A long-period grid population distribution construction method, system, device and medium
By acquiring the initial dataset and reconstructing the urban land use distribution using FLUS and H-FLUS models, and then combining it with the random forest model, the uncertainty and consistency issues of long-term grid population distribution data were resolved, achieving high-precision grid population distribution mapping.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SUN YAT SEN UNIV
- Filing Date
- 2023-08-03
- Publication Date
- 2026-05-29
Smart Images
Figure CN117076588B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of geographic information science and technology, and in particular to a method, system, device and medium for constructing long-term grid population distribution. Background Technology
[0002] Long-term, high-resolution gridded population distribution data is crucial for understanding the impact of human activities on the natural environment. Furthermore, gridded population distribution data is a key foundational resource for assessing the exposure and vulnerability of human societies to various natural disasters, including floods, sea-level rise, droughts, heat waves, infectious diseases, landslides, and earthquakes. Simultaneously, gridded population distribution data is also essential for assessing and monitoring progress toward the United Nations Sustainable Development Goals (SDGs), with over 73 SDG indicators requiring gridded population distribution data as input. Therefore, constructing long-term, high-resolution gridded population distribution data is of great significance for deepening our understanding of the interaction between humans and nature and for addressing various social and environmental challenges.
[0003] In recent decades, the revolution in geospatial data availability has greatly facilitated the development of gridded population spatial distribution data. Several well-known global gridded population distribution datasets based on remote sensing data have been produced, including GWP, GRUMP, GHS-POP, Landscan, and WorldPop. Satellite remote sensing data can capture large-scale population distribution patterns and is an important auxiliary data for gridded population distribution mapping. However, the limited time span of satellite imagery makes long-term gridded population mapping from historical to future periods very difficult. Current gridded population distribution data products are concentrated on the last few decades for which satellite observation data is available. To capture long-term population dynamics, it is necessary to produce gridded population data spanning history, the present, and the future.
[0004] While some studies have constructed future and historical grid population data, the lack of dynamic ancillary data means that these studies often rely on static or no ancillary data, leading to significant uncertainty in the results. Furthermore, existing studies are typically based on simple empirical methods, making it difficult to model the complex nonlinear relationships between various factors and population distribution. Additionally, current observational, historical, and future grid population distribution modeling often uses different data and methods, resulting in a lack of consistency between the data and hindering long-term population dynamics analysis. Summary of the Invention
[0005] The purpose of this invention is to propose a method, system, device, and medium for constructing long-term grid population distribution maps, so as to effectively improve the accuracy of grid population distribution mapping.
[0006] To achieve the above objectives, in a first aspect, embodiments of the present invention provide a method for constructing a long-term grid-based population distribution, the method comprising:
[0007] Obtain an initial urban land use distribution dataset, an initial population distribution dataset, and a driving factor dataset, wherein the driving factor dataset includes elevation, slope, distance to the river, distance to the city center, and distance to the main road;
[0008] An urban land use distribution model is established and trained based on the initial urban land use distribution dataset. The driving factor dataset is then input into the trained urban land use distribution model to obtain urban land use development suitability data. The urban land use development suitability data and the initial urban land use distribution dataset are then input into the FLUS model to construct the urban land use distribution.
[0009] A population distribution model is established and trained based on the initial population distribution dataset and the initial urban land distribution dataset. The urban land distribution and the initial population distribution dataset are then input into the trained population distribution model to construct a long-term grid population distribution.
[0010] Further, the step of inputting the urban land development suitability data and the initial urban land distribution dataset into the FLUS model includes:
[0011] The urban land development suitability data and the initial urban land distribution dataset of historical years are input into the H-FLUS model to reconstruct the historical urban land distribution. The H-FLUS model is obtained by removing the pixel with the lowest total probability from the FLUS model. The total probability is the product of urban land development suitability and neighborhood effect.
[0012] Further, the step of inputting the driving factor dataset into the trained urban land use distribution model to obtain urban land use development suitability data includes:
[0013] The intermediate layer node data is obtained by multiplying the network weights connecting the input layer nodes and the intermediate layer nodes in the urban land use distribution model by the driving factor data.
[0014] The output layer node data is obtained by multiplying the network weights connecting the intermediate layer nodes and the output layer nodes in the urban land use distribution model by the sigmoid value of the intermediate layer node data.
[0015] The sigmoid value of the output layer node data is calculated as urban land use suitability data.
[0016] Furthermore, obtaining the driving factor dataset includes:
[0017] The slope, elevation, distance to the river, distance to the city center, and distance to the main road of each pixel are calculated using GIS software.
[0018] Furthermore, obtaining the driving factor dataset includes:
[0019] The driving factor dataset is converted into raster data, normalized, and then resampled to a preset spatial resolution.
[0020] Further, the step of establishing and training a population distribution model based on the initial population distribution dataset and the initial urban land use distribution dataset includes:
[0021] A population distribution model is established based on a random forest model, and the population distribution model is trained based on the initial population distribution dataset, the initial urban land use distribution dataset, and the driving factor dataset.
[0022] Further, the step of establishing and training an urban land use distribution model based on the initial urban land use distribution dataset includes:
[0023] The total urban land use volume shared with the socio-economic path database will be used as a constraint to terminate the training.
[0024] Secondly, embodiments of the present invention provide a long-term grid-based population distribution construction system, the system comprising:
[0025] The data acquisition module is used to acquire an initial urban land use distribution dataset, an initial population distribution dataset, and a driving factor dataset. The driving factor dataset includes elevation, slope, distance to the river, distance to the city center, and distance to the main road.
[0026] The urban land use distribution construction module is used to establish and train an urban land use distribution model based on the initial urban land use distribution dataset, input the driving factor dataset into the trained urban land use distribution model to obtain urban land use development suitability data, and input the urban land use development suitability data and the initial urban land use distribution dataset into the FLUS model to construct the urban land use distribution.
[0027] The population distribution construction module is used to build and train a population distribution model based on the initial population distribution dataset. The urban land use distribution and the initial population distribution dataset are input into the trained population distribution model to construct a long-term grid population distribution.
[0028] Thirdly, embodiments of the present invention also provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.
[0029] Fourthly, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method.
[0030] This invention provides a method, system, device, and medium for constructing long-term grid-based population distribution maps. The method includes: acquiring an initial urban land use distribution dataset, an initial population distribution dataset, and a driving factor dataset; establishing and training an urban land use distribution model based on the initial urban land use distribution dataset; inputting the driving factor dataset into the trained urban land use distribution model to obtain urban land use development suitability data; inputting the urban land use development suitability data and the initial urban land use distribution dataset into a FLUS model to construct the urban land use distribution; establishing and training a population distribution model based on the initial population distribution dataset and the initial urban land use distribution dataset; inputting the urban land use distribution and the initial population distribution dataset into the trained population distribution model to construct a long-term grid-based population distribution map. This invention effectively improves the accuracy of grid-based population distribution mapping. Attached Figure Description
[0031] Figure 1 This is a flowchart illustrating a method for constructing a long-term grid population distribution according to an embodiment of the present invention;
[0032] Figure 2 This is a schematic diagram of the historical urban land reconstruction process based on H-FLUS provided in an embodiment of the present invention;
[0033] Figure 3 This is a schematic diagram of the long-term urban land use distribution simulation results of the study area provided in this embodiment of the invention;
[0034] Figure 4 This is a schematic diagram of the long-term grid population distribution construction results in the research area provided by an embodiment of the present invention;
[0035] Figure 5 This is a schematic diagram illustrating the accuracy verification results of the population prediction model and the population reconstruction model provided in the embodiments of the present invention;
[0036] Figure 6 This is a system block diagram of a long-term grid population distribution construction system provided in an embodiment of the present invention;
[0037] Figure 7 This is an internal structural diagram of a computer device provided in an embodiment of the present invention. Detailed Implementation
[0038] To make the objectives, technical solutions, and beneficial effects of this application clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Obviously, the embodiments described below are only part of the embodiments of the present invention and are used to illustrate the present invention, but are not intended to limit the scope of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.
[0039] In one embodiment, such as Figure 1 As shown, a method for constructing a long-term grid population distribution is provided, the method comprising:
[0040] S11. Obtain the initial urban land use distribution dataset, the initial population distribution dataset, and the driving factor dataset, wherein the driving factor dataset includes elevation, slope, distance to the river, distance to the city center, and distance to the main road;
[0041] In this embodiment, the process of obtaining the driving factor dataset is as follows: GIS software is used to calculate the slope, elevation, distance to the river, distance to the city center, and distance to the main road for each pixel. Specifically, GIS software is used to calculate the distance from each pixel within the area to the feature, generating a distance driving factor layer with a resolution of 1 km. Elevation data with a resolution of 1 km is collected, and GIS software is used to calculate the slope of each pixel within the area, generating a slope driving factor layer, ultimately obtaining five spatial driving factors. GIS software, or Geographic Information System software, is a technical system software that, with the support of computer hardware and software systems, collects, stores, manages, calculates, analyzes, displays, and describes geographic distribution data in the entire or part of the Earth's surface space. When obtaining the driving factor dataset, the dataset should also be converted into raster data, normalized, and resampled to a preset spatial resolution. This embodiment obtains driving factor data using GIS software and eliminates the scale effect through normalization, improving data reliability.
[0042] S12. Establish and train an urban land use distribution model based on the initial urban land use distribution dataset, input the driving factor dataset into the trained urban land use distribution model to obtain urban land use development suitability data; input the urban land use development suitability data and the initial urban land use distribution dataset into the FLUS model to construct the urban land use distribution.
[0043] To reconstruct historical urban land use in stages, this embodiment inputs the urban land development suitability data and the initial urban land distribution dataset from historical years into the H-FLUS model to reconstruct the historical urban land distribution. The H-FLUS model is obtained by removing the pixel with the lowest total probability from the FLUS model. The total probability is the product of urban land development suitability and neighborhood effect. Figure 2 As shown, Figure 2 'a' represents the urban land use distribution in year 't'. Figure 2 b calculates the total probability of each pixel in the urban land use distribution for year t. Figure 2 The city land distribution for year t-10 is obtained by removing the pixels with the lowest total probability. Specifically, H-FLUS is a method that eliminates n pixels from low to high probability in a development probability layer based on the ANN output of FLUS. It is a complement and extension of FLUS and is compatible with FLUS. In H-FLUS, the total probability TP of each pixel in the city land distribution map layer in the initial year t remaining urbanized in year t-10 is calculated. k Expressed as:
[0044] TP k =P k ×Ω k
[0045] Among them, P k The output of step S3 is the suitability of urban land development, Ω k The neighborhood effect represents the proportion of urban land in the neighborhood of a pixel mole. Ω k Represented as:
[0046]
[0047] Where d is the diameter of the mole neighborhood, and con is a conditional function that returns 1 if the cell (i,j) is of type city, otherwise it returns 0. For a given city area requirement, H-FLUS iterates through the current city using map layers to eliminate cells with the minimum total probability TP. k The H-FLUS model reconstructs historical urban land use distribution using pixels. For example, if the urban land area decreases by n pixels from year t to t-10, then the n urban pixels with the minimum total probability are eliminated. The H-FLUS model uses the same initial year urban land use distribution and urban development suitability layers as the FLUS model to drive the model, and the total probability equation of H-FLUS is similar to that of FLUS. This embodiment ensures the consistency of historical and future urban land use distribution simulation data through the H-FLUS model, which is compatible with the FLUS model.
[0048] In this embodiment, the process of inputting the driving factor dataset into the trained urban land use distribution model to obtain urban land use development suitability data is as follows: The network weights connecting the input layer nodes and intermediate layer nodes in the urban land use distribution model are multiplied by the driving factor data to obtain intermediate layer node data; the network weights connecting the intermediate layer nodes and output layer nodes in the urban land use distribution model are multiplied by the sigmoid value of the intermediate layer node data to obtain output layer node data; the sigmoid value of the output layer node data is calculated as urban land use development suitability data. Specifically, this can be represented as follows:
[0049]
[0050]
[0051] p k =sigmoid(net) k )
[0052] Where, p k This represents the probability of land use classification, where k represents urban and non-urban areas respectively, and p... 城市 This refers to the suitability of urban land use for development. i w represents the i-th spatial driving factor i,j w represents the network weight connecting input layer node i and intermediate layer node j. j,k This represents the network weight connecting intermediate layer node j and output layer node k. The sigmoid function is an activation function that compresses the network output to 0-1, thus avoiding gradient explosion while ensuring the network output has meaningful probability values. This embodiment uses the total urban land use data from the shared socioeconomic path database as a constraint to terminate training, effectively preventing the training process from continuing indefinitely. The preferred shared socioeconomic path database in this embodiment is the HYDE database; other shared socioeconomic path databases with total urban land use data can also be selected as needed. In this embodiment, both the FLUS model used for simulating future urban land use and the H-FLUS model used for reconstructing historical urban land use use the same urban development suitability and initial year urban land use distribution layers, further ensuring the consistency of long-term urban land use distribution data.
[0053] S13. Establish and train a population distribution model based on the initial population distribution dataset and the initial urban land distribution dataset, and input the urban land distribution and the initial population distribution dataset into the trained population distribution model to construct a long-term grid population distribution.
[0054] In this embodiment, a population distribution model is established based on a random forest model. The population distribution model is trained using the initial population distribution dataset, the initial urban land use distribution dataset, and the driving factor dataset. The future population prediction model is specifically represented by the following formula:
[0055] Pop t+10 =f p (env 0 Pop t Neigh (Pop) t Urban t+10 Neighborhood t+10 ))
[0056] The historical population reconstruction model is specifically expressed by the following formula:
[0057] Pop t-10 =f h (env 0 Pop t Neigh (Pop) t Urban t-10 Neighborhood t-10 ))
[0058] Where f p It's a population prediction model, fitted using a random forest model. 0 These represent socioeconomic and natural environmental factors influencing population distribution; due to data acquisition issues, these factors are static. t This represents the population distribution in year t. Due to the spatial dependence of population distribution, the population distribution in year t+10 is closely related to the population distribution in year t. t +10 This represents the urban land use distribution in year t+10. Cities are the main population centers, therefore the urban distribution in year t+10 is closely related to the population in year t+10. Neighborhood effect reflects the influence of neighborhood population and urban distribution. Random forest is an ensemble learning model composed of multiple decision trees, exhibiting low prediction bias and variance. The random forest regression equation is as follows:
[0059]
[0060] Among them, H t (X) is the regression result of a single decision tree, where X is the input feature, i.e., env. 0 Pop t Urban t+10 Neigh (Pop) t ) and Neigh (Urban)t+10 ), where T is the number of decision trees. y is the regression result, i.e., the population Population in year t+10. t +10 This embodiment uses the same input and output data as the future population prediction model through the random forest historical population reconstruction model, thus ensuring the consistency of population distribution data from history to the future.
[0061] In a specific embodiment, this invention focuses on the Shanghai metropolitan area in China, specifically a 2° region centered on Shanghai. The data used include: Worldpop grid population distribution data from 2000, 2010, and 2020, with a spatial resolution of 1 km; ESA CCI-LC urban land use distribution data from 1992, 2000, 2010, and 2020, with a spatial resolution of 300 m; natural environmental factors influencing urban land use and population distribution (elevation, slope, river vector datasets); and socioeconomic factors (city center, major roads vector datasets). All factors were converted to raster data, and values were normalized to [0,1] to eliminate scale effects. The spatial resolution of the data was uniformly resampled to 1 km.
[0062] We collected a vector dataset of gridded population distribution, urban land use distribution, and spatial driving factors that influence changes in urban land use and population distribution based on satellite observations for the year. Using GIS software, we calculated the distance from each pixel to features within the region, generating a distance driving factor layer with a resolution of 1 km. We also collected 1 km resolution elevation data and used GIS software to calculate the slope of each pixel within the region, generating a slope driving factor layer. Finally, we obtained five static spatial driving factors.
[0063] One hundred thousand samples were obtained using stratified random sampling from urban land use distribution data in 2010 and data on five driving factors. These samples were then used to train an artificial neural network model. 60% of the collected samples were used as training samples, 20% as validation samples, and 20% as test samples. Subsequently, all driving factor data were input into the trained model, which output an urban land use development suitability layer for the study area.
[0064] The urban land use distribution data from 2010 and the urban development suitability layer output from step 2 are input into the FLUS model to simulate the future SSP scenario, i.e., urban land use distribution under a shared socioeconomic pathway. The total urban land use data provided by the SSP database is used as a constraint to terminate the model simulation, and the model parameters are kept at their default values.
[0065] The urban land use distribution in 2010 and the urban development suitability layer output from step 2 were input into the H-FLUS model to reconstruct the historical urban land use distribution. The total urban land use data provided by the HYDE database was used as a constraint to terminate the model simulation.
[0066] Using 2010 as the base year, and employing grid population distributions from 2010 and 2020, urban land use distribution from 2020, and static spatial driving factors, a population prediction model was trained based on a random forest model. The future population prediction model is represented as follows:
[0067] Pop t+10 =f p (env 0 Pop t Neigh (Pop) t Urban t+10 Neighborhood t+10 ))
[0068] In this example, f p It's a random forest model, with 100 decision trees and the maximum feature count rule set to sqrt. 0 Spatial driving factors such as elevation, slope, distance to the river, distance to the city center, and distance to the main road. Pop t This is the population distribution in 2010, Urban. t+10 This shows the urban land use distribution in 2020. (Neighborhood Population) t These represent the total population within the 3, 5, and 7-mole neighborhoods of the grid in 2010. (Neighborhood / Urban) t+10 Pop represents the total urban land area within a 3-moole grid neighborhood. t+10 This is the population distribution in 2020.
[0069] Using 2010 as the base year, a population reconstruction model was trained based on a random forest model using the grid population distribution of 2000 and 2010, the urban land use distribution of 2000, and static spatial driving factors to reconstruct the population distribution of 2000. Since historical population distribution is less affected by socioeconomic factors, only static natural environmental factors are considered here compared to step 5.
[0070] Using the constructed future population simulation model and historical population reconstruction model, based on the long-term urban land use distribution, the system iteratively predicts the future population distribution and reconstructs the historical population distribution to build a long-term grid population distribution. Figure 3This invention presents the urban land distribution of the Shanghai metropolitan area under typical scenarios and years from 1870 to 2100. The historical urban land reconstruction model H-FLUS proposed in this invention can effectively combine with existing FLUS models to construct consistent urban land distribution data over a long period and capture the process of small settlements gradually evolving into large urban clusters. Figure 4 This invention presents the population grid distribution of the Shanghai metropolitan area under typical scenarios and years from 1870 to 2100. The method proposed in this invention can effectively construct consistent grid population distribution data over a long period and capture the spatiotemporal dynamics of population distribution. Figure 5 The accuracy analysis results of the machine learning-based future population prediction model and historical population reconstruction model constructed in this invention are presented. Both the future simulation and historical reconstruction models have high R-values. 2 The low %RMSE indicates that the model has high fitting accuracy.
[0071] Compared with existing technologies, this invention effectively unifies the historical reconstruction and future simulation of population distribution, avoiding inconsistencies caused by separate modeling. By combining simulations of urban land use distribution data closely related to population distribution with high-performance machine learning techniques, this method effectively improves the accuracy of grid-based population distribution mapping. The constructed long-term consistent high-resolution grid-based population distribution data will provide important scientific references for disaster exposure assessment, dual-carbon targets, and the achievement of sustainable development goals.
[0072] Based on the above-described method for constructing long-term grid-based population distribution, this invention also provides a system for constructing long-term grid-based population distribution, such as... Figure 6 As shown, the system includes:
[0073] Data acquisition module 1 is used to acquire an initial urban land use distribution dataset, an initial population distribution dataset, and a driving factor dataset. The driving factor dataset includes elevation, slope, distance to the river, distance to the city center, and distance to the main road.
[0074] The urban land use distribution construction module 2 is used to establish and train an urban land use distribution model based on the initial urban land use distribution dataset, input the driving factor dataset into the trained urban land use distribution model to obtain urban land use development suitability data, and input the urban land use development suitability data and the initial urban land use distribution dataset into the FLUS model to construct the urban land use distribution.
[0075] Population distribution construction module 3 is used to build and train a population distribution model based on the initial population distribution dataset, and input the urban land use distribution and the initial population distribution dataset into the trained population distribution model to construct a long-term grid population distribution.
[0076] For specific limitations regarding a long-term grid-based population distribution construction system, please refer to the limitations of a long-term grid-based population distribution construction method described above, which will not be repeated here. Each module in the above system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.
[0077] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method described above.
[0078] Figure 7 This diagram illustrates the internal structure of a computer device in one embodiment, which may specifically be a terminal or a server. The computer device includes a processor, memory, a network interface, a display, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. The display screen may be a liquid crystal display (LCD) or an e-ink display. The input devices may be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device's casing, or an external keyboard, touchpad, or mouse.
[0079] Those skilled in the art will understand that Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specifically, the computing device may include more or fewer components than shown in the diagram, or combine certain components, or have the same component arrangement.
[0080] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.
[0081] In summary, this invention provides a method, system, device, and medium for constructing long-term grid-based population distribution. The method includes: acquiring an initial urban land use distribution dataset, an initial population distribution dataset, and a driving factor dataset; establishing and training an urban land use distribution model based on the initial urban land use distribution dataset; inputting the driving factor dataset into the trained urban land use distribution model to obtain urban land use development suitability data; inputting the urban land use development suitability data and the initial urban land use distribution dataset into a FLUS model to construct the urban land use distribution; establishing and training a population distribution model based on the initial population distribution dataset and the initial urban land use distribution dataset; inputting the urban land use distribution and the initial population distribution dataset into the trained population distribution model to construct a long-term grid-based population distribution. This invention effectively improves the accuracy of grid-based population distribution mapping.
[0082] The various embodiments in this specification are described in a progressive manner. For directly identical or similar parts of the embodiments, refer to each other. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.
[0083] The embodiments described above are merely preferred embodiments of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various improvements and substitutions without departing from the technical principles of this invention, and these improvements and substitutions should also be considered within the scope of protection of this application. Therefore, the scope of protection of this patent application should be determined by the scope of the claims.
Claims
1. A method for constructing a long-term grid-based population distribution, characterized in that, The method includes: Obtain an initial urban land use distribution dataset, an initial population distribution dataset, and a driving factor dataset, wherein the driving factor dataset includes elevation, slope, distance to the river, distance to the city center, and distance to the main road; An urban land use distribution model is established and trained based on the initial urban land use distribution dataset. The driving factor dataset is then input into the trained urban land use distribution model to obtain urban land use development suitability data. The urban land use development suitability data and the initial urban land use distribution dataset are then input into the FLUS model to construct the urban land use distribution. The step of inputting the urban land development suitability data and the initial urban land distribution dataset into the FLUS model includes: The urban land development suitability data and the initial urban land distribution dataset of historical years are input into the H-FLUS model to reconstruct the historical urban land distribution. The H-FLUS model is obtained by removing the pixel with the lowest total probability from the FLUS model. The total probability is the product of urban land development suitability and neighborhood effect. The step of inputting the driving factor dataset into the trained urban land use distribution model to obtain urban land use development suitability data includes: The intermediate layer node data is obtained by multiplying the network weights connecting the input layer nodes and the intermediate layer nodes in the urban land use distribution model by the driving factor data. The network weights connecting the intermediate layer nodes and the output layer nodes in the urban land use distribution model are multiplied by the data of the intermediate nodes. The value is used to obtain the output layer node data; Calculate the output layer node data The value serves as data on the suitability of urban land use development; A population distribution model is established and trained based on the initial population distribution dataset and the initial urban land distribution dataset. The urban land distribution and the initial population distribution dataset are then input into the trained population distribution model to construct a long-term grid population distribution.
2. The method for constructing a long-term grid-based population distribution according to claim 1, characterized in that, The acquisition of the driver factor dataset includes: The slope, elevation, distance to the river, distance to the city center, and distance to the main road of each pixel are calculated using GIS software.
3. The method for constructing a long-term grid-based population distribution according to claim 1, characterized in that, The acquisition of the driver factor dataset includes: The driving factor dataset is converted into raster data, normalized, and then resampled to a preset spatial resolution.
4. The method for constructing a long-term grid-based population distribution according to claim 1, characterized in that, The step of establishing and training a population distribution model based on the initial population distribution dataset and the initial urban land use distribution dataset includes: A population distribution model is established based on a random forest model, and the population distribution model is trained based on the initial population distribution dataset, the initial urban land use distribution dataset, and the driving factor dataset.
5. The method for constructing a long-term grid-based population distribution according to claim 1, characterized in that, The step of establishing and training an urban land use distribution model based on the initial urban land use distribution dataset includes: The total urban land use volume shared with the socio-economic path database will be used as a constraint to terminate the training.
6. A long-term grid-based population distribution construction system, characterized in that, The system includes: The data acquisition module is used to acquire an initial urban land use distribution dataset, an initial population distribution dataset, and a driving factor dataset. The driving factor dataset includes elevation, slope, distance to the river, distance to the city center, and distance to the main road. The urban land use distribution construction module is used to establish and train an urban land use distribution model based on the initial urban land use distribution dataset, input the driving factor dataset into the trained urban land use distribution model to obtain urban land use development suitability data, and input the urban land use development suitability data and the initial urban land use distribution dataset into the FLUS model to construct the urban land use distribution. The step of inputting the urban land development suitability data and the initial urban land distribution dataset into the FLUS model includes: The urban land development suitability data and the initial urban land distribution dataset of historical years are input into the H-FLUS model to reconstruct the historical urban land distribution. The H-FLUS model is obtained by removing the pixel with the lowest total probability from the FLUS model. The total probability is the product of urban land development suitability and neighborhood effect. The step of inputting the driving factor dataset into the trained urban land use distribution model to obtain urban land use development suitability data includes: The intermediate layer node data is obtained by multiplying the network weights connecting the input layer nodes and the intermediate layer nodes in the urban land use distribution model by the driving factor data. The network weights connecting the intermediate layer nodes and the output layer nodes in the urban land use distribution model are multiplied by the data of the intermediate nodes. The value is used to obtain the output layer node data; Calculate the output layer node data The value serves as data on the suitability of urban land use development; The population distribution construction module is used to build and train a population distribution model based on the initial population distribution dataset. The urban land use distribution and the initial population distribution dataset are input into the trained population distribution model to construct a long-term grid population distribution.
7. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements a long-term grid population distribution construction method according to any one of claims 1 to 5.
8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements a long-term grid population distribution construction method according to any one of claims 1 to 5.