Method and device for optimizing carbon sink benefit of green land

Abstract

The embodiment of the application provides a green land carbon sink benefit optimization method and device. The method comprises the following steps: obtaining carbon sink data of a plurality of functional partitions in a research area; determining the vegetation carbon sink benefit of each functional partition based on the carbon sink data; calculating the importance percentage of each item in the influence factor of the functional partition on the vegetation carbon sink benefit; representing the degree of influence of the influence factor on the green land carbon sink by the importance percentage, quantifying the carbon sink influence factor, determining the key factor from the plurality of influence factors according to the importance percentage, calculating the local standardized regression coefficient of the key factor to analyze the spatial influence of the key factor on the vegetation carbon sink benefit, and determining the optimization strategy of the green land carbon sink benefit of the functional partition according to the key factor and the analysis result, so that the optimization strategy of the green land carbon sink benefit can be formulated for different functional partitions, and the comprehensive improvement of the carbon sink capacity of urban green land can be realized.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of ecological environment design, and in particular, to a green land carbon sink benefit optimization method and device. BACKGROUND

[0002] Under the premise of global climate change, reducing carbon emissions and increasing carbon have become the focus of international attention.

[0003] Carbon emission reduction and carbon increase are important directions for sustainable development of cities. At present, the related research on carbon sink capacity of green land mainly takes ecological source land with weak human intervention as the main carrier. However, as the only carbon sink carrier in the urban ecological system, the related research on carbon sink capacity of urban green land is relatively less, and the key influencing factors and linear relationship of carbon sink under different urban functional areas are not clear.

[0004] Therefore, how to quantify the influencing factors of urban green land carbon sink and optimize the carbon sink benefit to comprehensively improve the carbon sink capacity of urban green land is a technical problem to be solved at present. SUMMARY

[0005] Therefore, the embodiments of the present application provide a green land carbon sink benefit optimization method and device to at least solve or alleviate the above problems.

[0006] According to an aspect of the embodiments of the present application, a green land carbon sink benefit optimization method is provided, which comprises:

[0007] obtaining a data set related to carbon sink in a research area, the data set comprising data of a plurality of functional subareas, the plurality of functional subareas being obtained by dividing the research area according to urban functions;

[0008] for each functional subarea, determining a vegetation carbon sink benefit of the functional subarea according to the data of the functional subarea; obtaining a set of influencing factors of the functional subarea, the set of influencing factors comprising a plurality of influencing factors, the plurality of influencing factors comprising natural environmental factors and human activity characteristic factors; determining an importance percentage of each influencing factor on the vegetation carbon sink benefit, and determining a key factor from the plurality of influencing factors according to the importance percentage; fitting the key factor and the vegetation carbon sink benefit, determining a local standardized regression coefficient of the key factor according to a fitting result, the local standardized regression coefficient being used to represent a relationship between the influencing factor and the vegetation carbon sink benefit in space; analyzing a spatial influence of the key factor on the vegetation carbon sink benefit according to the local standardized regression coefficient, obtaining an analysis result, the analysis result being used to represent spatial heterogeneity of the vegetation carbon sink benefit; determining an optimization strategy of green land carbon sink benefit of the functional subarea according to the key factor and the analysis result.

[0009] According to another aspect of the embodiments of the present application, there is provided a device for optimizing green land carbon sink benefit, the device comprising:

[0010] an acquisition module configured to acquire a carbon sink related dataset in a study area, the dataset comprising data of a plurality of functional subareas, the plurality of functional subareas being obtained by dividing the study area according to urban functions;

[0011] a processing module configured to, for each of the functional subareas, determine a vegetation carbon sink benefit of the functional subarea according to data of the functional subarea, acquire a set of influence factors of the functional subarea, the set of influence factors comprising a plurality of influence factors, the plurality of influence factors comprising natural environmental factors and human activity characteristic factors, determine a percentage of importance of each of the influence factors to the vegetation carbon sink benefit, and determine a key factor from the plurality of influence factors according to the percentage of importance, fit the key factor and the vegetation carbon sink benefit, determine a local standardized regression coefficient of the key factor according to a fitting result, the local standardized regression coefficient being used to represent a relationship between the influence factor and the vegetation carbon sink benefit in space, analyze a spatial influence of the key factor on the vegetation carbon sink benefit according to the local standardized regression coefficient, and obtain an analysis result, the analysis result being used to represent spatial heterogeneity of the vegetation carbon sink benefit, and determine an optimization strategy of a green land carbon sink benefit of the functional subarea according to the key factor and the analysis result.

[0012] According to another aspect of the embodiments of the present application, there is provided an electronic device, the electronic device comprising a processor, a memory, a communication interface and a communication bus, the processor, the memory and the communication interface completing communication with each other through the communication bus, the memory being used to store at least one executable instruction, the executable instruction being executed by the processor to implement the method for optimizing green land carbon sink benefit provided in the above aspect.

[0013] According to another aspect of the embodiments of the present application, there is provided a computer storage medium, the computer storage medium storing a computer program, the computer program being executed by a processor to implement the method for optimizing green land carbon sink benefit provided in the above aspect.

[0014] According to another aspect of the embodiments of the present application, there is provided a computer program product, the computer program product comprising computer instructions, the computer instructions instructing a computing device to execute the method for optimizing green land carbon sink benefit provided in the above aspect.

[0015] According to the green land carbon sink benefit optimization method provided in the embodiments of the present application, a data set related to carbon sink in a research area is obtained, the data set including data of multiple functional divisions, for each functional division, vegetation carbon sink benefit of the functional division is determined based on data of the functional division; then, a set of influence factors of the functional division is obtained, the influence factors including natural environment factors and human activity characteristic factors, importance percentages of each item in the influence factors to the vegetation carbon sink benefit are determined, the importance percentages are used to represent the degree of influence of the influence factors on the green land carbon sink, the quantification of the influence factors of the green land carbon sink is realized, then, key factors are determined from the multiple influence factors according to the importance percentages, and local standardized regression coefficients of the key factors are calculated, the spatial influence of the key factors on the vegetation carbon sink benefit is analyzed according to the local standardized regression coefficients, and optimization strategies of the green land carbon sink benefit of the functional division are determined according to the key factors and the corresponding analysis results, the optimization strategies of the green land carbon sink benefit are formulated for different functional divisions, the influence factors not only include the natural environment factors, but also consider the human activity characteristic factors, and the comprehensive improvement of the urban green land carbon sink capacity can be realized. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art according to these drawings.

[0017] Figure 1 is a schematic diagram of an exemplary system to which an embodiment of the present application is applied;

[0018] Figure 2 is a flowchart of a green land carbon sink benefit optimization method of an embodiment of the present application;

[0019] Figure 3 is a schematic diagram of an importance percentage of an influence factor to a green land carbon sink benefit of an embodiment of the present application;

[0020] Figure 4 is a marginal effect curve of each influence factor to a green land carbon sink capacity of each functional division of an embodiment of the present application;

[0021] Figure 5 is a schematic diagram of a structural equation model of a spatial variation influence of an urban green land carbon sink capacity of a functional division of an embodiment of the present application;

[0022] Figure 6 is a schematic diagram of a green land carbon sink benefit optimization device of an embodiment of the present application;

[0023] Figure 7 is a schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] The present application is described in the following based on embodiments, but the present application is not limited to these embodiments only. In the following detailed description of the present application, some specific details are thoroughly described for the purpose of complete understanding of the present application by the persons skilled in the art. The present application can also be completely understood without these details. In order to avoid obscuring the essence of the present application, well-known methods, procedures, processes are not described in detail. In addition, the drawings are not necessarily drawn to scale.

[0025] First, some of the nouns or terms appearing in the description of the embodiments of the present application are applicable to the following explanations.

[0026] Carbon sink refers to the ability of plants to absorb carbon dioxide from the atmosphere through photosynthesis and fix it in vegetation and soil. Carbon sink capacity refers to the change in carbon storage in forest carbon pools over a certain period of time. It helps to achieve the goal of reducing greenhouse gas concentrations in the atmosphere and increasing carbon storage. Forests, grasslands, and agricultural ecosystems are important carbon sinks, and play an important role in global climate change by absorbing and storing carbon dioxide.

[0027] Vegetation carbon sink benefit refers to the comprehensive value derived from the absorption of carbon dioxide from the atmosphere by vegetation through photosynthesis and its fixation in biomass (such as tree trunks, branches and leaves) or soil, thereby reducing the concentration of greenhouse gases. It not only includes direct carbon sequestration effects, but also covers ecological, economic and social benefits resulting therefrom. Among them, ecological benefits include climate regulation (such as reducing local temperature), biodiversity protection, water conservation, and soil erosion reduction; economic benefits are achieved through carbon trading (such as forestry carbon trading projects) to achieve economic benefits, or through ecological compensation mechanisms to obtain financial support; social benefits include improving the living environment, raising public environmental awareness, and promoting sustainable development.

[0028] Figure 1 An exemplary computer system is shown. The computer system includes an electronic device 101 and a server 102.

[0029] The electronic device 101 installs and runs an application program, and the application program is suitable for the green land carbon sink benefit optimization method provided by the embodiments of the present application. For example, during the execution of the green land carbon sink benefit optimization method, the relevant parameters are set through the application program.

[0030] The electronic device 101 is connected to the server 102 through a wireless network or a wired network. The server 102 includes at least one of a server, a plurality of servers, a cloud computing platform, and a virtualization center. The server 102 is configured to support the running of an application on the electronic device, that is, to provide a background service for the application running on the electronic device 101. For example, a database is established on the server 102 to store a carbon sink dataset related to a research area and a factor set of each functional subarea. For another example, the server 102 can also perform all or part of the optimization step of the green land carbon sink benefit.

[0031] Optionally, the server 102 undertakes the main computing work, and the electronic device 101 undertakes the secondary computing work; or the server 102 undertakes the secondary computing work, and the electronic device 101 undertakes the main computing work; or the server 102 and the electronic device 101 adopt a distributed computing architecture to perform collaborative computing. In the embodiments of the present application, the electronic device undertakes the main computing work is taken as an example for description.

[0032] Based on the above system, the embodiments of the present application provide a green land carbon sink benefit optimization method. Taking the method executed by the electronic device as an example, the method is described in detail through the following embodiments.

[0033] Figure 2 FIG. 1 is a flowchart of a green land carbon sink benefit optimization method according to an embodiment of the present application. As shown in FIG. 1, the method includes the following steps: Figure 2

[0034] In step 210, a carbon sink dataset related to a research area is obtained, and the dataset includes data of a plurality of functional subareas, which are obtained by dividing the research area according to city functions.

[0035] The database stores a carbon sink dataset related to a research area. When used later, the carbon sink data is obtained from the database, which includes meteorological data, NDVI (Normalized Difference Vegetation Index), land use data, and social and economic data.

[0036] ​The dataset is pre-acquired and stored in a database. For example, the meteorological data can be acquired from multiple ground meteorological stations in the study area and its surrounding areas. For example, data from 30 ground meteorological stations in the study area and its surrounding areas can be acquired. For the meteorological data provided by each meteorological station, the meteorological data of each station is interpolated by using a meteorological interpolation model to generate a spatial distribution of meteorological data matched with each meteorological station, and the spatial distributions of meteorological data corresponding to multiple meteorological stations are combined to obtain a smooth climate data map, which is used to show the spatial distribution of the climate in the study area. In the absence of direct observation data, the estimated value of the location can be calculated by the interpolation model according to the existing measurement data. For example, the above-mentioned meteorological interpolation model can be an ANU surface interpolation (ANU surface interpolation) model based on a spline function.

[0037] NDVI can use 30-meter resolution multispectral images of Landsat 8 and Sentinel-2 satellites of Earth Explorer or Copernicus Open Access Center, and perform radiation correction, atmospheric correction, cloud detection and other processing on the multispectral image to match the target resolution. For example, the target resolution can be 20-meter resolution, or 30-meter resolution, or 35-meter resolution, etc.

[0038] Land use data can use data in a certain period of time, such as land data sets from 1999 to 2019.

[0039] Socioeconomic data can use open source data, such as road data provided by Open Street Map software. The socioeconomic data includes GDP (Gross Domestic Product), population density, and road data. The socioeconomic data needs to be projected and resampled to generate raster data of GDP, population density, and road density of the study area in multiple periods.

[0040] After obtaining the above data related to carbon sinks, the data is normalized to eliminate the scale difference between variables and determine the reasonable influence weight of each variable on the result during model training, thereby improving the stability and accuracy of the model. The data normalization process is as follows: remote sensing data (such as NDVI) is stored as raster data in a specified format, such as GeoTIFF format raster data; non-raster data (such as population density, GDP) is first converted to raster data by interpolation or rasterization. Then, based on the raster data, the minimum value, maximum value, mean value and standard deviation of each variable are determined for modeling analysis of the remote sensing model.

[0041] The research area can be a city, and each city is planned with multiple functional zones. The planning of the functional zones takes into account multiple aspects, such as historical development, geographical factors, economic demand, transportation and infrastructure, and planning strategies. For example, the functional zones of City A include the whole city, the central city, the development zone, and the ecological protection zone. For another example, the functional zones of City B include the central business district, the residential area, the industrial zone, the cultural and educational zone, and the ecological protection zone. The data related to carbon sinks in the data set can be stored in correspondence with the functional zones.

[0042] In step 220, for each functional zone, the vegetation carbon sink benefit of the functional zone is determined according to the data of the functional zone; a set of influence factors of the functional zone is obtained, the set of influence factors includes multiple influence factors, and the multiple influence factors include natural environmental factors and human activity characteristic factors; the importance percentage of each influence factor on the vegetation carbon sink benefit is determined, and a key factor is determined from the multiple influence factors according to the importance percentage; the key factor and the vegetation carbon sink benefit are fitted, and the local standardized regression coefficient of the key factor is determined according to the fitting result; the spatial influence of the key factor on the vegetation carbon sink benefit is analyzed according to the local standardized regression coefficient, and an analysis result is obtained; and the optimization strategy of the green space carbon sink benefit of the functional zone is determined according to the key factor and the analysis result.

[0043] The local standardized regression coefficient is used to represent the relationship between the influence factor and the vegetation carbon sink benefit in space. The analysis result is used to represent the spatial heterogeneity of the vegetation carbon sink benefit.

[0044] a) determining the vegetation carbon sink benefit of the functional zone according to the data of the functional zone, including: determining the NPP (Net Primary Productivity, vegetation net primary productivity) of the functional zone according to the remote sensing data of the functional zone; and determining the vegetation carbon sink benefit of the functional zone according to the NPP of the functional zone.

[0045] For example, for each grid of the functional zone after gridding, the NPP corresponding to the grid is calculated according to the remote sensing data of the grid; the vegetation carbon sink benefit of the grid is determined according to the NPP corresponding to the grid, and finally the vegetation carbon sink benefit on each grid in the entire functional zone is obtained.

[0046] The calculation of NPP can use a remote sensing model, for example, the CASA (Carnegie-Ames-Stanford Approach, process-based terrestrial ecosystem) model to calculate NPP.

[0047] For example, the calculation formula of NPP is:

[0048]

[0049] In the formula, This represents the carbon uptake of vegetation at a single pixel value x and time t (day), expressed in grams of organic carbon per square meter per day (gC / m²). 2 · a). In the formula, Indicates the value of a single cell and time The photosynthetically active radiation absorbed by plants on a single day is measured in megajoules per square meter (MJ / m²) per unit time. 2 / t), Indicates the value of a single cell and time (Daily) solar energy utilization efficiency.

[0050] This application embodiment also uses the medium-resolution imaging spectrometer NPP dataset to verify the accuracy of the calculation results given by the CASA model. For example, based on the above net primary productivity of vegetation, the 500-meter resolution MOD17A3 data is resampled to the target resolution, for example, the 500-meter resolution MOD17A3 data is resampled to 30-meter resolution. Multiple samples (e.g., 1000 samples) are selected within the study area to evaluate the correlation between the MOD17A3 data NPP and the CASA model estimate. Here, vegetation cover data, climate data, and time series data are raster data, and a pixel is the smallest unit of raster data.

[0051] Alternatively, the carbon sequestration benefit of vegetation can be expressed as vegetation carbon sequestration capacity. The formula for calculating vegetation carbon sequestration capacity is:

[0052] ;

[0053] In the formula, the coefficients 1.62 and 0.45 are obtained through the carbon fixation process and carbon conversion efficiency of vegetation. UGCSC represents the vegetation carbon sink capacity, which indicates the average carbon sink per unit time, expressed in grams per square meter per unit time (g / m²). 2 / t).

[0054] b) Obtain the set of impact factors for the functional zones. The set of impact factors includes multiple impact factors, including natural environmental factors and human activity characteristic factors.

[0055] The set of factors affecting the carbon sink benefit of green space in the functional division is established. First, preliminary screening is needed to establish a pool of candidate influencing factors based on multi-source data fusion. The pool of candidate influencing factors includes natural environmental factors and human activity characteristic factors. The natural environmental factors include climate influencing factors and non-climate natural influencing factors. Therefore, the pool of candidate factors mainly includes climate influencing factors, non-climate natural influencing factors, and human activity characteristic factors. Other categories of influencing factors, such as landscape factors, can also be included. Based on literature research and domain knowledge, a dynamic weight screening algorithm (DWSA) is used to screen factors that are theoretically strongly related to the carbon sink benefit of the functional division. For example, based on the weights of literature, data quality scores, and acquisition costs, the dynamic factor weights are calculated to screen factors that are theoretically strongly related to the carbon sink benefit. Through data cleaning, interpolation, and normalization processing, the temporal and spatial resolution of each factor is ensured to match the remote sensing data such as NDVI. The correlation between each factor and urban carbon sink capacity is evaluated through correlation analysis, and factors with low correlation or redundancy are removed to construct the set of influencing factors.

[0056] For example, a plurality of candidate influencing factors of the functional division are obtained. The literature support, data quality score, and acquisition cost corresponding to each candidate influencing factor are obtained. The dynamic factor weight of each candidate influencing factor is calculated, and the calculation formula of the dynamic factor weight is as follows: The candidate influencing factors with a dynamic factor weight greater than a weight threshold are added to the set of influencing factors of the functional division. Wherein, W DF is the dynamic factor weight, LS is the literature support, R DQ is the data quality score, and AC is the acquisition cost.

[0057] Optionally, the plurality of influencing factors include at least part of the following: annual average temperature, annual average precipitation, solar radiation; slope, elevation; population density, gross domestic product, human footprint index, green coverage rate, nighttime light index; landscape connectivity index, landscape shape index, landscape patch density.

[0058] Among them, the annual average temperature, the annual average precipitation, and the solar radiation belong to the climate influencing factors; the slope and the elevation belong to the non-climate influencing factors; the population density, the gross domestic product, the human footprint index, and the green coverage rate belong to the human activity characteristic factors; the landscape connectivity index (CONTAG), the landscape shape index (LSI), and the landscape patch density belong to the landscape indicators, i.e., the landscape factors.

[0059] The calculation formulas (based on the Anusplin interpolation model) of the monthly average temperature (TEM) and the monthly precipitation (PRE) are as follows:

[0060] ;

[0061] ;

[0062] wherein, represents the longitude of the location of the pixel, represents the latitude of the location of the pixel, represents the monthly average temperature of the pixel obtained by interpolation, represents the monthly average precipitation of the pixel obtained by interpolation, and the meteorological monthly temperature and meteorological monthly precipitation observation values are expanded to each pixel of the research area using the ANUsplin interpolation model.

[0063] The average value SR of daily solar radiation is calculated by the following formula:

[0064] ;

[0065] wherein, represents the solar radiation intensity on the jth day, and the unit is watt per square meter (W / m 2 ).

[0066] The slope calculation formula (based on digital elevation model data) is:

[0067] ;

[0068] wherein, Slope is the slope, z is the elevation value, / and / respectively represent the elevation change rate along the X and Y directions. The elevation value corresponding to each pixel is extracted by digital elevation model (DEM), and the elevation value on each pixel represents the altitude of the location.

[0069] The population density (PD) calculation formula is:

[0070] ;

[0071] wherein, Population is the population quantity of a single pixel, and Area is the area of a single pixel region, and the unit is square kilometer (km 2 ).

[0072] The GDP calculation formula (combined with remote sensing geographic boundary data distributed to grid cells) is:

[0073] ; ​

[0074] In the formula, For the GDP of the i-th single pixel, Let be the area of ​​the i-th single pixel. The total area of ​​a single pixel. It represents the total GDP.

[0075] The formula for calculating Land Use Functions (LUF) is as follows:

[0076] ;

[0077] In this formula, m represents the total number of land use types in the landscape. For the frequency or intensity of land use change in category J, The area represents the pixel area of ​​land use category J, and Area represents the total pixel area. The land use change frequency (such as the number of times farmland is converted to building land) is extracted from multi-time period remote sensing classification data.

[0078] The formula for calculating the nighttime light index is:

[0079] ;

[0080] In the formula, Let be the nighttime light intensity value of pixel 𝑖, and n be the total number of pixels in the study area. The nighttime light index can directly reflect the spatial distribution and intensity of human activities.

[0081] The Human Footprint Index is calculated based on the aforementioned nighttime light intensity index, population density, GDP, and land use intensity, as shown below. The calculation formula is:

[0082] ;

[0083] In the formula, NLI is the nighttime light index, GDP is the gross domestic product per unit area, and LUF is the land use intensity. , , , This is the weighting factor.

[0084] Green coverage rate The calculation formula is:

[0085] ;

[0086] In the formula, Green Area is the sum of the areas of pixels with NDVI higher than the set threshold, and Total Area is the total pixel area.

[0087] Connectivity Index The connectivity between different types of patches in the landscape is measured by the following formula:

[0088] ;

[0089] wherein, is the boundary frequency between the types and , and m is the total number of land use types in the landscape.

[0090] The landscape shape index LPI is calculated by the following formula:

[0091] ;

[0092] wherein, represents the area of the largest patch in the landscape, represents the total area of the entire study area.

[0093] The patch density PD is calculated by the following formula:

[0094] ;

[0095] wherein, N is the total number of patches, and A is the total area of the landscape.

[0096] The above landscape indexes can be obtained by importing the NDVI and land use data into the landscape pattern analysis software Fragstats and calculating by the software.

[0097] c) determining the importance percentage of each influencing factor on the vegetation carbon sink benefit, and determining the key factor from the multiple influencing factors according to the importance percentage.

[0098] Optionally, the multiple influencing factors are sorted according to the importance percentage, and the key factor meeting the screening condition is determined according to the sorting; for example, the multiple influencing factors are sorted in order from high to low according to the importance percentage, and the influencing factors ranked in the first S positions are determined as the key factors, S is a positive integer greater than 1. Alternatively, the influencing factors with an importance percentage greater than or equal to a percentage threshold value can also be determined as the key factors.

[0099] Optionally, the importance percentage of each influencing factor on the vegetation carbon sink benefit is calculated by a boosted regression tree (BRT) model, and then the key factor is determined from the multiple influencing factors according to the importance percentage. For example, the vegetation carbon sink benefit is input into the boosted regression tree model, and the importance percentage of each influencing factor on the vegetation carbon sink benefit is output by the boosted regression tree model.

[0100] Before using the enhanced regression tree model, firstly, the carbon sink capacity is taken as the dependent variable, and the multiple influence factors are taken as the independent variables to construct the enhanced regression tree model; the vegetation carbon sink benefit of each grid after the functional division is rasterized as a characteristic variable, the characteristic variable is input into the enhanced regression tree model, the enhanced regression tree model is optimized through cross-validation, and the optimized enhanced regression tree model is obtained. Specifically, the process of determining the key factors affecting the carbon sink capacity includes: taking the vegetation carbon sink benefit (such as UGCSC) of the whole region, central city, development zone and ecological protection zone as the dependent variable, and taking the slope, altitude, temperature, rainfall, solar radiation, human footprint index, green coverage rate, landscape shape index, connectivity index and patch density as the independent variable, an enhanced regression tree model is established, and the model fitting can be realized by calling the “caret”, “gbm” and “dismo” packages in R4.1.0 software. For the optimization training of the model, a self-supervised learning training method is adopted.

[0101] Then, based on the enhanced regression tree model (or the optimized enhanced regression tree model), the number of times of each influence factor as a branch node in the decision tree is counted, and the error reduction amount of the model in predicting each influence factor as a branch node is calculated; the number of times of each influence factor as a branch node and the corresponding error reduction amount are weighted and summed, and the weighted sum is normalized to the importance percentage corresponding to each influence factor.

[0102] Finally, the importance percentage can be directly output by the enhanced regression tree model as the key factor ranking result affecting the vegetation carbon sink benefit. For example, the key factor ranking result is screened and output by the enhanced regression tree model. After the weighted sum, the importance value is obtained, and after the normalization processing, the importance percentage is obtained. The calculation formula of the normalization is:

[0103] ;

[0104] the importance percentage of the qth factor, the original importance value of the qth factor, the original importance value of the kth factor, and Q is the total number of factors.

[0105] After obtaining the importance percentage, dynamic weight correction is also performed, and the calculation formula of the correction is:

[0106] =0.8× +0.2× ;

[0107] wherein, represents the updated importance value of the qth factor in this round of iteration, represents the importance value of the qth factor in the last round of iteration, The historical importance values ​​of each factor in the previous iteration. Indicates the first The importance values ​​of each factor are newly obtained in the current calculation.

[0108] like Figure 3 As shown, the percentage (%) of importance of the following climate change and human activity variables to the urban green space carbon sink capacity (UGCSC) of different functional zones in the study area was calculated based on the enhanced regression tree model. Note: BJ refers to the entire area; CUZ refers to the central urban area; DEZ refers to the development zone; ECZ refers to the ecological protection zone.

[0109] Based on the regression tree model, the following results were obtained: In the entire study area of ​​City A (BJ), slope was the most influential factor affecting carbon sink benefits, accounting for 78.9% of the importance; in the Ecological Reserve Zone (ECZ), slope was the most influential factor, accounting for 73.8% of the importance; in the Development Zone (DEZ), altitude was the most influential factor, accounting for 79.9% of the importance; and in the Central Urban Area (CUZ), temperature was the most influential factor, accounting for 45.7% of the importance. These percentages of importance clearly identify the key dominant factors, guiding urban green space construction. The key dominant factor is the one among the key factors that has the most significant impact on carbon sink benefits.

[0110] For example, the aforementioned key factors could be influencing factors where multicollinearity does not exist.

[0111] d) Fit the key factors to the vegetation carbon sink benefits, and determine the local standardized regression coefficients of the key factors based on the fitting results.

[0112] Optionally, the vegetation carbon sink benefit is weighted by a Gaussian kernel function to construct a spatial weight matrix of the vegetation carbon sink benefit, the spatial weight matrix being used to measure spatial similarity and influence degree between different geographical units, the geographical unit being a grid after functional zoning and rasterization; Akaike Information Criterion (AIC) or Bayesian Information Criterion (BIC) is used to calculate an optimal bandwidth value of each key factor, the optimal bandwidth value being used to optimize regression fitting of the key factor to the vegetation carbon sink benefit; based on the spatial weight matrix and the optimal bandwidth value of each key factor, a Multi-scale Geographically Weighted Regression (MGWR) model is used to fit the key factor to the vegetation carbon sink benefit to obtain a fitting result corresponding to each key factor; for each key factor, a local regression coefficient corresponding to each grid is extracted from the fitting result, and the local regression coefficient is standardized to obtain a local standardized regression coefficient of the key factor.

[0113] The spatial weight matrix of the vegetation carbon sink benefit influences the identification of spatial heterogeneity between variables and fitting effect of the Multi-scale Geographically Weighted Regression model, and the spatial weight matrix can be used to test the fitting effect. The optimal bandwidth value of each influencing factor can optimize regression fitting of the influencing factor to the vegetation carbon sink benefit. Therefore, before fitting the key factor to the vegetation carbon sink benefit, the spatial data of the vegetation carbon sink benefit is weighted by a Gaussian kernel function to construct a spatial weight matrix, and AIC or BIC is used as a bandwidth selection basis to determine the optimal bandwidth value of each key factor, so as to ensure that the MGWR model can reflect spatial heterogeneity and scale difference between variables.

[0114] Subsequently, for each key factor, the MGWR model is used to fit and analyze the vegetation carbon sink benefit and the key factor based on the spatial weight matrix and the optimal bandwidth value of the key factor to obtain a fitting result of the key factor; from the fitting result of the key factor, a local regression coefficient corresponding to each grid is extracted, and the local regression coefficient is standardized to obtain a local standardized regression coefficient of the key factor in different grids.

[0115] e) analyzing spatial influence of the key factor on the vegetation carbon sink benefit according to the local standardized regression coefficient to obtain an analysis result.

[0116] For each key factor, based on the local standardized regression coefficient of the key factor, the spatial heterogeneity of the key factor on the vegetation carbon sink benefit is identified, the identification result is obtained, and the spatial quantitative evaluation of the influence strength of the key factor is completed, and the evaluation result is obtained. The identification result and the evaluation result are included in the analysis result.

[0117] In steps d) and e), the spatial influence of the key factor on the vegetation carbon sink benefit is determined by using the MGWR model, the spatial heterogeneity of the key factor on the vegetation carbon sink benefit is identified, and the spatial quantitative evaluation of the influence strength of each factor is completed.

[0118] Optionally, first, vegetation carbon sink benefit spatial data and spatial distribution data of a plurality of potential influencing factors in a target region are acquired. A Gaussian kernel function is selected as a spatial weighting function to perform spatial weighting processing on the carbon sink benefit data. A spatial weight matrix is constructed according to the spatial position relationship, to measure the spatial similarity and potential spatial influence degree between any two geographical units. The expression of the spatial weight matrix is as follows:

[0119] ;

[0120] wherein, is the spatial weight between unit and unit , is the spatial distance between unit and unit , is a bandwidth parameter.

[0121] To determine the optimal bandwidth value of each key factor, AIC or BIC is used for bandwidth optimization. For the spatial response scale of different variables, the optimal bandwidth value is determined by traversing different bandwidth values to minimize the AIC / BIC value, so that the model can reflect the scale difference and heterogeneity characteristics of the variables in space.

[0122] Based on the constructed spatial weight matrix and the optimal bandwidth value of each key factor, an MGWR model is constructed to fit and analyze the relationship between the vegetation carbon sink benefit and the plurality of key factors. The form of the MGWR model is as follows:

[0123] ;

[0124] wherein, is the vegetation carbon sink benefit value of unit α, is a local intercept term at the spatial position , which represents the basic value or background level of the vegetation carbon sink benefit when all the key factors in the study region are zero at the position of the αth unit, the value of the kth key factor unit , represents the regression coefficient of the key factor k at position , with a bandwidth of , , represents a certain position of the unit , is the corresponding residual term of the unit , and Q is the total number of key factors.

[0125] According to the fitting results of the model, the local regression coefficients of each key factor in each geographical grid unit are extracted. In order to achieve comparability and visual analysis, all coefficients are standardized by using the following formula:

[0126] ;

[0127] wherein and are the mean and standard deviation of all local regression coefficients of the key factor k, represents the local regression coefficient of the key factor k at position , , which is the local regression coefficient of the unit at position after standardization, that is, the local standardized regression coefficient of the unit at position .

[0128] Finally, according to the spatial distribution of the local standardized regression coefficient, the influence intensity of the key factor on the carbon sink benefit of vegetation in different regions is identified. This process can be further deepened through spatial visualization maps, cluster analysis or local significance test, etc., so as to support the subsequent optimization of green space and the formulation of natural-based solution strategies.

[0129] In the embodiments of the present application, the following results are obtained based on the MGWR calculation: the local standardized regression coefficient of the carbon sink benefit of vegetation is calculated by the multi-scale geographical weighted regression model, and the spatial heterogeneity of the independent variable (i.e. the key factor) on the dependent variable (i.e. the carbon sink benefit of vegetation) is quantified. Based on this method, the spatial heterogeneity influence of each key factor on the carbon sink capacity of urban green space is scientifically identified and quantified, which provides a scientific basis and technical support for optimizing the layout of urban green infrastructure and improving the carbon sink benefit.

[0130] f) determining the optimization strategy of the carbon sink benefit of green space in the functional division according to the key factors and the analysis results.

[0131] Optionally, the change trend between the key factor and the vegetation carbon sink benefit is analyzed, a range in which the key factor has a significant influence on the vegetation carbon sink benefit is determined, and a threshold range of the key factor is obtained; a structural equation model is constructed according to the key factor and the threshold range of the key factor, and an influence path and an action strength of the key factor on the vegetation carbon sink benefit are generated through the structural equation model; and an optimization strategy of the carbon sink benefit of the functional division is determined according to the influence path, the action strength and the analysis result.

[0132] For example, a key dominant factor is determined from the key factors, for example, a key factor with the largest importance percentage is determined as the dominant factor; a change trend between the key dominant factor and the vegetation carbon sink benefit is analyzed, a range in which the key dominant factor has a significant influence on the vegetation carbon sink benefit is determined, and a threshold range of the key dominant factor is obtained; a structural equation model is constructed according to the key dominant factor and the threshold range of the key dominant factor, and an influence path and an action strength of the key dominant factor on the vegetation carbon sink benefit are generated through the structural equation model; and an optimization strategy of the carbon sink benefit of the functional division is determined according to the influence path, the action strength and the analysis result.

[0133] For example, a change trend between the key factor and the vegetation carbon sink benefit is analyzed according to a partial dependence plot between the key factor and the vegetation carbon sink benefit. Specifically, based on the existing results, a partial dependence plot (PDP) in the BRT model is used to analyze the marginal effect of the climate change and the human activity key indicators (i.e., the human activity characteristic factors) on the green land carbon sink capacity of the functional division. The target of the partial dependence plot is to show the marginal influence of a single factor (or a group of factors) on the model prediction result. By fixing other factors and gradually changing the value of the target factor, the change of the model prediction value (such as the carbon sink capacity) is observed, so as to analyze the action of the target factor.

[0134] For the target factor and the model prediction , the partial dependence value is calculated as:

[0135] ;

[0136] In the formula, the target factor is the kth key factor, indicates the value of the other factors, is a prediction function of the model, and S is the total number of the key factors. The partial dependence plot shows the mean trend of the model prediction value when the value of the target factor changes. The partial dependence value here can be used to represent the marginal effect of the climate change and the human activity key indicators on the green land carbon sink capacity of the functional division corresponding to the target key dominant factor.

[0137] Based on partial dependency plots, the changing trends between key dominant factors and vegetation carbon sequestration benefits in each functional zone within the study area were determined. The significance intervals and critical points of diminishing marginal effects of each key dominant factor influencing carbon sequestration capacity were estimated. Taking carbon sequestration capacity as an example to represent carbon sequestration benefits, such as... Figure 4 As shown, the graph illustrates the changing trends of key factors and vegetation carbon sequestration capacity across the entire city of A. For the "slope" within the 0-30° range, values ​​of 0°, 5°, 10°, etc., are set, while the values ​​of other factors remain unchanged. The partial dependency plot shows that the marginal effect of slope on carbon sequestration capacity is significant within the 0-13.5° range. When the slope reaches 13.5°, the incremental effect tends to plateau, and when the slope is between 13.5° and 30°, the marginal effect weakens. Therefore, this embodiment of the application determines the optimal slope threshold to be 13.5°. Based on this method, the threshold range of key dominant factors is scientifically determined, providing theoretical support and optimization basis for improving urban green space carbon sequestration capacity and guiding green space construction.

[0138] Optionally, the aforementioned PDP is output by an enhanced regression tree model. For example, vegetation carbon sequestration benefits are input into the enhanced regression tree model, which outputs a partial dependency graph between influencing factors and vegetation carbon sequestration benefits. Then, based on the partial dependency graph between key factors and vegetation carbon sequestration benefits, the changing trends between key factors and vegetation carbon sequestration benefits are analyzed.

[0139] Subsequently, with the goal of increasing carbon sequestration and reducing carbon emissions, a structural equation model was constructed by combining key dominant factors and their threshold ranges, and optimization strategies were proposed for the carbon sequestration benefits of different functional zones within the study area.

[0140] Alternatively, the structural equation model is:

[0141] ;

[0142] in, Let T be the coefficient matrix of the interaction paths between endogenous latent variables, and let T be the coefficient matrix of the influence paths between exogenous and endogenous latent variables. For random interference terms, As an endogenous latent variable, The exogenous latent variable is the vegetation carbon sequestration benefit, while the endogenous latent variable is the key factor.

[0143] To investigate the integrated mechanisms of UGCSC's interaction with climate change, human activities, and landscape patterns, a structural equation model (SEM) framework can be constructed using the SPSS plugin AMOS. SEM uses a system of linear equations to represent the relationship between measured variables and latent variables. Latent variables cannot be directly observed and must be estimated using the measured variables.

[0144] The structural equation model-measurement model is as follows:

[0145] ;

[0146] ;

[0147] In the structural equation model, X and Y are the exogenous and endogenous measurement variables, respectively; , These are exogenous latent variables and endogenous latent variables, respectively. for exist Factor loading matrix on; for exist Factor loading matrix on; , To account for measurement error. In the embodiments of this application, the exogenous latent variables are climate factors (average annual temperature, precipitation, solar radiation), topographic factors (slope, altitude), human activity characteristic factors (green coverage, nighttime light index), landscape factors (connectivity index (CONTAG), landscape shape index (LSI), and patch density (PD)), and the endogenous latent variable is carbon sink capacity UGCSC.

[0148] Evaluation indicators where the measured variables are latent variables include slope, altitude, temperature, solar radiation, rainfall, human footprint index, green coverage, landscape shape index, CONTAG, and patch density. For example, latent variables... (Climate factors) are measured variables (Average annual temperature) (Average annual precipitation) (Solar radiation) is reflected in the model. The measurement model associates existing manifest variables (such as climate factors, topographic factors, and human activity characteristics) with their corresponding latent variables. Climate factors, as exogenous latent variables, are measured by manifest variables such as annual mean temperature, precipitation, and solar radiation; topographic factors, as exogenous latent variables, are measured by manifest variables such as slope and altitude; and human activity characteristics, as exogenous latent variables, are measured by manifest variables such as green space coverage and nighttime light index. Carbon sequestration capacity, as an endogenous latent variable, is characterized by manifest variables such as vegetation cover, normalized difference vegetation index (NDVI), and net primary productivity (NPP). Based on theory and research questions, the causal paths of the latent variables are defined using structural equation modeling.

[0149] Structural Equation Model - The structural model is:

[0150] ;

[0151] Structural Equation Modeling - In Structural Modeling is an action path coefficient matrix between endogenous latent variables; is an influence path coefficient matrix of exogenous latent variables and endogenous latent variables. is a random disturbance term. In the embodiment of the present application, the latent variables are terrain, climate, human activity, landscape pattern, etc. In the structural model, the path relationship of the exogenous latent variables to the endogenous latent variables is defined to reveal the influence path and action strength of the climate factor, the terrain factor and the human activity characteristic factor on the carbon sink capacity. For example, the climate factor quantifies the marginal effect of the annual average temperature and the precipitation change on the carbon sink capacity through its path coefficient; the human activity characteristic factor quantifies the comprehensive action of the green coverage rate and the night light index on the carbon sink capacity through the path coefficient.

[0152] The simulation result is tested by using the standardized residual, the standardized root mean square residual (SRMR), the root mean square error of approximation (RMSEA), the chi-square degree of freedom ratio (Chi-square / df), the goodness of fit index (GFI) and the comparative fit index (CFI). Generally, when Chi-square / df<3, P<0.05, RMSEA<0.05, SRMR<0.08, GFI>0.95 and CFI>0.95, the model fitting is good. When the goodness of fit index of the structural model of the present application meets the requirement, the path relationship between the terrain, the climate, the human activity, the landscape pattern and the UGCSC is obtained, the main influence factors (i.e. the key dominant factors) of the UGCSC change are determined according to the significance and the size of the standardized path coefficient, the influence direction of each key factor on the UGCSC change, i.e. the positive influence and the negative influence, is determined according to the sign of the standardized path coefficient, the influence mechanism of each key factor on the UGCSC change is obtained according to the path relationship from the independent variable to the dependent variable, so that the direct influence and the indirect influence of each influence factor on the UGCSC are determined, and the total influence of each influence factor on the UGCSC is obtained. Based on the total influence of each influence factor on the UGCSC, the targeted carbon sink benefit optimization strategy can be proposed for different functional divisions of the research region.

[0153] ​In this embodiment, SPSS (Statistical Product and Service Solutions) plug-in AMOS (Analyze of Moment Structures) is used to construct three SEM structural equation model frameworks for different functional areas in A City, i.e., a structural equation model of the relationship between the carbon sink capacity variation coefficient of urban green land in the central city and its driving factors, a structural equation model of the relationship between the carbon sink capacity of urban green land in the development zone and its driving factors, and a structural equation model of the relationship between the carbon sink capacity of urban green land in the ecological protection zone and its driving factors. Each functional area corresponds to a respective structural equation model.

[0154] In this embodiment, the structural equation model (SEM) is used to calculate the following results:

[0155] The structural equation model is used to calculate the determination coefficient (R 2 ) of the carbon sink capacity, and the total explanatory ability of the exogenous latent variable (climatic factor, topographic factor, and human activity characteristic factor) to the endogenous latent variable carbon sink capacity is quantified. In this embodiment, it is shown that the exogenous latent variable defined in the model can explain 69% of the spatial variation coefficient of the carbon sink capacity of urban green land in the central city of A City, 86% of the spatial variation coefficient of the carbon sink capacity of urban green land in the development zone of A City, and 83% of the spatial variation coefficient of the carbon sink capacity of urban green land in the ecological protection zone of A City.

[0156] The structural equation model of the relationship between the carbon sink capacity variation coefficient of urban green land in the central city of A City and its driving factors is taken as an example for illustration:

[0157] Among the variables, slope, altitude, temperature, and solar radiation showed highly significant positive correlations with carbon sink capacity (p < 0.05). The exogenous latent variable topography, composed of altitude and slope, had the greatest impact on carbon sink capacity (standardized path total coefficient was 0.61, direct impact 0.61, indirect impact 0.10). Rainfall, human footprint index, and connectivity index also showed highly significant positive correlations with carbon sink capacity (p < 0.05). Although the measured variables green space coverage, landscape shape index, and patch density showed significant negative correlations with their latent variables (p < 0.05), the exogenous latent variables human activity characteristics and landscape model factors showed highly significant negative correlations with carbon sink capacity (p < 0.05, direct impact of human activity characteristics -0.23, indirect impact -0.36; direct impact of landscape model factors -0.16, indirect impact -0.04). Therefore, green space coverage, landscape shape index, patch density, and carbon sink capacity showed significant positive correlations (p < 0.05). Therefore, in the central urban area of ​​City A, the optimization of urban green space can be guided by setting a topographic factor at an optimization threshold. Two key points of this invention are that, through quantitative analysis of the influencing factor set, it can be found that some measured variables are negatively correlated with exogenous latent variables, yet still contribute to urban carbon sink capacity. For example, the higher the green coverage rate, the more restricted the space for human activities, thus inhibiting carbon sink capacity. However, due to the interaction between green coverage rate and human activities, green coverage rate and carbon sink capacity are positively correlated; the higher the green coverage rate, the better the carbon sink benefit.

[0158] In summary, the optimization strategies and their basis for the three structural models are as follows:

[0159] Based on the training results of the model for the central urban area of ​​City A, such as Figure 5 As shown (the numbers next to the arrows are the standardized path coefficients), in the model of the central urban area of ​​City A, the topographic factor has the greatest impact on carbon sink capacity (the total standardized path coefficient is 0.61, with a direct impact of 0.61 and an indirect impact of 0.10). The optimization strategy is to strengthen the control of the optimization thresholds for slope and altitude in the topographic shaping process. The slope optimization threshold for urban green space construction in the central urban area of ​​City A is 15.5°, and the altitude optimization threshold is 250 meters (m).

[0160] Based on the model training results of City A Development Zone, topographic factors have the greatest impact on carbon sink capacity (the total coefficient of the standardized path is 0.99, with a direct impact of 0.99 and an indirect impact of 0.17). The optimization strategy is to strengthen the control of the optimization thresholds for slope and altitude in topographic shaping. The optimization threshold for slope in urban green space construction in City A Development Zone is 9.7°, and the optimization threshold for altitude is 1205.7m.

[0161] Based on the model training results of the ecological protection zone in City A, the landscape model factor has the greatest impact on carbon sink capacity (standardized path total coefficient is 0.26, direct impact is 0.26, and indirect impact is 0.32), and the optimization strategy is to improve the connectivity index, and the optimization threshold of the connectivity index is 71.5.

[0162] In summary, the green space carbon sink benefit optimization method provided in the embodiment obtains a data set related to carbon sink in a research area, the data set includes data of multiple functional divisions, for each functional division, determines the vegetation carbon sink benefit of the functional division based on the data of the functional division; then, obtain a set of influence factors of the functional division, which includes natural environment factors and human activity characteristic factors in the set of influence factors, determine the importance percentage of each item in the influence factors to the vegetation carbon sink benefit, express the degree of influence of the influence factors on the green space carbon sink through the importance percentage, realize the quantification of the influence factors of the green space carbon sink, and then determine the key factors from the multiple influence factors according to the importance percentage, and calculate the local standardized regression coefficient of the key factors, analyze the spatial influence of the key factors on the vegetation carbon sink benefit according to the local standardized regression coefficient, and determine the optimization strategy of the green space carbon sink benefit of the functional division according to the key factors and the corresponding analysis results, formulate the optimization strategy of the green space carbon sink benefit for different functional divisions, and the influence factors not only include natural environment factors, but also consider human activity characteristic factors, which can realize the comprehensive improvement of the urban green space carbon sink capacity.

[0163] Corresponding to the method embodiment, Figure 6 A schematic diagram of a green space carbon sink benefit optimization device is shown. As Figure 6 shown, the green space carbon sink benefit optimization device includes:

[0164] The acquisition module 401 is configured to acquire a data set related to carbon sink in a research area, and the data set includes data of multiple functional divisions, and the multiple functional divisions are obtained by dividing the research area according to city functions.

[0165] The processing module 402 is configured to determine, for each functional partition, vegetation carbon sink benefits of the functional partition according to data of the functional partition; obtain a set of influence factors of the functional partition, the set of influence factors including a plurality of influence factors, the plurality of influence factors including natural environmental factors and human activity characteristic factors; determine importance percentages of the influence factors on the vegetation carbon sink benefits, and determine a key factor from the plurality of influence factors according to the importance percentages; fit the key factor and the vegetation carbon sink benefits, and determine a local standardized regression coefficient of the key factor according to a fitting result, the local standardized regression coefficient being used to represent a relationship between the key factor and the vegetation carbon sink benefits in space; analyze spatial influence of the key factor on the vegetation carbon sink benefits according to the local standardized regression coefficient, and obtain an analysis result, the analysis result being used to represent spatial heterogeneity of the vegetation carbon sink benefits; and determine an optimization strategy of green space carbon sink benefits of the functional partition according to the key factor and the analysis result.

[0166] In some embodiments, the processing module 402 determines the optimization strategy of the carbon sink benefits of the functional partition according to the key factor, including: analyzing a change trend between the key factor and the vegetation carbon sink benefits, determining an interval in which the key factor has a significant influence on the vegetation carbon sink benefits, and obtaining a threshold range of the key factor; constructing a structural equation model according to the key factor and the threshold range, and generating an influence path and an action strength of the key factor on the vegetation carbon sink benefits through the structural equation model; and determining the optimization strategy of the carbon sink benefits of the functional partition according to the influence path, the action strength, and the analysis result.

[0167] In some embodiments, the structural equation model is: ; wherein, is a path coefficient matrix between endogenous latent variables, T is an influence path coefficient matrix of exogenous latent variables and endogenous latent variables, is a random disturbance term, is an endogenous latent variable, is an exogenous latent variable, the endogenous latent variable is the vegetation carbon sink benefits, and the exogenous latent variable is the key factor.

[0168] In some embodiments, the processing module 402 analyzes the change trend between the key factor and the vegetation carbon sink benefits, including: inputting the vegetation carbon sink benefits into an enhanced regression tree model, outputting a partial dependence plot between the influence factor and the vegetation carbon sink benefits through the enhanced regression tree model; and analyzing the change trend between the key factor and the vegetation carbon sink benefits according to the partial dependence plot between the key factor and the vegetation carbon sink benefits.

[0169] In some embodiments, the processing module 402 determines the importance percentage of each impact factor on the vegetation carbon sink benefit, and determines the key factor from the plurality of impact factors according to the importance percentage, including: inputting the vegetation carbon sink benefit into the enhanced regression tree model, outputting the importance percentage of each impact factor on the vegetation carbon sink benefit through the enhanced regression tree model; ranking the plurality of impact factors according to the importance percentage, and determining the key factor meeting the screening condition according to the ranking.

[0170] In some embodiments, the processing module 402 is further configured to construct an enhanced regression tree model by taking the carbon sink capacity as the dependent variable and the plurality of impact factors as the independent variable; determine the vegetation carbon sink benefit of each grid after gridding the functional zoning as a characteristic variable, input the characteristic variable into the enhanced regression tree model, and optimize the enhanced regression tree model through cross-validation to obtain an optimized enhanced regression tree model.

[0171] In some embodiments, the processing module 402 outputs the importance percentage of each impact factor on the vegetation carbon sink benefit through the enhanced regression tree model, including: based on the enhanced regression tree model, counting the number of times each impact factor is used as a branch node in the decision tree, and calculating the error reduction amount of the model predicting each impact factor as a branch node; weighted sum the number of times each impact factor is used as a branch node and the corresponding error reduction amount, and normalize the weighted sum to the importance percentage corresponding to each impact factor.

[0172] In some embodiments, the processing module 402 obtains the impact factor set of the functional zoning, including: obtaining a plurality of candidate impact factors of the functional zoning; obtaining the literature support degree, data quality score and acquisition cost corresponding to each candidate impact factor; calculating the dynamic factor weight of each candidate impact factor, and the calculation formula of the dynamic factor weight W is as follows: ; candidate impact factors with a dynamic factor weight greater than a weight threshold are added to the impact factor set of the functional zoning; wherein, W DF is the dynamic factor weight, LS is the literature support degree, R DQ is the data quality score, and AC is the acquisition cost.

[0173] In some embodiments, the plurality of impact factors include at least part of: annual average temperature, annual average precipitation, solar radiation; slope, altitude; population density, gross domestic product, human footprint index, green coverage rate; landscape connectivity index, landscape shape index, landscape patch density.

[0174] In some embodiments, the processing module 402 fits the multiple impact factors and the vegetation carbon sink benefit, determines the local standardized regression coefficient of the key factor according to the fitting result, including: performing weighted processing on the vegetation carbon sink benefit by a Gaussian kernel function, constructing a spatial weight matrix of the vegetation carbon sink benefit, the spatial weight matrix is used to measure the spatial similarity and the influence degree between different geographical units, the geographical unit refers to a grid after the functional partition is rasterized; adopting Akaike information criterion or Bayesian information criterion, calculating an optimal bandwidth value of each key factor, the optimal bandwidth value is used to optimize the regression fitting of the key factor on the vegetation carbon sink benefit; based on the spatial weight matrix and the optimal bandwidth value of each key factor, fitting the key factor and the vegetation carbon sink benefit by a multi-scale geographical weighted regression model, obtaining the fitting result corresponding to each key factor; for each key factor, extracting the local regression coefficient corresponding to each grid according to the fitting result, and performing standardized processing on the local regression coefficient, obtaining the local standardized regression coefficient of the key factor.

[0175] It should be noted that the green space carbon sink benefit optimization device of the embodiment is used to implement the corresponding green space carbon sink benefit optimization method in the foregoing method embodiment, and has the beneficial effects of the corresponding method embodiment, which will not be described here.

[0176] Figure 7 is a schematic block diagram of an electronic device provided by an embodiment of the present application. Embodiments of the present application do not limit the specific implementation of the electronic device. For example, the electronic device can be a server or a terminal. As shown in Figure 7 The electronic device can include a processor 502, a communications interface 504, a memory 506, and a communications bus 508. Among them:

[0177] The processor 502, the communications interface 504, and the memory 506 complete mutual communication through the communications bus 508.

[0178] The communications interface 504 is used to communicate with other electronic devices or servers.

[0179] The processor 502 is used to execute the program 510, and specifically can execute the related steps in any of the foregoing green space carbon sink benefit optimization method embodiments.

[0180] Specifically, the program 510 can include program code, and the program code includes computer operation instructions.

[0181] The processor 502 can be a CPU, or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present application. The one or more processors included in the smart device can be the same type of processor, such as one or more CPUs; or different types of processors, such as one or more CPUs and one or more ASICs.

[0182] RISC-V is an open-source instruction set architecture based on the principle of reduced instruction set (RISC), which can be applied to various aspects such as single-chip microcomputers and FPGA chips. Specifically, it can be applied in the fields of Internet of Things security, industrial control, mobile phones, personal computers, etc. Due to the consideration of small size, speed, and low power consumption in the design, it is particularly suitable for modern computing devices such as warehouse-scale computers, high-end mobile phones, and small embedded systems. With the rise of artificial intelligence Internet of Things (AIoT), RISC-V instruction set architecture has received more and more attention and support, and is expected to become the next generation of widely used CPU architecture.

[0183] The computer operation instructions in the embodiments of the present application can be computer operation instructions based on the RISC-V instruction set architecture, and correspondingly, the processor 502 can be designed based on the RISC-V instruction set. Specifically, the chip of the processor in the electronic device provided by the embodiments of the present application can be a chip designed based on the RISC-V instruction set, which can execute executable code based on the configured instructions, and thus implement the optimization method of green carbon sink benefit in the above embodiments.

[0184] The memory 506 is used to store programs 510. The memory 506 can include a high-speed RAM memory, and can also include a non-volatile memory such as at least one disk memory.

[0185] The programs 510 can be specifically used to make the processor 502 execute the optimization method of green carbon sink benefit in any of the preceding embodiments.

[0186] The specific implementation of each step in the programs 510 can refer to the corresponding description in the corresponding steps and units of any of the optimization methods of green carbon sink benefit embodiments described above, and will not be repeated here. Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the devices and modules described above can refer to the corresponding process description in the preceding method embodiments, which will not be repeated here.

[0187] The present application also provides a computer-readable storage medium storing instructions for causing a machine to perform the method of optimizing the carbon sink benefit of green land as described herein. Specifically, a system or apparatus equipped with a storage medium on which a software program code for realizing the functions of any of the above-described embodiments is stored, and a computer (or CPU or MPU) of the system or apparatus can be provided, and the computer (or CPU or MPU) is caused to read out and execute the program code stored in the storage medium.

[0188] In this case, the program code read out from the storage medium can realize the functions of any of the above-described embodiments by itself, and therefore the program code and the storage medium storing the program code constitute a part of the present application.

[0189] Embodiments of the storage medium for providing the program code include a floppy disk, a hard disk, a magneto-optical disk, an optical disk (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), a magnetic tape, a non-volatile memory card, and a ROM. Alternatively, the program code can be downloaded from a server computer via a communication network.

[0190] The embodiments of the present application also provide a computer program product including computer instructions instructing a computing device to perform any corresponding operation of the above-described method embodiments.

[0191] It should be noted that, according to the needs of implementation, each component / step described in the embodiments of the present application can be split into more components / steps, or two or more components / steps or parts of the operation of the components / steps can be combined into a new component / step, to achieve the purpose of the embodiments of the present application.

[0192] The above-described method according to the embodiments of the present application can be implemented in hardware, firmware, or as software or computer code that can be stored in a recording medium such as a CD ROM, a RAM, a floppy disk, a hard disk, or a magneto-optical disk, or downloaded through a network and originally stored in a remote recording medium or a non-transitory machine-readable medium and then stored in a local recording medium, so that the method described herein can be processed by such software on a recording medium using a general-purpose computer, a special-purpose processor, or programmable or special-purpose hardware such as an ASIC or an FPGA. It can be understood that the computer, processor, microprocessor controller, or programmable hardware includes a storage component (for example, RAM, ROM, flash memory, etc.) that can store or receive software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the method described herein. Furthermore, when a general-purpose computer accesses the code for implementing the method shown herein, the execution of the code will convert the general-purpose computer into a special-purpose computer for executing the method shown herein.

[0193] Those skilled in the art can understand that the units and method steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software manner depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to realize the described functions for each specific application, but such implementation should not be considered as beyond the scope of the embodiments of the present application.

[0194] The above implementation is only used to illustrate the embodiments of the present application, and is not intended to limit the embodiments of the present application. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the embodiments of the present application. Therefore, all equivalent technical solutions belong to the scope of the embodiments of the present application, and the patent protection scope of the embodiments of the present application should be defined by the claims.

Claims

1. A method for optimizing the carbon sequestration benefits of green spaces, characterized in that, The method includes: Obtain a dataset of carbon sinks within the study area, the dataset including data from multiple functional zones, which are obtained by dividing the study area according to urban functions; For each functional zone, based on the data of the functional zone, the vegetation carbon sequestration benefit of the functional zone is determined; an influencing factor set for the functional zone is obtained, which includes multiple influencing factors, including natural environmental factors and human activity characteristic factors; the importance percentage of each influencing factor to the vegetation carbon sequestration benefit is determined, and key factors are identified from the multiple influencing factors based on the importance percentage; the key factors are fitted to the vegetation carbon sequestration benefit, and the local standardized regression coefficients of the key factors are determined based on the fitting results, which are used to characterize the spatial relationship between the influencing factors and the vegetation carbon sequestration benefit; the spatial impact of the key factors on the vegetation carbon sequestration benefit is analyzed based on the local standardized regression coefficients, and the analysis results are used to characterize the spatial heterogeneity of the vegetation carbon sequestration benefit; an optimization strategy for the green space carbon sequestration benefit of the functional zone is determined based on the key factors and the analysis results. The process of fitting the key factors with the vegetation carbon sink benefits and determining the local standardized regression coefficients of the key factors based on the fitting results includes: weighting the vegetation carbon sink benefits using a Gaussian kernel function to construct a spatial weight matrix for the vegetation carbon sink benefits, where the spatial weight matrix measures the spatial similarity and degree of influence between different geographic units, and the geographic unit refers to the grid after the functional partition is rasterized; calculating the optimal bandwidth value for each key factor using the Akaike information criterion or the Bayesian information criterion, where the optimal bandwidth value is used to optimize the regression fitting of the key factors with the vegetation carbon sink benefits; fitting the key factors with the vegetation carbon sink benefits using a multi-scale geographic weighted regression model based on the spatial weight matrix and the optimal bandwidth values ​​of each key factor to obtain the fitting results corresponding to each key factor; and for each key factor, extracting the local regression coefficients corresponding to each grid based on the fitting results and standardizing the local regression coefficients to obtain the local standardized regression coefficients of the key factor.

2. The method according to claim 1, characterized in that, The optimization strategy for determining the carbon sequestration benefits of the functional zones based on the key factors and the analysis results includes: Analyze the changing trends between the key factors and the vegetation carbon sequestration benefits, determine the intervals in which the key factors have a significant impact on the vegetation carbon sequestration benefits, and obtain the threshold range of the key factors. A structural equation model is constructed based on the key factors and their threshold ranges, and the influence path and intensity of the key factors on vegetation carbon sequestration benefits are generated through the structural equation model. Based on the influence path, the intensity of the effect, and the analysis results, an optimization strategy for the carbon sequestration benefits of the functional zone is determined.

3. The method according to claim 2, characterized in that, The structural equation model is as follows: ； in, Let T be the coefficient matrix of the interaction paths between endogenous latent variables, and let T be the coefficient matrix of the influence paths between exogenous and endogenous latent variables. For random interference terms, As an endogenous latent variable, The term "exogenous latent variable" refers to the exogenous latent variable, while "endogenous latent variable" refers to the vegetation carbon sequestration benefit. The term "exogenous latent variable" refers to the key factor.

4. The method according to claim 2, characterized in that, The analysis of the changing trends between the key factors and the vegetation carbon sequestration benefits includes: The vegetation carbon sequestration benefits are input into the enhanced regression tree model, and the enhanced regression tree model outputs a partial dependency graph between the influencing factors and the vegetation carbon sequestration benefits. Based on the partial dependency graph between the key factors and the vegetation carbon sink benefits, the changing trends between the key factors and the vegetation carbon sink benefits are analyzed.

5. The method according to claim 1, characterized in that, The determination of the percentage importance of each of the aforementioned influencing factors to the vegetation carbon sequestration benefit, and the identification of key factors from the multiple influencing factors based on the percentage importance, includes: The vegetation carbon sequestration benefits are input into an enhanced regression tree model, and the enhanced regression tree model outputs the percentage importance of each of the influencing factors to the vegetation carbon sequestration benefits. The multiple influencing factors are ranked according to their importance percentage, and the key factors that meet the screening criteria are determined based on the ranking.

6. The method according to claim 5, characterized in that, The method further includes: The enhanced regression tree model is constructed by using carbon sink benefits as the dependent variable and the aforementioned multiple influencing factors as independent variables. The vegetation carbon sequestration benefit of each grid after the functional partitioning is rasterized is determined as a feature variable. The feature variable is input into the enhanced regression tree model, and the enhanced regression tree model is optimized through cross-validation to obtain the optimized enhanced regression tree model.

7. The method according to claim 6, characterized in that, The percentage of importance of each of the influencing factors to the vegetation carbon sequestration benefit output by the enhanced regression tree model includes: Based on the enhanced regression tree model, the number of times each of the aforementioned influencing factors acts as a branch node in the decision tree is counted, and the error reduction of the model prediction for each of the aforementioned influencing factors as a branch node is calculated; the number of times each of the aforementioned influencing factors acts as a branch node and the corresponding error reduction are weighted and summed, and the weighted sum is normalized to the importance percentage corresponding to each of the aforementioned influencing factors.

8. The method according to claim 1, characterized in that, Obtaining the set of influence factors for the functional partition includes: Obtain multiple candidate influence factors for the functional partition; Obtain the literature support, data quality score, and acquisition cost for each candidate impact factor; Calculate the dynamic factor weights of each candidate impact factor. The formula for calculating the dynamic factor weights is as follows: ； Candidate impact factors whose dynamic factor weights are greater than the weight threshold are added to the impact factor set of the functional partition. Among them, W DF For dynamic factor weights, LS represents literature support, and R... DQ The data quality is scored, and AC represents the acquisition cost.

9. The method according to claim 1, characterized in that, The multiple influencing factors include at least some of the following: Average annual temperature, average annual precipitation, solar radiation; Slope, altitude; Population density, GDP, human footprint index, green space coverage; Landscape connectivity index, landscape shape index, and landscape patch density.

10. An optimization device for green space carbon sequestration benefits, characterized in that, The device includes: The acquisition module is used to acquire a dataset of carbon sinks within the study area. The dataset includes data from multiple functional zones, which are obtained by dividing the study area according to urban functions. The processing module is configured to: determine the vegetation carbon sequestration benefit of each functional zone based on the data of that functional zone; acquire a set of influencing factors for each functional zone, including multiple influencing factors such as natural environmental factors and human activity characteristic factors; determine the percentage importance of each influencing factor to the vegetation carbon sequestration benefit, and identify key factors from the multiple influencing factors based on the percentage importance; fit the key factors to the vegetation carbon sequestration benefit, and determine the local standardized regression coefficients of the key factors based on the fitting results, the local standardized regression coefficients being used to characterize the spatial relationship between the influencing factors and the vegetation carbon sequestration benefit; analyze the spatial impact of the key factors on the vegetation carbon sequestration benefit based on the local standardized regression coefficients, and obtain analysis results, the analysis results being used to characterize the spatial heterogeneity of the vegetation carbon sequestration benefit; and determine an optimization strategy for the green space carbon sequestration benefit of the functional zone based on the key factors and the analysis results. The processing module fits the key factors with the vegetation carbon sink benefits and determines the local standardized regression coefficients of the key factors based on the fitting results. This includes: weighting the vegetation carbon sink benefits using a Gaussian kernel function to construct a spatial weight matrix for the vegetation carbon sink benefits. This spatial weight matrix measures the spatial similarity and influence between different geographic units, where each geographic unit refers to a rasterized grid of the functional partition. The module also calculates the optimal bandwidth value for each key factor using the Akaike Information Criterion or the Bayesian Information Criterion. This optimal bandwidth value optimizes the regression fit of the key factors to the vegetation carbon sink benefits. Based on the spatial weight matrix and the optimal bandwidth values ​​of each key factor, a multi-scale geographic weighted regression model is used to fit the key factors with the vegetation carbon sink benefits to obtain the fitting results for each key factor. For each key factor, local regression coefficients corresponding to each grid are extracted based on the fitting results, and these local regression coefficients are standardized to obtain the local standardized regression coefficients of the key factor.

Images