Wetland hydrologic connectivity partitioning method

Abstract

The wetland hydrological connectivity zoning method disclosed in this invention includes the following steps: S1, acquiring basic data; S2, based on the basic data, constructing and calibrating a hydrodynamic model to obtain the daily spatiotemporal distribution of water level for a hydrological year; S3, based on the simulation results, extracting water level change process data within the inundated area to form a water level-time curve; S4, normalizing the water level-time curve; S5, calculating water level change characteristic values ​​based on the normalized curve; S6, performing cluster analysis on the water level change characteristic values ​​to obtain initial classification results; S7, displaying the initial classification results according to spatial coordinates to identify hydrological connectivity patches; S8, based on the principal components of the water level sequence and spatial distance of each patch, constructing a comprehensive distance matrix and performing hierarchical clustering to delineate hydrological connectivity zones. This invention solves the problem that existing wetland hydrological connectivity zoning methods are unable to accurately determine the specific boundaries of each zone.

Description

Technical Field

[0001] This invention belongs to the technical field of wetland hydrological connectivity zoning, specifically relating to wetland hydrological connectivity zoning methods. Background Technology

[0002] With rapid socio-economic development, human activities are increasingly impacting marsh wetlands. In their natural state, marsh wetlands are flat with gentle slopes, forming relatively complete and unified hydrological units. However, human activities have gradually fragmented these intact water surfaces through farmland, irrigation canals, dikes, and roads. The synergistic effect of water conservancy projects and the original wetland topography has led to the gradual obstruction of internal hydrological connectivity. Wetlands have transformed from previously unified hydrological units into several relatively independent sub-regions with relatively strong internal hydrological connectivity but obstructed inter-regional hydrological connections. This reduced hydrological connectivity weakens the diffusion and exchange of matter, energy, and ecological information, leading to ecosystem fragmentation and impacting the wetland's ecological functions such as flood regulation, water purification, and biodiversity maintenance.

[0003] Currently, commonly used methods for zoning hydrological connectivity in marsh wetlands include remote sensing imagery, in-situ testing, chemical tracing, and connectivity index methods. Each method has the following limitations: remote sensing imagery lacks sufficient accuracy in extracting water surface data from marsh wetlands with high vegetation cover; in-situ testing is not suitable for large-scale monitoring and struggles to provide specific boundaries for each zone; chemical tracing has high data acquisition costs; and connectivity index methods focus on connectivity between wetlands and external rivers, lacking detailed characterization of internal wetland zoning boundaries. Existing methods generally fail to provide spatially continuous boundaries for hydrological connectivity zones within wetlands. Summary of the Invention

[0004] The purpose of this invention is to provide a wetland hydrological connectivity zoning method, which solves the problem that existing swamp wetland hydrological connectivity zoning methods are unable to accurately give the specific boundaries of each zone.

[0005] The technical solution adopted in this invention is a wetland hydrological connectivity zoning method, comprising the following steps: S1. Obtain basic data; S2. Based on the basic data, construct and calibrate the hydrodynamic model to obtain the daily spatiotemporal distribution of water level for a hydrological year; S3. Based on the simulation results, extract the water level change process data in the flooded area to form a water level-time curve; S4. Normalize the water level-time curve; S5. Calculate the characteristic values ​​of water level changes based on the normalized curve; S6. Perform cluster analysis on the water level change characteristic values ​​to obtain the initial classification results; S7. Display the initial classification results according to spatial coordinates to identify hydrological connectivity patches; S8. Based on the principal components of the water level sequence and spatial distance of each patch, construct a comprehensive distance matrix and perform hierarchical clustering to delineate hydrological connectivity zones.

[0006] The invention is further characterized by: The basic data in S1 include daily inflow, daily precipitation, daily evaporation, digital elevation model (DEM) data, land use cover (LULC) data, outlet water level-discharge curves, and water level monitoring data collected from 3 to 4 water level monitoring stations deployed in the study area for one hydrological year.

[0007] S2 specifically refers to: S2.1 Based on DEM data, the study area is divided into several unstructured grids, the coordinates of the center point of each grid are obtained, and the surface elevation is interpolated to each grid, with the grid as the calculation unit; S2.2. Construct a hydrodynamic model based on the two-dimensional planar shallow water equation, with the daily inflow as the upper boundary and the water level-discharge curve as the lower boundary. Input the daily precipitation and evaporation, and estimate the Manning roughness coefficient based on LULC data. The coefficient is taken as 0.6~0.8 for reed marshland areas and 0.03~0.04 for river and lake areas. Simulate the spatiotemporal distribution of daily water level for one hydrological year. S2.3. The hydrodynamic model is calibrated and verified using water level monitoring data. The verification indicators meet the requirements of Nash efficiency coefficient NSE≥0.95 and root mean square error RMSE≤0.05 m.

[0008] The specific method for extracting water level change data within the flooded area in S3 is as follows: daily water level data of the center points of all calculation units in the flooded area are extracted from the hydrodynamic model simulation results using computer language to form a water level-time change curve dataset, while recording the spatial coordinates of each calculation unit. In S4, the normalization process uses the Max-min standardization method to map the original water level data to the [0,1] interval through a linear transformation, so that the maximum value of the original data is mapped to 1 and the minimum value is mapped to 0, thereby eliminating the difference in the absolute magnitude of the water level and shifting the focus of the analysis to the curve shape characteristics.

[0009] S5 specifically involves: randomly selecting the normalized water level-time curve of a computational unit in the flooded area as the baseline curve; calculating the correlation coefficient between the normalized curves of the remaining computational units and the baseline curve; using the correlation coefficient as the water level change characteristic value of the corresponding grid; where the characteristic value of the baseline curve itself is 1; to evaluate the impact of random selection of the baseline curve, multiple different random baseline curves are selected for repeated calculations, and the adjusted RAND index (ARI) is used to evaluate the consistency of different clustering results. When the mean ARI is greater than 0.60, the partitioning results are considered robust.

[0010] S6 specifically refers to: S6.1. The K-means algorithm is adopted, with the water level change characteristic value of each calculation unit as input and the clustering objective function as the sum of squared errors (SSE). The calculation is performed according to equation (1). (1); in, C i For the first i One cluster; c i For the first i The centroids of each cluster; dist The Euclidean distance between two samples in the dataset; SSE Kmeans This is the sum of squared errors for all samples in the dataset; S6.2 The initial number of clusters is set to 5 to 10 based on the size of the study area to form the initial classification results.

[0011] In S7, after the initial classification results are displayed in a two-dimensional space according to coordinates, the points of the same category are automatically separated because they are not connected in space, and the number of hydrological connectivity patches formed is greater than or equal to the number of cluster categories in S6.

[0012] In S8, the principal components of the water level sequence were extracted using global principal component analysis (PCA). The first three principal components, PC1, PC2, and PC3, were used to represent the overall water level, seasonal changes, and short-term fluctuations, respectively. The mean score of all grids within each patch on PC1 to PC3 was used as the hydrological feature vector of that patch, and the standard deviation of this score was used to evaluate the effectiveness of the initial clustering classification in S6. In S8, the comprehensive distance matrix is ​​constructed by weighted summation of hydrological distance and spatial distance between patches. The hydrological distance is the Mahalanobis distance between patch feature vectors, and the spatial distance is the minimum boundary distance between the convex hulls of each patch. After normalizing the two types of distances respectively, the comprehensive distance matrix is ​​constructed by weighted linear combination according to the weight coefficients.

[0013] The formula for constructing the integrated distance matrix in S8 is shown in equation (2); (2); in, D ij PCA The normalized Mahalanobis distance. D ij spatial The minimum boundary distance of the normalized convex hull, weighted by coefficients. α and β satisfy α + β =1, default value α = β =0.5; when α When the value is between 0.3 and 0.7, the adjusted RAND index (ARI) of the partitioning results is ≥0.60, indicating high stability.

[0014] In S8, hierarchical clustering uses the Ward minimum variance method, which uses the comprehensive distance matrix as input for clustering and automatically determines the optimal number of partitions by maximizing the silhouette coefficient, forming the final hydrological connectivity partitions without the need for manual intervention in the partition boundaries.

[0015] The beneficial effects of this invention are: The wetland hydrological connectivity zoning method provided by this invention determines the internal hydrological connectivity of wetlands by analyzing the temporal similarity and spatial continuity of water level changes in each grid, and provides specific zoning boundaries. This method is simple to operate, can provide specific spatial boundaries for zoning, and can be applied to various marsh wetlands, providing technical support for the optimized deployment of wetland ecological monitoring stations, the identification of hydrological connectivity barriers, and the management of ecological water replenishment. Attached Figure Description

[0016] Figure 1 This is a map showing the location of the Zhalong Wetland research area and the distribution of its outlets and water level monitoring points, as shown in Embodiment 6 of the present invention. Figure 2 This is a schematic diagram of the unstructured mesh generation of the hydrodynamic model of the study area in Embodiment 6 of the present invention; Figure 3 This is a schematic diagram of the first K-means clustering result in Embodiment 6 of the present invention; Figure 4 This is a spatial distribution map of hydrological connectivity patches in Embodiment 6 of the present invention; Figure 5 This is a flowchart of the PCA-spatial constraint hierarchical clustering process in Embodiment 6 of the present invention; Figure 6 This is a schematic diagram of the second clustering result in Embodiment 6 of the present invention; Figure 7 This is a schematic diagram of the water level change process in hydrological connectivity zone 1 in embodiment 6 of the present invention; Figure 8 This is a schematic diagram of the water level change process in hydrological connectivity zone 2 in embodiment 6 of the present invention; Figure 9 This is a schematic diagram of the water level change process in hydrological connectivity zones 3 and 4 in Embodiment 6 of the present invention; Figure 10 This is a schematic diagram of the water level change process in hydrological connectivity zone 5 in embodiment 6 of the present invention; Figure 11 This is a schematic diagram of the water level change process in hydrological connectivity zone 6 in embodiment 6 of the present invention; Figure 12 This is a schematic diagram of the water level change process in hydrological connectivity zone 7 in embodiment 6 of the present invention. Detailed Implementation

[0017] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0018] The wetland hydrological connectivity zoning method provided by this invention includes the following steps: S1. Obtain basic data; The basic data include daily inflow, daily precipitation, daily evaporation, digital elevation model (DEM) data, land use cover (LULC) data, outlet water level-discharge curves, and water level monitoring data collected from 3 to 4 water level monitoring stations deployed in the study area for one hydrological year. S2. Based on the basic data, construct and calibrate the hydrodynamic model to obtain the daily spatiotemporal distribution of water level for a hydrological year; Specifically: S2.1 Based on DEM data, the study area is divided into several unstructured grids, the coordinates of the center point of each grid are obtained, and the surface elevation is interpolated to each grid, with the grid as the calculation unit; S2.2. Construct a hydrodynamic model based on the two-dimensional planar shallow water equation, with the daily inflow as the upper boundary and the water level-discharge curve as the lower boundary. Input the daily precipitation and evaporation, and estimate the Manning roughness coefficient based on LULC data. The coefficient is taken as 0.6~0.8 for reed marshland areas and 0.03~0.04 for river and lake areas. Simulate the spatiotemporal distribution of daily water level for one hydrological year. S2.3. The hydrodynamic model is calibrated and verified using water level monitoring data. The verification indicators meet the requirements of Nash efficiency coefficient NSE≥0.95 and root mean square error RMSE≤0.05 m. S3. Based on the simulation results, extract the water level change process data in the flooded area to form a water level-time curve; The specific method for extracting water level change data within the flooded area is as follows: daily water level data of the center points of all calculation units in the flooded area are extracted from the hydrodynamic model simulation results using computer language to form a water level-time change curve dataset, while recording the spatial coordinates of each calculation unit. S4. Normalize the water level-time curve; The normalization process uses the Max-min standardization method to map the original water level data to the [0, 1] interval through linear transformation, so that the maximum value of the original data is mapped to 1 and the minimum value is mapped to 0, thereby eliminating the difference in the absolute magnitude of water level and shifting the focus of analysis to the curve shape characteristics. S5. Calculate the characteristic values ​​of water level changes based on the normalized curve; Specifically, the normalized water level-time curve of a computational unit is randomly selected from the flooded area as the baseline curve. The correlation coefficient between the normalized curves of the remaining computational units and the baseline curve is calculated. The correlation coefficient is used as the water level change characteristic value of the corresponding grid, where the characteristic value of the baseline curve itself is 1. To evaluate the impact of random selection of the baseline curve, multiple different random baseline curves are selected and the calculation is repeated. The adjusted RAND index (ARI) is used to evaluate the consistency of different clustering results. When the mean ARI is greater than 0.60, the partitioning results are considered robust. S6. Perform cluster analysis on the water level change characteristic values ​​to obtain the initial classification results; Specifically: S6.1. The K-means algorithm is adopted, with the water level change characteristic value of each calculation unit as input and the clustering objective function as the sum of squared errors (SSE). The calculation is performed according to equation (1). (1); in, C i For the first i One cluster; c i For the first i The centroids of each cluster; dist The Euclidean distance between two samples in the dataset; SSE Kmeans This is the sum of squared errors for all samples in the dataset; S6.2 The initial number of clusters is set to 5 to 10 based on the size of the study area to form the initial classification results; S7. Display the initial classification results according to spatial coordinates to identify hydrological connectivity patches; After the initial classification results are displayed in a two-dimensional space according to coordinates, the points of the same category are automatically separated because they are not connected in space, and the number of hydrological connectivity patches formed is greater than or equal to the number of cluster categories in S6. S8. Based on the principal components and spatial distances of the water level sequences of each patch, construct a comprehensive distance matrix and perform hierarchical clustering to delineate hydrological connectivity zones. The principal components of the water level sequence were extracted using global principal component analysis (PCA). The first three principal components, PC1, PC2, and PC3, were used to represent the overall water level, seasonal changes, and short-term fluctuations, respectively. The mean score of all grids within each patch on PC1 to PC3 was used as the hydrological feature vector of that patch, and the standard deviation of this score was used to evaluate the effectiveness of the initial clustering classification in S6. In S8, the comprehensive distance matrix is ​​constructed by weighted summation of hydrological distance and spatial distance between patches. The hydrological distance is the Mahalanobis distance between patch feature vectors, and the spatial distance is the minimum boundary distance between the convex hulls of each patch. After normalizing the two types of distances respectively, the comprehensive distance matrix is ​​constructed by weighted linear combination according to the weight coefficients. The formula for constructing the comprehensive distance matrix is ​​shown in equation (2); (2); in, D ij PCA The normalized Mahalanobis distance. D ij spatial The minimum boundary distance of the normalized convex hull, weighted by coefficients. α and β satisfy α + β =1, default value α = β =0.5; when α When the value is between 0.3 and 0.7, the adjusted RAND index (ARI) of the partitioning results is ≥0.60, indicating high stability. Hierarchical clustering uses Ward's minimum variance method, with the comprehensive distance matrix as input for clustering, and automatically determines the optimal number of partitions by maximizing the silhouette coefficient, forming the final hydrological connectivity partitions without the need for manual intervention in the partition boundaries.

[0019] Example 1 The wetland hydrological connectivity zoning method proposed in this embodiment includes the following steps: S1. Obtain basic data; S2. Based on the basic data, construct and calibrate the hydrodynamic model to obtain the daily spatiotemporal distribution of water level for a hydrological year; S3. Based on the simulation results, extract the water level change process data in the flooded area to form a water level-time curve; S4. Normalize the water level-time curve; S5. Calculate the characteristic values ​​of water level changes based on the normalized curve; S6. Perform cluster analysis on the water level change characteristic values ​​to obtain the initial classification results; S7. Display the initial classification results according to spatial coordinates to identify hydrological connectivity patches; S8. Based on the principal components of the water level sequence and spatial distance of each patch, construct a comprehensive distance matrix and perform hierarchical clustering to delineate hydrological connectivity zones.

[0020] Example 2 The wetland hydrological connectivity zoning method proposed in this embodiment includes the following steps: S1. Obtain basic data; The basic data include daily inflow, daily precipitation, daily evaporation, digital elevation model (DEM) data, land use cover (LULC) data, outlet water level-discharge curves, and water level monitoring data collected from 3 to 4 water level monitoring stations deployed in the study area for one hydrological year. S2. Based on the basic data, construct and calibrate the hydrodynamic model to obtain the daily spatiotemporal distribution of water level for a hydrological year; Specifically: S2.1 Based on DEM data, the study area is divided into several unstructured grids, the coordinates of the center point of each grid are obtained, and the surface elevation is interpolated to each grid, with the grid as the calculation unit; S2.2. Construct a hydrodynamic model based on the two-dimensional planar shallow water equation, with the daily inflow as the upper boundary and the water level-discharge curve as the lower boundary. Input the daily precipitation and evaporation, and estimate the Manning roughness coefficient based on LULC data. The coefficient is taken as 0.6~0.8 for reed marshland areas and 0.03~0.04 for river and lake areas. Simulate the spatiotemporal distribution of daily water level for one hydrological year. S2.3. The hydrodynamic model is calibrated and verified using water level monitoring data. The verification indicators meet the requirements of Nash efficiency coefficient NSE≥0.95 and root mean square error RMSE≤0.05 m. S3. Based on the simulation results, extract the water level change process data in the flooded area to form a water level-time curve; The specific method for extracting water level change data within the flooded area is as follows: daily water level data of the center points of all calculation units in the flooded area are extracted from the hydrodynamic model simulation results using computer language to form a water level-time change curve dataset, while recording the spatial coordinates of each calculation unit. S4. Normalize the water level-time curve; The normalization process uses the Max-min standardization method to map the original water level data to the [0, 1] interval through linear transformation, so that the maximum value of the original data is mapped to 1 and the minimum value is mapped to 0, thereby eliminating the difference in the absolute magnitude of water level and shifting the focus of analysis to the curve shape characteristics. S5. Calculate the characteristic values ​​of water level changes based on the normalized curve; S6. Perform cluster analysis on the water level change characteristic values ​​to obtain the initial classification results; S7. Display the initial classification results according to spatial coordinates to identify hydrological connectivity patches; S8. Based on the principal components of the water level sequence and spatial distance of each patch, construct a comprehensive distance matrix and perform hierarchical clustering to delineate hydrological connectivity zones.

[0021] Example 3 The wetland hydrological connectivity zoning method proposed in this embodiment includes the following steps: S1. Obtain basic data; The basic data include daily inflow, daily precipitation, daily evaporation, digital elevation model (DEM) data, land use cover (LULC) data, outlet water level-discharge curves, and water level monitoring data collected from 3 to 4 water level monitoring stations deployed in the study area for one hydrological year. S2. Based on the basic data, construct and calibrate the hydrodynamic model to obtain the daily spatiotemporal distribution of water level for a hydrological year; Specifically: S2.1 Based on DEM data, the study area is divided into several unstructured grids, the coordinates of the center point of each grid are obtained, and the surface elevation is interpolated to each grid, with the grid as the calculation unit; S2.2. Construct a hydrodynamic model based on the two-dimensional planar shallow water equation, with the daily inflow as the upper boundary and the water level-discharge curve as the lower boundary. Input the daily precipitation and evaporation, and estimate the Manning roughness coefficient based on LULC data. The coefficient is taken as 0.6~0.8 for reed marshland areas and 0.03~0.04 for river and lake areas. Simulate the spatiotemporal distribution of daily water level for one hydrological year. S2.3. The hydrodynamic model is calibrated and verified using water level monitoring data. The verification indicators meet the requirements of Nash efficiency coefficient NSE≥0.95 and root mean square error RMSE≤0.05 m. S3. Based on the simulation results, extract the water level change process data in the flooded area to form a water level-time curve; The specific method for extracting water level change data within the flooded area is as follows: daily water level data of the center points of all calculation units in the flooded area are extracted from the hydrodynamic model simulation results using computer language to form a water level-time change curve dataset, while recording the spatial coordinates of each calculation unit. S4. Normalize the water level-time curve; The normalization process uses the Max-min standardization method to map the original water level data to the [0, 1] interval through linear transformation, so that the maximum value of the original data is mapped to 1 and the minimum value is mapped to 0, thereby eliminating the difference in the absolute magnitude of water level and shifting the focus of analysis to the curve shape characteristics. S5. Calculate the characteristic values ​​of water level changes based on the normalized curve; Specifically, the normalized water level-time curve of a computational unit is randomly selected from the flooded area as the baseline curve. The correlation coefficient between the normalized curves of the remaining computational units and the baseline curve is calculated. The correlation coefficient is used as the water level change characteristic value of the corresponding grid, where the characteristic value of the baseline curve itself is 1. To evaluate the impact of random selection of the baseline curve, multiple different random baseline curves are selected and the calculation is repeated. The adjusted RAND index (ARI) is used to evaluate the consistency of different clustering results. When the mean ARI is greater than 0.60, the partitioning results are considered robust. S6. Perform cluster analysis on the water level change characteristic values ​​to obtain the initial classification results; S7. Display the initial classification results according to spatial coordinates to identify hydrological connectivity patches; S8. Based on the principal components of the water level sequence and spatial distance of each patch, construct a comprehensive distance matrix and perform hierarchical clustering to delineate hydrological connectivity zones.

[0022] Example 4 The wetland hydrological connectivity zoning method proposed in this embodiment includes the following steps: S1. Obtain basic data; The basic data include daily inflow, daily precipitation, daily evaporation, digital elevation model (DEM) data, land use cover (LULC) data, outlet water level-discharge curves, and water level monitoring data collected from 3 to 4 water level monitoring stations deployed in the study area for one hydrological year. S2. Based on the basic data, construct and calibrate the hydrodynamic model to obtain the daily spatiotemporal distribution of water level for a hydrological year; Specifically: S2.1 Based on DEM data, the study area is divided into several unstructured grids, the coordinates of the center point of each grid are obtained, and the surface elevation is interpolated to each grid, with the grid as the calculation unit; S2.2. Construct a hydrodynamic model based on the two-dimensional planar shallow water equation, with the daily inflow as the upper boundary and the water level-discharge curve as the lower boundary. Input the daily precipitation and evaporation, and estimate the Manning roughness coefficient based on LULC data. The coefficient is taken as 0.6~0.8 for reed marshland areas and 0.03~0.04 for river and lake areas. Simulate the spatiotemporal distribution of daily water level for one hydrological year. S2.3. The hydrodynamic model is calibrated and verified using water level monitoring data. The verification indicators meet the requirements of Nash efficiency coefficient NSE≥0.95 and root mean square error RMSE≤0.05 m. S3. Based on the simulation results, extract the water level change process data in the flooded area to form a water level-time curve; The specific method for extracting water level change data within the flooded area is as follows: daily water level data of the center points of all calculation units in the flooded area are extracted from the hydrodynamic model simulation results using computer language to form a water level-time change curve dataset, while recording the spatial coordinates of each calculation unit. S4. Normalize the water level-time curve; The normalization process uses the Max-min standardization method to map the original water level data to the [0, 1] interval through linear transformation, so that the maximum value of the original data is mapped to 1 and the minimum value is mapped to 0, thereby eliminating the difference in the absolute magnitude of water level and shifting the focus of analysis to the curve shape characteristics. S5. Calculate the characteristic values ​​of water level changes based on the normalized curve; Specifically, the normalized water level-time curve of a computational unit is randomly selected from the flooded area as the baseline curve. The correlation coefficient between the normalized curves of the remaining computational units and the baseline curve is calculated. The correlation coefficient is used as the water level change characteristic value of the corresponding grid, where the characteristic value of the baseline curve itself is 1. To evaluate the impact of random selection of the baseline curve, multiple different random baseline curves are selected and the calculation is repeated. The adjusted RAND index (ARI) is used to evaluate the consistency of different clustering results. When the mean ARI is greater than 0.60, the partitioning results are considered robust. S6. Perform cluster analysis on the water level change characteristic values ​​to obtain the initial classification results; Specifically: S6.1. The K-means algorithm is adopted, with the water level change characteristic value of each calculation unit as input and the clustering objective function as the sum of squared errors (SSE). The calculation is performed according to equation (1). (1); in, C i For the first i One cluster; c i For the first i The centroids of each cluster; dist The Euclidean distance between two samples in the dataset; SSE Kmeans This is the sum of squared errors for all samples in the dataset; S6.2 The initial number of clusters is set to 5 to 10 based on the size of the study area to form the initial classification results; S7. Display the initial classification results according to spatial coordinates to identify hydrological connectivity patches; After the initial classification results are displayed in a two-dimensional space according to coordinates, the points of the same category are automatically separated because they are not connected in space, and the number of hydrological connectivity patches formed is greater than or equal to the number of cluster categories in S6. S8. Based on the principal components of the water level sequence and spatial distance of each patch, construct a comprehensive distance matrix and perform hierarchical clustering to delineate hydrological connectivity zones.

[0023] Example 5 The wetland hydrological connectivity zoning method proposed in this embodiment includes the following steps: S1. Obtain basic data; The basic data include daily inflow, daily precipitation, daily evaporation, digital elevation model (DEM) data, land use cover (LULC) data, outlet water level-discharge curves, and water level monitoring data collected from 3 to 4 water level monitoring stations deployed in the study area for one hydrological year. S2. Based on the basic data, construct and calibrate the hydrodynamic model to obtain the daily spatiotemporal distribution of water level for a hydrological year; Specifically: S2.1 Based on DEM data, the study area is divided into several unstructured grids, the coordinates of the center point of each grid are obtained, and the surface elevation is interpolated to each grid, with the grid as the calculation unit; S2.2. Construct a hydrodynamic model based on the two-dimensional planar shallow water equation, with the daily inflow as the upper boundary and the water level-discharge curve as the lower boundary. Input the daily precipitation and evaporation, and estimate the Manning roughness coefficient based on LULC data. The coefficient is taken as 0.6~0.8 for reed marshland areas and 0.03~0.04 for river and lake areas. Simulate the spatiotemporal distribution of daily water level for one hydrological year. S2.3. The hydrodynamic model is calibrated and verified using water level monitoring data. The verification indicators meet the requirements of Nash efficiency coefficient NSE≥0.95 and root mean square error RMSE≤0.05 m. S3. Based on the simulation results, extract the water level change process data in the flooded area to form a water level-time curve; The specific method for extracting water level change data within the flooded area is as follows: daily water level data of the center points of all calculation units in the flooded area are extracted from the hydrodynamic model simulation results using computer language to form a water level-time change curve dataset, while recording the spatial coordinates of each calculation unit. S4. Normalize the water level-time curve; The normalization process uses the Max-min standardization method to map the original water level data to the [0, 1] interval through linear transformation, so that the maximum value of the original data is mapped to 1 and the minimum value is mapped to 0, thereby eliminating the difference in the absolute magnitude of water level and shifting the focus of analysis to the curve shape characteristics. S5. Calculate the characteristic values ​​of water level changes based on the normalized curve; Specifically, the normalized water level-time curve of a computational unit is randomly selected from the flooded area as the baseline curve. The correlation coefficient between the normalized curves of the remaining computational units and the baseline curve is calculated. The correlation coefficient is used as the water level change characteristic value of the corresponding grid, where the characteristic value of the baseline curve itself is 1. To evaluate the impact of random selection of the baseline curve, multiple different random baseline curves are selected and the calculation is repeated. The adjusted RAND index (ARI) is used to evaluate the consistency of different clustering results. When the mean ARI is greater than 0.60, the partitioning results are considered robust. S6. Perform cluster analysis on the water level change characteristic values ​​to obtain the initial classification results; Specifically: S6.1. The K-means algorithm is adopted, with the water level change characteristic value of each calculation unit as input and the clustering objective function as the sum of squared errors (SSE). The calculation is performed according to equation (1). (1); in, C i For the first i One cluster; c i For the first i The centroids of each cluster; dist The Euclidean distance between two samples in the dataset; SSE Kmeans This is the sum of squared errors for all samples in the dataset; S6.2 The initial number of clusters is set to 5 to 10 based on the size of the study area to form the initial classification results; S7. Display the initial classification results according to spatial coordinates to identify hydrological connectivity patches; After the initial classification results are displayed in a two-dimensional space according to coordinates, the points of the same category are automatically separated because they are not connected in space, and the number of hydrological connectivity patches formed is greater than or equal to the number of cluster categories in S6. S8. Based on the principal components and spatial distances of the water level sequences of each patch, construct a comprehensive distance matrix and perform hierarchical clustering to delineate hydrological connectivity zones. The principal components of the water level sequence were extracted using global principal component analysis (PCA). The first three principal components, PC1, PC2, and PC3, were used to represent the overall water level, seasonal changes, and short-term fluctuations, respectively. The mean score of all grids within each patch on PC1 to PC3 was used as the hydrological feature vector of that patch, and the standard deviation of this score was used to evaluate the effectiveness of the initial clustering classification in S6. In S8, the comprehensive distance matrix is ​​constructed by weighted summation of hydrological distance and spatial distance between patches. The hydrological distance is the Mahalanobis distance between patch feature vectors, and the spatial distance is the minimum boundary distance between the convex hulls of each patch. After normalizing the two types of distances respectively, the comprehensive distance matrix is ​​constructed by weighted linear combination according to the weight coefficients. The formula for constructing the comprehensive distance matrix is ​​shown in equation (2); (2); in, D ij PCA The normalized Mahalanobis distance. D ij spatial The minimum boundary distance of the normalized convex hull, weighted by coefficients. α and β satisfy α + β =1, default value α = β =0.5; when α When the value is between 0.3 and 0.7, the adjusted RAND index (ARI) of the partitioning results is ≥0.60, indicating high stability. Hierarchical clustering uses Ward's minimum variance method, with the comprehensive distance matrix as input for clustering, and automatically determines the optimal number of partitions by maximizing the silhouette coefficient, forming the final hydrological connectivity partitions without the need for manual intervention in the partition boundaries.

[0024] Example 6 The wetland hydrological connectivity zoning method proposed in this embodiment takes the core area of ​​Zhalong Wetland in Heilongjiang Province as an example. The study area is bordered by the G10 Suiman Expressway to the north, the Binzhou Railway to the south, the Eighth Branch of the Central Yin Nenjiang Water Diversion Project to the west, and the Ninth Branch of the Central Yin Nenjiang Water Diversion Project to the east, forming a relatively independent hydrological unit. The method includes the following steps: First, daily inflow, daily precipitation, and daily evaporation data for one hydrological year were collected in the study area; 30×30 m resolution DEM data were obtained from the Zhalong Nature Reserve Administration; 30×30 m resolution LULC data were obtained; the water level-discharge relationship curve at the outlet of the study area was obtained; and four water level monitoring points were set up in the study area, such as... Figure 1 As shown, water level monitoring data was collected; A hydrodynamic model of the study area was constructed based on two-dimensional planar shallow water equations. The model divides the study area into 158,085 unstructured grids according to digital terrain elevation, such as... Figure 2 As shown, the upper boundary of the model is the daily runoff of water in the study area, and the lower boundary is the water level-discharge relationship curve of the four outlets. The model is input with precipitation and evaporation data, and the model is calibrated using water level monitoring data at the four points. The simulation yielded the spatial distribution of daily water level in the study area over one hydrological year. The water level-time variation curves and spatial coordinates of 5032 points in the inundated area were extracted using a computer language program.

[0025] Max-min standardization was used to linearly transform the curve data, mapping the original data to the range of 0 to 1, thus normalizing the curves. A water level-time curve at a randomly selected point was chosen as the baseline curve. The water level-time curves of the other 5031 points were fitted to the baseline curve, and the correlation coefficient was calculated as the characteristic value of water level change at that point. The characteristic value of water level change for the baseline curve was 1. To ensure the stability of the results, repeated calculations were performed using different random baseline curves, and the consistency of the clustering results under different baseline curves was compared to verify the stability of the partitioning results. The water level change characteristic values ​​of each location were clustered using the K-means algorithm. The locations were divided into 6 categories based on their water level change characteristic values, and then spatially distributed according to their coordinates. Locations of the same category that were not spatially connected were automatically separated, ultimately forming 24 hydrological connectivity patches. Figure 3 As shown, the distribution of its water level change characteristic values ​​is as follows: Figure 4 As shown; Excessive zoning is detrimental to wetland administrative management. The number of wetland hydrological connectivity zones should ideally not exceed 10. The second clustering performs global PCA on the standardized daily water level sequences of all grids, extracting PC1 (overall water level), PC2 (seasonal variation), and PC3 (short-term fluctuations). The mean PC score of each grid within a patch is calculated as the patch feature. The Mahalanobis distance (hydrological distance) and the minimum boundary distance of the convex hull (spatial distance) between patches are calculated, and then weighted linearly combined (… α = β =0.5) Construct a comprehensive distance matrix, and use the Ward hierarchical clustering method and silhouette coefficient to automatically determine the optimal number of partitions. The research method is as follows: Figure 5 As shown; Ultimately, seven hydrological connectivity zones were formed, such as Figure 6 As shown, the distribution of water level change characteristic values ​​within each zone is as follows: Figure 7-12As shown, the water level change characteristic values ​​of points within partitions 3 and 4 exhibit significant overlap, but are not spatially connected; therefore, they are considered two partitions and do not require merging. At this point, the water level change characteristic values ​​of points within each partition are similar, the water level change processes are similar, and they are spatially connected, indicating strong hydrological connectivity among points within each partition. Ultimately, the Zhalong Wetland study area is divided into 7 hydrological connectivity partitions.

Claims

1. A wetland hydrological connectivity zoning method, characterized in that, include: S1. Obtain basic data; S2. Based on the basic data, construct and calibrate the hydrodynamic model to obtain the daily spatiotemporal distribution of water level for a hydrological year; S3. Based on the simulation results, extract the water level change process data in the flooded area to form a water level-time curve; S4. Normalize the water level-time curve; S5. Calculate the characteristic values ​​of water level changes based on the normalized curve; S6. Perform cluster analysis on the water level change characteristic values ​​to obtain the initial classification results; S7. Display the initial classification results according to spatial coordinates to identify hydrological connectivity patches; S8. Based on the principal components of the water level sequence and spatial distance of each patch, construct a comprehensive distance matrix and perform hierarchical clustering to delineate hydrological connectivity zones.

2. The wetland hydrological connectivity zoning method according to claim 1, characterized in that, The basic data mentioned in S1 includes daily inflow, daily precipitation, daily evaporation, digital elevation model (DEM) data, land use cover (LULC) data, outlet water level-discharge curves, and water level monitoring data collected from 3 to 4 water level monitoring stations deployed in the study area for one hydrological year.

3. The wetland hydrological connectivity zoning method according to claim 1, characterized in that, Specifically, S2 is: S2.1 Based on DEM data, the study area is divided into several unstructured grids, the coordinates of the center point of each grid are obtained, and the surface elevation is interpolated to each grid, with the grid as the calculation unit; S2.

2. Construct a hydrodynamic model based on the two-dimensional planar shallow water equation, with the daily inflow as the upper boundary and the water level-discharge curve as the lower boundary. Input the daily precipitation and evaporation, and estimate the Manning roughness coefficient based on LULC data. The coefficient is taken as 0.6~0.8 for reed marshland areas and 0.03~0.04 for river and lake areas. Simulate the spatiotemporal distribution of daily water level for one hydrological year. S2.

3. The hydrodynamic model is calibrated and verified using water level monitoring data. The verification indicators meet the requirements of Nash efficiency coefficient NSE≥0.95 and root mean square error RMSE≤0.05 m.

4. The wetland hydrological connectivity zoning method according to claim 1, characterized in that, The specific method for extracting water level change data in the flooded area as described in S3 is as follows: extract the daily water level data of the center points of all calculation units in the flooded area from the hydrodynamic model simulation results using computer language, form a water level-time change curve dataset, and record the spatial coordinates of each calculation unit at the same time. The normalization process described in S4 uses the Max-min standardization method to map the original water level data to the [0,1] interval through a linear transformation, so that the maximum value of the original data is mapped to 1 and the minimum value is mapped to 0, thereby eliminating the difference in the absolute magnitude of the water level and shifting the focus of the analysis to the curve shape characteristics.

5. The wetland hydrological connectivity zoning method according to claim 1, characterized in that, S5 specifically involves: randomly selecting the normalized water level-time curve of a computational unit from the flooded area as the baseline curve; calculating the correlation coefficient between the normalized curves of the remaining computational units and the baseline curve; using the correlation coefficient as the water level change characteristic value of the corresponding grid; where the characteristic value of the baseline curve itself is 1; to evaluate the impact of random selection of the baseline curve, multiple different random baseline curves are selected for repeated calculations, and the consistency of different clustering results is evaluated using the Adjusted Rand Index (ARI); when the mean ARI is greater than 0.60, the partitioning results are deemed robust.

6. The wetland hydrological connectivity zoning method according to claim 1, characterized in that, Specifically, S6 is: S6.

1. The K-means algorithm is adopted, with the water level change characteristic value of each calculation unit as input and the clustering objective function as the sum of squared errors (SSE). The calculation is performed according to equation (1). （1）； in, C i For the first i One cluster; c i For the first i The centroids of each cluster; dist The Euclidean distance between two samples in the dataset; SSE Kmeans This is the sum of squared errors for all samples in the dataset; S6.2 The initial number of clusters is set to 5 to 10 based on the size of the study area to form the initial classification results.

7. The wetland hydrological connectivity zoning method according to claim 1, characterized in that, In step S7, after the initial classification results are displayed in a two-dimensional space according to coordinates, the points of the same category are automatically separated because they are not connected in space, and the number of hydrological connectivity patches formed is greater than or equal to the number of cluster categories in step S6.

8. The wetland hydrological connectivity zoning method according to claim 1, characterized in that, The water level sequence principal components described in S8 are extracted by global principal component analysis (PCA) to extract the first three principal components, PC1, PC2, and PC3, which respectively represent the overall water level level, seasonal changes, and short-term fluctuations. The mean score of all grids in each patch on PC1 to PC3 is used as the hydrological feature vector of that patch, and the standard deviation of this score is used to evaluate the effectiveness of the initial clustering classification in S6. The comprehensive distance matrix described in S8 is constructed by weighted summation of hydrological distance and spatial distance between patches; the hydrological distance is the Mahalanobis distance between patch feature vectors, and the spatial distance is the minimum boundary distance between the convex hulls of each patch; after normalizing the two types of distances respectively, the comprehensive distance matrix is ​​constructed by weighted linear combination according to the weight coefficients.

9. The wetland hydrological connectivity zoning method according to claim 8, characterized in that, The formula for constructing the comprehensive distance matrix described in S8 is shown in equation (2); （2）； in, D ij PCA The normalized Mahalanobis distance. D ij spatial The minimum boundary distance of the normalized convex hull, weighted by coefficients. α and β satisfy α + β =1, default value α = β =0.5; when α When the value is between 0.3 and 0.7, the adjusted RAND index (ARI) of the partitioning results is ≥0.60, indicating high stability.

10. The wetland hydrological connectivity zoning method according to claim 8, characterized in that, The hierarchical clustering described in S8 uses the Ward minimum variance method, which uses the comprehensive distance matrix as input for clustering and automatically determines the optimal number of partitions by maximizing the silhouette coefficient, forming the final hydrological connectivity partitions without the need for manual intervention in the partition boundaries.

Images