A method for ecological restoration of caragana planting based on remote sensing and intelligent monitoring

Abstract

The present application relates to the technical field of ecological restoration, and more particularly to a caragana laying ecological restoration method based on remote sensing and intelligent monitoring. The method comprises the following steps: obtaining multispectral remote sensing images and ground survey data of the target grassland, and extracting vegetation coverage, soil erosion intensity and surface roughness; evaluating the degradation degree of the target grassland according to the vegetation coverage, soil erosion intensity and surface roughness, and determining the grassland degradation zoning map; determining the caragana laying parameters based on the grassland degradation zoning map; leveling and cleaning the target grassland; laying caragana on the pretreated target grassland surface according to the caragana laying parameters, and collecting monitoring images of the laying area in the target grassland by using a UAV at a preset period. The present application can accurately identify the degradation level and spatial difference of the grassland, thereby providing scientific and reasonable parameter settings for caragana laying, significantly improving the pertinence and effectiveness of the restoration measures, and optimizing the resource utilization efficiency.

Description

Technical Field

[0001] This invention relates to the field of ecological restoration technology, and in particular to an ecological restoration method for laying Caragana korshinskii based on remote sensing and intelligent monitoring. Background Technology

[0002] In the field of grassland ecological restoration, existing restoration schemes suffer from significant deficiencies in precision. Traditional methods rely primarily on manual field surveys and experience-based judgments when determining restoration areas and application parameters, lacking refined and quantitative assessments of multi-dimensional indicators such as vegetation cover and soil wind erosion intensity. This extensive decision-making approach makes it difficult for restoration measures to match the actual degradation level and spatial variations of the grassland, thus hindering the improvement of restoration effectiveness and the efficiency of resource input.

[0003] Furthermore, the ability to dynamically monitor and regulate the restoration process is lacking. Traditional effectiveness assessments often rely on periodic manual inspections or single remote sensing data sources, resulting in long monitoring cycles, high costs, and difficulty in achieving continuous, high-frequency monitoring of large-scale restoration areas. This leads to an inability to grasp the dynamics of vegetation restoration and the effectiveness of wind erosion control in a timely and accurate manner, making it difficult to dynamically adjust strategies based on restoration progress and limiting the closed-loop optimization and adaptive management of the ecological restoration process. Summary of the Invention

[0004] Therefore, it is necessary for the present invention to provide an ecological restoration method for laying Caragana korshinskii based on remote sensing and intelligent monitoring, so as to solve at least one of the above-mentioned technical problems.

[0005] To achieve the above objectives, a method for ecological restoration by laying Caragana korshinskii trees based on remote sensing and intelligent monitoring includes the following steps:

[0006] Step S1: Acquire multispectral remote sensing images and ground survey data of the target grassland, and extract vegetation cover, soil wind erosion intensity and surface roughness;

[0007] Step S2: Assess the degree of degradation of the target grassland based on vegetation cover, soil wind erosion intensity and surface roughness, and determine the grassland degradation zoning map;

[0008] Step S3: Determine the Caragana korshinskii laying parameters based on the grassland degradation zoning map; level and clean the target grassland;

[0009] Step S4: Lay Caragana korshinskii on the pretreated target grassland surface according to the Caragana korshinskii laying parameters, and use a drone to collect monitoring images of the laying area in the target grassland at a preset cycle. The monitoring images include monitoring multispectral images and oblique photography images.

[0010] Step S5: Extract the vegetation index and wind erosion modulus of the paving area based on the monitoring images, evaluate the restoration effect, generate a restoration effect evaluation report, and adjust the Caragana korshinskii paving parameters or paving area according to the restoration effect evaluation report.

[0011] The beneficial effects of this invention are as follows:

[0012] On the one hand, by integrating multispectral remote sensing and ground survey data, a high-precision quantitative assessment of grassland degradation has been achieved. This enables accurate identification of grassland degradation levels and spatial differences, thus providing scientifically sound parameter settings for Caragana korshinskii paving, significantly improving the targeting and effectiveness of restoration measures, and optimizing resource utilization efficiency.

[0013] On the other hand, by using drones to collect multispectral and oblique photographic images of the paving area at preset intervals, dynamic monitoring of the restoration process was achieved. This allows for real-time monitoring of the dynamic changes in vegetation restoration and wind erosion control, enabling timely adjustments to restoration strategies. This effectively solves the problem of insufficient dynamic monitoring in traditional methods and achieves closed-loop optimization and adaptive management of ecological restoration.

[0014] On the other hand, vegetation indices and wind erosion moduli are extracted from monitoring images to assess the restoration effect and generate an assessment report. Based on the assessment results, the Caragana korshinskii laying parameters or laying area are dynamically adjusted. This feedback adjustment mechanism ensures that restoration measures always adapt to the actual recovery state of the grassland, further enhancing the overall effectiveness and sustainability of ecological restoration. Attached Figure Description

[0015] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description taken in conjunction with the accompanying drawings:

[0016] Figure 1 A schematic flowchart illustrating the steps of an embodiment of an ecological restoration method for laying Caragana korshinskii based on remote sensing and intelligent monitoring is shown.

[0017] Figure 2 A detailed flowchart illustrating the steps of extracting surface roughness in step S1 of one embodiment is shown.

[0018] Figure 3 A schematic diagram of grassland degradation zoning is shown as an example. Detailed Implementation

[0019] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0020] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0021] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0022] To achieve the above objectives, please refer to Figures 1 to 3 This invention provides an ecological restoration method for laying Caragana korshinskii based on remote sensing and intelligent monitoring, comprising the following steps:

[0023] Step S1: Acquire multispectral remote sensing images and ground survey data of the target grassland, and extract vegetation cover, soil wind erosion intensity and surface roughness;

[0024] Step S2: Assess the degree of degradation of the target grassland based on vegetation cover, soil wind erosion intensity and surface roughness, and determine the grassland degradation zoning map;

[0025] Step S3: Determine the Caragana korshinskii laying parameters based on the grassland degradation zoning map; level and clean the target grassland;

[0026] Step S4: Lay Caragana korshinskii on the pretreated target grassland surface according to the Caragana korshinskii laying parameters, and use a drone to collect monitoring images of the laying area in the target grassland at a preset cycle. The monitoring images include monitoring multispectral images and oblique photography images.

[0027] Step S5: Extract the vegetation index and wind erosion modulus of the paving area based on the monitoring images, evaluate the restoration effect, generate a restoration effect evaluation report, and adjust the Caragana korshinskii paving parameters or paving area according to the restoration effect evaluation report.

[0028] Preferably, step S1 involves extracting vegetation cover, including:

[0029] Extract reflectance values ​​in the red and near-infrared bands from multispectral remote sensing images;

[0030] Vegetation coverage is calculated based on reflectance values ​​in the red and near-infrared bands.

[0031] The formula for calculating vegetation cover (NDVI formula) is as follows:

[0032] ;

[0033] in, For vegetation coverage, For near-infrared reflectivity, This refers to the reflectivity in the red light band.

[0034] In this embodiment of the invention, multispectral images are imported into image processing software, and the red light band (wavelength approximately 0.63-0.69 micrometers) and the near-infrared band (wavelength approximately 0.76-0.90 micrometers) are selected to calculate the reflectance values ​​for these two bands respectively. The reflectance values ​​are calculated based on the spectral reflectance characteristics of the images and reflect the reflectivity of vegetation in different wavelength bands.

[0035] In addition, the image processing software used in this case can be ENVI, ArcGIS, QGIS, and ERDAS Imagine, and this case does not impose any restrictions on this.

[0036] In one implementation of this invention, it is assumed that in a multispectral remote sensing image of a certain area, the red band reflectance value is 0.3 and the near-infrared band reflectance value is 0.7. Based on the above-mentioned formula for calculating vegetation cover, the calculation process for vegetation cover Z is as follows: =0.7 and Substituting 0.3 into the formula, we get... .

[0037] It is important to note that vegetation cover is calculated based on the difference in reflectance between the vegetation in the red and near-infrared wavelengths. Red light reflectance is primarily influenced by chlorophyll absorption, while near-infrared reflectance mainly reflects the health and density of the vegetation. In healthy vegetation, near-infrared reflectance is typically higher than red light reflectance. Therefore, a vegetation cover Z of 0.4 indicates that the reflectance of the vegetation in the near-infrared band is significantly higher than that in the red band, which generally signifies high vegetation density and good health. Specifically, a vegetation cover of 0.4 means that 40% of the area is covered by vegetation.

[0038] Preferably, step S1 involves extracting soil wind erosion intensity, including:

[0039] Extract the average reflectance in the visible light band and the average brightness temperature in the thermal infrared band of the multispectral remote sensing image.

[0040] In this embodiment of the invention, multispectral remote sensing images are imported into image processing software, and the visible light band and thermal infrared band are selected to calculate the average reflectance and average brightness temperature of these two bands, respectively. The average reflectance reflects the reflection characteristics of the Earth's surface in the visible light band, while the average brightness temperature reflects the thermal radiation characteristics of the Earth's surface.

[0041] The wind erosion sensitivity index is calculated based on average reflectance and average brightness temperature; the formula for calculating the wind erosion sensitivity index is as follows:

[0042] ;

[0043] in, The wind erosion sensitivity index The average reflectance in the visible light band. The average brightness temperature in the thermal infrared band. It is the angle between the local average annual wind direction and the surface slope aspect;

[0044] In one implementation of this invention, it is assumed that in a multispectral remote sensing image of a certain area, the average reflectance in the visible light band is 0.25, the average brightness temperature in the thermal infrared band is 300K, and the angle between the local annual average wind direction and the surface slope aspect is 30 degrees. Based on the above formula for calculating the wind erosion sensitivity index, the calculation process for the wind erosion sensitivity index (ESI) is as follows: , and Substituting into the formula, we get The calculated result is ESI = (−0.05) × 26.85 × 0.5 = −6.7125.

[0045] It is important to note that the Erosion Sensitivity Index (ESI) is assessed by comprehensively considering the average reflectance in the visible light band, the average brightness temperature in the thermal infrared band, and the angle between the local annual average wind direction and the surface slope aspect. In this example, the average reflectance in the visible light band... The value is 0.25, below the threshold of 0.3, indicating that the area has sparse vegetation cover, relatively bare ground, and is susceptible to wind erosion. The average brightness temperature in the thermal infrared band... The K value is 300K, higher than 273.15K, indicating a higher surface temperature in the region. This is likely due to sparse vegetation cover, leaving the soil directly exposed to sunlight, which in turn raises surface temperatures and exacerbates wind erosion. Furthermore, the angle between the local average annual wind direction and the surface slope aspect... A value of 30 degrees indicates that the wind direction and the slope of the land surface have a certain angle, which allows the wind to act more effectively on the land surface, increasing the risk of wind erosion. Taking all these factors into account, the wind erosion sensitivity index (ESI) is -6.7125, and a negative value indicates that the area has a high sensitivity to wind erosion.

[0046] Extract wind erosion pit density and average annual wind speed from ground survey data;

[0047] In this embodiment of the invention, wind erosion pit density and average annual wind speed data are obtained through ground surveys. Specifically, a field survey is conducted in the target area to record the number and distribution of wind erosion pits, and the wind erosion pit density is calculated. At the same time, the average annual wind speed of the area can be obtained through weather stations or anemometers. .

[0048] Soil wind erosion intensity is calculated based on wind erosion sensitivity index, wind erosion pit density, and annual average wind speed.

[0049] The formula for calculating soil wind erosion intensity is as follows:

[0050] ;

[0051] in, For soil wind erosion intensity, Density of wind erosion pits This represents the average annual wind speed.

[0052] In one implementation of this invention, it is assumed that in a ground survey of a certain area, the density of wind erosion pits is 20 per square kilometer, and the average annual wind speed is 5 meters per second. Based on the above formula for calculating soil wind erosion intensity, the soil wind erosion intensity... The calculation process is as follows: ESI = −6.7125, and Substituting into the formula, we get The calculation result is .

[0053] It should be noted that in this embodiment of the invention, the standard for determining the intensity of soil wind erosion is based on the absolute value of the soil wind erosion intensity. Specifically, the soil wind erosion intensity... The larger the absolute value of the wind erosion intensity, the stronger the wind erosion and the more severe the soil erosion. Generally speaking, when the absolute value of the soil wind erosion intensity exceeds a certain threshold, the wind erosion situation in the area can be considered relatively severe, requiring ecological restoration measures. For example, based on empirical data and field survey results, a threshold can be set, such as | A value >1000 indicates severe wind erosion, requiring priority for ecological restoration. In this example, the soil wind erosion intensity... The value is -2796.875, and its absolute value is much greater than 1000, so it can be determined that the soil erosion in this area is relatively serious.

[0054] Preferably, step S1 involves extracting surface roughness, including:

[0055] Step S131: In the multispectral remote sensing image, the gray-level co-occurrence matrix is ​​calculated using a sliding window with a preset range of pixels, and the contrast parameter and homogeneity parameter are extracted.

[0056] In one implementation of this invention, it is assumed that a 3×3 sliding window is used to analyze the texture features of a multispectral remote sensing image of a certain region. The specific operation is as follows: A 3×3 window containing 9 pixels is selected on the multispectral remote sensing image. Further, the gray-level co-occurrence matrix of the pixels within this window is calculated. The gray-level co-occurrence matrix describes the gray-level value relationship between pixels in the image. Through this matrix, contrast and homogeneity parameters can be extracted. For example, the contrast parameter calculated using a 3×3 sliding window is 0.8, and the homogeneity parameter is 0.2. The contrast parameter measures the degree of difference between different gray-level values ​​in the image, while the homogeneity parameter measures the uniformity of gray-level values ​​in the image.

[0057] In this example, the contrast parameter is 0.8, indicating significant variations in grayscale values ​​and complex textures in the image. The homogeneity parameter is 0.2, suggesting uneven distribution of grayscale values ​​and noticeable texture variations. These parameter values ​​indicate that the surface of this area may be relatively rough, with considerable texture variation.

[0058] Step S132: Standardize the contrast ratio parameter and homogeneity parameter to generate standard contrast ratio parameter and standard homogeneity parameter;

[0059] In this embodiment of the invention, the mean of each parameter value is subtracted and then divided by its standard deviation to obtain the standard contrast parameter and the standard homogeneity parameter.

[0060] In one implementation of this invention, it is assumed that the mean of the contrast parameter 0.8 is 0.5 and the standard deviation is 0.3; the mean of the homogeneity parameter 0.2 is 0.4 and the standard deviation is 0.1. Through standardized calculation, the standard contrast parameter is obtained as follows: The standard homogeneity parameter is .

[0061] Step S133: Using the standard ratio parameter and standard homogeneity parameter as input, the first principal component is extracted as the texture roughness index by dimensionality reduction through principal component analysis.

[0062] In this embodiment of the invention, a data matrix is ​​formed by combining standard contrast parameters and standard homogeneity parameters. Further, the covariance matrix of this data matrix is ​​calculated, reflecting the correlation between the various parameters. Further, the eigenvalues ​​and eigenvectors of the covariance matrix are calculated. The eigenvalues ​​represent the variance contribution rate of each principal component, and the eigenvectors represent the direction of the principal components. Principal components with a cumulative variance contribution rate greater than or equal to 85% are selected; typically, the first principal component has the highest variance contribution rate, therefore, the first principal component is selected as the texture roughness index.

[0063] In one implementation of this invention, it is assumed that the standard contrast parameter is 1.0 and the standard homogeneity parameter is -2.0. These two parameter values ​​are combined into a data matrix, as shown below: Next, calculate the covariance matrix of this data matrix. The formula for calculating the covariance matrix is: ;in, It is each sample, It is the sample mean. It is the sample size. It represents a vector The transpose of . In this example, there is only one sample, so the covariance matrix is: Next, we calculate the eigenvalues ​​and eigenvectors of the covariance matrix. The formulas for calculating the eigenvalues ​​and eigenvectors are as follows: ;in, Let I be the eigenvalue and I be the identity matrix. Solving this equation yields two eigenvalues: their corresponding eigenvectors are: , ; Eigenvalues The first principal component has the highest variance contribution rate, therefore it is chosen as the texture roughness index. The direction of the first principal component is determined by the feature vector. Decision. By projecting the raw data onto this principal component direction, the texture roughness index is obtained: Therefore, the texture roughness index of this area is 2.235, which takes into account the effects of contrast and homogeneity and reflects the roughness of the surface texture in this area.

[0064] Step S134: Combine the measured values ​​of surface protrusion height in the ground survey data, and correct the texture roughness index using a preset linear regression model to determine the surface roughness.

[0065] The preset linear regression model is as follows:

[0066] ;

[0067] in, For surface roughness, The texture roughness index. This represents the measured height of the surface uplift.

[0068] In this embodiment of the invention, the measured value of the surface protrusion height is obtained from ground survey data. Further, the texture roughness index and the measured value of the surface protrusion height are substituted into a preset linear regression model to calculate the corrected surface roughness.

[0069] In one implementation of this invention, the texture roughness index (TRI) is assumed to be 2.235, and the measured height of the surface protrusion is 8 cm. These values ​​are substituted into a preset linear regression model to calculate the corrected surface roughness. The specific calculation process is as follows: Calculation Adjustment factor: ;calculate Correction items: ; Calculate the corrected surface roughness: Substitute specific values: Therefore, the corrected surface roughness of this region It is 2.0371.

[0070] Preferably, step S2, determining the grassland degradation zoning map, includes:

[0071] Input vegetation cover, soil wind erosion intensity and surface roughness into the preset classification model, and output the preliminary classification results of grassland degradation level;

[0072] In this embodiment of the invention, a preset classification model is used to assess the degradation level of grassland. This model is based on three key indicators: vegetation cover, soil wind erosion intensity, and surface roughness. Specifically, the values ​​of these indicators are input into the classification model, which then outputs a preliminary classification result of the grassland degradation level for each pixel based on preset rules and thresholds. These preliminary classification results divide the grassland into different degradation levels, such as mild degradation, moderate degradation, and severe degradation.

[0073] In one implementation of this invention, it is assumed that the vegetation cover is 0.4, the soil wind erosion intensity is -2796.875, and the surface roughness is 2.0371. These values ​​are input into a preset classification model, and the model outputs that the grassland degradation level of the area is moderately degraded based on its internal rules and thresholds.

[0074] It is important to note that the pre-defined classification model is constructed through the following steps: Historical data from multiple grassland degradation areas are collected, including indicators such as vegetation cover, soil wind erosion intensity, and surface roughness. Each data point is labeled with a degradation level (e.g., mild, moderate, and severe) based on historical records, while outliers and missing values ​​are removed to ensure data integrity and accuracy. Vegetation cover, soil wind erosion intensity, and surface roughness are selected as key features, and all feature values ​​are normalized to the same range (e.g., 0 to 1) to eliminate dimensional differences between different indicators. A suitable classification algorithm, such as decision tree, random forest, support vector machine (SVM), or neural network, is selected. The classification model is trained using the labeled data, and model parameters are adjusted to optimize classification performance. Cross-validation or independent test sets are then used to verify the model's accuracy and generalization ability, ensuring stable performance across different datasets. Based on the validation results, model parameters are further adjusted to optimize model performance, and the trained model is deployed in real-world applications for real-time assessment of grassland degradation levels.

[0075] Using the pixels in the raster of the preliminary classification results as the basic unit, continuous degradation regions of the same level are identified according to the preset clustering density threshold and the preset clustering radius.

[0076] In one implementation of this invention, it is assumed that the preset clustering density threshold is 5 pixels / square kilometer and the clustering radius is 100 meters. A clustering algorithm is used to identify continuous degradation regions of the same level. For example, if a region contains more than 5 moderately degraded areas, and these pixels are connected within a 100-meter radius, then this region will be identified as a continuous moderately degraded region.

[0077] It should be noted that the clustering algorithms that can be used in this case include, but are not limited to, DBSCAN (density-based spatial clustering algorithm), K-Means (K-means clustering algorithm) and Mean Shift (mean shift clustering algorithm).

[0078] The degradation levels of adjacent areas are merged, and when the difference in degradation level between adjacent areas does not exceed 1 level, they are divided into the same zone;

[0079] In this embodiment of the invention, the degradation level of adjacent regions is checked. If the difference in degradation level between adjacent regions does not exceed one level, these regions are merged into one partition. For example, if a slightly degraded region is adjacent to a moderately degraded region, these two regions can be merged into one partition.

[0080] In one implementation of this invention, it is assumed that a slightly degraded region is adjacent to a moderately degraded region. According to the rules, the degradation level difference between these two regions is 1 level, therefore they can be merged into one partition. This partition will contain all pixels from both regions and will be treated as a single entity in subsequent analysis.

[0081] Assign a unique partition identifier to each merged partition and record the geographical boundary coordinates of the partition;

[0082] In one implementation of this invention, it is assumed that the merged partition includes mildly and moderately degraded areas with geographic boundary coordinates ranging from (100, 200) to (300, 400). A unique identifier, such as "Partition 1," is assigned to this partition, and its boundary coordinates are recorded. This information will be stored in a GIS system.

[0083] Spatial registration is performed between the partition identifier and the spatial resolution of the multispectral remote sensing image to generate a partition boundary layer;

[0084] In one implementation of this invention, it is assumed that the spatial resolution of the multispectral remote sensing image is 1 meter per pixel. The boundary coordinates of the partitions are registered with the spatial resolution of the image to generate a partition boundary layer. This layer will display the boundary of each partition and will be consistent with the spatial position of the multispectral remote sensing image.

[0085] By combining the distribution of sampling points in the ground survey data, the zoning boundaries in the zoning boundary layer are corrected, and a grassland degradation zoning map is output. The grassland degradation zoning map includes the degradation level, area, geographic coordinates and zoning identifier of each zoning.

[0086] In one implementation of this invention, it is assumed that the sampling points in the ground survey data are distributed within the merged zoning areas. Based on the location and data of these sampling points, the boundaries in the zoning boundary layer are corrected. For example, if the actual degradation level of a certain area shown by the sampling points is inconsistent with the level in the zoning map, the boundary of that area can be adjusted. The final output grassland degradation zoning map will contain detailed information for each zoning area, such as degradation level, area, geographic coordinates, and zoning identifier.

[0087] Preferably, in step S3, determining the Caragana korshinskii laying parameters based on the grassland degradation zoning map includes:

[0088] The density parameters of Caragana korshinskii are set according to the degradation level in the grassland degradation zoning map;

[0089] In one implementation of this invention, it is assumed that a grassland degradation zoning map shows a certain area as slightly degraded. According to preset rules, the Caragana korshinskii density parameter for slightly degraded areas is set to 3 Caragana korshinskii plants per square meter. Slightly degraded areas typically have a good vegetation base, and a lower Caragana korshinskii density can effectively promote vegetation recovery while avoiding excessive competition.

[0090] In another implementation of this invention, it is assumed that the grassland degradation zoning map shows a certain area as moderately degraded. According to preset rules, the Caragana korshinskii density parameter for the moderately degraded area is set to 5 Caragana korshinskii saplings per square meter. The vegetation cover in the moderately degraded area is low, requiring a medium density of Caragana korshinskii to provide sufficient coverage and protection, promoting vegetation recovery.

[0091] In another implementation of this invention, it is assumed that a grassland degradation zoning map shows a certain area as severely degraded. According to preset rules, the density parameter of Caragana korshinskii in the severely degraded area is set to 8 Caragana korshinskii plants per square meter. The vegetation cover in the severely degraded area is extremely low, and soil erosion is severe, requiring a high density of Caragana korshinskii to provide sufficient cover and fixation, prevent further soil erosion, and promote vegetation recovery.

[0092] The surface debris cover type of the target grassland was determined based on multispectral remote sensing imagery. The surface debris cover type includes shrub residue, construction waste and plastic waste.

[0093] In one implementation of this invention, it is assumed that multispectral remote sensing imagery shows shrub debris and plastic waste in the target grassland. Reflectance values ​​in the red band (wavelength approximately 0.63-0.69 micrometers), near-infrared band (wavelength approximately 0.76-0.90 micrometers), and thermal infrared band (wavelength approximately 10-12 micrometers) are extracted from the multispectral remote sensing imagery. For example, it is assumed that in a certain area, the red band reflectance is 0.3, the near-infrared band reflectance is 0.7, and the thermal infrared band reflectance is 0.2. Further, a spectral feature library is established based on the known spectral characteristics of shrub debris, construction waste, and plastic waste. For example, shrub debris has high reflectance in the near-infrared band and low reflectance in the red band; plastic waste has high reflectance in the red band and low reflectance in the thermal infrared band; construction waste has high reflectance in the thermal infrared band and low reflectance in the near-infrared band. The extracted reflectance values ​​are compared with the features in the spectral feature library to identify different types of surface debris. For example, if a certain area has a reflectance value of 0.3 in the red light band, 0.7 in the near-infrared band, and 0.2 in the thermal infrared band, its characteristics match those of shrub debris, and it is therefore identified as shrub debris. Through this comparison, the distribution area of ​​each type of debris is determined. For example, shrub debris is mainly distributed in the northern part of the grassland, while plastic waste is scattered in the southern part. In the GIS system, the identified debris types are marked on their corresponding geographical locations, generating a debris distribution map. Finally, the distribution information of surface debris is output, including the distribution area and coverage area of ​​each type of debris.

[0094] Calculate the coverage area percentage of different types of debris based on multispectral remote sensing images;

[0095] In this embodiment of the invention, the distribution range of each type of debris on the image is determined; further, the coverage area of ​​each type of debris is calculated; the coverage area of ​​each type of debris is divided by the total area of ​​the target grassland to obtain the coverage area ratio of each type of debris.

[0096] When the coverage area of ​​shrub residues exceeds the preset first percentage threshold, the Caragana density parameter is reduced according to the preset adjustment range.

[0097] In this embodiment of the invention, if the coverage area of ​​shrub residue exceeds a first threshold, such as 20%, the density parameter of Caragana korshinskii is reduced by a certain percentage, such as by 10%. This can avoid competition between Caragana korshinskii and shrub residue, and improve the effect of ecological restoration.

[0098] When the coverage area of ​​construction waste or plastic waste exceeds the preset second percentage threshold, the density parameter of Caragana korshinskii is increased according to the preset adjustment range.

[0099] In this embodiment of the invention, if the coverage area of ​​construction waste or plastic waste exceeds a second threshold, such as 10%, the density parameter of Caragana korshinskii is increased by a certain percentage, such as 20%. This can enhance the covering and fixing effect of Caragana korshinskii on these debris.

[0100] Among them, the preset first proportion threshold is greater than the preset second proportion threshold.

[0101] Preferably, step S3, which involves leveling and cleaning the target grassland, further includes determining the cleaning method, which includes:

[0102] The distribution data of surface debris is determined based on multispectral remote sensing images, and the proportion of debris coverage area is determined based on multispectral remote sensing images and surface debris distribution data.

[0103] In this embodiment of the invention, reflectance values ​​of the red, near-infrared, and thermal infrared bands are extracted from multispectral remote sensing images. Based on the spectral characteristics of each type of debris, the distribution area of ​​surface debris is identified. The debris coverage area of ​​each area is calculated, and the debris coverage area is divided by the total area of ​​the target grassland to obtain the debris coverage area percentage.

[0104] When the area covered by debris exceeds the preset area percentage threshold, a rotary tiller is used to clean the surface of the target grassland.

[0105] In this embodiment of the invention, an area percentage threshold is set, for example, 20%. If the area covered by a certain type of debris exceeds this threshold, it indicates that there is a lot of debris in the area, and a rotary tiller is needed for deep cleaning.

[0106] When the area covered by debris is less than or equal to a preset area percentage threshold, a laser leveler is used to level the surface of the target grassland.

[0107] In this embodiment of the invention, if the coverage area of ​​a certain type of debris does not exceed a preset area percentage threshold, it indicates that there are few debris in the area, and a laser leveler can be used to level the surface.

[0108] Preferably, step S5 involves extracting the vegetation index and wind erosion modulus of the paving area based on monitoring images and evaluating the restoration effect, including:

[0109] Perform radiometric and geometric corrections on the monitored multispectral images, output corrected multispectral images, and perform orthophoto projection and mosaicking to output orthophoto maps;

[0110] In this embodiment of the invention, it is assumed that the monitored multispectral image has certain radiometric and geometric distortions. Radiometric correction and geometric correction are performed on the image using image processing software. The corrected multispectral image is converted into an orthophoto image, where each pixel corresponds to a specific location on the ground. Multiple orthophoto images are mosaicked to generate a complete orthophoto map.

[0111] In one implementation of this invention, it is assumed that ENVI software is used for radiometric and geometric correction. The specific steps are as follows: Open ENVI software and load the monitored multispectral image. Select "Preprocessing" > "Radiometric Correction" > "Flat Field Correction" for radiometric correction. During the correction process, select an appropriate sensor model and atmospheric model, such as using the MODTRAN model for atmospheric correction. The software will automatically calculate and apply radiometric correction parameters to eliminate radiometric distortion in the image. Further, perform geometric correction by selecting "Preprocessing" > "Geometric Correction" > "Georeferencing," selecting a reference image or map with known geographic coordinates as a baseline, and selecting control points manually or automatically to align the image to be corrected with the reference image. The software will calculate geometric correction parameters based on the control points and apply these parameters for correction, ensuring that the geometry of the image is consistent with the actual land surface. Next, perform orthorectification. Select "Preprocessing" > "Geometric Correction" > "Orthorectification," and input the interior and exterior orientation elements of the image. These parameters can be obtained from the image's metadata. Choose an appropriate projection method, such as UTM projection, to ensure that each pixel corresponds to a specific location on the ground. The software will generate an orthophoto image where each pixel has accurate geographic coordinates. Then, perform image mosaicking. Select "Preprocessing" > "Mosaicking" > "Mosaic," load all the orthophoto images to be mosaicked, and choose a suitable mosaicking method, such as pixel-value-based mosaicking or feature-based mosaicking. The software will automatically stitch multiple orthophoto images into a complete orthophoto map.

[0112] Determine the vegetation coverage after restoration based on orthophoto images;

[0113] In one implementation of this invention, it is assumed that the red band reflectance of a certain region in an orthophoto image is 0.3, and the near-infrared band reflectance is 0.7. The vegetation cover of this region is calculated using the NDVI formula: The vegetation cover in this area is 0.4.

[0114] A three-dimensional model of the paving area was constructed based on oblique photogrammetry, and the volume and density of wind erosion pits were determined based on the three-dimensional model of the paving area.

[0115] In one implementation of this invention, it is assumed that oblique photogrammetry images cover the entire paving area. Using 3D modeling software, the acquired oblique photogrammetry images are synthesized into a detailed 3D model. In the 3D model, the volume of each wind erosion pit is identified and measured, and the number of wind erosion pits per unit area, i.e., the wind erosion pit density, is calculated. For example, assuming 10 wind erosion pits are identified in an area of ​​100 square meters, and the average volume of each wind erosion pit is 0.5 cubic meters, then the wind erosion pit density is 0.1 pits / square meter, and the average wind erosion pit volume is 0.5 cubic meters.

[0116] Calculate the wind erosion modulus after repair based on the volume and density of the wind erosion pits.

[0117] The formula for calculating the wind erosion modulus after repair is as follows:

[0118] ;

[0119] in, It is the wind erosion modulus after repair. It is the volume of the wind erosion pit. It is the density of wind erosion pits. It refers to the area of ​​the region.

[0120] In one implementation of this invention, it is assumed that the volume of the wind erosion pit is 0.5 cubic meters, the density of wind erosion pits is 0.1 pits / square meter, and the area is 100 square meters. The wind erosion modulus after repair is calculated according to the formula: The wind erosion modulus is 0.0005, indicating that the wind erosion effect after repair is weak and the repair effect is good.

[0121] Obtain the initial vegetation index and initial wind erosion modulus of the target grassland;

[0122] In one implementation of this invention, it is assumed that the initial vegetation index of the target grassland is 0.2 and the initial wind erosion modulus is 0.001. These initial data are obtained through monitoring and evaluation before the restoration project begins, serving as a benchmark for subsequent effect evaluation. An initial vegetation index of 0.2 indicates low vegetation coverage before restoration, and an initial wind erosion modulus of 0.001 indicates strong wind erosion before restoration.

[0123] The vegetation coverage change rate was determined based on the vegetation coverage after restoration and the initial vegetation index, and the wind erosion modulus change rate was determined based on the wind erosion modulus after restoration and the initial wind erosion modulus.

[0124] In this embodiment of the invention, the vegetation coverage change rate is calculated by comparing the restored vegetation coverage with the initial vegetation index. Simultaneously, the wind erosion modulus change rate is calculated by comparing the restored wind erosion modulus with the initial wind erosion modulus.

[0125] The specific formulas are: Vegetation coverage change rate = (Restored vegetation coverage - Initial vegetation index) / Initial vegetation index, Wind erosion modulus change rate = (Restored wind erosion modulus - Initial wind erosion modulus) / Initial wind erosion modulus.

[0126] The restoration effect is assessed based on the rate of change in vegetation cover and the rate of change in wind erosion modulus, and a restoration effect assessment report is generated.

[0127] Preferably, the restoration effect is evaluated based on the rate of change in vegetation cover and the rate of change in wind erosion modulus, including:

[0128] When the rate of change of vegetation coverage is greater than the preset first rate of change threshold and the rate of change of wind erosion modulus is greater than the preset second rate of change threshold, the restoration effect is determined to be in line with expectations and no further adjustment of restoration measures is required.

[0129] In one implementation of this invention, it is assumed that the preset first rate of change threshold is 50%, and the second rate of change threshold is 30%. Through monitoring and calculation, the vegetation coverage change rate is found to be 60%, and the wind erosion modulus change rate is 40%. Since both of these rates of change exceed the preset thresholds, the restoration effect is determined to be satisfactory, and no further adjustment of the restoration measures is required. When the vegetation coverage change rate is less than or equal to the preset first rate of change threshold or the wind erosion modulus change rate is less than or equal to the preset second rate of change threshold, the restoration effect is determined to be unsatisfactory, and further adjustment of the restoration measures is required.

[0130] In one implementation of this invention, it is assumed that monitoring and calculation show a vegetation cover change rate of 40% and a wind erosion modulus change rate of 20%. Since the vegetation cover change rate does not reach the 50% threshold and the wind erosion modulus change rate does not reach the 30% threshold, the restoration effect is deemed unsatisfactory. Further analysis is needed to determine the reasons, such as improper implementation of restoration measures or changes in environmental conditions. Based on the analysis results, restoration measures are adjusted, such as increasing vegetation planting density or improving irrigation conditions, to improve the restoration effect.

[0131] Preferably, step S4 involves using a drone to collect monitoring images of the laying area at a preset cycle, including:

[0132] The drone is equipped with a multispectral camera and an oblique photography camera. The multispectral camera covers the visible light and near-infrared bands, and the oblique photography camera acquires high-resolution images in K directions, where K is a positive integer.

[0133] In one implementation of this invention, K is assumed to be 4, meaning the oblique photography camera acquires images from four directions. Specifically, during the flight of the UAV, the multispectral camera simultaneously acquires images in the visible and near-infrared bands, while the oblique photography camera acquires high-resolution images from the front, rear, left, and right directions respectively.

[0134] Set the drone flight parameters, which include flight altitude, flight speed, and image overlap.

[0135] In one implementation of this invention, the flight altitude of the UAV is set to 100 meters to ensure that the resolution and coverage of the images are moderate; the flight speed is set to 5 meters per second to ensure the continuity of the images and the acquisition efficiency; and the image overlap is set to 60% to ensure that there is sufficient overlap between adjacent images.

[0136] Based on the UAV flight parameters and preset cycle, a flight path is planned to generate a flight path covering the paved area. During each flight, multispectral images and oblique photography images are collected and recorded as monitoring images.

[0137] In one implementation of this invention, it is assumed that the preset cycle is once a week, and the monitoring area is a 1-square-kilometer grassland. Specifically, using UAV flight planning software, the flight altitude of 100 meters, flight speed of 5 meters per second, image overlap of 60%, and the geographic coordinates of the grassland are input to generate a flight path covering the entire grassland. The UAV automatically flies once a week according to the planned path, collecting multispectral and oblique photographic images. These images are stored in the UAV's storage device and downloaded to a ground workstation for further analysis and processing after the flight.

[0138] Most importantly, after step S5, the following is also included:

[0139] A knowledge graph for ecological restoration through the application of Caragana korshinskii was constructed. The knowledge graph includes Caragana korshinskii species characteristic nodes, grassland degradation characteristic nodes, and restoration effect evaluation nodes. The nodes are connected through causal and correlational relationships.

[0140] Most importantly, the construction of the ecological restoration knowledge graph for Caragana korshinskii paving includes:

[0141] Step S61: Obtain historical restoration project data and extract Caragana species characteristic data from the historical restoration project data. Caragana species characteristic data includes drought resistance, root depth and growth rate.

[0142] In one implementation of this invention, it is assumed that data from three historical restoration projects are extracted from a database. These projects are located in different geographical regions and record the growth of Caragana korshinskii under different soil moisture and climatic conditions. Analysis of this data reveals that Caragana korshinskii exhibits high drought tolerance in arid regions, with an average root depth of 1.5 meters and an annual growth rate of approximately 30 centimeters.

[0143] Step S62: Extract degradation feature data from the grassland degradation zoning map. The degradation feature data includes soil moisture content, wind erosion rate, and vegetation type.

[0144] In one implementation of this invention, it is assumed that the grassland degradation zoning map shows three regions with different degrees of degradation. The first region has low soil moisture content, a high wind erosion rate, and vegetation mainly composed of drought-resistant herbaceous plants; the second region has moderate soil moisture content, a low wind erosion rate, and diverse vegetation types; the third region has high soil moisture content, a low wind erosion rate, and abundant vegetation types.

[0145] Step S63: Extract effect characteristic data from the restoration effect assessment report. The effect characteristic data includes vegetation index improvement rate, wind erosion modulus reduction rate, and Caragana survival rate.

[0146] In one implementation of this invention, it is assumed that the restoration effect evaluation report shows the restoration effects in three areas. In the first area, the vegetation index increased by 20%, the wind erosion modulus decreased by 30%, and the survival rate of Caragana korshinskii was 90%; in the second area, the vegetation index increased by 30%, the wind erosion modulus decreased by 40%, and the survival rate of Caragana korshinskii was 95%; in the third area, the vegetation index increased by 40%, the wind erosion modulus decreased by 50%, and the survival rate of Caragana korshinskii was 98%.

[0147] Step S64: Construct a knowledge graph of ecological restoration through Caragana spp. paving based on Caragana spp. species characteristic data, degradation characteristic data, and effect characteristic data.

[0148] In one implementation of this invention, Neo4j is assumed to be used as the knowledge graph construction tool. Extracted data on the species characteristics, degradation features, and effects of *Caragana korshinskii* are input into Neo4j, and nodes and relationships are defined using Cypher, the query language for graph databases. For example, the causal relationship between the "drought resistance" node and the "soil moisture content" node, and the association between the "vegetation index improvement rate" node and the "Caragana korshinskii survival rate" node are defined, ultimately generating the knowledge graph.

[0149] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0150] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A method for ecological restoration by laying Caragana korshinskii trees based on remote sensing and intelligent monitoring, characterized in that, Includes the following steps: Step S1: Acquire multispectral remote sensing images and ground survey data of the target grassland, and extract vegetation cover, soil wind erosion intensity and surface roughness; Step S2: Assess the degree of degradation of the target grassland based on vegetation cover, soil wind erosion intensity and surface roughness, and determine the grassland degradation zoning map; Step S3: Determine the Caragana korshinskii laying parameters based on the grassland degradation zoning map; Level and clear the target grassland; The parameters for determining Caragana korshinskii laying based on grassland degradation zoning maps include: The density parameters of Caragana korshinskii are set according to the degradation level in the grassland degradation zoning map; The surface debris cover type of the target grassland was determined based on multispectral remote sensing imagery. The surface debris cover type includes shrub residue, construction waste and plastic waste. Calculate the coverage area percentage of different types of debris based on multispectral remote sensing images; When the coverage area of ​​shrub residues exceeds the preset first percentage threshold, the Caragana density parameter is reduced according to the preset adjustment range. When the coverage area of ​​construction waste or plastic waste exceeds the preset second percentage threshold, the density parameter of Caragana korshinskii is increased according to the preset adjustment range. Among them, the preset first proportion threshold is greater than the preset second proportion threshold; Step S4: Lay Caragana korshinskii on the pretreated target grassland surface according to the Caragana korshinskii laying parameters, and use a drone to collect monitoring images of the laying area in the target grassland at a preset cycle. The monitoring images include monitoring multispectral images and oblique photography images. Step S5: Extract the vegetation index and wind erosion modulus of the paving area based on monitoring images, evaluate the restoration effect, generate a restoration effect evaluation report, and adjust the Caragana korshinskii paving parameters or paving area according to the restoration effect evaluation report. The extraction of vegetation index and wind erosion modulus of the paving area based on monitoring images and the evaluation of restoration effect include: Perform radiometric and geometric corrections on the monitored multispectral images, output corrected multispectral images, and perform orthophoto projection and mosaicking to output orthophoto maps; Determine the vegetation coverage after restoration based on orthophoto images; A three-dimensional model of the paving area was constructed based on oblique photogrammetry, and the volume and density of wind erosion pits were determined based on the three-dimensional model of the paving area. Calculate the wind erosion modulus after repair based on the volume and density of the wind erosion pits. Obtain the initial vegetation index and initial wind erosion modulus of the target grassland; The vegetation coverage change rate was determined based on the vegetation coverage after restoration and the initial vegetation index, and the wind erosion modulus change rate was determined based on the wind erosion modulus after restoration and the initial wind erosion modulus. The restoration effect is assessed based on the rate of change in vegetation cover and the rate of change in wind erosion modulus, and a restoration effect assessment report is generated.

2. The ecological restoration method for laying Caragana korshinskii based on remote sensing and intelligent monitoring according to claim 1, characterized in that, Step S1 extracts vegetation cover, including: Extract reflectance values ​​in the red and near-infrared bands from multispectral remote sensing images; Vegetation coverage is calculated based on reflectance values ​​in the red and near-infrared bands.

3. The ecological restoration method for laying Caragana korshinskii based on remote sensing and intelligent monitoring according to claim 1, characterized in that, Step S1 extracts soil wind erosion intensity, including: Extract the average reflectance in the visible light band and the average brightness temperature in the thermal infrared band of the multispectral remote sensing image. The wind erosion sensitivity index is calculated based on average reflectance and average brightness temperature; the formula for calculating the wind erosion sensitivity index is as follows: ； in, The wind erosion sensitivity index The average reflectance in the visible light band. The average brightness temperature in the thermal infrared band. It is the angle between the local average annual wind direction and the surface slope aspect; Extract wind erosion pit density and average annual wind speed from ground survey data; Soil wind erosion intensity is calculated based on wind erosion sensitivity index, wind erosion pit density, and annual average wind speed.

4. The ecological restoration method for laying Caragana korshinskii based on remote sensing and intelligent monitoring according to claim 1, characterized in that, Step S1 extracts surface roughness, including: Step S131: In the multispectral remote sensing image, the gray-level co-occurrence matrix is ​​calculated using a sliding window with a preset range of pixels, and the contrast parameter and homogeneity parameter are extracted. Step S132: Standardize the contrast ratio parameter and homogeneity parameter to generate standard contrast ratio parameter and standard homogeneity parameter; Step S133: Using the standard ratio parameter and standard homogeneity parameter as input, the first principal component is extracted as the texture roughness index by dimensionality reduction through principal component analysis. Step S134: Combine the measured values ​​of surface protrusion height in the ground survey data, and correct the texture roughness index using a preset linear regression model to determine the surface roughness.

5. The ecological restoration method for laying Caragana korshinskii based on remote sensing and intelligent monitoring according to claim 1, characterized in that, Step S2 involves determining the grassland degradation zoning map, including: Input vegetation cover, soil wind erosion intensity and surface roughness into the preset classification model, and output the preliminary classification results of grassland degradation level; Using the pixels in the raster of the preliminary classification results as the basic unit, continuous degradation regions of the same level are identified according to the preset clustering density threshold and the preset clustering radius. The degradation levels of adjacent areas are merged, and when the difference in degradation level between adjacent areas does not exceed 1 level, they are divided into the same zone; Assign a unique partition identifier to each merged partition and record the geographical boundary coordinates of the partition; Spatial registration is performed between the partition identifier and the spatial resolution of the multispectral remote sensing image to generate a partition boundary layer; By combining the distribution of sampling points in the ground survey data, the zoning boundaries in the zoning boundary layer are corrected, and a grassland degradation zoning map is output. The grassland degradation zoning map includes the degradation level, area, geographic coordinates and zoning identifier of each zoning.

6. The ecological restoration method for laying Caragana korshinskii based on remote sensing and intelligent monitoring according to claim 5, characterized in that, Step S3 involves leveling and clearing the target grassland, which also includes determining the clearing method. The determination of the clearing method includes: The distribution data of surface debris is determined based on multispectral remote sensing images, and the proportion of debris coverage area is determined based on multispectral remote sensing images and surface debris distribution data. When the area covered by debris exceeds the preset area percentage threshold, a rotary tiller is used to clean the surface of the target grassland. When the area covered by debris is less than or equal to a preset area percentage threshold, a laser leveler is used to level the surface of the target grassland.

7. The ecological restoration method for laying Caragana korshinskii based on remote sensing and intelligent monitoring according to claim 6, characterized in that, The effectiveness of restoration was assessed based on the rate of change in vegetation cover and the rate of change in wind erosion modulus, including: When the rate of change of vegetation coverage is greater than the preset first rate of change threshold and the rate of change of wind erosion modulus is greater than the preset second rate of change threshold, the restoration effect is determined to be in line with expectations and no further adjustment of restoration measures is required. When the rate of change of vegetation coverage is less than or equal to the preset first rate of change threshold or the rate of change of wind erosion modulus is less than or equal to the preset second rate of change threshold, the restoration effect is determined to be unsatisfactory and further adjustments to the restoration measures are required.

8. The ecological restoration method for laying Caragana korshinskii based on remote sensing and intelligent monitoring according to claim 1, characterized in that, Step S4 involves using a drone to collect monitoring images of the laying area at preset intervals, including: The drone is equipped with a multispectral camera and an oblique photography camera. The multispectral camera covers the visible light and near-infrared bands, and the oblique photography camera acquires high-resolution images in K directions, where K is a positive integer. Set the drone flight parameters, which include flight altitude, flight speed, and image overlap. Based on the UAV flight parameters and preset cycle, a flight path is planned to generate a flight path covering the paved area. During each flight, multispectral images and oblique photography images are collected and recorded as monitoring images.

Images