A bottom-up intelligent identification method of national land space based on grid

Abstract

The present invention provides a bottom-up intelligent identification method for land space based on grid, including: step 1, obtaining study area data and reclassifying point of interest data; step 2, identifying the initial land use type based on remote sensing image features; step 3, using the kernel density estimation algorithm to perform spatial estimation on the reclassified point of interest data to obtain the subdivided land use type; step 4, constructing a main function grid based on the subdivided land use type and determining the optimal scale of the grid, identifying the main function of the grid, and obtaining an estimated main function grid; step 5, using an unsupervised classification algorithm to obtain the initial land space pattern; step 6, performing spatial overlay analysis on the estimated main function grid and the initial land space pattern to obtain the final land space pattern. This application effectively solves the current problem of difficulty in unifying land space identification, monitoring and management, and lays the foundation for digital governance of land space.

Description

Technical Field

[0001] The present invention relates to the technical field of geography and planning, and in particular to a bottom-up intelligent identification method of national land space based on a grid. Background Art

[0002] National land space is an important carrier of the national economy, society and natural ecosystem, and is vital to the development of the country and the lives of the people. With the rapid advancement of urbanization and economic development, the growth of population and the continuous acceleration of the urbanization process have brought unprecedented challenges to the planning, management and governance of national land space. On the one hand, there are problems of "over-planning and over-construction" in the planning and construction of cities. A large amount of land is used to build residential, commercial and infrastructure, resulting in waste of land resources and damage to the ecological environment. On the other hand, rural land resources are facing multiple pressures from agricultural modernization, urbanization and industrialization, and the carrying capacity of rural land is gradually declining. In some areas, there are problems such as farmland occupation, village expansion, and abandoned cultivated land, which seriously affect the development of the rural economy and the lives of farmers. In addition, with the acceleration of industrialization and urbanization, the problems of land pollution and ecological environment destruction have become increasingly prominent, seriously affecting the overall layout of national land space and the stability of the ecological environment. Therefore, strengthening the digital governance of national land space and improving the utilization efficiency and protection level of national land space have become urgent needs for the current management and planning of national land space in my country.

[0003] The core of digital governance of land space is to accurately identify, dynamically monitor and scientifically plan land space through advanced information technology, so as to achieve rational use and sustainable development of land space. Digital governance can not only improve the utilization efficiency of land resources and promote economic and social development, but also protect the ecological environment, maintain national security, and promote urban-rural integration. Therefore, studying the methods of digital governance of land space has important practical significance and far-reaching historical significance for promoting land space management and planning.

[0004] Land space identification is the basis and premise of digital governance of land space, and its accuracy and effectiveness directly affect the quality and effect of land space management and planning. However, due to the complexity and diversity of land space, traditional land space identification methods have many limitations and shortcomings. First, traditional land space identification methods are often based on single data source and single scale data for analysis and judgment, which cannot fully and accurately reflect the diversity and complexity of land space; second, traditional land space identification methods are often based on artificial experience and subjective judgment, lack of scientificity and objectivity, and are easily affected by human factors; third, traditional land space identification methods are often static and cannot monitor and analyze the changes and evolution of land space in a timely and dynamic manner. Therefore, it is urgent to study and develop new land space identification methods to improve the efficiency of acquisition, analysis and utilization of land space information and provide more powerful support for digital governance of land space. Summary of the invention

[0005] In view of the shortcomings of current land and space identification methods, the purpose of this application is to provide a bottom-up intelligent land and space identification method based on a grid, which can realize intelligent identification, monitoring and management of land and space, and the method has good robustness and high reliability.

[0006] Specifically, the present invention proposes a grid-based bottom-up land space intelligent identification method, and the grid-based bottom-up land space intelligent identification method comprises the following steps:

[0007] Step 1: Obtain the remote sensing images, night light data, daily temperature data, daily rainfall data, standardized precipitation evaporation index, digital elevation model data, biodiversity data, 10 types of soil physical and chemical property data, gridded future population spatial distribution data, gridded future GDP spatial distribution data, ground sample data, point of interest data and road network data of the study area throughout the year; reclassify the types of point of interest data and obtain the reclassified point of interest data;

[0008] Step 2: Generate initial land use types using the random forest model using remote sensing images and ground sample data;

[0009] Step 3, using the kernel density estimation algorithm to perform spatial estimation on the reclassified POI data, and combining the initial land use type, night light data and road data to obtain the subdivided land use type;

[0010] Step 4: construct a main function grid based on the subdivided land use types and determine the optimal scale and main function type of the grid, use the kernel density estimation algorithm to perform spatial estimation on the main function grid, and obtain an estimated main function grid;

[0011] Step 5, using unsupervised classification algorithm to obtain the initial national land space pattern;

[0012] Step 6: Perform spatial overlay analysis on the estimated main functional grid and the initial national land spatial pattern, and use the erosion and dilation morphological processing algorithm to obtain the final national land spatial pattern.

[0013] Furthermore, in step 1, the POI data are reclassified into corresponding types in the "Guidelines for Classification of Land and Sea Use for National Land Space Survey, Planning and Use Control" based on the POI type.

[0014] Furthermore, in step 2, the following steps are also included:

[0015] Step 21, collecting remote sensing image features widely used in land use classification, wherein the remote sensing image features include spectral features, index features and texture features;

[0016] Step 22, randomly select 70% of the ground sample data as training data, and add remote sensing images to train the random forest model;

[0017] Step 23, the remaining 30% of ground sample data is used to evaluate the classification accuracy of the random forest model and identify the initial land use type.

[0018] Furthermore, in step 3, the following steps are also included:

[0019] Step 31, using a kernel density estimation algorithm and setting a search radius in the estimation algorithm, spatially estimating the reclassified point of interest data to obtain an estimated land use type of the point of interest;

[0020] Step 32, constructing a buffer zone of the road, the size of which can be set according to the level of road design, and converting it into raster data;

[0021] Step 33, spatially overlaying and analyzing the initial land use type, the land use type estimated by the points of interest, the night light data, and the road buffer zone to obtain the subdivided land use type.

[0022] Furthermore, in step 31, the search radius is a multiple of the resolution of the remote sensing image to determine whether all reclassified POI data are correctly estimated; if not, the search radius in the algorithm is expanded until the requirements are met.

[0023] Furthermore, in step 33, the scale of the subdivided land use type is consistent with the scale of the remote sensing image; the subdivided land use includes two categories: mixed land use type and single land use type.

[0024] Furthermore, in step 4, the following steps are also included:

[0025] Step 41, constructing a plurality of candidate main function grids of different scales, wherein the candidate main function grids contain grids of no less than 30 subdivided land use types;

[0026] Step 42, determining the main function type of each candidate main function grid according to the main function type with the largest proportion in the candidate main function grid;

[0027] The main functional types are divided according to the subdivided land use types;

[0028] If the proportion of a certain main function type in the candidate main function grid exceeds 50%, the grid is judged to have the main function, otherwise, it is judged to be a multifunctional grid;

[0029] Step 43, determine the rationality of the grid scale according to the proportion of the candidate main function grids with the main function in the study area; if the proportion of the grids with the main function is less than 90%, it indicates that the scale of the current candidate main function grid is too large, and return to step 41 to update the scale; when the proportion of the grids with the main function is greater than or equal to 90%, it indicates that the current candidate main function grid scale meets the composite requirements, and identify the main function grid and multifunctional grid in the entire study area based on the current scale;

[0030] Step 44, using a kernel density estimation algorithm to perform spatial estimation on the candidate main functional grids to obtain an estimated main functional grid.

[0031] Furthermore, in step 5, the following steps are also included:

[0032] Step 51, taking the comprehensive soil physical and chemical property index, 10°C annual accumulated temperature, annual rainfall, slope, biodiversity data, gridded future population spatial distribution data, and gridded future GDP spatial distribution data as variables for normalization processing;

[0033] Step 52, adopting the strategy of "birds of a feather flock together" and using an unsupervised classification machine learning algorithm to preliminarily divide the national land space pattern; this embodiment takes the K-Means unsupervised classification machine learning algorithm as an example, and the parameter "DistanceMetric" and the number of iterations are set to: "sqEuclidean" and 10 times respectively. After the calculation is completed, the preliminary national land space pattern can be obtained.

[0034] Furthermore, in step 51, the 10 types of soil physical and chemical property data include: soil organic matter, total nitrogen, total phosphorus, total potassium, alkaline nitrogen, available phosphorus, available potassium, bulk density, soil pH, and cation exchange capacity;

[0035] The comprehensive soil physical and chemical property index is calculated based on 10 types of soil physical and chemical property data; the comprehensive soil physical and chemical property index reclassifies the soil with reference to the "Technical Regulations for the Second National Soil Survey";

[0036] When the soil physical and chemical property index value S i When it is less than or equal to level X1, the calculation formula is as follows:

[0037] P i =S i ÷X1(S i ≤X1;P i ≤1)

[0038] Among them, P i represents the standardized value of soil physical and chemical properties i, S i represents the measured value of soil physical and chemical property index i, and X1 represents the first-level classification standard value;

[0039] When the soil physical and chemical property index value S i When it is greater than level X1 and less than or equal to level X2, the calculation formula is as follows:

[0040]

[0041] Among them, P i represents the standardized value of soil physical and chemical properties i, S i represents the measured value of soil physical and chemical property index i, X1 and X2 represent the first and second level classification standard values ​​respectively;

[0042] When the soil physical and chemical property index value S i When it is greater than level X2 and less than or equal to level X3, the calculation formula is as follows:

[0043]

[0044] Among them, P i represents the standardized value of soil physical and chemical properties i, S i represents the measured value of soil physical and chemical property index i, X2 and X3 represent the second and third level classification standard values ​​respectively;

[0045] When the soil physical and chemical property index value S i When it is greater than level X3, the calculation formula is as follows:

[0046] P i =3(S i >X3)

[0047] Among them, P i represents the standardized value of soil physical and chemical properties i, Si represents the measured value of soil physical and chemical property index i, and X3 represents the third-level classification standard value;

[0048] The comprehensive soil physical and chemical property index was calculated based on the standardized 10 soil physical and chemical property data. The calculation formula is as follows:

[0049]

[0050] Among them, Fer represents the comprehensive soil physical and chemical property index, P i Represents the standardized soil physical and chemical properties.

[0051] Furthermore, in step 51, the 10°C accumulated temperature calculation includes the following steps:

[0052] The daily temperature data is used to calculate the 10°C accumulated temperature. The calculation formula is as follows:

[0053]

[0054] Among them, AAT10 represents 10℃ accumulated temperature, i represents the start date of 10℃ effective active accumulated temperature, n represents the end date of 10℃ effective active accumulated temperature, Tem i Represents a temperature value greater than 10°C;

[0055] The annual rainfall calculation comprises the following steps:

[0056] The annual total rainfall is calculated using the daily rainfall data using the following formula:

[0057]

[0058] Where TPre represents annual rainfall, n represents the number of days per year, Pre i A numerical value representing daily rainfall;

[0059] The slope calculation comprises the following steps:

[0060] The slope is calculated using DEM data. The calculation formula is as follows:

[0061] Slope=(H i -H j )÷(D i -D j )×100%

[0062] Among them, Slope represents the slope, H i and H j are the heights of point i and point j respectively, D i and D j are the distances between point i and point j respectively.

[0063] The beneficial effects achieved by the present invention are:

[0064] The present invention can accurately identify land use types by fusing multi-source heterogeneous data.

[0065] The present invention adopts the idea of ​​grid management of national land space, which is conducive to the unified management and supervision of national land space, can obtain richer and more accurate spatial data and information, and is conducive to more accurate monitoring and management of resource utilization and environmental conditions in each small area.

[0066] While effectively overcoming administrative boundary barriers, the present invention can also ensure effective management within the existing administrative framework, and can guarantee the rationality, scientificity and operability of national land space governance. The method is highly robust and the structure is stable and reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a schematic flowchart of a grid-based bottom-up land space intelligent identification method according to an embodiment of the present application. DETAILED DESCRIPTION

[0068] The technical solution of the present invention is described in more detail below in conjunction with the accompanying drawings. The present invention includes but is not limited to the following embodiments.

[0069] like Figure 1 As shown, the present invention proposes a bottom-up intelligent identification method for national land space based on a grid, the method comprising:

[0070] Step 1: Obtain the study area data, calculate the comprehensive soil physical and chemical property index, 10°C accumulated temperature, annual rainfall and slope, and reclassify the data of interest points;

[0071] The data of the study area include remote sensing images of the study area throughout the year, night light data, daily temperature data, daily rainfall data, standardized precipitation evaporation index, digital elevation model data, biodiversity data, 10 types of soil physical and chemical properties data, gridded future population spatial distribution data, gridded future GDP spatial distribution data, ground sample data, point of interest data and road network data.

[0072] Referring to the "Guidelines for Classification of Land and Sea Use for National Land Space Survey, Planning and Use Control" issued by the Ministry of Natural Resources of China in November 2020, the POI data are divided into the types proposed in the guidelines. The types in the guidelines include primary, secondary and tertiary categories, which can be divided according to the complexity of the POI types in the study area.

[0073] This example uses the first-level class as an example to reclassify the POI data to obtain the reclassified POI data, as shown in Table 1 below:

[0074] Table 1 Types of points of interest and their corresponding reclassification names

[0075]

[0076] Step 2: Collect remote sensing image features for land use classification, train machine learning models, and identify initial land use types;

[0077] In step 2, the following steps are also included:

[0078] Step 21, collecting remote sensing image features widely used in land use classification, including spectral features, index features and texture features;

[0079] Step 22, random forest is a reliable, accurate and efficient machine learning model that has been widely used in land use classification; 70% of the ground sample data is randomly selected as training data, and remote sensing images are added to train the random forest model;

[0080] Step 23, the remaining 30% of ground sample data is used to evaluate the classification accuracy of the random forest model and identify the initial land use type.

[0081] In this embodiment, the initial land use types include cultivated land, garden land, forest land, grassland, wetland, water body, bare land and construction land.

[0082] Step 3: Use the kernel density estimation algorithm to perform spatial estimation on the reclassified POI data, and further subdivide the land use type by combining the initial land use type, night light data, and road network data;

[0083] In step 3, the following steps are also included:

[0084] Step 31, using the kernel density estimation algorithm and setting the search radius in the estimation algorithm, spatially estimating the reclassified point of interest data to obtain the estimated land use type of the point of interest; the search radius is a multiple of the resolution of the remote sensing image, and it is determined that all the reclassified point of interest data have been correctly estimated; if no incorrect estimation is obtained, the search radius in the algorithm is expanded until the requirements are met.

[0085] In this embodiment, the kernel density estimation algorithm uses the function provided by ArcGIS software, and manually sets the search radius. For example, if the remote sensing image is 10m, the search radius can be set to 10m, 20m, 30m, 40m, 50m, 60m, 70m, 80m, 90m, 100m...; the search radius is set to 10m at the beginning, and the search radius is gradually adjusted according to the results until the requirements are met.

[0086] Step 32, constructing a buffer zone of the road using the road network data, the size of the buffer zone can be set according to the level of road design, and converting it into raster data;

[0087] Step 33, spatially overlaying and analyzing the initial land use type, the estimated land use type of the points of interest, the night light data and the road buffer, to obtain the subdivided land use type; wherein the night light data divides the construction land into urban and rural areas, and the value of the night light data in the urban area is not less than 2; the scale of the subdivided land use type is consistent with the scale of the remote sensing image; this step is actually to further subdivide the construction land in the initial land use type;

[0088] In this embodiment, the subdivided land use types include cultivated land, gardens, forests, grasslands, wetlands, water bodies, bare land, urban residential land, rural residential land, agricultural facility construction land, public administration and public service land, commercial service land, industrial and mining land, warehousing land, transportation land, green space and open space land, and special land.

[0089] At the same time, the subdivided land use types are divided into two categories: mixed land use types and single land use types; single land use types include cultivated land, gardens, woodlands, grasslands, water bodies, wetlands, etc.; mixed land use types refer to a land use grid that contains multiple uses, for example, a land use grid contains both urban residential areas and commercial land use types.

[0090] Step 4: construct a main function grid based on the subdivided land use types and determine the optimal scale of the grid, identify the main function of the grid, and use the kernel density estimation algorithm to perform spatial estimation on the main function grid to obtain an estimated main function grid;

[0091] In step 4, the following steps are also included:

[0092] Step 41, construct candidate main function grids of different scales; the scale of the grid should be larger than the scale of the subdivided land use type; in addition, the scale of the grid conforms to the integer principle, which is convenient for operation and application in the actual national land space planning scheme, such as 500×500m, 1000×1000m, 1500×1500m, 2000×2000m, 2500×2500m and 3000×3000m, etc.; at the same time, the prerequisite of the minimum sample size of spatial analysis must be observed to ensure that the candidate main function grid contains no less than 30 grids of land use types;

[0093] Step 42, determining the main function according to the land use type with the largest proportion in each candidate main function grid; if the proportion of a certain land use type in the grid exceeds 50%, the grid is determined to have the main function, otherwise, it is determined to be a multifunctional grid; for example, if the proportion of cultivated land in the grid exceeds 50%, the grid is determined to have the main function, and it is an agricultural production function; the main function of the spatial grid is analyzed from the perspective of the "three life" function, and the types of main functions and their corresponding land use types are shown in Table 2 below; at the same time, each main function can be further divided according to urban and rural areas, such as rural life functions and urban life functions, and can also be combined according to the structure of land use in the grid, such as rural life-agricultural production multifunctionality;

[0094] Table 2 Main function types and their corresponding land use types

[0095]

[0096] Step 43, determine the rationality of the grid scale according to the proportion of grids with main functions in the study area; if the proportion of grids with main functions is less than 90%, it indicates that the scale of the current candidate main function grid is too large, and return to step 41 to update the scale; when the proportion of grids with main functions is greater than or equal to 90%, it indicates that the current candidate main function grid scale meets the composite requirements, and identify the main function grid and multifunctional grid in the entire study area based on the current scale;

[0097] Step 44, in order to obtain the future main functional grid, a kernel density estimation algorithm is used to perform spatial estimation on the candidate main functional grid obtained in step 43 to obtain an estimated main functional grid;

[0098] In this embodiment, the kernel density estimation algorithm uses the function provided by ArcGIS software; the estimated main functional grid will be used to calculate the national land spatial pattern; since the proportion of multifunctional grids is less than 10% and they are scattered in space, the multifunctional grids are not included in the calculation in this step;

[0099] Step 5, using unsupervised classification algorithm to obtain the initial national land space pattern;

[0100] In step 5, the following steps are also included:

[0101] Step 51, normalizing the variables collected and calculated in step 1, including comprehensive soil physical and chemical property index, 10°C accumulated temperature, annual rainfall, slope, biodiversity data, gridded future population spatial distribution data, gridded future GDP spatial distribution data, standardized precipitation evaporation index, etc.;

[0102] Step 12, calculate the comprehensive soil physical and chemical property index, 10°C accumulated temperature, annual rainfall and slope. The details are as follows:

[0103] (1) Calculation of comprehensive soil physical and chemical properties

[0104] Soil physical and chemical property data include: soil organic matter, total nitrogen, total phosphorus, total potassium, alkaline nitrogen, available phosphorus, available potassium, bulk density, soil pH, and cation exchange capacity; comprehensive soil physical and chemical property indexes are calculated using 10 soil physical and chemical property data. The comprehensive soil physical and chemical property indexes refer to the "Technical Regulations for the Second National Soil Survey" to reclassify the soil. The detailed standards are shown in the following table:

[0105] Table 1 Standard values ​​of soil property classification

[0106]

[0107]

[0108] Based on the soil property classification standard, the numerical standardization is carried out in the following way, and the calculation method is as follows:

[0109] When the soil physical and chemical property index value S i When it is less than or equal to level X1, the calculation formula is as follows:

[0110] P i =S i ÷X1(S i ≤X1;P i ≤1)

[0111] Among them, P i represents the standardized value of soil physical and chemical properties i, S i represents the measured value of soil physical and chemical property index i, and X1 represents the first-level classification standard value.

[0112] When the soil physical and chemical property index value S i When it is greater than level X1 and less than or equal to level X2, the calculation formula is as follows:

[0113]

[0114] Among them, P i represents the standardized value of soil physical and chemical properties i, S i represents the measured value of soil physical and chemical property index i, and X1 and X2 represent the first and second grade classification standard values ​​respectively.

[0115] When the soil physical and chemical property index value S i When it is greater than level X2 and less than or equal to level X3, the calculation formula is as follows:

[0116]

[0117] Among them, P i represents the standardized value of soil physical and chemical properties i, S i represents the measured value of soil physical and chemical property index i, and X2 and X3 represent the second and third level classification standard values ​​respectively.

[0118] When the soil physical and chemical property index value S i When it is greater than level X3, the calculation formula is as follows:

[0119] P i =3(S i >X3)

[0120] Among them, P i represents the standardized value of soil physical and chemical properties i, S i represents the measured value of soil physical and chemical property index i, and X3 represents the third-level classification standard value.

[0121] The comprehensive soil physical and chemical property index was calculated based on the standardized 10 soil physical and chemical property data. The calculation formula is as follows:

[0122]

[0123] Among them, Fer represents the comprehensive soil physical and chemical property index, P i Represents the standardized soil physical and chemical properties.

[0124] (2) Calculation of accumulated temperature at 10°C

[0125] The daily temperature data is used to calculate the 10°C accumulated temperature. The calculation formula is as follows:

[0126]

[0127] Among them, AAT10 represents 10℃ accumulated temperature, a represents the start date of 10℃ effective active accumulated temperature, n represents the end date of 10℃ effective active accumulated temperature, Tem a Represents a temperature value greater than 10℃.

[0128] (3) Calculation of annual rainfall

[0129] The annual total rainfall is calculated using the daily rainfall data using the following formula:

[0130]

[0131] Where TPre represents annual rainfall, m represents the number of days per year, and Pre b A numerical value representing daily rainfall.

[0132] (4) Slope calculation

[0133] The slope is calculated using digital elevation model data. The calculation formula is as follows:

[0134] Slope=(H c -H d )÷(D c -D d )×100%

[0135] Among them, Slope represents the slope, H c and H d are the heights of point c and point d, respectively. c and D d are the distances between point c and point d respectively.

[0136] Step 52, adopting the strategy of "birds of a feather flock together" and using an unsupervised classification machine learning algorithm to preliminarily divide the national land space pattern; this embodiment takes the K-Means unsupervised classification machine learning algorithm as an example, and the parameter "DistanceMetric" and the number of iterations are set to: "sqEuclidean" and 10 times respectively. After the calculation is completed, the preliminary national land space pattern can be obtained; in actual applications, other unsupervised classification algorithms can also be replaced according to the classification effect.

[0137] Step 6: Perform spatial overlay analysis on the estimated main functional grid and the initial national land spatial pattern, use the erosion and dilation morphological processing algorithm to smooth the boundaries of the national land spatial pattern, and obtain the final national land spatial pattern.

[0138] Compared with traditional planning methods and existing conventional methods, the grid-based bottom-up national land space intelligent identification method provided in this embodiment has been proved through experiments that grid-based management of national land space is conducive to the unified management and supervision of national land space, and can obtain richer and more accurate spatial data and information, which is conducive to more accurate monitoring and management of resource utilization and environmental conditions in each small area.

[0139] Through the technical solution in this application, the cascading effects of national land space elements, structures, functions and patterns are scientifically identified, revealing the multi-scale, multi-level and multi-process transmission mechanism of national land space identification, laying a solid foundation for the digital governance of national land space.

[0140] The steps in this application can be adjusted in order, combined, and deleted according to actual needs.

[0141] The present invention is not limited to the above-mentioned specific implementation modes. A person skilled in the art may implement the present invention in various other specific implementation modes according to the embodiments and the disclosure of the drawings. Therefore, any design that adopts the design structure and concept of the present invention and makes some simple transformations or changes shall fall within the scope of protection of the present invention.

Claims

1. A bottom-up intelligent identification method for national land space based on grid, characterized in that: The grid-based bottom-up land space intelligent identification method comprises the following steps: Step 1: Obtain the remote sensing images, night light data, daily temperature data, daily rainfall data, standardized precipitation evaporation index, digital elevation model data, biodiversity data, 10 types of soil physical and chemical property data, gridded future population spatial distribution data, gridded future GDP spatial distribution data, ground sample data, point of interest data and road network data of the study area throughout the year; reclassify the types of point of interest data and obtain the reclassified point of interest data; Step 2: Generate initial land use types using the random forest model using remote sensing images and ground sample data; Step 3, using the kernel density estimation algorithm to perform spatial estimation on the reclassified POI data, and combining the initial land use type, night light data and road data to obtain the subdivided land use type; Step 4: construct a main function grid based on the subdivided land use types and determine the optimal scale and main function type of the grid, use the kernel density estimation algorithm to perform spatial estimation on the main function grid, and obtain an estimated main function grid; Step 5, using unsupervised classification algorithm to obtain the initial national land space pattern; Step 6: Perform spatial overlay analysis on the estimated main functional grid and the initial national land spatial pattern, and use the erosion and dilation morphological processing algorithm to obtain the final national land spatial pattern; In step 3, the following steps are also included: Step 31, using a kernel density estimation algorithm and setting a search radius in the estimation algorithm, spatially estimating the reclassified point of interest data to obtain an estimated land use type of the point of interest; Step 32, constructing a buffer zone of the road, the size of which can be set according to the level of road design, and converting it into raster data; Step 33, performing spatial overlay analysis on the initial land use type, the land use type estimated by the points of interest, the night light data and the road buffer to obtain the subdivided land use type; In step 4, the following steps are also included: Step 41, constructing a plurality of candidate main function grids of different scales, wherein the candidate main function grids contain grids of no less than 30 subdivided land use types; Step 42, determining the main function type of each candidate main function grid according to the main function type with the largest proportion in the candidate main function grid; The main functional types are divided according to the subdivided land use types; If the proportion of a certain main function type in the candidate main function grid exceeds 50%, the grid is judged to have the main function, otherwise, it is judged to be a multifunctional grid; Step 43, determine the rationality of the grid scale according to the proportion of the candidate main function grids with the main function in the study area; if the proportion of the grids with the main function is less than 90%, it indicates that the scale of the current candidate main function grid is too large, and return to step 41 to update the scale; when the proportion of the grids with the main function is greater than or equal to 90%, it indicates that the current candidate main function grid scale meets the composite requirements, and identify the main function grid and multifunctional grid in the entire study area based on the current scale; Step 44, using a kernel density estimation algorithm to perform spatial estimation on the candidate main functional grids to obtain an estimated main functional grid; In step 5, the following steps are also included: Step 51, taking the comprehensive soil physical and chemical property index, 10°C annual accumulated temperature, annual rainfall, slope, biodiversity data, gridded future population spatial distribution data, and gridded future GDP spatial distribution data as variables for normalization processing; Step 52, adopting the strategy of "birds of a feather flock together", using the K-Means unsupervised classification machine learning algorithm to preliminarily divide the national land space pattern; the parameter "Distance Metric" and the number of iterations are set to: "sqEuclidean" and 10 times respectively, and the initial national land space pattern can be obtained after the calculation is completed.

2. The bottom-up intelligent identification method of national land space based on grid according to claim 1 is characterized in that: In step 1, the POI data are reclassified into the corresponding types in the "Guidelines for Classification of Land and Sea Use for National Land Space Survey, Planning and Use Control" based on the POI types.

3. The bottom-up intelligent identification method of national land space based on grid according to claim 1 is characterized in that: In step 2, the following steps are also included: Step 21, collecting remote sensing image features widely used in land use classification, wherein the remote sensing image features include spectral features, index features and texture features; Step 22, randomly select 70% of the ground sample data as training data, and add remote sensing images to train the random forest model; Step 23, the remaining 30% of ground sample data is used to evaluate the classification accuracy of the random forest model and identify the initial land use type.

4. The bottom-up intelligent identification method of national land space based on grid according to claim 1 is characterized in that: In step 31, the search radius is a multiple of the resolution of the remote sensing image to determine whether all reclassified POI data are correctly estimated; if not, the search radius in the algorithm is expanded until the requirements are met.

5. The bottom-up intelligent identification method of national land space based on grid according to claim 1 is characterized in that: In step 33, the scale of the subdivided land use type is consistent with the scale of the remote sensing image; the subdivided land use includes two categories: mixed land use type and single land use type.

6. The bottom-up intelligent identification method of national land space based on grid according to claim 1 is characterized in that: In step 51, the 10 types of soil physical and chemical property data include: soil organic matter, total nitrogen, total phosphorus, total potassium, alkaline nitrogen, available phosphorus, available potassium, bulk density, soil pH, and cation exchange capacity; The comprehensive soil physical and chemical property index is calculated based on 10 types of soil physical and chemical property data; the comprehensive soil physical and chemical property index reclassifies the soil with reference to the "Technical Regulations for the Second National Soil Survey"; When the soil physical and chemical property index value S i When it is less than or equal to level X1, the calculation formula is as follows: P i =S i ÷X1(S i ≤X1;P i ≤1) Among them, P i represents the standardized value of soil physical and chemical properties i, S i represents the measured value of soil physical and chemical property index i, and X1 represents the first-level classification standard value; When the soil physical and chemical property index value S i When it is greater than level X1 and less than or equal to level X2, the calculation formula is as follows: Among them, P i represents the standardized value of soil physical and chemical properties i, S i represents the measured value of soil physical and chemical property index i, X1 and X2 represent the first and second level classification standard values ​​respectively; When the soil physical and chemical property index value S i When it is greater than level X2 and less than or equal to level X3, the calculation formula is as follows: Among them, P i represents the standardized value of soil physical and chemical properties i, S i represents the measured value of soil physical and chemical property index i, X2 and X3 represent the second and third level classification standard values ​​respectively; When the soil physical and chemical property index value S i When it is greater than level X3, the calculation formula is as follows: P i =3 (S i >X3) Among them, P i represents the standardized value of soil physical and chemical properties i, S i represents the measured value of soil physical and chemical property index i, and X3 represents the third-level classification standard value; The comprehensive soil physical and chemical property index was calculated based on the standardized 10 soil physical and chemical property data. The calculation formula is as follows: Among them, Fer represents the comprehensive soil physical and chemical property index, P i Represents the standardized soil physical and chemical properties.

7. The bottom-up intelligent identification method of national land space based on grid according to claim 1 is characterized in that: In step 51, the 10°C accumulated temperature calculation includes the following steps: The daily temperature data is used to calculate the 10°C accumulated temperature. The calculation formula is as follows: Among them, AAT10 represents 10℃ accumulated temperature, i represents the start date of 10℃ effective active accumulated temperature, n represents the end date of 10℃ effective active accumulated temperature, Tem i Represents a temperature value greater than 10°C; The annual rainfall calculation comprises the following steps: The annual total rainfall is calculated using the daily rainfall data using the following formula: Where TPre represents annual rainfall, n represents the number of days per year, Pre i A numerical value representing daily rainfall; The slope calculation comprises the following steps: The slope is calculated using DEM data. The calculation formula is as follows: Slope=(H i -H j )÷(D i -D j )×100% Among them, Slope represents the slope, H i and H j are the heights of point i and point j respectively, D i and D j are the distances between point i and point j respectively.

Images