleaned data and number them in sequence, and coordinate all the three-level indicator factor data in the collected indicator system.

[0054] As a further technical solution of the present invention, the step D comprises the steps of:

[0055] (1) Estimating clustering trends: Import the data after the three-level single factor numbering into Python and use the Hopkins statistic to test whether each data set has a non-random structure, that is, the spatial randomness of the variable distribution. If the calculation results show that a data set does not have any non-random structure, then the clustering algorithm is not suitable; if the results show that the data set has a non-random structure, then the data set can be clustered and the process goes to step (2).

[0056] (2) Determine the optimal k value: The rationality of the k value in the k-means algorithm will directly affect the final clustering results. The elbow method is used to determine the optimal k value for each data set. By repeatedly training multiple k-means models and selecting different k values, the critical K value with good clustering performance in each data set is found.

[0057] (3) Cluster analysis: According to the k value determined in step (2), K-means cluster analysis was performed on each data set of the three-level indicator elements to obtain k data intervals and cluster result maps for a total of 18 data sets from C1 to C18, corresponding to different levels of rural landscape ecological sensitivity.

[0058] (4) Determine clustering quality: Use the Silhouette Coefficient method to examine the inter-cluster separation and intra-cluster compactness of each clustering result to evaluate the quality of the resulting cluster. The Silhouette Coefficient ranges from -1 to 1. The closer the value is to 1, the better the clustering effect. If the value is negative, it indicates that the clustering is unsuccessful and it is necessary to return to step (2) and reselect the k value.

[0059] As a further technical solution of the present invention, the step E comprises the steps of:

[0060] Since the single-factor clustering results obtained in step D have different dimensions and orders of magnitude, the min-max normalization method is applied to normalize the deviation of the results so that all clustering results are distributed in the interval [0,1] and dimensionless.

[0061] As a further technical solution of the present invention, the step F comprises the steps of:

[0062] (1) Use the range standardization method to standardize the evaluation indicators selected in step A. Suppose there are m objects to be evaluated and n evaluation indicators. Establish the original data matrix R = {X ij} m×n, where i = 1, 2, 3, ..., n; j = 1, 2, 3, ..., m, and the actual value of evaluation unit i on the jth indicator is X ij Assume that the standardized value of evaluation unit i on the jth indicator is I ij , then for the positive indicator, For negative indicators, Among them, X min 、X max are the minimum and maximum values before normalization.

[0063] (2) Calculate the weight of each indicator by entropy method. Calculate the proportion of the jth indicator of evaluation unit i Where i = 1, 2, 3, ..., n; j = 1, 2, 3, ..., m. Calculate the entropy value of the jth indicator of evaluation unit i The difference coefficient of the jth indicator is d i =1-e j , from which we can get the weight of the j-th indicator

[0064] As a further technical solution of the present invention, the step G comprises the steps of:

[0065] In each landscape ecological unit, the 18 standardized single-factor clustering results obtained in step E were combined with the weights of the three-level indicators C determined in step F to calculate the division results of the secondary indicators of each unit, including seven secondary indicators: "B1 terrain", "B2 land use", "B3 water body", "B4 habitat sensitivity", "B5 soil vulnerability", "B6 man-made landscape", and "B7 natural landscape".

[0066] As a further technical solution of the present invention, the step H comprises the steps of:

[0067] Repeat the K-means cluster analysis steps in steps DE for the sensitivity data results of the secondary elements in each landscape ecological unit. Combined with the weights of the secondary indicators B determined in step F, the sensitivity level classification results of the primary indicators are calculated, including the three primary indicators of "A1 stability", "A2 restoration", and "A3 landscape value".

[0068] As a further technical solution of the present invention, the step I comprises the steps of:

[0069] Repeat the K-means cluster analysis steps in steps D and D for the sensitivity data of the primary elements within each landscape ecological unit. Combined with the weights of the primary indicators A determined in step F, the ecological sensitivity level of each landscape ecological unit is calculated. The final ecological sensitivity zone is divided into five intervals: non-sensitive, slightly sensitive, moderately sensitive, highly sensitive, and extremely sensitive, and is visualized using a unified base map.

[0070] The following further discloses a specific example. In this embodiment, a method for dividing rural landscape ecological units and ecologically sensitive areas based on a clustering algorithm (K-means) is characterized in that the clustering algorithm (K-means) is:

[0071] The K-means algorithm uses distance as the similarity indicator. It uses the sum of squared errors from sample points to the center of the category as the evaluation criterion for clustering quality. It uses continuous iteration to minimize the sum of squared errors of the overall classification.

[0072] The method for dividing the rural landscape ecological sensitive areas is as follows:

[0073] Step 1: Establish an ecological landscape sensitivity evaluation system suitable for sensitive areas of plain villages. Summarize the indicator evaluation systems in the literature, select and re-stratify them to obtain the indicator evaluation system in the table below.

[0074] Table 1 Classification system of ecological sensitivity indicators

[0075]

[0076] Step 2: Divide the study area into several 100x100 square meter grids. Basic information about the study area will be collected, including administrative divisions, land use, water temperature and climate, elevation, and vegetation cover. Land use is categorized as construction land and non-construction land. Construction land is further subdivided into cultural facilities, retail, hotels, other commercial facilities, village construction, and roads. Non-construction land is categorized into water areas and agricultural and forestry land. Hydroclimate is categorized into hydrological and climatic elements. The presence of surface runoff determines data collection for hydrological elements, including water level, flow, hydrology, sediment content, water quality, and flow velocity. Climatic elements include temperature characteristics, precipitation characteristics, wind direction and force, and precipitation. Elevation data should also be recorded along with slope aspect and gradient. Vegetation cover refers to the percentage of vegetation vertically projected onto the ground relative to the total area of the survey area. Vegetation species should also be recorded along with data collection.

[0077] The collected basic information data is sorted and processed, the grid is subjected to secondary analysis, the grid is overlaid for analysis, the results are output visually, and the classification results are distinguished using color blocks of different grayscales.

[0078] Merge adjacent grids of the same grayscale, divide them into ecological units based on the analysis results, and give different units systematic names for easy subsequent use.

[0079] Step 3: Collect the three-level data required within the research scope and pre-process the data of various ecological factors, including data cleaning and data integration. After obtaining data that can be effectively processed, organize the three-level single-factor data samples required for dividing the research area;

[0080] (1) According to the three-level elements listed in the ecological landscape sensitivity evaluation index system of plain rural sensitive areas, data is collected for each grid separately, and the data is comprehensively arranged using Excel, and unnecessary data is deleted, and the required data is organized into the required coordinate form. (e.g. Figure 2 、 Figure 3 )

[0081] (2) Data cleaning: Delete irrelevant and duplicate data from the original data; Based on the geographic information data obtained after the field survey, including the data of "elevation", "slope", "land use type", "vegetation coverage", "vegetation type", "water body area", "water quality", "shoreline", "buffer zone length", "river water quality", "endangered species", "overall habitat continuity", "natural factors leading to soil fragility", "human factors leading to soil fragility", "village historical buildings", "number of ancient trees and famous trees", "special topography", and "direction of important landscape nodes" in the evaluation system; the null values in the data of the three-level indicators need to be deleted, such as the null values of "water body area", "water quality", "shoreline", and "buffer zone length", which need to be deleted, that is, the ecological unit does not contain these four three-level elements.

[0082] (3) Data numbering: Sort and number the data, and coordinate the data of "elevation", "slope", "land use type", "vegetation coverage", "vegetation type", "water body area", "water quality", "shoreline", "buffer zone length", "river water quality", "endangered species", "overall habitat continuity", "natural factors leading to soil fragility", "human factors leading to soil fragility", "village historical buildings", "number of ancient trees and famous trees", "special topography", and "direction of important landscape nodes". For example: coordinate the elevation data (such as Figure 3 ).

[0083] Step 4: Perform K-means cluster analysis on the data samples of the three-level single factor, and evaluate that the optimal k value is 5, dividing the data into 5 categories;

[0084] (1) Import the three-level single factor data into Python, select appropriate k values and perform K-means cluster analysis on the three-level factor data. Take k as 3, 4, and 5, and evaluate the classification of the data under different k values (such as Figure 4 ), check whether the data is classified reasonably, whether the cluster center is located at or close to the inflection point, etc. The evaluation shows that the best k value is 5, so the data is divided into 5 categories, corresponding to the five levels of non-sensitive, slightly sensitive, moderately sensitive, highly sensitive, and extremely highly sensitive;

[0085] (2) The large amount of data for each element is divided into 5 categories for clustering algorithm analysis, and the numerical interval of each category is obtained. The appropriate number of categories is selected to obtain the clustering result diagram of the data of "elevation", "slope", "land use type", "vegetation coverage", "vegetation type", "water body area", "water quality", "shoreline", "buffer zone length", "river water quality", "endangered species", "overall habitat continuity", "natural factors leading to soil fragility", "human factors leading to soil fragility", "village historical buildings", "number of ancient trees and famous trees", "special topography", and "direction of important landscape nodes". For example, the clustering result of elevation is obtained, and the visualization result is Figure 4 The graph shows the intervals for each level, which means the data intervals for each sensitivity. Interval 1 is -0.500-0.000, Interval 2 is 0-35.372, Interval 3 is 346.270-358.510, Interval 4 is 358.527-369.000, and Interval 5 is 369.013-442.617.

[0086] Step 5: Assign a value to each level of each single factor result obtained in step C;

[0087] The sensitivity of each factor is divided into 5 levels, namely non-sensitive, slightly sensitive, moderately sensitive, highly sensitive, and extremely highly sensitive, and each interval is assigned a sensitivity value (1, 3, 5, 7, 9).

[0088] Step 6: Calculate the weight of each indicator using the entropy method;

[0089] The range standardization formula is applied to standardize the 18 evaluation indicators, and the corresponding attributes are obtained to obtain the matrix of the corresponding proportions. The entropy value of each factor and the corresponding difference coefficient are calculated, and finally the corresponding weight is obtained:

[0090] Elevation accounts for 0.078%, slope accounts for 0.071, land use type accounts for 0.050, vegetation coverage accounts for 0.064, vegetation type accounts for 0.028, area accounts for 0.014, water quality accounts for 0.078, coastline accounts for 0.035, buffer zone length accounts for 0.014, river water quality accounts for 0.042, endangered species account for 0.047, overall habitat continuity accounts for 0.052, natural factors account for 0.022, human factors account for 0.120, village historical buildings account for 0.142, the number of ancient and famous trees accounts for 0.086, special topography and landforms account for 0.017, and the orientation of important passing nodes accounts for 0.040.

[0091] Step 6: Divide the study area into several 100x100 square meter grids. Collect basic information about the study area, including administrative divisions, land use, water temperature and climate, elevation, and vegetation cover. Land use is divided into construction land and non-construction land. Construction land is further subdivided into cultural facilities, retail commercial land, hotel land, other commercial facilities, village construction land, and road land. Non-construction land is divided into water areas and agricultural and forestry land. Hydroclimate is divided into hydrological and climatic elements. The presence of surface runoff determines data collection for hydrological elements, including water level, flow, hydrology, sediment content, water quality, and flow velocity. Climatic elements include temperature characteristics, precipitation characteristics, wind direction and force, and precipitation. Elevation data should also be recorded along with slope aspect and gradient. Vegetation cover refers to the percentage of vegetation vertically projected onto the ground relative to the total area of the survey area. Vegetation species should also be recorded along with data collection.

[0092] The collected basic information data is sorted and processed, the grid is subjected to secondary analysis, the grid is overlaid for analysis, the results are output visually, and the classification results are distinguished using color blocks of different grayscales.

[0093] Merge adjacent grids of the same grayscale, divide them into ecological units based on the analysis results, and give different units systematic names for easy subsequent use.

[0094] Step 7: After counting the results of the third-level single-factor elements in each ecological unit, K-means cluster analysis is performed again based on the weights to obtain the division results of the second-level elements of each unit;

[0095] The assignment results of the third-level elements were calculated and analyzed to obtain the sensitivity values of the second-level elements, including seven second-level indicators: "topography", "geology", "water body", "ecological sensitivity", "soil vulnerability", "artificial landscape" and "natural landscape". After obtaining their sensitivity values, the sensitivity of the second-level elements was still determined according to 1-3 for non-sensitive, 3-5 for slightly sensitive, 5-7 for moderately sensitive, 7-9 for highly sensitive and 9 for extremely highly sensitive.

[0096] Step 8: Repeat the K-means cluster analysis steps for the secondary elements in each unit to obtain the sensitivity level of the primary elements in each unit;

[0097] The sensitivity data results of the secondary elements in each ecological unit are calculated according to the weights, and the data are classified and sorted to obtain the sensitivity values of the primary elements in each unit, including three primary indicators: "stability", "restoration" and "landscape value".

[0098] Step 9: Neutralize and analyze to derive the ecological sensitivity level of each ecological unit and visualize it using a unified base map;

[0099] Neutralization analysis, based on the results to determine the sensitivity level of the first-level indicators, and then to obtain the ecological sensitivity level of each ecological unit (such as Figure 6 ) and visualized with a unified base map (such as Figure 7 ).

[0100] Taking elevation data analysis as an example, when k is 3, the data interval 1 is -0.5-0, interval 2 is -0.5, 35.327, 346.27-349.89, and interval 3 is 350-442.617. Interval 1 contains a large amount of data with an elevation of 0 and a very small amount of data with an elevation of -0.5, but does not contain all -0.5 data. Interval 2 still contains some data with a value of -0.5. The same data is separated into two different data intervals, which is an error. In addition, the data contained in interval 2 are three numerical intervals with large differences. The cluster center value point deviates from the existing data itself. The map shows that the second center point is located at an elevation of about 200, and there is no such data in the data set. It is close to a number, so it is unreasonable; when k is 4, the data interval 1 is -0.5-0, interval 2 is -0.5, 35.327, 346.27-346.30, interval 3 is 348.03-360, and interval 4 is 360.052-442.617. Interval 1 contains a large number of data with a value of 0 and a very small number of data with a value of -0.5. Interval 2 contains another part of data with a value of -0.5, 35.372 and 346.27-346.30. These three groups of data are quite different, but the center points of the clusters in the third and fourth intervals are at or near the inflection points, and the classification is more reasonable; when k is 6, there are many cluster centers and more offset inflection points. It is concise and more scientific without k=5.

[0101] The following figure Figure 5The clustering situation when k=5 is as follows: interval 1 is -0.500-0.000, interval 2 is 0-35.372, interval 3 is 346.270-358.510, interval 4 is 358.527-369.000, and interval 5 is 369.013-442.617. When k=5, interval 1 and interval 2 are clearly distinguished, and the data distribution is reasonable. The cluster centers are all located at or close to the data points; the cluster centers of the 3rd, 4th, and 5th intervals are also located at or close to the inflection points. From this test, it can be concluded that 5 is the optimal k value.

[0102] Taking a certain location as an example, this figure shows the result of clustering elevation data. After being divided into five categories, the intervals for each level can be obtained, and thus the data intervals corresponding to each sensitivity can be obtained. Interval 1 is -0.500-0.000, Interval 2 is 0-35.372, Interval 3 is 346.270-358.510, Interval 4 is 358.527-369.000, and Interval 5 is 369.013-442.617. These five intervals correspond to the five levels of sensitivity: non-sensitive, slightly sensitive, moderately sensitive, highly sensitive, and extremely sensitive.

[0103] Taking Huanglongshan as an example, after data analysis, the sensitivity zoning of some ecological units in a certain area is as follows.

[0104] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0105] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention, and such changes and modifications fall within the scope of the invention as claimed.

Claims

1. A method for dividing rural landscape ecological sensitive areas based on K-MEANS clustering algorithm, characterized in that: The following steps are involved: Step A: Establish a rural ecological landscape sensitivity evaluation index system that can effectively identify rural ecologically sensitive areas; Step B, divide the research area into several basic landscape ecological units; Step C: Collect the three-level indicator data required for the rural landscape ecological sensitivity analysis within the research scope, and sort out the three-level single factor data samples required for the classification of landscape ecological sensitive areas; Step D, performing K-means cluster analysis on the data samples of the three-level single factors, including estimating the clustering trend, selecting different k values and determining the optimal k value through testing and measuring the clustering quality; Step E: performing data standardization on the single-factor clustering results obtained in step C; Step F, using the entropy method to calculate the weights of indicators at each level in the indicator grading system; Step G: Taking the landscape ecological unit as the unit, the standardized single factor clustering results obtained in step D are combined with the weights determined in step E to calculate the division results of the secondary indicators of each unit; In step H, the K-means cluster analysis steps are repeated for the secondary elements, and the sensitivity level of the primary elements in each unit is obtained by combining the weights; Step I: Repeat the K-means cluster analysis steps for the first-level elements, combine the weights to obtain the final ecologically sensitive zoning of each landscape ecological unit, and visualize it using a unified base map; Described step D comprises the following steps: (1) Use the Hopkins statistic to test whether each data set has a non-random structure, that is, the spatial randomness of the variable distribution; if the result shows that the data set has a non-random structure, then the data set can be clustered and enter step (2); (2) Determine the optimal k value: The rationality of the k value in the k-means algorithm will directly affect the final clustering results. Use the elbow rule to determine the optimal k value for each data set. By repeatedly training multiple k-means models and selecting different k values, find the critical K value with good clustering performance in each data set. (3) According to the k value determined in step (2), K-means cluster analysis is performed on each data set of the three-level indicator elements to obtain k data intervals and clustering result maps of each data set, corresponding to different rural landscape ecological sensitivity levels; After obtaining the k data intervals and clustering result graphs of each data set, the silhouette coefficient method is used to examine the inter-cluster separation and intra-cluster compactness of each clustering result to evaluate the quality of the resulting clusters.

2. The method for dividing rural landscape ecological sensitive areas based on the K-MEANS clustering algorithm according to claim 1 is characterized in that: The first-level indicators of the index system include: stability, A2 restoration and A3 landscape value.

3. The method for dividing rural landscape ecological sensitive areas based on the K-MEANS clustering algorithm according to claim 1 is characterized in that: The secondary indicators of the indicator system include: topography, land use, water bodies, habitat sensitivity, soil fragility, man-made landscape and natural landscape.

4. The method for dividing rural landscape ecological sensitive areas based on the K-MEANS clustering algorithm according to claim 1 is characterized in that: The three-level indicators of the index system include: elevation, slope, land use type, vegetation coverage, vegetation type, water area, water quality, water coastline, water buffer zone width, biodiversity, endangered species, habitat continuity, natural factors, human factors, village historical buildings, ancient trees and famous trees, special topography and landforms, and orientation of important landscape nodes.

5. The method for dividing rural landscape ecological sensitive areas based on the K-MEANS clustering algorithm according to claim 1 is characterized in that: In the step C, after collecting the three-level indicator data, each of the three-level indicator data is preprocessed, and the preprocessing includes data cleaning and data integration.

6. The method for dividing rural landscape ecological sensitive areas based on the K-MEANS clustering algorithm according to claim 1 is characterized in that: In the step C, the three-level indicator data are processed, each layer is overlaid and analyzed in ArcGIS, the results are output visually, different color blocks are used to distinguish the classification results, and the landscape ecological units are divided according to the analysis results.

Images