Soil erosion monitoring method and system based on multi-source remote sensing and terrain factors

Abstract

The application discloses a soil and water loss monitoring method and system based on multi-source remote sensing and terrain factors, and particularly relates to the technical field of soil and water loss remote sensing monitoring, and is used for solving the problem that in the existing soil and water loss monitoring, it is difficult to fuse digital elevation data, multi-source remote sensing images and hydrodynamic factors in a unified time axis and spatial structure under the scale of slope unit and small watershed. By dividing the small watershed and the slope unit based on the digital elevation data, the terrain normalized reflectance feature, the terrain attribute and the hydrodynamic factor are combined into the joint feature on the scale of the slope unit, and the joint feature difference value of the historical reference period, the continuous abnormal link screening along the flow direction and the region growing aggregation erosion patch constrained by the terrain and the channel are combined, so that the soil and water loss erosion unit and the erosion patch are stably and continuously identified under the condition of the complex terrain and the multi-source remote sensing, and the effect of improving the response consistency of the monitoring result to the actual erosion process is achieved.

Description

Technical Field

[0001] This invention relates to the field of remote sensing monitoring technology for soil erosion, specifically to a method and system for monitoring soil erosion based on multi-source remote sensing and topographic factors. Background Technology

[0002] In the field of soil and water conservation monitoring, existing technologies rely on methods such as sample plot surveys and fixed monitoring stations to obtain erosion information. While these methods can provide relatively reliable observation results in local areas, they are costly to deploy, have limited coverage, and have long update cycles, making it difficult to continuously reflect the erosion evolution process at the slope and small watershed scales over large-scale, long-term time series. On the other hand, with the development of remote sensing and geographic information technology, soil and water conservation monitoring methods based on single satellite images or a small number of temporal images have been applied. These methods typically use regular grids or administrative units as basic spatial units and identify erosion-sensitive areas through indicators such as vegetation indices, bare land indices, and empirical models. However, these methods often use pixels or coarse-scale partitions as analysis units, focusing more on depicting "prone areas" in terms of planar distribution. They lack the ability to finely characterize erosion patches that develop along topographic confluence paths and are insufficient to support the refined management needs at the slope and small watershed scales.

[0003] In existing technologies, some schemes attempt to overlay digital elevation models, slope, aspect, and other topographic factors with multi-source remote sensing images to correct radiance or improve classification accuracy. However, these methods typically remain at the level of auxiliary static factors: First, the differences between multi-source remote sensing images in terms of spatial reference frame, imaging geometry, and observation time are often only addressed through simple resampling or stitching, without fully considering the systematic impact of topographic relief, aspect, and solar altitude angle on observed radiance. This results in non-negligible radiance biases at the slope scale for images from different sources and at different times. Second, in terms of spatial unit division, most methods still use regular grids or administrative units as analysis units, using only slope and aspect as pixel attributes. They lack spatial structure expressions that take slope units as basic units and consider the continuity of confluence direction, making it difficult to organically link topographic morphology, confluence path, and erosion process. Third, in the temporal dimension, existing monitoring methods mostly use a small number of temporal images for difference or threshold discrimination, lacking a mechanism to construct a slope baseline state based on long-term series and identify abnormal changes on this basis. Parameter thresholds are also mostly set empirically, resulting in weak version management and traceability.

[0004] In summary, existing soil erosion monitoring technologies combining multi-source remote sensing and topographic factors lack the ability to construct traceable joint features on a unified time axis based on slope units and small watersheds, while simultaneously integrating topographic factors, image spectral information, and hydrodynamic conditions. Furthermore, these technologies are insufficient to stably identify continuously eroding slope units and their aggregated erosion patches along the confluence direction, thereby forming a structured monitoring indicator system at the slope and small watershed scales. How to stably and continuously characterize soil erosion patches developing along confluence paths at the slope unit and small watershed scales under complex terrain and multi-source remote sensing conditions, and to form a time-series erosion indicator with unified caliber and traceable versions, remains a technical problem to be solved in this field. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for monitoring soil erosion based on multi-source remote sensing and topographic factors, in order to solve the problems mentioned in the background.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for monitoring soil erosion based on multi-source remote sensing and topographic factors, comprising:

[0007] S1. Obtain regional digital elevation data, divide the watershed and slope units according to the confluence direction, slope, aspect and curvature, and register the watershed and upstream and downstream slope units corresponding to the slope units;

[0008] S2. Based on the slope unit boundary, resample the multi-source remote sensing images to the elevation data raster, correct the radiance according to the raster slope, aspect and imaging geometry, and statistically analyze the terrain normalized reflectance characteristics according to the slope unit.

[0009] S3. Using slope units as indexes, the terrain normalized reflectivity features, slope aspect and hydrodynamic factors are combined into joint features. Neighborhoods are constructed according to terrain similarity and the joint features are smoothed.

[0010] S4. Using the joint characteristics of the historical baseline period as the baseline, calculate the difference between the joint characteristics of the target period and the baseline, and select adjacent slope units whose difference reaches the threshold along the confluence direction as candidate erosion slope units.

[0011] S5. Using candidate erosion slope units as seeds, regional growth is carried out in the small watershed along the confluence direction according to the constraints of slope, aspect and elevation. Growth is terminated when the growth area reaches the watershed boundary and the location of the abrupt change in aspect. Erosion patches are delineated in combination with the gully network.

[0012] S6. Calculate the changes in erosion patch area, slope, and topographic normalized reflectance according to the small watershed number, and register the calculation results as time series data of slope unit erosion index and small watershed erosion index.

[0013] Furthermore, S1 includes:

[0014] Digital elevation data is constructed based on digital elevation model raster, and the version number is registered in the terrain data version table;

[0015] Perform interpolation and smoothing processing on digital elevation data, and register the interpolation window size, height difference threshold, and smoothing parameters in the parameter configuration table;

[0016] The confluence direction is derived based on the elevation difference between adjacent grid cells. The slope, aspect and curvature are calculated. Small watersheds and slope units are divided according to the classification rules set in the parameter configuration table. Each slope unit is assigned a slope unit number and its corresponding small watershed number. Upstream and downstream slope units are identified. The above numbers and the topographic data version number are registered in the topographic unit index table.

[0017] Furthermore, S2 includes:

[0018] Register the imaging time and imaging geometric parameters for multi-source remote sensing images in the image metadata table;

[0019] The monitoring platform performs cloud volume discrimination based on the cloud volume ratio threshold in the parameter configuration table to remove images with excessive cloud volume;

[0020] Using the slope unit boundaries in the topographic unit index table as constraints, multi-source remote sensing images are resampled to digital elevation data raster units using an interpolation method that preserves the total radiation. Based on the slope, aspect, imaging time, and imaging geometry parameters of the raster units, topographic radiometric correction is performed to obtain the raster topographic normalized reflectance. Topographic normalized reflectance characteristics are statistically analyzed on a slope unit basis, and the slope unit number and imaging time label are registered in the slope spectral feature table.

[0021] Furthermore, S3 includes:

[0022] Using the slope unit as the index, at the monitoring time step, the terrain normalized reflectance is read from the slope spectral feature table, the small watershed number, slope grade, slope aspect azimuth segment and curvature interval are read from the terrain unit index table, and the hydrodynamic factors corresponding to the monitoring time step are read from the hydrodynamic service. The above features are combined into a joint feature vector.

[0023] Based on the preset slope difference threshold, slope aspect segment and curvature difference interval, a terrain similarity neighborhood is constructed. Within the terrain similarity neighborhood, a representative statistical range is determined according to the distribution of joint feature vectors. The weight of joint feature components that deviate from the representative statistical range is reduced to obtain smooth joint features. The smooth joint features, along with the slope unit number, time stamp, terrain similarity neighborhood identification number and rule version number, are registered in the joint feature table.

[0024] Furthermore, S4 includes:

[0025] Based on historical rainfall records, remote sensing time series, and water conservation management ledgers, the monitoring platform selects continuous monitoring periods for each slope unit that do not exceed the heavy rainfall threshold and have not been registered for construction disturbances as historical baseline periods.

[0026] Within the historical baseline period, joint features are read from the joint feature table according to the monitoring time step, representative values ​​and fluctuation ranges of each dimension are calculated, and the representative values ​​and fluctuation ranges are registered as the baseline of the slope unit in the baseline table, and associated with the slope unit number, the start and end time of the baseline period, the amount of data involved in the statistics and the rule version number.

[0027] Furthermore, the monitoring platform obtains the joint features of each slope unit from the joint feature table within the observation window according to the time step based on the observation window length, the number of consecutive judgments, and the difference threshold. It then compares these features with the baseline representative value to obtain the difference sequence. When the number of consecutive judgments that the difference exceeds the difference threshold reaches the number of consecutive judgments, the slope unit is marked as a suspected erosion unit.

[0028] Based on the confluence direction relationship recorded in the topographic unit index table, the monitoring platform searches for upstream and downstream slope units adjacent to the suspected erosion unit along the confluence direction. When the upstream slope unit, the suspected erosion unit, and the downstream slope unit are all marked and the abnormal time steps overlap, the continuous slope units along the confluence direction are marked as candidate erosion slope unit links.

[0029] Furthermore, S5 includes:

[0030] During the patch aggregation stage, the monitoring platform sets the candidate erosion slope units identified in the change identification stage as growth seeds. Within the small watershed to which the growth seed belongs, the watershed boundary is defined according to the topographic unit index table. Along the confluence direction, the growth link is extended according to the slope grade set, slope aspect azimuth segment, and elevation decrease constraint.

[0031] Under the condition that the joint feature table and hydrodynamic record are complete, the downstream slope unit is incorporated into the growth region;

[0032] The expansion terminates when the growth link reaches the watershed boundary.

[0033] The expansion terminates when the difference between the slope aspect azimuth segment of the downstream slope unit and the slope aspect azimuth segment of the current slope unit exceeds the azimuth difference threshold.

[0034] By combining the gully network, slope units are aggregated and merged to form erosion patches.

[0035] Furthermore, S6 includes:

[0036] The monitoring platform determines the area of ​​erosion patches and the average slope based on the coverage and slope of the slope unit, and determines the change in topographic normalized reflectance based on the joint feature difference of the target period and the topographic normalized reflectance of the baseline period.

[0037] Slope unit erosion index is formed by summarizing the area of ​​erosion patches, average slope, and changes in topographic normalized reflectance according to the slope unit number.

[0038] Small watershed erosion indices are formed by summarizing the area of ​​erosion patches, average slope, and changes in topographic normalized reflectance according to the small watershed number.

[0039] The erosion indices of slope units and small watersheds are written into the erosion index time series table in the order of monitoring time stamps, and the small watershed number, slope unit number, monitoring time stamp and rule version number are used as monitoring idempotent keys.

[0040] On the other hand, the present invention provides a soil and water loss monitoring system based on multi-source remote sensing and topographic factors, including:

[0041] The terrain unit division module is used to acquire regional digital elevation data, divide the raster units into small watersheds and slope units based on the confluence direction, slope, aspect and curvature of the raster units, and register the small watersheds, upstream slope units and downstream slope units corresponding to each slope unit.

[0042] The image radiometric feature construction module is used to resample multi-source remote sensing images to elevation data raster units according to the slope unit boundary, perform topographic radiometric correction on the raster radiometric amount according to the raster slope, aspect and imaging geometry parameters, and statistically analyze the topographic normalized reflectance features according to the slope unit.

[0043] The joint feature construction module is used to combine the terrain normalized reflectivity features, slope aspect and hydrodynamic factors into joint features using slope units as indexes, construct terrain similarity neighborhoods based on terrain similarity, and perform smoothing processing on the joint features within the terrain similarity neighborhoods.

[0044] The change identification module is used to calculate the difference between the joint features of the target time period and the baseline using the joint features of the historical reference period as the baseline, and select adjacent slope units whose difference reaches the threshold along the confluence direction and mark them as candidate erosion slope units.

[0045] The erosion patch delineation module is used to perform regional growth along the confluence direction within a small watershed based on the candidate erosion slope units as seeds, according to the constraints of slope, aspect and elevation. Growth is terminated when the growth area reaches the watershed boundary and the location of the abrupt change in aspect, and erosion patches are delineated in combination with the gully network.

[0046] The erosion index generation module is used to calculate the area, slope, and topographic normalized reflectance changes of erosion patches according to the small watershed number, and to register the calculation results as slope unit erosion index and small watershed erosion index, and to form time series data.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] 1. By dividing small watersheds and slope units based on digital elevation data, the topographic normalized reflectance characteristics, topographic attributes and hydrodynamic factors are combined into joint features at the slope unit scale. Combined with the joint feature difference of the historical reference period, the screening of continuous anomaly links along the confluence direction and the regional growth aggregation of erosion patches constrained by topography and gullies, the system can stably and coherently identify soil erosion units and erosion patches under complex terrain and multi-source remote sensing conditions, and improve the consistency of monitoring results with the actual erosion process.

[0049] 2. By generating structured erosion indicators at three levels—erosion patches, slope units, and small watersheds—and constructing monitoring idempotent keys using small watershed numbers, slope unit numbers, monitoring time stamps, and rule version numbers, and combining these with the correlation records between threshold rules and baseline versions, a traceable and comparable time series indicator system for soil and water loss monitoring is formed, facilitating subsequent supervision, assessment, and decision-making applications. Attached Figure Description

[0050] Figure 1 This is a flowchart illustrating the method for monitoring soil erosion based on multi-source remote sensing and topographic factors according to the present invention.

[0051] Figure 2 This is a schematic diagram of the soil and water loss monitoring system based on multi-source remote sensing and topographic factors according to the present invention. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] Example 1: Figure 1 A flowchart illustrating the soil erosion monitoring method based on multi-source remote sensing and topographic factors of this invention is provided. The method includes:

[0054] S1. Obtain regional digital elevation data, divide the watershed and slope units according to the confluence direction, slope, aspect and curvature, and register the watershed and upstream and downstream slope units corresponding to the slope units;

[0055] S2. Based on the slope unit boundary, resample the multi-source remote sensing images to the elevation data raster, correct the radiance according to the raster slope, aspect and imaging geometry, and statistically analyze the terrain normalized reflectance characteristics according to the slope unit.

[0056] S3. Using slope units as indexes, the terrain normalized reflectivity features, slope aspect and hydrodynamic factors are combined into joint features. Neighborhoods are constructed according to terrain similarity and the joint features are smoothed.

[0057] S4. Using the joint characteristics of the historical baseline period as the baseline, calculate the difference between the joint characteristics of the target period and the baseline, and select adjacent slope units whose difference reaches the threshold along the confluence direction as candidate erosion slope units.

[0058] S5. Using candidate erosion slope units as seeds, regional growth is carried out in the small watershed along the confluence direction according to the constraints of slope, aspect and elevation. Growth is terminated when the growth area reaches the watershed boundary and the location of the abrupt change in aspect. Erosion patches are delineated in combination with the gully network.

[0059] S6. Calculate the changes in erosion patch area, slope, and topographic normalized reflectance according to the small watershed number, and register the calculation results as time series data of slope unit erosion index and small watershed erosion index.

[0060] The technical connections and implementation logic of the six steps are as follows:

[0061] S1 acquires regional digital elevation data, divides small watersheds and slope units based on confluence direction, slope, aspect, and curvature, and registers upstream and downstream relationships, establishing a unified topographic structure framework and flow topology for all subsequent slope unit-based calculations. S2, under the boundary constraints of the slope units defined in S1, resamples multi-source remote sensing images to elevation data raster, corrects radiance according to raster slope, aspect, and imaging geometry parameters, and then statistically analyzes the topographic normalized reflectance characteristics at the slope unit scale, ensuring a one-to-one correspondence between remote sensing information and topographic units. S3 uses slope units as indices, combines the topographic normalized reflectance characteristics obtained in S2 with the slope, aspect, and hydrodynamic factors from S1 to form joint features, constructs neighborhoods based on topographic similarity, and smooths the joint features to suppress noise fluctuations while maintaining spatial structure constraints. S4 uses the joint features from the historical baseline period as a baseline to calculate... The difference between the target time period joint features and the baseline is used to screen adjacent slope units whose difference reaches a threshold along the confluence direction, and these units are marked as candidate erosion slope units, realizing integrated change identification based on time series and water flow path; S5 uses the candidate erosion slope units obtained in S4 as seeds to perform regional growth along the confluence direction within the small watershed according to slope, aspect and elevation constraints. The expansion is terminated when the growth area reaches the watershed boundary and the location of the abrupt change in aspect. In combination with the gully network, erosion patches are delineated, and discrete slope units are organized into spatial patches with clear boundaries; S6 further statistically analyzes the changes in area, slope and topographic normalized reflectance of erosion patches according to the small watershed number, generating slope unit erosion index and small watershed erosion index respectively, and forming time series data, thereby continuously quantifying the soil and water loss process at the slope unit scale and the small watershed scale.

[0062] S1. Obtain regional digital elevation data, divide the area into small watersheds and slope units based on confluence direction, slope, aspect, and curvature, and register the corresponding small watersheds and upstream and downstream slope units for each slope unit. The specific implementation is as follows:

[0063] During the deployment phase of the monitoring platform, digital elevation data covering the target monitoring area is preferably obtained first from legally obtained surveying and mapping results. The digital elevation data is a digital elevation model grid that records the surface elevation values ​​in the form of a regular grid. The elevation unit is meters. A unified plane projection coordinate system and a unified elevation datum are used. The grid resolution represents the horizontal distance between the centers of adjacent grid cells and can be set to a fixed value between five and thirty meters. In the preferred embodiment, it is set to about ten meters. Each set of digital elevation data is registered in the terrain data version table with a unique version number, acquisition date, data source, surveying accuracy level, and coordinate system description for subsequent traceability and comparison.

[0064] The monitoring platform preferably performs coordinate alignment and boundary rectification on the digital elevation data, aligning data from different sources to a unified coordinate frame. A buffer zone of at least one grid width is added to the outer edge of the administrative boundary. In a preferred embodiment, the buffer zone width can be set to one to three grid widths to ensure complete coverage of the catchment area when deducing the subsequent confluence direction. For grid cells with missing measurements, the elevation values ​​of adjacent grid cells within a preset window are used for interpolation. For grid cells whose elevation difference with surrounding grid cells within a preset neighborhood window exceeds a preset height difference threshold, they are identified as isolated elevation peaks and their elevation is weakened using a neighborhood smoothing method. The interpolation window size, height difference threshold, and smoothing parameters are registered in the parameter configuration table and correspond to the terrain data version number, so that the digital elevation data remains spatially continuous and smooth.

[0065] After completing the above quality control, the monitoring platform derives the maximum slope direction of each grid cell based on the elevation difference between adjacent grid cells, defining this direction as the confluence direction of that grid cell; it calculates the slope of each grid cell based on the elevation change rate, defining the slope as the local tilt angle of the ground surface relative to the horizontal plane; it calculates the slope aspect based on the azimuth angle of the maximum slope direction relative to true north, defining the slope aspect as the plane pointing direction of the maximum slope direction; and it calculates the curvature based on the elevation relationship between the grid cell and its surrounding grid cells, defining curvature as an index reflecting the degree of unevenness of the ground surface. The monitoring platform divides the slope according to fixed levels pre-set in the parameter configuration table, and the number of slope levels and boundary angles remain unchanged within the same monitoring version; it divides the slope aspect according to pre-set fixed azimuth segments, each azimuth segment corresponding to a defined angle range, and the azimuth segment division rules remain unchanged within the same monitoring version, and are registered in the parameter configuration table.

[0066] Subsequently, the monitoring platform automatically identifies natural watersheds based on the confluence direction relationship of each grid cell. The set of grid cells in which the confluence paths ultimately point to the same outlet position is defined as a small watershed. The spatial boundary of the small watershed is determined by a closed loop formed by ridgelines that no longer converge outward along the confluence direction. Each small watershed is assigned a unique small watershed number and recorded in the small watershed information table. Within each small watershed, the monitoring platform merges grid cells based on slope grade, slope aspect, curvature range, and confluence direction continuity. A group of adjacent grid cells with the same slope grade, slope difference not exceeding a preset slope tolerance, slope aspect in the same direction, curvature value falling within a preset curvature range, and forming a continuous path along the confluence direction are grouped into a slope unit. A slope unit is defined as a basic topographic unit with relatively consistent slope, slope aspect, and curvature in space, where water flows converge along a single dominant direction. Specific values ​​for slope tolerance and curvature range can be set according to the regional topographic type. In a preferred embodiment, the slope tolerance can be set to not exceed a preset slope tolerance threshold, which is registered in the parameter configuration table. The curvature range can be limited to a given value range based on historical statistical results. These parameters are also registered in the parameter configuration table and associated with the version number.

[0067] For each slope unit obtained from the division, the monitoring platform assigns a unique slope unit number, determines the watershed number to which the slope unit belongs, and identifies the upstream and downstream slope unit sets directly connected to the slope unit in the watershed direction based on the runoff path of the grid units within the slope unit. The upstream slope unit set consists of adjacent slope units that transport runoff to the slope unit along the watershed direction, and the downstream slope unit set consists of adjacent slope units that receive runoff from the slope unit along the watershed direction.

[0068] The sub-basin number, slope unit number, upstream slope unit set, downstream slope unit set, slope grade, slope aspect segment, curvature interval, and terrain data version number are registered in the terrain unit index table in a structured field format. The terrain unit index table can be implemented using a relational database table, columnar storage table, or key-value storage method, as long as it can support querying the sub-basin and upstream and downstream slope units to which the slope unit belongs by the slope unit number, and ensure the uniqueness of the number within the same version. In a preferred embodiment, the monitoring platform can divide the task into batches according to sub-basins to complete the above-mentioned elevation processing, confluence direction derivation, and slope unit division in parallel, so as to complete the terrain structure modeling of the entire area within a limited time window. For local areas with severe elevation missing or quality that does not meet the set conditions, the corresponding range can be marked as an area with incomplete terrain information in the terrain unit index table. The results of this area will not be used in subsequent analysis steps that rely on slope units, while the mark will be retained for manual verification.

[0069] The acquisition and use of digital elevation data comply with the relevant regulations on the management of surveying and mapping results and information security. When accessing and storing terrain data, the monitoring platform preferably limits the access scope through access control. In terms of hardware and software implementation, centralized servers, cloud computing platforms, or devices with equivalent computing and storage capabilities can be used. As long as the ability to divide small watersheds and slope units based on digital elevation data and form a terrain unit index table with version information is achieved, it is considered an equivalent implementation of this step.

[0070] S2. Based on the slope unit boundaries, resample the multi-source remote sensing images to elevation data raster, correct the radiance according to the raster slope, aspect, and imaging geometry, and statistically analyze the topographic normalized reflectance characteristics according to the slope unit. The specific implementation is as follows:

[0071] On the image service side, preferably, multi-source remote sensing images covering the target area are acquired from legally obtained remote sensing data sources according to the monitoring year. Multi-source remote sensing images are surface reflection information from different satellite platforms, airborne platforms, or other imaging platforms, with different band combinations, spatial resolutions, and imaging times, but covering the same geographical area. When acquiring each image, metadata such as imaging time, solar altitude angle, solar azimuth angle, observation angle, platform type, and original spatial resolution are registered, and a unique version number is assigned to the image in the image metadata table. The platform type, time precision, and geometric precision caliber recorded in the image metadata table are consistent with the aforementioned topographic data version number.

[0072] The monitoring platform preferably first unifies the coordinates of the multi-source remote sensing images, converting the spatial reference system of the images into a plane projection coordinate system and elevation datum consistent with the digital elevation data. Based on this, cloud cover is determined for each image. Cloud cover determination can be based on brightness thresholds, spectral thresholds, or cloud masks provided by the supplier to determine the cloud coverage area, and the ratio of cloud coverage area to effective imaging area is calculated. Images with cloud cover ratios exceeding a preset cloud cover ratio threshold are marked as images with excessive cloud cover. The preset cloud cover ratio threshold is registered in the parameter configuration table. In a preferred embodiment, it can be set to no more than a certain percentage of the total area. Images with excessive cloud cover do not participate in the current round of monitoring calculations. For images with cloud cover that do not exceed the limit but have local cloud shadows, multi-source remote sensing images of the same area in adjacent time periods can be compared in time and space. The cloud shadow area is replaced with the reflection information of the corresponding position in the image of the adjacent time period. The time interval range of the replacement operation, the acceptable imaging time difference, and the replacement priority are registered in the parameter configuration table to control the temporal consistency of the supplementary coverage.

[0073] After cloud cover control and coverage enhancement are completed, the monitoring platform resamples the multi-source remote sensing images of each scene onto the corresponding raster units of the digital elevation data within a specified time window. The specified time window can be set to a time range that does not exceed a preset time threshold, which is recorded in the parameter configuration table. In a preferred embodiment, it can be a time range from several hours to one day after the arrival of the image. During the resampling process, the slope unit boundary in the aforementioned terrain unit index table is used as a constraint to map the image pixels to the raster units under the elevation raster coordinate system. Preferably, an interpolation method that preserves the total radiation is adopted to distribute the radiation value of the original pixel to multiple raster units according to the neighborhood weight, so that the radiation distribution at the slope unit boundary is continuous and the overall radiation energy of the slope unit remains conserved.

[0074] To eliminate the influence of slope, aspect, and imaging geometry on the observed radiance, the monitoring platform uses the slope, aspect, and solar altitude angle, solar azimuth angle, and observation angle parameters registered in the image metadata table corresponding to each grid cell to perform topographic radiance correction on the grid radiance. The observed values ​​affected by slope orientation, shadow, and observation angle are converted into topographic normalized reflectance under a unified reference plane. In this embodiment, topographic normalized reflectance is defined as the dimensionless reflectance value describing the spectral characteristics of surface materials after eliminating the effects of slope and aspect geometry. The combination of geometric parameters used for correction and the method of taking the values ​​of correction coefficients are registered in the parameter configuration table and associated with the image version number.

[0075] For each slope unit, the monitoring platform aggregates the topographic normalized reflectance of all grid units within that slope unit at the same imaging time. Preferably, the median or a representative value within a preset quantile interval is used as the topographic normalized reflectance feature of the slope unit at that time across each band, to reduce the impact of local noise on the overall slope characteristics. The upper and lower bounds of the quantile interval can be set based on historical sample statistics; in a preferred embodiment, typical values ​​within the interquartile range can be selected. The aforementioned topographic normalized reflectance feature, along with the slope unit number, imaging time stamp, image source platform type, and image version number, is recorded in the slope spectral feature table. The slope spectral feature table stores multiple time-series entries in a structured field format, ensuring that subsequent steps can quickly obtain the topographic normalized reflectance features of the slope unit across each band based on the slope unit number and time stamp.

[0076] The entire image processing chain is preferably divided into task batches and run in parallel according to small watersheds. The monitoring platform can simultaneously perform image resampling and topographic radiometric correction for different small watersheds when resources permit, and set a maximum allowable processing delay for each image. The maximum allowable processing delay can be set to not exceed a preset processing delay threshold. In a preferred embodiment, the time range corresponding to the preset processing delay threshold can be set to several hours. When an image fails to complete resampling and topographic radiometric correction within the time threshold, its status is marked as pending recalculation in the scheduling record, and the image is added to the recalculation queue for recalculation when computing resources are available. After the recalculation result is generated, the original mark is overwritten, while the processing log and version number are retained for subsequent quality verification.

[0077] Through the above process, multi-source remote sensing images are unified in terms of spatial reference system and radiation aperture to a topographic normalized reflectance feature space with slope units as the basic unit, providing a consistent and traceable image feature basis for subsequent joint feature construction, change identification and erosion patch delineation based on slope units.

[0078] S3. Using slope units as indices, the terrain normalized reflectivity features, slope aspect, and hydrodynamic factors are combined into a joint feature. Neighborhoods are constructed based on terrain similarity, and the joint feature is smoothed. The specific implementation is as follows:

[0079] During the joint feature construction phase, the monitoring platform uses the aforementioned slope units as the basic spatial units and indexing criteria, integrating spectral, topographic, and hydrodynamic information on a unified time axis. Specifically, at a given monitoring time step, the platform first retrieves the topographic normalized reflectance characteristics of the corresponding slope unit in each band from the slope spectral feature table based on the slope unit number and time stamp. Simultaneously, based on the same slope unit number, it obtains topographic attributes such as the sub-basin number, slope grade, specific slope value, slope aspect, and curvature interval of the slope unit from the topographic unit index table. Then, through hydrodynamic services, it summarizes the rainfall, maximum rainfall intensity, previous cumulative rainfall, and other hydrodynamic factors introduced as needed within the time step for the sub-basin to which the slope unit belongs, on a time scale consistent with the time step. In this implementation, hydrodynamic factors can be understood as meteorological and hydrological quantities reflecting the dynamic conditions for runoff and erosion on the slope. Specific types and statistical criteria are registered in the parameter configuration table and remain unchanged within the same monitoring version.

[0080] The monitoring platform combines the aforementioned spectral features, topographic attributes, and hydrodynamic factors into a joint feature vector according to a pre-agreed field order. In this embodiment, the joint feature vector is a set of ordered values ​​used to simultaneously describe the spectral state, topographic state, and hydrodynamic state of surface materials under a single slope unit and a single time step. To ensure temporal consistency, the time step can be set to an hourly or daily scale. In a preferred embodiment, the small watershed area range can be set to a range of one square kilometer to tens of square kilometers, with the specific range registered in the parameter configuration table. The time step is set to one hour or one day, and the specific value of the time step and the small watershed area range are registered in the parameter configuration table and associated with the rule version number.

[0081] After obtaining the joint feature vector of each slope unit at a certain time step, the monitoring platform constructs a terrain similarity neighborhood within the same small watershed based on terrain attributes. In this embodiment, the terrain similarity neighborhood is defined as a set of slope units within the small watershed that are similar to the target slope unit in terms of slope, aspect, and curvature. The degree of similarity is limited by a preset terrain similarity threshold: the absolute value of the slope difference is limited to within the preset slope difference threshold, the aspect difference is limited to the angle range corresponding to adjacent azimuth segments, and the curvature difference is limited to within the preset curvature difference range. The above three thresholds are registered in the parameter configuration table in the form of terrain similarity thresholds and remain unchanged under the same rule version. Preferably, the monitoring platform can also set a lower limit and an upper limit for the number of slope units included in the terrain similarity neighborhood to avoid statistical instability due to an excessively small neighborhood or excessive smoothing of features due to an excessively large neighborhood.

[0082] For each slope unit, after constructing a corresponding topographically similar neighborhood, the monitoring platform smooths the joint feature vector of the slope unit and other slope units within that neighborhood. A representative statistical range is determined by the distribution of joint features of multiple slope units within the neighborhood. For dimensions that significantly deviate from this statistical range, their weight in the neighborhood statistics is reduced, thereby weakening the impact of occasional noise or local anomalies on the joint features. Specific smoothing strategies can include weighted averaging, quantile truncation, or other numerical strategies that meet the neighborhood consistency requirements. The relevant parameters and weighting rules are recorded in the parameter configuration table. The smoothed joint features retain the main change characteristics of the target slope unit at that time step and maintain relative spatial coordination with topographically similar slopes within the same small watershed.

[0083] After smoothing is completed, the monitoring platform registers the joint features of each slope unit at each time step in the joint feature table in the form of slope unit number, time stamp, joint feature vector value, identification number of the corresponding terrain similar neighborhood, and rule version number. The joint feature table adopts a structured field storage method, which can support quick retrieval of the corresponding joint features by slope unit number and time stamp, and supports tracing the terrain similarity threshold, time step setting and hydrodynamic factor caliber used at that time by rule version number.

[0084] Preferably, the monitoring platform generates a new rule version number each time it modifies the terrain similarity threshold, time step, or hydrodynamic factor type, and retains the joint features of the old version in the joint feature table. The version number distinguishes the joint features generated under different rule combinations, thereby clearly distinguishing the differences between different versions in subsequent analysis and quality verification processes. This enables the joint feature construction process to be traceable and reproducible, providing a stable and consistent feature basis for subsequent change identification and erosion patch delineation.

[0085] S4. Using the joint characteristics of the historical baseline period as the baseline, calculate the difference between the joint characteristics of the target period and the baseline, and select adjacent slope units whose difference reaches a threshold along the confluence direction as candidate erosion slope units. The specific implementation is as follows:

[0086] During the change identification phase, the monitoring platform establishes a long-term traceable historical baseline for each slope unit based on the aforementioned joint feature table and topographic unit index table, and identifies abnormal changes related to soil erosion within the target time period. To this end, the platform first determines the historical baseline period for each slope unit based on historical rainfall records, remote sensing time series, and soil and water conservation management records for the monitoring area. The historical baseline period is defined as a continuous monitoring period prior to a given monitoring year that is no less than a preset threshold for the number of monitoring cycles (this threshold is registered in the parameter configuration table), during which no rainfall events exceeding the preset heavy rainfall threshold occur and no large-scale excavation, road construction, mining, or other construction disturbance events are registered in the soil and water conservation management records. The heavy rainfall threshold can be set based on multi-year rainfall statistics for the area, and is no less than the daily rainfall or time-period rainfall intensity corresponding to the preset return period (the return period is registered in the parameter configuration table). Construction disturbance information can be periodically imported by the soil and water conservation supervision department. The platform registers the heavy rainfall threshold, disturbance event type, and screening rules in the parameter configuration table, keeping them unchanged within the same rule version.

[0087] After determining the historical baseline period for each slope unit, the platform retrieves the joint feature records of the slope unit within the baseline period according to the monitoring time step, and statistically analyzes the representative values ​​and natural fluctuation ranges of the joint features of each dimension. The representative values ​​are preferably the median, mean, or typical values ​​within the quantile interval. The fluctuation range can be determined by the quantile interval, fluctuation amplitude, or other robust statistical indicators. The above representative values ​​and fluctuation ranges are registered together as the baseline of the slope unit, and the start and end times of the baseline period, the amount of data involved in the statistics, and the rule version number applicable at that time are recorded to form a baseline table. Each record in the baseline table can be uniquely identified by the slope unit number and the rule version number.

[0088] When the monitoring and dispatch service issues a target time period based on short-term rainfall forecasts, water conservation supervision plans, or emergency notifications, the monitoring platform reads the observation window length, consecutive judgment count, and difference threshold corresponding to the current rule version number from the threshold rule set to form a change identification rule. The observation window length is defined as a fixed time length that traces forward or backward from the target time period as the center or starting point, and can be set to a range of several days to several weeks. The consecutive judgment count is defined as the lower limit of the number of time steps within the observation window that exceed the difference threshold. The difference threshold is determined according to the baseline fluctuation range by a preset multiple. For example, the threshold for the joint feature difference of a certain dimension can be set as the difference between the upper limit of the baseline fluctuation range of that dimension and the representative value multiplied by a preset multiple factor. The multiple factor is registered in the parameter configuration table.

[0089] Within the aforementioned observation window, the platform reads the joint feature vector for each slope unit according to the time step for the target period, and subtracts the baseline representative value of the corresponding dimension from the vector dimension by dimension to obtain a joint feature difference sequence ordered by the time step. In the difference sequence, the platform focuses on several dimensions that reflect the increase in bare soil exposure, the decrease in vegetation cover, and the enhancement of hydrodynamics. The components of these dimensions in the joint feature vector have been fixed in the parameter configuration table during rule design to ensure that the meaning is consistent in different running batches.

[0090] For each slope unit, the platform counts the number of time steps within which the difference of the above key dimensions exceeds the difference threshold consecutively. When the consecutive number reaches or exceeds the preset consecutive judgment number, the slope unit is marked as a suspected erosion unit in the target time period, and the slope unit number, target time period mark, triggered key dimension, occurrence number, baseline version number and threshold rule version number are recorded in the change identification result record.

[0091] To link the abnormal changes of a single slope unit with the erosion process along the water flow path, the monitoring platform further calls the topographic unit index table. Based on the confluence direction relationship of the grids inside the slope unit, it searches for the upstream and downstream slope unit sets directly adjacent to the suspected erosion unit along the confluence direction. Within the observation window, it determines whether these upstream and downstream slope units also have records where the key dimension difference exceeds the corresponding difference threshold, and records the abnormal time step of these slope units in the difference sequence.

[0092] If, on a continuous slope link along the confluence direction, the upstream slope unit, the suspected erosion unit, and the downstream slope unit all meet the condition of reaching the determination number of consecutive occurrences within the same observation window, and the abnormal time steps overlap in time, the platform will uniformly mark all continuous slope units that meet the condition on the confluence link as candidate erosion slope unit links. In this embodiment, a candidate erosion slope unit is defined as a set of slope units that simultaneously exhibit continuous abnormal changes in the joint feature difference and are spatially continuous along the confluence direction.

[0093] For slope units that only occasionally exceed the difference limit at certain time steps within the observation window but do not form a continuous anomaly, the platform only records the abnormal points in the difference sequence and does not include them in the candidate erosion slope units; for slope units that cannot calculate the difference threshold due to missing data, insufficient baseline samples or rule version changes, the platform marks their status in the change identification result record and does not participate in the candidate erosion slope unit determination.

[0094] The settings for observation window length, number of consecutive judgments, and difference threshold are uniformly registered in the threshold rule set, corresponding one-to-one with the rule version number. Each time the threshold rule is adjusted or the time step is adjusted, the platform generates a new rule version number and retains the old version of the threshold rule set and change identification result record. This allows the platform to trace the base period, baseline statistical method, observation window settings, and difference threshold caliber used at the time of the change identification rule and result during subsequent quality inspection or regulatory audit.

[0095] The list of suspected erosion units and the list of candidate erosion slope unit links generated during the change identification stage can be stored in the change identification result table. Each record in the change identification result table is associated with the slope unit number, watershed number, target time period marker, anomaly dimension marker, rule version number, and judgment status, serving as input for the subsequent regional growth and erosion patch delineation stages. This forms a structured expression of soil and water loss-related abnormal changes on the time axis and spatial structure, and constitutes a complete chain of evidence through rule version management and baseline version management.

[0096] S5. Using candidate erosion slope units as seeds, regional growth is performed within the small watershed along the confluence direction based on slope, aspect, and elevation constraints. Growth terminates when the growth area reaches the watershed boundary and the location of abrupt aspect change. Erosion patches are then delineated in conjunction with the gully network. The specific implementation is as follows:

[0097] During the patch aggregation stage, the monitoring platform uses the candidate erosion slope units obtained in the aforementioned change identification stage as the starting point for regional expansion. Within each small watershed, it gradually expands along the confluence direction according to unified topographic and hydrological constraints, forming erosion patches with clear spatial boundaries and internal structures. Specifically, for a set of candidate erosion slope units within a small watershed, the platform first reads the watershed boundary, the upstream and downstream relationships between slope units, and the spatial relationship with the gully network from the topographic unit index table. Each candidate erosion slope unit is defined as a growth seed, and the slope unit containing the growth seed is included in the initial growth area. Within this area, a growth link is maintained, arranged in an orderly manner along the confluence direction. In this embodiment, the growth link refers to the sequence of slope units connected sequentially according to the confluence direction, used to reflect potential erosion paths.

[0098] During the growth process, the platform limits the search range to the boundary of the small watershed to which the growth seed belongs. The boundary of the small watershed is based on the watershed range recorded in the topographic unit index table. For each current slope unit on the growth link, the platform queries the set of downstream slope units along its confluence direction. For each candidate downstream slope unit, the platform checks whether the slope grade of the slope unit falls into the preset slope grade set, whether the slope aspect segment is consistent with the slope aspect segment of the current slope unit, and whether the elevation decreases relative to the current slope unit along the confluence direction. At the same time, the platform checks whether the slope unit has complete joint features and hydrodynamic factor records in the joint feature table and hydrodynamic records. Only when all the above conditions are met is the slope unit included in the growth area and added to the end of the growth link. The slope grade set, slope aspect segment division, elevation decrease judgment threshold, and requirements for the completeness of joint features and hydrodynamic factors are all registered in the parameter configuration table in the form of regional growth rules and bound to the rule version number, remaining unchanged within the same rule version. For downstream slope units that do not meet the above conditions, the platform will not include them in the growth area and will mark the specific reasons in the internal status, such as the slope aspect has crossed to another orientation segment, the elevation no longer decreases, or key information is missing, so that they can be used in subsequent quality analysis.

[0099] As the growth link extends along the confluence direction, when the growth area reaches the watershed boundary of the small watershed recorded in the topographic unit index table, the platform determines that further downstream expansion will cross the current small watershed range, and stops the expansion in that direction. When it is detected that the slope orientation segment of the downstream slope unit is inconsistent with the slope orientation segment of the previous slope unit in the current growth link and the difference exceeds the preset orientation difference threshold, it is considered that the slope orientation has changed abruptly, and it no longer continues to extend along that direction, and the patch boundary closes in this direction.

[0100] To characterize the coupling relationship between the erosion process and natural channels, the platform simultaneously invokes a pre-constructed channel network during the growth process. In this embodiment, the channel network is a set of linear water system elements extracted based on digital elevation data and hydrodynamic information. Each channel records its geometric path, classification type, and intersection relationship with slope units. When the growth link expands, the platform spatially overlays the set of slope units covered by the current growth link with the channel network to identify the intersection points of the growth link and the channel. When a growth link intersects with a high-level channel at multiple locations, it can be marked as the main erosion path. Growth links that intersect with low-level channels or only intersect within a short distance can be marked as branches. The erosion path type can be pre-defined according to the channel level, number of intersections, and intersection location in the parameter configuration table and registered in the erosion patch table as a patch type field.

[0101] Within the same small watershed, the platform sequentially executes the aforementioned expansion process on all candidate erosion slope units as growth seeds. It merges the slope unit sets covered by each growth link, defining a set of slope units that satisfy the regional growth rules, are continuous along the confluence direction, and are spatially connected as an erosion patch. Each erosion patch is assigned a unique patch number, and the following information is recorded in the erosion patch table: patch number, watershed number, list of included slope unit numbers, growth seed slope unit number, growth link length, growth termination reason, patch type, formation time stamp, and rule version number. For slope units that satisfy the slope, aspect, and elevation constraints along the confluence direction in a certain growth cycle but are missing from the joint feature table or hydrodynamic record, the platform skips these slope units in that regional growth cycle, excluding them from the corresponding erosion patch. Simultaneously, it marks these slope units as lacking information in the operation log and erosion patch table to prompt focused attention during subsequent on-site verification or data supplementation.

[0102] To control computing resources and ensure timely results, the platform sets time windows for scheduling tasks throughout the entire patch aggregation process. In each round of monitoring tasks, the maximum allowable time is limited from the start of regional growth to the completion of growth of all candidate erosion slope units in the current batch of watersheds. The upper limit of the time window can be set to not exceed a preset processing delay threshold, which corresponds to the total time of several time steps and is registered in the parameter configuration table as the processing delay threshold field. When multiple watersheds have candidate erosion slope units at the same time, the platform constructs a task queue by watershed and sets task priorities according to the importance level, candidate erosion area, or monitoring priority of the watershed, and executes the patch aggregation tasks of multiple watersheds in batches according to priority.

[0103] For small watershed tasks that fail to complete regional growth within the set time window, the platform marks them as partially completed or pending recalculation in the scheduling record, registers the growth links that have not yet completed growth and their corresponding growth seeds in the recalculation queue, resumes execution when computing resources are available, and retains complete information such as rule version number, growth status and interruption time in the erosion patch table and logs to ensure the traceability of the patch aggregation process in terms of time and rules.

[0104] In this way, the patch aggregation stage expands the candidate erosion slope units into erosion patches with clear spatial boundaries and type markings under topographic and hydrological constraints. This provides a clear spatial unit basis for subsequent calculation of erosion area, slope and topographic normalized reflectance changes by patch and for constructing erosion indicators at the slope unit and small watershed scales.

[0105] S6. Calculate the changes in erosion patch area, slope, and topographically normalized reflectance according to the small watershed number, and record the calculation results as time series data of slope unit erosion index and small watershed erosion index. The specific implementation is as follows:

[0106] During the indicator generation and release phase, the monitoring platform constructs a structured erosion indicator system at three levels—erosion patches, slope units, and small watersheds—based on the aforementioned topographic unit index table, elevation data, joint feature table, and erosion patch table, and generates traceable monitoring records on a unified timeline. For each erosion patch, the platform first obtains the raster range and elevation value covered by the corresponding slope unit from the topographic unit index table and digital elevation data, based on the slope unit number list registered in the erosion patch table. It then sums the areas of all raster units within the patch's coverage area according to a preset raster area to obtain the planar area of ​​the erosion patch. Within the same area, it calculates the average slope weighted by area based on the slope value, and records the obtained area and average slope as the patch's geometric attributes.

[0107] Subsequently, the platform retrieves the joint feature difference sequence of the erosion patch-covered slope unit within the target time period from the joint feature table and change identification results. Combined with the representative value of the topographic normalized reflectance of the slope unit at the baseline period, it calculates the change in topographic normalized reflectance of each slope unit within the patch. Preferably, the change in topographic normalized reflectance of each slope unit is weighted and summarized according to the area occupied by the slope unit in the patch to obtain a representative change that characterizes the overall vegetation degradation or bare soil exposure of the erosion patch. In a preferred embodiment, a similar area-weighted summary can also be performed on the relevant dimensions reflecting hydrodynamic changes, and the above results are registered together as patch spectral and hydrodynamic change indicators.

[0108] At the slope unit scale, the platform collects records of all erosion patches that a slope unit participates in during the monitoring period, using the slope unit as an index. It statistically analyzes the number of times the slope unit is covered by erosion patches within a year or multiple years, the ratio of the cumulative area of ​​participating patches to its own area, the average slope of participating patches, and the change in the normalized reflectance of representative terrain. These quantities are combined into slope unit erosion indices according to preset rules. The content of the slope unit erosion indices may include erosion level, cumulative erosion area ratio, annual erosion frequency, average slope of representative patches, and typical vegetation change range. The erosion level can be set to several levels. In the preferred embodiment, it can be divided into light, moderate, and severe, based on the area ratio and change threshold.

[0109] At the small watershed scale, the platform summarizes the area, average slope, and patch-level change indicators of all erosion patches within the same small watershed according to the small watershed number. This yields small watershed erosion indicators such as the proportion of eroded area in the small watershed to the total area, the proportion of area with strong erosion, the average slope range of typical patches, the range of typical vegetation changes, and the annual frequency of erosion. The proportion of area with strong erosion can be defined as the proportion of the projected area of ​​slope units with an erosion level not lower than the preset strong erosion level threshold to the total area of ​​the small watershed. The strong erosion level threshold is consistent with the erosion level classification threshold and is registered in the parameter configuration table. The specific composition of the small watershed erosion indicators is also registered in the parameter configuration table to ensure consistency between different monitoring batches.

[0110] The aforementioned slope unit erosion indicators and small watershed erosion indicators are written into the erosion indicator time series table in chronological order of monitoring time. The erosion indicator time series table uses a combination of small watershed number, slope unit number, monitoring time stamp, and rule version number as the monitoring idempotent key to identify duplicate tasks issued repeatedly in the same monitoring period, ensuring that only one valid indicator record is retained for the same time, the same spatial unit, and the same rule version. At the same time, a status field is set in the record to mark the operation status such as completed, pending recalculation, and missing information in a hierarchical numbering manner, which facilitates the upstream scheduling service and the downstream display service to recalculate tasks and filter results based on the status.

[0111] To verify the field applicability and stability of the indicator system, in representative deployments, samples can be drawn annually from all small watersheds and slope units at a preset ratio. Field surveys, aerial photo interpretation, or higher-resolution remote sensing images are used to verify the credibility of erosion patch boundaries, erosion levels, and related erosion indicators. Based on the number of verified samples, the pass rate, and the misjudgment rate, the difference threshold, observation window length, and erosion level classification threshold are adjusted. The adjusted rules and corresponding rule version numbers are archived together and linked with the erosion indicator records of the year to form a complete threshold evolution trajectory.

[0112] The entire monitoring system can be deployed on a centralized server, cloud computing platform, or other computing power platform with equivalent computing and storage capabilities. The process of each stage is completed through the call between terrain services, image services, hydrodynamic services, and monitoring services. All digital elevation data, multi-source remote sensing images, and hydrodynamic factors used are from channels that comply with relevant laws, regulations, and data management requirements. The platform limits the scope of data use through access control and permission management to ensure data security and compliance. In terms of hardware form and software architecture, functionally equivalent alternatives are allowed. As long as it has the ability to divide small watersheds and slope units with digital elevation data, construct topographic normalized reflectance features and joint features based on slope units, screen candidate erosion slope units along the confluence direction and aggregate them into erosion patches under topographic and gully constraints, and continuously register erosion indicators to form time series data according to small watershed number and slope unit number, it is considered to fall within the technical scope of this invention.

[0113] Example 2: Figure 2 A schematic diagram of the soil erosion monitoring system based on multi-source remote sensing and topographic factors of the present invention is given. The soil erosion monitoring system based on multi-source remote sensing and topographic factors includes:

[0114] The terrain unit division module is used to acquire regional digital elevation data, divide the raster units into small watersheds and slope units based on the confluence direction, slope, aspect and curvature of the raster units, and register the small watersheds, upstream slope units and downstream slope units corresponding to each slope unit.

[0115] The image radiometric feature construction module is used to resample multi-source remote sensing images to elevation data raster units according to the slope unit boundary, perform topographic radiometric correction on the raster radiometric amount according to the raster slope, aspect and imaging geometry parameters, and statistically analyze the topographic normalized reflectance features according to the slope unit.

[0116] The joint feature construction module is used to combine the terrain normalized reflectivity features, slope aspect and hydrodynamic factors into joint features using slope units as indexes, construct terrain similarity neighborhoods based on terrain similarity, and perform smoothing processing on the joint features within the terrain similarity neighborhoods.

[0117] The change identification module is used to calculate the difference between the joint features of the target time period and the baseline using the joint features of the historical reference period as the baseline, and select adjacent slope units whose difference reaches the threshold along the confluence direction and mark them as candidate erosion slope units.

[0118] The erosion patch delineation module is used to perform regional growth along the confluence direction within a small watershed based on the candidate erosion slope units as seeds, according to the constraints of slope, aspect and elevation. Growth is terminated when the growth area reaches the watershed boundary and the location of the abrupt change in aspect, and erosion patches are delineated in combination with the gully network.

[0119] The erosion index generation module is used to calculate the area, slope, and topographic normalized reflectance changes of erosion patches according to the small watershed number, and to register the calculation results as slope unit erosion index and small watershed erosion index, and to form time series data.

[0120] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0121] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0122] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center containing one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0123] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0124] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0125] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0126] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0127] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0128] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included 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.

[0129] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for monitoring soil erosion based on multi-source remote sensing and topographic factors, characterized in that, include: S1. Obtain regional digital elevation data, divide the watershed and slope units according to the confluence direction, slope, aspect and curvature, and register the watershed and upstream and downstream slope units corresponding to the slope units; S2. Based on the slope unit boundary, resample the multi-source remote sensing images to the elevation data raster, correct the radiance according to the raster slope, aspect and imaging geometry, and statistically analyze the terrain normalized reflectance characteristics according to the slope unit. S3. Using slope units as indexes, the terrain normalized reflectivity features, slope aspect and hydrodynamic factors are combined into joint features. Neighborhoods are constructed according to terrain similarity and the joint features are smoothed. S4. Using the joint characteristics of the historical baseline period as the baseline, calculate the difference between the joint characteristics of the target period and the baseline, and select adjacent slope units whose difference reaches the threshold along the confluence direction as candidate erosion slope units. S5. Using candidate erosion slope units as seeds, regional growth is carried out in the small watershed along the confluence direction according to the constraints of slope, aspect and elevation. Growth is terminated when the growth area reaches the watershed boundary and the location of the abrupt change in aspect. Erosion patches are delineated in combination with the gully network. S6. Calculate the changes in erosion patch area, slope, and topographic normalized reflectance according to the small watershed number, and register the calculation results as time series data of slope unit erosion index and small watershed erosion index.

2. The method for monitoring soil erosion based on multi-source remote sensing and topographic factors according to claim 1, characterized in that, S1 includes: Digital elevation data is constructed based on digital elevation model raster, and the version number is registered in the terrain data version table; Perform interpolation and smoothing processing on digital elevation data, and register the interpolation window size, height difference threshold, and smoothing parameters in the parameter configuration table; The confluence direction is derived based on the elevation difference between adjacent grid cells. The slope, aspect and curvature are calculated. Small watersheds and slope units are divided according to the classification rules set in the parameter configuration table. Each slope unit is assigned a slope unit number and its corresponding small watershed number. Upstream and downstream slope units are identified. The above numbers and the topographic data version number are registered in the topographic unit index table.

3. The method for monitoring soil erosion based on multi-source remote sensing and topographic factors according to claim 1, characterized in that, S2 include: Register the imaging time and imaging geometric parameters for multi-source remote sensing images in the image metadata table; The monitoring platform performs cloud volume discrimination based on the cloud volume ratio threshold in the parameter configuration table to remove images with excessive cloud volume; Using the slope unit boundaries in the topographic unit index table as constraints, multi-source remote sensing images are resampled to digital elevation data raster units using an interpolation method that preserves the total radiation. Based on the slope, aspect, imaging time, and imaging geometry parameters of the raster units, topographic radiometric correction is performed to obtain the raster topographic normalized reflectance. Topographic normalized reflectance characteristics are statistically analyzed on a slope unit basis, and the slope unit number and imaging time label are registered in the slope spectral feature table.

4. The method for monitoring soil erosion based on multi-source remote sensing and topographic factors according to claim 1, characterized in that, S3 include: Using the slope unit as the index, at the monitoring time step, the terrain normalized reflectance is read from the slope spectral feature table, the small watershed number, slope grade, slope aspect azimuth segment and curvature interval are read from the terrain unit index table, and the hydrodynamic factors corresponding to the monitoring time step are read from the hydrodynamic service. The above features are combined into a joint feature vector. Based on the preset slope difference threshold, slope aspect segment and curvature difference interval, a terrain similarity neighborhood is constructed. Within the terrain similarity neighborhood, a representative statistical range is determined according to the distribution of joint feature vectors. The weight of joint feature components that deviate from the representative statistical range is reduced to obtain smooth joint features. The smooth joint features, along with the slope unit number, time stamp, terrain similarity neighborhood identification number and rule version number, are registered in the joint feature table.

5. The method for monitoring soil erosion based on multi-source remote sensing and topographic factors according to claim 1, characterized in that, S4 include: Based on historical rainfall records, remote sensing time series, and water conservation management ledgers, the monitoring platform selects continuous monitoring periods for each slope unit that do not exceed the heavy rainfall threshold and have not been registered for construction disturbances as historical baseline periods. Within the historical baseline period, joint features are read from the joint feature table according to the monitoring time step, representative values ​​and fluctuation ranges of each dimension are calculated, and the representative values ​​and fluctuation ranges are registered as the baseline of the slope unit in the baseline table, and associated with the slope unit number, the start and end time of the baseline period, the amount of data involved in the statistics and the rule version number.

6. The method for monitoring soil erosion based on multi-source remote sensing and topographic factors according to claim 5, characterized in that: The monitoring platform obtains the joint features of each slope unit from the joint feature table within the observation window according to the time step, based on the observation window length, the number of consecutive judgments, and the difference threshold. It then compares these features with the baseline representative value to obtain the difference sequence. When the number of consecutive judgments that the difference exceeds the difference threshold reaches the number of consecutive judgments, the slope unit is marked as a suspected erosion unit. Based on the confluence direction relationship recorded in the topographic unit index table, the monitoring platform searches for upstream and downstream slope units adjacent to the suspected erosion unit along the confluence direction. When the upstream slope unit, the suspected erosion unit, and the downstream slope unit are all marked and the abnormal time steps overlap, the continuous slope units along the confluence direction are marked as candidate erosion slope unit links.

7. The method for monitoring soil erosion based on multi-source remote sensing and topographic factors according to claim 1, characterized in that, S5 include: During the patch aggregation stage, the monitoring platform sets the candidate erosion slope units identified in the change identification stage as growth seeds. Within the small watershed to which the growth seed belongs, the watershed boundary is defined according to the topographic unit index table. Along the confluence direction, the growth link is extended according to the slope grade set, slope aspect azimuth segment, and elevation decrease constraint. Under the condition that the joint feature table and hydrodynamic record are complete, the downstream slope unit is incorporated into the growth region; The expansion terminates when the growth link reaches the watershed boundary. The expansion terminates when the difference between the slope aspect azimuth segment of the downstream slope unit and the slope aspect azimuth segment of the current slope unit exceeds the azimuth difference threshold. By combining the gully network, slope units are aggregated and merged to form erosion patches.

8. The method for monitoring soil erosion based on multi-source remote sensing and topographic factors according to claim 1, characterized in that, S6 include: The monitoring platform determines the area of ​​erosion patches and the average slope based on the coverage and slope of the slope unit, and determines the change in topographic normalized reflectance based on the joint feature difference of the target period and the topographic normalized reflectance of the baseline period. Slope unit erosion index is formed by summarizing the area of ​​erosion patches, average slope, and changes in topographic normalized reflectance according to the slope unit number. Small watershed erosion indices are formed by summarizing the area of ​​erosion patches, average slope, and changes in topographic normalized reflectance according to the small watershed number. The erosion indices of slope units and small watersheds are written into the erosion index time series table in the order of monitoring time stamps, and the small watershed number, slope unit number, monitoring time stamp and rule version number are used as monitoring idempotent keys.

9. A soil erosion monitoring system based on multi-source remote sensing and topographic factors, used to implement the soil erosion monitoring method based on multi-source remote sensing and topographic factors as described in any one of claims 1-8, characterized in that, include: The terrain unit division module is used to acquire regional digital elevation data, divide the raster units into small watersheds and slope units based on the confluence direction, slope, aspect and curvature of the raster units, and register the small watersheds, upstream slope units and downstream slope units corresponding to each slope unit. The image radiometric feature construction module is used to resample multi-source remote sensing images to elevation data raster units according to the slope unit boundary, perform topographic radiometric correction on the raster radiometric amount according to the raster slope, aspect and imaging geometry parameters, and statistically analyze the topographic normalized reflectance features according to the slope unit. The joint feature construction module is used to combine the terrain normalized reflectivity features, slope aspect and hydrodynamic factors into joint features using slope units as indexes, construct terrain similarity neighborhoods based on terrain similarity, and perform smoothing processing on the joint features within the terrain similarity neighborhoods. The change identification module is used to calculate the difference between the joint features of the target time period and the baseline using the joint features of the historical reference period as the baseline, and select adjacent slope units whose difference reaches the threshold along the confluence direction and mark them as candidate erosion slope units. The erosion patch delineation module is used to perform regional growth along the confluence direction within a small watershed based on the candidate erosion slope units as seeds, according to the constraints of slope, aspect and elevation. Growth is terminated when the growth area reaches the watershed boundary and the location of the abrupt change in aspect, and erosion patches are delineated in combination with the gully network. The erosion index generation module is used to calculate the area, slope, and topographic normalized reflectance changes of erosion patches according to the small watershed number, and to register the calculation results as slope unit erosion index and small watershed erosion index, and to form time series data.

Images