Future cultivated land distribution prediction method and device, electronic equipment and storage medium

Abstract

The present application provides a kind of future cultivated land distribution prediction method, device, electronic equipment and storage medium, it is related to data processing technical field, the present application is first obtained multiple sets of land use products and cultivated land area statistical data, multiple sets of land use products include historical land use data product and multiple sets of future land use data product;Again, multiple sets of land use products are pretreated;Then, according to the cultivated land area calculation data and cultivated land area statistical data under the base year in the target cultivated land data after pretreatment, the precision sorting result of multiple sets of land use products is determined;Further, according to the precision sorting result and cultivated land area statistical data, the cultivated land distribution data in the target cultivated land data is fused and processed, and the future cultivated land distribution prediction result is determined.In this way, multiple sets of land use products are evaluated in precision, and multiple sets of land use products are fused based on cultivated land area statistical data, to realize high-precision cultivated land distribution prediction under future climate change scenario.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method, apparatus, electronic device, and storage medium for predicting future arable land distribution. Background Technology

[0002] With the continuous growth of the global population and the improvement of living standards, the demand for food will continue to increase. At the same time, climate change, environmental degradation, and increased construction land use will lead to a reduction in arable land area and a decline in arable land quality, seriously affecting agricultural production and national food security. Therefore, accurately predicting the future area and spatial distribution of arable land is crucial. It can help policymakers and stakeholders make correct decisions in land use planning, food production, and resource allocation, contributing to optimized crop layout, improved agricultural production efficiency, and ensuring national food security and sustainable development.

[0003] Existing research has conducted predictive studies on arable land area and spatial distribution under climate change scenarios based on the Shared Socioeconomic Pathways (SSPs) and Representative Concentration Pathways (RCPs) proposed by the Intergovernmental Panel on Climate Change (IPCC). For example, the Land-Use Harmonization 2 (LUH2) project developed a global land use data product with a spatial resolution of 0.25° from 850 AD to 2100 AD based on past land use transformation patterns. Furthermore, researchers have used methods such as Cellular Automata (CA), Markov Models, Patch-level Land Use Simulation (PLUS), and Future Land Use Simulation (FLUS) to generate land use data products with higher spatial resolution (1 km) for different future SSP-RCP scenarios.

[0004] The aforementioned future land use data products used different base year land use data products and land use change simulation models, resulting in certain differences in the future arable land distribution prediction results among the various data products. At the same time, there is a lack of evaluation on the consistency of future arable land prediction data products, which limits the application of future arable land prediction data products. Summary of the Invention

[0005] The purpose of this invention is to provide a method, apparatus, electronic device, and storage medium for predicting future arable land distribution, so as to improve the consistency and accuracy of future arable land distribution prediction.

[0006] In a first aspect, embodiments of the present invention provide a method for predicting future arable land distribution, including:

[0007] The system acquires multiple sets of land use products and arable land area statistics. These multiple sets of land use products include historical land use data products and multiple sets of future land use data products. The historical land use data products include land use data from historical years, and the future land use data products include land use data for predicted years under a preset climate change scenario. The arable land area statistics include arable land area statistics for a baseline year and arable land demand data for future years, where the future arable land demand data is the arable land area demand data under the stated climate change scenario.

[0008] The multiple sets of land use products are preprocessed to obtain target arable land data. The target arable land data includes arable land distribution data of the target study area under the target historical year corresponding to the historical land use data product, as well as arable land distribution data and arable land area calculation data of the target study area under the target prediction year and target climate change scenario corresponding to each set of future land use data products. The target historical year is the year of the previous cycle of the target prediction year, and the arable land area calculation data includes the arable land area calculation value at the target regional scale.

[0009] Based on the cultivated land area calculation data under the baseline year in the target cultivated land data and the cultivated land area statistics data under the baseline year, the accuracy ranking results of the multiple sets of land use products are determined; wherein, the baseline year is the starting year of the target prediction year;

[0010] Based on the accuracy ranking results and the cultivated land area statistics, the cultivated land distribution data in the target cultivated land data are fused to determine the future cultivated land distribution prediction results.

[0011] Secondly, embodiments of the present invention also provide a device for predicting future arable land distribution, comprising:

[0012] The data acquisition module is used to acquire multiple sets of land use products and arable land area statistics. The multiple sets of land use products include historical land use data products and multiple sets of future land use data products. The historical land use data products include land use data from historical years, and the future land use data products include land use data for predicted years under a preset climate change scenario. The arable land area statistics include arable land area statistics for a baseline year and arable land demand data for future years, wherein the arable land demand data for future years is the arable land area demand data under the stated climate change scenario.

[0013] The preprocessing module is used to preprocess the multiple sets of land use products to obtain target arable land data. The target arable land data includes arable land distribution data of the target study area under the target historical year corresponding to the historical land use data product, as well as arable land distribution data and arable land area calculation data of the target study area under the target prediction year and target climate change scenario corresponding to each set of future land use data products. The target historical year is the year of the previous cycle of the target prediction year, and the arable land area calculation data includes the arable land area calculation value at the target regional scale.

[0014] The accuracy determination module is used to determine the accuracy ranking result of the multiple sets of land use products based on the cultivated land area calculation data under the reference year in the target cultivated land data and the cultivated land area statistics data under the reference year; wherein, the reference year is the starting year of the target prediction year;

[0015] The result determination module is used to fuse the cultivated land distribution data in the target cultivated land data based on the accuracy sorting results and the cultivated land area statistics, and determine the future cultivated land distribution prediction results.

[0016] Thirdly, embodiments of the present invention also provide an electronic device, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the computer program to implement the future arable land distribution prediction method described in the first aspect.

[0017] Fourthly, embodiments of the present invention also provide a storage medium storing a computer program, wherein the computer program is executed by a processor to perform the future arable land distribution prediction method described in the first aspect.

[0018] The future arable land distribution prediction method, apparatus, electronic device, and storage medium provided in this invention, when predicting future arable land distribution, first acquire multiple sets of land use products and arable land area statistics. The multiple sets of land use products include historical land use data products and multiple sets of future land use data products. The historical land use data products include land use data from historical years, and the future land use data products include land use data for the predicted year under a preset climate change scenario. The arable land area statistics include arable land area statistics for a baseline year and arable land demand data for future years, where the future arable land demand data is arable land area demand data under a climate change scenario. Then, the multiple sets of land use products are preprocessed to obtain target arable land data, which includes historical land use data products. The system uses data on cultivated land distribution in the target study area for the corresponding historical year, as well as cultivated land distribution data and calculated cultivated land area data for the target study area under the target prediction year and target climate change scenario for each set of future land use data products. The target historical year is the year of the previous cycle before the target prediction year, and the calculated cultivated land area data includes the calculated cultivated land area at the target regional scale. Then, based on the calculated cultivated land area data for the base year in the target cultivated land data and the cultivated land area statistics data for the base year, the accuracy ranking results of multiple sets of land use products are determined. The base year is the starting year of the target prediction year. Subsequently, based on the accuracy ranking results and the cultivated land area statistics data, the cultivated land distribution data in the target cultivated land data are fused to determine the future cultivated land distribution prediction results. This process evaluates the accuracy of multiple sets of land use products and fuses them based on cultivated land area statistics data, achieving high-precision cultivated land distribution prediction under future climate change scenarios and improving the consistency and accuracy of future cultivated land distribution predictions. Attached Figure Description

[0019] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 A flowchart illustrating a method for predicting future arable land distribution provided in an embodiment of the present invention;

[0021] Figure 2 A flowchart illustrating another method for predicting future arable land distribution provided in an embodiment of the present invention;

[0022] Figure 3A flowchart illustrating the synthesis of future arable land distribution prediction data products using a confidence score table is provided in this embodiment of the invention.

[0023] Figure 4 This is a scatter plot of predicted and statistical values ​​of cultivated land area under different climate scenarios in 2020, provided by an embodiment of the present invention.

[0024] Figure 5 This is a schematic diagram of the structure of a future arable land distribution prediction device provided in an embodiment of the present invention;

[0025] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0026] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Currently, future land use data products use different base year land use data products and land use change simulation models, resulting in certain differences in the future arable land distribution prediction results among the various data products. At the same time, there is a lack of evaluation on the consistency of future arable land prediction data products. Based on this, the present invention provides a future arable land distribution prediction method, device, electronic device and storage medium, which can synthesize high-precision future arable land distribution prediction data products, improve the consistency and accuracy of future arable land distribution prediction, and improve computational efficiency.

[0028] Based on existing historical arable land data products (i.e., historical land use data products) and multiple sets of future arable land prediction data products (i.e., future land use data products), this invention establishes sorting rules based on the consistency and accuracy of future arable land prediction data products during data product synthesis, in order to generate more accurate future arable land distribution prediction data products. At the same time, a dedicated system has been developed to achieve rapid synthesis of future arable land distribution prediction data products and visualization of prediction results.

[0029] To facilitate understanding of this embodiment, a method for predicting future arable land distribution disclosed in this embodiment of the invention will first be described in detail.

[0030] This invention provides a method for predicting future arable land distribution, which can be executed by an electronic device with data processing capabilities. See also... Figure 1The diagram shows a flowchart of a method for predicting future arable land distribution. This method mainly includes the following steps S102 to S108:

[0031] Step S102: Obtain multiple sets of land use products and arable land area statistics.

[0032] Among them, multiple sets of land use products include historical land use data products and multiple sets of future land use data products. The historical land use data products include land use data for historical years, and the future land use data products include land use data for predicted years under preset climate change scenarios. The cultivated land area statistics include cultivated land area statistics for the baseline year and cultivated land demand data for future years. The cultivated land demand data for future years is cultivated land area demand data under climate change scenarios.

[0033] Land use products are derived from the interpretation of remote sensing image data or model simulations. Historical land use data products include, for example, information on cultivated land distribution in a specific region, year, geographic coordinate system, and projected coordinate system. Future land use data products include, for example, information on cultivated land distribution in a specific region, year, geographic coordinate system, projected coordinate system, and climate change scenario. Cultivated land distribution information indicates which pixels are classified as cultivated land. The pixel size is related to the resolution of the remote sensing image data used, such as 30 meters or 500 meters. Baseline year cultivated land area statistics include, for example, cultivated land area statistics for a specific region, year, and at different regional scales. Future year cultivated land demand data includes, for example, cultivated land area demand values ​​for a specific region, year, climate change scenario, and at a specific regional scale.

[0034] The forecast year includes a base year and future years. The base year is the starting year of the forecast, and the previous cycle is the historical year. For example, if the distribution of arable land is forecasted every 10 years, and the forecast year is 2020 (the base year), then the historical year is 2010, and future years can include 2030, 2040, 2050, etc. Climate change scenarios, also known as climate models, correspond to five climate change scenarios for a shared socioeconomic path (SSP) and four representative concentration paths (RCP). Each future land use data product may involve one or more of these climate change scenarios. It should be noted that this embodiment of the invention does not limit the number of climate change scenarios; in other embodiments, there may be fewer climate change scenarios or other types of climate change scenarios.

[0035] For statistical data on cultivated land area, the statistical data for the base year can be the cultivated land area statistics from the National Land Survey of the same year; if statistical data is lacking for future years, the cultivated land area statistics from the National Land Survey of the base year, calibrated with the LUH2 rate of change, can be used as the "statistical data" for cultivated land in future years, based on the Land Use Coordination Project 2 (LUH2) dataset.

[0036] Step S104: Preprocess multiple sets of land use products to obtain target arable land data.

[0037] The target arable land data includes arable land distribution data for the target study area under the target historical year corresponding to the historical land use data products, as well as arable land distribution data and arable land area calculation data for the target study area under the target prediction year and target climate change scenario corresponding to each set of future land use data products. The target historical year is the year of the cycle preceding the target prediction year, and the arable land area calculation data includes the calculated arable land area at the target regional scale. The target regional scale can be, but is not limited to, national, provincial, or county scales.

[0038] Through preprocessing, farmland data products with the same geographic coordinates and projection coordinates, year, spatial range, and climate change scenarios are generated, providing basic data for the synthesis of subsequent farmland prediction data products.

[0039] For cases where the target prediction year and target climate change scenario have been determined, step S104 above can be achieved through the following process: Obtain the target study area, target regional scale, target climate change scenario, target prediction year, and the target historical year corresponding to the target prediction year; perform coordinate system transformation on each set of land use products to obtain land use standard data under the preset benchmark geographic coordinate system and benchmark projected coordinate system corresponding to the land use products; extract arable land, trim the target study area, extract the target prediction year, extract the target climate change scenario, and calculate the arable land area at the target regional scale for the land use standard data corresponding to each set of future land use data products, to obtain arable land distribution data and arable land area calculation data for the target study area under the target prediction year and target climate change scenario corresponding to the future land use data products; extract arable land, trim the target study area, and extract the target historical year for the land use standard data corresponding to the historical land use data products, to obtain arable land distribution data for the target study area under the target historical year corresponding to the historical land use data products.

[0040] For cases where the target prediction year and target climate change scenario are not determined, step S104 above can be achieved through the following process: Obtain the target study area and target area scale; perform coordinate system transformation on each set of land use products to obtain land use standard data under the preset benchmark geographic coordinate system and benchmark projected coordinate system corresponding to the land use products; extract arable land, crop the target study area, and calculate the arable land area at the target area scale from the land use standard data corresponding to each set of future land use data products to obtain intermediate data corresponding to the future land use data products; based on the same prediction year and the same climate change scenario in the intermediate data corresponding to each set of future land use data products... The system identifies target climate change scenarios, target prediction years, and corresponding target historical years. For each future land use data product, the intermediate data is used to extract the target prediction year and target climate change scenario, resulting in the target prediction year, cultivated land distribution data for the target study area under the target climate change scenario, and cultivated land area calculation data. For historical land use data products, the system extracts cultivated land, trims the target study area, and extracts the target historical year from the land use standard data, resulting in cultivated land distribution data for the target study area under the target historical year. The target prediction year can be the same prediction year for each future land use data product, the target historical year is the year of the previous cycle, and the target climate change scenario can be the same climate change scenario for each future land use data product.

[0041] The aforementioned reference geographic coordinate system and reference projected coordinate system can be the geographic coordinate system and projected coordinate system corresponding to any set of land use products.

[0042] Step S106: Based on the calculated cultivated land area data for the base year in the target cultivated land data and the cultivated land area statistics data for the base year, determine the accuracy ranking results of multiple sets of land use products.

[0043] The base year is the starting year for the target prediction year. The final generated prediction of future arable land distribution is a synthesis of historical arable land classification data products (i.e., historical land use data products) and multiple sets of future arable land prediction data products (i.e., future land use data products). In the synthesis process, it is necessary to first determine the accuracy of different arable land data products.

[0044] Historical farmland classification data products based on remote sensing data interpretation are considered to have higher accuracy than future farmland prediction data products. The accuracy of future farmland prediction data products can be assessed by the consistency between the farmland prediction data of the baseline year and the farmland area statistics of the national land survey in the same year. Based on this, in some possible embodiments, the above step S106 can be implemented through the following process: Calculate the prediction accuracy value of the future land use data product in the baseline year based on the baseline year corresponding to each set of future land use data products in the target farmland data, the farmland area calculation data of the target study area under the target climate change scenario, and the farmland area statistics of the target study area in the baseline year; determine the accuracy ranking result of multiple sets of land use products based on the prediction accuracy values ​​of each set of future land use data products in the baseline year; among them, the historical land use data products have the highest accuracy.

[0045] Optionally, the accuracy value corresponding to the future land use data product can be calculated as follows: For each set of future land use data products in the target cultivated land data, the root mean square error (RMSE) is calculated for the base year, the calculated cultivated land area of ​​the target study area under the target climate change scenario, and the statistical data of the cultivated land area of ​​the target study area under the base year. This yields the prediction accuracy value of the future land use data product in the base year. It should be noted that the accuracy value corresponding to the future land use data product is not limited to the calculation method based on RMSE. In other embodiments, other methods for measuring accuracy can also be used, such as calculation methods based on absolute coefficients, mean square error, average absolute error, or average absolute error percentage.

[0046] Step S108: Based on the accuracy sorting results and cultivated land area statistics, the cultivated land distribution data in the target cultivated land data is fused to determine the future cultivated land distribution prediction results.

[0047] A confidence score table can be generated based on the consistency and accuracy ranking results of cultivated land data products (i.e., land use products), and the ranking score of each pixel can be calculated. Using a country, province, or county as the administrative region, the cultivated land area within each pixel under different confidence scores is statistically analyzed. The cultivated land areas under different confidence scores are accumulated from highest to lowest until the accumulated cultivated land area most closely approximates the statistical value of the cultivated land area for that administrative region. Finally, multiple sets of cultivated land products are combined into a single future cultivated land distribution prediction data product. Based on this, step S108 can include the following sub-steps 1 to 6:

[0048] Sub-step 1: Generate a confidence score table based on the accuracy ranking results; in the confidence score table, the more land use products that are predicted to be cultivated land, and the higher the accuracy ranking, the higher the confidence score of the pixel.

[0049] Sub-step 2: For each prediction year of the target prediction year, determine the confidence score of each pixel in the target study area under the prediction year based on the confidence score table and the cultivated land distribution data corresponding to the prediction year in the target cultivated land data.

[0050] Sub-step 3: Based on the confidence scores of each pixel in the target study area under the predicted year, calculate the cultivated land area of ​​each pixel under each confidence score to obtain the cultivated land area value corresponding to each confidence score under the predicted year.

[0051] Sub-step 4: According to the confidence scores from high to low, the cultivated land area values ​​corresponding to different confidence scores in the prediction year are accumulated sequentially to obtain the accumulated cultivated land area value in the prediction year.

[0052] Sub-step 5: When the cumulative cultivated land area value under the predicted year is closest to the corresponding cultivated land area statistical value in the cultivated land area statistical data, the cultivated land distribution result corresponding to the cumulative cultivated land area value under the predicted year is determined as the cultivated land distribution prediction result under the predicted year.

[0053] Sub-step 6: The predicted farmland distribution results for the predicted year are used as the farmland distribution data for the target historical year corresponding to the next cycle prediction year, so as to determine the farmland distribution prediction results for the next cycle prediction year.

[0054] In this way, we can obtain the predicted results of arable land distribution for each predicted year, that is, the predicted results of future arable land distribution.

[0055] The future arable land distribution prediction method provided in this invention first acquires multiple sets of land use products and arable land area statistics when predicting future arable land distribution. The multiple sets of land use products include historical land use data products and multiple sets of future land use data products. The historical land use data products include land use data from historical years, and the future land use data products include land use data for the predicted year under a preset climate change scenario. The arable land area statistics include arable land area statistics for a baseline year and arable land demand data for future years, where the future arable land demand data is arable land area demand data under a climate change scenario. Then, the multiple sets of land use products are preprocessed to obtain target arable land data, which includes target historical land use data corresponding to the historical land use data products. The system uses cultivated land distribution data for the target study area under the given year, as well as cultivated land distribution data and calculated cultivated land area data for the target study area under the target climate change scenario for each set of future land use data products. The target historical year is the year of the previous cycle before the target prediction year, and the calculated cultivated land area data includes the calculated cultivated land area at the target regional scale. Then, based on the calculated cultivated land area data for the base year in the target cultivated land data and the cultivated land area statistics data for the base year in the cultivated land area statistics data, the accuracy ranking results of multiple sets of land use products are determined. The base year is the starting year of the target prediction year. Subsequently, based on the accuracy ranking results and the cultivated land area statistics data, the cultivated land distribution data in the target cultivated land data are fused to determine the future cultivated land distribution prediction results. In this way, the accuracy of multiple sets of land use products is evaluated, and multiple sets of land use products are fused based on cultivated land area statistics data, achieving high-precision cultivated land distribution prediction under the future climate change scenario and improving the consistency and accuracy of future cultivated land distribution prediction.

[0056] Furthermore, the accuracy of the aforementioned future arable land distribution prediction results can be verified. Based on this, the method further includes: calculating the index value corresponding to the preset accuracy index based on the predicted arable land area at the target region scale corresponding to the base year in the future arable land distribution prediction results and the statistical value of arable land area at the target region scale corresponding to the base year in the arable land area statistics; and determining the accuracy verification result of the future arable land distribution prediction results based on the index value corresponding to the preset accuracy index. The preset accuracy index may include one or more of the following: absolute coefficient, root mean square error, and mean absolute error.

[0057] To facilitate understanding, the above-mentioned methods for predicting future arable land distribution will be explained in detail below.

[0058] To address the discrepancies in arable land distribution predictions across multiple sets of future land use data, this invention proposes a method and system that utilizes the consistency of pixel identification (whether a pixel represents arable land) and the accuracy of the data products from multiple future arable land prediction data products to generate a ranking table. This table is used to estimate the probability that a pixel represents arable land, ultimately synthesizing a more accurate arable land data product. This provides high-precision arable land distribution data products for future climate change scenarios. This invention also includes a dedicated system comprising a high-performance computer, data analysis software, and visualization tools to enable rapid analysis and synthesis of future arable land distribution prediction data products, as well as the visualization of the prediction results.

[0059] To achieve the above objectives, the main process of the technical solution adopted in the embodiments of the present invention is as follows: Figure 2 As shown, it mainly includes the following six steps:

[0060] S1. Collection of land use data products and cultivated land statistics. The collection and use of land use products for historical years, baseline years, and future years, as well as cultivated land statistics, required in this embodiment of the invention, are as follows:

[0061] S11: Collection of historical land use data products. Historical land use data products based on remote sensing data interpretation for the study area can be selected from publicly available online data. The historical land use data products are land use type classification products for the previous period of the base year, and the classification results include cultivated land. This data product is used to synthesize the cultivated land prediction data product for the base year.

[0062] S12: Collection of Land Use Data Products for Baseline and Future Years. Multiple sets of high spatial resolution land use data products for the study area for the baseline and future years were selected from publicly available online data. The classification results include cultivated land. This data product was used to synthesize future cultivated land distribution prediction data products.

[0063] S13: Cultivated Land Statistics. The statistical data for the base year uses the cultivated land area statistics from the National Land Survey of the same year; for future years where statistical data is lacking, the cultivated land area statistics from the National Land Survey of the base year, calibrated using the Land Use Coordination Project 2 (LUH2) dataset, are used as the cultivated land "statistics" for future years (i.e., cultivated land demand data for future years).

[0064] Future farmland demand data can be calculated using the following formula:

[0065]

[0066] Among them, A ij For province i, the "statistical value of cultivated land area" (i.e., the demand value of cultivated land area) in future year j, A ijThe statistical value of cultivated land area A from the national land survey in province i under the baseline year. ibase And the predicted cultivated land area L for year j in the LUH2 dataset. ij Compared with the predicted cultivated land area L in the baseline year ibase The rate of change is used for correction.

[0067] S2. Preprocessing of Land Use Data Products. This involves generating farmland data products with the same geographic and projected coordinates, year, spatial extent (i.e., study area), and climate change scenarios, providing foundational data for the subsequent synthesis of farmland prediction data products. In this embodiment, the preprocessing of historical land use data products includes only three sub-steps (S21-S23): geographic and projected coordinate transformation, farmland distribution extraction, and study area cropping. Future farmland prediction data products, however, also include five sub-steps (S21-S25) for extracting common prediction years and future climate change scenarios.

[0068] S21: Geographic Coordinate System and Projected Coordinate System Conversion. Determine the geographic coordinate system and projected coordinate system for each set of land use data products; select one data product as the baseline data product, and use its geographic coordinate system and projected coordinate system as the baseline geographic coordinate system and baseline projected coordinate system. For other data products, use the "Convert" or "Project" tool in the GIS software (the name of this tool may vary in different versions of GIS software; possible names include "Convert" or "Project") to convert them to the baseline geographic coordinate system and baseline projected coordinate system.

[0069] S22: Extract arable land distribution data and calculate arable land area. Identify the land use classification standards used in each set of land use products, extract the corresponding land use types for arable land from the land use products, generate arable land spatial distribution data products, and use the area statistics function of GIS software to calculate the corresponding arable land area at the national, provincial, or county level in each set of data.

[0070] S23: Study Area Clipping. Using the vector boundaries of the study area, the spatial distribution data of each cultivated land area are clipped to obtain data with a uniform spatial extent.

[0071] S24: Extract farmland data for common forecast years. Define the base year, time range, and time interval for each set of future land use products, extract the same years from each set of farmland forecast data products, and maintain consistency in the base year, time range, and time interval across multiple sets of farmland forecast data products.

[0072] S25: Extract farmland prediction data products based on common climate change scenarios. Identify the shared socio-economic development paths and representative concentration paths of climate change scenarios involved in each set of farmland prediction products, and extract the common climate change scenarios in each set of future farmland prediction data products.

[0073] S3. Accuracy Assessment of Farmland Data Products. The final farmland prediction result is a synthesis of historical farmland classification data products and multiple future farmland prediction data products. During the synthesis, the accuracy of different farmland data products needs to be determined first. Historical farmland classification data products based on remote sensing data interpretation are considered to have higher accuracy than future farmland prediction data products. The accuracy of future farmland prediction data products is assessed by the consistency between the farmland prediction data of the baseline year and the farmland area statistics from the National Land Survey of the same year. The accuracy comparison of future farmland prediction data products involves the following two sub-steps:

[0074] S31: Calculation of cultivated land area in the base year. At the national, provincial, or county scale (the choice of scale is related to the size of the study area; if the study area is a country, the scale can be a province, city, or county, etc.; if the study area is a province, the scale can be a city or county, etc.), calculate the cultivated land area for the base year of each set of cultivated land prediction products. The cultivated land area is statistically analyzed using the area statistics function of GIS software.

[0075] S32: Accuracy assessment of future arable land prediction data products in the base year. Taking the provincial scale as an example, the accuracy of future arable land prediction data products is tested using the arable land area statistics of each province from the National Land Survey in the base year. The root mean square error (RMSE) between the predicted value and the statistical value is calculated to quantify the prediction accuracy of each set of arable land area data.

[0076]

[0077] In the formula, P i A represents the predicted cultivated land area for province i in the future cultivated land prediction data product. ibase The arable land area is the statistical value of province i in the national land survey of the base year, and n is the number of provinces included in the survey.

[0078] S4. Sort and synthesize future arable land distribution prediction data products. Based on the consistency and accuracy sorting results of the arable land data products, generate a confidence score table and calculate the confidence score for each pixel. Using countries, provinces, or counties as administrative regions, statistically analyze the arable land area within pixels under different confidence scores. Calculate the arable land area under different confidence scores from highest to lowest until the accumulated arable land area most closely matches the statistical value of the arable land area for that administrative region. This determines the optimal confidence score for each administrative region. Finally, synthesize multiple sets of arable land products into a single future arable land distribution prediction data product. The synthesis process is illustrated below using three sets of future arable land prediction data products A, B, and C, and historical arable land data product D as examples. The specific process is as follows:

[0079] S41: Generate a confidence score table. Based on the consistency of whether a pixel is farmland (1 for farmland, 0 for non-farmland) across different farmland data products and the accuracy of the data products, a confidence score table is generated to estimate the probability that a pixel is farmland. Table 1 lists the confidence score tables generated using three sets of future farmland prediction data products (A, B, and C) and one set of historical farmland product D as examples. There are 16 possible ranking combinations (22) for a pixel to be farmland across the four farmland products A, B, C, and D. 4 If a pixel is classified as farmland in all four sets of farmland data products (A, B, C, and D), then that pixel is considered to have the highest probability of being farmland in the future, with a confidence score of 16. Figure 3 The first cell in the first row and first column); if a cell is farmland in any three of the four product sets A, B, C, and D, and according to the farmland data product precision D>B>C>A, the cell is farmland in farmland products B, C, and D, but not farmland in farmland product A, and the confidence score is 15 points (e.g., Figure 3 The probability of whether a pixel will become arable land in the future is estimated based on the confidence score table, and the arable land distribution confidence score map is obtained.

[0080] Table 1

[0081] 1 1 1 1 1 16 2 1 0 1 1 15 3 1 1 1 0 14 4 1 1 0 1 13 5 1 0 1 0 12 6 1 0 0 1 11 7 1 1 0 0 10 8 1 0 0 0 9 9 0 1 1 1 8 10 0 0 1 1 7 11 0 1 1 0 6 12 0 1 0 1 5 13 0 0 1 0 4 14 0 0 0 1 3 15 0 1 0 0 2 16 0 0 0 0 1

[0082] Note: In columns A, B, C, and D, 1 and 0 represent cultivated land and non-cultivated land, respectively. The accuracy of the cultivated land data product is D>B>C>A.

[0083] S42: Generate future arable land distribution prediction data products. Using countries, provinces, or counties as administrative regions, calculate the arable land area under different confidence scores. Accumulate these scores sequentially from highest to lowest until the accumulated arable land area most closely approximates the statistical value of arable land area for that administrative region. This determines the optimal confidence score for each administrative region, which is then used to synthesize the future arable land distribution prediction data products. Taking a province as an administrative region, and using the four sets of arable land products A, B, C, and D as examples, the process begins by calculating the arable land area within pixels with a confidence score of 16 (meaning all four data products A, B, C, and D are arable land). If the calculated arable land area within the administrative region is greater than or equal to the statistical value of arable land area, it indicates that the arable land area corresponding to the pixel with a confidence score of 16 is closest to the statistical value, and the confidence score of 16 is determined as the threshold for distinguishing whether a pixel in that administrative region is arable land. Conversely, if the confidence score is lower, the arable land area within pixels with a confidence score of 15 is further accumulated, and so on, until the accumulated arable land area is closest to the statistical value. The iteration ends, and this confidence score is determined as the threshold for distinguishing whether a pixel in that administrative region is arable land, used to synthesize future arable land distribution prediction data products. Following the above steps, arable land prediction data products for the base year and different future years are synthesized sequentially. The arable land area for the base year uses the arable land area statistics data (A) from the National Land Survey of the same year. ibase For future years where there is a lack of arable land area statistics, the arable land area statistics from the benchmark national land survey calibrated with the LUH2 dataset are used as the “arable land area statistics”.

[0084] For example, such as Figure 3 As shown, a farmland distribution confidence score map can be generated from the predicted farmland distribution maps A, B, and C and the historical farmland distribution map D. The farmland area (i.e., the cumulative farmland area value, also known as the predicted value) of pixels with a confidence score ≥ F is calculated, where the initial value of F is 16. The F value is determined based on Min(predicted value, actual value), thus obtaining the future farmland distribution prediction data product based on the F value. When determining the F value, it is necessary to determine whether the predicted value is closest to the corresponding actual value (i.e., the statistical value of farmland area). If not, F is set to F-1, and the step of calculating the farmland area of ​​pixels with a confidence score ≥ F is re-executed. If yes, the current F is used as the threshold for distinguishing whether a pixel is farmland.

[0085] S5. Accuracy Verification of Future Cultivated Land Distribution Forecast Data Products. Taking the 2020 provincial scale as an example, evaluate the difference between the predicted cultivated land area in the synthesized future cultivated land distribution forecast data products and the cultivated land statistics from the Third National Land Survey, and calculate the absolute coefficient R. 2 Accuracy is evaluated using root mean square error (RMSE) and mean absolute error (MAE).

[0086]

[0087]

[0088]

[0089] In the formula, P i For the future arable land distribution prediction data product, A represents the predicted arable land area of ​​province i in 2020. ibase This refers to the statistical value of cultivated land area for province i in the Third National Land Survey. is the average cultivated land area of ​​all provinces in the Third National Land Survey, and n is the number of provinces included in the survey.

[0090] S6. Visualization of future arable land distribution prediction data products. This system includes:

[0091] S61: High-performance computer. A documentation document used for preprocessing, accuracy evaluation, sorting, and synthesis of multiple land use products involved in the embodiments of this invention, as well as for storing related data.

[0092] S62: Data analysis software. Used to achieve (1) preprocessing of multiple sets of land use data, including geographic and projected coordinate transformation, year, spatial range, climate change scenario, and arable land data extraction; (2) accuracy evaluation of multiple sets of arable land prediction data products and generation of confidence score tables; (3) sorting and synthesizing future arable land distribution prediction data products; (4) accuracy verification of future arable land distribution prediction data products.

[0093] S63: Visualization tool. Used to display synthesized future arable land distribution prediction data products, and can output a map of future arable land distribution.

[0094] The embodiments of the present invention have the following innovative features:

[0095] (1) Based on the results of the Third National Land Survey, this invention evaluates the accuracy of the distribution of cultivated land in multiple existing future land use data products.

[0096] (2) Based on the consistency and accuracy of multiple sets of future land use prediction results and the inertia of cultivated land (historical cultivated land data products), this invention designs a method to estimate the probability of a pixel being cultivated land, and finally synthesizes a prediction data product of cultivated land distribution under different future climate change scenarios.

[0097] The embodiments of the present invention have the following beneficial effects:

[0098] This invention proposes a method and system for generating future arable land distribution prediction data products, which are used to synthesize high-precision future arable land distribution prediction data products, improve the consistency and accuracy of future arable land distribution prediction, and enhance computational efficiency.

[0099] To facilitate understanding, the above-mentioned method for predicting future arable land distribution will be introduced as an example below.

[0100] (1) Perform S1 to collect multiple sets of land use products that need to be used, including land use data product D for the historical year (2010), three sets of future land use data products A, B and C, and the cultivated land area statistics of the third national land survey for the baseline year (2020).

[0101] (2) Execute S2 to preprocess the four sets of land use data products, including geographic and projected coordinate transformation, unification of year, spatial range and climate change scenarios, and extraction of cultivated land. Combined with the land use prediction products selected in this embodiment, the multiple sets of data are finally unified as follows: Albers_Conical_Equal_Area projected coordinate system, WGS-1984 geographic coordinate system, time range 2020-2100 (interval of 10 years), and part of China's land area. The climate change scenarios are SSP126, SSP245, SSP370, SSP434 and SSP534.

[0102] (3) Perform S3 and use the statistical data of cultivated land area from the third national land survey in the base year (2020) to test the predicted value of cultivated land area in 2020 in the three sets of future cultivated land prediction data products A, B and C. The historical cultivated land data in D is the cultivated land data product interpreted from remote sensing data in 2010. The final accuracy result is D>B>C>A.

[0103] (4) Execute S4 to generate a confidence score table based on the consistency and accuracy of the cultivated land data products, and synthesize a future cultivated land distribution prediction data product. Based on the confidence score table, calculate the predicted area of ​​cultivated land at different confidence scores at the provincial level, accumulate the confidence scores from high to low, select the confidence score that is closest to the statistical value of cultivated land area of ​​the administrative region as the threshold for distinguishing cultivated land, and synthesize a future cultivated land distribution prediction data product.

[0104] (5) Execute S5 to verify the accuracy of the synthesized cultivated land distribution prediction data product. Taking the baseline year (2020) as an example, the scatter plot of the predicted cultivated land area of ​​each province and the statistical value of cultivated land area from the Third National Land Survey under different climate scenarios in 2020 (a:SSP119, b:SSP126, c:SSP245, d:SSP434, e:SSP534) is shown below. Figure 4 As shown, in this embodiment, the predicted cultivated land area at the provincial administrative level under the five climate change scenarios is compared with the cultivated land area statistics from the Third National Land Survey. The synthesized cultivated land prediction data product is as follows: R 2 >0.91, RMSE <14,400 hectares, MAE <0.998 million hectares.

[0105] (6) Execute S6, using a future arable land distribution prediction data product generation system to realize the rapid synthesis and statistical analysis of arable land distribution prediction data products, as well as the visualization of prediction results.

[0106] Corresponding to the above-described method for predicting future arable land distribution, this invention also provides a device for predicting future arable land distribution. (See attached image) Figure 5 The diagram shows a structural schematic of a future arable land distribution prediction device, which includes:

[0107] The data acquisition module 501 is used to acquire multiple sets of land use products and arable land area statistics. The multiple sets of land use products include historical land use data products and multiple sets of future land use data products. The historical land use data products include land use data from historical years, and the future land use data products include land use data for predicted years under a preset climate change scenario. The arable land area statistics include arable land area statistics for a baseline year and arable land demand data for future years, wherein the arable land demand data for future years is the arable land area demand data under the aforementioned climate change scenario.

[0108] The preprocessing module 502 is used to preprocess the multiple sets of land use products to obtain target arable land data. The target arable land data includes arable land distribution data of the target study area under the target historical year corresponding to the historical land use data product, as well as arable land distribution data and arable land area calculation data of the target study area under the target prediction year and target climate change scenario corresponding to each set of future land use data products. The target historical year is the year of the previous cycle of the target prediction year, and the arable land area calculation data includes the arable land area calculation value at the target regional scale.

[0109] The accuracy determination module 503 is used to determine the accuracy ranking result of the multiple sets of land use products based on the cultivated land area calculation data under the reference year in the target cultivated land data and the cultivated land area statistics data under the reference year; wherein, the reference year is the starting year of the target prediction year;

[0110] The result determination module 504 is used to perform fusion processing on the cultivated land distribution data in the target cultivated land data based on the precision sorting result and the cultivated land area statistics, and determine the future cultivated land distribution prediction result.

[0111] Furthermore, in some possible embodiments, the preprocessing module 502 described above is specifically used for:

[0112] Obtain the target study area, target regional scale, target climate change scenario, target predicted year, and the target historical year corresponding to the target predicted year;

[0113] For each set of land use products, coordinate system transformation is performed to obtain the land use standard data under the preset benchmark geographic coordinate system and benchmark projection coordinate system corresponding to the land use product;

[0114] For each set of future land use data products, the standard land use data corresponding to the future land use data products are processed by extracting arable land, cropping the target study area, extracting the target prediction year, extracting the target climate change scenario, and calculating the arable land area at the target regional scale, to obtain the arable land distribution data and arable land area calculation data of the target prediction year, the target climate change scenario, and the target study area corresponding to the future land use data products.

[0115] The arable land is extracted from the land use standard data corresponding to the historical land use data product, the target study area is cropped, and the target historical year is extracted to obtain the arable land distribution data of the target study area under the target historical year corresponding to the historical land use data product.

[0116] Furthermore, in some other possible embodiments, the preprocessing module 502 described above is specifically used for:

[0117] Obtain the target study area and the scale of the target area;

[0118] For each set of land use products, coordinate system transformation is performed to obtain the land use standard data under the preset benchmark geographic coordinate system and benchmark projection coordinate system corresponding to the land use product;

[0119] For each set of future land use data products, the standard land use data corresponding to the future land use data products are processed by extracting cultivated land, cropping the target study area, and calculating the cultivated land area at the target area scale to obtain intermediate data corresponding to the future land use data products.

[0120] Based on the same predicted year and the same climate change scenario in the intermediate data corresponding to each set of future land use data products, determine the target climate change scenario, the target predicted year, and the target historical year corresponding to the target predicted year.

[0121] For each set of future land use data products, the intermediate data corresponding to the target prediction year and the target climate change scenario are extracted to obtain the target prediction year, the cultivated land distribution data and the cultivated land area calculation data of the target study area under the target climate change scenario corresponding to the future land use data products;

[0122] The arable land is extracted from the land use standard data corresponding to the historical land use data product, the target study area is cropped, and the target historical year is extracted to obtain the arable land distribution data of the target study area under the target historical year corresponding to the historical land use data product.

[0123] Furthermore, the aforementioned accuracy determination module 503 is specifically used to: calculate the prediction accuracy value of the future land use data product corresponding to the base year for each set of the future land use data products in the target cultivated land data, the cultivated land area calculation data of the target study area under the target climate change scenario, and the cultivated land area statistics data of the target study area under the base year;

[0124] Based on the prediction accuracy values ​​of each set of future land use data products under the base year, the accuracy ranking of the multiple sets of land use products is determined; among them, the historical land use data products have the highest prediction accuracy.

[0125] Furthermore, the aforementioned accuracy determination module 503 is also used to: calculate the root mean square error of the base year, the calculated farmland area of ​​the target study area under the target climate change scenario, and the farmland area statistics of the target study area under the base year for each set of future land use data products in the target farmland data, so as to obtain the prediction accuracy value of the future land use data product under the base year.

[0126] Furthermore, the above result determination module 504 is specifically used for:

[0127] Based on the accuracy ranking results, a confidence score table is generated; wherein, in the confidence score table, the more land use products that predict the pixels to be cultivated land, and the higher the accuracy ranking, the higher the confidence score of the pixels.

[0128] For each prediction year of the target prediction year, the confidence score of each pixel in the target study area under the prediction year is determined according to the confidence score table and the cultivated land distribution data corresponding to the prediction year in the target cultivated land data;

[0129] Based on the confidence scores of each pixel in the target study area under the predicted year, the cultivated land area of ​​each pixel under each confidence score is calculated to obtain the cultivated land area value corresponding to each confidence score under the predicted year.

[0130] The cultivated land area values ​​corresponding to different confidence scores in the predicted year are accumulated sequentially according to the confidence scores from high to low to obtain the accumulated cultivated land area value in the predicted year.

[0131] When the cumulative cultivated land area value in the predicted year is closest to the corresponding cultivated land area statistical value in the cultivated land area statistical data, the cultivated land distribution result corresponding to the cumulative cultivated land area value in the predicted year is determined as the cultivated land distribution prediction result in the predicted year.

[0132] The predicted farmland distribution results for the predicted year are used as the farmland distribution data for the target historical year corresponding to the next cycle prediction year, so as to determine the predicted farmland distribution results for the next cycle prediction year.

[0133] Furthermore, the aforementioned device also includes:

[0134] The accuracy verification module is used to: calculate the index value corresponding to the preset accuracy index based on the predicted value of cultivated land area at the target area scale corresponding to the base year in the future cultivated land distribution prediction results and the statistical value of cultivated land area at the target area scale corresponding to the base year in the cultivated land area statistics; and determine the accuracy verification result of the future cultivated land distribution prediction results based on the index value corresponding to the preset accuracy index.

[0135] The future arable land distribution prediction device provided in this embodiment has the same implementation principle and technical effect as the aforementioned future arable land distribution prediction method embodiment. For the sake of brevity, any parts not mentioned in the future arable land distribution prediction device embodiment can be referred to the corresponding content in the aforementioned future arable land distribution prediction method embodiment.

[0136] like Figure 6 As shown, an electronic device 600 provided in this embodiment of the invention includes: a processor 601, a memory 602 and a bus. The memory 602 stores a computer program that can run on the processor 601. When the electronic device 600 is running, the processor 601 and the memory 602 communicate through the bus. The processor 601 executes the computer program to realize the above-mentioned method for predicting the distribution of future arable land.

[0137] Specifically, the memory 602 and processor 601 mentioned above can be general-purpose memory and processor, without any specific limitations here.

[0138] This invention also provides a storage medium storing a computer program, which, when executed by a processor, performs the future arable land distribution prediction method described in the preceding method embodiments. The storage medium includes various media capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory (ROM), RAM, magnetic disk, or optical disk.

[0139] In all examples shown and described herein, any specific values ​​should be interpreted as merely exemplary and not as limitations; therefore, other examples of exemplary embodiments may have different values.

[0140] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0141] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0142] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0143] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0144] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting future arable land distribution, characterized in that, include: The system acquires multiple sets of land use products and arable land area statistics. These multiple sets of land use products include historical land use data products and multiple sets of future land use data products. The historical land use data products include land use data from historical years, and the future land use data products include land use data for predicted years under a preset climate change scenario. The arable land area statistics include arable land area statistics for a baseline year and arable land demand data for future years, where the future arable land demand data is the arable land area demand data under the stated climate change scenario. The multiple sets of land use products are preprocessed to obtain target arable land data. The target arable land data includes arable land distribution data of the target study area under the target historical year corresponding to the historical land use data product, as well as arable land distribution data and arable land area calculation data of the target study area under the target prediction year and target climate change scenario corresponding to each set of future land use data products. The target historical year is the year of the previous cycle of the target prediction year, and the arable land area calculation data includes the arable land area calculation value at the target regional scale. Based on the cultivated land area calculation data under the baseline year in the target cultivated land data and the cultivated land area statistics data under the baseline year, the accuracy ranking results of the multiple sets of land use products are determined; wherein, the baseline year is the starting year of the target prediction year; Based on the precision ranking results and the cultivated land area statistics, the cultivated land distribution data in the target cultivated land data is fused to determine the future cultivated land distribution prediction results, including: generating a confidence score table based on the precision ranking results; wherein, in the confidence score table, the more land use products predicted as cultivated land there are, and the higher the precision ranking, the higher the confidence score of that pixel; for each prediction year of the target prediction year, based on the confidence score table and the cultivated land distribution data corresponding to the prediction year in the target cultivated land data, the confidence score of each pixel in the target study area under the prediction year is determined; based on the confidence score of each pixel in the target study area under the prediction year, the confidence score under each confidence score is calculated separately. The cultivated land area is calculated by using the pixel cultivated land area to obtain the cultivated land area value corresponding to each confidence score in the predicted year; the cultivated land area values ​​corresponding to different confidence scores in the predicted year are accumulated sequentially according to the order of the confidence scores from high to low to obtain the accumulated cultivated land area value in the predicted year; when the accumulated cultivated land area value in the predicted year is closest to the corresponding cultivated land area statistical value in the cultivated land area statistical data, the cultivated land distribution result corresponding to the accumulated cultivated land area value in the predicted year is determined as the cultivated land distribution prediction result in the predicted year; the cultivated land distribution prediction result in the predicted year is determined as the cultivated land distribution data for the target historical year corresponding to the next cycle prediction year, so as to determine the cultivated land distribution prediction result corresponding to the next cycle prediction year.

2. The method for predicting future arable land distribution according to claim 1, characterized in that, The preprocessing of the multiple sets of land use products to obtain target arable land data includes: Obtain the target study area, target regional scale, target climate change scenario, target predicted year, and the target historical year corresponding to the target predicted year; For each set of land use products, coordinate system transformation is performed to obtain the land use standard data under the preset benchmark geographic coordinate system and benchmark projection coordinate system corresponding to the land use product; For each set of future land use data products, the standard land use data corresponding to the future land use data products are processed by extracting arable land, cropping the target study area, extracting the target prediction year, extracting the target climate change scenario, and calculating the arable land area at the target regional scale, to obtain the arable land distribution data and arable land area calculation data of the target prediction year, the target climate change scenario, and the target study area corresponding to the future land use data products. The arable land is extracted from the land use standard data corresponding to the historical land use data product, the target study area is cropped, and the target historical year is extracted to obtain the arable land distribution data of the target study area under the target historical year corresponding to the historical land use data product.

3. The method for predicting future arable land distribution according to claim 1, characterized in that, The preprocessing of the multiple sets of land use products to obtain target arable land data includes: Obtain the target study area and the scale of the target area; For each set of land use products, coordinate system transformation is performed to obtain the land use standard data under the preset benchmark geographic coordinate system and benchmark projection coordinate system corresponding to the land use product; For each set of future land use data products, the standard land use data corresponding to the future land use data products are processed by extracting cultivated land, cropping the target study area, and calculating the cultivated land area at the target area scale to obtain intermediate data corresponding to the future land use data products. Based on the same predicted year and the same climate change scenario in the intermediate data corresponding to each set of future land use data products, determine the target climate change scenario, the target predicted year, and the target historical year corresponding to the target predicted year. For each set of future land use data products, the intermediate data corresponding to the target prediction year and the target climate change scenario are extracted to obtain the target prediction year, the cultivated land distribution data and the cultivated land area calculation data of the target study area under the target climate change scenario corresponding to the future land use data products; The arable land is extracted from the land use standard data corresponding to the historical land use data product, the target study area is cropped, and the target historical year is extracted to obtain the arable land distribution data of the target study area under the target historical year corresponding to the historical land use data product.

4. The method for predicting future arable land distribution according to claim 1, characterized in that, The step of determining the precision ranking results of the multiple sets of land use products based on the cultivated land area calculation data under the reference year in the target cultivated land data and the cultivated land area statistics data under the reference year includes: Based on the base year corresponding to each set of future land use data products in the target arable land data, the calculated arable land area of ​​the target study area under the target climate change scenario, and the statistical data of arable land area of ​​the target study area under the base year, the prediction accuracy value of the future land use data product under the base year is calculated. Based on the prediction accuracy values ​​of each set of future land use data products under the base year, the accuracy ranking of the multiple sets of land use products is determined; among them, the historical land use data products have the highest prediction accuracy.

5. The method for predicting future arable land distribution according to claim 4, characterized in that, The step of calculating the prediction accuracy value of the future land use data product in the base year based on the base year corresponding to each set of future land use data products in the target arable land data, the calculated arable land area data of the target study area under the target climate change scenario, and the statistical data of arable land area of ​​the target study area in the base year includes: The root mean square error is calculated for each set of future land use data products in the target cultivated land data, corresponding to the base year, the calculated cultivated land area of ​​the target study area under the target climate change scenario, and the cultivated land area statistics of the target study area under the base year, to obtain the prediction accuracy value of the future land use data product under the base year.

6. The method for predicting future arable land distribution according to claim 1, characterized in that, After fusing the cultivated land distribution data in the target cultivated land data according to the precision sorting result and the cultivated land area statistics to determine the future cultivated land distribution prediction result, the future cultivated land distribution prediction method further includes: Based on the predicted value of cultivated land area at the target area scale corresponding to the base year in the future cultivated land distribution prediction results and the statistical value of cultivated land area at the target area scale corresponding to the base year in the cultivated land area statistics, the index value corresponding to the preset accuracy index is calculated. The accuracy verification result of the future arable land distribution prediction result is determined based on the index value corresponding to the preset accuracy index.

7. A device for predicting future arable land distribution, characterized in that, include: The data acquisition module is used to acquire multiple sets of land use products and arable land area statistics. The multiple sets of land use products include historical land use data products and multiple sets of future land use data products. The historical land use data products include land use data from historical years, and the future land use data products include land use data for predicted years under a preset climate change scenario. The arable land area statistics include arable land area statistics for a baseline year and arable land demand data for future years, wherein the arable land demand data for future years is the arable land area demand data under the stated climate change scenario. The preprocessing module is used to preprocess the multiple sets of land use products to obtain target arable land data. The target arable land data includes arable land distribution data of the target study area under the target historical year corresponding to the historical land use data product, as well as arable land distribution data and arable land area calculation data of the target study area under the target prediction year and target climate change scenario corresponding to each set of future land use data products. The target historical year is the year of the previous cycle of the target prediction year, and the arable land area calculation data includes the arable land area calculation value at the target regional scale. The accuracy determination module is used to determine the accuracy ranking results of the multiple sets of land use products based on the cultivated land area calculation data under the reference year in the target cultivated land data and the cultivated land area statistics data under the reference year; wherein, the reference year is the starting year of the target prediction year; The result determination module is used to fuse the cultivated land distribution data in the target cultivated land data according to the precision sorting result and the cultivated land area statistics, and determine the future cultivated land distribution prediction result. The result determination module is specifically used for: generating a confidence score table based on the accuracy ranking results; wherein, in the confidence score table, the more land use products predicted as cultivated land in a pixel, and the higher the accuracy ranking, the higher the confidence score of that pixel; for each prediction year of the target prediction year, determining the confidence score of each pixel in the target study area under the prediction year based on the confidence score table and the cultivated land distribution data corresponding to the prediction year in the target cultivated land data; and calculating the cultivated land area of ​​each pixel under each confidence score in the target study area under the prediction year, based on the confidence score of each pixel under the prediction year, to obtain each confidence score under the prediction year. The corresponding cultivated land area value; according to the confidence score from high to low, the cultivated land area values ​​corresponding to different confidence scores in the prediction year are accumulated sequentially to obtain the accumulated cultivated land area value in the prediction year; when the accumulated cultivated land area value in the prediction year is closest to the corresponding cultivated land area statistical value in the cultivated land area statistical data, the cultivated land distribution result corresponding to the accumulated cultivated land area value in the prediction year is determined as the cultivated land distribution prediction result in the prediction year; the cultivated land distribution prediction result in the prediction year is determined as the cultivated land distribution data in the target historical year corresponding to the next cycle prediction year, so as to determine the cultivated land distribution prediction result corresponding to the next cycle prediction year.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, characterized in that, When the processor executes the computer program, it implements the method for predicting the future distribution of arable land as described in any one of claims 1-6.

9. A storage medium storing a computer program, characterized in that, The computer program, when executed by the processor, performs the method for predicting the future distribution of arable land as described in any one of claims 1-6.

Images