Cultivated land abandoned risk comprehensive assessment and prediction method, equipment and medium

Abstract

The invention discloses a farmland abandoned risk comprehensive assessment and prediction method and device and a medium, and relates to the technical field of computers, and the method comprises the steps: obtaining a data source of a to-be-detected region, obtaining an abandoned farmland extraction result through data preprocessing, time sequence data synthesis and machine classification learning, further determining abandoned farmland pattern spots of the to-be-detected region, and obtaining abandoned farmland image spots of the to-be-detected region. Determining the abandoned farmland rate according to the abandoned farmland pattern spots, carrying out the assignment of the abandoned farmland risk indexes according to the abandoned farmland rate, carrying out the comprehensive evaluation of the abandoned farmland risk, and determining the abandoned farmland risk grade; determining a regression coefficient based on a cultivated land abandoned rate and a cultivated land abandoned risk index assignment result by using a logistic regression model; determining the contribution rate of the abandoned farmland risk index to the abandoned farmland based on the regression coefficient; and predicting the abandoned farmland risk based on the regression coefficient and the contribution rate. According to the invention, scientific, effective, comprehensive and real-time comprehensive assessment and prediction can be carried out on the abandoned farmland risk.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, device and medium for comprehensive assessment and prediction of farmland abandonment risk. Background Art

[0002] The advancement of farmland abandonment remediation work mainly relies on the transmission of hierarchical instructions and the reporting of data from the grassroots level, in order to build a list and account of abandoned farmland, but the authenticity of the data in this way is difficult to accurately verify. At present, there is still a lack of scientific and effective technical means in risk monitoring and early warning. Furthermore, there are limitations in monitoring abandoned farmland based on time-series open source remote sensing images, and it is necessary to combine high-resolution remote sensing images, professional data, etc. to further improve the monitoring accuracy and form an engineering application solution. Summary of the invention

[0003] The purpose of this application is to provide a method, equipment and medium for comprehensive assessment and prediction of farmland abandonment risk, which can conduct scientific, effective, comprehensive and real-time comprehensive assessment and prediction of farmland abandonment risk.

[0004] To achieve the above objectives, this application provides the following solutions: In the first aspect, the present application provides a method for comprehensive assessment and prediction of farmland abandonment risk, including: Obtain remote sensing image data, digital elevation model data and land change survey maps of the area to be measured; Preprocessing the remote sensing image data of the area to be measured based on the digital elevation model data; Determining a remote sensing index based on the remote sensing image data of the area to be measured; the remote sensing index includes: a normalized vegetation index, a normalized water index and a soil index; constructing a time series data set based on the remote sensing index; Construct abandoned farmland extraction model based on machine learning classifier; Inputting the remote sensing image data of the area to be measured and the time series data set into the abandoned farmland extraction model to obtain the abandoned farmland extraction result; Determine the abandoned farmland map patches in the area to be measured based on the abandoned farmland extraction result, the remote sensing image of the area to be measured and the land change survey map patches; Determining the abandonment rate of cultivated land based on the abandoned cultivated land map patches in the tested area; Obtaining risk indicators affecting cultivated land abandonment, and assigning values ​​to the risk indicators based on the cultivated land abandonment rate; the risk indicators affecting cultivated land abandonment include: cultivated land slope, cultivated land quality, farming distance and plot area; Based on the result of the farmland abandonment risk index assignment, a comprehensive assessment is made on the farmland abandonment risk of the test area to determine the farmland abandonment risk level of the test area; Using a logistic regression model, based on the cultivated land abandonment rate and the cultivated land abandonment risk index assignment result, a regression coefficient is determined; Determine the contribution rate of farmland abandonment risk index to farmland abandonment based on the regression coefficient; The risk of abandoned farmland in the tested area is predicted based on the regression coefficient and the contribution rate.

[0005] Optionally, the preprocessing includes: orthorectification, radiation calibration, atmospheric correction, image fusion and mosaic cutting.

[0006] Optionally, constructing a time series data set based on the remote sensing index includes: A time series data set is constructed based on the remote sensing index using a natural month as a unit and using the maximum synthesis method.

[0007] Optionally, a model for extracting abandoned farmland is constructed based on a machine learning classifier, including: Acquire remote sensing image sample data and time series data sets and different land types corresponding to the remote sensing image sample data and time series data sets to form training sample pairs; Taking the remote sensing image sample data and the time series data set in the training sample pair as input and taking the different land types corresponding to the remote sensing image sample data and the time series data set as output, a machine learning classifier is trained to obtain a trained machine learning classifier; The trained machine learning classifier is used as the abandoned farmland extraction model.

[0008] Optionally, the abandoned farmland rate is expressed as: ; in, is the rate of abandoned farmland, For the The area of ​​abandoned farmland in For the The base year cultivated land area corresponding to the year.

[0009] Optionally, based on the result of assigning the risk index of abandoned farmland, a comprehensive assessment is made on the risk of abandoned farmland in the tested area to determine the risk level of abandoned farmland in the tested area, including: Based on the evaluation results of the risk index of abandoned cultivated land, Comprehensively evaluate the risk of abandoned farmland in the tested area to determine the risk level of abandoned farmland in the tested area; To reduce the risk of abandonment, is the value assignment result of different cultivated land abandonment risk indicators, is the number of risk indicators of abandoned farmland.

[0010] Optionally, the logistic regression model is expressed as: ; in, The rate of abandoned farmland of Logistic Regression model, For the regression coefficients, is the number of risk indicators of abandoned farmland, is the occurrence rate of the event, is the result of assigning different risk indicators of abandoned cultivated land, For hour The regression coefficient calculated in .

[0011] Alternatively, the contribution rate of the farmland abandonment risk index to farmland abandonment is expressed as: ; in, is the contribution rate of farmland abandonment risk index to farmland abandonment, is the regression coefficient The estimated value of for The variance estimate of For hour The regression coefficient calculated in .

[0012] In a second aspect, the present application provides a computer device comprising: 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 any of the above-mentioned methods for comprehensive assessment and prediction of farmland abandonment risks.

[0013] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above-mentioned methods for comprehensive assessment and prediction of farmland abandonment risks.

[0014] According to the specific embodiments provided in this application, this application has the following technical effects: The present application provides a method, device and medium for comprehensive assessment and prediction of cultivated land abandonment risk. Based on remote sensing data, the method combines digital elevation model data and land change survey maps to determine the risk level of cultivated land abandonment in the survey area, so as to solve the problem that cultivated land abandonment data depends on reporting from the grassroots level and the authenticity of the data is difficult to accurately verify, thereby enabling real-time and scientific comprehensive assessment of cultivated land abandonment risk; by using a logistic regression model to obtain the regression coefficient and the contribution rate of the cultivated land abandonment risk index to the cultivated land abandonment, the cultivated land abandonment risk is predicted, thereby achieving effective and comprehensive prediction of the cultivated land abandonment risk. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 A schematic flow chart of a method for comprehensive assessment and prediction of farmland abandonment risk provided in one embodiment of the present application; Figure 2 A schematic diagram of abandoned farmland before extraction in a sample area provided in an embodiment of the present application; Figure 3 A schematic diagram of abandoned farmland after extraction in a sample area provided in an embodiment of the present application; Figure 4 A schematic diagram of a map of abandoned farmland provided in one embodiment of the present application; Figure 5 A schematic diagram of the analysis results of the relationship between the abandonment rate and the abandonment risk index provided in one embodiment of the present application; Figure 6 A schematic diagram of the risk level distribution of abandoned farmland provided in an embodiment of the present application; Figure 7 It is a structural diagram of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0018] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0019] In an exemplary embodiment, the present application provides a method for comprehensive assessment and prediction of farmland abandonment risk, such as Figure 1 As shown, the method includes the following steps 100 to 111. Among them: Step 100, obtaining remote sensing image data, digital elevation model data and land change survey map of the area to be measured.

[0020] Step 101: pre-process the remote sensing image data based on the digital elevation model data.

[0021] Step 102: Determine a remote sensing index based on the remote sensing image data.

[0022] Step 103: construct a time series data set based on remote sensing indices.

[0023] Step 104, constructing an abandoned farmland extraction model based on a machine learning classifier; inputting the preprocessed remote sensing image data and time series data set into the abandoned farmland extraction model to obtain an abandoned farmland extraction result.

[0024] Step 105, based on the abandoned farmland extraction results, the pre-processed remote sensing images and the land change survey patches, determine the abandoned farmland patches in the area to be tested.

[0025] Step 106, determining the abandoned farmland rate based on the abandoned farmland map patches in the area to be tested.

[0026] Step 107, obtaining the risk index affecting the abandoned farmland, and assigning a value to the risk index of abandoned farmland based on the abandoned farmland rate.

[0027] Step 108, based on the result of assigning the risk index of abandoned farmland, comprehensively evaluate the risk of abandoned farmland in the tested area and determine the risk level of abandoned farmland in the tested area.

[0028] Step 109, using a logistic regression model, based on the cultivated land abandonment rate and the cultivated land abandonment risk index assignment result, determine the regression coefficient.

[0029] Step 110, determining the contribution rate of the farmland abandonment risk index to farmland abandonment based on the regression coefficient.

[0030] Step 111, predicting the risk of abandoned farmland based on the regression coefficient and the contribution rate.

[0031] The implementation of steps 100 to 111 above can achieve scientific extraction of abandoned farmland, and further realize effective, comprehensive and real-time comprehensive assessment and prediction of the risk of abandoned farmland.

[0032] In another exemplary embodiment of the present application, in order to scientifically extract abandoned farmland, the data source is remote sensing image data, digital elevation model (DEM), land change survey map and other data results. Based on this, the data sources such as remote sensing image data, digital elevation model data and land change survey map obtained in the above step 100 of the present application need to meet the following requirements: Remote sensing image data includes high-resolution remote sensing image data and medium- and low-resolution remote sensing image data. High-resolution remote sensing image data needs to collect images that meet monitoring requirements and cover multiple periods of key phenological periods, with at least green light band (520-570nm), red light band (620-760nm) and near-infrared band (760-1100nm) to meet the basic needs of vegetation monitoring; medium- and low-resolution remote sensing image data should try to collect Sentinel-2 multispectral image results and Landsat series multispectral image results in the past two years. Among them, it is recommended to choose L2A products for Sentinel-2 and Collection2 products for Landsat; the current situation of DEM data meets monitoring requirements, and the higher the spatial resolution, the better; the current situation of land change survey maps must meet monitoring requirements.

[0033] In another exemplary embodiment of the present application, in order to improve the quality of the data obtained in the present application, the original remote sensing image data needs to be preprocessed. Since the medium and low resolution remote sensing image data are generally image results, it mainly refers to the use of remote sensing image processing software to preprocess the collected high-resolution multi-source optical remote sensing image data. The specific steps are as follows: Step 1: Using the regional remote sensing image results and digital elevation model data as reference data, orthorectify and geometrically calibrate the high-resolution remote sensing image data to make it have geographic coordinates consistent with the actual situation. After correction, the geodetic coordinate error of flat and hilly land is less than or equal to 1 pixel, and the geodetic coordinate error of mountainous and high mountainous land is less than or equal to 2 pixels.

[0034] Step 2: Perform radiometric calibration on the corrected image, converting the digital quantization value of the remote sensing image into a physically meaningful atmospheric apparent radiation brightness value, in order to eliminate the sensor's own error and determine the accurate radiation value at the sensor's entrance pupil. Here, it is necessary to select the corresponding radiometric calibration coefficient according to different sensors to perform radiometric calibration on the remote sensing image.

[0035] Step 3: The image data after radiometric calibration also needs to be atmospherically corrected to eliminate the radiometric error caused by the atmospheric influence and invert the true reflectivity of the ground objects. It should be made clear that the panchromatic band cannot be atmospherically corrected.

[0036] Step 4: Image fusion, mosaicking and cropping. The multispectral image and the panchromatic image are fused. The panchromatic enhancement (Pansharping) fusion method is recommended for image fusion. After fusion, the entire remote sensing image is mosaicked and cropped according to the scope of the area to be measured to form an orthophoto consistent with the scope of the area to be measured. In order to ensure that the image spots at the boundary are extracted correctly, the cropping range needs to be expanded by at least 100 meters outside the scope of the area to be measured.

[0037] In another exemplary embodiment of the present application, in order to solve the problem that abandoned farmland data depends on grassroots reporting and the authenticity of the data is difficult to be accurately verified, and to achieve scientific extraction of abandoned farmland, this embodiment provides specific steps for extracting abandoned farmland based on the acquired data, including: Step 1: Calculate remote sensing indexes based on low- to medium-resolution long-term series remote sensing image data. Remote sensing indices include the Normalized Difference Vegetation Index (NDVI), the Normalized Difference Water Index (NDWI1), and the Soil Index (MBI). When multiple data sources are involved, preprocessing operations such as assimilation and resampling can be performed before index calculation.

[0038] Step 2: The calculation results of different remote sensing indices for multiple periods each month need to be synthesized pixel by pixel to construct a maximum value dataset for each year and month. When multiple data sources are involved, images with higher spatial resolution are used first in the synthesis process, and images with lower spatial resolution are used to fill in the holes.

[0039] Step 3: Based on the yearly and monthly summary results of different remote sensing indices, the significant differences in different remote sensing indices between normal cultivated land and abandoned cultivated land are used to establish abandoned cultivated land extraction rules.

[0040] Step 4: Refer to high-resolution optical remote sensing images, select training samples for abandoned farmland, dry land, paddy field, woodland, bare land, water surface and other land types, establish a sample library, use random forest classifier or other machine learning algorithms to train the sample library, and complete the remote sensing extraction of abandoned land in the test area based on the training results.

[0041] In another exemplary embodiment of the present application, the implementation process of step 105 of the present application can be described as follows: Based on the remote sensing extraction results of abandoned land, with reference to high-resolution optical remote sensing images, the cultivated land patches, farmland-to-forest patches and bulldozer areas in the land change survey data are superimposed to screen out the abandoned cultivated land patches in the test area, thereby further improving the monitoring and identification accuracy.

[0042] In actual application, on-site verification and validation can be carried out as needed based on the remote sensing extraction results of abandoned land.

[0043] In another exemplary embodiment of the present application, a method for determining the abandoned farmland rate based on the abandoned farmland map patches in the test area is provided. The abandoned farmland rate is the ratio of the abandoned land area or the number of land plots to the total cultivated land area or the number of land plots, which is the proportion of abandoned farmland in the region and can reflect the degree of abandoned farmland in the region to a certain extent. The abandoned farmland rate is expressed as: .

[0044] in, is the rate of abandoned farmland, For the The area of ​​abandoned farmland in For the The base year cultivated land area corresponding to the year.

[0045] In another exemplary embodiment of the present application, a method for obtaining an index affecting the risk of abandoned farmland and assigning a value to the risk index of abandoned farmland based on the abandoned farmland rate is provided, wherein: In actual application, based on the principles of scientificity, operability, people-orientedness, and combination of commonality and individuality in indicator construction, indicators that affect the risk of abandoned farmland are selected under the premise that the quality of cultivated land in the region is improved, the amount of cultivated land is increased, the total amount of construction land does not increase, the ecological protection red line is not broken, the ecological environment is improved, and the negative list is not touched.

[0046] This application focuses on the land attributes that affect agricultural output levels and farming costs. For the factors that affect agricultural operating income and costs, the two groups of land output and labor costs with the greatest impact are selected. Considering the difficulty of data acquisition, the factors affecting land output are selected as the slope of cultivated land and the quality of cultivated land; the factors affecting labor costs are selected as the distance to cultivation and the area of ​​the plot. According to the relationship between the indicators and the abandonment rate of cultivated land, the indicators are assigned values, and the scores are between 0 and 1. The higher the score, the greater the risk of abandonment, and the lower the score, the smaller the risk of abandonment.

[0047] The definitions and reference assignment methods of the four farmland abandonment risk indicators are as follows.

[0048] ① Slope of cultivated land: The slope of cultivated land reflects the quality of the land and the difficulty of using machinery. The average slope value of the assessed plot is taken. The greater the slope of the plot, the higher the rate of abandoned cultivated land and the higher the corresponding indicator value.

[0049] ② Cultivated land quality: Cultivated land quality is an evaluation of the ability of cultivated land fertility, soil health and field infrastructure to meet the sustainable output and quality safety of agricultural products. The worse the cultivated land quality, the higher the cultivated land abandonment rate and the higher the corresponding indicator value.

[0050] ③ Cultivation distance: Cultivation distance refers to the commuting distance of farmers to the assessed plots for cultivation, reflecting the difficulty of commuting and the consumption of labor. It is the straight-line distance from the nearest road to the center of the assessed plot. The greater the cultivation distance, the higher the abandonment rate of cultivated land and the higher the corresponding indicator value.

[0051] ④ Plot area: reflects the degree of fragmentation of the plot, the degree of labor consumption in farming and commuting, and the difficulty of using machinery; the smaller the plot area, the higher the rate of abandoned farmland, and the higher the corresponding indicator value.

[0052] In another exemplary embodiment of the present application, a method for comprehensively evaluating the risk of abandoned farmland in the tested area based on the result of assigning the risk index of abandoned farmland is provided, and determining the risk level of abandoned farmland in the tested area is as follows: Based on the selected farmland abandonment risk indicators and assignment results, a comprehensive evaluation of the farmland abandonment risk can be carried out, so that the overall farmland abandonment risk in the region can be analyzed and evaluated, and the risk levels can be divided.

[0053] In practical applications, the comprehensive evaluation of the risk of abandoned farmland is based on the arithmetic average method based on the evaluation of each individual item. Comprehensive evaluation model for abandonment risk: .

[0054] in, To reduce the risk of abandonment, is the result of assigning different risk indicators of abandoned cultivated land, is the number of risk indicators of abandoned farmland.

[0055] In this application, each cultivated land plot has 4 indicators, and the total score of the 4 indicators of each plot is calculated. According to the scores of all plots and the proportion of plots in each score range, the abandonment risk of cultivated land plots is divided into three levels: low, medium and high.

[0056] In another exemplary embodiment of the present application, a method is provided for determining a regression coefficient based on the cultivated land abandonment rate and the cultivated land abandonment risk index assignment results by using a logistic regression model; determining the contribution rate of the cultivated land abandonment risk index to the cultivated land abandonment according to the regression coefficient; and predicting the cultivated land abandonment risk by combining the regression coefficient and the contribution rate, wherein: In practical applications, the logistic regression model is introduced into the spatial concept to analyze the multivariate relationship between abandonment risk and multiple risk indicators affecting farmland abandonment, and to calculate the contribution rate of each indicator to farmland abandonment. , so as to effectively predict the impact of changes in different risk indicators on the probability of farmland abandonment, so as to make targeted adjustments to relevant policies. Substitute into the model for regression calculation, the logistic regression model is as follows: .

[0057] .

[0058] in, The rate of abandoned farmland of Logistic Regression model, For the regression coefficients, is the number of risk indicators of abandoned farmland, is the occurrence rate of the event, The results of assigning values ​​to different risk indicators of abandoned cultivated land. It is used for hypothesis testing, measuring the difference between the information provided by the sample data and the parameter value under the null hypothesis, and is used to evaluate the contribution of each indicator to farmland abandonment. It is based on the asymptotic normal distribution characteristics of the parameter estimator. In the case of large samples, its distribution approximately follows the chi-square ( ) distribution. For a single regression coefficient , The calculation formula is: .

[0059] in, is the contribution rate of farmland abandonment risk index to farmland abandonment, is the regression coefficient The estimated value of for The variance estimate of For hour The regression coefficient calculated in .

[0060] In an exemplary embodiment, this embodiment takes Jintang County, Sichuan as a case to specifically explain the method of comprehensive assessment and prediction of farmland abandonment risk.

[0061] (1) Collect data: The collected data are: 169 high-resolution multispectral original images from January 2022 to December 2023, covering 20 periods of the survey area, with a spatial resolution better than 2 meters; 42 periods of Sentinel-2 MSI L2A images and 31 periods of Landsat Collection 2 products from January 2022 to December 2023; collected the 1:50,000 standard DEM results of the Sichuan region of the Real Three-Dimensional China Construction Project, with a grid spacing of 10 meters, CGCS2000 national geodetic coordinate system, Gauss-Krüger projection 6° zoning, and the overall potential in 2019; collected the 2023 land change survey data results of Jintang County and the corresponding planting attribute table.

[0062] (2) Multispectral image preprocessing: The multispectral original images collected in this embodiment include: high-resolution multispectral images and medium- and low-resolution multispectral images.

[0063] The preprocessing process of high-resolution multispectral images is as follows: through orthorectification, radiation calibration, atmospheric correction, image fusion and mosaic cutting, the output coordinate system is the CGS2000 Albers double standard latitude projection and the multi-phase 4-band image results covering the area to be measured.

[0064] The preprocessing process of medium and low resolution multispectral images is as follows: Due to the differences in the orbits and sensor wavelengths of different satellites, the surface radiation values ​​obtained are inconsistent. Here, the least squares regression parameters are used to regress the bands of Sentinel-2MSI and Landsat7 ETM+ to the same level as Landsat8 OLI. The bands needed are blue, green, red, near infrared and short-wave infrared, among which the assimilation parameters of MSI are 0.8729, 0.9621, 0.9103, 0.9701 and 0.9668, and the assimilation parameters of ETM+ are 0.8707, 0.8707, 0.9175, 0.9281 and 0.9414.

[0065] In addition, there is a difference between the spatial resolution of Landsat images (30m) and the spatial resolution of Sentinel-2 images (10m), so the assimilated Landsat images need to be resampled to 10m and all images need to be cropped according to the scope of the area to be measured.

[0066] (3) Determine remote sensing index and construct time series data set: The remote sensing index is calculated for the preprocessed images. Taking the natural month as the unit, the maximum value synthesis method is used to select the maximum value of each index in the month as the value of the monthly time series dataset. In the synthesis process, Sentinel-2 images with higher spatial resolution are used first, and Landsat images are used to fill the image holes to form a monthly time series dataset of the area to be tested.

[0067] (4) Extraction of abandoned farmland: First, the training samples are prepared: a sample area is selected in the test area, and the abandoned farmland in the sample area is extracted as follows Figure 2 As shown in the figure, with reference to the vector map data of land change survey results and high-resolution multispectral satellite image results, the vector map of the sample area was reclassified into abandoned land, dry land, paddy field, forest land, water surface, bare land and other land types through visual interpretation, which served as sample data and result verification data for target area screening. A total of 1,383 abandoned farmland map spots and 8,573 non-abandoned farmland map spots (including 317 paddy fields and 8,256 dry land) were obtained. The abandoned farmland map spots in the sample area are as follows: Figure 3As shown, the ratio of sample data to verification data is 7:3.

[0068] Based on the constructed time series dataset and training samples, random forest classifier and support vector machine classifier were used to train the sample data respectively. The time series image dataset and training samples were input into the classifier training, and the classification accuracy of abandoned farmland was evaluated using the validation data. Referring to the accuracy evaluation results (Table 1), the random forest classifier training model was finally selected to complete the extraction of abandoned farmland in the test area.

[0069] Table 1 Accuracy evaluation results of different classifiers

[0070] Among them, OA in Table 1 is the overall accuracy, which refers to the proportion of samples correctly classified by the classifier to the total number of samples; Kappa is the proportion of errors reduced by classification compared with completely random classification, which is used to measure the consistency of the classifier.

[0071] (5) Abandoned farmland map: The abandoned farmland extraction results are used as the target area. With reference to the high-resolution optical remote sensing image results, combined with the selected land change survey cultivated land patches, manual judgment is performed to eliminate the fallow patches, the patches of returning farmland to forests, and the bulldozing areas, and the abandoned cultivated land patches in the test area are screened out to further improve the accuracy of cultivated land abandonment monitoring. Figure 4 shown.

[0072] (6) Analysis on the relationship between the rate of abandoned farmland and indicators affecting the risk of abandoned farmland: Based on the extracted abandoned farmland data, the abandonment rate of cultivated land at the plot scale and area scale was calculated, and its correspondence with four abandonment risk indicators (cultivated land slope, cultivated land quality, farming distance, and plot area) was analyzed to find the critical point of the mutation value. According to the analysis results, the abandonment risk indicator of cultivated land was conveniently assigned. The analysis results of the relationship between the abandonment rate and the abandonment risk indicator are shown in Figure 2. Figure 5 As shown, Figure 5 Part (a) is a schematic diagram of the analysis results of abandonment rate and cultivated land slope. Figure 5 Part (b) is a schematic diagram of the analysis results of abandonment rate and cultivated land quality. Figure 5 Part (c) is a schematic diagram of the analysis results of abandonment rate and cultivation distance. Figure 5 Part (d) is a schematic diagram of the analysis results of abandonment rate and land area.

[0073] (7) Assignment of risk indicators for abandoned cultivated land: Slope assignment of cultivated land: According to detailed land survey data, the slope of cultivated land is divided into five levels: 0°-2°, 2°-6°, 6°-15°, 15°-25° and greater than 25°. Cultivated land of each slope level has different suitability and production capacity. 6°, 15° and 25° are obvious turning points. The abandonment rate of cultivated land with a slope of less than 6° fluctuates around 12%; when the slope of cultivated land exceeds 15°, the abandonment rate of cultivated land exceeds 20%; when the slope of cultivated land exceeds 25°, the abandonment rate of cultivated land reaches more than 40%. According to this rule, cultivated land with a slope of less than 15° is assigned a low score of 0.4, a slope of 15° to 25° is assigned a score of 0.7, and a slope greater than 25° is assigned a high score of 0.9.

[0074] Assignment of arable land quality: According to the direct relationship between the arable land grade and the standard grain production capacity, the arable land grade is divided into three levels. Among them, the basic soil fertility of the first to third grades of arable land is relatively high, and there are basically no obstacles. The good condition of this part of arable land is of great significance to ensuring food security, and there is basically no risk of abandonment. The environmental and climatic conditions of the arable land of the fourth to sixth grades are basically suitable, the construction of farmland infrastructure has a certain foundation, and the obstacles are not obvious. It is the key area and important breakthrough for increasing grain production in the future. By improving the basic soil fertility and farmland infrastructure of these arable lands, the grain output can be significantly increased, and the risk of abandonment is relatively low. The basic soil fertility of the arable land of the seventh to ninth grades is relatively poor, and the production obstacles are prominent. It is difficult to get fundamental improvement in a short period of time, and the risk of abandonment is high. According to this rule, the first to third grades of arable land are assigned a low score of 0.4, the fourth to sixth grades of arable land are assigned a score of 0.7, and the seventh to ninth grades of arable land are assigned a high score of 0.9.

[0075] Assignment of tillage distance: According to its relationship with the abandonment rate of cultivated land, 200 meters and 500 meters are two turning points. The abandonment rate of cultivated land within a tillage distance of 200 meters fluctuates mainly within the range of 5%, the abandonment rate of cultivated land between 200 and 500 meters fluctuates mainly within the range of 10% to 20%, and the abandonment rate of cultivated land with a tillage distance greater than 600 meters increases sharply to more than 20%. According to this rule, a low score of 0.4 is assigned to cultivated land plots with a tillage distance of less than 200 meters, a score of 0.7 is assigned to cultivated land plots with a tillage distance of 200 to 500 meters, and a high score of 0.9 is assigned to cultivated land plots with a tillage distance of more than 500 meters.

[0076] Land area assignment: According to its relationship with the rate of abandoned farmland, land area of ​​200 square meters and 600 square meters are two turning points. When the land area is less than 200 square meters, the rate of abandoned farmland is 80%, and the rate of abandoned farmland greater than 600 square meters is between 1% and 4%. According to this rule, a high score of 0.9 is assigned to a land area less than 200 square meters, a score of 0.7 is assigned to a land area between 200 and 600 square meters, and a low score of 0.4 is assigned to a land area greater than 600 square meters.

[0077] (8) Comprehensive assessment and classification of abandonment risks: Based on the above description, each cultivated land plot has 4 indicators, and the score of each indicator is between 0.4 and 0.9, so as to calculate the total score of each plot, with the highest score being 3.6 and the lowest score being 1.6. After all the parameters are input into the comprehensive evaluation model, the comprehensive score is obtained. According to the comprehensive score, the abandonment risk of cultivated land plots is divided into three levels: low, medium and high. A score less than 1.9 is considered low abandonment risk cultivated land, 1.9 to 2.2 is considered medium abandonment risk cultivated land, and a score greater than 2.2 is considered high abandonment risk cultivated land. The abandonment risk level of cultivated land in the test area is determined. The distribution of the abandonment risk level of cultivated land in this case is as follows: Figure 6 shown.

[0078] (9) Risk prediction of abandoned cultivated land: Substitute the dependent variable (abandonment rate of cultivated land) and the independent variable (risk index of abandoned cultivated land) into the logistic regression model to calculate the risk of abandoned cultivated land. Logistic The regression model parameters are shown in Table 2. The regression coefficient Indicates the correlation between each variable (driving factor) and abandoned farmland, and the statistic It is used to evaluate the contribution rate of each variable to event prediction and predict the risk of abandoned farmland based on the contribution rate. All of them are less than 0.01, indicating that the regression model has extremely significant statistical significance.

[0079] Table 2 Contribution rate of risk driving factors of cultivated land abandonment in the tested area

[0080] From the analysis of Table 2, we can see that the contribution rates of the four indicators are plot area, cultivated land slope, cultivation distance and cultivated land grade from high to low. Among the driving factors of agricultural output level, cultivated land slope has the greatest impact. For paddy field and dry land abandonment, the regression coefficients are all positive, indicating that the steeper the slope, the less conducive it is to mechanized farming. Due to high cost and low output, abandonment is prone to occur. Among the driving factors of labor cost, plot area has the greatest impact and is an important driving factor of cultivated land abandonment. The regression coefficient is negative, indicating that cultivated land abandonment is negatively correlated with plot area. The smaller the plot, the more difficult it is to mechanize agricultural farming, the less conducive it is to large-scale operation, and the greater the probability of cultivated land abandonment. In general, with the accelerated advancement of urbanization and the continuous development of social economy, a large number of agricultural labor forces have been separated, and cultivated land with long distance, small area, poor infrastructure, high altitude and steep slope is prone to abandonment.

[0081] In summary, the present application has the following advantages over the known technologies: (1) This application improves the accuracy of monitoring abandoned farmland: it makes up for the limitations of using time-series open source remote sensing images for monitoring abandoned farmland, combines high-resolution remote sensing images, and makes full use of professional data to further improve monitoring accuracy, thereby forming an engineering application solution.

[0082] (2) Forming a technical solution for monitoring and early warning of farmland abandonment risks: Based on the monitoring of farmland abandonment, this application identifies the risk of farmland abandonment at the plot scale, and forms a normalized risk monitoring and early warning technical solution that matches the governance requirements.

[0083] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 7 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store comprehensive assessment and prediction data of cultivated land abandonment risk. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a comprehensive assessment and prediction method of cultivated land abandonment risk is implemented.

[0084] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0085] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0086] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0087] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0088] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0089] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0090] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for comprehensive assessment and prediction of farmland abandonment risk, characterized in that: The comprehensive assessment and prediction method of farmland abandonment risk includes: Obtain remote sensing image data, digital elevation model data and land change survey maps of the area to be measured; Preprocessing the remote sensing image data of the area to be measured based on the digital elevation model data; Determining a remote sensing index based on the remote sensing image data of the area to be measured; the remote sensing index includes: a normalized vegetation index, a normalized water index and a soil index; constructing a time series data set based on the remote sensing index; Construct abandoned farmland extraction model based on machine learning classifier; Inputting the remote sensing image data of the area to be measured and the time series data set into the abandoned farmland extraction model to obtain the abandoned farmland extraction result; Determine the abandoned farmland map patches in the area to be measured based on the abandoned farmland extraction result, the remote sensing image of the area to be measured and the land change survey map patches; Determining the abandonment rate of cultivated land based on the abandoned cultivated land map patches in the tested area; Obtaining risk indicators affecting cultivated land abandonment, and assigning values ​​to the risk indicators based on the cultivated land abandonment rate; the risk indicators affecting cultivated land abandonment include: cultivated land slope, cultivated land quality, farming distance and plot area; Based on the value assignment result of the cultivated land abandonment risk index, a comprehensive assessment is made on the cultivated land abandonment risk of the tested area to determine the cultivated land abandonment risk level of the tested area; Determine the regression coefficient based on the value assignment results of the cultivated land abandonment rate and the cultivated land abandonment risk index by using a logistic regression model; Determining the contribution rate of the cultivated land abandonment risk index to the cultivated land abandonment based on the regression coefficient; The risk of abandoned farmland is predicted based on the regression coefficient and the contribution rate.

2. The method for comprehensive assessment and prediction of farmland abandonment risk according to claim 1, characterized in that: The preprocessing includes: orthorectification, radiation calibration, atmospheric correction, image fusion and mosaic cutting.

3. The method for comprehensive assessment and prediction of farmland abandonment risk according to claim 1, characterized in that: A time series data set is constructed based on the remote sensing index, including: A time series data set is constructed based on the remote sensing index using a natural month as a unit and using the maximum synthesis method.

4. The method for comprehensive assessment and prediction of farmland abandonment risk according to claim 1, characterized in that: Abandoned farmland extraction model is built based on machine learning classifiers, including: Acquire remote sensing image sample data and time series data sets and different land types corresponding to the remote sensing image sample data and time series data sets to form training sample pairs; Taking the remote sensing image sample data and the time series data set in the training sample pair as input and taking the different land types corresponding to the remote sensing image sample data and the time series data set as output, a machine learning classifier is trained to obtain a trained machine learning classifier; The trained machine learning classifier is used as the abandoned farmland extraction model.

5. The method for comprehensive assessment and prediction of farmland abandonment risk according to claim 1, characterized in that: The abandoned farmland rate is expressed as: ； in, is the rate of abandoned farmland, For the The area of ​​abandoned farmland in For the The base year cultivated land area corresponding to the year.

6. The method for comprehensive assessment and prediction of farmland abandonment risk according to claim 1, characterized in that: Based on the value assignment result of the cultivated land abandonment risk index, a comprehensive assessment is conducted on the cultivated land abandonment risk of the tested area to determine the cultivated land abandonment risk level of the tested area, including: Based on the evaluation results of the risk index of abandoned cultivated land, Conduct a comprehensive assessment of the risk of abandoned farmland in the area to be tested, and determine the risk level of abandoned farmland in the area to be tested; To reduce the risk of abandonment, is the value assignment result of different cultivated land abandonment risk indicators, is the number of risk indicators of abandoned farmland.

7. The method for comprehensive assessment and prediction of farmland abandonment risk according to claim 1, characterized in that: The logistic regression model is expressed as: ； in, The rate of abandoned farmland of Logistic Regression model, For the regression coefficients, is the number of risk indicators of abandoned farmland, is the occurrence rate of the event, is the result of assigning different risk indicators of abandoned cultivated land, For hour The regression coefficient calculated in .

8. The method for comprehensive assessment and prediction of farmland abandonment risk according to claim 7, characterized in that: The contribution rate of cultivated land abandonment risk index to cultivated land abandonment is expressed as: ； in, is the contribution rate of farmland abandonment risk index to farmland abandonment, is the regression coefficient The estimated value of for The variance estimate of For hour The regression coefficient calculated in .

9. A computer device comprising: 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 method for comprehensive assessment and prediction of farmland abandonment risk according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the comprehensive assessment and prediction method for the risk of abandoned farmland as described in any one of claims 1 to 8.

Images