A carbon reserve distribution prediction and evaluation method based on conflict detection and multi-objective optimization

Abstract

The present application relates to a kind of carbon reserves distribution forecast evaluation method based on conflict detection and multi-objective optimization, belong to geographic information technology field.It includes the following steps: first, construct logic regression model, carry out logic regression analysis and obtain corresponding regression coefficient, then in turn calculate the occurrence probability of each land type, local advantage index and conflict intensity index;Then, carry out multi-objective constraint condition setting, based on multi-objective constraint, the coordination of land use conflict is carried out, and the optimized land use distribution pattern is obtained;Finally, the land use data before and after coordination and carbon pool data are input into InVEST model, and the corresponding carbon reserves distribution before and after optimization is obtained.The present application improves the existing land use conflict identification and coordination method and carbon sink distribution optimization method by designing a framework combining land use conflict identification and coordination method and InVEST model, and reveals the influence of land conflict on regional carbon reserves.

Description

TECHNICAL FIELD

[0001] The present application relates to a carbon reserve distribution prediction evaluation method based on conflict detection and multi-objective optimization, belonging to the field of geographic information technology. BACKGROUND

[0002] Carbon storage refers to the carbon captured and stored by terrestrial ecosystems in the form of soil organic matter, dead vegetation and living vegetation biomass. It is a widely used indicator in ecosystem service assessment and is considered as one of the most critical measurement parameters for assessing the response of terrestrial ecosystem productivity and ecological resilience to climate change. Land use change is an important indicator reflecting the impact of human activities on the earth's surface system. It can change the structure and function of the ecosystem, and can cause changes in terrestrial carbon storage by affecting vegetation cover and changes in soil carbon storage by changing soil environment. It is a major factor affecting the changes in carbon storage and carbon cycle of terrestrial ecosystems.Land use conflict can be defined as the competition and conflict of interest among stakeholders in the process of land use due to land use patterns and structure, the change of land use spatial pattern will not only change the dynamic of land comprehensive utilization, but also change the nature of vegetation on the ground, destroy the connectivity of ecological landscape, make the regional vegetation coverage and primary ecosystem productivity decline, thus change the regional vegetation carbon storage and carbon balance, and further cause the decline of ecosystem function and ecosystem service value; the existing research on land use conflict can be divided into three categories, the first category focuses on case study of developed countries, the second category focuses on urban land use conflict in developing countries, and the third category focuses on the type of conflict in developed countries and suburban areas, such as the literature "Urban land use efficiency in Ethiopia: An assessment of urban land use sustainability in Addis Ababa" found that the expansion of urban construction in Ethiopia led to land conflict, which damaged the land use efficiency of the city, and there were serious problems related to low land use efficiency in almost all expansion boundaries, the literature "Spatial and temporal analyses of potential land use conflict under the constraints of water resources in the middle reaches of the Heihe River" constructed a multi-index evaluation system of land use competitiveness for the middle reaches of the Heihe River, and obtained the spatio-temporal pattern of potential land use conflict, the literature "How to reconcile land use conflicts in mega urban agglomeration? A scenario-based study in the Beijing-Tianjin-Hebei region, China" takes the Beijing-Tianjin-Hebei urban agglomeration as an example, constructs a spatial comprehensive conflict index (SCCI) to identify and evaluate land use conflict, however, few existing researches link urban land conflict, land cover change and carbon storage together, through identifying and coordinating land conflict, combining with multiple sustainable development goal scenarios to simulate land use change, conflict and its impact on terrestrial carbon pool, so the above is the problem and improvement direction of the method proposed in the present application.

[0003] In summary, identifying and resolving land use conflicts, optimizing urban land use, and reducing the negative impacts of intense land use changes are key to rational use of land resources, increasing regional carbon sinks, and achieving urban sustainability. Researching land use conflicts and optimizing to restore ecosystem carbon storage levels, maintain carbon storage balance, and improve ecological benefits is an urgent task and is of great significance to sustainable development. SUMMARY

[0004] The purpose of the present application is to provide a land use pattern optimization and regional carbon storage distribution optimization method, aiming to solve the problem of linking urban land conflicts, land cover changes and carbon storage, by identifying and coordinating land conflicts, combining multiple sustainable development goal scenarios to simulate land use changes, conflicts and their impact on terrestrial carbon pools.

[0005] To achieve the above-mentioned purpose, the technical scheme of the present application is: a carbon storage distribution prediction and evaluation method based on conflict detection and multi-objective optimization, which can effectively identify land use conflict areas, then coordinate under multi-objective constraints, and further combine carbon sinks to radiate the optimization results of land use patterns to regional carbon storage distribution optimization; improve the identification accuracy of land use conflict areas and main carbon sink areas; improve the method of land use conflict coordination. It includes the following steps:

[0006] S1: build the target variables, feature variables and covariates required for logistic regression analysis;

[0007] S2: Perform logistic regression analysis and obtain the corresponding regression coefficients, calculate the occurrence probability of each land type according to the regression coefficients, then calculate the local advantage index of each land type according to the occurrence probability, and finally calculate the conflict intensity index of each land type according to the local advantage index;

[0008] S3: Based on the local advantage index and the recognized land development intensity livable line, warning line, and basic land use standards for food security and ecological safety, set multi-objective constraint conditions;

[0009] S4: Coordinate land use conflicts based on multi-objective constraints to obtain optimized land use data;

[0010] S5: Input the land use data before coordination and the optimized land use data and carbon pool data into the InVEST carbon storage evaluation model to obtain the corresponding carbon storage distribution data before and after coordination, and realize the prediction of the optimized carbon storage distribution.

[0011] The step S1 specifically includes:

[0012] S1.1: Extract historical land use change as target variable from land use data, reclassify land space types in land use data into three categories, specifically building space, agricultural space, and ecological space, and represent them with different grid values respectively, and then generate three target variables according to the differences in grid values of different periods, specifically building space change, agricultural space change, and ecological space change, and identify whether changes have occurred;

[0013] S1.2: Feature variables are composed of factors that affect land use conversion, including temperature, precipitation, population, GDP, DEM, nighttime light, major road distribution, slope, and slope direction, and the grid data of the factors are downloaded and then resampled and reprojected into the same resolution and coordinate system as the land use data;

[0014] S1.3: Introduce three corresponding classification covariates according to the categories of the target variables to identify land space types.

[0015] The step S2 specifically includes:

[0016] S2.1: Use linear logistic regression model for logistic regression modeling to obtain regression coefficients of three target variables for feature variables and classification covariates respectively;

[0017] S2.2: Calculate the occurrence probability of three land spaces based on the regression coefficients of logistic regression, and the calculation formula is as follows:

[0018]

[0019] where g represents the grid position, and the subscripts c, a, and e represent building space, agricultural space, and ecological space respectively, P cg , P ag , and P eg represent the occurrence probabilities of building space, agricultural space, and ecological space at grid g respectively; α is the regression coefficient of the corresponding influencing factor; α c , α a , and α e are the regression coefficients of the corresponding influencing factors in the regression of building, agricultural, and ecological spaces; X ig represents the grid value of the influencing factor of land space i at grid g, X cg , X ag , and X eg are the grid values of the influencing factors in the regression process of building, agricultural, and ecological spaces respectively; DV c , DV a , and DV e represent the classification covariants of building, agricultural, and ecological spaces respectively; β c , β a , and β e are the regression coefficients of DVc , DV a , DV e , the regression coefficient; W is the stable ecological space, C is the unchanged stable building area; Neighborhood is the neighborhood of each land use type on each grid, and the specific calculation method is as follows:

[0020]

[0021] Neighborhood h,k represents the neighborhood aggregation degree of land use space h, k represents the kth h spatial patch in the neighborhood of a certain grid, n is the total number of patches in the neighborhood, S h,k represents the area of the kth h spatial patch in the neighborhood, and the probability values of the three land types are written into a three-channel grid image, each channel representing the probability value of the occurrence of one land type;

[0022] S2.3: Calculate the local dominance index of each land type based on the probability of occurrence, and the calculation formula is as follows:

[0023]

[0024] where, P c,i , P a,i , P e,i represent the average probability of occurrence of building, agricultural and ecological space in the neighborhood range of grid i, LDI j represents the ratio of the probability of occurrence of j space in its neighborhood range to the total probability of occurrence, which is the local dominance index, and the LDI j value is written into a three-channel grid image, each channel representing the local dominance index of one land type;

[0025] S2.4: Calculate the conflict intensity index based on the local dominance index, and the calculation formula is as follows:

[0026]

[0027] WCI op represents the weighted conflict intensity between o space and p space, o and p can be any one of building, agricultural and ecological space but cannot take the same value at the same time, LDI o and LDI p represent the local dominance index of o space and p space in their neighborhood range, and the total conflict intensity TWCI on the grid is:

[0028] TWCI = WCI ce + WCI ca + WCI ae

[0029] where WCIce ca ae WCI, WCI, WCI are the conflict intensity indexes between building space and ecological space, building space and agricultural space, and agricultural space and ecological space, respectively, for measuring the spatial distribution and intensity of land conflict, and WCI, WCI, WCI, WCI are the conflict intensity indexes of the four land types, respectively. ce ca ae The values of WCI, WCI, WCI and TWCI are written into a four-channel raster image, and the first three channels represent the conflict intensity indexes of one land type, and the last channel is the total conflict intensity.

[0030] The step S3 specifically comprises:

[0031] S3.1: Calculate the construction land livable line area and warning line area of the research area according to the internationally recognized land development intensity livable line and warning line;

[0032] S3.2: Obtain the ecological protection red line area and the cultivated land protection task area of the research area;

[0033] S3.3: Set multiple simulation scenarios;

[0034] S3.4: Arrange the three local advantage indexes in descending order according to the size of the respective raster values, and count the number of grids corresponding to each value.

[0035] The step S4 specifically comprises:

[0036] S4.1: Perform multi-objective conflict coordination by scenario, first select the raster numbers of agricultural and ecological space land that guarantee food security and ecological security basic land according to the local advantage indexes of agricultural space and ecological space in descending order of value, until the total area of the raster reaches the constraint value, and the selected raster is marked as the basic agricultural and ecological space that all scenarios have, and the selected raster no longer participates in subsequent coordination, and a mark column is set to mark it;

[0037] S4.2: Select the basic building space raster according to the size of the local advantage index of the building space, select the livable line or warning line of the building space according to the scenario, and select the basic building land raster of each scenario according to the size of the local advantage index of the building space in descending order of value, and mark it by setting a mark column;

[0038] S4.3: For the remaining raster, compare the local advantage indexes of agricultural space and ecological space, and convert the larger value to the corresponding land type, and for scenarios that need to highlight the importance of agricultural or ecological space, convert them all to the corresponding agricultural space or ecological space;

[0039] ​​​​S4.4: Assign corresponding land grids to each scenario, set a corresponding scenario column, and select corresponding basic land for food security, basic land for ecological security, and basic building land grids for each scenario in combination with the marker column, and then distribute the remaining grids according to S4.3;

[0040] S4.5: Read the values of the scenario column data and write them into three-channel grid images, wherein each channel in each image represents a land type, and the number of images is equal to the number of corresponding simulation scenarios.

[0041] The step S5 specifically comprises:

[0042] S5.1: Obtain carbon pool data, and divide carbon pool types into four types that meet the InVEST model, including above-ground biomass carbon pool, underground biomass carbon pool, soil carbon pool and dead organic matter carbon pool, and the calculation method of carbon storage of the InVEST model is as follows:

[0043] C i =C i,above +C i,below +C i,soil +C i,dead

[0044]

[0045] i is the i-th land use type; C i is the carbon density of the i-th land type; C i,above , C i,below , C i,soil , C i,dead are the above-ground biomass carbon density, underground biomass carbon density, soil organic matter carbon density and dead organic matter carbon density of the i-th land type respectively; C total is the total carbon storage of the whole region; A i is the total area of the i-th land use type; q is the number of land type classifications;

[0046] S5.2: Input the real land use spatial data and carbon pool data before coordination into the InVEST model to obtain real carbon storage distribution grid data, input the land use spatial data and carbon pool data optimized based on the multi-target constraint condition into the InVEST model to obtain carbon storage distribution prediction grid data under the coordinated multi-scenarios, and compare and analyze the land use distribution and the corresponding carbon storage distribution before and after coordination.

[0047] The beneficial effects of the present application are:

[0048] (1) The framework of the present application can reveal the characteristics of urban land use conflicts and their influence on urban carbon dynamics;

[0049] (2) The carbon storage distribution optimization method of the present application can help identify the main carbon sink area of the research area, and provide scientific basis for effective management of regional carbon storage and carbon balance;

[0050] (3) The present application simulates the influence of land use conflict and coordination on urban carbon dynamics, links land use conflict, land cover change and carbon sink of ecosystem services, provides new ideas for regional carbon sink research, and supports the analysis of urban carbon sink potential.

[0051] The method and results of the present application can also be used for reference by other cities of the same type, and provide reference basis for regional green low-carbon development, sustainable development and future land use policy making. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 is a carbon storage distribution prediction and evaluation method based on conflict detection and multi-objective optimization of the present application;

[0053] Figure 2 is an example of probability diagram of three kinds of land use space of the present application;

[0054] Figure 3 is an example of local advantage index diagram of three kinds of land use space of the present application;

[0055] Figure 4 is an example of conflict intensity diagram of three kinds of land use space and the whole of the present application;

[0056] Figure 5 is a land use conflict coordination process diagram of the present application;

[0057] Figure 6 is an example of land use diagram under multiple scenarios after land use conflict coordination of the present application;

[0058] Figure 7 is an example of carbon storage distribution diagram before and after land use conflict coordination of the present application, Figure 7 (a) is a carbon storage distribution diagram under multiple scenarios after coordination, Figure 7 (b) is a carbon storage distribution diagram without coordination. DETAILED DESCRIPTION

[0059] The specific embodiments of the present application will be further described below in combination with the drawings and specific examples:

[0060] As shown in Figure 1 , it is a carbon storage distribution prediction and evaluation method based on conflict detection and multi-objective optimization.

[0061] The embodiment uses two-phase land use raster data from Esri Sentinel-2 Land Cover, which is reclassified into three land use types (raster value 1 represents building, 2 represents agriculture, and 3 represents ecological land) in ArcGIS 10.2, resampled to 30m resolution, and spatial reference set to WGS_1984_UTM_zone_48N; the influencing factors include slope direction, slope, DEM, population, GDP, annual average precipitation, annual average temperature, night light, and main road 9, which are also resampled and projected into raster data with the same land use data; then the land use data and influencing factor data raster values are exported to an excel table, each row representing a raster in the original image, and three columns of target variables are generated based on the differences between the two-phase land use data, namely the building space change, agricultural space change, and ecological space change columns, whose values are 0 or 1, 0 representing no change and 1 representing change; based on the target variables, three corresponding classification covariate columns are set, and then the influencing factor columns are standardized if the values in the influencing factor columns are too different or there are too many extreme values; then the LogisticRegression in the python machine learning package sklearn is used for logistic regression analysis to obtain the regression coefficients of the characteristic variable columns for the three target variables.

[0062] Next, the occurrence probability of each grid of the three types of land space is calculated based on the regression coefficients of the logistic regression, taking the building space occurrence probability as an example, the calculation method is as follows:

[0063]

[0064] Where g represents the grid position, subscript c represents the building space, P cg represents the probability of the occurrence of the building space at grid g; a is the regression coefficient of the corresponding influencing factor; a c is the regression coefficient of the corresponding influencing factor in the building space regression; X ig represents the grid value of the influencing factor of the i land space at grid g, i takes the values of building space, agricultural space, and ecological space, X cg is the grid value of the influencing factor in the building space regression process; DV c represents the classification covariate of the building space; b c is the regression coefficient of DV c ; W is the stable ecological space such as reservoirs, large lakes, etc., and C is the unchanged stable building area; Neighborhood is the neighborhood of each type of land use at each grid, and the specific calculation method is as follows:

[0065]

[0066] Neighborhood c,k represents the neighborhood aggregation degree of building space, k represents the kth building space patch in the neighborhood of a certain grid, n is the total number of patches in the neighborhood, S c,k represents the area of the kth building space patch in the neighborhood; the specific calculation of Neighborhood is realized by using a sliding window algorithm on the second-stage land use grid data. A 3x3 window is moved across the entire image from top to bottom and left to right, and each time the window is moved, the values of each land type within the window are read and counted, and the neighborhood is calculated according to the formula, and the result is written into an excel table. The occurrence probability is calculated according to the formula by reading the required values in the excel table to calculate the occurrence probability of the three land types in each grid, and the value is written as a new grid, which is visualized as Figure 2 .

[0067] Next, the local dominance index of each land type is calculated based on the occurrence probability, and the calculation formula is as follows:

[0068]

[0069] where P c,i , P a,i , and P e,i represent the average occurrence probability of building, agricultural, and ecological space in the neighborhood of grid i, respectively, and LDI c represents the ratio of the occurrence probability of building space in its neighborhood to the total occurrence probability, i.e., the local dominance index, and LDI c is calculated by applying a sliding window algorithm to the occurrence probability grid image generated in the previous step, calculating the local dominance index of the three land types in each grid, and writing the value into a new grid image, which is visualized as Figure 3 ; then the conflict intensity index is calculated based on the local dominance index, and the calculation formula is as follows:

[0070]

[0071] WCI ce represents the weighted conflict intensity between building space and ecological space, and LDI e represents the ratio of the occurrence probability of ecological space in its neighborhood to the total occurrence probability, i.e., the local dominance index, and WCI ce is calculated by applying a sliding window algorithm to the newly generated local dominance index grid image, calculating the conflict intensity index of the three land types in each grid, and writing the value into a new grid image, which is visualized as Figure 4 .

[0072] In the next step of the multi-objective constraint setting step, the construction land livable line area and warning line area of the study area were calculated according to the internationally recognized land development intensity of 20% as the livable line and 30% as the warning line. Finally, four scenarios were set, namely natural development, food security, ecological protection, and high urbanization scenarios.

[0073] In the next step, the conflict coordination process of multi-objectives was carried out according to the scenario, as shown in Figure 5 . First, the four scenarios need to meet the two constraints of basic land for food security and basic land for ecological security. Therefore, according to the local advantage index of agricultural space and the local advantage index of ecological space, the grid number of basic agricultural and ecological space for food security and ecological security was taken out from large to small according to the value, until the total area of the grid reached the constraint value. The taken-out grid was used as the basic agricultural and ecological space for the four scenarios, and the taken-out grid was no longer involved in the subsequent coordination. In the excel table, two columns of mark column were set to mark them as 1. Then, according to the size of the local advantage index of building space, the basic building space grid was selected. Except for the high urbanization scenario using the land development intensity warning line, the other three scenarios took the livable line. Similarly, according to the local advantage index of building space, the basic building land grid of each scenario was selected from large to small according to the value. In the excel table, another column of mark column was set to mark them as 1. Then, for the remaining grids, the food security scenario converted them all into agricultural land, and the ecological security scenario converted them all into ecological land. While in the natural development scenario and the high urbanization scenario, the local advantage index of agricultural space and the local advantage index of ecological space were compared, and the one with the larger value was converted into the corresponding land type. Then, the corresponding land grid was allocated for each scenario. In the excel table, four columns corresponding to the four scenarios were set, and combined with the mark column, the grid of the basic food security land, the basic ecological security land, and the basic building land for each scenario was selected. Then, the remaining grids were allocated according to the rules. Finally, the values in the four scenario columns were written as new grid data, and the land use distribution data after coordination in the four scenarios was obtained, as shown in Figure 6 .

[0074] In the last step, the carbon pool data was obtained, as shown in Table 1. According to the reclassified land types, the carbon density of the six land types was combined into the carbon density of construction, agricultural, and ecological land. Specifically, building land was classified as building space, farmland was classified as agricultural space, and the remaining four types were classified as ecological space. Then, the second period land use grid data and carbon pool data were input into the InVEST model to obtain the real carbon storage distribution grid data. The optimized multi-scenario land use grid data and carbon pool data were input into the InVEST model to obtain the carbon storage distribution grid data after coordination in the multi-scenario. The visualized results are shown in Figure 7 , and Table 2 shows the comparison of carbon storage values before and after coordination.

[0075] Table 1 Carbon density (mg / hm2) of different land types -2 )

[0076]

[0077] Table 2 Comparison of carbon storage before and after coordination

[0078]

[0079] The specific embodiments of the present application are described in detail above with reference to the accompanying drawings, but the present application is not limited to the above-described embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the spirit of the present application.

Claims

1. A carbon storage distribution prediction evaluation method based on conflict detection and multi-objective optimization, characterized by: S1: constructing target variables, feature variables and covariates required for logistic regression analysis; S2: performing logistic regression analysis and obtaining corresponding regression coefficients, calculating the occurrence probability of each land type according to the regression coefficients, then calculating the local advantage index of each land type according to the occurrence probability, and finally calculating the conflict intensity index of each land type according to the local advantage index; S3: setting multi-objective constraint conditions based on local advantage index and recognized land development intensity livable line, warning line, food security and ecological safety basic land standard; The multi-objective constraint conditions specifically include: S3.1: calculating the construction land livable line area and warning line area of the study area according to the internationally recognized land development intensity livable line and warning line; S3.2: obtaining the ecological protection red line area and cultivated land protection task area of the study area; S3.3: setting multiple simulation scenarios; S3.4: arranging the three local advantage indexes in descending order according to their grid values, and counting the number of grids corresponding to each value; S4: coordinating land use conflicts based on multi-objective constraints to obtain optimized land use data; S5: inputting the land use data before coordination, the optimized land use data and the carbon pool data into the InVEST carbon storage evaluation model to obtain the carbon storage distribution data before and after coordination, and realizing the prediction of the optimized carbon storage distribution; The step S2 specifically includes: S2.1: using a linear logistic regression model to perform logistic regression modeling to obtain the regression coefficients of the three target variables with respect to the feature variables and classification covariates; S2.2: calculating the occurrence probability of three land spaces based on the regression coefficients of logistic regression, the calculation formula is as follows: ； ； ； Where g represents the grid position, and the subscripts c, a, and e represent architectural space, agricultural space, and ecological space, respectively. cg P ag P eg These represent the probabilities of architectural space, agricultural space, and ecological space appearing on grid g, respectively; α is the regression coefficient of the corresponding influencing factor; α c α a α e X represents the regression coefficients of the influencing factors corresponding to the regression of architectural, agricultural, and ecological spaces; ig X represents the raster value of the influencing factor on land space i on raster g. cg X ag X eg These are the raster values ​​of the influencing factors in the regression process of architectural, agricultural, and ecological spaces, respectively. DV c , DV a , DV e represent the categorical covariates for buildings, agriculture, and ecological space, respectively; β c , β a , β e are the regression coefficients for DV c , DV a , DV e , respectively; W is the stable ecological space, C is the unchanged stable building area; Neighborhood is the neighborhood of each land use type in each grid, which is calculated as follows: ； Neighborhood h,k The neighborhood aggregation degree of land use space h, k represents the kth h spatial patch in the neighborhood of a certain grid, n is the total number of patches in the neighborhood, S h,k The area of the kth h spatial patch in the neighborhood, and the probability values of the three land types are written into a three-channel grid image, and each channel represents the probability value of the occurrence of a land type. S2.3: Calculate the local dominance index of each land type based on the occurrence probability, LDI j The ratio of the occurrence probability of j space in its neighborhood range to the total occurrence probability is the local dominance index, j is any one of building, agriculture, and ecological space. LDI j values are written to a three-channel raster image, each channel representing a local dominance index for one land type; S2.4: calculating the conflict intensity index based on the local advantage index, the calculation formula is as follows: ； WCI op represent the weighted conflict intensity between o-space and p-space, o, p are any one of building, agriculture, ecological space but cannot take the same value at the same time, LDI o and LDI p respectively represent the local dominance index of o-space and p-space within their neighborhood range, and the total conflict intensity TWCI on the grid is: ； Among them WCI ce WCI ca WCI ae These are conflict intensity indices for built space and ecological space, built space and agricultural space, and agricultural space and ecological space, respectively, used to measure the spatial distribution and intensity of land conflicts. ce WCI ca WCI ae The TWCI values ​​are written into a four-channel raster image, with the first three channels representing conflict intensity indices for a land type and the last channel representing total conflict intensity.

2. The carbon stock distribution forecast evaluation method based on conflict detection and multi-objective optimization according to claim 1, characterized in that, The step S1 specifically includes: S1.1: extracting historical land use change from land use data as target variable, reclassifying land space types in land use data into three categories, specifically building space, agricultural space and ecological space, and representing them with different grid values, then generating three target variables according to the differences between grid values in different periods, specifically building space change, agricultural space change and ecological space change, and identifying whether there is a change; S1.2: the feature variables are composed of factors that affect land use change, including temperature, precipitation, population, GDP, DEM, night light, main road distribution, slope, slope direction, and the grid data of the factors are downloaded and then resampled and reprojected to the same resolution and coordinate system as the land use data; S1.3: introducing corresponding three classification covariates according to the categories of target variables to identify land space types.

3. The method for carbon reserve distribution forecast evaluation based on conflict detection and multi-objective optimization according to claim 1, characterized in that, The step S4 specifically includes: S4.1: Multi-objective conflict coordination is carried out according to different scenarios. First, the number of grids of agricultural and ecological space land for ensuring food safety and ecological safety is taken out from large to small according to the value of the local advantage index of agricultural space and the local advantage index of ecological space, until the total area of the grid reaches the constraint value. The taken-out grid is taken as the basic agricultural and ecological space of all scenarios, and the taken-out grid no longer participates in subsequent coordination, and a mark column is set to mark it; S4.2: The basic building space grid is selected according to the size of the local advantage index of the building space. The livable line or warning line of the building space is selected according to the scenario. The basic building land grid of each scenario is selected according to the size of the local advantage index of the building space, and a mark column is set to mark it; S4.3: For the remaining grids, compare the local advantage index of agricultural space and the local advantage index of ecological space, and convert the value to the corresponding land type. For the scenarios that need to highlight the importance of agricultural or ecological space, they are all converted into the corresponding agricultural space or ecological space; S4.4: Assign the corresponding land grid to each scenario, set the corresponding scenario column, and select the corresponding grid of the basic land for ensuring food safety, the basic land for ensuring ecological safety, and the basic building land for each scenario in combination with the mark column, and then distribute the remaining grids according to S4.3; S4.5: Read the value of the scenario column data, and write it into three-channel grid images, wherein each channel in each image represents a land type, and the number of images is the number of corresponding simulation scenarios.

4. The method for carbon reserve distribution forecast evaluation based on conflict detection and multi-objective optimization according to claim 1, characterized in that, The step S5 specifically comprises: S5.1: Obtain carbon pool data, and divide the carbon pool type into four carbon pools conforming to the InVEST model, including aboveground biomass carbon pool, underground biomass carbon pool, soil carbon pool and dead organic matter carbon pool. The calculation method of carbon storage of the InVEST model is as follows: ； ； i is the i-th land use type; C i Ci is the carbon density of the i-th land type; C i,above , C i,below , C i,soil , C i,dead respectively are the aboveground biomass carbon density, belowground biomass carbon density, soil organic matter carbon density and dead organic matter carbon density of the i-th land type; C total A is the total carbon storage of the whole region; A i Qi is the total area of the i-th land use type; q is the number of land type classification; S5.2: Input the real land use space data and carbon pool data before coordination into the InVEST model to obtain the real carbon storage distribution grid data, input the land use space data and carbon pool data optimized based on the multi-objective constraint condition into the InVEST model to obtain the carbon storage distribution prediction grid data after coordination under multiple scenarios, and compare and analyze the land use distribution before and after coordination and the corresponding carbon storage distribution.

Images