Water energy and grain associated ecosystem service classification early warning method and device

Abstract

The invention relates to a water energy and grain associated ecosystem service classification early warning method and device, belongs to the technical field of ecological resource regulation and control, and solves the problem that the existing ecosystem service evaluation is not associated with a water-energy-grain association relationship, and the water-energy-grain association relationship is brought into ecosystem service simulation prediction, so that the water energy-grain association relationship is not associated with the water energy-grain association relationship. Therefore, the problem of lack of accurate and effective monitoring and early warning in implementation of the ecological system service of the water energy grain is solved. On the basis of the land utilization type spatial distribution probability and land utilization data of the adjusted DLS model, constructing a comprehensive simulation and prediction framework of water production service, carbon sequestration service, grain production service, water resource demand, carbon emission and grain consumption in multiple scenes, and according to the evaluation result of the comprehensive evaluation model and the classification mapping relation, calculating the comprehensive evaluation result of the DLS model. And determining an ecological system service early warning result of the set grid unit, so that accurate and effective classification information and early warning information are provided for implementation of the ecological system service of the water energy grains, and the level of the ecological system service of the water energy grains is conveniently determined.

Description

Technical Field

[0001] This application relates to the field of ecological resource regulation technology, and in particular to a method and device for classifying and early warning of water, energy and food-related ecosystem services. Background Technology

[0002] Ecosystems are the foundation of human survival and development, and water, energy, and food production are fundamental resources closely linked to ecosystem services. A complex interplay exists between water, energy, food, and ecosystem services. On the one hand, food production relies on water consumption during irrigation and agricultural machinery energy consumption; on the other hand, water, energy, and food, as individual resources, can cause varying degrees of disturbance to the ecological environment during their development and utilization, directly or indirectly affecting ecosystem services. Therefore, developing an ecosystem service simulation and prediction method based on the water-energy-food relationship not only helps explore its connection with ecosystem services from a water-energy-food perspective but also helps resolve mismatches and conflicts among water, energy, food, and ecosystem services, contributing to regional ecological protection and high-quality development.

[0003] With the development of food security strategies, exploring future ecosystem services from the perspective of the water-energy-food system is of great significance for further addressing the trade-offs and synergies among water, energy, food, and ecosystem services, and promoting sustainable regional economic and social development. Faced with current development realities, food production, water resources, sustainable energy utilization, and ecosystem services are highly interconnected and complex. Previous ecosystem service assessments focused on ecosystem service types and measurements, formation mechanisms, and flows, without linking ecosystem services to the water-energy-food relationship or incorporating this relationship into ecosystem service simulation and prediction. This has resulted in a lack of accurate and effective monitoring and early warning for the implementation of water, energy, and food ecosystem services. Summary of the Invention

[0004] Based on the above analysis, the embodiments of the present invention aim to provide a method and apparatus for classifying and warning of water-energy-food related ecosystem services, in order to solve the problems that previous ecosystem service assessments did not link with the water-energy-food relationship, and that the water-energy-food relationship was not incorporated into ecosystem service simulation and prediction, resulting in a lack of accurate and effective monitoring and early warning of the implementation of water-energy-food ecosystem services.

[0005] This application provides a method for classifying and early warning of water, energy, and food-related ecosystem services, including the following steps:

[0006] Call the DLS model and adjust the basic construction function of the DLS model to adjust the linear relationship of the basic construction function and the land supply and demand balance model at the raster scale;

[0007] Obtain simulation data from the established climate model and input it into the adjusted DLS model to calculate the spatial distribution probability of land use types and land use data;

[0008] Based on the adjusted grid-scale land supply and demand balance model, the water production service supply and water resource demand of the set grid cells are calculated.

[0009] Based on land use data, calculate the supply of carbon sequestration services and the demand for carbon emissions.

[0010] Based on the adjusted grid-scale land supply and demand balance model, the supply of food production services and the demand for food are calculated for each grid cell.

[0011] The spatial distribution probability of land use type, water production service supply, water resource demand, carbon sequestration service supply, carbon emission demand, food production service supply, and food demand are input into the comprehensive evaluation model. Based on the evaluation results and classification mapping relationship of the comprehensive evaluation model, the ecosystem service early warning results of the grid unit are determined.

[0012] The water, energy, and food-related ecosystem service classification and early warning method of this application embodiment is based on the spatial distribution probability of land use type and land use data of the adjusted DLS model. It constructs a comprehensive simulation and prediction framework for water production services, carbon sequestration services, food production services, as well as water resource demand, carbon emissions, and food consumption under multiple scenarios. According to the evaluation results and classification mapping relationship of the comprehensive evaluation model, the early warning results of ecosystem services for the set grid unit are determined, providing accurate and effective classification information and early warning information for the implementation of water, energy, and food ecosystem services.

[0013] As one optional embodiment, the method further includes the following steps:

[0014] The elevation data of the set grid cells is determined based on the adjusted DLS model;

[0015] Based on the altitude data, the set grid cells are layered to determine the layered data;

[0016] A random forest regression model is constructed, with grid cells as partitioned data, to predict service demand by partitioning and stratifying the evaluation results, and to obtain the demand prediction results.

[0017] The supply of water production services, water resource demand, carbon sequestration service supply, carbon emission demand, food production service supply, food demand, and demand forecast results are imported into a large language model. Based on the derivation results of the demand forecast results from the large language model, the actual demand for ecosystem service classification and early warning is used.

[0018] As one optional embodiment, the process of adjusting the linear relationship of the basic construction functions includes the following steps:

[0019] We screened nonlinear driving factors, quantified their importance using random forest, and eliminated redundant variables.

[0020]

[0021] Where, N tree The total number of decision trees in the random forest; Importance j Score the importance of the j-th driving factor; Let be the out-of-bag error of the tree in lesson t. The error is the result of randomly permuting the j-th feature.

[0022] A spatial graph convolutional network is used to model and construct a land use spatial map G = (V, E), where nodes V are raster cells, edges E are defined by spatial adjacency relationships, and node features include driving factors. The feature propagation formula for the graph convolutional layer is as follows:

[0023]

[0024] in, This represents an adjacency matrix with self-joins; A is the original adjacency matrix; I is the identity matrix; H is the degree matrix; (l) W represents the node feature matrix of the l-th layer graph convolution; (l) Let be the trainable weight matrix of the l-th layer; σ is the non-linear activation function.

[0025] As one optional embodiment, the spatial distribution probability of land use types is calculated as follows:

[0026]

[0027] in, Let i be the probability of the k-th land use type appearing in grid cell i, i.e., the spatial distribution probability of land use type. The probability of the k-th land use type predicted by the random forest; α represents the probability of the k-th land use type output by the spatial graph convolutional network; α is a dynamic weighting coefficient used to balance the output contributions of the random forest and the spatial graph convolutional network.

[0028] Let the probability of the k-th land use type appearing in raster cell i based on the local action factor be . The probability of raster cell i exhibiting the kth land use type based on the neighborhood interaction factor is: Then the probability of land use combinations and The following results were obtained from the nonlinear constraint model of land use type distribution at the grid scale:

[0029]

[0030] Among them, X il This represents the l-th driving factor on grid cell i. Define the spatial autocorrelation factor as the regression coefficient. r is the influence coefficient of the spatial autocorrelation factor;

[0031] Through multiple iterations, the spatial allocation of land area requirements is achieved, as shown in the following formula:

[0032]

[0033] Among them, L i,k S represents the total probability of land use during the spatial allocation of land supply and demand at the raster scale. k C represents the stability parameter for the k-th land use type. k Indicators representing the competitiveness of different land use types.

[0034] As one optional embodiment, the process of calculating the water production service supply and water resource demand of a given grid cell based on the adjusted grid-scale land supply and demand balance model is as follows:

[0035]

[0036] PET (i) =K C(i) ×ET 0(i) ;

[0037]

[0038] AWC i =Min(Rest_layer_depth) (i) ,Root_depth (i) )×PAWC (i) ;

[0039] AET (i) =Min(K) C(i) ×ET 0(i) ,P (i) );

[0040] Where i represents the grid cell of the region, and can represent the set grid cell; S wy P represents the average annual water production of grid cell i, i.e., the water production service supply. (i) The annual precipitation for grid cell i; AET (i) and PET (i) The following are the annual actual evapotranspiration and potential evapotranspiration of grid cell i, respectively; KC (i) ET is the crop evapotranspiration coefficient.0(i) AWC is the reference crop evapotranspiration for grid cell i. (i) Water content available to plants; W (i) Non-physical parameters characterizing natural climate-soil properties; Z is an empirical coefficient reflecting local precipitation patterns and hydrogeological characteristics; Rest_layer_depth (i) Root_depth represents the maximum soil depth. (i) Indicates root depth; PAWC (i) The vegetation available water content of grid cell i;

[0041] D wy =D i,agr +D i,ind +D i,dom +D i,eco ;

[0042] D i,agr =Agr×A i,aps ;

[0043] D i,ind =Ind×G i ;

[0044] D i,dom =Dom×P i ;

[0045] D i,eco =Aec×A i,es ;

[0046] Among them, D wy It refers to the total water demand, i.e., water resource requirements; D i,agr D i,ind D i,dom and D i,eco These represent the total agricultural, industrial, domestic, and ecological water consumption of grid cell i, respectively; Agr represents the agricultural water consumption per unit area; A i,aps The agricultural production space area of ​​grid cell i; Ind represents the industrial water consumption per unit of GDP; G i Represents the regional GDP of grid cell i; Dom represents per capita domestic water consumption; P i Represents the population of grid cell i; Aec represents the water consumption per unit area of ​​ecological space; A i,es This represents the ecological space area of ​​grid cell i.

[0047] As one optional embodiment, the process of calculating the supply of carbon sequestration services and the demand for carbon emissions based on land use data is as follows:

[0048] S cs =C above +Cbelow +C soil +C dead ;

[0049] Among them, S cs Total carbon storage, i.e., carbon sequestration service supply; C above For aboveground biochar, C below underground biochar, C soil Soil organic carbon, C dead Dead organic matter;

[0050] D cs =D percarbon ×ρ pop ;

[0051] Among them, D cs This refers to the demand for carbon sequestration, i.e., the demand for carbon emissions; D percarbon Per capita carbon emissions; ρ pop To set the population density of the grid cells.

[0052] As one optional embodiment, the process of calculating the supply and demand of food production services for a given grid cell, based on the adjusted grid-scale land supply and demand balance model, is as follows:

[0053] S fp =(NDVI) x / NDVI sum )×FP sum

[0054] D fp =D perfood ×ρ pop ;

[0055] Where i represents the grid cell of the region, and can be used to define the grid cell; S fp The grain output allocated to grid cell i, i.e., the grain production service supply; FP sum Total grain output; NDVI x NDVI is the NDVI value of grid cell i; sum D is the sum of the NDVI values ​​of cultivated land in the study area; fp D represents the food demand for grid cell i, i.e., the food requirement. perfood This refers to the per capita food demand.

[0056] As one optional embodiment, the process of inputting the spatial distribution probability of land use types, water production service supply, water resource demand, carbon sequestration service supply, carbon emission demand, food production service supply, and food demand into a comprehensive evaluation model, and determining the ecosystem service early warning results for the set grid units based on the evaluation results and classification mapping relationships of the comprehensive evaluation model, is as follows:

[0057]

[0058] Where Score represents the coordinating score, ranging from [0,1]; γ m This represents the weight of the m-th gradient boosting tree, optimized through cross-validation; GBT m S represents the predicted output of a single gradient boosting tree; wy S represents the supply of water production services. cs S represents the supply of carbon sequestration services. fp D represents the supply of food production services; wy Indicates water resource demand; D cs Indicates carbon emission demand; D fp Indicates food demand; It represents the spatial distribution probability of land use types and is used to calculate the weights of various ecosystem services;

[0059] Based on the coordinated score, the K-means++ clustering algorithm is used to automatically classify the warning level:

[0060]

[0061] Where Cluster(D) represents the early warning result of ecosystem services; μ a Let be the a-th cluster center.

[0062] As one of the optional embodiments, the simulation data of the climate model is set to CMIP6 data.

[0063] This application also provides a water-energy-food related ecosystem service classification and early warning device, including:

[0064] The model adjustment module is used to call the DLS model and adjust the basic construction function of the DLS model in order to adjust the linear relationship of the basic construction function and the land supply and demand balance model at the raster scale.

[0065] The land parameter calculation module is used to acquire simulation data from the set climate model and input the adjusted DLS model to calculate the spatial distribution probability of land use types and land use data.

[0066] The water parameter calculation module is used to calculate the water production service supply and water resource demand of a given grid cell based on the adjusted grid-scale land supply and demand balance model.

[0067] The carbon parameter calculation module is used to calculate the supply of carbon sequestration services and the demand for carbon emissions based on land use data.

[0068] The grain parameter calculation module is used to calculate the supply of grain production services and the demand for grain in a given grid cell based on the adjusted grid-scale land supply and demand balance model.

[0069] The classification and early warning module is used to input the spatial distribution probability of land use type, water production service supply, water resource demand, carbon sequestration service supply, carbon emission demand, food production service supply, and food demand into the comprehensive evaluation model. Based on the evaluation results of the comprehensive evaluation model and the classification mapping relationship, the module determines the ecosystem service early warning results for the set grid units.

[0070] The water, energy, and food-related ecosystem service classification and early warning device of this application embodiment constructs a comprehensive simulation and prediction framework for water production services, carbon sequestration services, food production services, water resource demand, carbon emissions, and food consumption under multiple scenarios based on the spatial distribution probability of land use types and land use data of the adjusted DLS model. According to the evaluation results of the comprehensive evaluation model and the classification mapping relationship, the ecosystem service early warning results of the set grid unit are determined, providing accurate and effective classification information and early warning information for the implementation of water, energy, and food ecosystem services, and facilitating the determination of the level of water, energy, and food ecosystem services.

[0071] At least one embodiment of this application also provides a data control device, including:

[0072] One or more memories that store computer-executable instructions non-transitory;

[0073] One or more processors are configured to run computer-executable instructions, wherein the computer-executable instructions are executed by the one or more processors to implement the water, energy, and food-related ecosystem service classification and early warning method according to any embodiment of this application.

[0074] The aforementioned data control device, based on the spatial distribution probability of land use types and land use data of the adjusted DLS model, constructs a comprehensive simulation and prediction framework for water production services, carbon sequestration services, food production services, as well as water resource demand, carbon emissions, and food consumption under multiple scenarios. According to the evaluation results and classification mapping relationship of the comprehensive evaluation model, it determines the early warning results of ecosystem services for the set grid units, providing accurate and effective classification and early warning information for the implementation of water, energy, and food ecosystem services, and facilitating the determination of the level of water, energy, and food ecosystem services.

[0075] At least one embodiment of this application also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, implement the water, energy, and food-related ecosystem service classification and early warning method according to any embodiment of this application.

[0076] The aforementioned non-transient computer-readable storage medium, based on the spatial distribution probability of land use types and land use data of the adjusted DLS model, constructs a comprehensive simulation and prediction framework for water production services, carbon sequestration services, food production services, as well as water resource demand, carbon emissions, and food consumption under multiple scenarios. Based on the evaluation results and classification mapping relationship of the comprehensive evaluation model, it determines the early warning results of ecosystem services for designated grid units, providing accurate and effective classification and early warning information for the implementation of water, energy, and food ecosystem services, and facilitating the determination of the level of water, energy, and food ecosystem services. Attached Figure Description

[0077] Figure 1 A flowchart of a water-energy-food-related ecosystem service classification and early warning method according to an embodiment of the application;

[0078] Figure 2 A flowchart of a preferred embodiment of a water-energy-food-related ecosystem service classification and early warning method;

[0079] Figure 3 A structural diagram of a water-energy-food-related ecosystem service classification and early warning device module according to an embodiment of the application;

[0080] Figure 4 A schematic block diagram of a data control device provided by the present invention;

[0081] Figure 5 This is a schematic diagram of a non-transitory computer-readable storage medium provided by the present invention. Detailed Implementation

[0082] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the described embodiments of this application without creative effort are within the scope of protection of this application.

[0083] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms "first," "second," and similar terms used in this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0084] To keep the following description of the embodiments of this application clear and concise, detailed descriptions of some known functions and components have been omitted.

[0085] This application provides a method for classifying and early warning of water, energy, and food-related ecosystem services.

[0086] Figure 1 Here is a flowchart of a water, energy, and food-related ecosystem service classification and early warning method according to an embodiment of an application, such as... Figure 1 As shown, a water-energy-food-related ecosystem service classification and early warning method according to an embodiment of the application includes steps S100 to S105:

[0087] S100, call the DLS model and adjust the basic construction function of the DLS model to adjust the linear relationship of the basic construction function and the land supply and demand balance model at the raster scale;

[0088] S101, Obtain simulation data from the set climate model, input the adjusted DLS model to calculate the spatial distribution probability of land use types and land use data;

[0089] S102, based on the adjusted grid-scale land supply and demand balance model, calculates the water production service supply and water resource demand of the set grid cells;

[0090] S103, based on land use data, calculates the supply of carbon sequestration services and the demand for carbon emissions.

[0091] S104, based on the adjusted grid-scale land supply and demand balance model, calculates the supply of food production services and the demand for food in the set grid cells.

[0092] S105 inputs the spatial distribution probability of land use type, water production service supply, water resource demand, carbon sequestration service supply, carbon emission demand, food production service supply, and food demand into the comprehensive evaluation model. Based on the evaluation results and classification mapping relationship of the comprehensive evaluation model, the ecosystem service early warning results of the set grid units are determined.

[0093] In steps S100 to S105 of this embodiment, the data source for each step is the simulation data of the established climate model. The established climate model includes a complete database, and the data required for each calculation process in steps S100 to S105 can be retrieved from the database. The established climate model may be a global climate model (GCM).

[0094] As one of the preferred embodiments, the simulation data of the climate model is set as CMIP6 data, that is, the monthly climate data of the atmospheric circulation model adopts the World Climate Research Program (WCRP) Coupled Model Intercomparison Project Phase 6 (CMIP6) dataset.

[0095] Preferably, in this embodiment of the application, eight high-resolution GCMs model datasets were selected to analyze the future changes in ecosystem services oriented towards water-energy-food linkages from 2040 to 2060 under SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenarios. These eight GCMs model datasets are BCC-CSM2-MR, CESM2, INM-CM5-0, GFDL-ESM4, MR-IESM2, MPI-ESM-HR, EC-Earth3-Veg, and EC-EARTH3. To ensure that all involved models have equal weights in the multi-model statistics, only the dataset "r1i1p1f1" from the first implementation of each model was used, i.e., the first implementation (r1), the first initialization (i1), the first physical process (p1), and the first forcing (f1). The data from these models have good performance in simulating precipitation in East Asia.

[0096] Since climate models are imperfect representations of reality, downscaling and bias correction are necessary before data application. First, statistical downscaling is used to interpolate monthly values ​​of different climate variables on the GCM grid to their corresponding stations. Then, historical observation data from stations between 1980 and 2015 are used to correct biases in the downscaled data, which serves as the simulation data for the climate model established in this application.

[0097] The DLS model, or Dynamic of Land System (DLS) model, is based on the theory of regional land use structure change equilibrium and the theory of grid-scale land use type distribution constraints. It comprehensively considers the driving mechanisms of natural control factors and socio-economic factors on land use change, realizing the dynamic simulation of regional land use structure changes and pattern evolution. DLS fully considers the driving characteristics of regional land use type distribution and land supply and demand balance characteristics, reflecting the systematic, mechanistic, balanced, and convergent nature of the regional land system.

[0098] Clarifying the interaction between land use type distribution and its driving factors is a crucial foundation for dynamic simulation of the land system. To extract the main driving factors influencing regional land use type distribution and to describe the complex nonlinear spatial relationships during land system changes, the DLS model incorporates various discrete factors from the natural environment and socio-economic factors, constructing a nonlinear constraint model of land use type distribution at the grid scale. This model overcomes the limitation imposed by previous studies using linear regression or correlation analysis, which often require the dependent variable to be continuous, and fully reveals the driving mechanism of land use type distribution at the micro-unit level.

[0099] In traditional DLS models, the nonlinear constraint model for land use type distribution at the grid scale is constructed using the Logistic function. This application improves upon the traditional DLS model by employing a Random Forest-Graph Convolutional Network (RF-GCN) hybrid model instead of the Logistic function, thereby overcoming the dependence of the traditional Logistic function on linear relationships.

[0100] The adjusted DLS model consists of the following three modules:

[0101] Scenario Analysis Module: Used to construct demand changes for various land use types in simulated years under future development trends. Based on possible future economic, industrial, policy and technological developments, it designs a series of scenario assumptions for land use structure changes.

[0102] The land use type distribution-driven analysis module is used to measure the spatial relationship between influencing factors such as natural environmental factors, climate factors, location factors, and socio-economic factors and land use type distribution, and to characterize the control and transformation effects of these driving factors on land use.

[0103] Spatial Allocation Module: Based on conditions such as land use conversion probability, land use conversion rules, and land use conversion area restrictions, the module effectively allocates the spatial demand changes of various land use types in simulated years through the grid-scale land supply and demand balance.

[0104] As one optional embodiment, the process of adjusting the linear relationship of the basic construction functions includes the following steps:

[0105] Screening for nonlinear driving factors, using random forest (RF) to quantify the importance of driving factors, and eliminating redundant variables:

[0106]

[0107] Where, N tree This represents the total number of decision trees in the random forest. A larger number of decision trees increases the model's resistance to overfitting, but also increases computational cost. Importance j Score the importance of the j-th driving factor; Let be the out-of-bag error of the tree in lesson t. The value represents the error after randomly permuting the j-th feature; the higher the importance score, the greater the contribution of the driving factor to the land use type distribution.

[0108] A spatial graph convolutional network is used to model and construct a land use spatial map G = (V, E), where nodes V are raster cells, edges E are defined by spatial adjacency relationships, and node features include driving factors such as precipitation, slope, and GDP. The feature propagation formula for the graph convolutional layer is as follows:

[0109]

[0110] in, This represents an adjacency matrix with self-joins; A is the original adjacency matrix; I is the identity matrix; H is the degree matrix, used to prevent gradient explosion or vanishing during feature propagation; (l) W represents the node feature matrix of the l-th layer graph convolution; (l) σ is the trainable weight matrix of the l-th layer; σ is a non-linear activation function used to enhance the non-linear expressive power of the model.

[0111] As one optional embodiment, the spatial distribution probability of land use types is calculated as follows:

[0112]

[0113] in, Let i be the probability of the k-th land use type appearing in grid cell i, i.e., the spatial distribution probability of land use type. The probability of the k-th land use type predicted by the random forest; α represents the probability of the k-th land use type output by the spatial graph convolutional network; α is a dynamic weighting coefficient used to balance the output contributions of the random forest and the spatial graph convolutional network.

[0114] In the adjusted DLS model, changes in land supply and demand are a dynamic convergence process that maintains overall balance. The regional-scale land supply and demand balance is obtained from the annual land use demand predicted by the scenario analysis module, while the raster-scale land supply and demand balance requires a high-precision raster-scale land supply and demand balance model.

[0115] The grid-scale land supply and demand balance model is an empirical statistical model with typical spatial explicitness. By considering the effects of local driving factors within each grid and the neighborhood effects of driving factors from surrounding grids on the central grid, it meticulously depicts the competition process among land use types. Its core is the spatial allocation process of land demand at the microscale.

[0116] Therefore, let the probability of the k-th land use type appearing in grid cell i based on the local action factor be... The probability of raster cell i exhibiting the kth land use type based on the neighborhood interaction factor is: Then the probability of land use combinations and The following results were obtained from the nonlinear constraint model of land use type distribution at the grid scale:

[0117]

[0118] Among them, X il This represents the l-th driving factor on grid cell i, which can be a local driving factor or a neighborhood driving factor. Define the spatial autocorrelation factor as the regression coefficient. r is the influence coefficient of the spatial autocorrelation factor.

[0119] The changes in the land system evolution pattern depend not only on the probability of land use combinations, but also on factors such as land use stability parameters and competitiveness indicators of different land use types, which constitute the total land use probability L in spatial pattern simulation. i,k .

[0120] Through multiple iterations, the spatial allocation of land area requirements is achieved, as shown in the following formula:

[0121]

[0122] Among them, L i,k S represents the total probability of land use during the spatial allocation of land supply and demand at the raster scale. k C represents the stability parameter for the k-th land use type. k The competitiveness index for different land use types depends on the difference between the demand area and the current actual area for each land use type.

[0123] Among them, the water supply is estimated based on the principle of water balance, that is, the water supply is the amount of water after the actual evapotranspiration is the amount of rainfall in a certain grid cell.

[0124] As one optional embodiment, the process of calculating the water production service supply and water resource demand of a given grid cell based on the adjusted grid-scale land supply and demand balance model is as follows:

[0125]

[0126] PET (i) =K C(i) ×ET 0(i) ;

[0127]

[0128] AWC i =Min(Rest_layer_depth) (i) ,Root_depth (i) )×PAWC (i) ;

[0129] AET (i) =Min(K) C(i) ×ET 0(i) ,P (i) );

[0130] Where i represents the grid cell of the region, and can represent the set grid cell; S wy P represents the average annual water production (mm) of grid cell i, i.e., the water production service supply. (i) The annual precipitation (mm) for grid cell i; AET (i) and PET (i) The following are the annual actual evapotranspiration (mm) and potential evapotranspiration (mm) for grid cell i, respectively; KC (i) ET is the crop evapotranspiration coefficient. 0(i) The reference crop evapotranspiration (mm) for grid cell i; AWC (i) Plant-available water content (mm); W (i) Non-physical parameters characterizing natural climate-soil properties; Z is an empirical coefficient reflecting local precipitation patterns and hydrogeological characteristics; Rest_layer_depth (i) Maximum soil depth (mm); Root_depth (i) Indicates root depth (mm); PAWC (i) Let be the vegetation available water content of grid cell i.

[0131] D wy =D i,agr +Di,ind +D i,dom +D i,eco ;

[0132] D i,agr =Agr×A i,aps ;

[0133] D i,ind =Ind×G i ;

[0134] D i,dom =Dom×P i ;

[0135] D i,eco =Aec×A i,es ;

[0136] Among them, D wy It is the total water demand (m³) 3 / hm 2 ( ), namely, water resource demand; D i,agr D i,ind D i,dom and D i,eco The total water consumption (m³) for agriculture, industry, domestic use, and ecological use in grid cell i are respectively. 3 Agr represents the amount of water used for agricultural production per unit area (m²). 3 / hm 2 A i,aps The agricultural production space area (hm²) of grid cell i 2 Ind represents industrial water consumption per unit of GDP (m³). 3 / 10 4 Yuan); G i Represents the regional GDP (10) of grid cell i 4 Yuan / km 2 Dom represents the average domestic water consumption per capita (m³). 3 / person); P i Represents the population (people / km) of raster cell i. 2 Aec represents the water consumption per unit area of ​​ecological space (m³). 3 / hm 2 A i,es Represents the ecological space area (hm) of grid cell i 2 ).

[0137] Since energy use is largely concentrated in carbon emissions, carbon sequestration services are selected as the supply side to represent the water-energy-food relationship. On the supply side, based on land use data, the total regional carbon stock is estimated using four carbon pools: aboveground biomass, belowground biomass, soil carbon, and dead organic carbon. The carbon density data is revised based on average annual precipitation and average annual temperature.

[0138] As one optional embodiment, the process of calculating the supply of carbon sequestration services and the demand for carbon emissions based on land use data is as follows:

[0139] S cs =C above +C below +C soil +C dead ;

[0140] Among them, S cs Total carbon storage (t / hm) 2 C, i.e., the supply of carbon sequestration services; above Aboveground biomass carbon (t / hm) 2 ), C below underground biomass carbon (t / hm) 2 ), C soil Soil organic carbon (t / hm) 2 ), C dead Dead organic matter (t / hm) 2 ).

[0141] On the demand side, the demand for carbon sequestration services is estimated by multiplying per capita carbon emissions by population density. Population density is calculated using 1km×1km grid data; per capita carbon emissions are obtained from China's carbon emission inventory, covering various human activities such as agriculture, industry, heating, and transportation.

[0142] D cs =D percarbon ×ρ pop ;

[0143] Among them, D cs Carbon sequestration demand (t / hm) 2 ), namely carbon emission demand; D percarbon Carbon emissions per capita (t / person); ρ pop To set the population density (people / km) of the grid cells 2 ).

[0144] Food supply services are not only the material basis for human survival but also a crucial ecosystem service provision based on arable land. Since there is a significant linear relationship between crop yield and NDVI (Non-Distant Land Value Interchange), NDVI effectively reflects crop productivity. Total food production is allocated according to the ratio of grid NDVI to the total NDVI of arable land to determine the food supply capacity of each grid and thus estimate the supply in the study area. Food demand is estimated by multiplying per capita food demand by population density.

[0145] As one optional embodiment, the process of calculating the supply and demand of food production services for a given grid cell, based on the adjusted grid-scale land supply and demand balance model, is as follows:

[0146] S fp =(NDVI) x / NDVI sum )×FP sum

[0147] D fp =D perfood ×ρ pop ;

[0148] Where i represents the grid cell of the region, and can be used to define the grid cell; S fp The grain yield (t / hm) allocated to grid cell i 2 ), namely, the supply of food production services; FP sum Total grain output (t); NDVI x NDVI is the NDVI value of grid cell i; sum D is the sum of the NDVI values ​​of cultivated land in the study area; fp The food demand (t / hm) for grid cell i 2 ), that is, food demand; D perfood This represents the per capita food demand (person / kg).

[0149] In this embodiment, a machine learning algorithm is introduced to construct a dynamic comprehensive evaluation system for water-energy-food interconnected ecosystem services. Machine learning algorithms can overcome the limitations of traditional models that rely on linear weighting assumptions and struggle to capture complex nonlinear relationships.

[0150] First, regarding the supply of water production services (S) wy Carbon sequestration service supply (S) cs ), Supply of grain production services (S) fp ), water resource demand (D wy ), carbon emission demand (D cs Food demand (D) fp ) and spatial distribution probability of land use types The data is standardized to eliminate dimensional differences.

[0151]

[0152] Among them, X norm σ is the normalized value; μ is the mean, and σ is the standard deviation.

[0153] Key driving factors are selected using Random Forest (RF), and a comprehensive evaluation function is constructed by combining it with Gradient Boosting Tree (GBT).

[0154]

[0155] Here, "Score" represents the coordination score, ranging from [0,1], with a larger value indicating a more coordinated system. m This represents the weight of the m-th gradient boosting tree, optimized through cross-validation; GBT m This represents the predicted output of a single gradient boosting tree.

[0156] Based on the coordinated score output by the model, the K-means++ clustering algorithm is used to automatically classify the warning level:

[0157]

[0158] Where Cluster(D) represents the early warning result of ecosystem services; μ a Let be the a-th cluster center.

[0159] The classification criteria for warning level D are shown in Table 1:

[0160] Table 1. Classification of Coupling Coordination Degree Levels

[0161]

[0162] For example, setting a 20% increase in the probability of conversion from cultivated land, forestry, grassland, and water resources to construction land, and a 30% decrease in the probability of conversion from construction land back to forestry and grassland as early warning thresholds, an adjusted DLS model is used to simulate land use change. CMIP6 data and land use prediction data based on the improved DLS are used to drive the model to predict future trends in water production services, carbon sequestration services, and food production services, and to calculate the coupling coordination degree. Simultaneously, tiered early warning thresholds are set based on the coupling coordination degree level, triggering corresponding early warning levels according to the prediction results. This provides accurate and effective classification and early warning information for the implementation of water, energy, and food ecosystem services, facilitating the determination of the level of these ecosystem services.

[0163] As one of the preferred embodiments, Figure 2 A flowchart of a preferred embodiment of a water-energy-food-related ecosystem service classification and early warning method is shown below. Figure 2 As shown, a preferred embodiment of the water-energy-food-related ecosystem service classification and early warning method further includes the following steps:

[0164] S200, based on the adjusted DLS model, determines the elevation data of the set grid cells;

[0165] S201, Based on the altitude data, perform layering processing on the set grid cells to determine the layered data;

[0166] S202, Construct a random forest regression model, using grid cells as partitioned data, to partition and stratify the service demand prediction of the evaluation results, and obtain the demand prediction results;

[0167] S203 imports the supply of water production services, water resource demand, carbon sequestration service supply, carbon emission demand, food production service supply, food demand, and demand forecast results into a large language model. Based on the derivation results of the demand forecast results from the large language model, the actual demand for ecosystem service classification and early warning is used.

[0168] Based on the partitioning of the set grid cells, the altitude of each set grid cell is determined based on its altitude data, and then layered processing is performed according to the altitude.

[0169] As one optional embodiment, the elevation range of all set grid cells is divided into a set number of elevation ranges, the elevation ranges are standardized, and the standardized data is used as hierarchical data.

[0170] Therefore, altitude has a profound impact on demand forecasting within the ecosystem service system. Researching altitude and stratifying it can provide a scientific basis for the optimal allocation and scheduling management of ecosystem service demand forecasting. Service demand forecasting based on zoning and stratification of evaluation results includes the following characteristics:

[0171] (1) Altitude characteristics.

[0172] High altitudes: Water, energy, and food activities slow down, and the demand for ecosystem services decreases significantly.

[0173] At low altitudes, water, energy, and food activities are more active, and the demand for ecosystem services increases significantly.

[0174] (2) Partition characteristics.

[0175] The demand for ecosystem services varies regionally depending on the location of the grid cells.

[0176] Based on this, the training dataset is constructed as follows:

[0177] H={(x (1) ,y 1 ),(x (2) ,y (2) ),...,(x (N) ,y (N) )};

[0178] Where x (i) It sets the feature vector of i samples in the raster unit, y (i) This is the corresponding hierarchical data.

[0179] Based on the above training dataset, random forests and decision trees are constructed for random forest prediction. The prediction process of the demand prediction results is consistent with the random forest application in steps S100 to S105, ensuring the accuracy of subsequent large language model derivation.

[0180] Among them, the large language model can adopt a distillation model such as the Deepseek large language model, which can be used to localize and train historical water production service supply, water resource demand, carbon sequestration service supply, carbon emission demand, food production service supply, food demand and demand forecast results, and perform secondary derivation of demand forecast results as actual demand for ecosystem service classification and early warning.

[0181] Preferably, during the training process of the large language model, historical water production service supply, water resource demand, carbon sequestration service supply, carbon emission demand, food production service supply, food demand, demand forecast results and actual demand are used for training and deployment to constrain the derivation logic of the large language model.

[0182] The water, energy, and food-related ecosystem service classification and early warning method of this application embodiment is based on the spatial distribution probability of land use type and land use data of the adjusted DLS model. It constructs a comprehensive simulation and prediction framework for water production services, carbon sequestration services, food production services, as well as water resource demand, carbon emissions, and food consumption under multiple scenarios. According to the evaluation results of the comprehensive evaluation model and the classification mapping relationship, the early warning results of ecosystem services for the set grid units are determined, providing accurate and effective classification and early warning information for the implementation of water, energy, and food ecosystem services, and facilitating the determination of the level of water, energy, and food ecosystem services.

[0183] This application also provides a water, energy, and food-related ecosystem service classification and early warning device.

[0184] Figure 3 This is a structural diagram of a water-energy-food-related ecosystem service classification and early warning device module according to an embodiment of an application, such as... Figure 3 As shown, a water-energy-food related ecosystem service classification and early warning device according to one embodiment of the application includes:

[0185] Model adjustment module 100 is used to call the DLS model and adjust the basic construction function of the DLS model in order to adjust the linear relationship of the basic construction function and the land supply and demand balance model at the raster scale.

[0186] The land parameter calculation module 101 is used to acquire simulation data of the set climate model and input the adjusted DLS model to calculate the spatial distribution probability of land use type and land use data.

[0187] The water parameter calculation module 102 is used to calculate the water production service supply and water resource demand of a set grid cell based on the adjusted grid-scale land supply and demand balance model.

[0188] Carbon parameter calculation module 103 is used to calculate the supply of carbon sequestration services and carbon emission demand based on land use data;

[0189] The grain parameter calculation module 104 is used to calculate the supply of grain production services and the demand of grain for a set grid cell based on the adjusted grid-scale land supply and demand balance model.

[0190] The classification and early warning module 105 is used to input the spatial distribution probability of land use type, water production service supply, water resource demand, carbon sequestration service supply, carbon emission demand, food production service supply and food demand into the comprehensive evaluation model. Based on the evaluation results of the comprehensive evaluation model and the classification mapping relationship, the system determines the ecosystem service early warning results for the set grid units.

[0191] The water, energy, and food-related ecosystem service classification and early warning device of this application embodiment constructs a comprehensive simulation and prediction framework for water production services, carbon sequestration services, food production services, water resource demand, carbon emissions, and food consumption under multiple scenarios based on the spatial distribution probability of land use types and land use data of the adjusted DLS model. According to the evaluation results of the comprehensive evaluation model and the classification mapping relationship, the ecosystem service early warning results of the set grid unit are determined, providing accurate and effective classification information and early warning information for the implementation of water, energy, and food ecosystem services, and facilitating the determination of the level of water, energy, and food ecosystem services.

[0192] At least one embodiment of this application also provides a data control device. Figure 4 This is a schematic block diagram of a data control device provided for at least one embodiment of this application. For example, such as... Figure 4 As shown, the data control device 20 may include one or more memories 200 and one or more processors 201. The memories 200 are used to store computer-executable instructions non-transitory; the processors 201 are used to run the computer-executable instructions, which, when run by the processors 201, can cause the processors 201 to perform one or more steps in the water, energy, and food-related ecosystem service classification and early warning method according to any embodiment of this application.

[0193] For the specific implementation details and explanations of each step in the water-energy-food-related ecosystem service classification and early warning method, please refer to the relevant content in the above-mentioned embodiments of the water-energy-food-related ecosystem service classification and early warning method, which will not be repeated here. It should be noted that Figure 4The components of the data control device 20 shown are merely exemplary and not limiting. The data control device 20 may have other components depending on the actual application requirements.

[0194] In one embodiment, the processor 201 and the memory 200 can communicate directly or indirectly with each other. For example, the processor 201 and the memory 200 can communicate via a network connection. The network can include a wireless network, a wired network, and / or any combination of wireless and wired networks; this application does not limit the type and function of the network. Alternatively, the processor 201 and the memory 200 can also communicate via a bus connection. The bus can be a Peripheral Component Interconnect Standard (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. For example, the processor 201 and the memory 200 can be located at a remote data server (cloud) or a distributed energy system (local), or at a client (e.g., a mobile device such as a mobile phone). For example, the processor 201 can be a central processing unit (CPU), a tensor processor (TPU), or a graphics processing unit (GPU), etc., with data processing and / or instruction execution capabilities, and can control other components in the data control device 20 to perform desired functions. The central processing unit (CPU) can be an x86 or ARM architecture, etc.

[0195] In one embodiment, memory 200 may include any combination of one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, erasable programmable read-only memory (EPROM), portable compact disc read-only memory (CD-ROM), USB memory, flash memory, etc. One or more computer-executable instructions may be stored on the computer-readable storage medium, and processor 201 may execute these computer-executable instructions to implement various functions of data control device 20. Various application programs and various data, as well as various data used and / or generated by the application programs, may also be stored in memory 200.

[0196] It should be noted that the data control device 20 can achieve similar technical effects to the aforementioned water, energy and food-related ecosystem service classification and early warning method, and the repetition will not be repeated.

[0197] At least one embodiment of this application also provides a non-transitory computer-readable storage medium. Figure 5 This is a schematic diagram of a non-transitory computer-readable storage medium provided for at least one embodiment of this application. For example, such as... Figure 5As shown, one or more computer-executable instructions 301 may be stored non-transitory on the non-transitory computer-readable storage medium 30. For example, when the computer-executable instructions 301 are executed by a computer, they may cause the computer to perform one or more steps in the water, energy, and food-related ecosystem service classification and early warning method according to any embodiment of this application.

[0198] In one embodiment, the non-transitory computer-readable storage medium 30 can be applied to the data control device 20 described above, for example, it can be the memory 200 in the data control device 20.

[0199] In one embodiment, the description of the non-transitory computer-readable storage medium 30 can be found in the description of the memory 200 in the embodiment of the data control device 20, and will not be repeated hereafter.

[0200] It should be noted that the memory 200 stores different non-transient computer-executable instructions, and the data control device 20 corresponds to the firmware upgrade device. When the computer-executable instructions are run by the processor 201, the processor 201 can perform one or more steps in the water, energy and food-related ecosystem service classification and early warning method according to any embodiment of this application.

[0201] The following points should be noted regarding this application:

[0202] (1) The accompanying drawings of the embodiments of this application only involve the structures involved in the embodiments of this application. Other structures can be referred to the general design.

[0203] (2) For clarity, the thickness and dimensions of layers or structures are enlarged in the accompanying drawings used to describe embodiments of the invention. It will be understood that when an element such as a layer, film, region, or substrate is referred to as being “above” or “below” another element, the element may be “directly” located “above” or “below” the other element, or there may be intermediate elements present.

[0204] (3) Where there is no conflict, the embodiments and features in the embodiments of this application can be combined with each other to obtain new embodiments. The above are only specific implementations of this application, but the protection scope of this application is not limited thereto, and the protection scope of this application shall be determined by the protection scope of the claims.

[0205] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0206] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for classifying and providing early warning of water, energy, and food-related ecosystem services, characterized in that, Including the following steps: The DLS model is invoked and its basic construction function is adjusted to adjust the linear relationship of the basic construction function with the raster-scale land supply and demand balance model. Obtain simulation data from the established climate model and input it into the adjusted DLS model to calculate the spatial distribution probability of land use types and land use data; Based on the adjusted grid-scale land supply and demand balance model, the water production service supply and water resource demand of the set grid cells are calculated. Based on the aforementioned land use data, calculate the supply of carbon sequestration services and the demand for carbon emissions. Based on the adjusted grid-scale land supply and demand balance model, the supply of food production services and the demand for food are calculated for each grid cell. The spatial distribution probability of the land use type, the supply of water production services, the demand for water resources, the supply of carbon sequestration services, the demand for carbon emissions, the supply of food production services, and the demand for food are input into a comprehensive evaluation model. Based on the evaluation results and classification mapping relationship of the comprehensive evaluation model, the ecosystem service early warning results of the set grid units are determined.

2. The method for classifying and early warning of water, energy, and food-related ecosystem services according to claim 1, characterized in that, It also includes the following steps: The elevation data of the defined raster cell is determined based on the adjusted DLS model; Based on the altitude data, the set grid unit is layered to determine the layered data; A random forest regression model is constructed, and the service demand is predicted by partitioning and stratifying the evaluation results using the set grid cells as partitioned data, so as to obtain the demand prediction results. The supply of water production services, the demand for water resources, the supply of carbon sequestration services, the demand for carbon emissions, the supply of food production services, the demand for food, and the demand forecast results are imported into a large language model. The derivation results of the demand forecast results from the large language model are used as the actual demand for ecosystem service classification and early warning.

3. The method for classifying and early warning of water, energy, and food-related ecosystem services according to claim 1, characterized in that, The process of adjusting the linear relationship of the basic building functions includes the following steps: We screened nonlinear driving factors, quantified their importance using random forest, and eliminated redundant variables. Where, N tree The total number of decision trees in the random forest; Importance j Score the importance of the j-th driving factor; Let be the out-of-bag error of the tree in lesson t. The error is the result of randomly permuting the j-th feature. A spatial graph convolutional network is used to model and construct a land use spatial map G = (V, E), where nodes V are raster cells, edges E are defined by spatial adjacency relationships, and node features include driving factors. The feature propagation formula for the graph convolutional layer is as follows: in, This represents an adjacency matrix with self-joins; A is the original adjacency matrix; I is the identity matrix; H is the degree matrix; (l) W represents the node feature matrix of the l-th layer graph convolution; (l) Let be the trainable weight matrix of the l-th layer; σ is the non-linear activation function.

4. The method for classifying and early warning of water, energy, and food-related ecosystem services according to claim 3, characterized in that, The spatial distribution probability of the land use type is calculated as follows: in, The probability of the k-th land use type appearing in grid cell i is the spatial distribution probability of the land use type. The probability of the k-th land use type predicted by the random forest; α represents the probability of the k-th land use type output by the spatial graph convolutional network; α is a dynamic weighting coefficient used to balance the output contributions of the random forest and the spatial graph convolutional network. Let the probability of the k-th land use type appearing in raster cell i based on the local action factor be . The probability of raster cell i exhibiting the kth land use type based on the neighborhood interaction factor is: Then the probability of land use combinations and The following results were obtained from the nonlinear constraint model of land use type distribution at the grid scale: Among them, X il This represents the l-th driving factor on grid cell i. Define the spatial autocorrelation factor as the regression coefficient. r is the influence coefficient of the spatial autocorrelation factor; Through multiple iterations, the spatial allocation of land area requirements is achieved, as shown in the following formula: Among them, L i,k S represents the total probability of land use during the spatial allocation of land supply and demand at the raster scale. k C represents the stability parameter for the k-th land use type. k Indicators representing the competitiveness of different land use types.

5. The method for classifying and early warning of water, energy, and food-related ecosystem services according to claim 1, characterized in that, Based on the adjusted grid-scale land supply and demand balance model, the process of calculating the water production service supply and water resource demand for a given grid cell is as follows: PET (i) =K C(i) ×ET 0(i) ; AWC i =Min(Rest_layer_depth (i) ,Root_depth (i) )×PAWC (i) ; Where i represents the grid cell of the region, and can represent the set grid cell; S wy P represents the average annual water production of grid cell i, i.e., the water production service supply. (i) The annual precipitation for grid cell i; AET (i) and PET (i) The following are the annual actual evapotranspiration and potential evapotranspiration of grid cell i, respectively; KC (i) ET is the crop evapotranspiration coefficient. 0(i) AWC is the reference crop evapotranspiration for grid cell i. (i) Water content available to plants; W (i) Non-physical parameters characterizing natural climate-soil properties; Z is an empirical coefficient reflecting local precipitation patterns and hydrogeological characteristics; Rest_layer_depth (i) Root_depth represents the maximum soil depth. (i) Indicates root depth; PAWC (i) The vegetation available water content of grid cell i; D wy =D i,agr +D i,ind +D i,dom +D i,eco ; D i,agr =Agr×A i,aps ; D i,ind =Ind×G i ; D i,dom =Dom×P i ; D i,eco =Aec×A i,es ; Among them, D wy It is the total water demand, i.e., the water resource demand; D i,agr D i,ind D i,dom and D i,eco These represent the total agricultural, industrial, domestic, and ecological water consumption of grid cell i, respectively; Agr represents the agricultural water consumption per unit area; A i,aps The agricultural production space area of ​​grid cell i; Ind represents the industrial water consumption per unit of GDP; G i Represents the regional GDP of grid cell i; Dom represents per capita domestic water consumption; P i Represents the population of grid cell i; Aec represents the water consumption per unit area of ​​ecological space; A i,es This represents the ecological space area of ​​grid cell i.

6. The method for classifying and early warning of water, energy, and food-related ecosystem services according to claim 1, characterized in that, Based on the aforementioned land use data, the process of calculating the supply of carbon sequestration services and the demand for carbon emissions is as follows: S cs =C above +C below +C soil +C dead ; Among them, S cs Total carbon storage, i.e., the supply of carbon sequestration services; C above For aboveground biochar, C below underground biochar, C soil Soil organic carbon, C dead Dead organic matter; D cs =D percarbon ×ρ pop ; Among them, D cs This refers to the demand for carbon sequestration, i.e., the carbon emission demand; D percarbon Per capita carbon emissions; ρ pop To set the population density of the grid cells.

7. The method for classifying and early warning of water, energy, and food-related ecosystem services according to claim 1, characterized in that, Based on the adjusted grid-scale land supply and demand balance model, the process of calculating the supply of food production services and the demand for food in a given grid cell is as follows: S fp =(NDVI x / NDVI sum )×FP sum D fp =D perfood ×ρ pop ; Where i represents the grid cell of the region, and can be used to define the grid cell; S fp The grain output allocated to grid cell i, i.e., the grain production service supply; FP sum Total grain output; NDVI x NDVI is the NDVI value of grid cell i; sum D is the sum of the NDVI values ​​of cultivated land in the study area; fp The food demand is the amount of food required for grid cell i, i.e., the food demand; D perfood This refers to the per capita food demand.

8. The method for classifying and early warning of water, energy, and food-related ecosystem services according to any one of claims 1 to 7, characterized in that, The process of inputting the spatial distribution probability of the land use type, the supply of water production services, the demand for water resources, the supply of carbon sequestration services, the demand for carbon emissions, the supply of food production services, and the demand for food into a comprehensive evaluation model, and determining the ecosystem service early warning results for the set grid units based on the evaluation results and classification mapping relationships of the comprehensive evaluation model, is as follows: Where Score represents the coordinating score, ranging from [0,1]; γ m This represents the weight of the m-th gradient boosting tree, optimized through cross-validation; GBT m S represents the predicted output of a single gradient boosting tree; wy S represents the supply of water production services. cs S represents the supply of carbon sequestration services. fp Indicates the supply of food production services; D wy Indicates water resource demand; D cs Indicates carbon emission demand; D fp Indicates food demand; It represents the spatial distribution probability of land use types, used to calculate the weights of various ecosystem services; Based on the aforementioned coordination score, the K-means++ clustering algorithm is used to automatically classify the early warning level: Where Cluster(D) represents the early warning result of ecosystem services; μ a Let be the a-th cluster center.

9. The method for classifying and early warning of water, energy, and food-related ecosystem services according to any one of claims 1 to 7, characterized in that, The simulation data for the climate model is CMIP6 data.

10. A classification and early warning device for water, energy, and food-related ecosystem services, characterized in that, include: The model adjustment module is used to call the DLS model and adjust the basic construction function of the DLS model in order to adjust the linear relationship of the basic construction function and the raster-scale land supply and demand balance model. The land parameter calculation module is used to acquire simulation data from the set climate model and input the adjusted DLS model to calculate the spatial distribution probability of land use types and land use data. The water parameter calculation module is used to calculate the water production service supply and water resource demand of a given grid cell based on the adjusted grid-scale land supply and demand balance model. The carbon parameter calculation module is used to calculate the supply of carbon sequestration services and the demand for carbon emissions based on the land use data. The grain parameter calculation module is used to calculate the supply of grain production services and the demand for grain in a given grid cell based on the adjusted grid-scale land supply and demand balance model. The classification and early warning module is used to input the spatial distribution probability of the land use type, the supply of water production services, the demand for water resources, the supply of carbon sequestration services, the demand for carbon emissions, the supply of food production services, and the demand for food into a comprehensive evaluation model. Based on the evaluation results of the comprehensive evaluation model and the classification mapping relationship, the module determines the early warning results of ecosystem services for the set grid units.

Images