Land use layout prediction method and device and electronic equipment

Abstract

The application discloses a land use layout prediction method and device and electronic equipment. The method comprises the following steps: acquiring rasterized basic data, land use space constraint data, scenario driving data and historical land use data of a space region to be predicted; determining a historical land use type conversion relationship and a historical conversion frequency according to the historical land use data; for each space constraint item, determining a first constraint strength in combination with a space action range, a constraint attribute and a spatial relationship between the constraint attribute and each grid cell, and determining a dispersion weight according to a numerical distribution of the first constraint strength; determining an initial weight based on the historical conversion frequency, fusing the initial weight and the dispersion weight to obtain a fused weight, and calculating a weighted constraint strength of each grid cell; under different prediction scenarios, participating in development suitability probability, neighborhood influence parameter and conversion probability correction according to the weighted constraint strength, and generating a predicted land use layout. The application can convert overall policy constraints into dynamic constraint influences for different grid cells, and improve the accuracy of land use layout prediction under policy constraint conditions.

Description

Technical Field

[0001] This application relates to the field of land use technology, and in particular to a land use layout prediction method, apparatus, and electronic equipment. Background Technology

[0002] Land use layout prediction is an important technical means in national spatial planning, ecological environmental protection, urban expansion control, and regional resource allocation. With the development of remote sensing imagery, geographic information systems, and spatial data processing technologies, raster-based land use simulation methods are widely used to analyze the spatial change trends of land use types such as cultivated land, forest land, grassland, water areas, construction land, and unused land under different development scenarios. Existing land use prediction methods typically utilize historical land use data, natural environmental data, and socio-economic data, combined with land use type conversion probabilities, to simulate the spatial distribution of land use in future periods.

[0003] In existing technologies, land use simulation methods such as the PLUS model can predict the transformation relationships between different land use types through land expansion analysis and cellular automata processes. These methods typically extract driving factors from historical land use change samples, calculate the suitability of each grid cell for transitioning to different land use types, and combine neighborhood effects, random disturbances, and land use transition matrices to generate the land use layout for the prediction period. These methods have a certain application foundation in urban expansion simulation, ecological protection scenario analysis, and land use change prediction.

[0004] However, in actual land use planning scenarios, land use changes are influenced not only by natural environmental and socio-economic conditions, but also by spatial control requirements such as ecological protection red lines, permanent basic farmland, urban development boundaries, and functional zones. Existing land use forecasting methods, when incorporating such policy or planning constraints, typically treat them as overall constraint variables, unified transfer rules, or fixed limitations in simulations. This makes it difficult to reflect the differences in the impact of different constraints on different spatial locations, and also to demonstrate the varying degrees of influence of the same constraint on different grid cells.

[0005] Because policy constraints often act on the prediction area in a holistic, static, or constant manner during existing simulations, it is difficult for models to dynamically adjust the land use type conversion probability based on the spatial location, constraint attributes, historical conversion characteristics, and scenario conditions of different grid cells. For example, some grid cells within the same area may be strongly constrained by ecological protection red lines, while others may be located near urban development boundaries and have a high tendency for construction land expansion. If uniform constraint parameters are used, it is easy to cause a mismatch between the prediction results and the actual spatial control differences, thereby affecting the refinement and accuracy of land use layout prediction.

[0006] Therefore, it is necessary to provide a land use layout prediction method that can transform the spatial constraint data of land use that originally applied to the whole area into differentiated constraint parameters corresponding to each grid cell, and use these differentiated constraint parameters to correct the development suitability probability, neighborhood influence parameters, and conversion probability, so as to improve the spatial adaptability and accuracy of land use layout prediction under policy constraints. Summary of the Invention

[0007] This application provides a land use layout prediction method, apparatus, and electronic device to address the problem that the accuracy of land use layout prediction is low in the prior art because policy constraints that can only use constant influences on the whole area are applied.

[0008] According to embodiments of this application, this application provides a land use layout prediction method, comprising: acquiring rasterized basic data, land use spatial constraint data, scenario-driven data, and historical land use data corresponding to a spatial area to be predicted. The rasterized basic data includes location data, natural environment data, and socio-economic data of each raster unit within the spatial area to be predicted. The land use spatial constraint data characterizes constraint information on land use type conversion within the spatial area to be predicted and includes at least one spatial constraint term. The spatial constraint term is a spatial data item in the land use spatial constraint data used to characterize a restriction on land use type conversion. The scenario-driven data is used to characterize the socio-economic development affecting land use type conversion under different prediction scenarios. Based on development conditions and climate change conditions, historical land use data includes land use types corresponding to each grid cell at multiple historical moments. Based on the historical land use data, the historical land use type transformation relationships between adjacent historical moments for each grid cell are determined, and the historical transformation frequency between each land use type is determined based on these relationships. For each spatial constraint item in the land use spatial constraint data, based on the spatial scope, constraint attribute, and spatial relationship between the spatial constraint item and each grid cell, the first constraint strength of the spatial constraint item on each grid cell is determined. The first constraint strength characterizes the effect of the spatial constraint item on the land use type transformation of the corresponding grid cell. The constraint strength of the change is determined as follows: For each spatial constraint item, the dispersion statistic corresponding to the spatial constraint item is calculated based on the numerical distribution of the first constraint strength of the spatial constraint item across all grid cells, and the dispersion weight corresponding to each spatial constraint item is determined based on the dispersion statistic corresponding to each spatial constraint item. The dispersion weight is used to characterize the difference in the first constraint strength of the corresponding spatial constraint item across different grid cells. The initial weight corresponding to each spatial constraint item is determined based on the historical conversion frequency between different land use types. The initial weight is used to characterize the preset configuration proportion of the spatial constraint item in land use type conversion constraints. For each spatial constraint item, the initial weight and dispersion weight corresponding to the spatial constraint item are fused according to a preset ratio. For example, a weighted sum is performed to obtain the fusion weight corresponding to the spatial constraint term; for each grid cell, the first constraint strength of each spatial constraint term on that grid cell is multiplied by the corresponding fusion weight and summed to obtain the weighted constraint strength corresponding to that grid cell; for each prediction scenario, based on the scenario-driven data corresponding to the prediction scenario and the historical conversion frequency between each land use type, the scenario transfer rules between different land use types are determined; for each prediction scenario, based on the scenario-driven data corresponding to the prediction scenario, historical land use data, rasterized basic data, and the weighted constraint strength corresponding to each grid cell, the development suitability probability of each grid cell being converted from the current land use type to each candidate land use type is determined;For each prediction scenario, based on the current land use type of each grid cell, each candidate land use type, the distribution of land use types within the neighborhood of each grid cell, and the weighted constraint strength corresponding to each grid cell, the neighborhood influence parameter for the transformation of each grid cell from its current land use type to each candidate land use type is determined. For each prediction scenario, based on the development suitability probability, scenario transition rules, and neighborhood influence parameter corresponding to each grid cell, the initial conversion probability for the transformation of each grid cell from its current land use type to each candidate land use type is calculated. For each prediction scenario, based on the weighted constraint strength and... The scenario transition rule determines the probability correction coefficient for each grid cell corresponding to each candidate land use type, and uses this probability correction coefficient to correct the corresponding initial conversion probability, obtaining the constrained correction conversion probability for each grid cell to convert from the current land use type to each candidate land use type. For each prediction scenario, based on the constrained correction conversion probability corresponding to each grid cell, the predicted land use type corresponding to each grid cell in the target prediction period is determined from each candidate land use type. Based on the location data of each grid cell and the corresponding predicted land use type, the predicted land use layout of the spatial area to be predicted under that prediction scenario is generated.

[0009] According to embodiments of this application, location data includes at least one of grid number, grid center point coordinates, and grid boundary coordinates; natural environment data includes at least one of elevation, slope, aspect, average annual temperature, annual precipitation, soil type, distance from water system, and distance from road; socioeconomic data includes at least one of population density, regional GDP, nighttime light intensity, distance from settlement, distance from town center, and traffic accessibility data.

[0010] According to the embodiments of this application, land use types include at least two of cultivated land, forest land, grassland, water area, construction land, and unused land; the current land use type is the land use type of the corresponding grid cell at the start of the prediction, and the candidate land use type is the land use type that the corresponding grid cell is allowed to be determined in the target prediction period.

[0011] According to the embodiments of this application, the spatial constraints include at least one of the following: functional zone constraints, ecological protection red line constraints, urban development boundary constraints, and permanent basic farmland constraints.

[0012] According to the embodiments of this application, for each spatial constraint item in the land use spatial constraint data, the first constraint intensity of the spatial constraint item on each grid cell is determined based on the spatial scope of the spatial constraint item, the constraint attribute, and the spatial relationship between the spatial constraint item and each grid cell. This includes: when the spatial constraint item is a main functional area constraint item, the first constraint intensity corresponding to the grid cell is determined based on the main functional area type to which the grid cell belongs; when the spatial constraint item is an ecological protection red line constraint item, the first constraint intensity corresponding to the grid cell is determined based on whether the grid cell is located within the ecological protection red line; when the spatial constraint item is an urban development boundary constraint item, the first constraint intensity corresponding to the grid cell is determined based on the spatial distance between the grid cell and the urban development boundary; when the spatial constraint item is a permanent basic farmland constraint item, the first constraint intensity corresponding to the grid cell is determined based on whether the grid cell is located within the permanent basic farmland or based on whether the grid cell continuously corresponds to the cultivated land type at multiple historical moments.

[0013] According to the embodiments of this application, the constraint attributes include at least one of the following: prohibited conversion attribute, restricted conversion attribute, priority maintenance attribute, development guidance attribute, and distance decay attribute; wherein, the prohibited conversion attribute is used to characterize the corresponding spatial constraint item restricting the raster cell from transferring into or out of a specified land use type, and the distance decay attribute is used to characterize the change in the restriction intensity of the corresponding spatial constraint item on the raster cell with spatial distance.

[0014] According to embodiments of this application, the dispersion statistics include at least one of variance, standard deviation, coefficient of variation, range, interquartile range, mean absolute deviation, and difference coefficient determined based on information distribution.

[0015] According to an embodiment of this application, determining the dispersion weight of each spatial constraint term based on the dispersion statistics of each spatial constraint term includes: performing same-scale processing on the dispersion statistics of each spatial constraint term to obtain the same-scale dispersion of each spatial constraint term; and performing a ratio calculation between the same-scale dispersion of each spatial constraint term and the sum of the same-scale dispersions of all spatial constraint terms to obtain the dispersion weight of that spatial constraint term.

[0016] According to the embodiments of this application, the initial weight corresponding to each spatial constraint item is determined based on the historical conversion frequency between each land use type, including: extracting the target historical conversion frequency corresponding to the spatial constraint item from the historical conversion frequency between each land use type according to the land use type conversion combination restricted by each spatial constraint item; determining the initial weight corresponding to the spatial constraint item based on the target historical conversion frequency, so that the initial weight corresponding to the spatial constraint item with the higher the target historical conversion frequency is.

[0017] According to an embodiment of this application, for each spatial constraint term, the initial weight and dispersion weight corresponding to the spatial constraint term are weighted and summed according to a preset fusion ratio to obtain the fusion weight corresponding to the spatial constraint term. This includes: weighting and summing the initial weight and dispersion weight corresponding to the spatial constraint term according to a preset first fusion coefficient and a preset second fusion coefficient to obtain the candidate fusion weight corresponding to the spatial constraint term; and normalizing the candidate fusion weights corresponding to each spatial constraint term to obtain the fusion weight corresponding to each spatial constraint term.

[0018] According to an embodiment of this application, after obtaining the weighted constraint strength corresponding to each grid cell, the method further includes: smoothing and correcting the weighted constraint strength corresponding to each grid cell based on the weighted constraint strength distribution of each grid cell in the neighborhood range, to obtain the corrected weighted constraint strength; wherein, the corrected weighted constraint strength is used to determine at least one of the development suitability probability, neighborhood influence parameter, and probability correction coefficient.

[0019] According to embodiments of this application, the predicted scenarios include at least two of the following: natural development scenarios, economic development scenarios, ecological protection scenarios, low-carbon optimization scenarios, and baseline scenarios; the scenario-driven data include at least one of the following: population change data, economic development data, transportation construction data, climate change data, and land use conversion preference data.

[0020] According to the embodiments of this application, for each predicted scenario, scenario transfer rules between different land use types are determined based on the scenario-driven data corresponding to the predicted scenario and the historical conversion frequency between each land use type. This includes: determining the basic transfer permission relationship between different land use types based on the historical conversion frequency between each land use type; and enhancing, suppressing, or prohibiting the basic transfer permission relationship based on the scenario-driven data corresponding to the predicted scenario to obtain scenario transfer rules.

[0021] According to the embodiments of this application, for each predicted scenario, based on the scenario-driven data, historical land use data, rasterized basic data, and the weighted constraint strength corresponding to each raster unit, the development suitability probability of each raster unit being converted from the current land use type to each candidate land use type is determined. This includes: determining historical conversion samples based on historical land use data, including raster unit samples that have undergone land use type conversion and raster unit samples that have not undergone land use type conversion; determining the development suitability score corresponding to each candidate land use type based on the historical conversion samples, rasterized basic data, scenario-driven data, and the weighted constraint strength corresponding to each raster unit; and performing probabilistic processing on the development suitability scores corresponding to each candidate land use type to obtain the development suitability probability of each raster unit being converted from the current land use type to each candidate land use type.

[0022] According to the embodiments of this application, for each prediction scenario, based on the current land use type of each grid cell, each candidate land use type, the distribution of land use types within the neighborhood of each grid cell, and the weighted constraint strength corresponding to each grid cell, the neighborhood influence parameters for converting each grid cell from its current land use type to each candidate land use type are determined. This includes: for each grid cell, counting the number of neighboring grid cells or the proportion of neighboring area belonging to each candidate land use type within the neighborhood of the grid cell; determining the basic neighborhood influence parameters for converting the grid cell from its current land use type to each candidate land use type based on the number of neighboring grid cells or the proportion of neighboring area; and adjusting the basic neighborhood influence parameters according to the weighted constraint strength corresponding to the grid cell to obtain the neighborhood influence parameters.

[0023] According to the embodiments of this application, for each predicted scenario, based on the development suitability probability, scenario transition rules, and neighborhood influence parameters corresponding to each grid cell, the initial conversion probability of each grid cell from the current land use type to each candidate land use type is calculated, including: determining the allowable conversion identifier and scenario adjustment coefficient for the conversion from the current land use type to the candidate land use type according to the scenario transition rules; and performing a product operation or weighted operation on the development suitability probability, neighborhood influence parameters, allowable conversion identifier, and scenario adjustment coefficient to obtain the initial conversion probability.

[0024] According to the embodiments of this application, for each predicted scenario, based on the weighted constraint strength corresponding to each grid cell and the scenario transition rules, the probability correction coefficient corresponding to each candidate land use type for each grid cell is determined, including: determining the constraint response type for the conversion of the current land use type to the candidate land use type according to the scenario transition rules; when the constraint response type is a restricted transition type, determining a probability correction coefficient less than or equal to 1 based on the weighted constraint strength of the corresponding grid cell; when the constraint response type is an allowed retention type, determining a probability correction coefficient equal to 1 based on the weighted constraint strength of the corresponding grid cell; when the constraint response type is a guided transition type, determining a probability correction coefficient greater than or equal to 1 based on the weighted constraint strength of the corresponding grid cell.

[0025] According to the embodiments of this application, the initial conversion probability is corrected using a probability correction coefficient to obtain the constraint-corrected conversion probability of each grid cell from the current land use type to each candidate land use type. This includes: multiplying the initial conversion probability corresponding to each candidate land use type by the corresponding probability correction coefficient to obtain the candidate correction probability corresponding to the candidate land use type; and normalizing each candidate correction probability corresponding to the same grid cell to obtain the constraint-corrected conversion probability of the grid cell from the current land use type to each candidate land use type.

[0026] According to the embodiments of this application, for each prediction scenario, based on the constraint correction conversion probability corresponding to each grid cell, the predicted land use type corresponding to each grid cell in the target prediction period is determined from each candidate land use type, including: selecting the candidate land use type with the highest constraint correction conversion probability from each candidate land use type as the predicted land use type corresponding to the corresponding grid cell in the target prediction period; or, randomly sampling based on the constraint correction conversion probability corresponding to each candidate land use type, and using the sampled candidate land use type as the predicted land use type corresponding to the corresponding grid cell in the target prediction period.

[0027] According to the embodiments of this application, when the target prediction period includes multiple prediction times, for each prediction scenario, the predicted land use type corresponding to the previous prediction time is taken as the current land use type for the next prediction time, and the determination of development suitability probability, determination of neighborhood influence parameters, calculation of initial conversion probability, calculation of constraint correction conversion probability and determination of predicted land use type are repeatedly executed until the predicted land use layout corresponding to the last prediction time in the target prediction period is obtained.

[0028] According to an embodiment of this application, before generating the predicted land use layout of the spatial area to be predicted under the prediction scenario, the method further includes: predicting the land use layout corresponding to the second historical time based on the land use type corresponding to the first historical time in the historical land use data; comparing the predicted land use layout corresponding to the second historical time with the land use type corresponding to the second historical time in the historical land use data to obtain a prediction accuracy evaluation result; and performing land use layout prediction for the target prediction period if the prediction accuracy evaluation result meets the preset accuracy conditions.

[0029] According to an embodiment of this application, after generating the predicted land use layout of the spatial area to be predicted under the prediction scenario, the method further includes: obtaining the corrected carbon density parameter corresponding to each land use type; calculating the predicted carbon storage value of each grid cell under the corresponding prediction scenario based on the predicted land use type and the corrected carbon density parameter corresponding to each grid cell in the predicted land use layout; and generating the carbon storage prediction result of the spatial area to be predicted under the prediction scenario based on the location data of each grid cell and the corresponding predicted carbon storage value.

[0030] According to the embodiments of this application, obtaining the corrected carbon density parameters corresponding to each land use type includes: obtaining the basic carbon density parameters corresponding to each land use type, wherein the basic carbon density parameters include at least one of aboveground biomass carbon density, belowground biomass carbon density, soil carbon density, and dead organic matter carbon density; and correcting the basic carbon density parameters according to the climate change conditions under the corresponding predicted scenario to obtain the corrected carbon density parameters corresponding to each land use type.

[0031] According to an embodiment of this application, the carbon storage prediction value of each grid cell under the corresponding prediction scenario is calculated based on the predicted land use type and the corrected carbon density parameter corresponding to each grid cell in the predicted land use layout. This includes: determining the corrected carbon density parameter corresponding to the grid cell based on the predicted land use type; and calculating the carbon storage prediction value corresponding to the grid cell based on the corrected carbon density parameter and the area data corresponding to the grid cell.

[0032] According to an embodiment of this application, after generating the carbon storage prediction result for the spatial area to be predicted under the prediction scenario, the method further includes: determining the carbon storage response relationship with the carbon storage prediction value corresponding to each grid cell as the target quantity, and the land use structure data, socio-economic data, and weighted constraint strength corresponding to each grid cell as input quantities; determining the nonlinear response relationship between the weighted constraint strength and the carbon storage prediction value based on the carbon storage response relationship; and dividing the weighted constraint strength into multiple response intervals based on the nonlinear response relationship.

[0033] According to an embodiment of this application, the weighted constraint strength is divided into multiple response intervals based on the nonlinear response relationship, including: determining at least one interval dividing point based on the rate of change, curvature, or marginal contribution change of the carbon storage prediction value with the weighted constraint strength in the nonlinear response relationship; and dividing the weighted constraint strength into at least two of the following intervals based on the at least one interval dividing point: low response interval, sensitive response interval, high-efficiency response interval, and response saturation interval.

[0034] According to the embodiments of this application, after dividing the weighted constraint intensity into multiple response intervals, the method further includes: determining the feedback control coefficient corresponding to each grid cell based on the response interval to which the weighted constraint intensity of each grid cell belongs; re-correcting the probability correction coefficient or constraint correction conversion probability corresponding to each grid cell based on the feedback control coefficient; and re-determining the predicted land use type corresponding to each grid cell in the target prediction period based on the re-corrected constraint correction conversion probability, so as to obtain the optimized predicted land use layout.

[0035] According to an embodiment of this application, after obtaining the optimized predicted land use layout, the method further includes: recalculating the predicted carbon storage value based on the optimized predicted land use layout to obtain the optimized carbon storage prediction result; and outputting at least one of the following: the optimized predicted land use layout, the optimized carbon storage prediction result, the weighted constraint strength corresponding to each grid cell, the feedback control coefficient corresponding to each grid cell, and the comparison results between different prediction scenarios.

[0036] This application also provides a land use layout prediction device, comprising: a data acquisition module for acquiring rasterized basic data, land use spatial constraint data, scenario-driven data, and historical land use data corresponding to the spatial area to be predicted. The rasterized basic data includes location data, natural environment data, and socioeconomic data of each raster unit within the spatial area to be predicted. The land use spatial constraint data represents constraint information on land use type conversion within the spatial area to be predicted and includes at least one spatial constraint term. The spatial constraint term is a spatial data item in the land use spatial constraint data used to represent a restriction on land use type conversion. The scenario-driven data is used to represent the social factors affecting land use type conversion under different prediction scenarios. The system considers economic development conditions and climate change conditions. Historical land use data includes land use types corresponding to each grid cell at multiple historical moments. The historical conversion frequency determination module is used to determine the historical land use type conversion relationship between adjacent historical moments for each grid cell based on the historical land use data, and to determine the historical conversion frequency between each land use type based on the historical land use type conversion relationship. The first constraint strength determination module is used to determine the first constraint strength of each spatial constraint item in the land use spatial constraint data based on the spatial scope of the spatial constraint item, the constraint attribute, and the spatial relationship between the spatial constraint item and each grid cell. The discrete weight determination module is used to calculate the discrete statistical value corresponding to each spatial constraint item based on the numerical distribution of the first constraint strength of the spatial constraint item across all raster cells, and to determine the discrete weight corresponding to each spatial constraint item based on the discrete statistical value of each spatial constraint item; the initial weight determination module is used to determine the initial weight corresponding to each spatial constraint item based on the historical conversion frequency between each land use type; the fusion weight determination module is used to calculate the fusion weight corresponding to each spatial constraint item by weighting and summing the initial weight and discrete weight of the spatial constraint item according to a preset fusion ratio; the weighted constraint strength determination module is used to determine the discrete weight corresponding to each raster cell. The unit calculates the weighted constraint strength of each spatial constraint term by multiplying its first constraint strength by the corresponding fusion weight and summing the results. The scenario transition rule determination module determines the scenario transition rules between different land use types for each predicted scenario, based on the scenario-driven data and historical conversion frequencies between land use types. The development suitability probability determination module determines the development suitability probability of each raster unit for each predicted scenario, based on the scenario-driven data, historical land use data, rasterized basic data, and the weighted constraint strength of each raster unit.The neighborhood influence parameter determination module, for each prediction scenario, determines the neighborhood influence parameters for each grid cell to transform from its current land use type to each candidate land use type, based on the current land use type of each grid cell, each candidate land use type, the distribution of land use types within the neighborhood of each grid cell, and the weighted constraint strength corresponding to each grid cell. The initial conversion probability calculation module, for each prediction scenario, calculates the initial conversion probability for each grid cell to transform from its current land use type to each candidate land use type, based on the development suitability probability, scenario transition rules, and neighborhood influence parameters corresponding to each grid cell. The constraint correction conversion probability calculation module, for each prediction scenario, calculates the initial conversion probability for each grid cell to transform from its current land use type to each candidate land use type, based on the development suitability probability, scenario transition rules, and neighborhood influence parameters. The weighted constraint strength and scenario transition rules corresponding to each grid cell determine the probability correction coefficient for each candidate land use type. This probability correction coefficient is then used to correct the initial conversion probability, resulting in the constraint-corrected conversion probability of each grid cell transitioning from its current land use type to its candidate land use type. The predicted layout generation module, for each prediction scenario, determines the predicted land use type for each grid cell in the target prediction period from among the candidate land use types based on the constraint-corrected conversion probability of each grid cell. Based on the location data of each grid cell and its corresponding predicted land use type, it generates the predicted land use layout for the spatial area to be predicted under that prediction scenario.

[0037] This application also provides an electronic device, including a processor and a memory, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, it implements any of the aforementioned land use layout prediction methods.

[0038] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the aforementioned land use layout prediction methods.

[0039] The land use layout prediction method, apparatus, and electronic equipment provided in this application acquire rasterized basic data, land use spatial constraint data, scenario-driven data, and historical land use data of the spatial area to be predicted. First, based on historical land use type conversion relationships, the historical conversion frequency between each land use type is determined. Then, for different spatial constraint items in the land use spatial constraint data, the first constraint strength on each raster unit is determined according to its spatial scope, constraint attribute, and spatial relationship with each raster unit. Furthermore, the dispersion weight is determined based on the numerical distribution of the first constraint strength across all raster units. This is combined with the initial weight obtained based on the historical conversion frequency to form a fusion weight, thereby obtaining a weighted constraint strength for each raster unit that reflects the combined effect of multiple spatial constraint items. Based on this, this application incorporates the weighted constraint strength into... Determining the development suitability probability, neighborhood influence parameters, and initial conversion probability under different prediction scenarios allows land use spatial constraints to no longer be treated as a constant overall variable affecting the entire prediction area. Instead, they are subdivided into grid-level dynamic influence quantities that vary with grid location, spatial constraint distribution, constraint attributes, historical conversion characteristics, and prediction scenarios. This addresses the problem in existing technologies using land use models such as PLUS to predict land use distribution, where policy constraints are typically treated as an overall variable and applied to all layout predictions in a constant manner, making it impossible to flexibly adjust the land use conversion probabilities of different regions under policy constraints. This allows the conversion probabilities of different grid units between different candidate land use types to be corrected based on the actual differences in spatial constraints, improving the spatial refinement and accuracy of land use layout prediction under policy constraints.

[0040] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0041] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart of an embodiment of the land use layout prediction method provided in this application; Figure 2 A schematic diagram of a structure of an embodiment of the land use layout prediction device provided in this application; Figure 3 A schematic diagram of the structure of an embodiment of the electronic device provided in this application. Detailed Implementation

[0042] Exemplary embodiments of the present application will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this application will be thorough and complete, and will fully convey the scope of the present application to those skilled in the art.

[0043] In the description of this application, it should be noted that the terms "first," "second," etc., are used only for descriptive purposes and should not be construed as indicating or implying relative importance, nor as limiting the order. Unless otherwise expressly specified and limited, the terms "connection," "communication connection," "coupled," etc., should be interpreted broadly. For example, they can refer to fixed connections or detachable connections; direct connections or indirect connections through intermediate components; mechanical connections, electrical connections, communication connections, or data connections. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0044] In this embodiment, the raster unit can be a basic spatial unit obtained by rasterizing the spatial area to be predicted. Land use types can include cultivated land, forest land, grassland, water area, construction land, and unused land. Spatial constraints can be spatial data items that can restrict or guide the conversion of land use types, such as functional zoning constraints, ecological protection red line constraints, urban development boundary constraints, and permanent basic farmland constraints. The prediction scenario can be set according to actual planning objectives, socio-economic development conditions, climate change conditions, or ecological protection objectives.

[0045] First Embodiment Land use layout prediction is a crucial technical component in national spatial planning, ecological protection pattern optimization, urban expansion management, and regional resource allocation. With the development of remote sensing, geographic information systems (GIS), and spatial analysis technologies, grid-based land use simulation methods are widely used to predict the spatial distribution changes of different land use types in the future. Existing techniques typically calculate the probability of different grid cells transitioning to different land use types based on historical land use data, natural environment data, socioeconomic data, and land use type conversion rules, and then generate the land use layout for the predicted period.

[0046] In existing technologies, land use simulation methods such as the PLUS model can combine land expansion analysis strategies and cellular automata simulation mechanisms to predict spatial processes such as construction land expansion, arable land change, and ecological land change. These methods can typically extract driving factors from historical change samples and generate land use prediction results using neighborhood effects, random perturbations, and transition matrices, thus having a certain application foundation in the field of land use change simulation.

[0047] However, in actual land use planning scenarios, land use change is not only affected by natural environmental and socio-economic conditions, but also by policy or planning constraints such as ecological protection red lines, permanent basic farmland, urban development boundaries, and functional zones. Existing technologies, when introducing such policy constraints, typically treat them as overall variables, uniform transfer rules, or fixed limitations applied to the entire prediction area. This makes it difficult to reflect the differentiated impacts of policy constraints on land use type transformation under different spatial locations, constraint types, historical transformation characteristics, and prediction scenarios.

[0048] For example, some grid cells may be located within ecological protection red lines, where the expansion of construction land should be subject to strong restrictions; some grid cells may be located near urban development boundaries, where the transfer of construction land may have high spatial suitability; and some grid cells may be located within permanent basic farmland, where the preservation of arable land should have a high priority. If existing models only use uniform policy constraint parameters to handle the above-mentioned different regions, it will be difficult to achieve flexible adjustments to land use layout under policy constraints, which may easily lead to a mismatch between the prediction results and actual spatial control requirements, thereby affecting the accuracy of the prediction.

[0049] Based on this, embodiments of this application provide a land use layout prediction method, apparatus, electronic device, and computer-readable storage medium. The method acquires rasterized basic data, land use spatial constraint data, scenario-driven data, and historical land use data corresponding to the spatial area to be predicted, and determines historical land use type conversion relationships and historical conversion frequencies. For different spatial constraint items, it determines a first constraint intensity by combining the spatial scope, constraint attributes, and spatial relationships between the spatial constraint items and each raster unit. Then, it determines the dispersion weight based on the numerical distribution of the first constraint intensity, and combines it with the initial weight determined based on the historical conversion frequency to obtain a fusion weight, thereby obtaining the weighted constraint intensity corresponding to each raster unit.

[0050] After obtaining the weighted constraint strength, this embodiment further incorporates the weighted constraint strength into the determination of development suitability probability, the determination of neighborhood influence parameters, and the correction of conversion probability. Thus, policy constraints or planning constraints that originally acted as a whole on the prediction area can be transformed into grid-level, differentiated, and dynamic constraint influence quantities. This allows the conversion probability of different grid units between different candidate land use types to be corrected according to the actual spatial constraint differences, thereby improving the spatial refinement and accuracy of land use layout prediction.

[0051] Therefore, the embodiments of this application can solve the problem that when using land use models such as PLUS to predict land use distribution in the prior art, policy constraints can only be applied as a whole variable in a constant manner to all layout predictions, and it is impossible to flexibly adjust the land use conversion probability of different regions under policy constraints. The embodiments of this application can subdivide policy constraints into dynamic effects for different regions, different spatial constraints, and different scenario conditions, thereby improving the flexibility, adaptability, and accuracy of policy constraints in land use layout prediction.

[0052] Second Embodiment This embodiment combines Figure 1 This application describes a land use layout prediction method provided in its embodiments. Figure 1 A flowchart of a land use layout prediction method according to an embodiment of this application is shown. This method can be executed by an electronic device, a server, a computer device, a geographic information processing platform, or a land use simulation system. The land use layout prediction method according to an embodiment of this application includes: S101, acquire the rasterized basic data, land use spatial constraint data, scenario-driven data and historical land use data corresponding to the spatial area to be predicted.

[0053] In step S101, rasterized basic data, land use spatial constraint data, scenario-driven data, and historical land use data of the spatial area to be predicted can be acquired. For example, in this embodiment, the spatial area to be predicted can be an urban area, a watershed area, an administrative region, an ecological function zone, a national land spatial planning unit, or other spatial areas that require land use layout prediction. The electronic device can acquire relevant data from remote sensing image databases, geographic information system databases, planning and control databases, statistical yearbook databases, climate scenario databases, or data files uploaded by users.

[0054] Rasterized base data can include location data, natural environmental data, and socioeconomic data for each raster cell within the spatial area to be predicted. Location data is used to determine the spatial location of the raster cell within the spatial area to be predicted, natural environmental data is used to characterize the natural geographical conditions that influence land use change, and socioeconomic data is used to characterize the driving forces of human activities, economic development, and transportation accessibility on land use change.

[0055] Land use spatial constraint data is used to characterize the constraints on land use type transformation within the spatial area to be predicted. This data may include one or more spatial constraint terms, each representing a limiting or guiding factor for land use type transformation. For example, ecological protection red lines can restrict the inflow of construction land, permanent basic farmland can restrict the outflow of arable land, urban development boundaries can guide the concentration of construction land within or near the boundaries, and functional zones can reflect the development or protection orientation of different spatial areas.

[0056] Scenario-driven data can be used to characterize socioeconomic development and climate change conditions that influence land use type transformation under different forecast scenarios. Historical land use data can include the land use types corresponding to each raster cell at multiple historical moments, and can be used to analyze historical transformation patterns between land use types.

[0057] Furthermore, in this embodiment, the location data includes at least one of grid number, grid center point coordinates, and grid boundary coordinates; the natural environment data includes at least one of elevation, slope, aspect, average annual temperature, annual precipitation, soil type, distance from water system, and distance from road; and the socio-economic data includes at least one of population density, regional GDP, nighttime light intensity, distance from settlement, distance from town center, and traffic accessibility data.

[0058] In practical implementation, data from different sources, with different spatial resolutions, and different coordinate systems can first be uniformly converted to the same spatial reference frame, and then resampled, cropped, and registered according to a preset raster scale, so that each data layer can correspond to the same raster unit. This ensures that various driving factors, constraint factors, and historical land use types are spatially consistent when calculating the transformation probability later.

[0059] Furthermore, in this embodiment, the land use type includes at least two of the following: cultivated land, forest land, grassland, water area, construction land, and unused land; the current land use type is the land use type of the corresponding grid cell at the start of the prediction, and the candidate land use type is the land use type that the corresponding grid cell is allowed to be determined in the target prediction period.

[0060] In practical applications, land use types can be further refined based on the characteristics of the study area and the accuracy of the data. For example, forest land can be further divided into wooded land, shrubland, and sparse forest; construction land can be further divided into urban construction land, rural residential land, and industrial and mining land. The current land use type can be determined by the land use raster map of the prediction start year, and candidate land use types can be determined based on the land use classification system, planning control requirements, or scenario transition rules.

[0061] Furthermore, in this embodiment, the spatial constraints include at least one of the following: functional zone constraints, ecological protection red line constraints, urban development boundary constraints, and permanent basic farmland constraints.

[0062] In practical implementation, the main functional area constraint can reflect the differences in functional zones such as optimized development, key development, restricted development, and prohibited development; the ecological protection red line constraint can reflect the intensity of ecological space protection; the urban development boundary constraint can reflect the boundary of urban construction space expansion; and the permanent basic farmland constraint can reflect the constraints on cultivated land protection. These spatial constraints can be used individually or in combination to enhance the adaptability of land use layout prediction to the actual requirements of national land space management.

[0063] Furthermore, in the embodiments of this application, the predicted scenarios include at least two of the following: natural development scenarios, economic development scenarios, ecological protection scenarios, low-carbon optimization scenarios, and baseline scenarios; the scenario-driven data include at least one of the following: population change data, economic development data, transportation construction data, climate change data, and land use conversion preference data.

[0064] In practice, the natural development scenario can primarily continue historical land use change trends; the economic development scenario can increase the tendency for construction land to be converted into land; the ecological protection scenario can strengthen the preservation of ecological land and restrictions on the conversion of construction land; and the low-carbon optimization scenario can adjust land use conversion preferences in conjunction with carbon storage or ecosystem service targets. By setting different prediction scenarios, multiple possible future land use layouts can be obtained, providing a basis for comparison in spatial planning decisions.

[0065] S102, Based on historical land use data, determine the historical land use type conversion relationship between adjacent historical moments for each grid cell, and determine the historical conversion frequency between each land use type based on the historical land use type conversion relationship.

[0066] In step S102, the land use raster data from multiple historical moments obtained in step S101 can be analyzed by overlaying each raster cell. For any raster cell, the land use type of that raster cell can be compared to see if it has changed between two adjacent historical moments, thereby determining the historical land use type conversion relationship corresponding to that raster cell.

[0067] For example, if a grid cell was cultivated land at the first historical moment and became construction land at the second historical moment, the conversion relationship between the two historical moments can be recorded as cultivated land to construction land. If a grid cell was forest land at both historical moments, it can be recorded as forest land remaining unchanged. By statistically analyzing all grid cells within the spatial area to be predicted, the historical conversion frequency between each land use type can be obtained.

[0068] Historical conversion frequency can be represented in the form of a conversion matrix. Rows in the conversion matrix represent land use types before conversion, columns represent land use types after conversion, and matrix elements represent the number of conversions, conversion area, or conversion ratio of corresponding land use type combinations within a historical period. This historical conversion frequency can be used to reflect the actual conversion activity between different land use types within the spatial area to be predicted.

[0069] In subsequent steps, the historical transformation frequency can be used to determine the initial weights corresponding to spatial constraints, and also to determine the scenario transition rules between land use types under different prediction scenarios. Therefore, the embodiments of this application enable the land use layout prediction process to have both a foundation in historical change patterns and the ability to dynamically adjust by combining spatial constraints and scenario-driven approaches.

[0070] S103, for each spatial constraint item in the land use spatial constraint data, determine the first constraint strength of the spatial constraint item on each grid cell according to the spatial range of action, constraint attribute and spatial relationship between the spatial constraint item and each grid cell.

[0071] In step S103, each spatial constraint item in the spatial constraint data obtained in step S101 can be used to determine the first constraint strength of the spatial constraint item on each grid cell based on its corresponding spatial range, constraint attribute, and spatial relationship with each grid cell. For example, in this embodiment, the spatial constraint item can correspond to different spatial ranges and constraint attributes. The spatial range can be a planar range, a linear boundary, the influence range of a point facility, or a constraint range formed by a combination of multiple spatial objects. The constraint attribute can represent constraint methods such as prohibiting conversion, restricting conversion, maintaining priority, development guidance, or distance attenuation.

[0072] For each spatial constraint, the first constraint strength can be determined based on the spatial relationship between the spatial constraint and each grid cell. The spatial relationship can include whether the grid cell is within the constraint range, the distance between the grid cell and the constraint boundary, the proximity between the grid cell and the constrained object, and the functional partition type to which the grid cell belongs. The first constraint strength can be represented numerically, for example, using a continuous value between 0 and 1, or it can be represented by a tiered value to indicate weak, medium, and strong constraints.

[0073] For example, for the ecological protection red line constraint, grid cells located within the ecological protection red line can be assigned a higher first constraint strength, while grid cells located outside the ecological protection red line can be assigned a lower first constraint strength. For the urban development boundary constraint, the first constraint strength can be determined based on the distance of the grid cell to the urban development boundary, so that grid cells within or near the boundary have different constraint or guidance strengths than grid cells far from the boundary.

[0074] The first constraint strength is not directly equivalent to the final transformation probability, but rather describes the strength of the effect of a single spatial constraint term on a single grid cell. By calculating the first constraint strength for each spatial constraint term separately, macroeconomic policy constraints or planning constraints can be transformed into grid-level constraint parameters with spatial differences.

[0075] Furthermore, in this embodiment of the application, for each spatial constraint item in the land use spatial constraint data, the first constraint intensity of the spatial constraint item on each grid cell can be determined according to the spatial scope of the spatial constraint item, the constraint attribute, and the spatial relationship between the spatial constraint item and each grid cell. This includes: when the spatial constraint item is a main functional area constraint item, determining the first constraint intensity corresponding to the grid cell according to the main functional area type to which the grid cell belongs; when the spatial constraint item is an ecological protection red line constraint item, determining the first constraint intensity corresponding to the grid cell according to whether the grid cell is located within the ecological protection red line; when the spatial constraint item is an urban development boundary constraint item, determining the first constraint intensity corresponding to the grid cell according to the spatial distance between the grid cell and the urban development boundary; when the spatial constraint item is a permanent basic farmland constraint item, determining the first constraint intensity corresponding to the grid cell according to whether the grid cell is located within the permanent basic farmland or according to whether the grid cell continuously corresponds to the cultivated land type at multiple historical moments.

[0076] In practical implementation, the functional zoning constraint can be assigned different constraint levels based on different functional zones. For example, higher restriction intensity can be set for the transfer of construction land in prohibited development areas, while lower restriction intensity or higher guidance intensity can be set for the transfer of construction land in key development areas. The ecological protection red line constraint can be determined by spatial overlay to determine whether the grid cell falls within the red line range, and the constraint intensity can be determined accordingly. The urban development boundary constraint can be determined by using a distance decay function. The permanent basic farmland constraint can be determined by overlaying the permanent basic farmland vector range with the grid cell, or by combining the farmland preservation status at multiple historical moments to determine the stability of farmland protection.

[0077] Furthermore, in the embodiments of this application, the constraint attributes include at least one of the following: prohibited conversion attribute, restricted conversion attribute, priority maintenance attribute, development guidance attribute, and distance decay attribute; wherein, the prohibited conversion attribute is used to characterize the corresponding spatial constraint item restricting the raster cell from entering or leaving the specified land use type, and the distance decay attribute is used to characterize the restriction intensity of the corresponding spatial constraint item on the raster cell as the spatial distance changes.

[0078] In practical implementation, the prohibition of conversion attribute can be used in scenarios such as prohibiting the conversion of land within ecological protection red lines to construction land and prohibiting the conversion of permanent basic farmland to non-agricultural construction land; the restriction of conversion attribute can indicate that a certain type of conversion is not completely prohibited but the conversion probability needs to be reduced; the priority preservation attribute can be used to encourage cultivated land, forest land or ecological land to maintain its original type; the development guidance attribute can be used for the agglomeration of construction land within urban development boundaries; the distance decay attribute can make the constraint strength change continuously with distance, thereby avoiding abrupt changes in simulation results caused by simple binary constraints.

[0079] S104. For each spatial constraint term, calculate the discreteness statistics corresponding to the spatial constraint term based on the numerical distribution of the first constraint strength of the spatial constraint term on all grid cells, and determine the discreteness weight corresponding to each spatial constraint term based on the discreteness statistics corresponding to each spatial constraint term.

[0080] In step S104, the first constraint strength obtained in step S103 can be used to calculate the dispersion statistics of each spatial constraint term obtained in step S101 based on the numerical distribution of the first constraint strength of the spatial constraint term across all grid cells. The dispersion weight of each spatial constraint term is then determined based on the dispersion statistics. For example, in this embodiment, the first constraint strength of each spatial constraint term across all grid cells can be collected to form a constraint strength sequence corresponding to that spatial constraint term. By analyzing the numerical distribution of this constraint strength sequence, it can be determined whether the spatial constraint term has significant spatial differences within the study area.

[0081] If the first constraint strength of a spatial constraint term varies significantly across different raster cells, it indicates that the spatial constraint term has a strong distinguishing effect on different regions; if the first constraint strength of a spatial constraint term is basically the same across all raster cells, it indicates that the spatial constraint term has a weak ability to express spatial differences. Dispersion statistics can be used to quantify this degree of difference.

[0082] After obtaining the dispersion statistics for each spatial constraint term, the dispersion weight can be determined based on these statistics. A larger dispersion weight indicates a more significant difference in constraint strength across different grid cells for the corresponding spatial constraint term, and thus a higher contribution to the overall constraint strength calculation.

[0083] By introducing dispersion weights, the embodiments of this application can avoid all spatial constraint terms participating in the comprehensive constraint strength calculation with the same weight, thereby enabling constraint terms with stronger spatial differentiation capabilities to play a more reasonable role in the land use layout prediction process.

[0084] Furthermore, in the embodiments of this application, the dispersion statistics include at least one of variance, standard deviation, coefficient of variation, range, interquartile range, mean absolute deviation, and difference coefficient determined based on the information distribution.

[0085] In practical implementation, variance and standard deviation can reflect the degree of fluctuation of the first constraint strength around the mean; the coefficient of variation can be used to eliminate the influence of differences in dimensions or mean; the range can reflect the span between the maximum and minimum values; the interquartile range can reduce the influence of extreme values; the mean absolute deviation can reflect the overall dispersion; and the coefficient of variation determined based on the information distribution can be used to measure the information difference in the constraint strength distribution. In practical applications, one or more dispersion statistics can be selected according to the data distribution characteristics.

[0086] Furthermore, in this embodiment of the application, the dispersion weight of each spatial constraint term can be determined based on the dispersion statistics of each spatial constraint term, including: scaling the dispersion statistics of each spatial constraint term to obtain the same-scale dispersion of each spatial constraint term; and performing a ratio calculation between the same-scale dispersion of each spatial constraint term and the sum of the same-scale dispersions of all spatial constraint terms to obtain the dispersion weight of that spatial constraint term.

[0087] In practical implementation, the same-scale processing can include range standardization, mean standardization, normalization, or other dimensionless processing. Same-scale processing allows for the comparison of different types of dispersion statistics at the same scale. Subsequently, dividing the same-scale dispersion corresponding to each spatial constraint term by the sum of the same-scale dispersions corresponding to all spatial constraint terms yields the dispersion weights, ensuring that the sum of the dispersion weights for each spatial constraint term is 1, facilitating subsequent fusion processing.

[0088] S105. Determine the initial weights of each spatial constraint term based on the historical conversion frequency between different land use types.

[0089] In step S105, the initial weights corresponding to each spatial constraint can be determined using the historical conversion frequencies between land use types determined in step S102. For example, in this embodiment, the initial weights can be used to characterize the preset configuration proportions of spatial constraint items in land use type conversion constraints. Different spatial constraint items typically play a role in different land use type conversion processes. For example, the permanent basic farmland constraint mainly affects the process of converting arable land out, the ecological protection red line constraint mainly affects the process of converting ecological land out or construction land in, and the urban development boundary constraint mainly affects the process of expanding construction land.

[0090] The activity level of land use type conversions in the spatial region to be predicted can be analyzed based on historical conversion frequency, and the importance of different spatial constraints can be further determined. If the land use type conversion restricted by a certain spatial constraint has occurred frequently in historical periods, it indicates that this type of conversion is relatively active in the region, and it is highly necessary to effectively constrain this conversion. Therefore, a higher initial weight can be set for the corresponding spatial constraint.

[0091] The initial weights can be obtained by directly normalizing the historical conversion frequency, or they can be determined by combining expert experience, planning and control rules, and the land use conversion matrix. These initial weights reflect the correspondence between spatial constraints and historical land use conversion processes.

[0092] By using historical conversion frequency to determine initial weights, the embodiments of this application enable the constraint weights to no longer rely entirely on manual setting, but to be determined in combination with historical land use change patterns, thereby improving the objectivity of weight setting.

[0093] Furthermore, in this embodiment of the application, the initial weight corresponding to each spatial constraint can be determined based on the historical conversion frequency between each land use type, including: extracting the target historical conversion frequency corresponding to the spatial constraint from the historical conversion frequency between each land use type based on the land use type conversion combination restricted by each spatial constraint; determining the initial weight corresponding to the spatial constraint based on the target historical conversion frequency, so that the initial weight corresponding to the spatial constraint with the higher the target historical conversion frequency is.

[0094] In practical implementation, a correspondence between spatial constraints and land use type conversion combinations can be established in advance. For example, the permanent basic farmland constraint can correspond to combinations such as conversion from cultivated land to construction land and conversion from cultivated land to unused land; the ecological protection red line constraint can correspond to combinations such as conversion from forest land, grassland, and water area to construction land; and the urban development boundary constraint can correspond to combinations of conversion from non-construction land to construction land. Then, the conversion frequency of the corresponding combinations is extracted from historical conversion frequencies, and the initial weights are calculated accordingly.

[0095] S106. For each spatial constraint term, the initial weight and dispersion weight corresponding to the spatial constraint term are weighted and summed according to a preset fusion ratio to obtain the fusion weight corresponding to the spatial constraint term.

[0096] In step S106, the dispersion weights obtained in step S104 and the initial weights obtained in step S105 can be weighted and summed according to a preset fusion ratio to obtain the fusion weights corresponding to the spatial constraint term. For example, in this embodiment, the initial weights reflect the relationship between the spatial constraint term and the frequency of historical land use conversions, while the dispersion weights reflect the spatial difference expression ability of the spatial constraint term across different grid cells. Both describe the contribution of the spatial constraint term from the perspectives of historical conversion importance and spatial difference significance, respectively.

[0097] The initial weights and dispersion weights can be weighted and summed according to a preset fusion ratio to obtain the fusion weights. The preset fusion ratio can be set according to the application scenario. For example, when more attention is paid to historical transformation patterns, the ratio of the initial weights can be increased; when more attention is paid to differences in spatial constraints, the ratio of the dispersion weights can be increased.

[0098] The fusion weight can be used as the comprehensive weight of each spatial constraint term when calculating the weighted constraint strength in the subsequent process. By using the fusion weight, the embodiments of this application can simultaneously consider the historical land use change patterns and the ability to express the differences in spatial constraint terms, avoiding the result deviation caused by determining the weight of spatial constraint terms based on only a single factor.

[0099] In practical implementation, the fusion weights corresponding to each spatial constraint term can be normalized so that the sum of the fusion weights of all spatial constraint terms is 1, so that they can participate in the weighted summation calculation in a unified manner.

[0100] Furthermore, in this embodiment of the application, for each spatial constraint term, the initial weight and dispersion weight corresponding to the spatial constraint term can be weighted and summed according to a preset fusion ratio to obtain the fusion weight corresponding to the spatial constraint term. This includes: weighting and summing the initial weight and dispersion weight corresponding to the spatial constraint term according to a preset first fusion coefficient and a preset second fusion coefficient to obtain the candidate fusion weight corresponding to the spatial constraint term; and normalizing the candidate fusion weights corresponding to each spatial constraint term to obtain the fusion weight corresponding to each spatial constraint term.

[0101] In practical implementation, the preset first fusion coefficient and the preset second fusion coefficient can either have a sum of 1, or they can be uniformly normalized after calculating the candidate fusion weights. Through the calculation and normalization of candidate fusion weights, the comparability of weights between different spatial constraint terms can be ensured, and the inconsistency of fusion weight scales can be avoided from affecting subsequent calculations.

[0102] S107, For each grid cell, the first constraint strength of each spatial constraint term on the grid cell is multiplied by the corresponding fusion weight and summed to obtain the weighted constraint strength corresponding to the grid cell.

[0103] In step S107, the constraint strengths obtained in step S103 and the fusion weights determined in step S106 can be multiplied and summed for each grid cell to obtain the weighted constraint strength corresponding to that grid cell. For example, in this embodiment, for any grid cell, the first constraint strength of each spatial constraint term on that grid cell can be obtained, and the fusion weight corresponding to each spatial constraint term can be obtained. By multiplying the first constraint strength of each spatial constraint term with the corresponding fusion weight and summing the results, the weighted constraint strength corresponding to that grid cell can be obtained.

[0104] Weighted constraint intensity is used to characterize the comprehensive restriction or guidance degree of multiple spatial constraints on the land use type transformation of the same grid unit. Compared with a single spatial constraint, weighted constraint intensity can comprehensively reflect the superimposed effect of multiple spatial constraints such as functional zones, ecological protection red lines, urban development boundaries, and permanent basic farmland.

[0105] For example, if a grid cell is located within both the ecological protection red line and a restricted development zone, its weighted constraint strength may be relatively high; conversely, if another grid cell is located within the urban development boundary and far from the ecological protection space, its constraint strength regarding the transfer of construction land may be relatively low or its guidance strength relatively high. By calculating the grid-level weighted constraint strength, the overall policy constraint can be transformed into dynamic constraint parameters specific to each grid cell.

[0106] The weighted constraint strength can be used to determine the probability of subsequent development suitability, the parameters of neighborhood influence, and the probability correction coefficient, so that policy constraints can run through multiple stages of land use layout prediction.

[0107] Furthermore, in this embodiment, after obtaining the weighted constraint strength corresponding to each grid cell, the method further includes: smoothing and correcting the weighted constraint strength corresponding to each grid cell based on the distribution of the weighted constraint strength in the neighborhood range, to obtain the corrected weighted constraint strength; wherein the corrected weighted constraint strength is used to determine at least one of the development suitability probability, neighborhood influence parameter, and probability correction coefficient.

[0108] In practical implementation, methods such as moving windows, neighborhood mean, median filtering, or distance-weighted averaging can be used to smooth the weighted constraint strength. Smoothing correction can reduce the abrupt impact of data overlay errors or rasterization errors near the spatial constraint boundary, making the weighted constraint strength more continuous and stable in space, thereby improving the spatial rationality of the prediction results.

[0109] S108. For each predicted scenario, based on the scenario-driven data corresponding to the predicted scenario and the historical conversion frequency between different land use types, determine the scenario transfer rules between different land use types.

[0110] In step S108, the historical conversion frequency obtained in step S102 and the scenario-driven data determined in step S101 can be used to determine scenario transition rules between different land use types for each prediction scenario. For example, in this embodiment, the prediction scenario can correspond to different socio-economic development conditions, ecological protection goals, or climate change conditions. Under different prediction scenarios, the conversion relationships between different land use types may differ. For example, under an economic development scenario, the possibility of converting non-construction land to construction land may increase; under an ecological protection scenario, the expansion of construction land to ecological land is more restricted, and the probability of maintaining forest land, grassland, and water areas may increase.

[0111] Scenario-based transition rules can include whether transitions between land use types are permitted, transition intensity adjustment coefficients, transition priorities, or transition restriction levels. Basic transition rules can be determined first based on historical transition frequencies, and then adjusted according to scenario-driven data corresponding to the predicted scenarios.

[0112] Through scenario transfer rules, the embodiments of this application can enable the same area to generate different land use layout prediction results under different prediction scenarios, thereby supporting multi-scenario comparison and planning decisions.

[0113] Scenario transition rules can be stored as matrices, rule tables, or parameter sets. For each combination of current land use type and candidate land use type, a transition flag, scenario adjustment coefficient, or constraint response type can be set to facilitate subsequent calculation of the initial transition probability and probability correction coefficient.

[0114] Furthermore, in this embodiment of the application, for each predicted scenario, scenario transfer rules between different land use types can be determined based on the scenario-driven data corresponding to the predicted scenario and the historical conversion frequency between each land use type. This includes: determining the basic transfer permission relationship between different land use types based on the historical conversion frequency between each land use type; and enhancing, suppressing, or prohibiting the basic transfer permission relationship based on the scenario-driven data corresponding to the predicted scenario to obtain scenario transfer rules.

[0115] In practical implementation, the frequency of historical conversions can be used to determine which land use type conversions have a historical basis within the region, and then adjustments can be made based on scenario-driven data. For example, under an ecological protection scenario, the conversion of arable land, forest land, grassland, and water areas into construction land can be suppressed; under an economic development scenario, the conversion of non-construction land into construction land around transportation corridors or within urban development boundaries can be enhanced; and under a low-carbon optimization scenario, the probability of forest land and grassland being retained or converted can be increased.

[0116] S109, For each prediction scenario, based on the scenario-driven data, historical land use data, rasterized basic data, and the weighted constraint strength corresponding to each raster unit, determine the development suitability probability of each raster unit from the current land use type to each candidate land use type.

[0117] In step S109, the scenario-driven data, historical land use data, rasterized basic data, and weighted constraint strength determined in step S107 obtained in step S101 can be used to determine the development suitability probability of each raster cell converting from its current land use type to each candidate land use type for each prediction scenario. For example, in this embodiment, the development suitability probability is used to characterize the degree of suitability of each raster cell converting from its current land use type to each candidate land use type under a given prediction scenario. This probability can comprehensively reflect natural environmental conditions, socio-economic conditions, historical land use change patterns, scenario-driven factors, and spatial constraint strength.

[0118] Machine learning models, statistical models, logistic regression models, random forest models, neural network models, or other suitability assessment models can be used to train development suitability models for different candidate land use types based on historical land use change samples and driving factor data. Model inputs can include rasterized basic data, scenario-driven data, and weighted constraint strengths; the output can be the development suitability probability corresponding to different candidate land use types.

[0119] Using the weighted constraint strength as an input parameter in determining the development suitability probability allows the model to consider the impact of policy constraints during the suitability assessment stage. For example, in ecological protection areas with higher weighted constraint strength, the probability of converting construction land to suitability can be reduced; while in areas with higher development guidance strength, the probability of converting construction land to suitability can be increased.

[0120] The development suitability probability can be normalized to ensure that the development suitability probabilities of each candidate land use type corresponding to the same grid cell are at a uniform scale, which facilitates the subsequent calculation of the initial conversion probability in combination with scenario transition rules and neighborhood influence parameters.

[0121] Furthermore, embodiments of this application can also, for each predicted scenario, determine the development suitability probability of each grid cell converting from the current land use type to each candidate land use type based on the scenario-driven data, historical land use data, rasterized basic data, and the weighted constraint strength corresponding to each grid cell. This includes: determining historical conversion samples based on historical land use data, including grid cell samples that have undergone land use type conversion and grid cell samples that have not undergone land use type conversion; determining the development suitability score corresponding to each candidate land use type based on the historical conversion samples, rasterized basic data, scenario-driven data, and the weighted constraint strength corresponding to each grid cell; and performing probabilistic processing on the development suitability scores corresponding to each candidate land use type to obtain the development suitability probability of each grid cell converting from the current land use type to each candidate land use type.

[0122] In practical implementation, raster cell samples that have undergone land use type conversion can be used as positive samples, while raster cell samples that have not undergone land use type conversion can be used as negative samples. By training the suitability evaluation model, a development suitability score corresponding to each candidate land use type can be obtained. Subsequently, a normalization function, logistic function, softmax function, or piecewise mapping function can be used to convert the development suitability score into a development suitability probability.

[0123] S110, for each prediction scenario, based on the current land use type of each grid cell, each candidate land use type, the distribution of land use types within the neighborhood of each grid cell, and the weighted constraint strength corresponding to each grid cell, determine the neighborhood influence parameter for converting the current land use type of each grid cell into each candidate land use type.

[0124] In step S110, the neighborhood influence parameters for converting the current land use type to each candidate land use type can be determined using the current land use type, each candidate land use type, the land use type distribution within the neighborhood of each grid cell obtained in step S101, and the weighted constraint strength determined in step S107. In this embodiment, the neighborhood influence parameter is used to characterize the impact of the surrounding land use type distribution on the land use conversion of the target grid cell. Land use changes often have spatial clustering characteristics; for example, the expansion of construction land often occurs around existing construction land, and changes in forest land or grassland may also be affected by the continuity of surrounding ecological land.

[0125] A neighborhood range can be set for each grid cell, such as a 3×3, 5×5, or 7×7 window, or the neighborhood range can be determined based on spatial distance. Then, the number of grid cells or the area ratio of each land use type within the neighborhood range are counted to determine the basic neighborhood influence parameter of the candidate land use type.

[0126] In this embodiment, the weighted constraint strength is also involved in determining the neighborhood influence parameter. This avoids the neighborhood expansion effect being mechanically amplified within strongly constrained areas. For example, if a grid cell has a high proportion of surrounding construction land, but the grid cell is located within an ecological protection red line, the weighted constraint strength can suppress the driving effect of the construction land neighborhood effect on the conversion probability of that grid cell.

[0127] Therefore, the embodiments of this application can achieve synergistic adjustment between neighborhood influence and policy constraints, so that the land use simulation results not only conform to the spatial expansion law, but also do not exceed the planning control constraints.

[0128] Furthermore, for each prediction scenario, this application embodiment can also determine the neighborhood influence parameters for converting each grid cell from its current land use type to each candidate land use type based on the current land use type of each grid cell, each candidate land use type, the distribution of land use types within the neighborhood of each grid cell, and the weighted constraint strength corresponding to each grid cell. This includes: for each grid cell, counting the number of neighboring grid cells or the proportion of neighboring area belonging to each candidate land use type within the neighborhood of the grid cell; determining the basic neighborhood influence parameters for converting the grid cell from its current land use type to each candidate land use type based on the number of neighboring grid cells or the proportion of neighboring area; and adjusting the basic neighborhood influence parameters according to the weighted constraint strength corresponding to the grid cell to obtain the neighborhood influence parameters.

[0129] In practical implementation, the basic neighborhood influence parameter can be determined by the area proportion of candidate land use types within the neighborhood, or it can be determined by combining distance attenuation weights. Neighboring grids closer to the target grid can be assigned higher weights. Subsequently, the basic neighborhood influence parameter can be strengthened or weakened based on the weighted constraint strength, allowing the neighborhood effect to be adjusted by spatial constraints.

[0130] S111, for each predicted scenario, calculate the initial conversion probability of each grid cell from the current land use type to each candidate land use type based on the development suitability probability, scenario transition rules and neighborhood influence parameters corresponding to each grid cell.

[0131] In step S111, the initial conversion probability of each grid cell changing from the current land use type to each candidate land use type can be calculated using the neighborhood influence parameters determined in step S110. For example, in the embodiments of this application, the initial conversion probability can be used to represent the probability that a certain grid cell changes from the current land use type to a candidate land use type before the final constraint correction is performed. The initial conversion probability can be obtained by combining the development suitability probability, scenario transition rules, and neighborhood influence parameters.

[0132] The development suitability probability reflects the development suitability of the grid cell itself under its natural environment, socio-economic conditions, and constraints; the scenario transition rule reflects the permissible conversion relationship and scenario preference between land use types under different prediction scenarios; and the neighborhood influence parameter reflects the spatial clustering influence of the surrounding land use distribution on the conversion process.

[0133] The initial conversion probability can be calculated using a product model, a weighted summation model, or other probabilistic fusion models. For example, for land use type combinations that are not allowed to be converted, the initial conversion probability can be set to 0 using scenario transition rules; for land use type combinations that are allowed to be converted and whose scenarios encourage conversion, their initial conversion probability can be increased.

[0134] The initial conversion probability can serve as the basis for subsequent probability adjustments. Since the initial conversion probability incorporates suitability, scenario, and neighborhood factors, it can reflect the basic likelihood of land use conversion more comprehensively.

[0135] Furthermore, for each predicted scenario, the embodiments of this application can calculate the initial conversion probability of each grid cell from the current land use type to each candidate land use type based on the development suitability probability, scenario transition rules, and neighborhood influence parameters corresponding to each grid cell. This includes: determining the allowable conversion identifier and scenario adjustment coefficient for the conversion from the current land use type to the candidate land use type according to the scenario transition rules; and performing a product operation or weighted operation on the development suitability probability, neighborhood influence parameters, allowable conversion identifier, and scenario adjustment coefficient to obtain the initial conversion probability.

[0136] In practice, the allowed conversion flag can be 0 or 1. When the allowed conversion flag is 0, the corresponding land use type conversion is prohibited; when the allowed conversion flag is 1, the corresponding conversion can participate in the probability calculation. The scenario adjustment coefficient can be greater than 1, equal to 1, or less than 1, and is used to enhance, maintain, or inhibit specific land use type conversions, respectively.

[0137] S112, for each prediction scenario, based on the weighted constraint strength and scenario transition rules corresponding to each grid cell, determine the probability correction coefficient of each grid cell corresponding to each candidate land use type, and use the probability correction coefficient to correct the corresponding initial conversion probability to obtain the constraint correction conversion probability of each grid cell from the current land use type to each candidate land use type.

[0138] In step S112, the initial conversion probability calculated in step S111 can be corrected based on the weighted constraint strength determined in step S107 and the scenario transition rule determined in step S108, resulting in the constraint-corrected conversion probability of each grid cell converting from the current land use type to each candidate land use type. In this embodiment, the probability correction coefficient is used to further constrain and correct the initial conversion probability based on the weighted constraint strength and the scenario transition rule. Unlike S109, where the weighted constraint strength is used as the suitability model input, this step can directly correct the candidate land use type at the conversion probability level.

[0139] For a given grid cell and a candidate land use type, the constraint response type corresponding to the transition can be determined based on the scenario transition rules. For example, a transition might be a restricted transition type, a permitted retention type, or a guided transition type. Then, the corresponding probability correction coefficient is determined by combining the weighted constraint strength of the grid cell.

[0140] If the candidate land use type is a restricted conversion type, the higher the weighted constraint strength, the smaller the corresponding probability correction coefficient can be, thereby reducing the conversion probability of that candidate land use type. If the candidate land use type is a permitted retention type, the probability correction coefficient can be close to or equal to 1 to maintain the stability of the original land use type. If the candidate land use type is a guided conversion type, the probability correction coefficient can be greater than or equal to 1 to enhance the conversion probability of the corresponding land use type.

[0141] Through this step, the embodiments of this application can reflect the impact of spatial constraints on different candidate land use types at the level of conversion probability, and realize differentiated probability correction under policy constraints.

[0142] Furthermore, in this embodiment, for each predicted scenario, the probability correction coefficient corresponding to each candidate land use type for each grid cell can be determined based on the weighted constraint strength of each grid cell and the scenario transition rule. This includes: determining the constraint response type for the conversion of the current land use type to the candidate land use type based on the scenario transition rule; when the constraint response type is a restricted transition type, determining a probability correction coefficient less than or equal to 1 based on the weighted constraint strength of the corresponding grid cell; when the constraint response type is a allowed retention type, determining a probability correction coefficient equal to 1 based on the weighted constraint strength of the corresponding grid cell; and when the constraint response type is a guided transition type, determining a probability correction coefficient greater than or equal to 1 based on the weighted constraint strength of the corresponding grid cell.

[0143] In practical implementations, restricting the transition type can use a decreasing function to determine the probability correction coefficient; for example, the higher the weighted constraint strength, the closer the probability correction coefficient is to 0. Guiding the transition type can use an increasing function to determine the probability correction coefficient; for example, the higher the weighted constraint strength, the stronger the guiding effect. Allowing the maintainable type can keep the probability correction coefficient at 1 to avoid unnecessary probability perturbations.

[0144] Furthermore, embodiments of this application can also use probability correction coefficients to correct the corresponding initial conversion probabilities to obtain the constraint-corrected conversion probabilities of each grid cell from the current land use type to each candidate land use type. This includes: multiplying the initial conversion probability corresponding to each candidate land use type by the corresponding probability correction coefficient to obtain the candidate correction probability corresponding to the candidate land use type; and normalizing the candidate correction probabilities corresponding to the same grid cell to obtain the constraint-corrected conversion probabilities of the grid cell from the current land use type to each candidate land use type.

[0145] In practical implementation, normalization ensures that the sum of the constraint correction transformation probabilities for each candidate land use type corresponding to the same raster cell is 1, facilitating subsequent maximum probability selection or random sampling. Normalization also avoids the problem of inconsistent probability scales caused by the amplification or reduction of probability correction coefficients.

[0146] S113. For each prediction scenario, based on the constraint correction conversion probability corresponding to each grid cell, determine the predicted land use type corresponding to each grid cell in the target prediction period from each candidate land use type, and generate the predicted land use layout of the spatial area to be predicted under the prediction scenario based on the location data of each grid cell and the corresponding predicted land use type.

[0147] In step S113, the predicted land use type for each grid cell in the target prediction period can be determined from the candidate land use types based on the constraint-corrected conversion probability determined in step S112. Then, based on the location data of each grid cell and its corresponding predicted land use type, the predicted land use layout of the spatial area to be predicted under this prediction scenario is generated. For example, in this embodiment, the predicted land use type for the target prediction period can be determined based on the constraint-corrected conversion probability corresponding to each grid cell. For each grid cell, the candidate land use type with the highest conversion probability can be selected from multiple candidate land use types, or random sampling can be performed based on the probability distribution to simulate the uncertainty in the land use change process.

[0148] After determining the predicted land use type for each raster cell, a predicted land use layout can be generated based on the location data of each raster cell. The predicted land use layout can be a raster map, vector map, spatial database layer, thematic map, or other forms of spatial distribution results.

[0149] For multiple prediction scenarios, predicted land use layouts under each scenario can be generated and spatial comparative analysis can be performed. For example, indicators such as the expansion range of construction land, the effect of ecological land protection, the area of ​​cultivated land preserved, and the degree of spatial agglomeration can be compared under different scenarios.

[0150] Through this step, the embodiments of this application can output land use layout prediction results that meet the dynamic effects of spatial constraints, providing data support for national spatial planning, ecological protection decision-making, urban expansion management and control and low-carbon development assessment.

[0151] Furthermore, in this embodiment of the application, for each prediction scenario, the predicted land use type corresponding to each grid cell in the target prediction period can be determined from each candidate land use type based on the constraint correction conversion probability corresponding to each grid cell. This includes: selecting the candidate land use type with the highest constraint correction conversion probability from each candidate land use type as the predicted land use type corresponding to the corresponding grid cell in the target prediction period; or, randomly sampling based on the constraint correction conversion probability corresponding to each candidate land use type, and using the sampled candidate land use type as the predicted land use type corresponding to the corresponding grid cell in the target prediction period.

[0152] In practical implementation, the maximum probability selection method yields relatively stable prediction results and is suitable for deterministic layout prediction scenarios. The random sampling method retains a certain degree of randomness and is suitable for simulating uncertain processes in land use change. In the random sampling method, the model can be run multiple times and the prediction results can be statistically analyzed to obtain the prediction stability or uncertainty distribution.

[0153] Furthermore, in the case where the target prediction period includes multiple prediction times, for each prediction scenario, the predicted land use type corresponding to the previous prediction time is used as the current land use type for the next prediction time, and the determination of development suitability probability, determination of neighborhood influence parameters, calculation of initial conversion probability, calculation of constraint correction conversion probability and determination of predicted land use type are repeatedly executed until the predicted land use layout corresponding to the last prediction time in the target prediction period is obtained.

[0154] In practical implementation, the target prediction period can include 2030, 2035, and 2040. The land use layout in 2030 can be predicted starting with the current year's land use type. Then, the 2030 prediction result can be used as the current land use type to predict the layout in 2035, and so on, iterating until the end of the target prediction period. This iterative approach can better reflect the phased evolution of land use change.

[0155] Furthermore, before generating the predicted land use layout of the spatial area to be predicted under the prediction scenario, this application embodiment may include: predicting the land use layout corresponding to the second historical time based on the land use type corresponding to the first historical time in the historical land use data; comparing the predicted land use layout corresponding to the second historical time with the land use type corresponding to the second historical time in the historical land use data to obtain the prediction accuracy evaluation result; and performing land use layout prediction for the target prediction period if the prediction accuracy evaluation result meets the preset accuracy conditions.

[0156] In practical implementation, historical backtesting can be used to evaluate model accuracy. For example, land use data from 2010 can be used as a starting point to predict land use layout in 2020, and the predicted results can be compared with the actual land use data from 2020. Evaluation indicators for prediction accuracy can include overall accuracy, Kappa coefficient, FoM index, user accuracy, and producer accuracy. The model should only be used to predict future periods when the prediction accuracy meets preset conditions, thereby improving the reliability of the results.

[0157] Furthermore, embodiments of this application may, after generating the predicted land use layout of the spatial area to be predicted under the prediction scenario, include: obtaining the corrected carbon density parameter corresponding to each land use type; calculating the predicted carbon storage value of each grid cell under the corresponding prediction scenario based on the predicted land use type and the corrected carbon density parameter corresponding to each grid cell in the predicted land use layout; and generating the carbon storage prediction result of the spatial area to be predicted under the prediction scenario based on the location data of each grid cell and the corresponding predicted carbon storage value.

[0158] In practice, corresponding carbon density parameters can be set for different land use types, and the carbon storage of each grid cell can be calculated in conjunction with the predicted land use layout. The carbon storage prediction results can be used to evaluate the impact of different land use layout schemes on regional carbon storage, ecosystem service functions, or low-carbon development goals.

[0159] In addition, in this embodiment of the application, the corrected carbon density parameters corresponding to each land use type can also be obtained, including: obtaining the basic carbon density parameters corresponding to each land use type, wherein the basic carbon density parameters include at least one of aboveground biomass carbon density, belowground biomass carbon density, soil carbon density and dead organic matter carbon density; and correcting the basic carbon density parameters according to the climate change conditions under the corresponding predicted scenario to obtain the corrected carbon density parameters corresponding to each land use type.

[0160] In practice, different land use types have different carbon densities. Climate change conditions may affect vegetation growth, soil organic carbon accumulation, and ecosystem carbon storage. Therefore, the basic carbon density parameters can be corrected based on temperature, precipitation, or climate scenario parameters to make the carbon storage prediction results more consistent with the corresponding prediction scenarios.

[0161] Furthermore, in this embodiment of the application, the carbon storage prediction value of each grid cell under the corresponding prediction scenario can be calculated based on the predicted land use type and the corrected carbon density parameter corresponding to each grid cell in the predicted land use layout. This includes: determining the corrected carbon density parameter corresponding to the grid cell based on the predicted land use type corresponding to the grid cell; and calculating the carbon storage prediction value corresponding to the grid cell based on the corrected carbon density parameter corresponding to the grid cell and the area data corresponding to the grid cell.

[0162] In practical implementation, if the grid cell area is fixed, the carbon storage of the grid cell can be obtained by multiplying the corrected carbon density parameter by the grid area. For grids or vectorized cells with inconsistent areas, the corresponding area data can be used for calculation, and the total carbon storage of the region can be obtained by summing over all grid cells.

[0163] Furthermore, after generating the carbon storage prediction results for the spatial area to be predicted under the prediction scenario, this application embodiment may also include: using the carbon storage prediction value corresponding to each grid cell as the target quantity, and using the land use structure data, socio-economic data, and weighted constraint strength corresponding to each grid cell as input quantities to determine the carbon storage response relationship; determining the nonlinear response relationship between the weighted constraint strength and the carbon storage prediction value based on the carbon storage response relationship; and dividing the weighted constraint strength into multiple response intervals based on the nonlinear response relationship.

[0164] In practical implementation, nonlinear regression models, random forest models, gradient boosting tree models, neural network models, or other machine learning models can be used to establish the carbon storage response relationship. Through feature contribution analysis, partial dependency analysis, or sensitivity analysis, the nonlinear response curve between the weighted constraint strength and the predicted carbon storage value can be identified, thereby discovering the influence of changes in constraint strength on changes in carbon storage.

[0165] Furthermore, in this embodiment of the application, the weighted constraint strength can be divided into multiple response intervals based on the nonlinear response relationship, including: determining at least one interval dividing point based on the rate of change, curvature, or marginal contribution change of the carbon storage prediction value with the weighted constraint strength in the nonlinear response relationship; and dividing the weighted constraint strength into at least two of the following intervals based on at least one interval dividing point: low response interval, sensitive response interval, high-efficiency response interval, and response saturation interval.

[0166] In practice, the inflection point can be determined by calculating the first-order rate of change, second-order curvature, or marginal contribution change of the response curve. After dividing the weighted constraint intensity into multiple response intervals based on the inflection point, it is possible to identify which constraint intensity ranges show a significant increase in carbon storage and which ranges offer lower marginal benefits from further increasing the constraint intensity.

[0167] Furthermore, the embodiments of this application may further include, after dividing the weighted constraint intensity into multiple response intervals, the following steps: determining the feedback control coefficient corresponding to each grid cell based on the response interval to which the weighted constraint intensity of each grid cell belongs; further correcting the probability correction coefficient or constraint correction conversion probability corresponding to each grid cell based on the feedback control coefficient; and re-determining the predicted land use type corresponding to each grid cell in the target prediction period based on the further corrected constraint correction conversion probability, so as to obtain the optimized predicted land use layout.

[0168] In practical implementation, the feedback control coefficient can be used to enhance the ecological protection or low-carbon optimization of grid cells within the high-efficiency response range, and also to avoid excessively increasing the constraint intensity within the response saturation range. Through feedback control, synergistic optimization between land use layout prediction and carbon storage response results can be achieved.

[0169] Furthermore, after obtaining the optimized predicted land use layout, the embodiments of this application may also include: recalculating the carbon storage prediction value based on the optimized predicted land use layout to obtain the optimized carbon storage prediction result; and outputting at least one of the following: the optimized predicted land use layout, the optimized carbon storage prediction result, the weighted constraint strength corresponding to each grid cell, the feedback control coefficient corresponding to each grid cell, and the comparison results between different prediction scenarios.

[0170] In practice, the output results can be presented in the form of maps, tables, raster files, vector files, or reports. The comparison results between different prediction scenarios can include indicators such as changes in land use area, intensity of construction land expansion, effectiveness of arable land protection, ecological land retention rate, and changes in carbon storage.

[0171] Third Embodiment This embodiment combines Figure 2 This application describes a land use layout prediction device provided in an embodiment. This device can be used to execute the land use layout prediction method in the second embodiment. The device can be deployed in a server, computer equipment, geographic information processing platform, planning support decision system, or other electronic equipment with data processing capabilities.

[0172] like Figure 2 As shown, the land use layout prediction device provided in this application embodiment includes: a data acquisition module 201, a historical conversion frequency determination module 202, a first constraint strength determination module 203, a dispersion weight determination module 204, an initial weight determination module 205, a fusion weight determination module 206, a weighted constraint strength determination module 207, a scenario transition rule determination module 208, a development suitability probability determination module 209, a neighborhood influence parameter determination module 210, an initial conversion probability calculation module 211, a constraint correction conversion probability calculation module 212, and a prediction layout generation module 213.

[0173] The data acquisition module 201 can be used to acquire rasterized basic data, land use spatial constraint data, scenario-driven data, and historical land use data corresponding to the spatial area to be predicted. Among them, the rasterized basic data includes the location data, natural environment data, and socio-economic data of each raster unit in the spatial area to be predicted. The land use spatial constraint data represents the constraint information of land use type transformation in the spatial area to be predicted and includes at least one spatial constraint term. The spatial constraint term is a spatial data item in the land use spatial constraint data used to represent a land use type transformation restriction. The scenario-driven data is used to represent the socio-economic development conditions and climate change conditions that affect land use type transformation under different prediction scenarios. The historical land use data includes the land use types corresponding to each raster unit at multiple historical moments.

[0174] The data acquisition module 201 can acquire data from external databases, remote sensing image platforms, geographic information systems, planning and control databases, climate scenario databases, user terminals, or local storage media. The data acquisition module 201 can also perform format conversion, coordinate unification, rasterization, cropping, and spatial registration on data from different sources.

[0175] The historical land use data acquired by the data acquisition module 201 can be sent to the historical conversion frequency determination module 202; the acquired land use spatial constraint data can be sent to the first constraint intensity determination module 203; the acquired scenario-driven data can be sent to the scenario transfer rule determination module 208 and the development suitability probability determination module 209; the acquired rasterized basic data and historical land use data can be sent to the development suitability probability determination module 209; and the acquired location data can be sent to the prediction layout generation module 213.

[0176] Furthermore, in this embodiment of the application, the location data acquired by the data acquisition module 201 includes at least one of grid number, grid center point coordinates, and grid boundary coordinates; the natural environment data includes at least one of elevation, slope, aspect, average annual temperature, annual precipitation, soil type, distance from water system, and distance from road; and the socio-economic data includes at least one of population density, regional GDP, nighttime light intensity, distance from settlement, distance from town center, and traffic accessibility data.

[0177] Furthermore, in this embodiment of the application, the land use types acquired or determined by the data acquisition module 201 include at least two of the following: cultivated land, forest land, grassland, water area, construction land, and unused land; the current land use type is the land use type of the corresponding grid cell at the start of the prediction, and the candidate land use type is the land use type that the corresponding grid cell is allowed to be determined in the target prediction period.

[0178] Furthermore, in this embodiment of the application, the spatial constraints acquired by the data acquisition module 201 include at least one of the following: main functional area constraints, ecological protection red line constraints, urban development boundary constraints, and permanent basic farmland constraints.

[0179] Furthermore, in this embodiment of the application, the prediction scenarios acquired by the data acquisition module 201 include at least two of the following: natural development scenarios, economic development scenarios, ecological protection scenarios, low-carbon optimization scenarios, and baseline scenarios; the scenario-driven data include at least one of the following: population change data, economic development data, transportation construction data, climate change data, and land use conversion preference data.

[0180] The historical conversion frequency determination module 202 can be used to determine the historical land use type conversion relationship between adjacent historical times for each grid cell based on historical land use data, and to determine the historical conversion frequency between each land use type based on the historical land use type conversion relationship.

[0181] The historical conversion frequency determination module 202 can perform overlay analysis on land use raster data from multiple historical moments and compare land use types between adjacent historical moments grid by grid. For raster cells where the land use type has changed, it can record the type before and after the conversion; for raster cells where the land use type has not changed, it can record the retention relationship.

[0182] The historical conversion frequency determination module 202 can generate a conversion matrix based on the conversion relationships of all raster cells. Elements in the conversion matrix can represent the number of conversions, conversion area, or conversion ratio between corresponding land use type combinations. The historical conversion frequency determination module 202 can send the historical conversion frequencies to the initial weight determination module 205 and the scenario transfer rule determination module 208.

[0183] The first constraint strength determination module 203 can be used to determine the first constraint strength of each spatial constraint item in the land use spatial constraint data based on the spatial range of action, constraint attribute and spatial relationship between the spatial constraint item and each grid cell. The first constraint strength is used to characterize the restriction strength of the spatial constraint item on the land use type conversion of the corresponding grid cell.

[0184] The first constraint strength determination module 203 can determine the first constraint strength based on spatial overlay, distance analysis, partition matching, proximity analysis, or historical state preservation analysis. For area constraint ranges, it can determine whether the grid cell is within the constraint range; for boundary constraints, it can calculate the distance between the grid cell and the boundary; for functional area constraints, it can determine the constraint strength level based on the functional area to which the grid cell belongs.

[0185] The first constraint strength determination module 203 can send the first constraint strength corresponding to each spatial constraint item to the discrete weight determination module 204 and the weighted constraint strength determination module 207. The discrete weight determination module 204 is used to calculate the discrete weight based on the spatial distribution of the first constraint strength, and the weighted constraint strength determination module 207 is used to combine the first constraint strength with the fusion weight to obtain the weighted constraint strength.

[0186] Furthermore, in this embodiment, when the spatial constraint item is a main functional area constraint item, the first constraint intensity determination module 203 determines the first constraint intensity corresponding to the grid unit according to the main functional area type to which the grid unit belongs; when the spatial constraint item is an ecological protection red line constraint item, it determines the first constraint intensity corresponding to the grid unit according to whether the grid unit is located within the ecological protection red line; when the spatial constraint item is an urban development boundary constraint item, it determines the first constraint intensity corresponding to the grid unit according to the spatial distance between the grid unit and the urban development boundary; when the spatial constraint item is a permanent basic farmland constraint item, it determines the first constraint intensity corresponding to the grid unit according to whether the grid unit is located within the permanent basic farmland, or according to whether the grid unit continuously corresponds to the cultivated land type at multiple historical moments.

[0187] Furthermore, in this embodiment, the constraint attributes used by the first constraint strength determination module 203 include at least one of the following: prohibition of conversion attribute, restriction of conversion attribute, priority maintenance attribute, development guidance attribute, and distance attenuation attribute; wherein, the prohibition of conversion attribute is used to characterize the corresponding spatial constraint item restricting the raster cell from entering or leaving the specified land use type, and the distance attenuation attribute is used to characterize the restriction strength of the corresponding spatial constraint item on the raster cell as the spatial distance changes.

[0188] The discrete weight determination module 204 can be used to calculate the discrete statistical value corresponding to each spatial constraint term based on the numerical distribution of the first constraint strength of the spatial constraint term on all grid cells, and determine the discrete weight corresponding to each spatial constraint term based on the discrete statistical value corresponding to each spatial constraint term. The discrete weight is used to characterize the difference in the first constraint strength of the corresponding spatial constraint term on different grid cells.

[0189] The dispersion weight determination module 204 can perform statistical analysis on the first constraint strength output by the first constraint strength determination module 203. For each spatial constraint term, the dispersion weight determination module 204 can form a constraint strength sequence based on the first constraint strength of the spatial constraint term across all grid cells, and calculate the dispersion of the sequence.

[0190] The dispersion weights determined by the dispersion weight determination module 204 can be sent to the fusion weight determination module 206. The larger the dispersion weight, the more obvious the spatial difference expression ability of the corresponding spatial constraint term in the spatial region to be predicted, and the greater its contribution to the comprehensive constraint calculation.

[0191] Furthermore, in this embodiment of the application, the dispersion statistics calculated by the dispersion weight determination module 204 include at least one of variance, standard deviation, coefficient of variation, range, interquartile range, mean absolute deviation, and difference coefficient determined based on information distribution.

[0192] Furthermore, in this embodiment, the dispersion weight determination module 204 performs same-scale processing on the dispersion statistics corresponding to each spatial constraint term to obtain the same-scale dispersion corresponding to each spatial constraint term; and performs a ratio calculation on the same-scale dispersion corresponding to each spatial constraint term and the sum of the same-scale dispersion corresponding to all spatial constraint terms to obtain the dispersion weight corresponding to that spatial constraint term.

[0193] The initial weight determination module 205 can be used to determine the initial weight of each spatial constraint item based on the historical conversion frequency between each land use type. The initial weight is used to characterize the preset configuration proportion of the spatial constraint item in the land use type conversion constraint.

[0194] The initial weight determination module 205 can receive the historical conversion frequency output by the historical conversion frequency determination module 202, and determine the initial weight based on the land use type conversion combinations affected by different spatial constraints. For land use type conversion combinations with high historical conversion frequency and significant influence from a certain spatial constraint, a higher initial weight can be set for the corresponding spatial constraint.

[0195] The initial weight determination module 205 can send the initial weights corresponding to each spatial constraint to the fusion weight determination module 206. Through this module, the determination of spatial constraint weights can be combined with historical land use change patterns, reducing the subjectivity caused by relying entirely on manual experience for weight assignment.

[0196] Furthermore, in this embodiment of the application, the initial weight determination module 205 extracts the target historical conversion frequency corresponding to the spatial constraint item from the historical conversion frequency between each land use type according to the land use type conversion combination restricted by each spatial constraint item; and determines the initial weight corresponding to the spatial constraint item according to the target historical conversion frequency, so that the initial weight corresponding to the spatial constraint item with the higher the target historical conversion frequency is greater.

[0197] The fusion weight determination module 206 can be used to perform a weighted summation of the initial weight and dispersion weight corresponding to each spatial constraint term according to a preset fusion ratio to obtain the fusion weight corresponding to the spatial constraint term.

[0198] The fusion weight determination module 206 can receive the discrete weights output by the discrete weight determination module 204 and the initial weights output by the initial weight determination module 205. The fusion weight determination module 206 can fuse the initial weights and discrete weights of each spatial constraint term according to a preset fusion ratio.

[0199] The fusion weights reflect both the relationship between spatial constraints and the frequency of historical land use conversions, as well as the ability of spatial constraints to express differences across different raster cells. The fusion weight determination module 206 can send the fusion weights to the weighted constraint strength determination module 207.

[0200] Furthermore, in this embodiment of the application, the fusion weight determination module 206 performs a weighted summation of the initial weight and dispersion weight corresponding to the spatial constraint term based on the preset first fusion coefficient and the preset second fusion coefficient to obtain the candidate fusion weight corresponding to the spatial constraint term; and performs normalization processing on the candidate fusion weights corresponding to each spatial constraint term to obtain the fusion weight corresponding to each spatial constraint term.

[0201] The weighted constraint strength determination module 207 can be used to multiply the first constraint strength of each spatial constraint term on the grid cell by the corresponding fusion weight and sum them up to obtain the weighted constraint strength corresponding to the grid cell.

[0202] The weighted constraint strength determination module 207 can receive the first constraint strength output by the first constraint strength determination module 203 and the fusion weight output by the fusion weight determination module 206. For each grid cell, the weighted constraint strength determination module 207 can perform a weighted summation of the first constraint strengths corresponding to multiple spatial constraint terms and the corresponding fusion weights to obtain the comprehensive constraint result.

[0203] The weighted constraint strength determination module 207 can send the weighted constraint strength to the development suitability probability determination module 209, the neighborhood influence parameter determination module 210, and the constraint correction conversion probability calculation module 212. Thus, the weighted constraint strength can participate in multiple key processing stages of land use layout prediction.

[0204] Furthermore, in this embodiment, the weighted constraint strength determination module 207 performs a smooth correction on the weighted constraint strength corresponding to each grid cell based on the weighted constraint strength distribution of each grid cell within its neighborhood, thereby obtaining the corrected weighted constraint strength; wherein, the corrected weighted constraint strength is used to determine at least one of the development suitability probability, neighborhood influence parameter, and probability correction coefficient.

[0205] The scenario transition rule determination module 208 can be used to determine the scenario transition rules between different land use types for each predicted scenario, based on the scenario-driven data corresponding to the predicted scenario and the historical transition frequency between each land use type.

[0206] The scenario transition rule determination module 208 can receive scenario-driven data output by the data acquisition module 201 and historical transition frequencies output by the historical transition frequency determination module 202. The scenario transition rule determination module 208 can establish basic transition relationships based on historical transition frequencies and adjust the basic transition relationships according to different prediction scenarios.

[0207] The scenario transition rule determination module 208 can send the scenario transition rules to the initial transition probability calculation module 211 and the constraint correction transition probability calculation module 212. The initial transition probability calculation module 211 can calculate the initial transition probability according to the scenario transition rules, and the constraint correction transition probability calculation module 212 can determine the probability correction coefficient according to the scenario transition rules.

[0208] Furthermore, in this embodiment of the application, the scenario transfer rule determination module 208 determines the basic transfer permission relationship between different land use types based on the historical conversion frequency between each land use type; and enhances, suppresses or prohibits the basic transfer permission relationship based on the scenario-driven data corresponding to the predicted scenario to obtain the scenario transfer rule.

[0209] The development suitability probability determination module 209 can be used to determine the development suitability probability of each grid cell from the current land use type to each candidate land use type for each prediction scenario, based on the scenario-driven data, historical land use data, rasterized basic data and the weighted constraint strength corresponding to each grid cell.

[0210] The development suitability probability determination module 209 can receive scenario-driven data, historical land use data, and rasterized basic data output by the data acquisition module 201, and receive the weighted constraint strength output by the weighted constraint strength determination module 207. The development suitability probability determination module 209 can use statistical models or machine learning models to calculate the development suitability probability corresponding to each candidate land use type.

[0211] The development suitability probability determination module 209 can send the development suitability probability to the initial transformation probability calculation module 211. By using the weighted constraint strength as an input factor, the development suitability probability determination module 209 can incorporate the influence of spatial constraints in the suitability evaluation stage.

[0212] Furthermore, in this embodiment, the development suitability probability determination module 209 determines historical conversion samples based on historical land use data. The historical conversion samples include raster cell samples that have undergone land use type conversion and raster cell samples that have not undergone land use type conversion. Based on the historical conversion samples, rasterized basic data, scenario-driven data, and the weighted constraint strength corresponding to each raster cell, the development suitability score corresponding to each candidate land use type is determined. The development suitability score corresponding to each candidate land use type is probabilistically processed to obtain the development suitability probability of each raster cell converting from the current land use type to each candidate land use type.

[0213] The neighborhood influence parameter determination module 210 can be used to determine the neighborhood influence parameters of each grid cell from the current land use type to each candidate land use type for each prediction scenario, based on the current land use type of each grid cell, each candidate land use type, the distribution of land use types within the neighborhood of each grid cell, and the weighted constraint strength corresponding to each grid cell.

[0214] The neighborhood influence parameter determination module 210 can receive the current land use type and related land use distribution data output by the data acquisition module 201, and receive the weighted constraint strength output by the weighted constraint strength determination module 207. The neighborhood influence parameter determination module 210 can statistically determine the quantity or area proportion of different land use types within the neighborhood based on a preset neighborhood window.

[0215] The neighborhood influence parameter determination module 210 can send the neighborhood influence parameters to the initial transformation probability calculation module 211. Since the determination process of the neighborhood influence parameters introduces weighted constraint strength, the neighborhood expansion effect can be adjusted by spatial constraint conditions.

[0216] Furthermore, in this embodiment, the neighborhood influence parameter determination module 210, for each grid cell, counts the number of neighboring grid cells or the proportion of neighboring area belonging to each candidate land use type within the neighborhood of the grid cell; based on the number of neighboring grid cells or the proportion of neighboring area, it determines the basic neighborhood influence parameter for converting the grid cell from the current land use type to each candidate land use type; and based on the weighted constraint strength corresponding to the grid cell, it adjusts the basic neighborhood influence parameter to obtain the neighborhood influence parameter.

[0217] The initial conversion probability calculation module 211 can be used to calculate the initial conversion probability of each grid cell from the current land use type to each candidate land use type for each prediction scenario, based on the development suitability probability, scenario transition rules and neighborhood influence parameters corresponding to each grid cell.

[0218] The initial transition probability calculation module 211 can receive the development suitability probability output by the development suitability probability determination module 209, the scenario transition rule output by the scenario transition rule determination module 208, and the neighborhood influence parameter output by the neighborhood influence parameter determination module 210. The initial transition probability calculation module 211 can generate the initial transition probability using product operation, weighted operation, or other probability fusion methods.

[0219] The initial transformation probability calculation module 211 can send the initial transformation probability to the constraint correction transformation probability calculation module 212. The initial transformation probability is used as the base probability for subsequent constraint correction.

[0220] In addition, in this embodiment of the application, the initial conversion probability calculation module 211 determines the allowable conversion identifier and scenario adjustment coefficient for the conversion of the current land use type to the candidate land use type according to the scenario transfer rules; and performs a product operation or weighted operation on the development suitability probability, the neighborhood influence parameter, the allowable conversion identifier and the scenario adjustment coefficient to obtain the initial conversion probability.

[0221] The constraint correction conversion probability calculation module 212 can be used to determine the probability correction coefficient of each grid cell corresponding to each candidate land use type for each prediction scenario based on the weighted constraint strength and scenario transition rules of each grid cell, and use the probability correction coefficient to correct the corresponding initial conversion probability to obtain the constraint correction conversion probability of each grid cell from the current land use type to each candidate land use type.

[0222] The constraint correction transition probability calculation module 212 can receive the weighted constraint strength output by the weighted constraint strength determination module 207, the scenario transition rule output by the scenario transition rule determination module 208, and the initial transition probability output by the initial transition probability calculation module 211. The constraint correction transition probability calculation module 212 can determine the probability correction coefficient according to the constraint response type corresponding to different candidate land use types.

[0223] The constraint correction transformation probability calculation module 212 can send the constraint correction transformation probability to the prediction layout generation module 213. This module can realize dynamic and differentiated correction of policy constraints at the probability level.

[0224] Furthermore, in this embodiment, the constraint correction conversion probability calculation module 212 determines the constraint response type for the conversion of the current land use type to the candidate land use type according to the scenario transfer rules; when the constraint response type is a restricted conversion type, a probability correction coefficient less than or equal to 1 is determined based on the weighted constraint strength of the corresponding grid cell; when the constraint response type is an allowed retention type, a probability correction coefficient equal to 1 is determined based on the weighted constraint strength of the corresponding grid cell; when the constraint response type is a guided conversion type, a probability correction coefficient greater than or equal to 1 is determined based on the weighted constraint strength of the corresponding grid cell.

[0225] In addition, in this embodiment of the application, the constraint correction conversion probability calculation module 212 multiplies the initial conversion probability corresponding to each candidate land use type with the corresponding probability correction coefficient to obtain the candidate correction probability corresponding to the candidate land use type; and normalizes each candidate correction probability corresponding to the same grid cell to obtain the constraint correction conversion probability of the grid cell from the current land use type to each candidate land use type.

[0226] The prediction layout generation module 213 can be used to determine the predicted land use type of each grid cell in the target prediction period from each candidate land use type for each prediction scenario, based on the constraint correction conversion probability corresponding to each grid cell, and generate the predicted land use layout of the spatial area to be predicted under the prediction scenario based on the location data of each grid cell and the corresponding predicted land use type.

[0227] The predictive layout generation module 213 can receive the constraint correction transformation probability output by the constraint correction transformation probability calculation module 212, and the location data output by the data acquisition module 201. For each grid cell, the predictive layout generation module 213 can determine the predicted land use type based on the constraint correction transformation probability, and then generate the spatial layout result according to the location data of the grid cell.

[0228] The predictive layout generation module 213 can output predicted land use layouts in raster, vector, map layer, or data table format. For multiple prediction scenarios, the predictive layout generation module 213 can generate predicted land use layouts for the corresponding prediction scenarios.

[0229] In addition, in this embodiment of the application, the prediction layout generation module 213 selects the candidate land use type with the highest constraint correction conversion probability from each candidate land use type as the predicted land use type corresponding to the corresponding grid cell in the target prediction period; or, it randomly samples according to the constraint correction conversion probability corresponding to each candidate land use type, and uses the sampled candidate land use type as the predicted land use type corresponding to the corresponding grid cell in the target prediction period.

[0230] Furthermore, in embodiments of this application, when the target prediction period includes multiple prediction times, the prediction layout generation module 213 can, for each prediction scenario, use the predicted land use type corresponding to the previous prediction time as the current land use type for the next prediction time, and make the relevant modules repeatedly execute the determination of development suitability probability, determination of neighborhood influence parameters, calculation of initial conversion probability, calculation of constraint correction conversion probability, and determination of predicted land use type until the predicted land use layout corresponding to the last prediction time in the target prediction period is obtained.

[0231] Furthermore, in this embodiment, based on the land use type corresponding to the first historical moment in the historical land use data, the land use layout corresponding to the second historical moment is predicted; the predicted land use layout corresponding to the second historical moment is compared with the land use type corresponding to the second historical moment in the historical land use data to obtain the prediction accuracy evaluation result; if the prediction accuracy evaluation result meets the preset accuracy conditions, the land use layout prediction for the target prediction period is performed.

[0232] Furthermore, in this embodiment of the application, after generating the predicted land use layout of the spatial area to be predicted under the prediction scenario, the corrected carbon density parameter corresponding to each land use type is obtained; based on the predicted land use type and corrected carbon density parameter corresponding to each grid cell in the predicted land use layout, the carbon storage prediction value of each grid cell under the corresponding prediction scenario is calculated; based on the location data of each grid cell and the corresponding carbon storage prediction value, the carbon storage prediction result of the spatial area to be predicted under the prediction scenario is generated.

[0233] In addition, in the embodiments of this application, the basic carbon density parameters corresponding to each land use type can be obtained. The basic carbon density parameters include at least one of aboveground biomass carbon density, belowground biomass carbon density, soil carbon density, and dead organic matter carbon density. According to the climate change conditions under the corresponding prediction scenario, the basic carbon density parameters are corrected to obtain the corrected carbon density parameters corresponding to each land use type.

[0234] Furthermore, in this embodiment, the corrected carbon density parameter corresponding to the grid cell can be determined based on the predicted land use type corresponding to the grid cell; and the predicted carbon storage value corresponding to the grid cell can be calculated based on the corrected carbon density parameter corresponding to the grid cell and the area data corresponding to the grid cell.

[0235] Furthermore, in this embodiment, after generating the carbon storage prediction results for the spatial area to be predicted under the prediction scenario, the carbon storage prediction value corresponding to each grid cell is used as the target quantity, and the land use structure data, socio-economic data, and weighted constraint strength corresponding to each grid cell are used as input quantities to determine the carbon storage response relationship; based on the carbon storage response relationship, the nonlinear response relationship between the weighted constraint strength and the carbon storage prediction value is determined; based on the nonlinear response relationship, the weighted constraint strength is divided into multiple response intervals.

[0236] Furthermore, in this embodiment, at least one interval dividing point can be determined based on the rate of change, curvature, or marginal contribution of the carbon storage prediction value with the weighted constraint strength in the nonlinear response relationship; based on at least one interval dividing point, the weighted constraint strength can be divided into at least two of the following: low response interval, sensitive response interval, high-efficiency response interval, and response saturation interval.

[0237] Furthermore, in this embodiment, the feedback control coefficient corresponding to each grid cell can be determined based on the response interval to which the weighted constraint strength of each grid cell belongs; the probability correction coefficient or constraint correction conversion probability corresponding to each grid cell can be corrected again based on the feedback control coefficient; and the predicted land use type corresponding to each grid cell in the target prediction period can be re-determined based on the corrected constraint correction conversion probability to obtain the optimized predicted land use layout.

[0238] In addition, in this embodiment of the application, the carbon storage prediction value can be recalculated based on the optimized predicted land use layout to obtain the optimized carbon storage prediction result; at least one of the following can be output: the optimized predicted land use layout, the optimized carbon storage prediction result, the weighted constraint strength corresponding to each grid cell, the feedback control coefficient corresponding to each grid cell, and the comparison results between different prediction scenarios.

[0239] It should be noted that the land use layout prediction device in this embodiment and the land use layout prediction method in the second embodiment belong to the same inventive concept. For details not described in this embodiment, please refer to the corresponding descriptions in the second embodiment; they will not be repeated here.

[0240] Fourth embodiment The above describes the internal functions and structure of a land use layout prediction device, which can be implemented as an electronic device. Figure 3 A schematic diagram illustrating the structure of an embodiment of the electronic device provided in this application. (See attached diagram.) Figure 3 As shown, the electronic device includes a memory 31 and a processor 32.

[0241] Memory 31 is used to store programs. In addition to the programs described above, memory 31 can also be configured to store various other data to support operation on the electronic device. Examples of this data include instructions for any application or method used to operate on the electronic device, contact data, phonebook data, messages, pictures, videos, etc.

[0242] The memory 31 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0243] Processor 32 is not limited to a central processing unit (CPU), but may also be a graphics processing unit (GPU), a field-programmable gate array (FPGA), an embedded neural network processor (NPU), or an artificial intelligence (AI) chip. Processor 32 is coupled to memory 31 and executes the program stored in memory 31. When the program runs, it executes the land use layout prediction method of Embodiment 2 described above.

[0244] Furthermore, such as Figure 3 As shown, the electronic device may also include other components such as a communication component 33, a power supply component 34, an audio component 35, and a display 36. Figure 3 The diagram only shows some components and does not mean that the electronic device includes only these components. Figure 3 The components shown.

[0245] Communication component 33 is configured to facilitate wired or wireless communication between electronic devices and other devices. The electronic devices can access wireless networks based on communication standards, such as WiFi, 3G, 4G, or 5G, or combinations thereof. In one exemplary embodiment, communication component 33 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 33 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0246] Power supply component 34 provides power to various components of the electronic device. Power supply component 34 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the electronic device.

[0247] Audio component 35 is configured to output and / or input audio signals. For example, audio component 35 includes a microphone (MIC) configured to receive external audio signals when the electronic device is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 31 or transmitted via communication component 33. In some embodiments, audio component 35 also includes a speaker for outputting audio signals.

[0248] Display 36 includes a screen, which may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen can be implemented as a touchscreen to receive input signals from a user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors can detect not only the boundaries of touch or swipe actions, but also the duration and pressure associated with the touch or swipe operation.

[0249] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0250] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A land use layout prediction method, characterized in that, include: The method involves acquiring rasterized basic data, land use spatial constraint data, scenario-driven data, and historical land use data corresponding to the spatial area to be predicted. The rasterized basic data includes the location data, natural environment data, and socio-economic data of each raster unit within the spatial area to be predicted. The land use spatial constraint data represents the constraint information on land use type conversion within the spatial area to be predicted and includes at least one spatial constraint term. The spatial constraint term is a spatial data item in the land use spatial constraint data used to represent a limitation on land use type conversion. The scenario-driven data is used to represent the socio-economic development conditions and climate change conditions that affect land use type conversion under different prediction scenarios. The historical land use data includes the land use types corresponding to each raster unit at multiple historical moments. Based on the historical land use data, determine the historical land use type conversion relationship between adjacent historical moments for each grid cell, and determine the historical conversion frequency between each land use type based on the historical land use type conversion relationship; For each spatial constraint item in the land use spatial constraint data, the first constraint intensity of the spatial constraint item on each grid cell is determined according to the spatial scope of the spatial constraint item, the constraint attribute and the spatial relationship between the spatial constraint item and each grid cell. The first constraint intensity is used to characterize the restriction intensity of the spatial constraint item on the land use type conversion of the corresponding grid cell. For each spatial constraint term, the discreteness statistics corresponding to the spatial constraint term are calculated based on the numerical distribution of the first constraint strength of the spatial constraint term on all grid cells, and the discreteness weight corresponding to each spatial constraint term is determined based on the discreteness statistics corresponding to each spatial constraint term. The discreteness weight is used to characterize the difference in the first constraint strength of the corresponding spatial constraint term on different grid cells. Based on the historical conversion frequency between each land use type, the initial weight corresponding to each spatial constraint is determined, wherein the initial weight is used to characterize the preset configuration proportion of the spatial constraint in the land use type conversion constraint; For each spatial constraint, the initial weight and the dispersion weight corresponding to the spatial constraint are weighted and summed according to a preset fusion ratio to obtain the fusion weight corresponding to the spatial constraint. For each grid cell, the first constraint strength of each spatial constraint term on that grid cell is multiplied by the corresponding fusion weight and summed to obtain the weighted constraint strength corresponding to that grid cell. For each predicted scenario, scenario transfer rules between different land use types are determined based on the scenario-driven data corresponding to the predicted scenario and the historical transition frequency between each land use type. For each predicted scenario, based on the scenario-driven data corresponding to the predicted scenario, the historical land use data, the rasterized basic data, and the weighted constraint strength corresponding to each raster cell, the development suitability probability of each raster cell being converted from the current land use type to each candidate land use type is determined. For each prediction scenario, based on the current land use type of each grid cell, each candidate land use type, the distribution of land use types within the neighborhood of each grid cell, and the weighted constraint strength corresponding to each grid cell, the neighborhood influence parameter for converting each grid cell from its current land use type to each candidate land use type is determined. For each predicted scenario, the initial conversion probability of each grid cell from the current land use type to each candidate land use type is calculated based on the development suitability probability corresponding to each grid cell, the scenario transition rule, and the neighborhood influence parameter. For each predicted scenario, based on the weighted constraint strength corresponding to each grid cell and the scenario transition rule, the probability correction coefficient of each grid cell corresponding to each candidate land use type is determined, and the initial conversion probability is corrected using the probability correction coefficient to obtain the constraint correction conversion probability of each grid cell from the current land use type to each candidate land use type. For each prediction scenario, based on the constraint correction conversion probability corresponding to each grid cell, the predicted land use type corresponding to each grid cell in the target prediction period is determined from each candidate land use type. Based on the location data of each grid cell and the corresponding predicted land use type, the predicted land use layout of the spatial area to be predicted under the prediction scenario is generated.

2. The land use layout prediction method according to claim 1, characterized in that, The dispersion weights of each spatial constraint term are determined based on the dispersion statistics of each spatial constraint term, including: The dispersion statistics corresponding to each spatial constraint term are scaled to obtain the same-scale dispersion of each spatial constraint term. The dispersion weight corresponding to each spatial constraint term is obtained by taking the ratio of the same-scale dispersion of each spatial constraint term to the sum of the same-scale dispersions of all spatial constraint terms.

3. The land use layout prediction method according to claim 1, characterized in that, For each predicted scenario, based on the scenario-driven data corresponding to that predicted scenario, the historical land use data, the rasterized basic data, and the weighted constraint strength corresponding to each raster cell, the development suitability probability of each raster cell being converted from the current land use type to each candidate land use type is determined, including: Historical conversion samples are determined based on the historical land use data. The historical conversion samples include raster cell samples that have undergone land use type conversion and raster cell samples that have not undergone land use type conversion. Based on the historical conversion samples, the rasterized basic data, the scenario-driven data, and the weighted constraint strength corresponding to each raster unit, the development suitability score corresponding to each candidate land use type is determined. The development suitability scores corresponding to each candidate land use type are probabilistically processed to obtain the development suitability probability of each grid cell from the current land use type to each candidate land use type.

4. The land use layout prediction method according to claim 1, characterized in that, For each prediction scenario, based on the current land use type of each grid cell, each candidate land use type, the distribution of land use types within the neighborhood of each grid cell, and the weighted constraint strength corresponding to each grid cell, the neighborhood influence parameters for converting each grid cell from its current land use type to each candidate land use type are determined, including: For each grid cell, count the number of neighboring grid cells belonging to each candidate land use type or the percentage of the neighboring area within the grid cell's neighborhood. Based on the number of neighboring grids or the proportion of neighboring area, determine the basic neighborhood influence parameters for converting the current land use type into each candidate land use type for the grid cell; The basic neighborhood influence parameter is adjusted based on the weighted constraint strength corresponding to the grid cell to obtain the neighborhood influence parameter.

5. The land use layout prediction method according to claim 1, characterized in that, After generating the predicted land use layout of the spatial area to be predicted under this prediction scenario, the method further includes: Obtain the corrected carbon density parameters corresponding to each land use type; Based on the predicted land use type corresponding to each grid cell in the predicted land use layout and the corrected carbon density parameter, calculate the predicted carbon storage value of each grid cell under the corresponding prediction scenario. Based on the location data of each grid cell and the corresponding carbon storage prediction value, the carbon storage prediction result of the spatial region to be predicted under this prediction scenario is generated.

6. The land use layout prediction method according to claim 5, characterized in that, After generating the carbon storage prediction results for the spatial region to be predicted under this prediction scenario, the method further includes: The carbon storage prediction value corresponding to each grid cell is used as the target value, and the land use structure data, socio-economic data and weighted constraint intensity corresponding to each grid cell are used as input values ​​to determine the carbon storage response relationship. Based on the carbon storage response relationship, determine the nonlinear response relationship between the weighted constraint strength and the predicted carbon storage value; Based on the nonlinear response relationship, the weighted constraint strength is divided into multiple response intervals.

7. The land use layout prediction method according to claim 6, characterized in that, After dividing the weighted constraint strength into multiple response intervals, the method further includes: Based on the response range to which the weighted constraint strength of each grid cell belongs, determine the feedback control coefficient corresponding to each grid cell; The probability correction coefficient or constraint correction conversion probability corresponding to each grid cell is further corrected based on the feedback control coefficient. Based on the revised constraint correction transformation probability, the predicted land use type of each grid cell in the target prediction period is re-determined to obtain the optimized predicted land use layout.

8. The land use layout prediction method according to claim 7, characterized in that, After obtaining the optimized predicted land use layout, it also includes: The carbon storage prediction value is recalculated based on the optimized predicted land use layout to obtain the optimized carbon storage prediction result. Output at least one of the following: the optimized predicted land use layout, the optimized predicted carbon storage, the weighted constraint strength of each grid cell, the feedback control coefficient of each grid cell, and the comparison results between different prediction scenarios.

9. A land use layout prediction device, characterized in that, include: The data acquisition module is used to acquire rasterized basic data, land use spatial constraint data, scenario-driven data, and historical land use data corresponding to the spatial area to be predicted. The rasterized basic data includes the location data, natural environment data, and socio-economic data of each raster unit in the spatial area to be predicted. The land use spatial constraint data represents the constraint information on the land use type conversion in the spatial area to be predicted and includes at least one spatial constraint item. The spatial constraint item is a spatial data item in the land use spatial constraint data used to represent a restriction on land use type conversion. The scenario-driven data is used to represent the socio-economic development conditions and climate change conditions that affect land use type conversion under different prediction scenarios. The historical land use data includes the land use types corresponding to each raster unit at multiple historical moments. The historical conversion frequency determination module is used to determine the historical land use type conversion relationship between adjacent historical moments for each grid cell based on the historical land use data, and to determine the historical conversion frequency between each land use type based on the historical land use type conversion relationship. The first constraint strength determination module is used to determine the first constraint strength of each spatial constraint item in the land use spatial constraint data based on the spatial range of action, constraint attribute and spatial relationship between the spatial constraint item and each grid cell. The first constraint strength is used to characterize the restriction strength of the spatial constraint item on the land use type conversion of the corresponding grid cell. The discrete weight determination module is used to calculate the discrete statistics of each spatial constraint item based on the numerical distribution of the first constraint strength of the spatial constraint item on all grid cells, and to determine the discrete weight of each spatial constraint item based on the discrete statistics of each spatial constraint item. The discrete weight is used to characterize the difference of the first constraint strength of the corresponding spatial constraint item on different grid cells. The initial weight determination module is used to determine the initial weight corresponding to each spatial constraint item based on the historical conversion frequency between each land use type, wherein the initial weight is used to characterize the preset configuration proportion of the spatial constraint item in the land use type conversion constraint; The fusion weight determination module is used to, for each spatial constraint term, perform a weighted summation of the initial weight and the dispersion weight corresponding to the spatial constraint term according to a preset fusion ratio to obtain the fusion weight corresponding to the spatial constraint term. The weighted constraint strength determination module is used to multiply the first constraint strength of each spatial constraint term on the grid cell by the corresponding fusion weight and sum them up to obtain the weighted constraint strength corresponding to the grid cell for each grid cell. The scenario transition rule determination module is used to determine the scenario transition rules between different land use types for each predicted scenario, based on the scenario-driven data corresponding to the predicted scenario and the historical transition frequency between each land use type. The development suitability probability determination module is used to determine the development suitability probability of each grid cell from the current land use type to each candidate land use type for each prediction scenario, based on the scenario-driven data corresponding to the prediction scenario, the historical land use data, the rasterized basic data, and the weighted constraint strength corresponding to each grid cell. The neighborhood influence parameter determination module is used to determine the neighborhood influence parameters of each grid cell from the current land use type to each candidate land use type for each prediction scenario, based on the current land use type of each grid cell, each candidate land use type, the distribution of land use types within the neighborhood of each grid cell, and the weighted constraint strength corresponding to each grid cell. The initial conversion probability calculation module is used to calculate the initial conversion probability of each grid cell from the current land use type to each candidate land use type for each predicted scenario, based on the development suitability probability corresponding to each grid cell, the scenario transition rule and the neighborhood influence parameter. The constraint correction conversion probability calculation module is used to determine the probability correction coefficient of each grid cell corresponding to each candidate land use type for each prediction scenario based on the weighted constraint strength and scenario transition rule of each grid cell, and to correct the corresponding initial conversion probability using the probability correction coefficient to obtain the constraint correction conversion probability of each grid cell from the current land use type to each candidate land use type. The predictive layout generation module is used to determine the predicted land use type of each grid cell in the target prediction period from each candidate land use type according to the constraint correction conversion probability corresponding to each grid cell for each prediction scenario, and generate the predicted land use layout of the spatial area to be predicted under the prediction scenario based on the location data of each grid cell and the corresponding predicted land use type.

10. An electronic device, characterized in that, It includes a processor and a memory, the memory storing a computer program that, when executed by the processor, implements the land use layout prediction method according to any one of claims 1 to 8.

Images