A method of soil erosion prediction

Abstract

The application provides a soil erosion prediction method, and belongs to the technical field of soil erosion. The method comprises the following steps: obtaining a target data set corresponding to each grid, obtaining a rainfall erosion force factor, a soil erosion factor, a slope length and slope gradient factor, a vegetation coverage factor and a water and soil conservation factor of each grid based on the target data set; calculating the soil erosion amount of each grid by using a RUSLE model based on the above factors; correcting the soil erosion amount of each grid in two adjacent grids based on the geological features and environmental features between the two adjacent grids to obtain the corrected soil erosion amount of each grid; obtaining the predicted soil erosion amount of each grid based on the corrected soil erosion amount of each grid and the soil erosion amount in each period in history; and obtaining the predicted soil erosion amount of a target region based on the predicted soil erosion amount of each grid. The application can improve the prediction accuracy of the soil erosion amount in the target region.

Description

Technical Field

[0001] This application belongs to the field of soil erosion technology, and more specifically, relates to a method for predicting soil erosion. Background Technology

[0002] Soil erosion is a global environmental problem. Especially against the backdrop of increased human engineering activities, frequent soil erosion events caused by extreme rainfall not only damage agricultural production but also significantly impact transportation, industry, and other sectors, posing a serious threat to regional ecological security and sustainable development. Therefore, accurately acquiring soil erosion data and predicting erosion trends have become core requirements in soil and water conservation planning and ecological risk prevention.

[0003] Existing soil erosion prediction technologies are mostly based on classical models, such as the Revised Universal Soil Loss Equation (RUSLE). This model estimates soil erosion rates using rainfall erosivity, soil erodibility factors, topographic factors, vegetation cover factors, and soil and water conservation factors. Furthermore, current technologies typically use the RUSLE model directly to predict soil erosion in a single area without a correction mechanism for the predicted erosion, resulting in low prediction accuracy and difficulty in meeting practical needs. Summary of the Invention

[0004] The purpose of this application is to provide a soil erosion prediction method that can improve the accuracy of soil erosion prediction in a target area.

[0005] A first aspect of this application provides a method for predicting soil erosion, comprising: Obtain the target dataset corresponding to each grid. Each grid is obtained after dividing the target area. The target area is the area where soil erosion prediction is to be performed. The target dataset includes data corresponding to geological parameters, topographic parameters, hydro-meteorological parameters, and ecological parameters. Based on the target dataset, the rainfall erosivity factor, soil erosion factor, slope length and slope factor, vegetation cover factor and soil and water conservation factor of each grid are obtained. Based on rainfall erosivity factor, soil erosion factor, slope length and gradient factor, vegetation cover factor and soil and water conservation factor, the soil erosion of each grid is calculated using the RUSLE model. Based on the geological and environmental characteristics between two adjacent grids, the soil erosion of each grid in the two adjacent grids is corrected to obtain the corrected soil erosion of each grid. Based on the corrected soil erosion of each grid and the soil erosion of each grid in each historical period, the predicted soil erosion of each grid is obtained, and based on the predicted soil erosion of each grid, the predicted soil erosion of the target area is obtained.

[0006] A second aspect of this application provides a soil erosion prediction device, comprising: The data acquisition unit is used to acquire the target dataset corresponding to each grid. Each grid is obtained by dividing the target area. The target area is the area to be predicted for soil erosion. The target dataset includes data corresponding to geological parameters, topographic parameters, hydro-meteorological parameters and ecological parameters respectively. The multi-factor calculation unit is used to obtain the rainfall erosivity factor, soil erosion factor, slope length and slope factor, vegetation cover factor and soil and water conservation factor of each grid based on the target dataset. The data calculation unit is used to calculate the soil erosion of each grid cell based on rainfall erosivity factor, soil erosion factor, slope length and gradient factor, vegetation cover factor and soil and water conservation factor using the RUSLE model. The data correction unit corrects the soil erosion of each grid cell in two adjacent grid cells based on the geological and environmental characteristics between the two adjacent grid cells to obtain the corrected soil erosion of each grid cell. The prediction unit is used to obtain the predicted soil erosion of each grid based on the corrected soil erosion of each grid and the soil erosion of each grid in each historical period, and to obtain the predicted soil erosion of the target area based on the predicted soil erosion of each grid.

[0007] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the soil erosion prediction method described above.

[0008] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the soil erosion prediction method described above.

[0009] The beneficial effects of the soil erosion prediction method provided in this application are as follows: This application can improve the prediction accuracy of soil erosion within a target area. This application divides the target area into multiple grids, first predicts the soil erosion within each grid, and then obtains the predicted soil erosion value for the entire target area based on the predicted values ​​of the soil erosion within each grid. By calculating independently for each grid and then summarizing the results, this application solves the problem of traditional methods treating the target area as a whole, which is prone to masking internal heterogeneity due to the "averaging effect," significantly improving the spatial refinement and local accuracy of soil erosion prediction.

[0010] For each grid cell, this application obtains an initial value of soil erosion for that grid cell in the current period based on the target dataset acquired within the grid and the RUSLE model. Then, it corrects the initial value of soil erosion based on the interaction between two adjacent grid cells to obtain a corrected soil erosion value. Based on the corrected soil erosion value and the soil erosion values ​​in each historical period, the application predicts the soil erosion value for that grid cell. This application corrects erosion based on the geological and environmental characteristics of two adjacent grid cells, effectively restoring the dynamic interaction mechanism of grids with different terrains and environments during the actual erosion process. This reduces prediction bias caused by neglecting cross-grid material and energy exchange, significantly enhancing the accuracy and reliability of soil erosion prediction. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A schematic flowchart of a soil erosion prediction method provided in an embodiment of this application; Figure 2 A schematic diagram of the distribution of adjacent grids provided in an embodiment of this application; Figure 3 This is a structural block diagram of a soil erosion prediction device provided in an embodiment of this application; Figure 4 This is a schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0013] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0014] To make the objectives, technical solutions, and advantages of this application clearer, the following description will be provided in conjunction with the accompanying drawings and specific embodiments.

[0015] Please refer to Figure 1 , Figure 1 This is a schematic flowchart of a soil erosion prediction method provided in an embodiment of this application. The method can be executed by an electronic device and may include: S101~S105.

[0016] S101: Obtain the target dataset corresponding to each raster.

[0017] Each grid is obtained by dividing the target area, which is the area where soil erosion prediction is to be performed. The target dataset includes data corresponding to geological parameters, topographic parameters, hydro-meteorological parameters, and ecological parameters.

[0018] In this embodiment, traditional methods typically apply the RUSLE model to an independent region to predict soil erosion within that region. However, considering the large size of a single independent region, directly calculating soil erosion amounts would be inaccurate and hinder relevant departments from making reasonable plans for soil and water conservation and ecological construction in that region. Therefore, this application employs a grid-based calculation method, dividing the target region into multiple grids. Soil erosion is predicted for each grid individually, and the predicted soil erosion result for the target region is obtained based on the prediction results of each grid.

[0019] In this embodiment, the target area is first determined. The target area is the region where soil erosion prediction is to be performed, such as city A, the coastline of an island B, or a mountainous area C. The target area is then divided into multiple grids according to certain rules, such as equally spaced latitude and longitude lines or squares with fixed side lengths. Each grid is an independent calculation unit, and each grid can be labeled to distinguish it from others.

[0020] The process involves acquiring the target dataset for each raster, which includes data corresponding to geological parameters, topographic parameters, hydro-meteorological parameters, and ecological parameters. Geological parameters include data on sand and gravel content, silt content, clay content, organic carbon content, rock compressive strength, porosity, fault density, and stratum dip angle. Topographic parameters include slope length, slope, aspect, curvature, and topographic moisture index. Hydro-meteorological parameters include the raster's average soil moisture content, the average soil moisture content within a preset time before rainfall, rainfall data, and wind data. Ecological parameters include satellite remote sensing data and location data. After preprocessing the satellite remote sensing data (such as radiometric and geometric corrections), ecological parameters such as vegetation type and vegetation cover can be extracted. Location data represents the geographic location of the current raster center, obtainable through satellite positioning or UAV positioning.

[0021] The data in the target dataset mentioned above is a key factor in predicting soil erosion in the target area. By obtaining the target dataset corresponding to each grid and processing the target dataset corresponding to each grid to obtain the predicted soil erosion for each grid, the predicted soil erosion for the target area can be obtained.

[0022] S102: Based on the target dataset, obtain the rainfall erosivity factor, soil erosion factor, slope length and slope factor, vegetation cover factor and soil and water conservation factor for each grid.

[0023] In this embodiment, rainfall erosivity factor, soil erosion factor, slope length and gradient factor, vegetation cover factor, and soil and water conservation factor are the input parameters of the RUSLE model. These factors quantify the soil erosion process from different dimensions.

[0024] The rainfall erosivity factor R can represent the potential erosion of soil by annual, monthly, or daily rainfall per unit area, reflecting the impact of rainfall itself on the soil. The rainfall erosivity factor can be derived from the hydrological and meteorological parameters in the target dataset using an empirical formula, which is the product of rainfall kinetic energy and the maximum 30-minute rainfall intensity. Alternatively, other influencing factors can be added to improve the empirical formula.

[0025] Soil erosion factor K, also known as soil erodibility factor, characterizes the soil's sensitivity to erosion, i.e., the amount of soil loss per unit R value. In this embodiment, it can be calculated based on geological parameters in the target dataset. For example, a calculation model for the soil erosion factor can be constructed based on one or more indicators such as gravel content, silt content, clay content, organic carbon content, soil organic matter thickness, and aggregate stability. There are multiple calculation models for the soil erosion factor; a suitable model can be selected based on the geological characteristics within the raster.

[0026] The slope length-gradient factor (LS) comprehensively reflects the impact of topography on erosion. It is typically determined by the combined effects of slope length (L) and slope gradient (S) among topographic parameters. In special topographic conditions, the combined effects of slope length (L), slope gradient (S), gully density coefficient, or other influencing parameters may also be present. A longer slope results in greater cumulative surface runoff and stronger erosion energy; a steeper slope leads to faster runoff velocity and a stronger ability to transport soil. The gully density coefficient is used to quantify the interception effect of gullies on runoff confluence in small watershed scenarios; a higher gully density coefficient indicates weaker erosion capacity.

[0027] The vegetation cover factor C is used to quantify the inhibitory effect of vegetation on soil erosion. The value ranges from 0 to 1. The vegetation cover factor C is calculated based on hydrological and meteorological parameters and ecological parameters. Among the ecological parameters, vegetation type and vegetation coverage rate directly affect the vegetation cover factor, while wind data among the hydrological and meteorological parameters affect vegetation coverage rate, and thus indirectly affect the vegetation cover factor.

[0028] The soil and water conservation factor P reflects the mitigation effect of human soil and water conservation measures on erosion, with a value ranging from 0 to 1 (1 indicates no measures, <1 indicates effective measures). Common measures include terracing (P≈0.5), contour planting (P≈0.7), and straw checkerboard sand fixation (P≈0.3). The P factor can be assigned a value based on historical data of measures, reflecting the degree of human intervention in the natural erosion process.

[0029] S103: Based on rainfall erosivity factor, soil erosion factor, slope length and gradient factor, vegetation cover factor and soil and water conservation factor, the soil erosion of each grid is calculated using the RUSLE model.

[0030] In this embodiment, the RUSLE model is a commonly used soil erosion prediction model, and its formula is as follows:

[0031] Where A represents soil erosion, R represents rainfall erosivity factor, K represents soil erosion factor, LS represents slope length and gradient factor, C represents vegetation cover factor, and P represents soil and water conservation factor. For each grid cell, multiple factors are calculated using the method described in the above embodiment, and then input into the RUSLE model to obtain the soil erosion amount for each grid cell. Because the soil erosion prediction model only considers the influence of each grid cell's own environmental factors on soil erosion, and does not consider the influence of the interaction between two adjacent grid cells on soil erosion, the soil erosion amount obtained based on the RUSLE model is the initial soil erosion amount.

[0032] S104: Based on the geological and environmental characteristics between two adjacent grids, the soil erosion of each grid in the two adjacent grids is corrected to obtain the corrected soil erosion of each grid.

[0033] In this embodiment, the soil erosion amount of two adjacent grids is corrected by the interaction between them to obtain the corrected soil erosion amount of each grid. During the soil erosion process, the two adjacent grids do not exist in isolation, but are spatially associated through the exchange of materials such as sediment and runoff, energy transfer, and environmental conditions such as topography, vegetation, geology, and wind force and direction.

[0034] Traditional independent grid calculations only consider the local conditions of a single grid, neglecting the dynamic interaction between two adjacent grids, leading to biases in soil erosion predictions. The topographic features and soil permeability of two adjacent grids jointly determine runoff paths and sediment transport patterns; the upstream grid directly influences the erosion input to the downstream grid. For example, if the upstream grid has a steep slope and poor soil permeability, surface runoff from rainfall will rapidly converge at the downstream grid. Even if the downstream grid has a gentle slope, it may still receive additional runoff energy from the upstream, resulting in erosion significantly higher than the independently calculated value. Sediment from erosion at the upstream grid is transported to the downstream grid with runoff, potentially covering the downstream surface through deposition, reducing vegetation cover, or altering soil texture. If the downstream grid's soil erosion calculation does not consider the upstream sediment input, its actual erosion load will be underestimated.

[0035] The protective effect of vegetation cover between two adjacent grids can be weakened or enhanced by wind force and direction, thus affecting the soil erosion of the two adjacent grids. For example, the protective effect of the windbreak forest belt planted on grid I (upstream) on grid J (downstream) is highly dependent on wind direction. If the wind direction is perpendicular to the forest belt, the sand interception efficiency of the forest belt can reach more than 80%; if the wind direction is parallel to the forest belt, the protective effect of the forest belt may drop to below 30%. In this case, the erosion of grid J needs to be corrected according to the actual wind direction to reflect the dynamic changes in the protective effectiveness of the forest belt.

[0036] In this embodiment, the geological and environmental characteristics between two adjacent grids profoundly influence the soil erosion of the two adjacent grids through mechanisms such as runoff transport, sediment exchange, dynamic transmission, and changes in protective effectiveness. Therefore, by correcting the soil erosion of each grid in two adjacent grids based on the geological and environmental characteristics between them, the corrected soil erosion can be obtained, enabling an accurate assessment of the soil erosion of each grid.

[0037] S105: Based on the corrected soil erosion of each grid and the soil erosion of each grid in each historical period, the predicted soil erosion of each grid is obtained, and based on the predicted soil erosion of each grid, the predicted soil erosion of the target area is obtained.

[0038] In this embodiment, the corrected soil erosion amount for each grid is the soil erosion amount for each grid in the current period. To predict the soil erosion amount for each grid in future periods, the soil erosion amount for each grid in each historical period is needed. For example, for grid A1, the corrected soil erosion amount in the current period is q1. Based on the time sequence, moving backward from the current period, the soil erosion amount for the first historical period is qt1, the soil erosion amount for the second historical period is qt2, the soil erosion amount for the third historical period is qt3, and so on. Based on the variation pattern of soil erosion amount in multiple periods, the predicted soil erosion amount for each grid is obtained. The implementation methods include trend extrapolation, periodic fluctuation, and machine learning model methods. The trend extrapolation method is used when historical data shows a stable linear trend, and future soil erosion amount can be predicted directly by extrapolating the slope. The periodic fluctuation method is used when historical data shows periodicity due to climate or human activities, and periodic features can be extracted and combined with the current period to predict future soil erosion amount. The machine learning model approach involves using historical data with significant non-linear characteristics. Soil erosion data from each historical period can be used as training data to train a Long Short-Term Memory (LSTM) network model or a Random Forest model, allowing it to learn time dependencies. The corrected soil erosion values ​​for each grid cell in the current period are then input into the trained LTM network model or Random Forest model to obtain the predicted soil erosion values ​​for each grid cell. Finally, the predicted soil erosion values ​​for each grid cell are weighted and summed to obtain the predicted soil erosion value for the target area.

[0039] As can be seen from the above, in order to improve the accuracy of soil erosion prediction in the target area, this application proposes a method that combines grid division with soil erosion data.

[0040] This application divides the target area into multiple grids, first predicts the soil erosion within each grid, and then obtains the predicted soil erosion value for the entire target area based on the predicted values ​​of soil erosion within each grid. By calculating and summarizing the data for each grid independently, this application solves the problem of traditional methods treating the target area as a whole, which is prone to masking internal heterogeneity due to the "averaging effect," and significantly improves the spatial refinement and local accuracy of soil erosion prediction.

[0041] For each grid cell, this application obtains an initial value of soil erosion for that grid cell in the current period based on the target dataset acquired within the grid and the RUSLE model. Then, it corrects the initial value of soil erosion based on the interaction between two adjacent grid cells to obtain a corrected soil erosion value. Based on the corrected soil erosion value and the soil erosion values ​​in each historical period, the application predicts the soil erosion value for that grid cell. This application corrects erosion based on the geological and environmental characteristics of two adjacent grid cells, effectively restoring the dynamic interaction mechanism of grids with different terrains and environments during the actual erosion process. This reduces prediction bias caused by neglecting cross-grid material and energy exchange, significantly enhancing the accuracy and reliability of soil erosion prediction.

[0042] In one embodiment, the data corresponding to the hydrometeorological parameters include the grid average soil moisture content, the average soil moisture content within a preset time before rainfall, and rainfall data; in this embodiment, the rainfall erosivity factor of each grid is obtained based on the target dataset, including: For each grid cell, the total kinetic energy of a single rainfall event and the maximum rainfall intensity per unit are obtained based on the rainfall data; the energy of a single rainfall event is obtained based on the total kinetic energy of a single rainfall event and the maximum rainfall intensity per unit. A power function is constructed based on the grid average soil moisture content and the average soil moisture content within a preset time before rainfall, and the soil moisture is obtained based on the power function. The rainfall erosivity factor for each grid cell is obtained based on the energy of a single rainfall event and soil moisture.

[0043] In this embodiment, the total kinetic energy of a single rainfall event is determined based on the total amount of rainfall in the rainfall data. The specific calculation formula is as follows: Where E represents the total kinetic energy of a single rainfall event, and P represents the total amount of rainfall in a single event. The maximum rainfall intensity per unit time in the rainfall data can be understood as the maximum 30-minute rainfall intensity, the maximum hourly rainfall intensity, or other maximum rainfall intensity per unit time. In this embodiment, it can be the maximum 30-minute rainfall intensity.

[0044] A power function is constructed based on the grid average soil moisture content and the average soil moisture content within a preset time period before rainfall. This includes constructing the power function using the ratio of the grid average soil moisture content to the average soil moisture content within a preset time period before rainfall, and the following formula: ,in, Indicates soil moisture. This indicates the average soil moisture content within a preset timeframe before rainfall. The grid average soil moisture content is represented by , and m represents the nonlinear response index of soil moisture, reflecting the saturation effect of moisture content on erosiveness; a value of 0.7 is typically used. In this embodiment, the preset time period can be 24 hours, 48 ​​hours, or other time intervals. The grid average soil moisture content and the average soil moisture content within the preset time period before rainfall in this embodiment can be obtained using equipment such as a soil moisture velocimeter and a tensiometer.

[0045] The rainfall erosivity factor for each grid cell is obtained based on the energy of a single rainfall event and soil moisture, including: the rainfall-soil synergistic effect coefficient, the number of rainfall events, the energy of a single rainfall event, and the soil moisture.

[0046] In this embodiment, the rainfall-soil synergistic effect coefficient k is used to characterize the differences in the coupling effect between single rainfall energy and soil moisture under different grid soil background conditions. When k ≥ 1, the soil within the grid is more sensitive to rainfall erosion. For example, in clay areas, increased moisture easily triggers surface crust damage, amplifying the erosion energy effect. When k < 1, the soil within the grid is more resistant to erosion. For example, in sandy soil areas, increased moisture has a weak promoting effect on erosion. The rainfall-soil synergistic effect coefficients differ for different soil types under different soil moisture conditions. This correspondence can be obtained through a soil type-soil moisture-synergistic effect coefficient mapping table.

[0047] In this embodiment, the number of rainfall events can be monthly rainfall events, annual rainfall events, or other representations. When the number of rainfall events is annual rainfall events, the rainfall erosivity factor represents the cumulative amount on an annual scale; when the number of rainfall events is monthly rainfall events, the rainfall erosivity factor represents the cumulative amount on a monthly scale.

[0048] The formula for calculating the rainfall erosivity factor is:

[0049] Where R represents the rainfall erosivity factor, k represents the rainfall-soil synergistic effect coefficient, N represents the number of rainfall events, E represents the total kinetic energy of a single rainfall event, and I represents the maximum rainfall intensity per unit. This indicates the average soil moisture content (by volume, %) within a preset time period before rainfall. This represents the average soil moisture content per square meter (by volume, %), where m represents the nonlinear response index of soil moisture. P represents the total amount of rainfall in a single event.

[0050] In this embodiment, existing technologies such as the Wischmeier empirical formula rely solely on the statistical characteristics of rainfall data, neglecting the dynamic modulation effect of prior soil moisture on rainfall erosivity. That is, soil moisture can indirectly alter the effective erosion amount of rainfall by affecting surface erosion resistance, such as porosity. This embodiment incorporates soil moisture, making the calculated rainfall erosivity factor more accurate. Furthermore, this embodiment introduces a rainfall-soil synergistic effect coefficient to adapt to differences in soil-rainfall coupling mechanisms across different grids or regions, resulting in a more accurate calculation of the rainfall erosivity factor reflecting the actual environment.

[0051] In one embodiment, the data corresponding to the ecological parameters includes satellite remote sensing data and location data; the vegetation cover factor of each grid is obtained based on the target dataset, including: Determine the vegetation cover factor dataset, which contains multiple location-based vegetation cover factors; For each grid cell, a target vegetation cover factor is selected from the vegetation cover factor dataset based on satellite remote sensing data and location data, and the target vegetation cover factor is used as the vegetation cover factor for each grid cell.

[0052] In this embodiment, satellite remote sensing data refers to surface observation data acquired through satellite sensors, including spectral information, temperature and humidity data, and image data. The vegetation status within the target grid, such as vegetation type, vegetation coverage, and vegetation health, can be retrieved using the spectral information. Location data consists of coordinate information used to identify geospatial locations, such as latitude and longitude, and UTM grid encoding. Location data allows for precise positioning of the absolute location of each grid cell.

[0053] In this embodiment, the vegetation cover factor dataset contains multiple vegetation cover factors based on location features. This can be understood as the vegetation cover factor dataset including multiple vegetation cover factors, and the location features can be geographical locations, such as latitude and longitude data. These vegetation cover factors are classified and stored based on geographical locations, that is, each geographical location corresponds to multiple different vegetation cover factors.

[0054] For each grid cell, firstly, multiple different vegetation cover factors corresponding to the current grid cell's geographical location are determined. Then, the vegetation characteristics of the current grid cell are retrieved using satellite remote sensing data. Based on the similarity or difference between the vegetation characteristics and multiple different vegetation cover factors, the target vegetation cover factor is obtained.

[0055] This embodiment determines the target vegetation cover factor based on satellite remote sensing data and location data, ensuring that it satisfies both the geographical location characteristics of the current grid and the actual vegetation status. This solves the problem that traditional methods of determining the target vegetation cover factor directly based on land use type cannot dynamically select based on the grid's geographical location, making the calculation results of the vegetation cover factor for each grid more accurate.

[0056] In one embodiment, each geographic location in the vegetation cover factor dataset corresponds to multiple vegetation cover factors, and the location data includes latitude and longitude data; the target vegetation cover factor is obtained by filtering from the vegetation cover factor dataset based on satellite remote sensing data and location data, including: Retrieve a set of candidate vegetation cover factors corresponding to the geographical location from the vegetation cover factor dataset based on latitude and longitude data; The vegetation characteristics of the current grid are obtained based on satellite remote sensing data. The vegetation characteristics include vegetation type, coverage and vegetation health. Calculate the matching degree between the vegetation features of the current raster and each factor in the candidate vegetation cover factor set, and obtain the target vegetation cover factor based on the matching degree of multiple factors.

[0057] In this embodiment, satellite remote sensing data can be used to infer the vegetation type, vegetation coverage, and vegetation health characteristics of the current grid through reflectance information in multispectral / hyperspectral bands. Specifically, the satellite remote sensing data is preprocessed with radiometric correction, atmospheric correction, and geometric correction to ensure the accuracy of the multispectral data; the reflectance of key bands in the preprocessed multispectral data is extracted to construct a spectral feature vector; and the vegetation type of the current grid is obtained based on the spectral feature vector using a random forest model, support vector machine, or deep learning model.

[0058] The normalized vegetation index is obtained by inversion based on the vegetation index method: Wherein, NDVI is the Normalized Difference Vegetation Index, NIR is the near-infrared reflectance, and RED is the red reflectance. Both near-infrared and red reflectance can be obtained by extracting the reflectance of key bands from the preprocessed multispectral data.

[0059] Vegetation cover is obtained based on the relationship between the normalized vegetation index and vegetation cover, namely: Where FVC represents vegetation cover and NDVI max This represents the maximum NDVI value in sparsely vegetated and densely vegetated areas. min This represents the minimum NDVI value in sparsely vegetated and densely vegetated areas.

[0060] Vegetation health reflects the growth status of vegetation and can be obtained by weighted summation of indicators such as leaf area index (LAI), chlorophyll content, and water content after dimensionality reduction using principal component analysis. LAI can be obtained by inversion from the red and near-infrared bands of the spectral spectrum; a decrease in LAI indicates inhibited vegetation growth. Chlorophyll content is obtained by inversion from the offset of the red band, and water content can be obtained by inversion from the short-wave infrared band.

[0061] After obtaining the vegetation type, cover, and vegetation health, the matching degree between the vegetation features of the current raster and each factor in the candidate vegetation cover factor set can be calculated based on the following formula:

[0062] in, Indicates the degree of matching. , and Indicates the weighting factor. Indicates the vegetation type of the current raster. This represents the vegetation type corresponding to the i-th vegetation cover factor. This indicates the vegetation cover of the current grid cell. This represents the vegetation coverage corresponding to the i-th vegetation cover factor. This indicates the current vegetation health status of the grid. This represents the vegetation health corresponding to the i-th vegetation cover factor.

[0063] This indicates the degree of difference between the vegetation features of the current raster and each factor in the candidate vegetation cover factor set. The smaller the degree of difference, the higher the matching degree.

[0064] As can be seen from the above, this embodiment first determines candidate vegetation cover factors using latitude and longitude data, breaking through the limitations of the traditional static table lookup method and dynamically adapting to vegetation characteristics at different locations. Then, vegetation type, vegetation coverage, and vegetation health are obtained by inverting based on satellite remote sensing data. Based on the above three parameters, the target vegetation cover factor with the smallest difference and the highest matching degree is selected from the candidate vegetation cover factors, which improves the matching accuracy between the target vegetation cover factor and the actual ecological scene and provides reliable input parameters for the subsequent calculation of soil erosion.

[0065] In one embodiment, the soil erosion amount of each grid in two adjacent grids is corrected based on the geological and environmental characteristics between the two adjacent grids to obtain the corrected soil erosion amount of each grid, including: Based on the terrain parameters between two adjacent grids, the slope difference, slope length ratio, and confluence path overlap between the two adjacent grids are obtained; based on the slope difference, slope length ratio, and confluence path overlap, the terrain connectivity factor is obtained using the terrain connectivity model. Wind data is obtained based on hydrological and meteorological parameters between two adjacent grids, and vegetation type data is obtained based on ecological parameters between two adjacent grids; environmental heterogeneity factors are obtained based on wind data and vegetation type data. The correction coefficient between two adjacent grids is obtained by integrating the topographic connectivity factor and the environmental heterogeneity factor; the soil erosion of the two adjacent grids is corrected based on the correction coefficient to obtain the corrected soil erosion of each grid.

[0066] In this embodiment, the terrain parameters include slope length, slope, and confluence path. The slope difference, slope length ratio, and confluence path overlap in this embodiment are all normalized data. The slope difference, slope length ratio, and confluence path overlap between two adjacent grid cells are obtained based on the terrain parameters between them, including: The slope difference between two adjacent grid cells is calculated based on the first formula, which is: ,in, This represents the normalized slope difference between two adjacent grid cells. This represents the slope difference between two adjacent grid cells. This represents the minimum slope difference between two adjacent grid cells within the target area. This represents the maximum slope difference between two adjacent grid cells within the target area.

[0067] The slope ratio between two adjacent grid cells is calculated based on the second formula, which is: ,in, This represents the normalized slope ratio between two adjacent grid cells. This indicates the ratio of the slope lengths between two adjacent grid cells. This represents the minimum slope length ratio between two adjacent grid cells within the target area. This represents the maximum value of the slope length ratio between two adjacent grid cells within the target area.

[0068] Normalized convergence path overlap .

[0069] Based on slope difference, slope length ratio, and convergence path overlap, a terrain connectivity factor is obtained using a terrain connectivity model. The terrain connectivity model is as follows: in, Represents the terrain connectivity factor. .

[0070] In one embodiment, wind data includes wind speed difference and wind direction angle, and vegetation type data includes vegetation coverage and erosion resistance coefficients corresponding to different vegetation types. Environmental heterogeneity factors are obtained based on the wind data and vegetation type data, including: The wind interaction term is calculated based on the relationship between wind speed difference and wind direction angle; the vegetation interaction term is calculated based on vegetation coverage and erosion resistance coefficient; the cross-factor modulation term is calculated based on wind data and vegetation type data; the environmental heterogeneity factor is obtained by multiplying the wind interaction term, vegetation interaction term and cross-factor modulation term.

[0071] In this embodiment, the wind interaction item is: Where 'a' represents the nonlinear effect coefficient of wind speed difference, which is a constant, defaulting to 0.1. Indicates wind speed difference, represents the wind direction angle, and b represents the wind direction angle response coefficient, which is a constant with a default value of 0.01. This represents the wind interaction term; the larger the value, the greater the interference of wind factors on runoff transmission.

[0072] The vegetation interaction item is: ,in, Indicates vegetation interaction items, represents vegetation coverage, c represents the erosion resistance coefficient scaling factor, which is a constant with a default value of 1.0, and d represents the coverage-erosion resistance synergy index, which is a constant with a default value of 0.5.

[0073] The cross-factor modulation term is: Where e represents the wind weakening coefficient of vegetation, which is a constant and defaults to 0.005, and f represents the vegetation blocking coefficient of wind, which defaults to 0.01. This represents the cross-factor modulation term; the larger the value, the stronger the inhibitory effect of vegetation factors on erosion.

[0074] Environmental heterogeneity factors are: , .

[0075] In this embodiment, a correction coefficient between two adjacent grid cells is obtained by fusing the terrain connectivity factor and the environmental heterogeneity factor. This correction coefficient is the result of multiplying the terrain connectivity factor and the environmental heterogeneity factor. (See reference...) Figure 2 The adjacent grids to grid 5 are grids 2, 4, 6, and 8. Assuming the correction coefficient between grids 2 and 5 is r1, between grids 4 and 5 is r2, between grids 6 and 5 is r3, and between grids 8 and 5 is r4, then the actual correction coefficient for grid 5 is the average of r1, r2, r3, and r4. The corrected soil erosion for grid 5 is ( (r1) × Soil erosion (initial soil erosion). In this embodiment, since grid 2 is above grid 5, it can be understood as upstream, so r1 is a negative value; grid 8 is below grid 5, so r4 is a positive value; r2 and r3 are also positive values.

[0076] As can be seen from the above, this application constructs a two-factor correction system by quantifying the topographic connectivity and environmental heterogeneity of adjacent grids, which can effectively capture the impact of spatial interactions on soil erosion. The topographic connectivity factor reveals the laws of runoff transfer and energy distribution, while the environmental heterogeneity factor reflects the dynamic adjustment of wind power and vegetation protection. The correction coefficient resulting from the fusion of these two factors overcomes the limitations of traditional isolated calculations. The corrected soil erosion data more closely reflects the actual erosion process, providing a reliable basis for accurately predicting future soil erosion in target areas.

[0077] In one embodiment, the predicted soil erosion of each grid is obtained based on the corrected soil erosion of each grid and the soil erosion of each grid in each historical period, including: For each grid cell, the soil erosion variation characteristics of that grid cell are obtained based on the amount of soil erosion in each period of the grid cell's history. The soil erosion variation characteristics include seasonal variation characteristics and specific weather condition variation characteristics. The soil erosion variation characteristics of the grid and the corrected soil erosion amount of the grid are input into the neural network model to obtain the predicted soil erosion amount of the grid.

[0078] In this embodiment, the corrected soil erosion amount for each grid is the soil erosion amount for each grid in the current period. The soil erosion amount for each grid in each historical period can be the soil erosion amount for each grid in each month, year, or two-year period. Soil erosion variation characteristics represent seasonal variation characteristics extracted from historical data, such as higher soil erosion during the rainy season than during the dry season, an average erosion amount during the rainy season being 40% higher than during the dry season in the past 5 years, and abnormal erosion caused by specific weather conditions, such as heavy rain or drought. The neural network model can be a Long Short-Term Memory (LSTM) network model or a Random Forest model. The neural network model is trained based on the soil erosion amount of each grid in each historical period. The model learns the mapping relationship between historical features and future erosion through training, and outputs the predicted erosion amount for the grid in the next period, comprehensively reflecting the influence of long-term trends and short-term fluctuations. The soil erosion variation characteristics of the current grid and the corrected soil erosion amount of the grid are input into the trained neural network model to obtain the predicted soil erosion amount for the grid in the next period.

[0079] As can be seen from the above, this embodiment deeply integrates the dynamic evolution patterns over time with the current corrected erosion amount by leveraging seasonal variation characteristics and specific weather condition variations, overcoming the limitations of traditional predictions that rely solely on the current state. The neural network model can non-linearly learn the correlation between historical trends and current characteristics, enabling the prediction results to both conform to long-term evolution trends and respond to short-term environmental changes, significantly improving the timeliness and reliability of soil erosion prediction.

[0080] Corresponding to the soil erosion prediction method in the above embodiments, Figure 3 This is a structural block diagram of a soil erosion prediction device provided according to an embodiment of this application. For ease of explanation, only the parts relevant to the embodiment of this application are shown. References Figure 3 The soil erosion prediction device 20 includes: a data acquisition unit 21, a multi-factor calculation unit 22, a data calculation unit 23, a data correction unit 24, and a prediction unit 25.

[0081] Among them, the data acquisition unit 21 is used to acquire the target dataset corresponding to each grid. Each grid is obtained after dividing the target area. The target area is the area to be predicted for soil erosion. The target dataset includes data corresponding to geological parameters, topographic parameters, hydro-meteorological parameters and ecological parameters respectively. The multi-factor calculation unit 22 is used to obtain the rainfall erosivity factor, soil erosion factor, slope length and slope factor, vegetation cover factor and soil and water conservation factor of each grid based on the target dataset. Data calculation unit 23 is used to calculate the soil erosion of each grid cell based on rainfall erosivity factor, soil erosion factor, slope length and slope factor, vegetation cover factor and soil and water conservation factor using the RUSLE model. Data correction unit 24 corrects the soil erosion of each grid in two adjacent grids based on the geological and environmental characteristics between the two adjacent grids to obtain the corrected soil erosion of each grid. The prediction unit 25 is used to obtain the predicted soil erosion of each grid based on the corrected soil erosion of each grid and the soil erosion of each grid in each historical period, and to obtain the predicted soil erosion of the target area based on the predicted soil erosion of each grid.

[0082] In one embodiment of this application, the data corresponding to the hydrological and meteorological parameters include the grid-averaged soil moisture content, the average soil moisture content within a preset time before rainfall, and rainfall data; the multi-factor calculation unit 22 is specifically used for: For each grid cell, the total kinetic energy of a single rainfall event and the maximum rainfall intensity per unit are obtained based on the rainfall data; the energy of a single rainfall event is obtained based on the total kinetic energy of a single rainfall event and the maximum rainfall intensity per unit. A power function is constructed based on the grid average soil moisture content and the average soil moisture content within a preset time before rainfall, and the soil moisture is obtained based on the power function. The rainfall erosivity factor for each grid cell is obtained based on the energy of a single rainfall event and soil moisture.

[0083] In one embodiment of this application, the power function is: ; in, Indicates soil moisture. This indicates the average soil moisture content within a preset timeframe before rainfall. represents the average soil moisture content of the grid, and m represents the nonlinear response index of soil moisture.

[0084] In one embodiment of this application, the multi-factor calculation unit 22 is specifically used for: Based on the energy of a single rainfall event and the soil moisture, the rainfall erosivity factor for each grid cell is calculated using the following formula: ; Where R represents the rainfall erosivity factor, k represents the rainfall-soil synergistic effect coefficient, N represents the number of rainfall events, E represents the total kinetic energy of a single rainfall event, and I represents the maximum rainfall intensity per unit. This indicates the average soil moisture content within a preset timeframe before rainfall. This indicates the average soil moisture content of the grid. The value represents soil moisture, and m represents the nonlinear response index of soil moisture. P represents the total amount of rainfall in a single event.

[0085] In one embodiment of this application, the data corresponding to the ecological parameters include satellite remote sensing data and location data; the multi-factor calculation unit 22 is specifically used for: Determine the vegetation cover factor dataset, which contains multiple location-based vegetation cover factors; For each grid cell, a target vegetation cover factor is selected from the vegetation cover factor dataset based on satellite remote sensing data and location data, and the target vegetation cover factor is used as the vegetation cover factor for that grid cell.

[0086] In one embodiment of this application, each geographical location in the vegetation cover factor dataset corresponds to multiple vegetation cover factors, and the location data includes latitude and longitude data; the multi-factor calculation unit 22 is specifically used for: Retrieve a set of candidate vegetation cover factors corresponding to the geographical location from the vegetation cover factor dataset based on latitude and longitude data; The vegetation characteristics of the current grid are obtained based on satellite remote sensing data. The vegetation characteristics include vegetation type, coverage and vegetation health. Calculate the matching degree between the vegetation features of the current raster and each factor in the candidate vegetation cover factor set, and obtain the target vegetation cover factor based on the matching degree of multiple factors.

[0087] In one embodiment of this application, the data correction unit 24 is specifically used for: Based on the terrain parameters between two adjacent grids, the slope difference, slope length ratio, and confluence path overlap between the two adjacent grids are obtained; based on the slope difference, slope length ratio, and confluence path overlap, the terrain connectivity factor is obtained using the terrain connectivity model. Wind data is obtained based on hydrological and meteorological parameters between two adjacent grids, and vegetation type data is obtained based on ecological parameters between two adjacent grids; environmental heterogeneity factors are obtained based on wind data and vegetation type data. The correction coefficient between two adjacent grids is obtained by integrating the topographic connectivity factor and the environmental heterogeneity factor; the soil erosion of the two adjacent grids is corrected based on the correction coefficient to obtain the corrected soil erosion of each grid.

[0088] In one embodiment of this application, the wind data includes wind speed difference and wind direction angle, and the vegetation type data includes vegetation coverage and erosion resistance coefficient corresponding to the vegetation type; the data correction unit 24 is specifically used for: Environmental heterogeneity factors were obtained based on wind and vegetation type data, including: The wind interaction term is calculated based on the relationship between wind speed difference and wind direction angle, and the vegetation interaction term is calculated based on vegetation coverage and erosion resistance coefficient. Calculate cross-factor modulation terms based on wind force data and vegetation type data; The environmental heterogeneity factor is obtained by multiplying the wind interaction term, vegetation interaction term, and cross-factor modulation term.

[0089] In one embodiment of this application, the prediction unit 25 is specifically used for: For each grid cell, the soil erosion variation characteristics of that grid cell are obtained based on the amount of soil erosion in each period of the grid cell's history. The soil erosion variation characteristics include seasonal variation characteristics and specific weather condition variation characteristics. The soil erosion variation characteristics of the grid and the corrected soil erosion amount of the grid are input into the neural network model to obtain the predicted soil erosion amount of the grid.

[0090] See Figure 4 , Figure 4 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Figure 4The electronic device 300 in this embodiment may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memories 304 store computer programs, including program instructions. The processors 301 execute the program instructions stored in the memories 304. Specifically, the processors 301 are configured to invoke the program instructions to perform the functions of the units in the above-described device embodiments, for example... Figure 3 The functions of the data acquisition unit 21, multi-factor calculation unit 22, data calculation unit 23, data correction unit 24, and prediction unit 25 are shown.

[0091] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0092] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0093] The memory 304 may include read-only memory and random access memory, and provides instructions and data to the processor 301. A portion of the memory 304 may also include non-volatile random access memory.

[0094] In specific implementations, the processor 301, input device 302, and output device 303 described in the embodiments of this application can execute the implementation method described in the soil erosion prediction method provided in the embodiments of this application, or they can execute the implementation method of the electronic device described in the embodiments of this application, which will not be repeated here.

[0095] In another embodiment of this application, a computer-readable storage medium is provided. This computer-readable storage medium stores a computer program, which includes program instructions. When executed by a processor, the program instructions implement all or part of the processes in the methods described above. Alternatively, the computer program can instruct related hardware to complete the process. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0096] The computer-readable storage medium can be an internal storage unit of the electronic device in any of the foregoing embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device. Furthermore, the computer-readable storage medium can include both internal and external storage units of the electronic device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0097] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0098] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the electronic devices and units described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0099] In the several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces or units, or it may be an electrical, mechanical, or other form of connection.

[0100] 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 the embodiments of this application, depending on actual needs.

[0101] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or one or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0102] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for predicting soil erosion, characterized in that, include: Obtain the target dataset corresponding to each grid, wherein each grid is obtained after dividing the target area, the target area is the area to be predicted for soil erosion, and the target dataset includes data corresponding to geological parameters, topographic parameters, hydro-meteorological parameters and ecological parameters respectively; Based on the target dataset, the rainfall erosivity factor, soil erosion factor, slope length and gradient factor, vegetation cover factor and soil and water conservation factor of each grid are obtained. Based on the rainfall erosivity factor, the soil erosion factor, the slope length and gradient factor, the vegetation cover factor, and the soil and water conservation factor, the soil erosion of each grid is calculated using the RUSLE model. Based on the geological and environmental characteristics between two adjacent grids, the soil erosion of each grid in the two adjacent grids is corrected to obtain the corrected soil erosion of each grid. Based on the corrected soil erosion of each grid and the soil erosion of each grid in each historical period, the predicted soil erosion of each grid is obtained, and based on the predicted soil erosion of each grid, the predicted soil erosion of the target area is obtained.

2. The method as described in claim 1, characterized in that, The data corresponding to the hydrological and meteorological parameters include grid average soil moisture content, average soil moisture content within a preset time before rainfall, and rainfall data; Based on the target dataset, the rainfall erosivity factor of each raster is obtained, including: For each grid cell, the total kinetic energy of a single rainfall event and the maximum rainfall intensity per unit are obtained based on the rainfall data, and the energy of a single rainfall event is obtained based on the total kinetic energy of the single rainfall event and the maximum rainfall intensity per unit. A power function is constructed based on the average soil moisture content of the grid and the average soil moisture content within a preset time before rainfall, and the soil moisture is obtained based on the power function. The rainfall erosivity factor for each grid cell is obtained based on the single rainfall energy and the soil moisture.

3. The method as described in claim 2, characterized in that, The power function is: ； in, Indicates soil moisture. This indicates the average soil moisture content within a preset timeframe before rainfall. represents the average soil moisture content of the grid, and m represents the nonlinear response index of soil moisture.

4. The method as described in claim 2, characterized in that, The method of obtaining the rainfall erosivity factor for each grid cell based on the single rainfall energy and the soil moisture includes: Based on the single rainfall energy and the soil moisture, the rainfall erosivity factor for each grid cell is calculated using the following formula: ； Where R represents the rainfall erosivity factor, k represents the rainfall-soil synergistic effect coefficient, N represents the number of rainfall events, E represents the total kinetic energy of a single rainfall event, and I represents the maximum rainfall intensity per unit. This indicates the average soil moisture content within a preset timeframe before rainfall. This indicates the average soil moisture content of the grid. The value represents soil moisture, and m represents the nonlinear response index of soil moisture. P represents the total amount of rainfall in a single event.

5. The method as described in claim 1, characterized in that, The data corresponding to the ecological parameters include satellite remote sensing data and location data; The vegetation cover factor for each raster is obtained based on the target dataset, including: A vegetation cover factor dataset is determined, which contains multiple vegetation cover factors based on location features; For each grid cell, a target vegetation cover factor is selected from the vegetation cover factor dataset based on the satellite remote sensing data and the location data, and the target vegetation cover factor is used as the vegetation cover factor for that grid cell.

6. The method as described in claim 5, characterized in that, In the vegetation cover factor dataset, each geographical location corresponds to multiple vegetation cover factors, and the location data includes latitude and longitude data; The process of obtaining target vegetation cover factors from the vegetation cover factor dataset based on the satellite remote sensing data and the location data includes: Based on the latitude and longitude data, a set of candidate vegetation cover factors corresponding to the geographical location is retrieved from the vegetation cover factor dataset; The vegetation characteristics of the current grid are obtained based on the satellite remote sensing data, and the vegetation characteristics include vegetation type, coverage and vegetation health. Calculate the matching degree between the vegetation features of the current grid and each factor in the candidate vegetation cover factor set, and obtain the target vegetation cover factor based on the matching degree of multiple factors.

7. The method as described in claim 1, characterized in that, Based on the geological and environmental characteristics between two adjacent grids, the soil erosion of each grid in the two adjacent grids is corrected to obtain the corrected soil erosion of each grid, including: Based on the terrain parameters between the two adjacent grids, the slope difference, slope length ratio, and confluence path overlap between the two adjacent grids are obtained; based on the slope difference, slope length ratio, and confluence path overlap, the terrain connectivity factor is obtained using the terrain connectivity model; Wind data is obtained based on the hydrological and meteorological parameters between the two adjacent grids; vegetation type data is obtained based on the ecological parameters between the two adjacent grids; environmental heterogeneity factors are obtained based on the wind data and the vegetation type data. The correction coefficient between two adjacent grids is obtained by fusing the terrain connectivity factor and the environmental heterogeneity factor; the soil erosion of the two adjacent grids is corrected based on the correction coefficient to obtain the corrected soil erosion of each grid.

8. The method as described in claim 7, characterized in that, The wind data includes wind speed difference and wind direction angle; the vegetation type data includes vegetation coverage and erosion resistance coefficient corresponding to the vegetation type. Environmental heterogeneity factors are obtained based on the wind data and the vegetation type data, including: The wind force interaction term is calculated based on the relationship between the wind speed difference and the wind direction angle, and the vegetation interaction term is calculated based on the vegetation coverage and the erosion resistance coefficient. Calculate the cross-factor modulation term based on the wind data and the vegetation type data; The environmental heterogeneity factor is obtained by multiplying the wind interaction term, the vegetation interaction term, and the cross-factor modulation term.

9. The method as described in claim 1, characterized in that, The predicted soil erosion for each grid cell is obtained based on the corrected soil erosion for each grid cell and the soil erosion for each historical period, including: For each grid cell, the soil erosion variation characteristics of that grid cell are obtained based on the amount of soil erosion in each period of the grid cell's history. The soil erosion variation characteristics include seasonal variation characteristics and specific weather condition variation characteristics. The soil erosion variation characteristics of the grid and the corrected soil erosion amount of the grid are input into the neural network model to obtain the predicted soil erosion amount of the grid.

Images