A method and system for constructing an urban bird ecological network

Abstract

This invention discloses a method and system for constructing urban bird ecological networks, belonging to the field of data analysis technology. The method includes: identifying potential suitable habitats using a maximum entropy model based on preprocessed bird observation records and two-dimensional environmental factor data; assessing habitat quality using land use type raster data based on an InVEST model; determining candidate ecological source areas based on bird potential suitable habitat information and habitat quality assessment results, and performing connectivity analysis to determine target ecological source areas; determining a trained CatBoost model based on preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data to determine ecological resistance surfaces; and determining minimum cost paths based on target ecological source areas and ecological resistance surfaces to determine bird ecological networks and ecological pinch points using current density values. This invention achieves refined identification and analysis of bird ecological networks in high-density urban environments.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and in particular to a method and system for constructing an urban bird ecological network. Background Technology

[0002] Existing urban ecological network research primarily focuses on green spaces or blue-green spaces, typically identifying ecological source areas and corridors based on land use types or landscape pattern indicators. This approach has been widely applied in areas such as ecological security pattern construction and national spatial planning. However, research specifically addressing bird ecological networks—a group highly dependent on spatial structure and environmental heterogeneity—remains relatively limited. Current studies largely concentrate on habitat suitability assessments or correlation analyses between landscape patterns and bird diversity, lacking a detailed characterization of bird ecological network structures. Especially in high-density urban settings, bird ecological network identification has yet to develop a mature, systematic, and scalable technical framework.

[0003] Currently, ecological network construction typically follows a technical framework of ecological source area identification, ecological resistance surface construction, and ecological corridor extraction. However, its application to bird ecological network construction still has significant shortcomings. First, ecological source area identification, as a fundamental step in subsequent network analysis, directly determines the reliability of the ecological network construction results. However, existing bird ecological source area identification methods often rely on single models or evaluation indicators, failing to comprehensively reflect actual bird distribution characteristics, habitat quality differences, and landscape structural integrity, leading to functional biases in the identification results. Second, existing ecological resistance surface construction is mainly based on two-dimensional surface factors, neglecting the impact of urban three-dimensional spatial structure on bird migration. Third, the construction of ecological resistance surfaces often employs empirical weighting or linear superposition methods, which are highly subjective and lack interpretability, making it difficult to objectively characterize the comprehensive impact of multiple factors on bird migration behavior, thus limiting the scientific rigor and stability of the ecological network construction results. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art. This invention provides a method and system for constructing urban bird ecological networks, which improves the reliability of bird ecological network construction and realizes the refined identification and analysis of bird ecological networks in high-density urban environments.

[0005] To address the aforementioned technical problems, this invention provides a method for constructing an urban bird ecological network, the method comprising: Bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data of the target area are acquired, and the bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data are preprocessed to obtain preprocessed bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data. Based on preprocessed bird observation records and two-dimensional environmental factor data, the maximum entropy model was used to identify potential suitable habitats and obtain information on potential suitable habitats for birds. Land use type raster data of the target area is acquired, and habitat quality is assessed using the InVEST model, which is a comprehensive assessment of ecosystem services and trade-offs, to obtain habitat quality assessment results. Candidate ecological sources are determined based on the information on potential suitable habitats for birds and the results of habitat quality assessment. Connectivity analysis is then performed on the candidate ecological sources to obtain an overall connectivity index and a potential connectivity index. Target ecological sources are then determined based on the overall connectivity index and the potential connectivity index. An ecological resistance index system is constructed based on preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data. A resistance sample set is generated based on the ecological resistance index system. The CatBoost model is trained based on the resistance sample set to obtain a trained CatBoost model. The ecological resistance surface is determined based on the trained CatBoost model. The minimum cost path is determined based on the target ecological source area and the ecological resistance surface, and a bird ecological network is constructed based on the minimum cost path using current density values. Ecological pinch points are also determined based on the minimum cost path using current density values.

[0006] Optionally, the preprocessing of the bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data to obtain preprocessed bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data includes: The bird observation record data is deduplicated to obtain deduplicated bird observation record data, and outlier removal is performed on the deduplicated bird observation record data to obtain preprocessed bird observation record data. The two-dimensional environmental factor data is subjected to projection unification processing to obtain two-dimensional environmental factor data after projection unification processing. The two-dimensional environmental factor data after projection unification processing is subjected to spatial registration processing to obtain two-dimensional environmental factor data after spatial registration processing. The two-dimensional environmental factor data after spatial registration processing is subjected to resolution resampling processing to obtain preprocessed two-dimensional environmental factor data. Spatial structure extraction processing is performed on the urban three-dimensional spatial structure factor data to obtain urban three-dimensional spatial structure factor data after spatial structure extraction processing. Then, coordinate unification processing is performed on the urban three-dimensional spatial structure factor data after spatial structure extraction processing to obtain preprocessed urban three-dimensional spatial structure factor data.

[0007] Optionally, the identification of potential suitable habitats based on preprocessed bird observation records and two-dimensional environmental factor data using a maximum entropy model to obtain information on potential suitable habitats for birds includes: Obtain a sample dataset, and use the bootstrap method to train the maximum entropy model several times through random sampling based on the sample dataset to obtain a trained maximum entropy model. The preprocessed bird observation records and two-dimensional environmental factor data are input into the trained maximum entropy model to predict the probability of habitat suitability and obtain the prediction results of habitat suitability probability. Based on the natural breakpoint method, potential suitable habitats are identified using the habitat suitability probability prediction results to obtain information on potential suitable habitats for birds.

[0008] Optionally, the InVEST integrated assessment model based on ecosystem services and trade-offs utilizes the land use type raster data to assess habitat quality and obtain habitat quality assessment results, including: The land use type raster data is input into the InVEST model for habitat quality index analysis to obtain the target habitat quality index. Based on the target habitat quality index, the natural breakpoint method is used to assess habitat quality and obtain the assessment results.

[0009] Optionally, the expression for the target habitat quality index is: , , in, R represents the degree of habitat degradation, and R represents the set of threat factors. Threat factor weights, To assess the sensitivity of different habitat types to threats, The strength of the threat factor at pixel y. Let be the distance decay function. Let x be the spatial distance between pixel x and the threatening pixel y. The target habitat quality index. denoted as habitat suitability value for the land use type to which the pixel belongs, and k is a half-saturation constant.

[0010] Optionally, the step of determining candidate ecological source areas based on the information on potential suitable habitats for birds and the results of habitat quality assessment, and performing connectivity analysis on the candidate ecological source areas to obtain an overall connectivity index and a potential connectivity index, and determining the target ecological source area based on the overall connectivity index and the potential connectivity index, includes: Based on the information on potential suitable habitats for birds and the results of habitat quality assessment, spatial overlay is performed to determine candidate ecological sources; Connectivity analysis of the candidate ecological source areas is performed using Conefor software to obtain the overall connectivity index and the potential connectivity index. A comprehensive patch importance index is calculated based on the overall connectivity index and the potential connectivity index, and the target ecological source area is determined based on the comprehensive patch importance index.

[0011] Optionally, the expression for the overall connectivity index is: , Wherein, IIC is the overall connectivity index. plaque area, plaque area, Find the shortest path between patches i and j. The total area of ​​the target region; The expression for the possible connectivity index is: , Where PC is the potential connectivity index. plaque area, plaque area, The probability of the maximum product between patch i and patch j. The total area of ​​the target region; The expression for the comprehensive plaque importance index is: , in, The overall importance index is defined by IIC (Integrated Connectivity Index) and PC (Possible Connectivity Index).

[0012] Optionally, determining the ecological resistance surface based on the trained CatBoost model includes: Feature vectors are extracted from the preprocessed two-dimensional environmental factor data and the three-dimensional spatial structure factor data of the city to obtain the target feature vector. The target feature vector is input into the trained CatBoost model to perform ecological resistance prediction analysis, obtain the target ecological resistance prediction value, and determine the ecological resistance surface based on the target ecological resistance prediction value.

[0013] Optionally, the step of determining the minimum cost path based on the target ecological source area and ecological resistance surface, constructing a bird ecological network based on the minimum cost path using current density values, and determining ecological pinch points based on the minimum cost path using current density values ​​includes: Based on the target ecological source area and ecological resistance surface, a grid adjacency relationship is constructed using the 8-neighborhood movement rule, and a cost-weighted distance grid is determined based on the grid adjacency relationship. The minimum cost path is then determined based on the cost-weighted distance grid. The current density value of the multi-source migration process is calculated based on the current density model, and key ecological corridors are determined based on the current density value. The bird ecological network is obtained by performing topological association based on the key ecological corridors and the minimum cost path. The local current contraction index is calculated based on the current density value, and the current contraction candidate region is determined based on the local current contraction index. The channel compression ratio of the current contraction candidate region is determined, and the spatial bottleneck candidate region is determined based on the channel compression ratio. The node vulnerability index of the spatial bottleneck candidate region is determined, and the ecological pinch point is determined based on the node vulnerability index.

[0014] In addition, the present invention also provides a system for constructing an urban bird ecological network, the system comprising: Data preprocessing module: used to acquire bird observation record data, two-dimensional environmental factor data and urban three-dimensional spatial structure factor data of the target area, and preprocess the bird observation record data, two-dimensional environmental factor data and urban three-dimensional spatial structure factor data to obtain preprocessed bird observation record data, two-dimensional environmental factor data and urban three-dimensional spatial structure factor data; Potential habitat identification module: This module is used to identify potential suitable habitats for birds based on preprocessed bird observation records and two-dimensional environmental factor data using a maximum entropy model, thereby obtaining information on potential suitable habitats for birds. Habitat quality assessment module: used to acquire land use type raster data of the target area, and use the land use type raster data to conduct habitat quality assessment based on the InVEST integrated assessment model of ecosystem services and trade-offs, and obtain habitat quality assessment results; Ecological source area determination module: used to determine candidate ecological source areas based on the information on potential suitable habitats for birds and the results of habitat quality assessment, and to perform connectivity analysis on the candidate ecological source areas to obtain the overall connectivity index and the potential connectivity index, and to determine the target ecological source area based on the overall connectivity index and the potential connectivity index; The resistance surface determination module is used to construct an ecological resistance index system based on preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data, generate a resistance sample set based on the ecological resistance index system, train the CatBoost model based on the resistance sample set to obtain the trained CatBoost model, and determine the ecological resistance surface based on the trained CatBoost model. Ecological network construction module: used to determine the minimum cost path based on the target ecological source area and ecological resistance surface, and to construct a bird ecological network based on the minimum cost path using current density values, and to determine ecological pinch points based on the minimum cost path using current density values.

[0015] In this embodiment of the invention, a maximum entropy model is used to identify potential suitable habitats based on preprocessed bird observation records and two-dimensional environmental factor data, improving the accuracy of potential suitable habitat identification. Habitat quality is assessed using the land use type raster data based on the InVEST model. Candidate ecological source areas are determined based on bird potential suitable habitat information and habitat quality assessment results. Connectivity analysis is performed on the candidate ecological source areas to obtain the overall connectivity index and the potential connectivity index. Target ecological source areas are determined based on the overall connectivity index and the potential connectivity index, avoiding functional bias caused by a single model or single indicator in determining ecological source areas, and improving the scientific rigor and functional accuracy of ecological source area identification. An ecological resistance index system is constructed based on preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data. A resistance sample set is generated based on the ecological resistance index system. The CatBoost model is trained based on the resistance sample set. The ecological resistance surface is determined based on the trained CatBoost model, which can more realistically reflect the three-dimensional spatial constraints faced by birds migrating in high-density urban environments, improving the closeness of the ecological resistance expression to the actual migration process. The minimum cost path is determined based on the target ecological source area and ecological resistance surface. Based on the minimum cost path, the bird ecological network is constructed using current density values. Based on the minimum cost path, the ecological pinch point is determined using current density values. This improves the reliability of bird ecological network construction and realizes the refined identification and analysis of bird ecological networks in high-density urban environments. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the method for constructing an urban bird ecological network in an embodiment of the present invention; Figure 2 This is a flowchart illustrating a method for constructing an urban bird ecological network according to another embodiment of the present invention; Figure 3 This is a schematic diagram of the structural composition of the urban bird ecological network construction system in an embodiment of the present invention; Figure 4This is an example diagram illustrating the construction effect of the ecological resistance surface in the study area in an embodiment of the present invention; Figure 5 This is an example diagram illustrating the construction effect of the bird ecological network in the study area in this embodiment of the invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating the method for constructing an urban bird ecological network according to an embodiment of the present invention. The method includes: S11: Acquire bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data of the target area, and preprocess the bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data to obtain preprocessed bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data. In the specific implementation of this invention, bird observation record data is deduplicated, and outlier removal is performed on the deduplicated bird observation record data to obtain preprocessed bird observation record data; two-dimensional environmental factor data undergoes projection unification processing, spatial registration processing is performed on the projection unification processing of the two-dimensional environmental factor data, and resolution resampling processing is performed on the spatial registration processing of the two-dimensional environmental factor data to obtain preprocessed two-dimensional environmental factor data; urban three-dimensional spatial structure factor data undergoes spatial structure extraction processing, and coordinate unification processing is performed on the spatial structure extracted urban three-dimensional spatial structure factor data to obtain preprocessed urban three-dimensional spatial structure factor data, providing accurate data support for subsequent analysis and processing.

[0020] S12: Based on the preprocessed bird observation record data and two-dimensional environmental factor data, the maximum entropy model is used to identify potential suitable habitats and obtain information on potential suitable habitats for birds. In the specific implementation of this invention, a sample dataset is obtained, and the maximum entropy model is trained several times by random sampling using the bootstrapping method based on the sample dataset to obtain a trained maximum entropy model; the preprocessed bird observation record data and two-dimensional environmental factor data are input into the trained maximum entropy model to predict the probability of habitat suitability, and the result of the habitat suitability probability prediction is obtained; the potential suitable habitat is identified using the natural breakpoint method based on the result of the habitat suitability probability prediction, and the information on potential suitable habitats for birds is obtained, thereby improving the accuracy of potential suitable habitat identification.

[0021] S13: Obtain land use type raster data of the target area, and use the land use type raster data to conduct habitat quality assessment based on the InVEST model of integrated assessment of ecosystem services and trade-offs, and obtain habitat quality assessment results. In the specific implementation of this invention, land use type raster data of the target area is acquired, and the land use type raster data is input into the Integrated Valuation of Ecosystem Services and Trade-offs (InVEST) model for habitat quality index analysis to obtain the target habitat quality index; based on the target habitat quality index, the natural discontinuity method is used to conduct habitat quality assessment to obtain habitat quality assessment results, providing a spatial basis for subsequent ecological network construction.

[0022] S14: Based on the information on potential suitable habitats for birds and the results of habitat quality assessment, candidate ecological sources are determined, and connectivity analysis is performed on the candidate ecological sources to obtain the overall connectivity index and the potential connectivity index. Based on the overall connectivity index and the potential connectivity index, the target ecological source is determined. In the specific implementation of this invention, candidate ecological source areas are determined by spatial overlay based on the information on potential suitable habitats for birds and the results of habitat quality assessment. Connectivity analysis is performed on the candidate ecological source areas using Conefor software to obtain an overall connectivity index and a potential connectivity index. A comprehensive patch importance index is calculated based on the overall connectivity index and the potential connectivity index, and the target ecological source area is determined based on the comprehensive patch importance index. This avoids functional bias caused by using a single model or single indicator to determine the ecological source area, and can comprehensively reflect the potential distribution characteristics of birds, habitat quality differences, and landscape structure integrity, thereby improving the scientific nature and functional accuracy of the ecological source area identification results and reducing functionally distorted false source area identification results.

[0023] S15: Construct an ecological resistance index system based on preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data, generate a resistance sample set based on the ecological resistance index system, train the CatBoost model based on the resistance sample set to obtain a trained CatBoost model, and determine the ecological resistance surface based on the trained CatBoost model. In the specific implementation of this invention, an ecological resistance index system is constructed based on preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data. A resistance sample set is generated based on the ecological resistance index system. The CatBoost model, a machine learning model based on gradient boosting decision trees, is trained based on the resistance sample set to obtain a trained CatBoost model. Feature vectors are extracted from the preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data to obtain target feature vectors. The target feature vectors are input into the trained CatBoost model for ecological resistance prediction value analysis to obtain target ecological resistance prediction values. Based on the target ecological resistance prediction values, an ecological resistance surface is determined. Compared with the resistance characterization method based solely on two-dimensional planar environmental factors, this method can more realistically reflect the three-dimensional spatial constraints faced by birds migrating in high-density urban environments, and improve the closeness of the ecological resistance expression to the actual migration process.

[0024] S16: Determine the minimum cost path based on the target ecological source area and ecological resistance surface, construct a bird ecological network based on the minimum cost path using current density values, and determine the ecological pinch point based on the minimum cost path using current density values.

[0025] In the specific implementation of this invention, based on the target ecological source area and ecological resistance surface, a grid adjacency relationship is constructed using the 8-neighborhood movement rule. A cost-weighted distance grid is determined based on this grid adjacency relationship, and a minimum cost path is determined based on the cost-weighted distance grid. The current density value of the multi-source migration process is calculated based on the current density model, and key ecological corridors are determined based on the current density value. Topological association is performed based on the key ecological corridors and the minimum cost path to obtain the bird ecological network. A local current contraction index is calculated based on the current density value, and current contraction candidate areas are determined based on the local current contraction index. The channel compression rate of the current contraction candidate region is used to determine the spatial bottleneck candidate region based on the channel compression rate. The node vulnerability index of the spatial bottleneck candidate region is determined, and the ecological pinch point is determined based on the node vulnerability index. A multi-path migration mechanism is introduced to construct a bird ecological network, which breaks through the limitations of the single optimal path assumption in the existing technology. It can identify multiple potential migration channels between source areas and further identify key nodes and bottleneck areas in the ecological network. This is conducive to accurately locating high-risk blocking areas in the bird migration process and provides more refined technical support for targeted habitat protection, ecological restoration and spatial management.

[0026] In this embodiment of the invention, a maximum entropy model is used to identify potential suitable habitats based on preprocessed bird observation records and two-dimensional environmental factor data, improving the accuracy of potential suitable habitat identification. Habitat quality is assessed using the land use type raster data based on the InVEST model. Candidate ecological source areas are determined based on bird potential suitable habitat information and habitat quality assessment results. Connectivity analysis is performed on the candidate ecological source areas to obtain the overall connectivity index and the potential connectivity index. Target ecological source areas are determined based on the overall connectivity index and the potential connectivity index, avoiding functional bias caused by a single model or single indicator in determining ecological source areas, and improving the scientific rigor and functional accuracy of ecological source area identification. An ecological resistance index system is constructed based on preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data. A resistance sample set is generated based on the ecological resistance index system. The CatBoost model is trained based on the resistance sample set. The ecological resistance surface is determined based on the trained CatBoost model, which can more realistically reflect the three-dimensional spatial constraints faced by birds migrating in high-density urban environments, improving the closeness of the ecological resistance expression to the actual migration process. The minimum cost path is determined based on the target ecological source area and ecological resistance surface. Based on the minimum cost path, the bird ecological network is constructed using current density values. Based on the minimum cost path, the ecological pinch point is determined using current density values. This improves the reliability of bird ecological network construction and realizes the refined identification and analysis of bird ecological networks in high-density urban environments.

[0027] Example 2 Please see Figure 2 , Figure 2 This is a flowchart illustrating a method for constructing an urban bird ecological network according to another embodiment of the present invention, the method comprising: S201: Acquire bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data of the target area, and preprocess the bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data to obtain preprocessed bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data; In the specific implementation of this invention, the preprocessing of the bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data to obtain preprocessed bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data includes: performing deduplication processing on the bird observation record data to obtain deduplicated bird observation record data, and performing outlier removal processing on the deduplicated bird observation record data to obtain preprocessed bird observation record data; performing projection unification processing on the two-dimensional environmental factor data to obtain projection unification processed two-dimensional environmental factor data, performing spatial registration processing on the projection unification processed two-dimensional environmental factor data to obtain spatial registration processed two-dimensional environmental factor data, and performing resolution resampling processing on the spatial registration processed two-dimensional environmental factor data to obtain preprocessed two-dimensional environmental factor data; performing spatial structure extraction processing on the urban three-dimensional spatial structure factor data to obtain spatial structure extracted urban three-dimensional spatial structure factor data, and performing coordinate unification processing on the spatial structure extracted urban three-dimensional spatial structure factor data to obtain preprocessed urban three-dimensional spatial structure factor data.

[0028] Specifically, bird observation records are acquired in the target area, which is the research area selected by relevant personnel, along with two-dimensional environmental factor data and urban three-dimensional spatial structure factor data. The bird observation records are collected from surveys and monitoring, public science platforms, or historical databases. The two-dimensional environmental factor data includes climatic factors (annual temperature range, wet season precipitation, highest temperature in the warmest month), topographic factors (elevation, slope, aspect), vegetation cover factors (normalized difference in vegetation index), anthropogenic disturbance factors (building density, nighttime light intensity, Euclidean distance from roads), and hydrological factors (Euclidean distance from water sources, wetlands). The urban three-dimensional spatial structure factor data includes global human settlement layer data and high-resolution vegetation height data. The bird observation records are then deduplicated to obtain deduplicated bird observation records. Outlier removal is then performed on the deduplicated bird observation records to obtain preprocessed bird observation records. This process of deduplicating and removing outliers from the original observation points reduces sampling bias and spatial autocorrelation, resulting in an effective set of bird distribution points for species distribution modeling.

[0029] The two-dimensional environmental factor data undergoes projection unification processing to obtain unified two-dimensional environmental factor data. The original coordinate system information of each environmental factor data is obtained and uniformly transformed to the standard projected coordinate system of the study area, preferably UTM projection or equal-area projection. Spatial registration processing is then performed on the unified two-dimensional environmental factor data to obtain spatially registered two-dimensional environmental factor data. A data layer with high geometric accuracy is selected as the reference image. Geometric correction is performed by setting 10–20 or more control points and using a first-order polynomial or affine transformation model, ensuring that the root mean square error after correction does not exceed 0.5 pixels, achieving pixel-level spatial alignment of multi-source data. Resolution resampling processing is then performed on the spatially registered two-dimensional environmental factor data to obtain preprocessed two-dimensional environmental factor data. All environmental factor layers are unified to the same spatial resolution, preferably 10–30 m. Continuous variables are resampled using bilinear interpolation, and categorical variables are resampled using the nearest neighbor method. This ensures that each environmental factor forms a standardized environmental factor dataset under a unified coordinate system, precise spatial location, and consistent pixel scale, for subsequent overlay analysis and modeling.

[0030] Spatial structure extraction processing is performed on urban 3D spatial structure factor data to obtain processed urban 3D spatial structure factor data. Urban 3D spatial structure features such as building height (BH), building volume (BV), building floor area ratio (FAR), and vegetation canopy height (VH) are extracted to characterize the constraint effect of urban vertical space on bird flight and migration activities. Coordinate unification processing is then performed on the processed urban 3D spatial structure factor data to obtain preprocessed urban 3D spatial structure factor data. Global human settlement layer data and high-resolution vegetation height data are then subjected to projection unification, spatial registration, and resolution resampling processing to achieve... The coordinate system and spatial resolution are consistent with the two-dimensional environmental factor data. Subsequently, three-dimensional structural indicators of urban buildings are extracted based on the average building height raster layer and building base raster layer in the global human settlement layer data. The building height factor directly uses the preprocessed average building height raster data, and its cell value represents the average building height (m). The building volume factor is calculated by multiplying the building base area by the average building height. The building floor area ratio is obtained by multiplying the building base area by the number of building floors and then dividing by the cell area. At the same time, the vegetation vertical structure factor is extracted based on the high-resolution vegetation canopy height raster, and its cell value represents the height of the vegetation canopy above the ground.

[0031] S202: Based on the preprocessed bird observation record data and two-dimensional environmental factor data, the maximum entropy model is used to identify potential suitable habitats and obtain information on potential suitable habitats for birds. In the specific implementation of this invention, the step of identifying potential suitable habitats based on preprocessed bird observation records and two-dimensional environmental factor data using a maximum entropy model to obtain potential suitable habitat information for birds includes: acquiring a sample dataset, and training the maximum entropy model several times using a bootstrapping method based on the sample dataset to obtain a trained maximum entropy model; inputting the preprocessed bird observation records and two-dimensional environmental factor data into the trained maximum entropy model to predict the probability of habitat suitability, and obtaining the probability prediction result of habitat suitability; and using the probability prediction result of habitat suitability based on the natural breakpoint method to identify potential suitable habitats and obtain potential suitable habitat information for birds.

[0032] Specifically, a sample dataset is obtained, with a 75% training set and a 25% test set ratio. Based on this dataset, the maximum entropy model is trained several times using a bootstrapping method with random sampling. While maintaining the same sample size, multiple resampled subsets are generated through random sampling with replacement. Simultaneously, background point data is randomly generated within the study area to mitigate overfitting, with an optimal number of 20,000 background points. An independent MaxEnt model is trained based on each resampled subset, and the habitat suitability probability of the study area is predicted to obtain the corresponding spatial probability distribution. The above resampling and modeling process is repeated multiple times, preferably 30–200 times.

[0033] The preprocessed bird observation records and two-dimensional environmental factor data are input into the trained maximum entropy model to predict the probability of habitat suitability. The prediction results are obtained, and the prediction results are statistically summarized at the pixel level. The average value of the habitat suitability probability of each pixel is calculated as the final output.

[0034] Based on the natural breakpoint method, potential suitable habitats are identified using the habitat suitability probability prediction results to obtain information on potential suitable habitats for birds. By minimizing intra-class variance and maximizing inter-class differences through optimization principles, the suitability probability is divided into five levels: hot spots, secondary hot spots, medium, secondary cold spots, and cold spots. Based on the preset probability threshold, potential suitable habitat areas for birds are extracted, thus obtaining information on potential suitable habitats for birds in the study area.

[0035] S203: Obtain land use type raster data of the target area, and use the land use type raster data to conduct habitat quality assessment based on the InVEST model of integrated assessment of ecosystem services and trade-offs, and obtain habitat quality assessment results; In the specific implementation of this invention, the InVEST model, which is a comprehensive assessment model based on ecosystem services and trade-offs, uses the land use type raster data to assess habitat quality and obtain habitat quality assessment results. This includes: inputting the land use type raster data into the InVEST model for habitat quality index analysis to obtain a target habitat quality index; and using the natural breakpoint method to assess habitat quality based on the target habitat quality index to obtain habitat quality assessment results.

[0036] Furthermore, the expression for the target habitat quality index is: , , in, R represents the degree of habitat degradation, and R represents the set of threat factors. Threat factor weights, To assess the sensitivity of different habitat types to threats, The strength of the threat factor at pixel y. Let be the distance decay function. Let x be the spatial distance between pixel x and the threatening pixel y. The target habitat quality index. denoted as habitat suitability value for the land use type to which the pixel belongs, and k is a half-saturation constant.

[0037] Specifically, land use type raster data for the target area is acquired and input into the InVEST model for habitat quality index analysis. Each land use type is assigned a corresponding habitat suitability value. Subsequently, a set of threat factors is defined, preferentially including land types with high human disturbance intensity such as construction land and bare land. Weights, maximum impact distances, and spatial decay function types (linear or exponential decay) are defined for each threat factor to clarify the sensitivity of different habitat types to threats. Based on this, the model calculates for each pixel using the habitat quality index formula, generating a habitat quality raster map with a continuous distribution of 0 to 1. Higher values ​​represent better habitat quality and lower threat levels, thus obtaining the target habitat quality index. The expression for the target habitat quality index is: , , in, R represents the degree of habitat degradation, and R represents the set of threat factors. Threat factor weights, To assess the sensitivity of different habitat types to threats, The strength of the threat factor at pixel y. Let be the distance decay function. Let x be the spatial distance between pixel x and the threatening pixel y. The target habitat quality index. is the habitat suitability value of the land use type to which the pixel belongs, k is a half-saturation constant, k is used to control the degree of degradation impact, and the preferred value range can be 0.3-0.7.

[0038] Based on the target habitat quality index, habitat quality is assessed using the natural breakpoint method to obtain habitat quality assessment results. Specifically, the habitat quality grid is classified using the natural breakpoint method to extract high-quality habitat areas, providing a spatial basis for subsequent ecological network construction.

[0039] S204: Based on the information on potential suitable habitats for birds and the results of habitat quality assessment, spatial overlay is performed to determine candidate ecological sources; In the specific implementation of this invention, based on the information on potential suitable habitats for birds and the results of habitat quality assessment, spatial overlay is performed to determine candidate ecological sources. The potential suitable habitat areas for birds and the high-quality habitat areas for birds in the results of habitat quality assessment are spatially overlaid and designated as foreground pixels for morphological spatial pattern analysis. Other areas are used as background, and the edge width is set to 1. Core patches with complete structure and high spatial continuity are extracted as candidate ecological sources.

[0040] S205: Based on Conefor software, perform connectivity analysis on the candidate ecological source areas to obtain the overall connectivity index and the potential connectivity index. Calculate the comprehensive patch importance index based on the overall connectivity index and the potential connectivity index, and determine the target ecological source area based on the comprehensive patch importance index. In the specific implementation of this invention, connectivity analysis is performed on the candidate ecological source areas using Conefor software to obtain the overall connectivity index and the potential connectivity index. The expression for the overall connectivity index is as follows: , Wherein, IIC is the overall connectivity index. plaque area, plaque area, Find the shortest path between patches i and j. The total area of ​​the target region; The expression for the possible connectivity index is: , Where PC is the potential connectivity index. plaque area, plaque area, The probability of the maximum product between patch i and patch j. The total area of ​​the target region.

[0041] The comprehensive patch importance index is calculated based on the overall connectivity index and the potential connectivity index. The expression for the comprehensive patch importance index is as follows: , in, The comprehensive patch importance index is used, where IIC is the overall connectivity index and PC is the potential connectivity index. Target ecological source areas are determined based on this comprehensive patch importance index; that is, patches with a comprehensive patch importance index greater than a preset threshold are selected as target ecological source areas. The preset threshold can be set to 0.9.

[0042] S206: Construct an ecological resistance index system based on preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data; generate a resistance sample set based on the ecological resistance index system; train the CatBoost model based on the resistance sample set to obtain a trained CatBoost model; extract feature vectors from the preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data to obtain target feature vectors. In the specific implementation of this invention, an ecological resistance index system is constructed based on preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data. Two-dimensional environmental factors such as land use type, vegetation coverage, climate factors, topography factors, and human disturbance factors are taken, and urban three-dimensional spatial structure factors such as building height, building volume, building plot ratio, and vegetation canopy height are introduced to construct a three-dimensional ecological resistance index system.

[0043] Based on the aforementioned ecological resistance index system, a resistance sample set is generated. Low-resistance samples (positive samples) are randomly generated within the bird's habitat, while high-resistance samples (negative samples) are generated in areas outside the habitat where human disturbance is high. The minimum distance between sample points is controlled to ensure spatial independence, forming a labeled training sample set. Subsequently, multi-dimensional feature extraction is performed. The spatial location of the sample points is spatially superimposed with the preprocessed two-dimensional environmental factor layer and three-dimensional spatial structure factor layer. A corresponding environmental feature vector is extracted for each sample, forming the model input dataset, which is the resistance sample set.

[0044] The CatBoost model is trained based on the aforementioned resistance sample set to obtain a trained CatBoost model. The sample data is randomly divided into a training set (70%-80%) and a test set (20%-30%). A gradient boosting decision tree-based machine learning model (CatBoost) is trained and fitted to the relationship of bird migration resistance. Hierarchical K-fold cross-validation and Bayesian optimization algorithms are used to optimize the model's key hyperparameters (tree depth, learning rate, number of iterations) to automatically learn the nonlinear influence of different environmental factors on bird migration resistance. The model's accuracy, recall, F1 score, and AUC are calculated using the test set to evaluate the model's performance. When the model performance meets a preset threshold (e.g., AUC ≥ 0.85), the resistance surface generation step begins. Feature vectors are extracted from the preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data to obtain target feature vectors. This involves extracting the complete two-dimensional environmental factor and three-dimensional spatial structure factor feature vectors from each grid cell as a sample point to be predicted.

[0045] S207: Input the target feature vector into the trained CatBoost model to perform ecological resistance prediction analysis, obtain the target ecological resistance prediction value, and determine the ecological resistance surface based on the target ecological resistance prediction value; In the specific implementation of this invention, the target feature vector is input into a trained CatBoost model for ecological resistance prediction analysis to obtain the target ecological resistance prediction value. Based on the target ecological resistance prediction value, an ecological resistance surface is determined and input into the trained CatBoost model. The model will output an ecological resistance prediction value for each pixel. (A higher value indicates greater resistance to passage), thus generating a continuous three-dimensional ecological resistance surface that reflects spatial heterogeneity. By adopting an ecological resistance assignment mechanism based on an interpretable machine learning model, it can characterize the nonlinear interaction between multiple factors, reduce the subjectivity and uncertainty brought about by traditional empirical weighting and linear superposition methods, improve the objectivity, stability and regional applicability of the ecological resistance surface construction results, and provide decision-making basis with clear physical meaning and quantitative constraints for ecological space optimization and planning management.

[0046] The effect of ecological resistance surface construction in the study area is as follows: Figure 4 As shown, a total of 249 ecological corridors were extracted, with a total length of 766.77 km. The longest corridor is 25.50 km. The bird ecological network exhibits an ecological pattern of "two horizontal and one vertical, multi-core driven, and corridors interwoven." Among them, there are 40 primary corridors, mainly distributed in the central and western regions, forming the main framework of the ecological network; 56 secondary corridors extend along multiple secondary green belts, supplementing the main corridors; and 153 tertiary corridors cover green patches and small wetlands around cities, with relatively fragile connectivity. In addition, 28 ecological fulcrums were identified, mainly concentrated near the central mountain corridor and the western water system corridor, and along the primary ecological corridors.

[0047] S208: Determine the minimum cost path based on the target ecological source area and ecological resistance surface, construct a bird ecological network based on the minimum cost path using current density values, and determine ecological pinch points based on the minimum cost path using current density values.

[0048] In the specific implementation of this invention, the process of determining the minimum cost path based on the target ecological source area and ecological resistance surface, constructing a bird ecological network based on the minimum cost path using current density values, and determining ecological pinch points based on the minimum cost path using current density values ​​includes: constructing grid adjacency relationships based on the target ecological source area and ecological resistance surface using 8-neighborhood movement rules, determining cost-weighted distance grids based on the grid adjacency relationships, and determining minimum cost paths based on the cost-weighted distance grids; calculating current density values ​​for the multi-source migration process based on the current density model, determining key ecological corridors based on the current density values, performing topological associations based on the key ecological corridors and minimum cost paths to obtain a bird ecological network; calculating a local current contraction index based on the current density values, and determining current contraction candidate areas based on the local current contraction index; determining the channel compression rate of the current contraction candidate areas, determining spatial bottleneck candidate areas based on the channel compression rate, determining the node vulnerability index of the spatial bottleneck candidate areas, and determining ecological pinch points based on the node vulnerability index.

[0049] Specifically, based on the target ecological source area and ecological resistance surface, a grid adjacency relationship is constructed using the 8-neighborhood movement rule. The 8-neighborhood movement rule typically refers to a pixel search or tracking algorithm based on 8-neighborhood connectivity, used for tasks such as boundary extraction, region growing, or path tracing. Its core idea is that the movement or exploration range of a pixel is limited to its eight neighboring pixels, including up, down, left, right, and the four diagonal directions. Based on the grid adjacency relationship, a cost-weighted distance grid is determined, and the cost-weighted distance grid (CWD) from any grid cell to the nearest ecological source area is calculated. This CWD is the cumulative sum of the products of the resistance values ​​of each grid cell along the path and the movement distance. The expression for the cost-weighted distance grid is: , in, Let i be the ecological resistance value of the i-th grid cell. It represents the distance traversed through the i-th grid cell.

[0050] Based on the cost-weighted distance grid, a minimum cost path is determined. At each step, the cell with the smallest cost-weighted distance in the neighborhood is selected until the originating source is reached, forming a minimum cost path (LCP) between source locations, which is theoretically the optimal single migration corridor. Finally, the extracted path is transformed into a grid or linear ecological connectivity skeleton for subsequent key corridor identification and ecological network construction. The expression for the minimum cost path is: , in, Let i be the ecological resistance value of the i-th grid cell. It represents the distance traversed through the i-th grid cell.

[0051] Based on the current density model, the current density value of the multi-source migration process is calculated. To overcome the limitation that a single path cannot reflect the randomness of bird dispersal, this invention further employs a current density model based on circuit theory to identify key ecological corridors, building upon the minimum cost path. First, all ecological source areas are considered as current source nodes. A unit current is applied between source pairs to simulate the current distribution in the resistance network, and the cumulative current density distribution grid of the study area is calculated. The current density value reflects the passability probability and path overlap of pixels during the multi-source migration process. The expression for the current density value is: , in, This indicates the migration process between source locations s and t via pixels. The current value, cumulative current density Used to characterize the passage probability and channel importance of a cell in a multi-path migration process. Based on the current density values, key ecological corridors were identified. Considering both the current density results and the spatial distribution of the minimum cost path, areas with current densities above the 80th percentile of the entire region, located within a 300-meter buffer zone of the minimum cost path, and with a continuous length of at least 500 meters were selected as key ecological corridors. A topological association was then performed on these key ecological corridors and the minimum cost path to obtain a bird ecological network. All the extracted key ecological corridors were then topologically associated with their source areas (source areas as network nodes, key ecological corridors as network edges) to construct an ecological network that characterizes the multi-path migration relationships of birds in a three-dimensional urban landscape. The construction effect of the bird ecological network is shown below. Figure 5 As shown, The local current contraction index (SCI) is calculated based on the current density value, and candidate regions for current contraction are determined based on the SCI. Using a 3×3 neighborhood window, the SCI is calculated from the current density grid obtained by circuit theory. The SCI is defined as the ratio of the current density of the central grid to the average current density of the neighborhood, to identify areas of concentrated current transmission. When the SCI is greater than 1.8 and the current density is in the top 70 percentile of the entire region, the region is identified as a candidate region for current contraction. The expression for the SCI is: , in, For grid The current contraction index, For grid current density, for The neighborhood set is 3×3, where n is the number of neighborhood grid cells.

[0052] The channel compression ratio of candidate areas for current contraction is determined, and based on this channel compression ratio, candidate areas for spatial bottlenecks are identified. To avoid misidentifying wide ecological corridors as pinch points, spatial structure analysis is then performed on the candidate areas, calculating the channel compression ratio CR, which is the ratio of the lateral width of the candidate area to its longitudinal principal axis length, to identify channels exhibiting narrow spatial structures. When the channel compression ratio is less than 0.35 and located within the 300-meter buffer zone of the minimum cost path between two ecological source areas, the area is identified as a candidate area for spatial bottlenecks. The expression for the channel compression ratio is: , in, Candidate region Compression ratio; The average width is horizontal. The main axis length is used. The node vulnerability index of candidate spatial bottleneck areas is determined, and ecological pinch points are identified based on this index. To verify the actual contribution of candidate areas to network connectivity, a connectivity perturbation verification mechanism is introduced. The ecological resistance value within the candidate area is increased to 1.5 times the original resistance to simulate habitat degradation, and current diffusion simulation is performed again to calculate the overall network connectivity after the perturbation. By comparing the changes in network connectivity before and after the perturbation, a node vulnerability index (NVI) is constructed, defined as the percentage decrease in overall connectivity before and after the perturbation. When the node vulnerability index is greater than 0.15, the area is identified as an ecological pinch point. The expression for the node vulnerability index is: , in, For the original network connectivity, This refers to the connectivity after the disturbance.

[0053] In this embodiment of the invention, a maximum entropy model is used to identify potential suitable habitats based on preprocessed bird observation records and two-dimensional environmental factor data, improving the accuracy of potential suitable habitat identification. Habitat quality is assessed using the land use type raster data based on the InVEST model. Candidate ecological source areas are determined based on bird potential suitable habitat information and habitat quality assessment results. Connectivity analysis is performed on the candidate ecological source areas to obtain the overall connectivity index and the potential connectivity index. Target ecological source areas are determined based on the overall connectivity index and the potential connectivity index, avoiding functional bias caused by a single model or single indicator in determining ecological source areas, and improving the scientific rigor and functional accuracy of ecological source area identification. An ecological resistance index system is constructed based on preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data. A resistance sample set is generated based on the ecological resistance index system. The CatBoost model is trained based on the resistance sample set. The ecological resistance surface is determined based on the trained CatBoost model, which can more realistically reflect the three-dimensional spatial constraints faced by birds migrating in high-density urban environments, improving the closeness of the ecological resistance expression to the actual migration process. The minimum cost path is determined based on the target ecological source area and ecological resistance surface. Based on the minimum cost path, the bird ecological network is constructed using current density values. Based on the minimum cost path, the ecological pinch point is determined using current density values. This improves the reliability of bird ecological network construction and realizes the refined identification and analysis of bird ecological networks in high-density urban environments.

[0054] Example 3 Please see Figure 3 , Figure 3 This is a schematic diagram of the structural composition of an urban bird ecological network construction system according to an embodiment of the present invention. The system includes: Data preprocessing module 31: used to acquire bird observation record data, two-dimensional environmental factor data and urban three-dimensional spatial structure factor data of the target area, and preprocess the bird observation record data, two-dimensional environmental factor data and urban three-dimensional spatial structure factor data to obtain preprocessed bird observation record data, two-dimensional environmental factor data and urban three-dimensional spatial structure factor data; Potential habitat identification module 32: used to identify potential suitable habitats based on preprocessed bird observation record data and two-dimensional environmental factor data using the maximum entropy model, and obtain information on potential suitable habitats for birds; Habitat quality assessment module 33: used to acquire land use type raster data of the target area, and to conduct habitat quality assessment using the land use type raster data based on the InVEST integrated assessment model of ecosystem services and trade-offs, and obtain habitat quality assessment results; Ecological source area determination module 34: used to determine candidate ecological source areas based on the information on potential suitable habitats for birds and the results of habitat quality assessment, and to perform connectivity analysis on the candidate ecological source areas to obtain an overall connectivity index and a potential connectivity index, and to determine the target ecological source area based on the overall connectivity index and the potential connectivity index; The resistance surface determination module 35 is used to construct an ecological resistance index system based on preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data, generate a resistance sample set based on the ecological resistance index system, train the CatBoost model based on the resistance sample set to obtain a trained CatBoost model, and determine the ecological resistance surface based on the trained CatBoost model. Ecological network construction module 36: used to determine the minimum cost path based on the target ecological source area and ecological resistance surface, and to construct a bird ecological network based on the minimum cost path using current density values, and to determine ecological pinch points based on the minimum cost path using current density values.

[0055] In the specific implementation of this invention, the specific implementation methods of the system items can be referred to the implementation methods of the above-mentioned method items, and will not be repeated here.

[0056] In this embodiment of the invention, a maximum entropy model is used to identify potential suitable habitats based on preprocessed bird observation records and two-dimensional environmental factor data, improving the accuracy of potential suitable habitat identification. Habitat quality is assessed using the land use type raster data based on the InVEST model. Candidate ecological source areas are determined based on bird potential suitable habitat information and habitat quality assessment results. Connectivity analysis is performed on the candidate ecological source areas to obtain the overall connectivity index and the potential connectivity index. Target ecological source areas are determined based on the overall connectivity index and the potential connectivity index, avoiding functional bias caused by a single model or single indicator in determining ecological source areas, and improving the scientific rigor and functional accuracy of ecological source area identification. An ecological resistance index system is constructed based on preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data. A resistance sample set is generated based on the ecological resistance index system. The CatBoost model is trained based on the resistance sample set. The ecological resistance surface is determined based on the trained CatBoost model, which can more realistically reflect the three-dimensional spatial constraints faced by birds migrating in high-density urban environments, improving the closeness of the ecological resistance expression to the actual migration process. The minimum cost path is determined based on the target ecological source area and ecological resistance surface. Based on the minimum cost path, the bird ecological network is constructed using current density values. Based on the minimum cost path, the ecological pinch point is determined using current density values. This improves the reliability of bird ecological network construction and realizes the refined identification and analysis of bird ecological networks in high-density urban environments.

[0057] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc.

[0058] Furthermore, the above provides a detailed description of the method and system for constructing an urban bird ecological network according to the embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for constructing an urban bird ecological network, characterized in that, The method includes: Bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data of the target area are acquired, and the bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data are preprocessed to obtain preprocessed bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data. Based on preprocessed bird observation records and two-dimensional environmental factor data, the maximum entropy model was used to identify potential suitable habitats and obtain information on potential suitable habitats for birds. Land use type raster data of the target area is acquired, and habitat quality is assessed using the InVEST model, which is a comprehensive assessment of ecosystem services and trade-offs, to obtain habitat quality assessment results. Candidate ecological sources are determined based on the information on potential suitable habitats for birds and the results of habitat quality assessment. Connectivity analysis is then performed on the candidate ecological sources to obtain an overall connectivity index and a potential connectivity index. Target ecological sources are then determined based on the overall connectivity index and the potential connectivity index. An ecological resistance index system is constructed based on preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data. A resistance sample set is generated based on the ecological resistance index system. The CatBoost model is trained based on the resistance sample set to obtain a trained CatBoost model. The ecological resistance surface is determined based on the trained CatBoost model. The minimum cost path is determined based on the target ecological source area and the ecological resistance surface, and a bird ecological network is constructed based on the minimum cost path using current density values. Ecological pinch points are also determined based on the minimum cost path using current density values.

2. The method for constructing an urban bird ecological network according to claim 1, characterized in that, The preprocessing of the bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data to obtain preprocessed bird observation record data, two-dimensional environmental factor data, and urban three-dimensional spatial structure factor data includes: The bird observation record data is deduplicated to obtain deduplicated bird observation record data, and outlier removal is performed on the deduplicated bird observation record data to obtain preprocessed bird observation record data. The two-dimensional environmental factor data is subjected to projection unification processing to obtain two-dimensional environmental factor data after projection unification processing. The two-dimensional environmental factor data after projection unification processing is subjected to spatial registration processing to obtain two-dimensional environmental factor data after spatial registration processing. The two-dimensional environmental factor data after spatial registration processing is subjected to resolution resampling processing to obtain preprocessed two-dimensional environmental factor data. Spatial structure extraction processing is performed on the urban three-dimensional spatial structure factor data to obtain urban three-dimensional spatial structure factor data after spatial structure extraction processing. Then, coordinate unification processing is performed on the urban three-dimensional spatial structure factor data after spatial structure extraction processing to obtain preprocessed urban three-dimensional spatial structure factor data.

3. The method for constructing an urban bird ecological network according to claim 1, characterized in that, The preprocessed bird observation records and two-dimensional environmental factor data are used to identify potential suitable habitats using a maximum entropy model, obtaining information on potential suitable habitats for birds, including: Obtain a sample dataset, and use the bootstrap method to train the maximum entropy model several times through random sampling based on the sample dataset to obtain a trained maximum entropy model. The preprocessed bird observation records and two-dimensional environmental factor data are input into the trained maximum entropy model to predict the probability of habitat suitability and obtain the prediction results of habitat suitability probability. Based on the natural breakpoint method, potential suitable habitats are identified using the habitat suitability probability prediction results to obtain information on potential suitable habitats for birds.

4. The method for constructing an urban bird ecological network according to claim 1, characterized in that, The InVEST integrated assessment model based on ecosystem services and trade-offs uses the land use type raster data to assess habitat quality and obtain habitat quality assessment results, including: The land use type raster data is input into the InVEST model for habitat quality index analysis to obtain the target habitat quality index. Based on the target habitat quality index, the natural breakpoint method is used to assess habitat quality and obtain the assessment results.

5. The method for constructing an urban bird ecological network according to claim 4, characterized in that, The expression for the target habitat quality index is: ， ， in, R represents the degree of habitat degradation, and R represents the set of threat factors. Threat factor weights, To assess the sensitivity of different habitat types to threats, The strength of the threat factor at pixel y. Let be the distance decay function. Let x be the spatial distance between pixel x and the threatening pixel y. The target habitat quality index. denoted as habitat suitability value for the land use type to which the pixel belongs, and k is a half-saturation constant.

6. The method for constructing an urban bird ecological network according to claim 1, characterized in that, The process of identifying candidate ecological source areas based on the information on potential suitable habitats for birds and the results of habitat quality assessment, performing connectivity analysis on the candidate ecological source areas to obtain an overall connectivity index and a potential connectivity index, and determining the target ecological source area based on the overall connectivity index and the potential connectivity index includes: Based on the information on potential suitable habitats for birds and the results of habitat quality assessment, spatial overlay is performed to determine candidate ecological sources; Connectivity analysis of the candidate ecological source areas is performed using Conefor software to obtain the overall connectivity index and the potential connectivity index. A comprehensive patch importance index is calculated based on the overall connectivity index and the potential connectivity index, and the target ecological source area is determined based on the comprehensive patch importance index.

7. The method for constructing an urban bird ecological network according to claim 6, characterized in that, The expression for the overall connectivity index is: ， Wherein, IIC is the overall connectivity index. plaque area, plaque area, Find the shortest path between patches i and j. The total area of ​​the target region; The expression for the possible connectivity index is: ， Where PC is the potential connectivity index. plaque area, plaque area, The probability of the maximum product between patch i and patch j. The total area of ​​the target region; The expression for the comprehensive plaque importance index is: ， in, The overall importance index is defined by IIC (Integrated Connectivity Index) and PC (Possible Connectivity Index).

8. The method for constructing an urban bird ecological network according to claim 1, characterized in that, The determination of the ecological resistance surface based on the trained CatBoost model includes: Feature vectors are extracted from the preprocessed two-dimensional environmental factor data and the three-dimensional spatial structure factor data of the city to obtain the target feature vector. The target feature vector is input into the trained CatBoost model to perform ecological resistance prediction analysis, obtain the target ecological resistance prediction value, and determine the ecological resistance surface based on the target ecological resistance prediction value.

9. The method for constructing an urban bird ecological network according to claim 1, characterized in that, The process of determining the minimum cost path based on the target ecological source area and ecological resistance surface, constructing a bird ecological network using current density values ​​based on the minimum cost path, and determining ecological pinch points using current density values ​​based on the minimum cost path includes: Based on the target ecological source area and ecological resistance surface, a grid adjacency relationship is constructed using the 8-neighborhood movement rule, and a cost-weighted distance grid is determined based on the grid adjacency relationship. The minimum cost path is then determined based on the cost-weighted distance grid. The current density value of the multi-source migration process is calculated based on the current density model, and key ecological corridors are determined based on the current density value. The bird ecological network is obtained by performing topological association based on the key ecological corridors and the minimum cost path. The local current contraction index is calculated based on the current density value, and the current contraction candidate region is determined based on the local current contraction index. The channel compression ratio of the current contraction candidate region is determined, and the spatial bottleneck candidate region is determined based on the channel compression ratio. The node vulnerability index of the spatial bottleneck candidate region is determined, and the ecological pinch point is determined based on the node vulnerability index.

10. A system for constructing an urban bird ecological network, characterized in that, The system includes: Data preprocessing module: used to acquire bird observation record data, two-dimensional environmental factor data and urban three-dimensional spatial structure factor data of the target area, and preprocess the bird observation record data, two-dimensional environmental factor data and urban three-dimensional spatial structure factor data to obtain preprocessed bird observation record data, two-dimensional environmental factor data and urban three-dimensional spatial structure factor data; Potential habitat identification module: This module is used to identify potential suitable habitats for birds based on preprocessed bird observation records and two-dimensional environmental factor data using a maximum entropy model, thereby obtaining information on potential suitable habitats for birds. Habitat quality assessment module: used to acquire land use type raster data of the target area, and use the land use type raster data to conduct habitat quality assessment based on the InVEST integrated assessment model of ecosystem services and trade-offs, and obtain habitat quality assessment results; Ecological source area determination module: used to determine candidate ecological source areas based on the information on potential suitable habitats for birds and the results of habitat quality assessment, and to perform connectivity analysis on the candidate ecological source areas to obtain the overall connectivity index and the potential connectivity index, and to determine the target ecological source area based on the overall connectivity index and the potential connectivity index; The resistance surface determination module is used to construct an ecological resistance index system based on preprocessed two-dimensional environmental factor data and urban three-dimensional spatial structure factor data, generate a resistance sample set based on the ecological resistance index system, train the CatBoost model based on the resistance sample set to obtain the trained CatBoost model, and determine the ecological resistance surface based on the trained CatBoost model. Ecological network construction module: used to determine the minimum cost path based on the target ecological source area and ecological resistance surface, and to construct a bird ecological network based on the minimum cost path using current density values, and to determine ecological pinch points based on the minimum cost path using current density values.

Images