Erosion gully density rapid estimation method and related equipment

Abstract

The invention relates to the technical field of regional soil erosion, in particular to an erosion gully density rapid estimation method and related equipment, after remote sensing image data of a target region is obtained, the region is divided into different types of prediction units according to landform unit characteristics, and influence factor data is extracted from each unit. And then inputting the data into a pre-trained target prediction model to obtain an erosion gully density prediction value corresponding to each prediction unit. And finally, based on the erosion gully density prediction values of all the prediction units, drawing an erosion gully density spatial distribution diagram of the target area.

Description

Technical Field

[0001] This invention relates to the field of regional soil erosion technology, specifically to a method for rapid estimation of erosion gully density and related equipment. Background Technology

[0002] The occurrence and development of gullies is a typical form of soil degradation on sloping farmland. Their density reflects the degree of land erosion and damage, serving as a crucial basis for formulating soil and water conservation measures. Currently, obtaining regional-scale gully density mainly relies on two methods: whole-sample surveys and sampling surveys. Whole-sample surveys generally rely on remote sensing interpretation combined with partial field verification, achieving large-scale coverage, but are costly and inefficient. Furthermore, they are limited by the image resolution available for large areas, making it difficult to identify small gullies. Sampling surveys are less costly and more accurate, but the results are limited to the sample area, requiring extrapolation to the entire region using interpolation or statistical methods. Existing interpolation and statistical methods are mostly based on linear assumptions, making it difficult to reveal the nonlinear relationship between gully density and factors such as slope and rainfall, resulting in insufficient regional prediction accuracy. In recent years, machine learning methods have been preliminarily applied in gully sensitivity analysis, demonstrating strong modeling capabilities. However, their application is currently mainly focused on classification problems, with insufficient applications for large-scale gully density estimation. Moreover, they have not fully integrated high-quality sampling data with regression-based artificial intelligence algorithms, making it difficult to achieve high-precision estimation of regional-scale gully density.

[0003] Therefore, it is necessary to develop a new, efficient, accurate, and low-cost method for estimating the density of regional-scale erosion gullies in order to overcome the limitations of existing technologies. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and related equipment for rapid estimation of erosion trench density in large-scale areas, in order to address the shortcomings of the prior art.

[0005] The objective of this invention is achieved through the following technical solutions: In a first aspect, the present invention provides a method for rapidly estimating the density of erosion trenches, comprising: Acquire remote sensing image data of the target area; divide the target area into different types of prediction units based on the geomorphic unit characteristics of the target area; extract the influencing factor data from each prediction unit; The influencing factor data extracted from each prediction unit is input into the pre-trained target prediction model to obtain the corresponding predicted value of erosion gully density; Based on the predicted erosion gully density values ​​of all predicted units, a spatial distribution map of erosion gully density in the target area is obtained.

[0006] As a further improvement of the present invention, the target area is divided according to the geomorphic unit characteristics of the target area to obtain different types of prediction units, including: When the topographic features are low mountain and hilly areas and plateau areas, the watershed boundary is extracted using the hydrological analysis tools in ArcGIS based on the DEM data, and the watershed boundary is used as the prediction unit. When the topographic features are characterized as plains, create a fishing net in ArcGIS to establish equal-area grid cells and generate rectangular boundaries as prediction cells.

[0007] As a further improvement of the present invention, the prediction unit is provided with a unique coded identifier; the coded identifier serves as a unique identification code for spatial location and is used to establish a correspondence between the prediction result and the actual geographical location.

[0008] As a further improvement of the present invention, the influencing factor data includes at least one of topographic factors, rainfall factors, vegetation factors, soil factors, land use factors, lithological factors, and management factors.

[0009] As a further improvement of the present invention, the target prediction model adopts one of the following: random forest, support vector machine, convolutional neural network and Transformer model.

[0010] As a further improvement of the present invention, the training process of the target prediction model includes: A dataset of erosion gully samples was obtained by stratified sampling within the study area; Construct a database of factors influencing gully density within a known region; Based on the geomorphic unit characteristics of the study area, the study area is divided into different prediction units; Using the erosion gully density in the erosion gully sample dataset as the response variable and the data of each input influencing factor as the explanatory variable, a target prediction model is constructed. The target prediction model was trained using an erosion gully sample dataset, and the hyperparameters in the target prediction model were updated using cross-validation. The SHAP value analysis method is used to screen the impact factor data. The screened impact factor data is used as the input variable of the prediction unit. The target prediction model is updated based on the screened impact factor data to obtain the optimal target prediction model.

[0011] As a further improvement of the present invention, the SHAP value analysis method includes: using game theory principles to calculate the contribution direction and contribution intensity of global influence factor data to the prediction results, as well as the contribution direction and contribution intensity of local influence factor data to the prediction results.

[0012] Secondly, the present invention provides a rapid estimation system for erosion trench density, used to implement the above-mentioned rapid estimation method for erosion trench density, comprising: The data acquisition module acquires remote sensing image data of the target area; The prediction unit segmentation module divides the target area into different types of prediction units based on the geomorphic unit characteristics of the target area. The impact factor data extraction module extracts impact factor data from each prediction unit. The erosion gully density prediction module and the erosion gully density prediction estimation module input the influencing factor data extracted from each prediction unit into the pre-trained target prediction model to obtain the corresponding erosion gully density prediction value. The spatial distribution map generation module generates a spatial distribution map of erosion gully density in the target area based on the predicted erosion gully density values ​​of all predicted units.

[0013] Thirdly, the present invention provides a computer-readable storage medium storing one or more programs, the one or more programs including instructions that, when executed by a computing device, cause the computing device to perform the above-described rapid estimation method for erosion trench density.

[0014] Fourthly, the present invention provides a computing device, comprising: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including steps for performing the above-described rapid estimation method for erosion trench density.

[0015] The beneficial effects of this invention are as follows: This invention provides a rapid estimation method for gully density applicable to cultivated land. It breaks through the limitations of traditional unified prediction across the entire area, and its regional modeling is more in line with geographical patterns, reducing prediction errors. After geomorphic unit division, the influencing factors input into the model (such as soil type and rainfall) are more representative, enhancing the model's ability to handle spatial heterogeneity. Combined with time-series remote sensing data, the dynamic changes in gully development can be captured, improving the timeliness of prediction. This invention achieves high efficiency, high accuracy, and strong practicality in gully prediction, providing a scientific decision-making tool for soil and water conservation, agricultural planning, and ecological restoration.

[0016] Furthermore, in the prediction stage, the study area is divided into prediction units of small watersheds or regular grids based on geomorphological differences, and environmental factor statistics are extracted within each unit as model input, which enhances the spatial adaptability and input accuracy of regional prediction.

[0017] Furthermore, this invention improves the predictive stability of the model in different density ranges by employing a sampling method that combines the classification of erosion trench density levels to structurally divide the samples and extracting training and test sets from each level.

[0018] Furthermore, by employing the SHAP value method, the importance of factors can be quantitatively analyzed, thereby clarifying the positive and negative impacts of each factor on the prediction results and their degree of contribution. This invention, through the SHAP value method, not only enhances the interpretability of the model but also provides a solid theoretical basis for the selection of key factors and the delineation of priority governance areas. Attached Figure Description

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

[0020] Figure 1 This is a schematic diagram of the rapid estimation method for erosion trench density in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of an electronic device in an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives and technical solutions of this invention clearer and easier to understand, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0022] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings and specific embodiments. The described embodiments are only some embodiments of the present invention, and not all embodiments.

[0023] Example 1 like Figure 1 As shown in the figure, this embodiment discloses a rapid estimation method for gully density, aiming to solve the problems of low efficiency, high cost, and difficulty in quickly obtaining spatial distribution information of gully density over large areas in traditional gully density estimation processes. This method combines remote sensing technology, geographic information system technology, and machine learning models to achieve rapid and accurate estimation of gully density in a target area, providing a scientific basis for soil and water conservation monitoring, ecological environment protection, and land resource management.

[0024] The main steps of the rapid estimation method for gully density include: acquiring remote sensing image data of the target area; dividing the target area into different types of prediction units based on the geomorphic unit characteristics of the target area; extracting influencing factor data from each prediction unit; inputting the extracted influencing factor data of each prediction unit into a pre-trained target prediction model to obtain the corresponding predicted value of gully density; and obtaining the spatial distribution map of gully density of the target area based on the predicted values ​​of gully density of all prediction units.

[0025] The acquired remote sensing image data should have high spatial and spectral resolution to meet the needs of subsequent extraction of influencing factors and identification of erosion gully information.

[0026] Remote sensing image data can be acquired through various means, such as satellite remote sensing platforms (e.g., Landsat series satellites, Sentinel series satellites, Gaofen series satellites, etc.) and airborne remote sensing platforms. After acquiring remote sensing image data, it needs to be preprocessed. Preprocessing steps include radiometric correction and geometric correction. Radiometric correction is used to eliminate image radiometric errors caused by factors such as the sensor itself and atmospheric scattering, ensuring the radiometric accuracy of the image. Geometric correction is used to eliminate image geometric distortions caused by factors such as satellite attitude and terrain undulations, giving the image accurate geographic coordinate information for overlay analysis with other geographic data.

[0027] The target region is divided into different types of prediction units based on its geomorphic features. The specific division method is as follows: When the topographic features are characterized as low hills and plateaus, watershed boundaries are extracted using DEM (Digital Elevation Model) data and hydrological analysis tools in ArcGIS, with these boundaries serving as prediction units. The specific process is as follows: First, acquire DEM data for the target area. The resolution of the DEM data is selected based on the size and accuracy requirements of the study area; higher resolution results in more accurate watershed boundary extraction. Then, load the DEM data into ArcGIS and perform depression filling to eliminate the influence of depressions in the DEM data on the calculation of water flow direction. Next, calculate the water flow direction. Based on the depression-filled DEM data, the D8 algorithm or other water flow direction calculation methods are used to determine the water flow direction for each raster. Finally, calculate the runoff accumulation. The runoff accumulation for each raster is calculated based on the water flow direction; a larger runoff accumulation indicates a more concentrated water flow at that raster. Finally, the watershed boundary is extracted based on the cumulative runoff volume. A cumulative runoff volume threshold is set, and areas with cumulative runoff volumes exceeding this threshold are classified as watersheds, thus obtaining the watershed boundary. This watershed boundary serves as the prediction unit for low mountain and hilly areas and plateau regions. The area of ​​the watershed boundary is controlled within the range of 1-10 km². 2 .

[0028] When the topographic features are characterized as plains, a fishing net is created in ArcGIS to establish equal-area grid cells, generating rectangular boundaries as prediction units. The specific procedure is as follows: Open ArcGIS software, load the remote sensing image or vector boundary data of the plain area, and determine the size of the grid cells. The size of the grid cells should be set according to the area of ​​the study area, the distribution characteristics of gullies, and the required estimation accuracy; for example, it can be set to 1km×1km, 2km×2km, etc. Then, using the "Create Fishing Net" tool in ArcGIS, input the extent of the study area, set the width and height of the grid cells, and generate equal-area grid cells. Each generated grid cell is the prediction unit for the plain area, and its rectangular boundary is used to define the spatial extent of the prediction unit.

[0029] Furthermore, each prediction unit in this embodiment is assigned a unique coded identifier, which can be in the form of numbers, letters, or a combination thereof. This coded identifier serves as a unique spatial identifier, establishing a correspondence between the prediction result and the actual geographical location. For example, the coded identifier can include information such as the administrative division code, row number, and column number of the prediction unit. Through this coded identifier, the predicted erosion gully density value of each prediction unit can be accurately associated with its actual geographical location, ensuring the spatial traceability of the prediction results and management efficiency.

[0030] The influencing factor data includes, but is not limited to, topographic factors, rainfall factors, vegetation factors, soil factors, land use factors, lithological factors, and management factors.

[0031] The topographic factors mainly include slope, slope length, aspect, relief, and slope variability. Slope indicates the degree of inclination of the land surface and can be obtained from DEM data using the slope calculation tool in ArcGIS; slope length refers to the horizontal distance from the top to the bottom of the slope and can be calculated by combining DEM data and water flow direction; aspect indicates the orientation of the slope and can be calculated from DEM data; relief refers to the difference between the maximum and minimum elevations within a certain area; and slope variability refers to the rate of change of slope, reflecting the degree of steepness of the terrain.

[0032] Rainfall factors mainly include annual average monthly rainfall and rainfall erosivity. Annual average monthly rainfall can be obtained from meteorological station observation data in and around the target area using spatial interpolation methods such as Kriging interpolation and inverse distance weighted interpolation. Rainfall erosivity is an indicator that measures the potential for rainfall to cause soil erosion, and can be calculated using appropriate formulas based on rainfall data (such as rainfall amount and rainfall intensity).

[0033] Vegetation factors mainly include vegetation cover and normalized difference vegetation index (NDVI). Vegetation cover refers to the percentage of the vertical projection area of ​​vegetation on the ground to the total area of ​​the statistical area, which can be calculated using remote sensing image data and methods such as pixel binarization. NDVI reflects the growth status and coverage of vegetation and can be calculated using the near-infrared and red bands of remote sensing images.

[0034] Soil factors mainly include soil type, soil texture, soil organic matter content, and soil erosion resistance. Soil type and soil texture can be obtained through soil maps of the target area; soil organic matter content can be obtained through soil sampling analysis combined with spatial interpolation methods; soil erosion resistance refers to the soil's ability to resist water erosion, which can be estimated through relevant soil physicochemical properties.

[0035] Land use factors are indicators that reflect the impact of different land use types on the development of gullies. They can be used to classify land use (such as cultivated land, forest land, grassland, construction land, etc.) based on remote sensing image data, and then assign corresponding factor values ​​to different land use types.

[0036] Lithological factors mainly reflect the influence of different rock types on the development of erosion gullies. Based on the geological map of the target area, the distribution range of different lithologies can be determined, and corresponding factor values ​​can be assigned to different lithologies.

[0037] Management factors mainly include soil and water conservation measures, land use management methods, and irrigation factors.

[0038] When extracting impact factor data, it is necessary to ensure that the spatial resolution and coordinate system of each impact factor data are consistent with the prediction unit, so as to accurately assign the impact factor data to each prediction unit.

[0039] The target prediction model in this embodiment is a machine learning regression model. The machine learning regression model employs one of the following: random forest, support vector machine, convolutional neural network, or Transformer model. The following uses the random forest model as an example to explain in detail the training process of the target prediction model.

[0040] An erosion gully sample dataset was obtained by performing stratified sampling with unequal probability within the study area. Stratified sampling divides the study area into different strata based on characteristics such as landform type and land use type, and then randomly selects sample units within each stratum according to a certain proportion. For each sample unit, information such as the length and number of erosion gullies was obtained through field surveys and remote sensing image interpretation, and the erosion gully density (the ratio of the total length of erosion gullies to the area of ​​the sample unit) was calculated, thus forming the erosion gully sample dataset. For example, within each stratum, the samples were divided into a 70% training set and a 30% test set ratio to ensure that the model has representative training data in different density ranges and to maintain the comprehensiveness and fairness of the test set.

[0041] Data on various influencing factors related to gully density within the study area were collected, including the aforementioned topographic factors, rainfall factors, vegetation factors, soil factors, land use factors, lithological factors, and management factors. These data were then preprocessed (e.g., data standardization, normalization, and handling of missing values) to construct a database of influencing factors on gully density within the known area.

[0042] Based on the geomorphic unit characteristics of the study area, the study area is divided into different prediction units using the same method as the prediction unit division of the target area.

[0043] Using the erosion gully density from the erosion gully sample dataset as the response variable and the input influencing factor data as explanatory variables, a random forest target prediction model is constructed. The random forest model is an ensemble learning algorithm that constructs multiple decision trees and then votes or averages the predictions from these trees to obtain the final prediction result, exhibiting high prediction accuracy and generalization ability.

[0044] The target prediction model was trained using an erosion trench sample dataset. During training, cross-validation methods (such as 5-fold cross-validation and 10-fold cross-validation) were employed to fine-tune the hyperparameters of the model. Hyperparameters included the number of decision trees, maximum tree depth, and minimum number of split samples. Through cross-validation, the hyperparameter combination that minimized the model's prediction error was selected to improve its predictive performance.

[0045] The Shapley Additive Explanations (SHAP) method was used to screen the impact factor data. SHAP analysis can explain the contribution of each impact factor to the model's prediction results; specifically, the importance of each factor was quantified by introducing the SHAP method. Based on game theory principles, the SHAP method can provide the direction and intensity of the positive and negative contributions of factors to the prediction results at both global and local levels, improving the model's interpretability and providing theoretical support for the identification and scientific governance of dominant factors.

[0046] Finally, multiple evaluation metrics were used to quantitatively assess the predictive performance, including: Root Mean Square Error (RMSE), which measures the overall deviation between predicted and actual values; Mean Absolute Error (MAE), which reflects the average level of prediction error; and Nash-Sutcliffe Efficiency (NSE), a comprehensive evaluation of the model's goodness of fit, with values ​​closer to 1 indicating better model performance. These metrics comprehensively reflect the model's predictive ability across the entire region, providing a basis for subsequent model selection and optimization.

[0047] Based on the SHAP value analysis results from the previous stage, the main factors that have a significant impact on the model's prediction results are prioritized as the final prediction input variables to improve the efficiency and robustness of regional inference. For the defined prediction units, the statistical feature values ​​of the selected factors are extracted within the spatial range of the prediction unit. These typically include the mean or main type values, and are used as the input features of that unit. Subsequently, the factor values ​​extracted from each prediction unit are sequentially input into the trained machine learning regression model to generate the corresponding predicted erosion trench density values.

[0048] The filtered influencing factor data extracted from each prediction unit are input into a pre-trained optimal target prediction model to obtain the corresponding predicted gully density values. Then, based on the coded identifier of each prediction unit, the predicted gully density values ​​are associated with their spatial locations. Using geographic information system software such as ArcGIS, spatial interpolation or direct plotting is employed to ultimately generate a spatial distribution map of gully density continuously covering the entire study area. Furthermore, a vector layer of gully density is obtained from this spatial distribution map. The results can be used for spatial visualization, density level classification, identification of priority remediation areas, and to provide quantitative support for regional-scale soil erosion control and resource allocation.

[0049] To fully demonstrate the effectiveness of the rapid estimation method for erosion gully density in Example 1, the following is a detailed explanation of the rapid estimation method for erosion gully density, using the Songnen Black Soil region of Northeast China as the study area.

[0050] Using the Songnen Black Soil Region of Northeast China as the study area, a stratified unequal probability systematic sampling method was employed to delineate small watershed units. Units with less than 5% cultivated land area were excluded, resulting in 853 valid cultivated land sample units. A cultivated land gully density sample dataset was obtained based on sub-meter level imagery and manual visual interpretation. Fifty-five typical small watersheds were selected as validation units for field surveys and UAV aerial photography to evaluate the accuracy of the interpreted gully density data.

[0051] Machine learning models were constructed based on 37 relevant factors across six categories: topography, precipitation, vegetation cover, soil, lithology, crops, and management. Topography primarily considered slope and slope length; precipitation considered 14 factors, including multi-year average precipitation, multi-year average storm rainfall, and monthly multi-year average monthly precipitation; vegetation cover considered 13 factors, including multi-year average vegetation cover and monthly multi-year average vegetation cover; soil considered 5 factors, including silt, sand, clay content, soil organic matter content, and soil type; lithology primarily considered lithology category factors; and crops and management primarily considered crop type and irrigation factors.

[0052] In areas dominated by terraces and low hills, small watershed units are established; in areas dominated by plains, units are divided into areas of 2 km². 2 The study area was ultimately divided into 85,516 small watersheds and grid units, with an average area of ​​approximately 2.5 km². The grid units were 2 km² × 1 km². 2 .

[0053] The 853 sampled units were divided into four groups based on their groove density values: 0, (0, 0.5], (0.5, 3], and (3, +∞]. Within each group, 30% of the sample units were randomly selected as the test dataset, and 70% as the training dataset, resulting in 595 training samples and 258 test samples. The final machine learning algorithm used was Random Forest, with the optimal parameter combination as follows: maximum tree depth (max_depth) = 7, minimum number of leaf node samples (min_samples_leaf) = 6, minimum number of samples within a split (min_samples_split) = 12, and number of trees (n_estimators) = 303. Under this parameter configuration, the model performed best on the test set, with NSE, MAE, and RMSE of 0.66, 0.25, and 0.49 for the training set, and 0.60, 0.30, and 0.58 for the test set.

[0054] By using the unique coded identifier (ID) of the prediction unit, the model prediction results are accurately associated with their corresponding spatial locations, thereby realizing the spatial expression of the predicted value of ditch density in cultivated land, and finally generating a spatial distribution map of ditch density in cultivated land in the Songnen Black Soil Region of Northeast China.

[0055] Example 2 This embodiment discloses a rapid erosion trench density estimation system for implementing the rapid erosion trench density estimation method described in Embodiment 1, including: The data acquisition module acquires remote sensing image data of the target area; The prediction unit segmentation module divides the target area into different types of prediction units based on the geomorphic unit characteristics of the target area. The impact factor data extraction module extracts impact factor data from each prediction unit. The erosion gully density prediction module and the erosion gully density prediction estimation module input the influencing factor data extracted from each prediction unit into the pre-trained target prediction model to obtain the corresponding erosion gully density prediction value. The spatial distribution map generation module generates a spatial distribution map of erosion gully density in the target area based on the predicted erosion gully density values ​​of all predicted units.

[0056] Example 3 In another embodiment of the present invention, a computer-readable storage medium is provided as a storage component within a terminal device, the function of which is to store programs and data. It should be noted that the computer-readable storage medium here encompasses not only the built-in storage components of the terminal device but also extended storage components supported by the device. Essentially, it is a tangible medium capable of containing or storing programs that can be invoked by or in conjunction with an instruction execution system, device, or apparatus. This storage medium provides storage areas for the terminal's operating system and stores one or more instructions suitable for processor loading and execution, which can constitute one or more computer programs containing program code.

[0057] Specifically, examples of computer-readable storage media (a non-exclusive list) include: electrical connections with one or more wires, portable disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, optical fibers, portable optical disc read-only memory, optical storage devices, magnetic storage devices, or any reasonable combination of the above types.

[0058] The storage medium may also include data signals propagated as part of a baseband portion or a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any reasonable combination of both. Furthermore, computer-readable storage medium may also refer to other readable media besides conventional readable storage media, capable of sending, propagating, or transmitting programs for use or operation by an instruction execution system, apparatus, or device. Program code on the storage medium can be transmitted via any suitable medium, including but not limited to wireless, wired, optical fiber, or any reasonable combination thereof.

[0059] The program code used to implement the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C. The execution modes of the program code include: running entirely on the user's computing device, running partially on the user's device as a standalone software package, running partially in a distributed manner on both the user's device and a remote computing device, or running entirely on a remote computing device or server. When a remote computing device is involved, the device can be connected to the user's computing device via any type of network such as a local area network (LAN) or a wide area network (WAN), or connected to an external computing device via the Internet through an Internet service provider.

[0060] The processor is capable of loading and executing one or more instructions stored in a computer-readable storage medium to implement the corresponding steps of the rapid estimation method for erosion trench density described in Example 1.

[0061] Example 4 Figure 2 This is a schematic diagram of a computer device provided according to an embodiment of the present invention.

[0062] Please see Figure 2 The terminal device is a computer device. In this embodiment, the computer device 60 includes a processor 61, a memory 62, and a computer program 63 stored in the memory 62 and executable on the processor 61. When executed by the processor 61, the computer program 63 implements the rapid estimation method for erosion trench density in this embodiment; to avoid repetition, these details are not elaborated here. Alternatively, when executed by the processor 61, the computer program 63 implements the functions of each model / unit in the computational system constituting the rapid estimation method for erosion trench density in this embodiment; to avoid repetition, these details are not elaborated here.

[0063] Computer device 60 can be a desktop computer, laptop, handheld computer, cloud server, or other computing device. Computer device 60 may include, but is not limited to, a processor 61 and a memory 62. Those skilled in the art will understand that... Figure 2 This is merely an example of computer device 60 and does not constitute a limitation on computer device 60. It may include more or fewer components than shown, or combine certain components, or different components. For example, computer device may also include input / output devices, network access devices, buses, etc.

[0064] The processor 61 may be a central processing unit (CPU), or other general-purpose processors, CPUs, graphics processing units (GPUs), 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, quantum computing-based data processing logic units, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0065] The memory 62 can be an internal storage unit of the computer device 60, such as a hard disk or RAM of the computer device 60. The memory 62 can also be an external storage device of the computer device 60, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., provided on the computer device 60.

[0066] Furthermore, the memory 62 may include both internal storage units of the computer device 60 and external storage devices. The memory 62 is used to store computer programs and other programs and data required by the computer device. The memory 62 can also be used to temporarily store data that has been output or will be output.

[0067] Any references to memory, databases, or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may 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), etc.

[0068] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

Claims

1. A method for rapid estimation of erosion trench density, characterized in that, include: Acquire remote sensing image data of the target area; The target area is divided into different types of prediction units based on the geomorphic unit characteristics of the target area; Impact factor data are extracted from each prediction unit; The influencing factor data extracted from each prediction unit is input into the pre-trained target prediction model to obtain the corresponding predicted value of erosion gully density; Based on the predicted erosion gully density values ​​of all predicted units, a spatial distribution map of erosion gully density in the target area is obtained.

2. The method for rapid estimation of erosion trench density according to claim 1, characterized in that, The target area is divided into different types of prediction units based on its geomorphic features, including: When the topographic features are low mountain and hilly areas and plateau areas, the watershed boundary is extracted using the hydrological analysis tools in ArcGIS based on the DEM data, and the watershed boundary is used as the prediction unit. When the topographic features are characterized as plains, create a fishing net in ArcGIS to establish equal-area grid cells and generate rectangular boundaries as prediction cells.

3. The method for rapid estimation of erosion trench density according to claim 2, characterized in that, The prediction unit is equipped with a unique coded identifier; the coded identifier serves as a unique identification code for spatial location and is used to establish the correspondence between the prediction result and the actual geographical location.

4. The method for rapid estimation of erosion trench density according to claim 1, characterized in that, The influencing factor data includes at least one of the following: topographic factors, rainfall factors, vegetation factors, soil factors, land use factors, lithological factors, and management factors.

5. The method for rapid estimation of erosion trench density according to claim 1, characterized in that, The target prediction model employs one of the following: random forest, support vector machine, convolutional neural network, or Transformer model.

6. The method for rapid estimation of erosion trench density according to claim 5, characterized in that, The training process of the target prediction model includes: A dataset of erosion gully samples was obtained by stratified sampling within the study area; Construct a database of factors influencing gully density within a known region; Based on the geomorphic unit characteristics of the study area, the study area is divided into different prediction units; Using the erosion gully density in the erosion gully sample dataset as the response variable and the data of each input influencing factor as the explanatory variable, a target prediction model is constructed. The target prediction model was trained using an erosion gully sample dataset, and the hyperparameters in the target prediction model were updated using cross-validation. The SHAP value analysis method is used to screen the impact factor data. The screened impact factor data is used as the input variable of the prediction unit. The target prediction model is updated based on the screened impact factor data to obtain the optimal target prediction model.

7. The method for rapid estimation of erosion trench density according to claim 6, characterized in that, The SHAP value analysis method includes: using game theory principles to calculate the contribution direction and intensity of global influence factor data to the prediction results, as well as the contribution direction and intensity of local influence factor data to the prediction results.

8. A rapid erosion trench density estimation system, used to implement the rapid erosion trench density estimation method according to any one of claims 1 to 7, characterized in that, include: The data acquisition module acquires remote sensing image data of the target area; The prediction unit segmentation module divides the target area into different types of prediction units based on the geomorphic unit characteristics of the target area. The impact factor data extraction module extracts impact factor data from each prediction unit. The erosion gully density prediction module and the erosion gully density prediction estimation module input the influencing factor data extracted from each prediction unit into the pre-trained target prediction model to obtain the corresponding erosion gully density prediction value. The spatial distribution map generation module generates a spatial distribution map of erosion gully density in the target area based on the predicted erosion gully density values ​​of all predicted units.

9. A computer-readable storage medium for storing one or more programs, characterized in that, The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform the rapid estimation method for erosion trench density as described in any one of claims 1 to 7.

10. A computing device, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including steps for performing the rapid estimation method for erosion trench density of any one of claims 1 to 7.

Images