An ecological function improvement-oriented land use scenario simulation method and system

Abstract

The application belongs to the technical field of land use planning and ecological environment protection, more specifically, relates to a land use scenario simulation method and system for ecological function improvement, through obtaining land use and driving factor data of a study area, dynamically evaluating habitat quality and analyzing spatio-temporal evolution characteristics; eliminating redundant collinear factors, accurately identifying key driving factors and their influence thresholds through multi-model coupling, solving the problems of core driving selection and quantitative constraint deficiency; then converting the thresholds into spatial ecological constraints, combining with regional policies to build differentiated land use scenarios, substituting into the calibrated spatial simulation model to predict future land use patterns; finally, calculating future habitat quality, positioning ecological risk areas, coupling ecological background grade to delimit control zones and develop differentiated management strategies, forming a complete optimization scheme; perfecting the connection between simulation results and land space governance, and forming a land use pattern optimization scheme for ecological function improvement.

Description

Technical Field

[0001] This application belongs to the field of land use planning and ecological environment protection technology, and more specifically, it relates to a land use scenario simulation method and system for improving ecological functions. Background Technology

[0002] my country has a rich variety of land use types and a complex structure. The evolution of spatial patterns has a profound impact on regional ecological security and sustainable development. With the rapid development of urbanization and industrialization, the scope of human activities is constantly expanding into ecologically vulnerable areas, leading to accelerated fragmentation of natural habitats, a continuous decline in ecosystem service functions, and increasingly prominent contradictions between humans and the land. This poses a great challenge to the spatial optimization of land use patterns. Therefore, it is urgent to construct a land use pattern optimization framework constrained by ecological security to provide scientific support for the coordinated development of regional human-land relations and sustainable development.

[0003] Land use scenario prediction is a technical method that combines regional natural and socio-economic conditions, sets up multiple development scenarios, and uses spatial simulation technology to predict the spatiotemporal evolution of land. This method has become an important tool for land use pattern analysis, helping planners identify spatial conflicts and ecological risks under different development paths, and providing a basis for coordinated protection and development and optimization of land spatial structure. Constructing land use prediction models can accurately capture the spatiotemporal evolution characteristics and patterns of land use, providing scientific predictions for the layout of national land spatial planning and the optimization of development and protection patterns. At the same time, it provides strong support for regional ecological security and the coordinated development of human-land relations, which is of great significance for improving national land spatial governance. However, land use scenario simulation is highly dependent on the scientific construction of a constraint system. Although current simulations have incorporated policy objectives that align with regional development, there are still shortcomings in the screening and quantitative constraints of core driving factors of land use evolution. How to construct prediction rules that couple core driving mechanisms and policy orientations and conform to the laws of ecological evolution is the core challenge for further improving the accuracy and decision support value of multi-scenario land use prediction.

[0004] Land use scenario simulation technology can predict future land spatiotemporal evolution in a controllable and low-cost manner, serving as an important tool for supporting scientific understanding and planning decisions in land space. Since the 20th century, nonlinear spatial simulation methods based on cellular automata have driven the development of this field, subsequently leading to the formation of multivariate simulation systems represented by CLUE-S, SLEUTH, and grey system coupling models, continuously enhancing the analytical capabilities for the complex dynamic mechanisms of land use change. In recent years, the FLUS model, which integrates artificial neural networks and adaptive competition mechanisms, has further improved the accuracy and stability of land use evolution simulation under the synergistic effects of multiple driving factors. However, existing land use simulation technologies still have significant limitations in the coupling of ecological processes. On the one hand, while most models incorporate policy guidance and core driving factors in scenario construction, their characterization of the nonlinear feedback mechanism of ecological processes remains insufficient, making it difficult to accurately reveal the dynamic coupling relationship between land use pattern evolution and ecosystem service functions. On the other hand, current research still needs to further improve standardized technical paths to promote the deep integration of multi-scale simulation results with the national land spatial planning and management system, so as to better embed them into actual governance processes such as policy formulation, scheme evaluation, and dynamic supervision. Based on this, constructing a land use scenario simulation framework that integrates ecological constraint mechanisms, relying on the analysis of core driving factors and the optimization of the FLUS model, has become a key direction that urgently needs to be broken through in the current research on sustainable governance of land spatial patterns. Summary of the Invention

[0005] In order to overcome the technical problems mentioned in the prior art, the present invention provides a land use scenario simulation method and system for improving ecological functions.

[0006] On the one hand, this invention provides a land use scenario simulation method for improving ecological functions, comprising the following steps: Obtain land use data and driving factor data for the study area; The habitat quality of the study area is dynamically assessed based on land use data to obtain habitat quality assessment results, and the spatiotemporal evolution characteristics of habitat quality are analyzed based on the habitat quality assessment results. After preprocessing the driving factor data and removing redundant and collinear factors, a set of driving factors is obtained. Using the habitat quality assessment results and the set of driving factors as input, a multi-model coupling approach is adopted to identify the key driving factors affecting habitat quality and their impact thresholds. The identified key driving factors and their impact thresholds are transformed into spatial constraint rules. Combined with regional development policy guidance, multiple differentiated land use scenarios are constructed and parameterized. The spatial constraint rules and scenario parameters are input into a calibrated and verified spatial simulation model. Using recent land use data of the study area as the base period, the spatial pattern of land use in future years is simulated and predicted. Based on the land use spatial patterns under various future scenarios, the future habitat quality pattern is calculated, and ecological risk areas are identified by comparing the habitat quality in the baseline period. The importance of the ecological baseline in the study area is assessed and classified. Through spatial overlay analysis, the ecological risk areas are coupled with the importance level of the ecological baseline to construct a diagnostic matrix, divide different types of ecological control zones, formulate differentiated spatial governance strategies for each ecological control zone, and form an optimized land use spatial pattern scheme oriented towards improving ecological functions.

[0007] This invention acquires land use and driving factor data of the study area, dynamically assesses habitat quality, and analyzes its spatiotemporal evolution characteristics, laying the foundation for ecological constraint construction. Subsequently, the driving factors are preprocessed and redundant collinear factors are removed. A multi-model coupling approach is used to accurately identify key driving factors affecting habitat quality and their impact thresholds, solving the problem of insufficient screening and quantification constraints for core driving factors, and constructing prediction rules coupled with core driving mechanisms. Next, the key driving factor thresholds are transformed into spatial constraint rules. Multiple differentiated land use scenarios are constructed and parameterized in conjunction with regional development policy guidance. These scenarios are then input into a calibrated and validated spatial simulation model to predict future land use patterns, compensating for the shortcomings of existing models in coupling ecological evolution patterns with policy guidance. Finally, by calculating future habitat quality, identifying ecological risk areas, coupling ecological baseline importance levels to classify ecological control zones, and formulating differentiated governance strategies, this not only depicts the dynamic coupling relationship between land use pattern evolution and ecosystem service functions but also improves the connection path between simulation results and the national land space governance process, ultimately forming a land use spatial pattern optimization scheme oriented towards ecological function enhancement.

[0008] Preferably, the driving factor data includes background-type driving factors and human activity-type driving factors; The natural background driving factors include at least one of the following: altitude, slope, topographic relief, lithology, annual precipitation, soil type, and distance from the river; The human activity-related driving factors include at least one of the following: nighttime light index, distance from road, distance from settlement, proportion of construction land, proportion of arable land, landscape fragmentation index, population density, and GDP.

[0009] Preferably, the habitat quality assessment result adopts a habitat quality assessment model, wherein the habitat quality assessment model is the habitat quality module of the InVEST model.

[0010] Preferably, the spatiotemporal evolution characteristic analysis of habitat quality includes spatial autocorrelation analysis, hot and cold spot detection, and transition matrix analysis, which are used to reveal the spatial clustering patterns, stable regions, and evolution trajectories of habitat quality, and to identify areas of persistently low habitat quality and areas of significant degradation.

[0011] Preferably, the collinearity factor is eliminated by Spearman correlation analysis to remove multicollinearity among the driving factors.

[0012] Preferably, the multi-model coupling includes the following steps: Based on the set of driving factors and habitat quality assessment results as input, the marginal contribution of each driving factor is quantified through an interpretable machine learning model, the global importance and direction of action of each driving factor are determined, and key driving factors and their impact thresholds are identified. By using the interaction detection of geographic detectors, and taking the set of driving factors and habitat quality assessment results as input, the explanatory power of the superposition of single driving factors is compared to determine the synergistic enhancement, inhibition or independent effect relationship between driving factors, which helps to accurately identify key driving factors and their impact thresholds.

[0013] Preferably, the construction of the multiple differentiated land use scenarios takes regional development policy guidance as input and includes at least three types, namely natural development scenario, ecological priority scenario, and urban-rural development and rural revitalization scenario. After parameter setting, the parameters corresponding to each scenario are output. The natural development scenario is based on the historical land use transfer patterns of the study area, without imposing additional ecological constraints, and outputs natural development scenario parameters. The ecological priority scenario is based on the results of key driving factors and impact thresholds, and incorporates rigid ecological constraint rules to improve the attractiveness of ecological land transfer and the resistance to transfer out, and outputs ecological priority scenario parameters. The urban and rural development and rural revitalization scenarios adhere to the ecological protection red line and the bottom line of permanent basic farmland, increase the probability of conversion of urban and rural construction land, guide agricultural land to transform into ecologically friendly and high-value-added land, and output urban and rural development and rural revitalization scenario parameters; The three scenario parameters mentioned above are used as inputs to the space simulation model.

[0014] Preferably, the spatial constraint rules take key driving factors and influence thresholds as inputs, map the influence thresholds of key driving factors into prohibitive or probabilistic rules for the conversion of different land use types, and output spatial constraint rules as one of the inputs to the spatial simulation model.

[0015] Preferably, the spatial overlay analysis uses the results of ecological risk areas and the results of ecological baseline importance levels as input for analysis; The ecological risk area results are obtained based on the following steps: Using the future habitat quality pattern results and habitat quality assessment results as inputs, the differences between the two are compared, and spatial units where habitat quality has declined significantly compared to the base period are identified as ecological risk areas, and the ecological risk area results are output. The results of the ecological baseline importance level were obtained based on the following steps: Using land use data of the study area as input, the potential habitat quality is calculated by removing the weight of human threats, and the potential habitat quality data is output. Then, the natural breakpoint method is used to divide the potential habitat quality into multiple levels and output the ecological background importance level results.

[0016] Preferably, the ecological management zoning includes the following steps: Based on the results of ecological risk areas and the importance level of ecological baseline, a diagnostic matrix is ​​constructed by coupling spatial overlay analysis, and the diagnostic matrix results are output. Based on the diagnostic matrix results, multiple ecological management zones are divided, and the ecological management zone results are output. These results are then used as input for formulating differentiated spatial governance strategies.

[0017] On the other hand, the present invention provides a land use scenario simulation system for improving ecological functions, including a memory and a processor. The memory stores a computer program, and the processor can call the computer program stored in the memory to execute the land use scenario simulation method for improving ecological functions described in the present invention.

[0018] The beneficial effects of this invention include: 1. This invention constructs a multi-model coupled analysis framework to overcome the shortcomings of single methods in identifying driving factors, revealing the complex coupling mechanism between natural and social systems behind habitat quality degradation. Furthermore, based on the dual dimensions of natural background constraints and human activity interference, it identifies key driving factors and their boundaries of influence on habitat quality degradation, thereby clarifying the impact of human activities on habitat quality and overcoming the limitations of fragmented explanations. Moreover, this invention proposes an analytical path of pattern analysis – factor screening – mechanism analysis, deconstructing the spatiotemporal evolution of land use and habitat quality. This analytical path provides a basis and support for subsequent ecological threshold parameterization and its embedding into land use scenario simulation and ecological protection spatial management rules.

[0019] 2. This invention constructs a multi-scenario land use simulation framework by coupling key driving factor constraints with policy objective parameterization. This not only realizes the transformation from macro-strategy to spatial pattern, but also embeds the intrinsic mechanism of regional ecological response into the model. This enables the analytical evaluation of the complex relationship between land use and ecological effects under policy intervention, providing a scientific and dynamic decision support tool for regional land space optimization and ecological sustainable management.

[0020] 3. This invention achieves dynamic early warning and zoning optimization of land use spatial ecological risks by constructing a risk-baseline-scenario coupled diagnostic model. This diagnostic model spatially overlays and diagnoses the risks of future habitat degradation under multiple scenarios with the potential of the current ecological baseline, accurately delineates multiple types of control zones, and formulates differentiated strategies for each type of control zone. This promotes the transformation of ecological planning from static blueprints to dynamic governance, and provides a directly applicable scientific decision-making tool for coordinating the contradictions between ecological protection and urban-rural development and achieving sustainable development goals. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application, 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 this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 The overall flowchart provided for embodiments of the present invention.

[0023] Figure 2 A flowchart of the coupled SHAP model and geographic detector provided in an embodiment of the present invention.

[0024] Figure 3 A flowchart for multi-scenario land use simulation with factor constraints and policy guidance provided for embodiments of the present invention.

[0025] Figure 4 Flowchart for ecological risk diagnosis and land use pattern optimization provided in embodiments of the present invention Detailed Implementation

[0026] To make the technical problems, solutions, and beneficial effects of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0027] Example 1 See Figure 1 As shown, this embodiment provides a land use scenario simulation method for improving ecological functions. Taking Southwest my country as the study area, this region is characterized by significant ecological vulnerability, substantial human-land relationship conflicts, and diverse terrain including mountains, plateaus, and basins. It also features rich and complex land use types, making it a typical region for conducting land use scenario simulations aimed at improving ecological functions. In this embodiment, 2020 is used as the base year (the baseline year for recent land use data), and 2035 is set as the future prediction year. The specific implementation steps are as follows: S1: Obtain land use data and driving factor data for the study area. The land use data uses Landsat 8 OLI / TIRS remote sensing images of the study area from 2015 and 2020. The data is preprocessed using ENVI 5.3 software (including radiometric correction, geometric correction, and atmospheric correction). Combined with Google Earth high-resolution imagery, field exploration data, and the "Land Use Status Classification", the supervised classification method (Support Vector Machine SVM) is used for interpretation to obtain the land use type map. The land types mentioned are mainly divided into six categories: cultivated land, forest land, grassland, water area, construction land, and unused land, covering all land use forms in the study area.

[0028] The driving factor data includes natural background driving factors and human activity driving factors. The two types of factors together constitute a complete driving system affecting the habitat quality of the study area. The natural background driving factors include altitude, slope, topographic relief, lithology, annual precipitation, soil type, and distance from the river; The human activity-related driving factors include nighttime light index, distance from roads, distance from settlements, proportion of construction land, proportion of arable land, landscape fragmentation index, population density, and GDP.

[0029] S2: Dynamically assess habitat quality in the study area based on land use data, and analyze the spatiotemporal evolution characteristics of habitat quality. This includes the following steps: In this embodiment, habitat quality assessment adopts the habitat quality module of the InVEST model. This model can combine land use type, threat factors and their impact range to quantify the spatial distribution and dynamic changes of habitat quality. By calculating the habitat suitability index and the threat factor stress index, the habitat quality index is finally obtained, with a value range of 0-1. The higher the index, the better the habitat quality.

[0030] In this embodiment, the habitat quality assessment includes the following steps: The land use data from 2015 and 2020 were used as input data for the InVEST model. Based on the actual conditions of the study area, threat factors (including construction land, cultivated land, and roads) and their weights were set (e.g., construction land weight 0.8, cultivated land weight 0.5, and roads weight 0.6). Habitat suitability was set for each land use type, for example, forest, grassland, and water suitability was 1.0, cultivated land suitability was 0.6, and construction land and unused land suitability was 0.1. The maximum impact distance of threat factors was set, such as 5km for construction land, 3km for cultivated land, and 2km for roads. The InVEST model was run to obtain the habitat quality assessment results (habitat quality index map) for the study area in 2015 and 2020.

[0031] As one possible implementation of this embodiment, the spatiotemporal evolution characteristics analysis of habitat quality adopts three methods: spatial autocorrelation analysis, hot and cold spot detection, and transition matrix analysis. The specific steps are as follows: Spatial autocorrelation analysis was performed using the spatial autocorrelation analysis tool in ArcGIS software to calculate the global Moran's I index and the local Moran's I index. The global Moran's I index was used to determine the overall spatial clustering characteristics of habitat quality in the study area (value range: -1 to -1, I>0 for positive clustering, I<0 for negative clustering, I=0 for random distribution). The local Moran's I index was used to identify the local clustering types of habitat quality (high-high clustering, high-low clustering, low-high clustering, low-low clustering), thereby revealing the spatial clustering patterns of habitat quality.

[0032] Hot and cold spot detection: Using the hot spot analysis tool (Getis-Ord Gi*) in ArcGIS software, the Gi* statistic for each pixel is calculated. Based on the significance level of the Gi* statistic (P<0.05), habitat quality is divided into hot spots (high habitat quality and significant clustering), cold spots (low habitat quality and significant clustering), and transitional areas (no significant clustering). This identifies stable areas (hot spots and cold spots) and evolution trends of habitat quality.

[0033] Transition Matrix Analysis: Using the overlay analysis tool in ArcGIS software, habitat quality index maps from 2015 and 2020 were overlaid to generate a habitat quality transition matrix. This matrix clearly reflects the number and proportion of transitions between different habitat quality levels (high, medium, and low, divided using the natural breakpoint method), thereby revealing the evolution trajectory of habitat quality and accurately identifying key spatial types such as persistently low-value areas (consistently low habitat quality from 2015 to 2020) and significantly degraded areas (medium / high habitat quality in 2015, deteriorating to low habitat quality in 2020), providing a clear direction for subsequent analysis of driving mechanisms.

[0034] In this embodiment, the InVEST model is used to achieve accurate quantitative assessment of habitat quality. Combined with three spatial analysis methods, the spatiotemporal evolution characteristics of habitat quality are comprehensively analyzed, and the clustering patterns, stable areas, and key degradation areas of habitat quality are clarified. This provides solid basic data support for the subsequent identification of key driving factors and the construction of ecological constraint rules, and solves the problem of incomplete characterization of habitat quality evolution patterns.

[0035] S3: Preprocess the driving factor data, including data cleaning, normalization, and collinearity removal; for example, data cleaning removes outliers from the driving factor data, such as using the 3σ principle and linear interpolation to fill in missing values, to ensure the integrity and accuracy of the data.

[0036] The collinearity elimination method employs Spearman correlation analysis to calculate the Spearman correlation coefficient between each driving factor. If the absolute value of the correlation coefficient between two driving factors is greater than or equal to a set threshold (e.g., 0.7), it is considered that there is severe collinearity between the two factors, and the factors with weak explanatory power are eliminated (which can be determined through preliminary regression analysis). Finally, after eliminating redundant and collinear factors, a set of driving factors containing 15 core indicators is obtained (7 natural background factors and 8 human activity factors), ensuring the independence and representativeness of the driving factor set.

[0037] Then, using a multi-model coupling approach, with habitat quality assessment results (2020 habitat quality index) and a set of driving factors as input, the key driving factors affecting habitat quality and their impact thresholds are identified. (See [link to relevant documentation]). Figure 2 As shown, the specific steps are as follows: In this embodiment, the multi-model coupling method uses the SHAP interpretable machine learning model coupled with the geographic detector; The SHAP model is an interpretable machine learning model that treats each driving factor as an independent player in a game. By calculating the marginal contribution (SHAP value) of each factor in the habitat quality prediction process, it quantifies the global importance and direction of influence of the factors, and identifies the nonlinear influence patterns and thresholds between factors and habitat quality. The specific steps are as follows: Using the set of driving factors as input features and the 2020 habitat quality index as input labels, an XGBoost regression model (for habitat quality prediction) was constructed. The model parameters were optimized using a grid search method to ensure the model's prediction accuracy. The trained XGBoost regression model is input into the SHAP model, and the SHAP value of each driving factor is calculated. A positive SHAP value indicates that the factor has a positive promoting effect on habitat quality, while a negative SHAP value indicates that the factor has an inhibitory effect on habitat quality. The larger the absolute value of the SHAP value, the higher the global importance of the factor.

[0038] Then, by using the SHAP dependency graph, the nonlinear relationship between each driving factor and the habitat quality index was analyzed to identify the impact threshold of key driving factors, that is, the critical value when the value of the driving factor changes abruptly. At this time, the rate of change of the SHAP value increases significantly. For example, when the distance from the river is ≤500m, the SHAP value is positive and the absolute value is large, which has a significant positive promoting effect on habitat quality. When the distance from the river is >500m, the SHAP value decreases rapidly and the positive promoting effect is significantly weakened. Therefore, the impact threshold of the distance from the river was determined to be 500m.

[0039] In this embodiment, a geographic detector is used to analyze the spatial interaction effects of driving factors, assisting in the accurate identification of key driving factors and their impact thresholds. Specifically, the geographic detector model calculates the explanatory power of driving factors to assess their impact on habitat quality. The explanatory power (q-value) ranges from 0 to 1; the larger the q-value, the stronger the explanatory power of the factor. The interaction detection capability can be compared by contrasting the q-value of a single driving factor with the q-value of the combined effect of two factors to determine the synergistic, inhibitory, or independent effects between the driving factors. The set of driving factors and the 2020 habitat quality index (divided into high, medium and low levels) are input into the geospatial detector model. The q value of each driving factor is calculated, and driving factors with q values ​​greater than or equal to a predetermined threshold (such as 0.1) are selected as preliminary key driving factors.

[0040] The initial key driving factors are combined in pairs, and the q-values ​​of the two factors are calculated. By comparing the q-values ​​of the individual factors, the interaction type is determined. If the q-value after superposition is greater than the maximum q-value of the individual factor, it is a synergistic enhancement effect; if the q-value after superposition is less than the minimum q-value of the individual factor, it is a synergistic inhibition effect; if the q-value after superposition is between the q-values ​​of the individual factors, it is an independent effect.

[0041] By combining the key driving factors and impact thresholds identified by the SHAP model, the rationality of the key driving factors is verified through the interaction results of the geospatial detector. At the same time, the impact thresholds are revised. For example, the SHAP model identifies altitude as a key driving factor with an impact threshold of 2500m. The geospatial detector analysis found that altitude and slope have a synergistic enhancement effect. When altitude > 2500m and slope > 25°, the inhibitory effect on habitat quality is significantly enhanced. Therefore, the impact threshold of altitude is revised to 2500m to ensure the scientific validity of the impact threshold.

[0042] In this embodiment, redundant and collinear factors are removed through data preprocessing to ensure the independence of the driving factor set. By coupling the SHAP model with the geographic detector, not only are the global importance, direction of action, and nonlinear influence threshold of the driving factors quantified, but the spatial interaction effects between factors are also analyzed. This solves the problems of inaccurate selection of driving factors and insufficient quantitative constraints in the prior art, and provides clear mechanistic support for the subsequent construction of spatial constraint rules.

[0043] S4: Based on the identified key driving factors and their impact thresholds as input, these are mapped into prohibitive or probabilistic rules for the transformation of different land types, forming spatial constraint rules. See [link to relevant documentation]. Figure 3 As shown, the details are as follows: Examples of prohibitive and probabilistic rules are as follows: Prohibited rules: In areas ≤500m from a river, the transfer of construction land and unused land is prohibited, and the transfer of forest land, grassland and water area is prohibited; in areas with an altitude ≤2500m and a slope ≤25°, the transfer of unused land is prohibited, and the transfer of construction land is restricted. These areas have excellent natural conditions and are the main distribution areas of arable land and forest land. Destructive development is restricted to ensure the needs of ecological and agricultural land use. Probabilistic rules: In areas with a nighttime light index ≤ 0.3 and a population density ≤ 100 people / km², increase the probability of conversion to forest land and grassland (e.g., increase by 30% compared to the baseline probability) and decrease the probability of conversion to construction land (e.g., decrease by 50% compared to the baseline probability). These areas have relatively weak human activity interference, are suitable for ecological land restoration, and can curb disorderly urban expansion. In areas where construction land accounts for ≤ 10%, increase the probability of conversion to arable land (e.g., increase by 20% compared to the baseline probability) and decrease the probability of conversion to construction land (e.g., decrease by 40% compared to the baseline probability). These areas have a good agricultural foundation, and can balance ecological protection and agricultural production, avoiding excessive encroachment of construction land on arable land.

[0044] In this embodiment, the spatial constraint rules are spatialized using ArcGIS software to generate a spatial constraint layer that maintains the same coordinate system and resolution as the land use data and driving factor data, making it easier to embed into subsequent spatial simulation models.

[0045] Furthermore, in this embodiment, guided by relevant policy documents such as the "Southwest Region Ecological Protection and High-Quality Development Plan" and the "Southwest Region Provincial Territorial Spatial Planning," three differentiated land use scenarios are constructed: a natural development scenario, an ecological priority scenario, and a urban-rural development and rural revitalization scenario. These three scenarios correspond to different development paths. The core of the parameterization setting is to reflect different policy orientations by adjusting the conversion probability, attractiveness of conversion, and resistance to conversion of land use types, as detailed below: The natural development scenario (baseline reference scenario) is based on the land use transfer matrix of the study area from 2015 to 2020. No additional ecological constraints are imposed. The conversion probability, attraction to conversion, and resistance to conversion of land use types are all set according to historical patterns. For example, the conversion probability from cultivated land to construction land is 2% / year, and the conversion probability from forest land to cultivated land is 1% / year. The inertial development path is simulated to compare the ecological effects of other scenarios. The parameters of the natural development scenario are output after parameterization.

[0046] Ecological Priority Scenario: Based on the results of key driving factors and impact thresholds, the constructed rigid ecological constraint rules are fully embedded. At the same time, the attractiveness of ecological land such as forest land, grassland, and water area to be transferred in is increased (e.g., 40% higher than the natural development scenario), and the resistance to their transfer out is increased (e.g., 60% higher than the natural development scenario). The attractiveness of construction land and unused land to be transferred in is decreased (e.g., 50% lower than the natural development scenario), and the resistance to their transfer out is increased (e.g., 50% higher than the natural development scenario). The spatial pattern under the strictest ecological protection policy is simulated, and the parameters of the ecological priority scenario are output after parameterization.

[0047] Urban and rural development and rural revitalization scenario: Adhere to the ecological protection red line and the bottom line of permanent basic farmland (do not break through the prohibitive constraints within the ecological protection red line and ensure that the area of ​​permanent basic farmland does not decrease), appropriately increase the conversion probability of construction land in urban periphery and rural areas (for example, increase it by 20% compared with the natural development scenario), guide the transformation of agricultural land to eco-friendly high-value-added land (such as ecological agricultural land and forest land) (for example, increase the conversion probability of cultivated land to forest land by 30% compared with the natural development scenario), while retaining the core ecological constraint rules, simulating the path of coordinated development and protection, and outputting the parameters of urban and rural development and rural revitalization scenario after parameterization.

[0048] In this embodiment, the FLUS model (integrating artificial neural networks and an adaptive competition mechanism) is used as a spatial simulation model to handle the land use evolution process under the synergistic effect of multiple driving factors. It simulates the conversion probability of land use types through an artificial neural network (ANN) and resolves the competition conflict between land use types through an adaptive inertial competition mechanism, ultimately achieving the simulation and prediction of land use spatial patterns. Specifically: First, the FLUS model was calibrated and validated. Using 2015 land use data as the base period and 2020 land use data as the validation period, the acquired set of driving factors and the constructed spatial constraint rules (without additional constraints under the natural development scenario) were input into the model. The key parameters of the model were adjusted (70% of the neural network training samples, 0.3 for the domain effect weight, and the conversion cost between land use types was set according to historical transfer patterns). The model was run to simulate the spatial pattern of land use in 2020. The simulation results were compared with the actual land use data in 2020. The Kappa coefficient and FoM coefficient (map feature fit) were used to evaluate the model accuracy. A Kappa coefficient ≥ 0.85 and a FoM coefficient ≥ 0.7 indicate that the model calibration is qualified and can be used to simulate future land use patterns.

[0049] Subsequently, using 2020 land use data as the base period, the constructed spatial constraint rules and three scenario parameters were input into the calibrated and validated FLUS model. The corresponding parameters of the model were adjusted (the conversion probability, attraction of conversion, and resistance of conversion need to be adjusted for the ecological priority scenario and the urban-rural development and rural revitalization scenario). The model was run to simulate and predict the land use spatial pattern under the three scenarios in 2035 (a future set year). The resulting land use type maps (spatial resolution 30m×30m) for the natural development scenario, ecological priority scenario, and urban-rural development and rural revitalization scenario in 2035 were output, providing a spatial data foundation for subsequent ecological risk diagnosis.

[0050] In this embodiment, the impact thresholds of key driving factors are transformed into operable spatial constraint rules, integrating the simulation process with ecological mechanisms. By constructing and parameterizing three differentiated scenarios, the transformation of macro-policy guidance into spatial patterns is realized. Through calibration and verification of the FLUS model, the accuracy of the simulation results is ensured, solving the problems of insufficient coupling of ecological processes and shallow embedding of policy guidance in existing simulation technologies. This provides multi-scenario, high-precision land use spatial data for subsequent ecological risk assessment and optimization scheme formulation.

[0051] S5: See also Figure 4 As shown, in this embodiment, the land use spatial patterns under three simulated scenarios in 2035 are used as input. The InVEST model habitat quality module, consistent with step S2, is used to calculate the future habitat patterns (habitat quality index map) under the three scenarios in 2035. Using the 2020 habitat quality assessment results as a reference, the difference between future habitat quality and the baseline habitat quality is compared. Spatial units where habitat quality has significantly decreased compared to the baseline (e.g., habitat quality index decrease ≥ 0.2) are identified as ecological risk areas, generating the ecological risk area results as follows: Land use data for three scenarios in 2035 were input into the InVEST model, and the model was run to obtain the future habitat quality index map for each scenario. Using the spatial analysis tools in ArcGIS software, the difference between the future habitat quality index and the habitat quality index in the base period of 2020 was calculated to obtain a habitat quality change map. The threshold for a significant decline in habitat quality was set at 0.2. Spatial units with a difference of ≤-0.2 were divided into ecological risk areas, generating ecological risk area results under three scenarios. The ecological risk area was the largest under the natural development scenario, the ecological risk area was the smallest under the ecological priority scenario, and the urban-rural development and rural revitalization scenario was in between.

[0052] Using 2020 land use data (base period) of the study area as input, the InVEST model's habitat quality module was used to remove the weights of all human-caused threats (construction land, cultivated land, roads) (setting the threat factor weights to 0), resulting in a potential habitat quality map that reflects the maximum ecological potential of each spatial unit based on its land use cover type. The natural discontinuity method was employed to classify the potential habitat quality into three levels: high, medium, and low, outputting the ecological background importance level results. An exemplary classification standard is as follows: High ecological baseline importance areas: potential habitat quality index ≥0.7, mainly forest land, grassland and water area, with great ecological potential, and are the core carriers of regional ecological security; Medium-level ecological background importance area: potential habitat quality index 0.4-0.7, mainly composed of cultivated land and sparse forest land, with moderate ecological potential and both ecological and agricultural functions; Low ecological background importance areas: potential habitat quality index <0.4, mainly areas of construction land and unused land, with small ecological potential, and are the main areas for human activities.

[0053] Finally, ArcGIS spatial overlay analysis technology was used to spatially intersect the ecological risk area results with the ecological baseline importance level results to construct a two-dimensional diagnostic matrix of ecological risk and ecological baseline. Based on this two-dimensional diagnostic matrix, four types of ecological management zones were divided, as follows: Emergency Restoration Zone: The spatial intersection of ecological risk areas and areas of high ecological importance; this area has a superior ecological foundation and is a key node in the ecological security network, possessing extremely high restoration value, but currently faces severe ecological degradation risks, making time extremely urgent; governance strategy: immediately terminate damaging activities (such as encroachment on construction land, excessive reclamation), initiate time-limited ecological restoration projects, and take measures such as returning farmland to forest, engineering pollution control, topography reshaping, and vegetation restoration to quickly rebuild key ecological functions; establish a normalized monitoring mechanism to regularly assess the restoration effect and ensure the gradual improvement of habitat quality.

[0054] Key Prevention Area: This refers to the portion of the high ecological baseline importance area that has not yet experienced degradation hotspots. This area currently boasts intact habitat quality and complete ecological functions, serving as the cornerstone for regional biodiversity conservation and ecosystem service provision. The core risk stems from future human activity disturbances. Governance Strategy: Implement preventative, comprehensive management, incorporating the entire area into the strictest control scope of the ecological protection red line, prohibiting any development or construction activities in principle. The management focus will shift to biodiversity management, habitat protection, and the maintenance of natural restoration processes. An ecological early warning mechanism will be established to promptly prevent potential ecological risks and solidify the ecological security baseline.

[0055] Optimization and Upgrading Zone: This area overlaps with ecological risk zones and zones of medium to low ecological importance. These zones are often located in densely populated urban-rural fringe areas or agricultural areas, with relatively weak ecological foundations but facing intense development pressures. They are at the forefront of the conflict between ecological protection and economic development. Governance Strategy: Implement guided ecological transformation through the planning and construction of ecological isolation green wedges and corridors, the promotion of ecological agriculture (such as straw return to the field and water-saving irrigation), and the implementation of green infrastructure and industrial guidance strategies such as ecological transformation of industrial zones (such as sewage treatment and vegetation greening). While ensuring reasonable development needs, minimize the ecological footprint and promote the green transformation of the region.

[0056] General Development Zone: Areas within the study area other than the three categories mentioned above; these areas have relatively low ecological constraints and serve as the main spatial carriers for urbanization, industrialization, and agricultural development. Governance Strategy: Advocating adaptive development based on ecological constraints, ensuring reasonable development needs while strictly adhering to the rigid constraints of national land spatial planning (such as the protection of permanent basic farmland and total control of construction land); comprehensively promoting low-impact development technologies (such as sponge city construction) and economical and intensive land use models to improve land use efficiency and achieve synergy between efficient use of national land space and ecological maintenance.

[0057] In this embodiment, the spatial scope and governance strategies of the above four types of ecological control zones are integrated, and the ecological effect assessment results under three scenarios are combined to form a land use spatial pattern optimization scheme for improving ecological functions. This scheme includes a future multi-scenario land use spatial distribution map, an ecological risk spatial distribution map, an ecological control zone planning map, and a differentiated spatial governance strategy system.

[0058] This embodiment identifies ecological risk areas under different development scenarios by analyzing future habitat quality patterns; by coupling the importance of ecological risk with the ecological baseline, a two-dimensional diagnostic matrix is ​​constructed, enabling the scientific delineation of ecological management zones; by formulating differentiated governance strategies, the contradiction between ecological protection and regional development is resolved, achieving a leap from simulation and prediction to spatial governance, and providing scientific support for the coordinated and sustainable development of human-land relations in the region.

[0059] Example 2 This embodiment provides a land use scenario simulation system for improving ecological functions, including a memory and a processor. The memory stores a computer program, and the processor can call the computer program stored in the memory to execute the land use scenario simulation method for improving ecological functions described in Embodiment 1.

[0060] The above are merely preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A land use scenario simulation method for improving ecological function, characterized in that, Includes the following steps: Obtain land use data and driving factor data for the study area; The habitat quality of the study area is dynamically assessed based on land use data to obtain habitat quality assessment results, and the spatiotemporal evolution characteristics of habitat quality are analyzed based on the habitat quality assessment results. After preprocessing the driving factor data and removing redundant and collinear factors, a set of driving factors is obtained. Using the habitat quality assessment results and the set of driving factors as input, a multi-model coupling approach is adopted to identify the key driving factors affecting habitat quality and their impact thresholds. The identified key driving factors and their impact thresholds are transformed into spatial constraint rules. Combined with regional development policy guidance, multiple differentiated land use scenarios are constructed and parameterized. The spatial constraint rules and scenario parameters are input into a calibrated and verified spatial simulation model. Using recent land use data of the study area as the base period, the spatial pattern of land use in future years is simulated and predicted. Based on the land use spatial patterns under various future scenarios, calculate the future habitat quality patterns and identify ecological risk areas by comparing them with the base period habitat quality. The study assesses and classifies the importance of the ecological baseline in the research area. Through spatial overlay analysis, ecological risk areas are coupled with the ecological baseline importance level to construct a diagnostic matrix. Different types of ecological control zones are divided, and differentiated spatial governance strategies are formulated for each ecological control zone to form an optimized land use spatial pattern scheme oriented towards improving ecological functions.

2. The land use scenario simulation method for improving ecological function as described in claim 1, characterized in that, The driving factor data includes background-type driving factors and human activity-type driving factors; The natural background driving factors include at least one of the following: altitude, slope, topographic relief, lithology, annual precipitation, soil type, and distance from the river; The human activity-related driving factors include at least one of the following: nighttime light index, distance from road, distance from settlement, proportion of construction land, proportion of arable land, landscape fragmentation index, population density, and GDP.

3. The land use scenario simulation method for improving ecological function as described in claim 1, characterized in that, The multi-model coupling includes the following steps: Based on the set of driving factors and habitat quality assessment results as input, the marginal contribution of each driving factor is quantified through an interpretable machine learning model, the global importance and direction of action of each driving factor are determined, and key driving factors and their impact thresholds are identified. By using the interaction detection of geographic detectors, and taking the set of driving factors and habitat quality assessment results as input, the explanatory power of the superposition of single driving factors is compared to determine the synergistic enhancement, inhibition or independent effect relationship between driving factors, which helps to accurately identify key driving factors and their impact thresholds.

4. The land use scenario simulation method for improving ecological function according to claim 1, characterized in that, The construction of the various differentiated land use scenarios takes regional development policy guidance as input and includes at least three types: natural development scenario, ecological priority scenario, and urban-rural development and rural revitalization scenario. After parameter setting, the parameters corresponding to each scenario are output. The natural development scenario is based on the historical land use transfer patterns of the study area, without imposing additional ecological constraints, and outputs natural development scenario parameters. The ecological priority scenario is based on the results of key driving factors and impact thresholds, and incorporates rigid ecological constraint rules to improve the attractiveness of ecological land transfer and the resistance to transfer out, and outputs ecological priority scenario parameters. The urban and rural development and rural revitalization scenarios adhere to the ecological protection red line and the bottom line of permanent basic farmland, increase the probability of conversion of urban and rural construction land, guide agricultural land to transform into ecologically friendly and high-value-added land, and output urban and rural development and rural revitalization scenario parameters; The three scenario parameters mentioned above are used as inputs to the space simulation model.

5. The land use scenario simulation method for improving ecological function according to claim 1, characterized in that, The spatial constraint rules take key driving factors and influence thresholds as inputs, map the influence thresholds of key driving factors into prohibitive or probabilistic rules for the conversion of different land use types, and output spatial constraint rules as one of the inputs to the spatial simulation model.

6. The land use scenario simulation method for improving ecological function according to claim 1, characterized in that, The spatial overlay analysis uses the results of ecological risk areas and the importance level of ecological baseline as inputs for analysis; The ecological risk area results are obtained based on the following steps: Using the future habitat quality pattern results and habitat quality assessment results as inputs, the differences between the two are compared, and spatial units where habitat quality has declined significantly compared to the base period are identified as ecological risk areas, and the ecological risk area results are output. The results of the ecological baseline importance level were obtained based on the following steps: Using land use data of the study area as input, the potential habitat quality is calculated by removing the weight of human threats, and the potential habitat quality data is output. Then, the natural breakpoint method is used to divide the potential habitat quality into multiple levels and output the ecological background importance level results.

7. The land use scenario simulation method for improving ecological function according to claim 6, characterized in that, The ecological management zoning includes the following steps: Based on the results of ecological risk areas and the importance level of ecological baseline, a diagnostic matrix is ​​constructed by coupling spatial overlay analysis, and the diagnostic matrix results are output. Based on the diagnostic matrix results, multiple ecological management zones are divided, and the ecological management zone results are output. These results are then used as input for formulating differentiated spatial governance strategies.

8. The land use scenario simulation method for improving ecological function according to claim 7, characterized in that, The ecological management zones include emergency restoration zones, key prevention zones, optimization and upgrading zones, and general development zones. The emergency restoration zone is the spatial intersection of ecological risk areas and areas of high ecological importance; The key prevention area refers to the part of the high ecological background importance area that has not yet experienced degradation hotspots; The optimization and improvement area is the overlapping area of ​​ecological risk areas and medium- and low-level ecological background importance areas; The general development zone refers to the area within the study area excluding the aforementioned emergency repair zone, key prevention zone, and optimization and improvement zone.

9. A land use scenario simulation system for improving ecological functions, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, and the processor is capable of calling the computer program stored in the memory to execute the land use scenario simulation method for improving ecological functions as described in any one of claims 1 to 8.

Images