Territorial space pattern simulation method cooperatively driven by climate change and human activity

Abstract

The invention relates to the technical field of territorial space simulation and prediction, and discloses a territorial space pattern simulation method cooperatively driven by climatic changes and human activities, which comprises the following steps: collecting and sorting related data of the climatic changes and the human activities; based on the collected data, utilizing a feature selection technology to identify driving factors influencing land utilization changes, and evaluating the influence degree of the driving factors so as to screen out an optimal driving model; on the basis of the optimal driving model, sorting the influence degree of each driving factor to construct a scene analysis framework of the land utilization change; simulating land utilization changes under different scenes by combining a land space pattern simulation model; in the simulation process, based on the sorting result of the driving factors in the optimal driving model, model input parameters are dynamically adjusted, a prediction result is continuously optimized through a feedback mechanism, the simulation precision can be improved, the influence of multiple factors can be predicted, and a scientific basis is provided for land utilization management and policy making.

Description

Technical Field

[0001] The present invention relates to the field of land space simulation and prediction technology, and more specifically, to a land space pattern simulation method driven by the synergy of climate change and human activities. Background Art

[0002] With the intensification of global climate change and the increasing intensity of human activities, the spatial pattern of land is undergoing significant changes. These changes not only affect the stability and sustainable development of regional ecosystems but also have profound impacts on land resource management, urban and rural planning, and ecological protection. Therefore, scientifically and effectively simulating and predicting changes in the spatial pattern of land and identifying the main drivers of climate change and human activities have become important directions in current land and spatial research.

[0003] Traditional land-use change simulation methods are mostly based on static or single-factor driving factors, rarely considering the synergistic effects between climate change and human activities, and thus struggle to accurately capture the complex dynamic processes driven by multiple factors. Furthermore, with the rapid development of big data technologies and machine learning methods, a growing number of studies are attempting to leverage multi-source data and intelligent algorithms to improve the accuracy and adaptability of land-use simulations. However, these studies still face limitations in identifying driving factors, building models, and conducting scenario analysis.

[0004] While existing simulation methods, such as the CA-Markov model, the CLUE-S model, and the PLUS model, can describe land use evolution to a certain extent, they generally suffer from issues such as insufficient response to driving factors, difficulty in dynamically adjusting input parameters, and poor model adaptability. In the context of climate change, complex and changing climate scenarios and highly uncertain human activity trends place even higher demands on traditional methods for modeling and prediction.

[0005] Therefore, how to effectively simulate the changing trends and spatial distribution characteristics of land use patterns under the interaction of multiple factors has become a technical problem that needs to be solved urgently. Summary of the Invention

[0006] The present invention provides a method for simulating the spatial pattern of land use driven by the synergy of climate change and human activities, aiming to solve the problem in the existing technology that it is difficult to effectively simulate the changing trends and spatial distribution characteristics of land use patterns under the interaction of multiple factors.

[0007] The present invention provides a method for simulating land spatial patterns driven by the synergy between climate change and human activities, comprising:

[0008] Collect and organize data related to climate change and human activities;

[0009] The data related to climate change and human activities include climate model data from the CMIP6 database and socioeconomic data from the SSP database;

[0010] Based on data related to climate change and human activities, feature selection techniques are used to identify the driving factors affecting land use change, evaluate the impact of each factor, and select the best driving model;

[0011] Based on the optimal driving model, the impact of each driving factor is ranked and a scenario analysis framework for land use change is constructed;

[0012] Combined with the national land spatial pattern simulation model, simulate land use changes under different scenarios;

[0013] During the simulation process, the model input parameters are dynamically adjusted according to the ranking results of the driving factors, and the prediction accuracy of the model is continuously optimized through a feedback mechanism.

[0014] Furthermore, the data related to climate change and human activities also include key climate factors that affect ecosystems and land use changes, such as temperature, precipitation and radiation, and other climate variables; the relevant data are used to assess the impact of climate change on land use.

[0015] Furthermore, the socioeconomic data include human activity-related indicators such as population growth rate, GDP growth rate and urbanization rate; the data are extracted from the SSP database and used to analyze the driving effect of human activities on land use change.

[0016] Furthermore, the feature selection technique includes a Lasso regression function, which selects features by minimizing an objective function:

[0017]

[0018] Among them, y i is the response variable, x ij is the characteristic variable, β j is the regression coefficient, λ is the regularization parameter, n is the number of samples, p is the number of features, and min() is the minimization objective function.

[0019] Furthermore, the multiple machine learning models include a multi-layer perceptron regression function, a Lasso regression function, a random forest regression function, and an XGBoost regression function; the multiple machine learning models are used to capture the nonlinear impact of climate change and human activities on land use change.

[0020] Furthermore, the climate change scenarios include low emission scenarios, medium emission scenarios and high emission scenarios; the climate change scenarios are used to simulate the impact of climate change under different emission pathways.

[0021] Furthermore, the human activity scenarios include three development scenarios: low growth, medium growth and high growth; the human activity scenarios are used to simulate the impact of different socio-economic development paths on land use.

[0022] Furthermore, the national land space pattern simulation model includes a grey prediction model, a Markov model and a PLUS model; the national land space pattern simulation model includes:

[0023]

[0024] Among them, x (0) Represents the original data sequence, x (1) represents the cumulative generation sequence, a represents the development coefficient, b represents the gray action, x (1) (k+1) represents the value of the cumulative generated sequence at the k+1th moment; x (0) (1) represents the value of the original data sequence at the first moment; The constant term in the grey prediction model represents the equilibrium state of the system; e -ak represents the exponential decay factor, which indicates the decay trend of the system over time; k represents the time step, which indicates the predicted time point.

[0025] Furthermore, the result verification step is also included:

[0026] A specific research area was selected and the simulation results of this method were compared with those of traditional methods to verify the effectiveness of the model.

[0027] Compare and analyze the prediction performance of different methods through accuracy evaluation indicators;

[0028] Based on the simulation results, the spatial pattern and trend of land use changes under different scenarios are analyzed to provide a scientific basis for regional land resource management and planning.

[0029] Furthermore, the application step is also included:

[0030] For specific research objectives, representative regions should be selected for case studies. The selected regions should take into account factors such as significant climate change characteristics, obvious human activity impacts, and diverse land use change types;

[0031] Using the same historical data, the method of the present invention and the traditional method were used for prediction. The prediction performance of the different methods was quantitatively evaluated using multiple accuracy evaluation indicators, and the applicability and advantages of the method of the present invention in different scenarios and different regions were analyzed.

[0032] Generate land use change prediction maps under different scenario combinations, identify hot spots and key driving factors of land use change, evaluate the potential effects of different land use policies, develop visualization tools to intuitively display simulation results, and provide scientific support for national land space planning, resource management and environmental protection.

[0033] The beneficial effects of the present invention are as follows: by comprehensively considering the synergistic driving effects of climate change and human activities, the present invention achieves high-precision simulation of national land spatial patterns. Specifically, by introducing multiple machine learning models, the present invention can effectively capture the nonlinear relationship between climate change and human activities, thereby improving the accuracy and reliability of land use change predictions. In addition, through multi-scenario analysis and a dynamically adjusted feedback mechanism, the present invention can comprehensively assess land use changes under different scenarios, providing a scientific basis for land use management and policy formulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a flow chart of a method for simulating land spatial pattern driven by the synergy of climate change and human activities provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0035] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that these embodiments are discussed solely to enable those skilled in the art to better understand and implement the subject matter described herein, and that the functions and arrangements of the elements discussed may be varied without departing from the scope of this specification. Various examples may omit, substitute, or add various processes or components as needed. Furthermore, features described in some examples may be combined in other examples.

[0036] At least one embodiment of the present invention discloses a method for simulating land spatial pattern driven by the synergy between climate change and human activities, such as Figure 1 Shown, including:

[0037] Example 1: Driving factor selection and modeling based on machine learning.

[0038] In this embodiment, the driving factors related to climate change and human activities are first screened and analyzed through data preprocessing and feature engineering.

[0039] Data preprocessing: Collect and organize data related to climate change and human activities, including:

[0040] Climate change data: Data on temperature, precipitation, and radiation climate variables were selected from the CMIP6 database;

[0041] Socioeconomic data: Population growth rate, GDP growth rate, and urbanization rate data were extracted from the SSP database;

[0042] Historical land use data: historical land use change information obtained based on remote sensing data.

[0043] Feature Engineering: Lasso regression feature selection technology was used to identify the driving factors most relevant to land-use change from the above data. By setting different regularization parameters λ, the feature variables were screened to retain the driving factors that were significantly related to land-use change.

[0044] Multi-model comparison and evaluation: We trained the selected driving factors using MLP regression, random forest regression, and XGBoost regression machine learning models. We evaluated the predictive performance of each model through 5-fold cross-validation, and selected the model with the highest predictive accuracy as the basis for subsequent analysis.

[0045] Ranking and Analysis of Driving Factors: Based on the best model, the impact of each driving factor was ranked. The ranking results showed that among climate change factors, changes in temperature and precipitation patterns had the most significant impacts on land use; among human activity factors, urbanization rate, population growth rate, and GDP growth rate had the most significant impacts.

[0046] Example 2: Multi-scenario national land spatial pattern simulation.

[0047] In this embodiment, based on the driving factor ranking results obtained in Example 1, a national land spatial pattern simulation under multiple scenarios is performed.

[0048] Scenario analysis framework construction:

[0049] Climate change scenarios: Three typical scenarios were selected: SSP126 (low emission scenario), SSP245 (medium emission scenario), and SSP585 (high emission scenario);

[0050] Human activity scenarios: Based on socioeconomic development forecasts, three development scenarios are set: low growth, medium growth, and high growth;

[0051] Through the combination of these two sets of scenarios, a total of 9 different future development scenarios are formed, providing input conditions for land use change simulation.

[0052] Model integration and multi-objective optimization: Combining three traditional models, namely the grey prediction model, the Markov model and the PLUS model, preliminary simulations were conducted under nine scenarios;

[0053] A multi-objective genetic algorithm is used for model integration, with prediction accuracy and operability set as optimization goals, to find the optimal solution among multiple goals;

[0054] By setting appropriate crossover rate, mutation rate and population size, the optimal model parameter combination is obtained after multiple iterations.

[0055] Dynamic adjustment and feedback mechanism: During the simulation process, the model input parameters are dynamically adjusted according to the ranking of driving factors obtained in Example 1;

[0056] After each simulation iteration, the model is revised based on the comparison between the simulation results and historical data;

[0057] Through multiple feedback adjustments, the model performance is continuously optimized and the prediction accuracy is improved.

[0058] Example 3: To verify the effectiveness of the method of the present invention, a certain area was selected as a research case, and the simulation results were compared and analyzed with the traditional method.

[0059] Case area selection: The karst area in southeastern Yunnan was selected as the study area, where climate change is obvious, human activities are increasing, and land use changes are significant.

[0060] Comparison of simulation results: The prediction results of the method of the present invention (combining machine learning with multi-scenario simulation) were compared with the independent prediction results of the traditional Markov model and the grey prediction model;

[0061] Compare and analyze the prediction performance of different methods through accuracy evaluation indicators (such as Kappa coefficient and overall accuracy);

[0062] The results show that the prediction accuracy of the method of the present invention is higher than that of traditional methods under different scenarios, especially when considering the complex interaction between climate change and human activities, the prediction accuracy is significantly improved.

[0063] Based on the simulation results, the spatial pattern and trend of land use change under different scenarios are analyzed;

[0064] Provide a scientific basis for regional land resource management and planning, and support decision makers in formulating adaptive policies and measures;

[0065] Through the visualization platform, the dynamics of land use changes under different scenarios can be intuitively displayed, making it easier for decision makers to understand and apply simulation results.

[0066] The above examples demonstrate the feasibility and effectiveness of the proposed method for simulating land spatial patterns driven by the synergy of climate change and human activities. This method not only improves the accuracy of land use change simulations but also provides technical support for land spatial planning and policy decision-making, demonstrating its broad application prospects.

[0067] like Figure 1As shown, this embodiment provides a driving factor selection and modeling method based on machine learning, including the following steps:

[0068] Step 1: Collect and organize data related to climate change and human activities;

[0069] Specifically, it involves obtaining CMIP6 climate model data (temperature, precipitation, and radiation), SSP socioeconomic scenario data (population growth rate, GDP growth rate, and urbanization rate), and historical land use remote sensing data; performing data standardization and spatial matching to form a set of driving factors at a unified spatiotemporal scale;

[0070] Step 2: Based on data related to climate change and human activities, use feature selection techniques to identify the impact of various driving factors on land use change and select the best driving model;

[0071] Step 3: Based on the optimal driving model, rank the impact of each driving factor to construct a scenario analysis framework for land use change;

[0072] Step 4: Combine the national land spatial pattern simulation model to simulate land use changes under different scenarios;

[0073] Step 5: During the simulation process of the national land spatial pattern simulation model, the model input is dynamically adjusted based on the ranking results of the driving factors by the optimal driving model, and the prediction accuracy is continuously improved through the feedback mechanism.

[0074] Specifically, Multilayer Perceptron Regression (MLP Regression): a feedforward neural network capable of learning nonlinear relationships;

[0075] Random forest regression: An ensemble learning method that improves prediction accuracy by building multiple decision trees and averaging them;

[0076] XGBoost regression: A gradient boosting-based ensemble learning method that can handle complex nonlinear relationships.

[0077] For each model, a 5-fold cross-validation method was used to evaluate its performance, and the root mean square error (RMSE), mean absolute error (MAE) and coefficient of determination (R 2 ) indicators to select the best performing model. Driving factor importance assessment is based on the best model, evaluating the importance of each driving factor to land use change. For random forest and XGBoost models, feature importance scores can be directly obtained; for MLP regression, feature importance can be assessed using sensitivity analysis. Finally, all driving factors are ranked by importance to provide a basis for subsequent simulations.

[0078] Example 2: Multi-scenario national land spatial pattern simulation method, the specific steps are as follows:

[0079] The scenario framework is constructed based on the combination of the Shared Socioeconomic Pathway (SSP) and Representative Concentration Pathway (RCP) in the IPCC Sixth Assessment Report to construct multiple scenarios:

[0080] SSP1-RCP2.6 (SSP126): Sustainable development scenario, low emission pathway;

[0081] SSP2-RCP4.5 (SSP245): intermediate development scenario, medium emission pathway;

[0082] SSP5-RCP8.5 (SSP585): Fossil fuel-intensive development scenario, high emissions pathway.

[0083] For each climate change scenario, different human activity scenarios are considered simultaneously, such as high, medium and low population growth and high, medium and low economic development, to form a scenario matrix.

[0084] Land use change prediction based on the grey prediction model uses the grey prediction model GM(1,1) to predict the area of ​​various land uses. The grey prediction model is a prediction method for small sample and incomplete information systems and is suitable for systems with uncertainty such as land use change. The prediction steps are as follows:

[0085] Performing the Accumulation Generation (AGO) process on the original data sequence to obtain a new sequence;

[0086] Establish grey differential equation model;

[0087] Solve the model parameters by the least squares method;

[0088] Build forecasting models and make forecasts;

[0089] The final prediction result is obtained through cumulative reduction.

[0090] Land use type conversion simulation based on Markov model uses Markov model to simulate the conversion probability between different land use types. The specific steps are as follows:

[0091] Based on historical land use data, calculate the conversion matrix between land use types in different periods;

[0092] Combined with the prediction results of the grey prediction model, the probability values ​​in the conversion matrix are adjusted;

[0093] According to the transformation matrix, the distribution of land use types in the future period is simulated;

[0094] Spatial allocation based on PLUS model The PLUS model is used to allocate the land use type area predicted by the Markov model to specific spatial locations;

[0095] The PLUS model takes into account multiple factors, including topography, distance, and neighborhood, enabling a more accurate simulation of the spatial distribution of land use. Step 5: Multi-Objective Optimization and Feedback Adjustment: A multi-objective genetic algorithm (MOGA) is used to optimize the simulation results. The optimization objectives include maximizing prediction accuracy, minimizing resource consumption, and maximizing ecological benefits.

[0096] The MOGA algorithm can find a balanced solution among multiple conflicting objectives. At the same time, a feedback adjustment mechanism is established: during each iteration, the model parameters are dynamically adjusted based on the importance of the driving factors identified by the machine learning model, further improving prediction accuracy.

[0097] Example 3: This example verifies and applies the proposed method.

[0098] For model validation, historical data of a certain region (e.g., 2000-2010) is selected as the training set, and data from 2010-2020 is selected as the validation set;

[0099] The prediction accuracy of the model is evaluated by comparing the consistency between the model prediction results and the actual land use changes. The evaluation indicators include overall accuracy, Kappa coefficient, producer accuracy, and user accuracy.

[0100] The method of the present invention is compared with traditional methods (such as simple Markov model and CA-Markov model) to verify the advantages of the method under the consideration of the synergistic driving of climate change and human activities.

[0101] The future scenario simulation is based on the verified model to simulate land use changes under different scenarios in the future (e.g., 2020-2050). The simulation results include:

[0102] Area change trends of land use types under different scenarios;

[0103] Spatial distribution patterns of land use types under different scenarios;

[0104] Contribution rate and impact pathway of different driving factors on land use change.

[0105] Decision support applications provide decision support for land use planning and management based on simulation results:

[0106] Identify hotspots and key drivers of land use change;

[0107] Analyze the trends and impacts of land use change under different policy scenarios;

[0108] Propose recommendations for land use regulation to adapt to climate change and optimize human activities.

[0109] The core steps of the proposed method for simulating the spatial pattern of land driven by the synergy between climate change and human activities include:

[0110] Data Collection and Collation: This method first obtains climate model data from the CMIP6 database, including temperature, precipitation, and radiation climate variables at different time scales and under different emissions scenarios. It also obtains socioeconomic data from the SSP database, including population growth rate, GDP growth rate, and urbanization rate indicators. These data are spatially aligned, temporally normalized, and quality-controlled to form a complete analytical dataset.

[0111] Driving Factor Assessment and Model Selection: Based on the collected data, this method uses feature selection techniques to identify the driving factors that have a significant impact on land-use change. Lasso regression is used to screen the driving factors. Comparative analysis is performed using multiple machine learning models (such as MLP regression, random forest regression, and XGBoost regression). The model with the best performance is ultimately selected as the optimal driving model.

[0112] Driving Factor Ranking and Scenario Framework Construction: Based on the optimal driving model, the impact of each driving factor was quantitatively ranked, and a scenario analysis framework for land use change was constructed based on this ranking. This framework fully considers the changing characteristics of climate change and human activities under different scenarios, providing a scientific basis for subsequent simulations.

[0113] Multi-scenario simulation and dynamic adjustment: Combined with national spatial pattern simulation models (such as the gray prediction model, Markov model, and PLUS model), land use change under different scenarios is simulated. During the simulation process, model inputs are dynamically adjusted based on the ranking results of driving factors. The feedback mechanism continuously corrects the prediction results to improve simulation accuracy.

[0114] The innovation of this method lies in its comprehensive consideration of the synergistic driving effects of climate change and human activities, and the realization of high-precision simulation of national land spatial patterns through the combination of machine learning and traditional models.

[0115] The climate change related data used in this invention mainly include the following key climate factors:

[0116] Temperature data: including annual average temperature, seasonal temperature changes, and extreme temperature event indicators. These data directly affect vegetation growth, agricultural production, and human settlement patterns, which in turn affect land use decisions.

[0117] Precipitation data: including annual precipitation, seasonal distribution of precipitation, and drought frequency indicators, which have important impacts on agricultural land use, water resource management, and ecosystem function.

[0118] Radiation data: including solar radiation intensity and radiation duration. These factors affect vegetation photosynthesis and energy balance, and indirectly affect land use patterns.

[0119] During the specific implementation process, the prediction data of the above-mentioned climate variables under different climate scenarios (SSP126, SSP245, SSP585) were obtained from the CMIP6 database, and the global-scale climate model data were converted into regional-scale high-resolution data through the spatial downscaling method for coupling analysis with the land use change model.

[0120] By analyzing the relationship between these key climate factors and land use changes, we can quantify the impact of climate change on different types of land use and provide a scientific basis for multi-scenario simulations.

[0121] The socioeconomic data used in this paper mainly include the following human activity related indicators:

[0122] Population growth rate: This indicator represents the relative increase in population within a specific area per unit time. This indicator directly reflects the changing demand for land resources due to population pressure and is a key factor influencing urban expansion and the increase in residential land use.

[0123] GDP growth rate: This indicator represents the relative growth of the total economic volume per unit time. This indicator reflects the level of economic development and changes in industrial structure, which directly affects land use intensity and land use type conversion.

[0124] Urbanization rate: This refers to the proportion of urban population to total population. This indicator reflects the speed and degree of urbanization and is directly related to the expansion of urban land and the conversion of rural land.

[0125] These data are extracted from the SSP database, which provides socioeconomic development forecasts under different shared socioeconomic pathways. In practical applications, socioeconomic data under different SSP scenarios (e.g., SSP1-SSP5) can be selected based on the characteristics of the study area to analyze the driving effect of human activities on land use change.

[0126] By establishing quantitative relationships between socioeconomic indicators and land use change, we can assess the impact of human activities on land use patterns and predict land use change trends under different socioeconomic development paths.

[0127] This paper uses Lasso regression as the core technology of feature selection to screen the most relevant driving factors by minimizing the objective function. The objective function of Lasso regression is:

[0128]

[0129] Among them, y i is the response variable, x ij is the characteristic variable, β j is the regression coefficient, λ is the regularization parameter, n is the number of samples, p is the number of features, and min() is the minimization objective function.

[0130] The key feature of Lasso regression is the introduction of the L1 regularization term; this compresses some unimportant feature coefficients to zero, thereby achieving automatic feature selection. In practical applications, the optimal lambda value is determined through cross-validation to balance the model's goodness of fit and complexity.

[0131] The implementation steps of feature selection are as follows:

[0132] Construct a feature matrix containing all candidate driving factors;

[0133] Set different λ values ​​and train multiple Lasso regression models;

[0134] Evaluate the model performance under different λ values ​​through cross-validation;

[0135] The model corresponding to the optimal λ value is selected, and the features corresponding to the non-zero coefficients are extracted as key driving factors.

[0136] Through Lasso regression feature selection, the driving factors that have a significant impact on land use change can be effectively identified, providing a scientific basis for subsequent modeling.

[0137] This paper uses a variety of machine learning models to capture the nonlinear impacts of climate change and human activities on land use change, mainly including the following models:

[0138] Multilayer Perceptron regression is a feedforward neural network consisting of an input layer, hidden layers, and an output layer. In this paper, the input variables of the MLP regression model are the driving factors, and the output variable is the land-use change indicator. MLP regression is trained using a backpropagation algorithm and can learn the complex nonlinear relationships between driving factors and land-use change. The advantage of MLP regression is that it can capture highly nonlinear patterns, making it particularly well-suited for handling complex interactions.

[0139] In addition to being used for feature selection, Lasso regression is also used as a modeling method for land use change prediction. By introducing L1 regularization, Lasso regression can control model complexity while providing an easily interpretable linear relationship, making it suitable for analyzing the direct impact of driving factors.

[0140] Random forest regression is an ensemble learning method that improves prediction accuracy by constructing multiple decision trees and averaging the results. The advantages of random forests are that they are insensitive to outliers, can handle high-dimensional data, and provide feature importance assessments. In this paper, random forests are used to assess the impact of various driving factors on land use change and make predictions based on this assessment.

[0141] XGBoost regression is a gradient-based ensemble learning method that sequentially constructs multiple weak learners and then weights them together to form a powerful predictive model. XGBoost's advantages lie in its efficiency, flexibility, and built-in regularization mechanism, which effectively prevents overfitting. In many cases, XGBoost excels in capturing the complex nonlinear relationships between climate change and the impact of human activities on land use.

[0142] These machine learning models use different algorithmic principles to capture the nonlinear impacts of climate change and human activities on land use change from different perspectives. In practical applications, cross-validation is used to compare the performance of different models and select the most suitable model for a specific research area and problem.

[0143] This paper considers different climate change scenarios to simulate the impact of climate change under different emission pathways. Specific scenarios include:

[0144] Low Emission Scenario (SSP1-RCP2.6, SSP126): This scenario assumes that global greenhouse gas emissions peak in the mid-21st century and then decline rapidly, limiting the global average temperature rise to well below 2°C by 2100. Under this scenario, countries take proactive emission reduction measures, implement sustainable development strategies, advance urbanization in an orderly manner, and effectively protect the ecological environment.

[0145] Medium Emissions Scenario (SSP2-RCP4.5, abbreviated as SSP245): This scenario assumes that global greenhouse gas emissions peak in the middle of the 21st century and then decline slowly, with the global average temperature rising by about 2.7°C by 2100. Under this scenario, socioeconomic development remains at a moderate level, technological progress and environmental policy implementation are uneven, and climate change adaptation policies are partially implemented.

[0146] High Emissions Scenario (SSP5-RCP8.5, SSP585): This scenario assumes continued growth in global greenhouse gas emissions, leading to a global average temperature rise of approximately 4.8°C by 2100. This scenario involves rapid economic development, high reliance on fossil fuels, inadequate environmental protection measures, and frequent extreme weather events caused by climate change.

[0147] During the simulation, climate prediction data for different scenarios, including temperature, precipitation, and radiation variables, were obtained from the CMIP6 database. By analyzing the changing trends of these climate variables under different scenarios and their impact on land use, the possible paths of future land use pattern changes can be predicted.

[0148] The consideration of different climate change scenarios enables this invention to assess the potential impacts of climate policies and emission reduction measures on land use change, providing a scientific basis for adaptive planning and policy making.

[0149] This paper considers different human activity scenarios to simulate the impact of different socioeconomic development paths on land use. Specific scenarios include:

[0150] Low Growth Scenario: This scenario assumes low population growth, GDP growth, and urbanization rates. In this scenario, economic development is slow, urban expansion is limited, and demand for land resources grows relatively moderately. This scenario is suitable for regions with aging populations or economic bottlenecks.

[0151] Medium Growth Scenario: This scenario assumes moderate population growth, GDP growth, and urbanization rates. Under this scenario, the economy develops steadily, urbanization continues, and land use change exhibits a relatively stable transition pattern. This scenario is suitable for most regions undergoing transition.

[0152] High Growth Scenario: This scenario assumes high population growth, GDP growth, and urbanization rates. This scenario involves rapid economic development and urban expansion, resulting in strong demand for land resources and a high rate of land use change. This scenario is suitable for rapidly developing emerging economies.

[0153] These human activity scenarios are constructed based on socioeconomic projections from the SSP database and adjusted based on the actual development conditions of the study area. During the simulation process, different human activity scenarios are combined with climate change scenarios to form coupled climate-socioeconomic scenarios, comprehensively assessing the possible pathways of future land use change.

[0154] Taking into account different human activity scenarios, the present invention can evaluate the impact of socioeconomic policies on land use and provide decision support for national spatial planning and sustainable development.

[0155] This paper uses a variety of national land spatial pattern simulation models, including grey prediction model, Markov model and PLUS model, to achieve a comprehensive simulation of land use change. Among them, the grey prediction model GM(1,1) is one of the core models, and its mathematical expression is:

[0156]

[0157] Among them, x (0) Represents the original data sequence, x (1) represents the cumulative generation sequence, a represents the development coefficient, b represents the gray action, x (1) (k+1) represents the value of the cumulative generated sequence at the k+1th moment; x (0) (1) represents the value of the original data sequence at the first moment; The constant term in the grey prediction model represents the equilibrium state of the system; e -ak represents the exponential decay factor, which indicates the decay trend of the system over time; k represents the time step, which indicates the predicted time point.

[0158] Grey prediction models are suitable for systems with small samples and incomplete information, and are particularly well-suited for uncertain phenomena such as land use change. In this paper, the grey prediction model is combined with the Markov model and the PLUS model to fully leverage the advantages of each model and achieve high-precision simulation of changes in the spatial pattern of land.

[0159] The results verification and application steps of the present invention are key steps to ensure the reliability and practicality of the model, and specifically include:

[0160] Study Area Selection: Based on specific research objectives, representative regions should be selected for case studies. Region selection should take into account factors such as significant climate change characteristics, significant human impacts, and diverse land use change types. For example, areas sensitive to climate change (such as coastal areas and ecologically fragile regions) or areas with high human activity intensity (such as rapidly urbanizing areas and concentrated agricultural areas) can be selected as case studies.

[0161] Model comparison and analysis: The method of the present invention is compared with the traditional method to verify the effectiveness of the model. The specific steps include:

[0162] Using the same historical data, the method of the present invention and traditional methods (such as a single Markov model and a CA-Markov model) are used for prediction respectively;

[0163] Compare the predicted results with actual land use change data;

[0164] The prediction performance of different methods is quantitatively evaluated through a variety of accuracy evaluation indicators (such as overall accuracy, Kappa coefficient, producer accuracy, and user accuracy);

[0165] Analyze the applicability and advantages of the method of the present invention in different scenarios and different regions.

[0166] Scenario analysis and decision support: Based on the validated model, simulate land use changes under multiple scenarios and apply the results to actual decision support. Specifically including:

[0167] Generate land use change prediction maps under different scenario combinations and analyze the spatial distribution and change trends of various land uses;

[0168] Identify hotspots and key drivers of land use change to provide key areas of focus for regional planning;

[0169] Evaluate the potential effects of different land use policies and provide scientific basis for decision makers;

[0170] Develop visualization tools to intuitively display simulation results and facilitate decision makers' understanding and application.

[0171] Through these result verification and application steps, the present invention not only proves its scientific nature and effectiveness, but also demonstrates its application value in actual planning and management, providing scientific support for national land space planning, resource management and environmental protection.

[0172] The above describes an embodiment of the present invention, but this embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Ordinary technicians in this field can also make more forms of the same embodiments based on the inspiration of this embodiment, all of which are protected by this embodiment.

Claims

1. A method for simulating land spatial patterns driven by the synergy between climate change and human activities, characterized by: The following steps are involved: Collect and organize data related to climate change and human activities; The data related to climate change and human activities include climate model data from the CMIP6 database and socioeconomic data from the SSP database; Based on data related to climate change and human activities, feature selection techniques are used to identify the driving factors affecting land use change, evaluate the impact of each factor, and select the best driving model; Based on the optimal driving model, the impact of each driving factor is ranked and a scenario analysis framework for land use change is constructed; Combined with the national land spatial pattern simulation model, simulate land use changes under different scenarios; During the simulation process, the model input parameters are dynamically adjusted according to the ranking results of the driving factors, and the prediction accuracy of the model is continuously optimized through a feedback mechanism.

2. The method for simulating land spatial pattern driven by synergistic climate change and human activities according to claim 1, characterized in that: The data related to climate change and human activities also include key climate factors such as temperature, precipitation and radiation, which affect ecosystems and land use changes; the relevant data are used to assess the impact of climate change on land use.

3. The method for simulating land spatial pattern driven by climate change and human activities according to claim 1, characterized in that: The socioeconomic data include human activity-related indicators such as population growth rate, GDP growth rate and urbanization rate; the data are extracted from the SSP database and used to analyze the driving effect of human activities on land use change.

4. The method for simulating land spatial pattern driven by synergistic climate change and human activities according to claim 1, characterized in that: The feature selection technique includes the Lasso regression function, which selects features by minimizing the objective function: Among them, y i is the response variable, x ij is the characteristic variable, β j is the regression coefficient, λ is the regularization parameter, n is the number of samples, p is the number of features, and min() is the minimization objective function.

5. The method for simulating land spatial pattern driven by synergistic climate change and human activities according to claim 1, characterized in that: The multiple machine learning models include a multi-layer perceptron regression function, a Lasso regression function, a random forest regression function, and an XGBoost regression function; the multiple machine learning models are used to capture the nonlinear impact of climate change and human activities on land use change.

6. The method for simulating land spatial pattern driven by synergistic climate change and human activities according to claim 1, characterized in that: The climate change scenarios include low emission scenarios, medium emission scenarios and high emission scenarios; the climate change scenarios are used to simulate the impact of climate change under different emission pathways.

7. The method for simulating land spatial pattern driven by synergistic climate change and human activities according to claim 1, characterized in that: The human activity scenarios include three development scenarios: low growth, medium growth and high growth; the human activity scenarios are used to simulate the impact of different socio-economic development paths on land use.

8. The method for simulating land spatial pattern driven by synergistic climate change and human activities according to claim 1, characterized in that: The national land space pattern simulation model includes a grey prediction model, a Markov model and a PLUS model; the national land space pattern simulation model includes: Among them, x (0) Represents the original data sequence, x (1) represents the cumulative generation sequence, a represents the development coefficient, b represents the gray action, x (1) (k+1) represents the value of the cumulative generated sequence at the k+1th moment; x (0) (1) represents the value of the original data sequence at the first moment; The constant term in the grey prediction model represents the equilibrium state of the system; e -ak represents the exponential decay factor, which indicates the decay trend of the system over time; k represents the time step, which indicates the predicted time point.

9. The method for simulating land spatial pattern driven by synergistic climate change and human activities according to claim 1, characterized in that: It also includes the result verification steps: A specific research area was selected and the simulation results of this method were compared with those of traditional methods to verify the effectiveness of the model. Compare and analyze the prediction performance of different methods through accuracy evaluation indicators; Based on the simulation results, the spatial pattern and trend of land use changes under different scenarios are analyzed to provide a scientific basis for regional land resource management and planning.

10. The method for simulating land spatial pattern driven by climate change and human activities according to claim 9, characterized in that: Also includes application steps: For specific research objectives, representative regions should be selected for case studies. The selected regions should take into account factors such as significant climate change characteristics, obvious human activity impacts, and diverse land use change types; Using the same historical data, the method of the present invention and the traditional method were used for prediction. The prediction performance of the different methods was quantitatively evaluated using multiple accuracy evaluation indicators, and the applicability and advantages of the method of the present invention in different scenarios and different regions were analyzed. Generate land use change prediction maps under different scenario combinations, identify hot spots and key driving factors of land use change, evaluate the potential effects of different land use policies, develop visualization tools to intuitively display simulation results, and provide scientific support for national land space planning, resource management and environmental protection.

Images