A regional ecological security pattern optimization method based on a multi-objective genetic algorithm

Abstract

The application relates to the field of ecological security pattern optimization methods, and relates to a regional ecological security pattern optimization method based on a multi-objective genetic algorithm, which comprises the following steps: acquiring various types of geographic spatial data of an ecological region; obtaining ecological importance, ecological sensitivity and landscape advantage index of the ecological region according to the various types of geographic spatial data, and determining an ecological source according to the three types of indexes; obtaining a modified resistance surface by superimposing the weights of various resistance factors of natural conditions and social economic activities according to the ecological region; determining ecological corridors between various ecological sources, ecological nodes and constructing an ecological security pattern according to each ecological source and the resistance surface; giving consideration to the two targets of food security and ecological benefit while optimizing the regional ecological security pattern according to the multi-objective genetic algorithm; and evaluating the optimization effect of the optimized ecological security pattern, so that the application provides effective data support for the multi-objective genetic algorithm optimization of the regional ecological security pattern.

Description

Technical Field

[0001] This invention relates to the field of ecological security pattern optimization methods, specifically a regional ecological security pattern optimization method based on a multi-objective genetic algorithm. Background Technology

[0002] An ecological security pattern is a model or structure for planning and constructing spatial layout to achieve ecological security. It emphasizes the integrity, connectivity, and balance of ecological functions to ensure the continuity and stability of ecosystem services. A rational and scientific ecological security pattern can effectively protect the health and stability of ecosystems; therefore, it is an important means and approach to achieving ecological security goals. Currently, processes such as climate change, unsustainable land-use change, and rapid urbanization have severely disrupted the balance of natural ecosystems, leading to a decline in regional ecosystem service functions, an increase in ecologically sensitive areas, and a deterioration in landscape patterns. This has resulted in significant changes to the ecological security pattern, thereby threatening ecological security. Therefore, there is an urgent need to optimize the regional ecological security pattern. Currently, there is a lack of effective and reasonable optimization methods for the ecological security pattern. Breakthrough results are scarce in the integrated protection and systematic governance of mountains, rivers, forests, fields, lakes, grasslands, and deserts. The focus is concentrated on optimizing the ecological security pattern for a single objective, neglecting the importance of coordinated development of multiple regional objectives and coordinated restoration of multiple elements. This results in a lack of reliable technical support for the protection and restoration of the regional ecological security pattern. In other words, the existing technology lacks a regional ecological security pattern optimization scheme based on multi-objective coordinated development. Therefore, this invention proposes a regional ecological security pattern optimization method based on a multi-objective genetic algorithm to address the above problems. Summary of the Invention

[0003] The purpose of this invention is to provide a method for optimizing regional ecological security patterns based on a multi-objective genetic algorithm, so as to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, a method for optimizing regional ecological security patterns based on a multi-objective genetic algorithm includes the following steps:

[0005] Step S1: Obtain various geospatial data of the ecological region;

[0006] Step S2: For ecological regions, based on various geospatial data, obtain the ecological importance, ecological sensitivity, and landscape dominance indices of the ecological regions, and determine the ecological source areas based on the three indices;

[0007] Step S3: For the ecological region, the modified resistance surface is obtained by weighting and superimposing various resistance factors based on natural conditions and socio-economic activities.

[0008] Step S4: Based on each ecological source area and resistance surface, determine the ecological corridors and ecological nodes between each ecological source area, and construct an ecological security pattern.

[0009] Step S5: Based on the multi-objective genetic algorithm, while optimizing the regional ecological security pattern, the two objectives of food security and ecological benefits are taken into account.

[0010] Step S6: Evaluate the optimization effect of the optimized ecological security pattern.

[0011] Preferably, the geographic data in step S1 includes multi-source data such as meteorological, soil, hydrological, land use, and remote sensing images with the same spatiotemporal scale.

[0012] Preferably, step S2 assesses three ecosystem service functions—habitat quality, water conservation, and soil retention—based on multi-source data such as climate and land use, and then weights them equally to obtain an ecological importance index. Based on multi-source remote sensing data, it calculates three sensitivity indices: nitrogen and phosphorus pollution sensitivity, arable land soil erosion sensitivity, and human disturbance sensitivity, and then weights them equally to obtain an ecological sensitivity index. It also uses land use data to estimate the degree of landscape fragmentation, landscape complexity, landscape aggregation connectivity, and landscape heterogeneity, and then weights them equally to obtain a landscape dominance index.

[0013] Preferably, in step S3, seven resistance factor indicators are selected: land use, altitude, NDVI, nighttime light, distance from road, distance from railway, and distance from water. The resistance factor levels are divided into five categories, and the resistance coefficients are set to 1-5. The weights are superimposed to obtain the ecological resistance surface of the ecological area.

[0014] Preferably, step S4, based on the ecological source areas and comprehensive resistance surfaces, and based on circuit theory principles, determines the corridors, pinch points, and obstacle points between ecological source areas using the following formula, and constructs an ecological security pattern.

[0015]

[0016] Where I represents the size of the ecological flow through the corridor; V represents the size of the ecological source measured throughout the corridor; and R represents the cumulative resistance of the corridor.

[0017] Preferred: In step S5, based on a multi-objective genetic algorithm, while optimizing the ecological security pattern, the two objectives of food security and ecological benefits are considered simultaneously. The specific steps are: (1) Setting the objective function.

[0018] First, the goals of the ecological security framework.

[0019] Max ESP =W ESP ×(A ESI +A ESS+A LDI (27)

[0020] Among them, ESP represents the ecological security pattern objective; W ESP As the weight of the ecological security pattern goal, A ESI As an ecological importance parameter, A ESS A is an ecological sensitivity parameter. LDI For landscape dominance parameters; ESP represents the ecological security pattern objective derived from A. ESI A ESS A LDI It consists of three parameters;

[0021] Secondly, the goal of food security.

[0022]

[0023] Among them, FP represents the food security target; W FP As a weighting of food security goals, NPP x Let NPP be the NPP value of the x-th pixel on cultivated land; P be the total grain output of a certain county / district; NPP sum This represents the total NPP (National Per Capita Preference) on agricultural land in a certain county / district.

[0024] Finally, ecological benefit goals.

[0025] Max ESV =W ESV ×(∑(VC i ×A i (29)

[0026] Among them, ESV represents the ecological benefit target; W ESV As a weighting of ecological benefits, VC i Let i be the ecological benefit value per unit area of ​​the i-th land use type; Ai is the area of ​​the i-th land use type.

[0027] (2) Condition constraints: Slopes with a gradient greater than 25° are prohibited from being converted into cultivated land, and slopes with a gradient less than 5° are prohibited from being converted into forest land; the conversion probability of water bodies is 0; condition constraints for each land use type.

[0028] (3) Quantity constraints: available land area constraints; forest coverage constraints; grassland area constraints; cultivated land area constraints; wetland area constraints.

[0029] (4) Multi-objective genetic algorithm design: quantitative optimization, based on the Non-dominated Sorting Genetic Algorithm II (NGSA-II) algorithm, combined with ecological security pattern objectives, food security objectives and ecological benefit objectives, to perform quantitative optimization of land use types in ecological areas;

[0030] Spatial optimization: Based on land use data from 2010 and 2020 for the ecological region and five driving factors (elevation, slope, precipitation, NDVI, and temperature), the Geosos-FLUS model was initialized and trained to produce a land use spatial optimization simulation model with a Kappa coefficient higher than 0.85.

[0031] (5) Ecological security pattern optimization: Based on the optimized land use pattern, an optimized ecological security pattern is constructed, thereby obtaining the regional ecological security pattern optimized by the multi-objective genetic algorithm.

[0032] Preferably, step S6 evaluates the optimization effect of the optimized ecological security pattern by comparing the optimized regional ecological security pattern obtained in S5 with the regional ecological security pattern before optimization, calculating the optimization effect, and calculating the relative optimization effect η. a η b The expressions are as follows:

[0033]

[0034] Compared with the prior art, the beneficial effects of this invention are as follows:

[0035] This invention replaces the traditional Markov model prediction in the second step of spatial optimization with the NGSA-II algorithm. While optimizing the ecological security pattern based on the NGSA-II algorithm, it also simultaneously optimizes the food security and ecological benefit objectives. The optimized values ​​of various land use types are then used in the third step for image simulation, thereby achieving multi-objective collaborative optimization in space. This realizes multi-objective optimization of the ecological security pattern, breaking through the limitation of traditional optimization that only considers the single objective of ecological security pattern. It highlights multi-objective dynamic optimization and the integrated protection and management of mountains, rivers, forests, fields, lakes, grasslands, and deserts. Attached Figure Description

[0036] Figure 1 This is a schematic diagram illustrating the overall process of the regional ecological security pattern optimization method based on a multi-objective genetic algorithm proposed in this invention.

[0037] Figure 2 This represents the land use pattern optimization map obtained based on the multi-objective genetic algorithm proposed in this embodiment of the invention;

[0038] Figure 3 This represents the ecological security pattern optimization diagram obtained based on the multi-objective genetic algorithm proposed in this embodiment of the invention;

[0039] Figure 4 The multi-objective genetic algorithm proposed in this invention has the effects of optimizing ecological source areas and ecological corridors.

[0040] Figure 5 The data provided are the constraint data for each land use type in the example. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] Example

[0043] Please see Figures 1-5 The figure shows a preferred embodiment of the present invention, a method for optimizing regional ecological security patterns based on a multi-objective genetic algorithm, comprising the following steps:

[0044] Step S1: Obtain various geospatial data of the ecological region;

[0045] Step S2: For ecological regions, based on various geospatial data, obtain the ecological importance, ecological sensitivity, and landscape dominance indices of the ecological regions, and determine the ecological source areas based on the three indices;

[0046] Step S3: For the ecological region, the modified resistance surface is obtained by weighting and superimposing various resistance factors based on natural conditions and socio-economic activities.

[0047] Step S4: Based on each ecological source area and resistance surface, determine the ecological corridors and ecological nodes between each ecological source area, and construct an ecological security pattern.

[0048] Step S5: Based on the multi-objective genetic algorithm, while optimizing the regional ecological security pattern, the two objectives of food security and ecological benefits are taken into account.

[0049] Step S6: Evaluate the optimization effect of the optimized ecological security pattern.

[0050] In this embodiment, the various types of geographic data in step S1 include multi-source data such as meteorological, soil, hydrological, land use, and remote sensing images with the same spatiotemporal scale.

[0051] In this embodiment, step S2 assesses three ecosystem service functions—habitat quality, water conservation, and soil retention—based on multi-source data such as climate and land use, and then weights them equally to obtain an ecological importance index. Specifically,

[0052] (1) Habitat quality (HQ): Based on the following formula (1), habitat quality is calculated by combining habitat suitability, the intensity of external threats, and sensitivity. The calculation formula is as follows:

[0053]

[0054] In the formula: Qxj represents the habitat quality of pixel x on land cover of type j, H j Habitat suitability for type J land cover.

[0055] (2) Water conservation (WR): Water conservation is obtained by subtracting the total annual runoff from the water production. The water production is estimated based on the following formula (2). The total annual runoff is estimated by assigning an average runoff coefficient value to each land use type and multiplying it by the annual precipitation of each pixel.

[0056]

[0057] In the formula: Y xj The annual water yield (m3) of grid cell x on land use type j; AET xj P represents the annual average evapotranspiration of grid cell x on land use type j. x The annual precipitation for grid cell x; AET xj / P x The Zhang coefficient is an algorithm based on the Budyko equation and simulated in conjunction with actual conditions.

[0058] Runoff ij =P ij ×C ij (3)

[0059] WR ij =Y xj -Runoff ij (4)

[0060] In the formula, WR ij It is LUCC j The water conservation capacity of each pixel in that year, Y xj It is LUCC j The annual water output (mm) of each pixel, Runoff is LUCC. j Annual runoff (mm / a) of the land surface in each pixel, P ij Indicates LUCC j The annual precipitation (mm) for each pixel in that year.

[0061] C j The surface runoff coefficient for land use type j is shown in Table 1.

[0062] Table 1 Average runoff coefficient data

[0063]

[0064]

[0065] (3) Soil retention (SR): The amount of erosion reduction and sediment retention is estimated by the following formula (5). Finally, the amount of soil retention is obtained by subtracting the amount of sediment retention from the amount of soil loss.

[0066] SEDRET = PKLS x -USLE x (5)

[0067] In the formula: SEDRET x This represents the soil retention capacity of grid x for a certain land use type, using PKLS. x USLEx represents the potential soil loss of grid x in a certain land use type, while USLEx represents the actual erosion of grid x in a certain land use type, i.e., the soil erosion under vegetation cover and soil and water conservation measures.

[0068] In this invention, the three types of ecosystem service function indicators contribute equally to ecological importance. Therefore, habitat quality, water conservation, and soil retention are normalized. The cumulative ecological importance score (ESI) of each grid is the sum of the scores of the three types of ecosystem service functions, calculated as follows:

[0069] ESI=NorHQ(i,t)+NorWR(i,t)+NorSR(i,t) (6)

[0070] In the formula: NorHQ(i,t), NorWR(i,t), and NorSR(i,t) are the normalized values ​​of habitat quality, water conservation, and soil retention scores for grid i in year t, respectively.

[0071] Furthermore, step S2, based on multi-source remote sensing data, calculates three sensitivity indices: nitrogen and phosphorus pollution sensitivity, arable land soil erosion sensitivity, and human disturbance sensitivity, and then weights them equally to obtain an ecological sensitivity index, specifically including:

[0072] (1) Nitrogen and phosphorus pollution sensitivity (NDR): The nitrogen and phosphorus output of each land use type was estimated based on the following equations (7), (8) and (9).

[0073] ALV x =HSS x ×POL x (7)

[0074]

[0075] λ x =log(∑ U Y U (9)

[0076] In the formula, ALV x POL is the load value adjusted at grid cell x. x HSS is the output coefficient of grid cell x. x This represents the hydrological sensitivity score at grid cell x. λ is the average runoff index of the catchment area. x is the runoff index at grid cell x, and is the sum of the water production of all grid cells in the catchment area.

[0077] (2) Farmland Soil Erosion Sensitivity (FP): The sensitivity of farmland soil erosion is characterized by food supply, and combined with NPP to achieve a reasonable distribution of food production on farmland in each county of the ecological region, specifically:

[0078]

[0079] In the formula: P is the grain yield (kg) of the x-th pixel; NPP is the NPP value of the x-th pixel on cultivated land; P is the total grain yield of a certain county; NPP is the sum of NPP of a certain county on agricultural land.

[0080] (3) Human Disturbance Sensitivity (HDI): Based on multi-source data, the pressure of population concentration, land use intensity, traffic network pressure and power consumption pressure are estimated; and the cumulative human disturbance sensitivity index is obtained by weighting them equally.

[0081] The population density pressure calculation uses 1km×1km resolution raster data of population density from the LandScan global population database. For grids with more than 1,000 people per square kilometer, the interference intensity score is set at 10 points. 2 The following grid uses a logarithmic method to determine the population density stress score, thus obtaining the population agglomeration stress index for the ecological region.

[0082]

[0083] Among them, PDI s Population density is the pressure of population concentration. Pd represents population density. PDI s The value range is [0, 10]. The higher the score, the stronger the interference.

[0084] Land use intensity is calculated by assigning values ​​to the relative pressure exerted on the natural ecosystem by each land use form, resulting in the land use intensity index of the ecological region, as shown in Table 2.

[0085] Table 2. Quantitative Assignment of Pressure by Land Use Type

[0086]

[0087]

[0088] A land use intensity index is constructed by calculating the area proportion of different land use types within a grid, which is used to describe the relative magnitude of land use risk in a region. The formula is as follows:

[0089]

[0090] Among them, LDI s Let n represent the land use intensity, and n be the total number of land use types contained in the grid.

[0091] P i S represents the percentage of the raster area occupied by land use type i. i The pressure is quantified into scores for land use types.

[0092] Traffic network pressure was calculated by assigning pressure values ​​to traffic networks with different buffer radii based on Euclidean distance, resulting in the traffic network pressure index for the ecological region, as shown in Table 3.

[0093] Table 3 Traffic pressure assignments based on Euclidean distance

[0094]

[0095] The electricity consumption pressure assessment converts nighttime light data into electricity consumption figures to characterize electricity consumption intensity. A power function fitting model is then constructed to fit the relationship between the electricity consumption data and nighttime light intensity values ​​of the ecological area, thereby obtaining the electricity consumption pressure of the ecological area.

[0096] E ij =DN sij a (13)

[0097] In the formula, E ij DN represents the total electricity consumption of city j in year i; sij ; represents the sum of DN values ​​for nighttime lighting data in city j in year i; a represents the predicted parameters of the power function fitting model.

[0098] Cumulative Anthropogenic Disturbance Intensity Calculation: Since the four types of anthropogenic disturbance indicators contribute equally to cumulative anthropogenic pressure, population concentration pressure, land use pressure, transportation network pressure, and electricity consumption pressure are normalized. The cumulative anthropogenic disturbance pressure score for each grid is the sum of the scores for the four types of anthropogenic pressure. Specifically,

[0099] HDI(i,t)=NorPDI s (i,t)+NorLDI s (i,t)+NorTDI s (i, t) + NorEDIs (i, t) (14)

[0100] Wherein, HDI(i,t) is the cumulative human-caused disturbance pressure score for grid i in year t.

[0101] NorPDI s (i,t), NorLDI s (i,t), NorTDI s (i, t), NorEDI s (i, t) are the normalized scores of population agglomeration pressure, land use pressure, transportation network pressure and electricity consumption pressure for grid i in year t, respectively.

[0102] This invention posits that the three types of ecological sensitivity indicators contribute equally to ecological sensitivity. Therefore, it normalizes the sensitivity to nitrogen and phosphorus pollution, arable land soil erosion, and human disturbance. The cumulative ecological sensitivity score for each grid cell is the sum of the scores for the three types of ecological sensitivity. Specifically,

[0103] ESS=NorNDR(i,t)+NorFP(i,t)+NorHDI(i,t) (15)

[0104] In the formula: NorNDR(i,t), NorFP(i,t), and NorHDI(i,t) are the normalized values ​​of the nitrogen and phosphorus pollution sensitivity, farmland soil erosion sensitivity, and human disturbance sensitivity scores of grid i in year t, respectively.

[0105] Furthermore, landscape indices can highly condense landscape pattern information, reflecting certain characteristics of its structural composition and spatial configuration. The landscape pattern index method can effectively measure regional landscape patterns and reflect changes in specific regional landscape pattern characteristics at different scales. It uses land use data to estimate the degree of landscape fragmentation, landscape complexity, landscape aggregation connectivity, and landscape heterogeneity, and then weights these parameters to obtain a landscape dominance index, specifically including…

[0106] Landscape fragmentation index (LFI): The landscape fragmentation index is obtained by weighting patch density (PD) and average patch area (AREA-MN). PD reflects the complexity of the landscape spatial structure; AREA-MN represents an average condition and can reflect the degree of landscape fragmentation in landscape structure analysis.

[0107]

[0108] In the formula: N is the number of patches contained in patch type i in the landscape; N is the total number of patches in the landscape; and A represents the total area of ​​the landscape in each ecological region.

[0109]

[0110] In the formula: A is the total area of ​​a certain patch type, and N is the total number of patches of this patch type.

[0111] LFI=NorPD(i,t)+NorAREA_MN(i,t) (18)

[0112] In the formula: NorPD(i,t) and NorAREA_MN(i,t) are the normalized values ​​of patch density and average patch area index score of grid i in year t, respectively.

[0113] Landscape Complexity Index (LCI): The Landscape Complexity Index is obtained by weighting Edge Density (ED) and Mean Segment Shape (MSI). ED is an important indicator reflecting the shape of the patches, indicating the degree to which the types are segmented. MSI reflects the overall complexity of the patch shapes of the landscape types.

[0114]

[0115] A higher ED value indicates a higher edge density, a greater degree of segmentation of landscape types, and a more dispersed layout.

[0116]

[0117] In the formula: MSI is used to characterize the overall complexity of the shape of landscape patch (MSI≥1). When all patches are square, MSI=1. The larger the value, the more irregular the patch shape.

[0118] LCI=NorED(i,t)+NorMSI(i,t) (21)

[0119] In the formula: NorED(i,t) and NorMSI(i,t) are the normalized values ​​of the edge density and average patch shape index score of grid i in year t, respectively.

[0120] The Landscape Aggregation Connectivity Index (LACI) is obtained by weighting the aggregation degree (AI) and the spread degree (CONTAG). The larger the AI ​​value, the greater the spatial aggregation of patch types. The higher the CONTAG value, the better the connectivity of a certain dominant patch type in the landscape. Conversely, the lower the CONTAG value, the more fragmented the landscape is.

[0121]

[0122] Where: g ii maxg is the number of pairs of adjacent pixels of the patch type. iiThis represents the maximum number of adjacent pixels for each patch type. It reflects the spatial clustering and connectivity of the patch type.

[0123]

[0124] In the formula: P i Let represent the area proportion of patch type i in the landscape, gik represent the number of adjacencies between pixels of patch types i and k based on the double-counting method, and m represent the number of patch types present in the landscape.

[0125] LAFI=NorAI(i,t)+NorCONTAG(i,t) (24)

[0126] In the formula: NorAI(i,t) and NorCONTAG(i,t) are the normalized values ​​of the clustering and spread scores of grid i in year t, respectively.

[0127] Landscape heterogeneity index (LHI): The landscape heterogeneity index is characterized by the Shannon diversity index (SHDI) to reflect the degree of landscape heterogeneity. The higher the diversity index, the richer the landscape types. As the difference in the area proportion of each landscape type increases, the diversity decreases.

[0128]

[0129] In the formula: the ratio of the number of patches to the area in a landscape type is represented by P. i This indicates the quantity of landscape elements and the specific proportion of each landscape element.

[0130] Based on the assessment results of ecological importance, ecological sensitivity, and landscape dominance of the ecological regions using equal weighting, they were divided into five levels according to the natural breakpoint method. A threshold method was then used to select areas larger than 200 km² from the first two levels. 2 Ecological patches; and ecological protection zones are superimposed to obtain the ecological source area of ​​the ecological region.

[0131] Step S3 selects seven resistance factor indicators: land use, altitude, NDVI, nighttime light, distance from road, distance from railway, and distance from water body. These resistance factors are categorized into five levels, with resistance coefficients set from 1 to 5 (see Table 4). The weighted summation yields the ecological resistance surface of the ecological region, specifically including…

[0132] (1) Land use types

[0133] Land use types affect regional information exchange, material and energy flow. The closer a land use pattern is to the ecological source, the stronger its ecological function and the smaller its ecological resistance. Conversely, the more human interference a land use type is subjected to, the greater its ecological resistance. Therefore, forest land is considered the land use type with the least ecological resistance, followed by grassland, water area, cultivated land, unused land, and building land, with resistance values ​​of 1, 2, 3, 4, and 5, respectively.

[0134] (2) Altitude

[0135] Altitude is used to reflect the topography of the ecological area. The higher the altitude, the higher the resistance value. The altitude is divided into five levels: <255m, 255m-491m, 491m-748m, 748m-1079m, and >1079m, with resistance values ​​of 1, 2, 3, 4, and 5, respectively.

[0136] (3) NDVI

[0137] NDVI, or Normalized Difference Vegetation Index, is a commonly used index to detect vegetation growth, vegetation cover, and dynamic changes in vegetation over time. NDVI is directly proportional to vegetation cover; the higher the NDVI value, the higher the vegetation cover, the higher the habitat quality, and the lower the ecological resistance. Using the natural breakpoint method, it is divided into five levels: 0-0.296, 0.296-0.384, 0.384-0.469, 0.469-0.563, and 0.563-0.804, with resistance values ​​of 5, 4, 3, 2, and 1, respectively.

[0138] (4) Night lighting

[0139] Nighttime light data is used to reflect the intensity of human activity in the ecological area. The higher the nighttime light data value, the greater the intensity of human interaction, the worse the ecological environment, and the greater the resistance value. Nighttime light is divided into five levels: 0, 0-15, 15-29, 29-47, and 47-63, with resistance values ​​of 1, 2, 3, 4, and 5, respectively.

[0140] (5) Distance from the road

[0141] The development of transportation reflects the intensity of human activities. The more developed the transportation, the more serious the interference with the ecosystem, and the greater the resistance to ecological flow. This seriously hinders the migration of organisms and the exchange of information. The selection of national highways, provincial highways, and expressways as factors affecting the ecological source areas is based on the distance from the road, which is divided into five levels: 0-1000m, 1000m-2000m, 2000m-3000m, 3000m-4000m, and >4000m. Their resistance values ​​are 5, 4, 3, 2, and 1, respectively.

[0142] (6) Distance from the railway

[0143] The distance from the railway is divided into 5 levels: 0-1000m, 1000m-2000m, 2000m-3000m, 3000m-4000m, and >4000m, with resistance values ​​of 5, 4, 3, 2, and 1 respectively.

[0144] (7) Distance from water

[0145] Water bodies are important pathways for the migration of matter, energy, and species. Therefore, the closer one is to a water body, the better the ecological quality and the lower the ecological resistance value. The distance from water bodies is divided into 5 levels: 0-1000m, 1000m-2000m, 2000m-3000m, 3000m-4000m, and >4000m, with resistance values ​​of 1, 2, 3, 4, and 5, respectively.

[0146] Table 4 Determination of Single-Factor Resistance Coefficient and Weight

[0147]

[0148]

[0149] S4. Based on the ecological source areas and comprehensive resistance surfaces, and based on the principles of circuit theory, the corridors, junctions, and obstacle points between ecological source areas are determined through equation (26), and an ecological security pattern is constructed.

[0150]

[0151] Where I represents the size of the ecological flow through the corridor; V represents the size of the ecological source measured throughout the corridor; and R represents the cumulative resistance of the corridor.

[0152] Step S5, based on a multi-objective genetic algorithm, optimizes the ecological security pattern while simultaneously considering both food security and ecological benefits. This is specifically carried out in the following steps:

[0153] (1) Objective function setting:

[0154] Ecological security pattern goals

[0155] Max ESP =W ESP ×(A ESI +A ESS +A LDI (27)

[0156] Among them, ESP represents the ecological security pattern objective; W ESP As the weight of the ecological security pattern goal, A ESI As an ecological importance parameter, A ESS A is an ecological sensitivity parameter. LDI For landscape dominance parameters; ESP represents the ecological security pattern objective derived from A.ESI A ESS A LDI It consists of three parameters.

[0157] Food security goals

[0158]

[0159] Among them, FP represents the food security target; W FP As a weighting of food security goals, NPP x Let NPP be the NPP value of the x-th pixel on cultivated land; P be the total grain output of a certain county / district, and NPP be... sum This represents the total NPP (National Per Capita Preference) on agricultural land in a certain county / district.

[0160] Ecological benefit goals

[0161] Max ESV =W ESV ×(∑(VC i ×A i (29)

[0162] Among them, ESV represents the ecological benefit target; W ESV As a weighting of ecological benefits, VC i Ecological benefit value per unit area for land use type i (yuan / hm²) 2 Ai represents the area (hm) of the i-th land use type. 2 ).

[0163] (2) Condition constraints:

[0164] Land with a slope greater than 25° may not be converted into farmland, and land with a slope less than 5° may not be converted into forest land.

[0165] The probability of water body conversion is 0;

[0166] Constraints for each land use type, such as Figure 5 As shown, 1 indicates that conversion is possible, and 0 indicates that conversion is not possible.

[0167] (3) Quantity constraints:

[0168] Available land area constraints

[0169]

[0170] Where A is the total area of ​​land use types, x is the number of land use types, and X i Let be the area of ​​the i-th land type.

[0171] Forest coverage constraints:

[0172] X forest land≥t×A forest land

[0173] Where t represents vegetation coverage, determined according to the 2035 long-term planning goals outline for ecological regions, A forest land This represents the current forest area of ​​the ecological region; simply input the data.

[0174] Grassland area constraints:

[0175] X graassland ≥A grassland

[0176] Among them, A grassland This represents the current grassland area of ​​the ecological region; simply input this value.

[0177] Farmland area constraints:

[0178] X farmland ≥A farmland

[0179] Among them, A farmland This represents the current arable land area of ​​the ecological region; simply input the data.

[0180] Wetland area constraints:

[0181] X wetland ≥t×A wetland

[0182] Where t represents the wetland protection and restoration rate, determined according to the 2035 long-term planning goals outline for ecological regions, A wetland This represents the current wetland area of ​​the ecological region; simply input the value.

[0183] (4) Multi-objective genetic algorithm design:

[0184] Quantitative optimization: This invention is based on the Non-dominated Sorting Genetic Algorithm II (NGSA-II) algorithm, which combines ecological security pattern objectives, food security objectives, and ecological benefit objectives to quantitatively optimize land use types in ecological regions.

[0185] Genetic algorithms mimic Darwin's theory of "survival of the fittest," using natural selection to coordinate relationships between objective functions and find the optimal solution to a mathematical function. The NGSA-II algorithm, also known as the second-generation non-dominated sorting genetic algorithm, improves upon genetic algorithms in initialization, inheritance, crossover, and mutation. Its advantages are threefold: first, it uses randomly generated land use data to enhance the algorithm's ability to converge to reasonable results; second, it introduces a single-parent crossover operator; and third, it adds patch mutation and constraint-guided mutation operators. Starting from a population containing multiple initial solutions (in genetic algorithms, solutions are also called individuals), it assigns a fitness value to each solution based on its quality. Then, mimicking the natural law of survival of the fittest, individuals with high fitness are selected to mate and produce offspring. Offspring individuals have a certain probability of crossover and mutation, thereby enriching the diversity of the population. Under normal circumstances, two parent individuals will produce two offspring to ensure that the population size remains unchanged in each generation. The offspring population continues to repeat the above process until a certain generation of the population meets the given termination condition, at which point one optimization is completed. The NGSA-II algorithm is a multi-objective optimization algorithm that is widely used in fields such as land use optimization. Land use optimization involves multiple objectives, such as economic benefits, environmental protection, and social benefits. Traditional single-objective optimization methods are difficult to consider these complex objectives simultaneously, while NGSA-II can effectively solve such problems.

[0186] Furthermore, the basic framework is as follows: for a population P of size N in the t-th generation... t First, following the basic genetic algorithm process, selection, crossover, and mutation are performed to generate the offspring population Q. t Next, P t and Q t Merging, denoted as R t And from R with 2N individuals t N individuals are selected from the population to form the next generation population P. t+1 The screening step is to select R... t The population is sorted using a non-dominated ranking system. Non-dominated ranking involves assigning a non-dominated rank to each solution. A solution is considered non-dominated if no solution in the solution set has a better objective function value. The non-dominated rank of a non-dominated solution is 1. After removing solutions with rank 1, the non-dominated rank of the remaining solutions in the population is 2, and so on, until all solutions are assigned a non-dominated rank. The set of solutions with non-dominated rank i is denoted as F. i Then from F i Start by assigning R in ascending order of non-dominant rank. t Individuals selected from the next generation of P t+1 Until the first one makes P t+1 The set F with a number of individuals greater than Nj In F j Select the individuals with the lowest population density to enter P. t+1 , making P t+1 The number of individuals in the sample reaches N.

[0187] Spatial optimization: Based on land use data from 2010 and 2020 for the ecological region and five driving factors (elevation, slope, precipitation, NDVI, and temperature), the Geosos-FLUS model was initialized and trained to produce a land use spatial optimization simulation model with a Kappa coefficient higher than 0.85.

[0188] Furthermore, the basic framework is as follows: based on the neural network algorithm (ANN), the suitability probability of various land use types within the research scope is obtained. Its core is to determine the probability of different land types appearing under the drive of natural and socio-economic factors in the early stage of the research. The number of pixels of each land use type in future years is predicted according to the Markov model. The future land use image is simulated by cellular automata based on the adaptive inertial mechanism. This mechanism can effectively handle the uncertainty and complexity of each land use type under the influence of different driving factors. The GeoSOS-FLUS model coupling set CA has the ability to handle the spatial changes of complex systems and the advantage of Markov in predicting the amount of spatial demand. Therefore, the model has high simulation accuracy and good simulation effect.

[0189] Optimization of the ecological security pattern:

[0190] Based on the optimized land use pattern, an optimized ecological security pattern is constructed, thereby obtaining a regional ecological security pattern optimized by a multi-objective genetic algorithm.

[0191] Step S6 evaluates the optimization effect of the optimized ecological security pattern by comparing the optimized regional ecological security pattern obtained in S5 with the regional ecological security pattern before optimization and calculating the optimization effect.

[0192] Calculate the relative optimization effect η a η b The expressions are as follows:

[0193]

[0194] Where, η a ,η b Representing the optimization effect of ecological source areas and the optimization effect of ecological corridors, respectively, Δr a ,Δr b These represent the changes in the number of source areas and corridors caused by the optimization effect of ecological source areas and the optimization effect of ecological corridors, respectively, where R is the current number of ecological source areas and corridors in the ecological region.

[0195] The above description, in conjunction with specific embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, several simple deductions or substitutions can be made without departing from the concept of the present invention, and all such deductions or substitutions should be considered to fall within the scope of protection defined by the claims submitted herein.

Claims

1. A method for optimizing regional ecological security patterns based on a multi-objective genetic algorithm, characterized in that, Includes the following steps: Step S1: Obtain various geospatial data of the ecological region; Step S2: For ecological regions, based on various geospatial data, obtain the ecological importance, ecological sensitivity, and landscape dominance indices of the ecological regions, and determine the ecological source areas based on the three indices; Step S3: For the ecological region, the modified resistance surface is obtained by weighting and superimposing various resistance factors based on natural conditions and socio-economic activities. Step S4: Based on each ecological source area and resistance surface, determine the ecological corridors and ecological nodes between each ecological source area, and construct an ecological security pattern. Step S5: Based on the multi-objective genetic algorithm, while optimizing the regional ecological security pattern, the two objectives of food security and ecological benefits are taken into account. Step S6: Evaluate the optimization effect of the optimized ecological security pattern; Step S5, based on a multi-objective genetic algorithm, optimizes the regional ecological security pattern while taking into account both food security and ecological benefits; the specific steps are as follows: (1) Setting the objective function: First, the ecological security pattern objective, Max ESP =W ESP ×(A ESI +A ESS +A LDI ) (27) Among them, ESP represents the ecological security pattern objective; W ESP As the weight of the ecological security pattern goal, A ESI As an ecological importance parameter, A ESS A is an ecological sensitivity parameter. LDI For landscape dominance parameters; ESP represents the ecological security pattern objective derived from A. ESI A ESS A LDI It consists of three parameters; Secondly, the goal of food security. Among them, FP represents the food security target; W FP As a weighting of food security goals, NPP x Let NPP be the NPP value of the x-th pixel on cultivated land; P be the total grain output of a certain county / district; NPP sum This represents the total NPP (National Per Capita Preference) on agricultural land in a certain county / district. Finally, ecological benefit goals. Max ESV =W ESV ×(∑(VC i ×A i )) (29) Among them, ESV represents the ecological benefit target; W ESV As a weighting of ecological benefits, VC i Let Ai be the ecological benefit value per unit area of ​​the i-th land use type; Ai is the area of ​​the i-th land use type. (2) Condition constraints: Slopes with a gradient greater than 25° are prohibited from being converted into cultivated land, and slopes with a gradient less than 5° are prohibited from being converted into forest land; the conversion probability of water bodies is 0; condition constraints for each land use type; (3) Quantity constraints: available land area constraints; forest coverage constraints; grassland area constraints; cultivated land area constraints; wetland area constraints; (4) Multi-objective genetic algorithm design: quantitative optimization, based on the NGSA-II algorithm, combined with ecological security pattern objectives, food security objectives and ecological benefit objectives, to perform quantitative optimization of land use types in ecological areas; Spatial optimization: Based on the land use data of the ecological region in 2010 and 2020 and five driving factors of elevation, slope, precipitation, NDVI and temperature, the Geosos-FLUS model was initialized and trained to produce a land use spatial optimization simulation model with a Kappa coefficient higher than 0.

85. (5) Ecological security pattern optimization: Based on the optimized land use pattern, an optimized ecological security pattern is constructed, thereby obtaining the regional ecological security pattern optimized by a multi-objective genetic algorithm. Step S6 evaluates the optimization effect of the optimized ecological security pattern by comparing the optimized regional ecological security pattern obtained in S5 with the unoptimized regional ecological security pattern, calculating the optimization effect, and calculating the relative optimization effect η. a η b The expressions are as follows: Where, η a ,η b Representing the optimization effect of ecological source areas and the optimization effect of ecological corridors, respectively, Δr a ,Δr b These represent the changes in the number of source areas and corridors caused by the optimization effect of ecological source areas and the optimization effect of ecological corridors, respectively, where R is the current number of ecological source areas and corridors in the ecological region.

2. The method for optimizing regional ecological security patterns based on a multi-objective genetic algorithm according to claim 1, characterized in that: The various types of geographic data mentioned in step S1 include multi-source data such as meteorological, soil, hydrological, land use, and remote sensing imagery with the same spatiotemporal scale.

3. The method for optimizing regional ecological security patterns based on a multi-objective genetic algorithm according to claim 1, characterized in that: Step S2 assesses three ecosystem service functions—habitat quality, water conservation, and soil retention—based on multi-source data such as climate and land use, and then weights them equally to obtain an ecological importance index. Based on multi-source remote sensing data, it calculates three sensitivity indices: nitrogen and phosphorus pollution sensitivity, arable land soil erosion sensitivity, and human disturbance sensitivity, and then weights them equally to obtain an ecological sensitivity index. Finally, it uses land use data to estimate the degree of landscape fragmentation, landscape complexity, landscape aggregation connectivity, and landscape heterogeneity, and then weights them equally to obtain a landscape dominance index.

4. The method for optimizing regional ecological security patterns based on a multi-objective genetic algorithm according to claim 1, characterized in that: In step S3, seven resistance factor indicators are selected: land use, altitude, NDVI, nighttime light, distance from road, distance from railway, and distance from water. The resistance factor levels are divided into five categories, and the resistance coefficients are set to 1-5. The weights are superimposed to obtain the ecological resistance surface of the ecological area.

5. The method for optimizing regional ecological security patterns based on a multi-objective genetic algorithm according to claim 1, characterized in that: Step S4, based on the ecological source areas and comprehensive resistance surfaces, and using circuit theory principles, determines the corridors, clamps, and obstacle points between ecological source areas through the following formula, and constructs an ecological security pattern. Where I represents the size of the ecological flow through the corridor; V represents the size of the ecological source measured throughout the corridor; and R represents the cumulative resistance of the corridor.

Images