A method and system for monitoring and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspection

Abstract

This invention proposes a method and system for monitoring and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspection, belonging to the field of water conservancy engineering monitoring technology. The method includes the following steps: S1: Divide the riverbank into segments and select the segments to be monitored; S2: Obtain historical data on hydrological conditions, erosion resistance, slope, rainfall intensity, and historical vegetation coverage of the monitored segments, and construct a riverbank erosion rate prediction model based on the historical data; S3: Plan the flight path of the UAV in the monitored segments and obtain orthophotos at different monitoring times; S4: Calculate the actual erosion rate of the segments based on the orthophotos at different monitoring times; S5: Based on the calculated actual erosion rate and the predicted erosion rate of the riverbank erosion rate prediction model at the corresponding time, assess the degree of riverbank erosion based on the difference and values ​​between the actual and predicted erosion rates.

Description

Technical Field

[0001] This invention relates to the field of water conservancy engineering monitoring technology, and in particular to a method and system for monitoring and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspection. Background Technology

[0002] River landforms are geomorphic features on the Earth's surface formed by erosion, transportation, and deposition, typically including headward erosion, downcutting, and lateral erosion. River landforms reflect the corresponding river hydrogeomorphic processes, which in turn influence river morphology. By assessing the impacts of natural and anthropogenic disturbances on river landforms, we can determine the river landform conditions and predict trends in their changes.

[0003] Riverbank erosion refers to the process by which a river, through its own dynamics and the sediment it carries, damages the riverbed, leading to the deepening and widening of the valley. Over time, this erosion can destabilize the slopes on both banks, posing safety hazards. Riverbank erosion is typically monitored using fixed video surveillance, but this method only focuses on specific areas and cannot provide a comprehensive picture of the overall erosion situation across an entire region.

[0004] Therefore, it is essential to provide a method and system for monitoring and assessing riverbank erosion based on drone inspections. By combining hydrological conditions with regular drone-based real-time photography of designated riverbank areas, the correlation between predicted erosion rates and actual erosion changes can be established, thereby improving the accuracy of riverbank erosion monitoring. Summary of the Invention

[0005] In view of this, the present invention proposes a riverbank erosion monitoring and assessment method and system based on UAV inspection, which scientifically segments the riverbank and prioritizes monitoring of high-risk areas, while using a combination of predictive models and UAV inspections to double-confirm the actual erosion status of high-risk areas.

[0006] On the one hand, the present invention provides a method for monitoring and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspections, comprising the following steps: S1: Divide the riverbank into segments and select the riverbank areas that need to be monitored; S2: Obtain historical data on hydrological conditions, erosion resistance, slope, rainfall intensity, and historical vegetation coverage of the monitored segmented riverbank areas, and construct a riverbank erosion rate prediction model based on historical data; S3: Plan the flight path of the UAV in the monitored segmented riverbank area and obtain orthophotos at different monitoring times; S4: Calculate the actual erosion rate of the segmented riverbank area based on orthophotos from different monitoring times; S5: Based on the calculated actual erosion rate and the predicted erosion rate of the riverbank erosion rate prediction model at the corresponding time, assess the degree of riverbank erosion according to the difference and value between the actual erosion rate and the predicted erosion rate.

[0007] Based on the above technical solutions, preferably, step S1 involves dividing the riverbank into river bends, river branching sections, straight sections, and sections with abrupt changes in river width; assigning a unique segment number to each segment, ensuring that each segment has the same length, and recording the location of each segment; and selecting riverbank areas near towns, farmland, or areas lacking vegetation for monitoring.

[0008] Preferably, step S2 involves acquiring historical data on hydrological conditions, erosion resistance, slope, rainfall intensity, and historical vegetation coverage of the monitored riverbank sections for more than five consecutive years, and constructing the first erosion component accordingly. E t Second erosion component E m and the third erosion component E s The first erosion component of the joint system E t Second erosion component E m and the third erosion component E s A model for predicting riverbank erosion rates was obtained. E = E t + mE m + E s , m The average number of times the non-zero output case of the second erosion component occurs within the monitoring period.

[0009] Preferably, the first erosion component E t By calculating the shear stress of the water flow With critical shear stress The difference, through the index a Attenuation, based on the erosion coefficient It was obtained by adjusting the proportions.

[0010] Preferably, the second erosion component E m The method involves selecting several sliding bodies in the monitored segmented riverbank area, establishing an instability judgment criterion based on the safety factor of each sliding body, and obtaining the second erosion component based on the output of the instability judgment criterion of the sliding body safety factor. E m .

[0011] Preferably, the third erosion component E s It is obtained by combining soil erodibility factors, rainfall, slope and vegetation index.

[0012] Preferably, step S3 involves the UAV carrying a camera, setting a monitoring period for the hydrological event, and acquiring images of the area where the landslide is located in the same segmented riverbank area at noon on each day during the monitoring period, according to a given fixed flight altitude, pixel resolution, focal length, flight path overlap range, and upper limit of ambient wind speed. The images are then stitched together to obtain an orthophoto map of the monitoring period based on a time series, and the orthophoto map of the area where the landslide is located on any day before the monitoring period is used as the reference orthophoto map.

[0013] Preferably, step S4 involves finding the same land features in orthophoto images from different monitoring periods, registering them to the same world coordinate system, pre-obtaining the correspondence between pixel size and the actual size in the world coordinate system, identifying the maximum changes in shoreline location and vegetation cover in the area where the landslide body is located, and comparing them with the shoreline location and vegetation cover in the baseline orthophoto image to calculate the actual erosion rate: 1) Marking the location of the landslide body in both the baseline orthophoto image and the monitoring period orthophoto image, and conducting water body detection within the adjacent rectangle of the landslide body. The adjacent rectangle extends from the top of the bank slope to the center of the river channel. The water body detection uses an improved water index method. MNDWI The threshold for the water index is [0, 1]. Then, a 3×3 kernel opening operation is performed on the water area to eliminate noise, and a 15×15 kernel closing operation is used to fill the holes. The pixel area within the water index area in the baseline orthophoto and monitoring period orthophoto are calculated and converted into the actual water area to obtain the area corresponding to the change in shoreline position; 2) The water index area in the adjacent rectangle is removed, and the normalized vegetation index is used in the remaining adjacent rectangle. NDVI Calculate the number of vegetation pixels and divide it by the total number of pixels outside the water surface to obtain the vegetation coverage in the baseline orthophoto and the monitoring period orthophoto. By comparison, obtain the change in vegetation coverage within the adjacent rectangle; 3) Actual erosion rate E f This is obtained by determining the area corresponding to the change in shoreline location, the mass of erosion per unit shoreline area, the change in vegetation cover within the adjacent rectangle, and the amount of erosion corresponding to the decrease in vegetation cover per unit area.

[0014] Preferably, step S5 involves, for each sliding body, dividing the first erosion component... E t and the third erosion component E s According to the width of the sliding bodyL The length of the monitored segmented riverbank area L After calculating the ratio to 0, the second erosion component of the sliding body during the monitoring period is added. mE m The predicted erosion rate of each sliding body during the monitoring period is obtained. , with the actual erosion rate of the sliding body E f Compare the results and obtain the relative error: 1) If the relative error does not exceed 15%, it indicates that the predicted erosion rate is within acceptable limits. Reliable, using predicted erosion rates 1) Assess the degree of erosion; 2) If the relative error is >15%, it indicates a predicted erosion rate. Unreliable; actual erosion rate should be used. E f Assess the degree of erosion; 3) If the erosion rate is <500, or the shoreline retreat speed does not exceed 0.1m / year, it is considered light erosion; if 500 < erosion rate ≤ 2000, and the shoreline retreat speed is greater than 0.1m / year but not more than 1.5m / year, it is considered moderate erosion; if the erosion rate is ≥ 2000, or the shoreline retreat speed is > 1.5m / year, it is considered severe erosion.

[0015] On the other hand, the present invention provides a riverbank erosion monitoring and assessment system based on unmanned aerial vehicle (UAV) inspection, for implementing the aforementioned riverbank erosion monitoring and assessment method based on UAV inspection, including: The riverbank segmentation module is used to segment the riverbank and select the segmented riverbank area that needs to be monitored as the object to be analyzed. The erosion rate prediction model construction module is connected to the riverbank segmentation module to acquire historical data on hydrological conditions, erosion resistance, slope, rainfall intensity, and historical vegetation coverage of the monitored riverbank segments, and to construct a riverbank erosion rate prediction model based on historical data. The drone monitoring module plans the flight path of the drone in the monitored segmented riverbank area and acquires orthophotos of the camera on the drone at different monitoring times; based on the orthophotos of the different monitoring times, it calculates the actual erosion rate of the segmented riverbank area. The analysis and evaluation module is connected to the erosion rate prediction model construction module and the UAV monitoring module, respectively. It is used to evaluate the degree of riverbank erosion based on the calculated actual erosion rate and the predicted erosion rate of the riverbank erosion rate prediction model at the corresponding time, and based on the difference and value between the actual erosion rate and the predicted erosion rate.

[0016] The present invention provides a method and system for monitoring and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspection, which has the following advantages compared with the prior art: (1) By scientifically segmenting the riverbank and screening high-risk areas for monitoring, it is beneficial to reduce the amount of data processing required for maintenance and monitoring; (2) A multi-component erosion prediction model was constructed using historical data of more than five years. The first erosion component considered the dynamic difference between the shear stress of the water flow and the critical shear stress. The second erosion component considered the instability analysis of the safety factor of the sliding body and the soil. The third erosion component considered the vegetation coverage under the influence of slope and rainfall, which together improved the accuracy and reliability of erosion prediction. (3) The UAV is equipped with a camera and preset flight parameters to automatically acquire orthophotos and collect images at noon every day during the monitoring period to ensure data consistency. Improved water body detection and morphological operations are used to obtain the water body area corresponding to the water body boundary and the vegetation coverage, so that the calculation of the actual erosion rate is closer to the real situation and provides a reliable data basis for subsequent assessment. (4) The assessment process compares the difference between the prediction model and the actual erosion rate to determine the degree of erosion by selecting the predicted or actual erosion rate value, so as to serve as the basis for maintenance personnel to carry out corresponding verification operations and reinforcement construction. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the steps of a method and system for monitoring and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspection, as described in this invention. Detailed Implementation

[0019] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] Riverbank erosion is typically monitored using fixed video surveillance, but this method only focuses on a specific area and cannot provide a comprehensive picture of the overall erosion changes across an area. Therefore, if... Figure 1 As shown, on the one hand, the present invention provides a method for monitoring and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspection, including the following steps: S1: Divide the riverbank into segments and select the riverbank segments that need to be monitored.

[0021] In this embodiment, step S1 involves dividing the riverbank into meandering sections, bifurcation sections, straight sections, and sections with abrupt changes in width. Each section is assigned a unique number, which can be sequentially numbered according to the direction of the riverbank's extension. Each section has the same length, and the location of each section is recorded. Monitoring is conducted on riverbank sections located near towns, farmland, or areas lacking vegetation.

[0022] In river bends, the water flow is subject to centrifugal force, resulting in higher velocity and stronger erosion of the riverbanks, especially in areas where the radius of curvature does not exceed three times the river's width. At river bifurcation points, the influx of tributaries increases flow and velocity downstream, intensifying erosion of the downstream banks. In straight sections of the river, the banks are eroded uniformly. At abrupt changes in river width, the water flow is compressed or widened, causing sudden shifts in velocity and flow, particularly in sections where the narrowing exceeds 20%, resulting in strong erosion of the banks.

[0023] In addition, the segmentation was also based on nearby facilities on the riverbank. Riverbanks near towns and farmland are related to human production and living areas and have high economic value. Riverbanks lacking vegetation have poor soil and water conservation capacity and are more susceptible to erosion. Therefore, these typical areas were combined into segments for key monitoring.

[0024] S2: Obtain historical data on hydrological conditions, erosion resistance, slope, rainfall intensity, and historical vegetation coverage of the monitored segmented riverbank areas, and construct a riverbank erosion rate prediction model based on historical data.

[0025] Step S2 involves acquiring historical data on hydrological conditions, erosion resistance, slope, rainfall intensity, and historical vegetation coverage for the monitored riverbank sections over a continuous period of more than five years, and constructing the first erosion component accordingly. E t Second erosion component E m and the third erosion component E s The first erosion component of the joint system E t Second erosion component E m and the third erosion component E s A model for predicting riverbank erosion rates was obtained. E = E t + mE m + E s , mThe average number of times the non-zero output case of the second erosion component occurs within the monitoring period is used in the riverbank erosion rate prediction model. E It is a simple numerical operation.

[0026] First erosion component E t By calculating the shear stress of the water flow With critical shear stress The difference, through the index a Attenuation, based on the erosion coefficient After adjusting the proportions, we get: Among them, water flow shear stress , The density of water, It is the acceleration due to gravity. Manning's roughness coefficient For water flow velocity, The average water depth of the river profile; critical shear stress. , c For soil cohesion, This is the vegetation stability coefficient; a positive value indicates vegetation expansion, while a negative value indicates vegetation decline. It is a positive accumulated temperature. The accumulated temperature is negative, and the cumulative accumulated temperature is... , The daily average temperature Base temperature is the minimum temperature at which vegetation begins to grow. Different plant types have different base temperatures, with tropical plants having the highest base temperature and temperate plants having the lowest. Temperature index It is a non-zero real number. The first erosion component reflects the effect of water flow on the natural erosion of the bank.

[0027] Second erosion component E m Several landslide bodies are selected in the monitored segmented riverbank area. These landslide bodies are based on digital elevation models, such as those obtained using LiDAR. Each landslide body is divided into several equally sized strips, each strip perpendicular to the sliding direction. Instability judgment criteria are established based on the safety factor of each landslide body. ,in t For the current time, For the first transient safety factor Time, The residual coefficient, For the density of soil, Let V be the volume of the sliding body. The disintegration rate is given. The first scenario of the instability criterion indicates no collapse has occurred, while the second scenario indicates a collapse has occurred. The transient safety factor is calculated using the following formula: The sliding body is obtained by dividing it into sections perpendicular to the predicted sliding direction. N Each strip, The weight of a single strip, Let be the angle between the direction of gravity of the strip and the normal to the sliding direction. The angle between the bottom surface of the strip and the horizontal plane. For the angle of soil shear strength, Pore ​​water pressure, b The width of the strip; For the iteration safety factor, , The initial value of the iterative safety factor is 1. The current value after the iterative safety factor update. Let be a non-negative real number, and the step size of each iteration does not exceed 0.05. "Iteration" means that iteration continues until the iteration termination condition is met, at which point iteration stops and the iteration safety coefficient obtained from the previous update is output. For ease of calculation, in this embodiment, let be a non-negative real number. and equal.

[0028] LiDAR acquisition of landslides involves constructing a digital elevation model (DEM) by combining dry season topographic data and flood season water depth data. It observes areas of annual average topographic, slope, and surface displacement changes within the DEM from different years. Areas where the slope first changes by more than 35°, areas where the annual surface displacement due to non-human factors exceeds 2 meters, and areas where ground-penetrating radar first detects cracks are designated as landslide zones. Topographic profiles of the landslide zones are extracted from the DEM to obtain the point set corresponding to the landslide, determining the center and radius of the landslide surface. The landslide is then divided into several blocks, and the center coordinates of the bottom surfaces of these blocks are obtained by setting parameters. b , and Define the orientation of the landslide blocks. The sliding body and sliding surface are used to determine the amount of bank collapse, i.e., the second erosion component. E m The connection is obvious. Establishing digital elevation models is a standard technique in this field and will not be elaborated upon here.

[0029] The sliding surface is usually assumed to be circular, defined by determining the center and radius of the sliding circle. The sliding surface is the failure interface where the sliding body separates from the stable rock / soil layer. Connecting the center of the sliding circle with the entry and exit points of the sliding surface forms a fan-shaped region. Along the sliding direction, such as parallel to the riverbank, the sliding body is divided into... NThe blocks are of equal width. A transient safety factor is used to determine whether a collapse has occurred.

[0030] Third erosion component E s It is obtained by comprehensively considering soil erodibility factors, rainfall, slope, and vegetation index: , As a soil erodibility factor, P This represents the average annual rainfall intensity, expressed in millimeters. Therefore, divide by 1000 to convert to meters. S For slope, For vegetation coverage, , This represents the maximum vegetation coverage under ideal conditions, which can be estimated at 85%-95%. The minimum vegetation cover under ideal conditions is obtained based on actual measurements during the winter frost period. k This is the vegetation growth rate constant, which is usually a positive real number at the riverbank. The accumulated temperature required for vegetation cover to increase to 50%. Parameter A , B and C The terms are exponential and are all non-zero real numbers. The third erosion classification reflects the riverbank erosion that may result from changes in vegetation cover.

[0031] S3: Plan the flight path of the UAV in the monitored segmented riverbank area and obtain orthophotos at different monitoring times.

[0032] Step S3 involves the UAV carrying a camera, setting the monitoring period for the hydrological event, and acquiring images of the area where the landslide is located in the same segment of the riverbank during the monitoring period at noon each day, according to a given fixed flight altitude, pixel resolution, focal length, flight path overlap range, and upper limit of ambient wind speed. The images are then stitched together to obtain an orthophoto map of the monitoring period based on a time series, and the orthophoto map of the area where the landslide is located on any day before the monitoring period is used as the reference orthophoto map.

[0033] Noon was chosen for each day because the sunlight is nearly direct, minimizing the shading of vegetation. By fixing the flight altitude, flight path, resolution, focal length, and flight path overlap, the acquired images are ensured to have minimal deviation, facilitating subsequent fusion and alignment processing.

[0034] S4: Calculate the actual erosion rate of the segmented riverbank area based on orthophotos from different monitoring times.

[0035] Step S4 involves finding the same land features in orthophotos from different monitoring periods, registering them to the same world coordinate system, pre-obtaining the correspondence between pixel size and the actual size in the world coordinate system, identifying the maximum changes in shoreline location and vegetation cover in the area where the landslide body is located, and comparing them with the shoreline location and vegetation cover in the baseline orthophoto image to calculate the actual erosion rate: 1) Mark the location of the landslide body in both the baseline orthophoto image and the monitoring period orthophoto image, and conduct water body detection within the adjacent rectangle of the landslide body. The adjacent rectangle extends from the top of the bank slope to the center of the river channel. The water body detection uses the improved water index method. MNDWI The threshold for the water index is [0, 1). Then, a 3×3 kernel opening operation is performed on the water area to eliminate noise, and a 15×15 kernel closing operation is used to fill the holes. The pixel area within the water index area in the baseline orthophoto and monitoring period orthophoto are calculated and converted into the actual water area to obtain the area corresponding to the change in shoreline position; 2) The water index area in the adjacent rectangle is removed, and the normalized vegetation index is used in the remaining adjacent rectangle. NDVI Calculate the number of vegetation pixels and divide it by the total number of pixels outside the water surface to obtain the vegetation coverage in the baseline orthophoto and the monitoring period orthophoto. By comparison, obtain the change in vegetation coverage within the adjacent rectangle; 3) Actual erosion rate E f Calculate using the following formula: ,in This represents the area corresponding to the change in the shoreline's position. Mass of eroded material per unit area , For soil bulk density, The thickness of the eroded layer, Soil porosity; This represents the change in vegetation cover within the adjacent rectangular area. The amount of erosion reduced per unit of vegetation cover. , The root system coefficient is 0.6-0.8 for herbaceous plants, 0.8-1.2 for shrubs, and 1.2-1.8 for trees. For effective root depth, the effective root depth is 0.3-0.5 cm for herbaceous plants, 0.8-1.2 cm for shrubs, and 1.2-2.5 cm for trees. The monitoring period can typically be set to one year.

[0036] Improved water index method MNDWI = ( Green - MIR ) / ( Green -MIR ), Green It is in the green light band. MIR Using the mid-infrared band, this method is effective for extracting water body boundaries, especially those of eutrophic water bodies. Normalized Difference Vegetation Index (NDVI) NDVI = ( NIR - Red ) / ( NIR + Red ), Red For the red channel band, NIR This is the near-infrared channel band. Healthy vegetation reflects more near-infrared and blue light, while absorbing more red and blue light.

[0037] S5: Based on the calculated actual erosion rate and the predicted erosion rate of the riverbank erosion rate prediction model at the corresponding time, assess the degree of riverbank erosion according to the difference and value between the actual erosion rate and the predicted erosion rate.

[0038] Step S5 involves, for each sliding body, calculating the first erosion component. E t and the third erosion component E s According to the width of the sliding body L The length of the monitored segmented riverbank area L After calculating the ratio to 0, the second erosion component of the sliding body during the monitoring period is added. mE m The predicted erosion rate of each sliding body during the monitoring period is obtained. , , with the actual erosion rate of the sliding body E f To make a comparison, given the relative error 1) If the relative error does not exceed 15%, it indicates that the predicted erosion rate is within acceptable limits. Reliable, using predicted erosion rates 1) Assess the degree of erosion; 2) If the relative error is >15%, it indicates a predicted erosion rate. Unreliable; actual erosion rate should be used. E f Assess the degree of erosion; 3) If the erosion rate is <500, or the shoreline retreat speed does not exceed 0.1m / year, it is considered light erosion; if 500 < erosion rate ≤ 2000, and the shoreline retreat speed is greater than 0.1m / year but not more than 1.5m / year, it is considered moderate erosion; if the erosion rate is ≥ 2000, or the shoreline retreat speed is > 1.5m / year, it is considered severe erosion.

[0039] The shoreline retreat rate here is defined as the annual minimum distance by which the shoreline in the monitoring orthophoto map deviates from the shoreline in the baseline orthophoto map.

[0040] Because the predicted erosion rate is based on historical data and does not account for extreme hydrological events, model failures, underreporting in historical data, or unreasonable parameters during the current monitoring period, there may be a certain deviation from the measured actual erosion rate. Therefore, when the deviation is too large, the actual measured erosion rate should prevail. A segmented riverbank area may include more than one landslide body; therefore, it is necessary to convert it to the corresponding landslide body area and calculate the predicted and actual erosion rates for that landslide body area separately.

[0041] For mild erosion, natural recovery is possible; for moderate erosion, flexible protection can be used, such as using willow cuttings along the riverbank in conjunction with soil-stabilizing netting to increase soil and water conservation capacity; for severe erosion, concrete blocks need to be added to the riverbank and gabions used for reinforcement.

[0042] On the other hand, the present invention provides a riverbank erosion monitoring and assessment system based on unmanned aerial vehicle (UAV) inspection, for implementing the aforementioned riverbank erosion monitoring and assessment method based on UAV inspection, specifically including: The riverbank segmentation module is used to segment the riverbank and select the segmented riverbank area that needs to be monitored as the object to be analyzed. The erosion rate prediction model construction module is connected to the riverbank segmentation module to acquire historical data on hydrological conditions, erosion resistance, slope, rainfall intensity, and historical vegetation coverage of the monitored riverbank segments, and to construct a riverbank erosion rate prediction model based on historical data. The drone monitoring module plans the flight path of the drone in the monitored segmented riverbank area and acquires orthophotos of the camera on the drone at different monitoring times; based on the orthophotos of the different monitoring times, it calculates the actual erosion rate of the segmented riverbank area. The analysis and evaluation module is connected to the erosion rate prediction model construction module and the UAV monitoring module, respectively. It is used to evaluate the degree of riverbank erosion based on the calculated actual erosion rate and the predicted erosion rate of the riverbank erosion rate prediction model at the corresponding time, and based on the difference and value between the actual erosion rate and the predicted erosion rate.

[0043] 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 and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspection, characterized in that, Includes the following steps: S1: Divide the riverbank into segments and select the riverbank areas that need to be monitored; S2: Obtain historical data on hydrological conditions, erosion resistance, slope, rainfall intensity, and historical vegetation coverage of the monitored segmented riverbank areas, and construct a riverbank erosion rate prediction model based on historical data; S3: Plan the flight path of the UAV in the monitored segmented riverbank area and obtain orthophotos at different monitoring times; S4: Calculate the actual erosion rate of the segmented riverbank area based on orthophotos from different monitoring times; S5: Based on the calculated actual erosion rate and the predicted erosion rate of the riverbank erosion rate prediction model at the corresponding time, assess the degree of riverbank erosion according to the difference and value between the actual erosion rate and the predicted erosion rate.

2. The method for monitoring and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspection according to claim 1, characterized in that, Step S1 involves dividing the riverbank into meandering sections, bifurcation sections, straight sections, and sections with abrupt changes in width; assigning a unique segment number to each segment, ensuring each segment has the same length, and recording the location of each segment; and selecting riverbank areas near towns, farmland, or areas lacking vegetation for monitoring.

3. The method for monitoring and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspection according to claim 2, characterized in that, Step S2 involves acquiring historical data on hydrological conditions, erosion resistance, slope, rainfall intensity, and historical vegetation coverage for the monitored riverbank sections over a continuous period of more than five years, and constructing the first erosion component accordingly. E t Second erosion component E m and the third erosion component E s The first erosion component of the joint system E t Second erosion component E m and the third erosion component E s A riverbank erosion rate prediction model was obtained. E = E t + mE m + E s , m The average number of times the non-zero output case of the second erosion component occurs within the monitoring period.

4. The method for monitoring and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspection according to claim 3, characterized in that, First erosion component E t By calculating the shear stress of the water flow With critical shear stress The difference, through the index a Attenuation, based on the erosion coefficient It was obtained by adjusting the proportions.

5. A method for monitoring and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspection according to claim 4, characterized in that, Second erosion component E m The method involves selecting several sliding bodies in the monitored segmented riverbank area, establishing an instability judgment criterion based on the safety factor of each sliding body, and obtaining the second erosion component based on the output of the instability judgment criterion of the sliding body safety factor. E m .

6. A method for monitoring and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspection according to claim 5, characterized in that, Third erosion component E s It is obtained by combining soil erodibility factors, rainfall, slope and vegetation index.

7. A method for monitoring and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspection according to claim 5, characterized in that, Step S3 involves the UAV carrying a camera, setting the monitoring period for the hydrological event, and acquiring images of the area where the landslide is located in the same segment of the riverbank during the monitoring period at noon each day, according to a given fixed flight altitude, pixel resolution, focal length, flight path overlap range, and upper limit of ambient wind speed. The images are then stitched together to obtain an orthophoto map of the monitoring period based on a time series, and the orthophoto map of the area where the landslide is located on any day before the monitoring period is used as the reference orthophoto map.

8. A method for monitoring and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspection according to claim 7, characterized in that, Step S4 involves finding the same land features in orthophotos from different monitoring periods, registering them to the same world coordinate system, pre-obtaining the correspondence between pixel size and the actual size in the world coordinate system, identifying the maximum changes in shoreline location and vegetation cover in the area where the landslide body is located, and comparing them with the shoreline location and vegetation cover in the baseline orthophoto image to calculate the actual erosion rate: 1) Mark the location of the landslide body in both the baseline orthophoto image and the monitoring period orthophoto image, and conduct water body detection within the adjacent rectangle of the landslide body. The adjacent rectangle extends from the top of the bank slope to the center of the river channel. The water body detection uses the improved water index method. MNDWI The threshold for the water index is [0, 1]. Then, a 3×3 kernel opening operation is performed on the water area to eliminate noise, and a 15×15 kernel closing operation is used to fill the holes. The pixel area within the water index area in the baseline orthophoto and monitoring period orthophoto are calculated and converted into the actual water area to obtain the area corresponding to the change in shoreline position; 2) The water index area in the adjacent rectangle is removed, and the normalized vegetation index is used in the remaining adjacent rectangle. NDVI The number of vegetation pixels is calculated and divided by the total number of non-water surface pixels to obtain the vegetation coverage in the baseline orthophoto and the monitoring period orthophoto. The change in vegetation coverage within the adjacent rectangle is obtained by comparison. 3) Actual erosion rate E f This is obtained by determining the area corresponding to the change in shoreline location, the mass of erosion per unit shoreline area, the change in vegetation cover within the adjacent rectangle, and the amount of erosion corresponding to the decrease in vegetation cover per unit area.

9. A method for monitoring and assessing riverbank erosion based on unmanned aerial vehicle (UAV) inspection according to claim 8, characterized in that, Step S5 involves, for each sliding body, calculating the first erosion component. E t and the third erosion component E s According to the width of the sliding body L The length of the monitored segmented riverbank area L After calculating the ratio to 0, the second erosion component of the sliding body during the monitoring period is added. mE m The predicted erosion rate of each sliding body during the monitoring period is obtained. E *, relative to the actual erosion rate of the sliding body E f Compare the results and obtain the relative error: 1) If the relative error does not exceed 15%, it indicates that the predicted erosion rate is within acceptable limits. E *Reliable, employs predicted erosion rate E *Assess the degree of erosion; 2) If the relative error is >15%, it indicates a predicted erosion rate. E *Unreliable; actual erosion rate is used. E f Assess the degree of erosion; if the erosion rate is <500, it is considered light erosion; if 500 < erosion rate ≤ 2000, it is considered moderate erosion; if the erosion rate is >2000, it is considered severe erosion.

10. A riverbank erosion monitoring and assessment system based on unmanned aerial vehicle (UAV) inspection, used to implement the riverbank erosion monitoring and assessment method based on UAV inspection as described in any one of claims 1-9, characterized in that, include: The riverbank segmentation module is used to segment the riverbank and select the segmented riverbank area that needs to be monitored as the object to be analyzed. The erosion rate prediction model construction module is connected to the riverbank segmentation module to acquire historical data on hydrological conditions, erosion resistance, slope, rainfall intensity, and historical vegetation coverage of the monitored riverbank segments, and to construct a riverbank erosion rate prediction model based on the historical data. The drone monitoring module plans the flight path of the drone in the monitored segmented riverbank area and acquires orthophotos of the camera on the drone at different monitoring times; based on the orthophotos of the different monitoring times, it calculates the actual erosion rate of the segmented riverbank area. The analysis and evaluation module is connected to the erosion rate prediction model construction module and the UAV monitoring module, respectively. It is used to evaluate the degree of riverbank erosion based on the calculated actual erosion rate and the predicted erosion rate of the riverbank erosion rate prediction model at the corresponding time, and based on the difference and value between the actual erosion rate and the predicted erosion rate.

Images