Dam slope landslide monitoring method using unmanned aerial vehicle to carry hyperspectral camera

Abstract

The invention relates to a method for monitoring the landslide of a dam slope by using an unmanned aerial vehicle to carry a hyperspectral camera. Core equipment is a multi-rotor unmanned aerial vehicle carrying a global positioning system and the hyperspectral camera. The method is mainly used for identifying a potential landslide area by monitoring vegetation change based on a hyperspectral imaging technology, and is mainly suitable for tropical island areas where vegetation is slightly influenced by climate, such as Hainan island. The unmanned aerial vehicle flies over vegetation of a dam slope at regular intervals, hyperspectral image data are obtained, and related spectral feature changes of the vegetation and soil are analyzed. Abnormal spectral change of vegetation is often related to geological instability, so that the abnormal spectral change can be used as an index of landslide early warning. According to the method, efficient, rapid and non-contact monitoring can be carried out on the large-area slope, high precision and real-time performance are achieved, and improvement of the accuracy and response speed of landslide disaster early warning is facilitated.

Description

Technical Field

[0001] The invention relates to the technical field of geological disaster monitoring, and in particular to a dam slope landslide monitoring method using an unmanned aerial vehicle equipped with a hyperspectral camera. Background Art

[0002] As an important water conservancy project facility, dams not only play an irreplaceable role in flood control, power generation and water resource allocation, but also serve as infrastructure to ensure the stable development of regional economy and society. The stability of the dam slope is directly related to the safe operation of the dam. However, factors such as geological activities, climate change and human activities may lead to slope instability and landslides. Especially after the construction of the dam, due to long-term water erosion, leakage and other problems, the risk of slope landslides has increased significantly. In severe cases, it may cause catastrophic consequences such as dam failure, causing immeasurable losses to downstream areas. Therefore, the monitoring and early warning of slope landslides have become one of the core links in dam safety management.

[0003] Planting vegetation on the slopes of dams has many benefits and functions. First, it can prevent soil erosion. The roots of vegetation can effectively consolidate the soil, reduce the possibility of soil being washed away by rainwater, and prevent slope collapse and erosion. Secondly, it can enhance the stability of the slope. The roots of vegetation increase the stability of the slope by combining with the soil, reducing the risk of landslides or collapses. In addition, it can reduce water pollution. Vegetation can act as a natural filter to intercept and absorb silt and pollutants carried by rainwater, reduce their entry into the water body of the dam, and thus protect water quality. Finally, it can also promote ecological diversity. Planting vegetation on the slopes of dams helps maintain the local ecological balance, provide biological habitats, and increase regional biodiversity.

[0004] At present, the monitoring technology of slope landslide mainly includes ground monitoring and remote sensing monitoring. Ground monitoring methods such as displacement sensors, inclinometers, and ground stress monitoring devices rely on the deployment of sensors in the landslide area to obtain physical information on the surface and underground. Although they can provide high-precision data, they have the following shortcomings: (1) Due to the geographical environment of the monitoring area, it is difficult to deploy them extensively in large-scale, complex terrain slope areas, and the monitoring coverage is limited; (2) The installation and maintenance costs of sensors are high, and they may be damaged by environmental factors after long-term operation, resulting in incomplete data or monitoring failure; (3) The monitoring data is not real-time enough, making it difficult to respond quickly to sudden landslide events. Remote sensing monitoring technologies such as satellite remote sensing and lidar can perform macroscopic observations over a large range, but they have problems with insufficient temporal and spatial resolution, making it difficult to capture subtle changes in landslides or provide continuous high-frequency monitoring. In addition, satellite remote sensing is greatly affected by weather conditions and cannot obtain effective data when blocked by clouds.

[0005] In this context, drone technology has gradually come into people's view. With its high flexibility, rapid deployment, low cost and high-resolution observation capabilities, drones have become an ideal method for landslide monitoring in complex terrain areas. In addition, drones can provide high-precision and high-resolution ground information collection by carrying different types of sensors. Compared with traditional ground sensors, drone monitoring not only saves the complicated deployment and maintenance work, but also allows non-contact monitoring in areas that are difficult to access for landslides, greatly improving the coverage and efficiency of monitoring.

[0006] At the same time, hyperspectral imaging technology has been widely used in the field of environmental monitoring in recent years. Hyperspectral imaging is an advanced remote sensing method that combines imaging and spectral technology. It can accurately record the spectral information of objects in multiple spectral bands. It can not only obtain images in the visible light range, but also capture information in the near-infrared and short-wave infrared bands that cannot be recognized by the human eye. Hyperspectral data has rich spectral characteristics and can reflect the chemical composition and physical state of substances. Therefore, it has broad application prospects in the fields of vegetation monitoring and geological exploration. In slope landslide monitoring, hyperspectral technology can directly reflect the changes in the soil spectrum map where the plant roots are located by monitoring the trace elements in the leaves of vegetation, and indirectly reflect the precursors of landslides. In addition, if a landslide occurs, the vegetation will slide down with the landslide soil, and the soil at the bottom of the original vegetation will be exposed. The spectral characteristics of the soil and vegetation are obviously different, which can reflect the occurrence of landslides.

[0007] Based on the combination of UAV and hyperspectral imaging technology, the present invention proposes a new slope landslide monitoring method, which is suitable for areas where vegetation is less affected by climate, especially for tropical island areas such as Hainan Island. By carrying a hyperspectral camera on the UAV and regularly collecting and analyzing high-resolution data on the dam slope, it is possible to monitor vegetation changes in the slope area and geological anomalies at its roots in real time. Compared with traditional landslide monitoring methods, this method can not only cover a wide monitoring area, but also accurately identify potential landslide precursors through hyperspectral imaging. By comparing and analyzing hyperspectral data at different time periods, the stability of the slope can be evaluated in real time, providing early warning information on landslides, and ensuring the safe operation of the dam.

[0008] In summary, the use of drones equipped with hyperspectral cameras for slope landslide monitoring is an innovative technical means with high efficiency, accuracy, automation and wide coverage, which is expected to significantly improve the monitoring capabilities of dam slopes and the level of disaster warning. Summary of the invention

[0009] The present invention aims to provide a dam slope landslide monitoring method based on UAV and hyperspectral imaging technology to overcome the problems of insufficient monitoring accuracy, high cost, limited coverage and poor real-time performance in the prior art. The method can efficiently and accurately obtain vegetation changes and soil anomaly information in the dam slope area, and through early identification and analysis of landslide precursors, it can realize real-time monitoring and early warning of landslide risks, which helps to improve the safety management level of the dam.

[0010] The technical solution adopted by the present invention is as follows:

[0011] A method for monitoring dam slope landslides using a drone equipped with a hyperspectral camera uses drones as data collection platforms to inspect the dam slope area regularly or on demand. The flight path and frequency of the drone can be adjusted according to the actual terrain of the dam and the landslide risk level. Compared with traditional ground monitoring equipment, drones can cover large areas with complex terrain, and are particularly suitable for dangerous areas that are difficult for traditional monitoring equipment to reach.

[0012] Hyperspectral cameras capture the reflectance spectra of vegetation in hundreds of bands, covering different spectral regions such as visible light, near infrared and short-wave infrared. The health of plants will show specific spectral characteristics in different bands. For example, healthy vegetation absorbs more light in the red band (600-700nm) and has a higher reflectivity in the near-infrared band (700-1000nm). Through this difference, the hyperspectral camera can record the reflectance spectrum curve of vegetation in each band. The Vegetation Index is calculated using hyperspectral data, and the reflectance of a specific band is used to measure the health of vegetation.

[0013] The following are several commonly used vegetation indices and their calculation methods:

[0014] 1) Normalized Difference Vegetation Index (NDVI):

[0015] NDVI is one of the most commonly used vegetation indices and is used to measure the density and health of vegetation. Healthy vegetation will generally have higher NDVI values. NDVI values ​​range from -1 to 1. High NDVI values ​​(close to 1) indicate a high density of healthy vegetation. Low NDVI values ​​(close to 0 or negative values) indicate sparse vegetation, soil, or water.

[0016] Calculation formula: NDVI = (NIR-Red) / (NIR+Red)

[0017] Among them: NIR: reflectivity in the near-infrared band (700-1000nm);

[0018] Red: reflectivity in the red light band (600-700nm);

[0019] 2) Enhanced Vegetation Index (EVI)

[0020] EVI adds corrections for the atmosphere and surface background on the basis of NDVI, and is suitable for use in areas with high vegetation coverage or strong atmospheric interference. EVI can better suppress the influence of soil and atmosphere, and is more effective in assessing vegetation conditions in areas with high vegetation coverage.

[0021] Calculation formula: EVI = 2.5 × (NIR-Red) / (NIR + C1 × Red-C2 × Blue + L)

[0022] Among them: NIR, Red, and Blue represent near-infrared reflectivity, red light reflectivity, and blue light reflectivity, respectively.

[0023] C1=6 and C2=7.5: atmospheric correction coefficients.

[0024] L=1: Parameter used to correct soil background.

[0025] 3) Ratio Vegetation Index (RVI)

[0026] RVI is the simplest vegetation index, which directly calculates the ratio of near-infrared and red light bands. The larger the RVI value, the denser the vegetation. RVI can quickly determine vegetation density, but it is easily affected by light and soil.

[0027] Calculation formula: RVI = NIR / Red

[0028] 4) Soil Adjusted Vegetation Index (SAVI)

[0029] SAVI adds a soil correction factor to NDVI, which is suitable for areas with sparse vegetation and a large impact of bare soil. By introducing L to correct the impact of bare soil, SAVI can more accurately reflect vegetation coverage.

[0030] Calculation formula: SAVI = ((1+L)×(NIR-Red)) / (NIR+Red+L)

[0031] Where: L: soil correction factor, usually 0.5.

[0032] Compare hyperspectral data from different periods to identify the spectral change trend of vegetation over time. By regularly collecting hyperspectral data from the same location, construct a time series and observe the fluctuation of vegetation index. If the NDVI or other vegetation index in a certain area continues to decline, it may indicate that the vegetation in the area is suffering from soil erosion, thus inferring the possibility of landslide on the dam slope.

[0033] The present invention has the following beneficial effects:

[0034] 1. The present invention uses a drone equipped with a hyperspectral camera to efficiently and accurately monitor the dam slope, which can detect landslide precursors in a timely manner and provide early warning. This highly sensitive monitoring method can significantly reduce the risk of disasters caused by landslides, avoid sudden disasters caused by dam landslides, and ensure the safety of life and property of people in the dam and downstream areas.

[0035] 2. Compared with traditional landslide monitoring methods, the present invention has lower operating costs. UAV inspection does not require the laying of complex sensor networks, which can reduce manpower, material and time costs. In addition, through early warning of landslides, economic losses caused by landslide disasters can be avoided, the maintenance and repair costs of dams can be reduced, and the service life of dams can be extended.

[0036] 3. Hyperspectral technology can indirectly reflect landslide precursors by monitoring vegetation conditions, avoiding direct damage to the environment. Compared with traditional sensor deployment, the method of the present invention will not interfere with the natural ecological environment of the dam slope, which helps to protect the local ecosystem.

[0037] 4. The present invention combines UAV and hyperspectral imaging technology, can flexibly cope with complex terrain conditions, and achieve high-precision monitoring of large areas. It provides a new monitoring method for indirect warning of landslides, which has strong technical innovation. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 Schematic diagram of a multi-rotor drone equipped with a hyperspectral camera;

[0039] Figure 2 Schematic diagram of the hyperspectral principle;

[0040] Figure 3 Schematic diagram of UAV monitoring of slope vegetation;

[0041] Figure 4 Schematic diagram of abnormal slope vegetation detected by drone.

[0042] In the figure: 1 Hyperspectral camera; 2 Global Positioning System (GPS); 3 Spectrum of plant leaves;

[0043] 4 soil spectrum; 5 multi-rotor drone; 6 vegetation; 7 dam; 8 vegetation on slope; 9 landslide area;

[0044] 10Sliding soil. DETAILED DESCRIPTION

[0045] The present invention provides a dam slope landslide monitoring method using an unmanned aerial vehicle equipped with a hyperspectral camera, which is suitable for tropical island areas where vegetation is less affected by climate. The specific implementation plan is as follows:

[0046] 1. UAV platform construction

[0047] This implementation uses a multi-rotor drone equipped with a high-resolution hyperspectral camera, such as Figure 1 . The UAV is equipped with a GPS navigation system 2 and a hyperspectral image acquisition module 1, and is capable of autonomous flight and real-time data transmission. Among them, a UAV with light weight and high flight stability is selected to ensure that it can adapt to complex slope terrain and has a certain wind resistance. The hyperspectral camera selects a hyperspectral camera with high spatial resolution and spectral resolution. The band range should cover visible light to near infrared (400-1000nm) so that it can collect tiny spectral changes in vegetation and soil. Use UAV flight control to plan the route to ensure that all key monitoring areas of the dam slope are covered. Reasonably set the flight altitude, speed and overlap rate (generally 70-80%) to ensure that the acquired image data is complete and there are no missed shots.

[0048] 2. Extract spectral features regularly

[0049] like Figure 3 During the flight, the UAV uses a hyperspectral camera to regularly collect hyperspectral image data of the slope vegetation 6 in the target area. The UAV is flown at a scheduled time to collect slope image data. To improve the timeliness and data quality of monitoring, the flight is usually carried out in the morning or afternoon when the light is uniform to reduce the impact of light changes. Data is collected regularly to construct a time series. It is recommended to obtain it once a month or quarter for time series analysis.

[0050] 3. Spectral analysis

[0051] Perform radiation correction, geometric correction and atmospheric correction on the acquired hyperspectral data. Ensure that the data collected each time has consistent spectral characteristics and spatial resolution, laying the foundation for subsequent analysis. Use calibration plates or standard spectral data to correct spectral information and reduce errors caused by factors such as light intensity and sensors. Perform geometric correction of images based on GPS and attitude data to ensure accurate spatial alignment of images at different times. Ensure the reliability of spectral data by removing the effects of water vapor, aerosols, etc. in the atmosphere on the spectrum.

[0052] Calculate the spectral index of each image data, such as the Normalized Difference Vegetation Index (NDVI), Water Index (NDWI), Soil Adjusted Vegetation Index (SAVI), etc.

[0053] NDVI: It is used to assess the health of vegetation. Healthy vegetation has a higher NDVI value. Vegetation stress or death may occur before a landslide, resulting in a decrease in NDVI.

[0054] NDWI: Used to monitor the moisture content of the slope area. Increased soil moisture will lead to an increased risk of landslides.

[0055] SAVI: Suitable for use in areas with sparse vegetation and can more accurately reflect the status of soil and vegetation.

[0056] By comparing data from different periods, the changing trends of key spectral indices can be monitored. If the NDVI value in a certain area decreases and the moisture index increases, it may indicate that vegetation is dead or soil moisture has increased, indicating that there is a potential risk of landslides in the area.

[0057] This inspection method uses a drone equipped with a hyperspectral camera to reflect the changes in the spectral graph of slope vegetation. It can predict the possibility of slope landslides to a certain extent and can also identify areas where landslides have already occurred, providing a method to ensure early warning and identification of dam slope landslides.

Claims

1. A method for monitoring dam slope landslide using an unmanned aerial vehicle equipped with a hyperspectral camera, characterized in that: Drones are used as data collection platforms, taking advantage of their high maneuverability and low-altitude flight capabilities to inspect the dam slope area regularly or on demand; the flight path and frequency of the drones can be adjusted according to the actual terrain of the dam and the level of landslide risk; hyperspectral cameras capture the reflectance spectrum of vegetation in hundreds of bands, which cover different spectral regions such as visible light, near-infrared and short-wave infrared; the health of plants will show specific spectral characteristics in different bands; through this difference, the hyperspectral camera records the reflectance spectral curve of vegetation in each band; hyperspectral data is used to calculate the vegetation index, and the reflectance of a specific band is calculated to measure the health of the vegetation.

2. The method for monitoring dam slope landslide using an unmanned aerial vehicle equipped with a hyperspectral camera according to claim 1, further comprising: The following commonly used vegetation indices and their calculation methods are used: 1) Normalized Difference Vegetation Index NDVI: NDVI is one of the most commonly used vegetation indices and is used to measure the density and health of vegetation. Healthy vegetation will typically have higher NDVI values. NDVI values ​​range from -1 to 1. High NDVI values ​​close to 1 indicate high density, healthy vegetation. Low NDVI values ​​close to 0 or negative values ​​indicate sparse vegetation, soil or water. Calculation formula: NDVI = (NIR-Red) / (NIR+Red) Among them: NIR: reflectivity in the near-infrared band (700-1000nm); Red: reflectivity in the red light band (600-700nm); 2) Enhanced Vegetation Index (EVI) EVI adds corrections for the atmosphere and surface background on the basis of NDVI, and is suitable for use in areas with high vegetation coverage or strong atmospheric interference. EVI can better suppress the influence of soil and atmosphere, and is more effective in assessing vegetation conditions in areas with high vegetation coverage. Calculation formula: EVI = 2.5 × (NIR-Red) / (NIR + C1 × Red-C2 × Blue + L) Where: NIR, Red, Blue represent near-infrared reflectivity, red light reflectivity, and blue light reflectivity, respectively; C1=6 and C2=7.5: atmospheric correction coefficients; L = 1: Parameter used to correct soil background; 3) Ratio Vegetation Index (RVI) RVI is the simplest vegetation index, which directly calculates the ratio of near infrared and red light bands. The larger the RVI value, the denser the vegetation. RVI can quickly determine vegetation density, but it is easily affected by light and soil. Calculation formula: RVI = NIR / Red 4) Soil Adjusted Vegetation Index (SAVI) SAVI adds a soil correction factor to NDVI, which is suitable for areas with sparse vegetation and a large impact of bare soil. By introducing L to correct the impact of bare soil, SAVI more accurately reflects vegetation coverage. Calculation formula: SAVI = ((1+L)×(NIR-Red)) / (NIR+Red+L) Where: L: soil correction factor, usually 0.5; By comparing the hyperspectral data of different periods, the spectral change trend of vegetation over time can be identified; by regularly collecting hyperspectral data at the same location, a time series is constructed to observe the fluctuation of vegetation index; if the NDVI or other vegetation index in a certain area continues to decline, it may indicate that the vegetation in the area is suffering from soil erosion, thus inferring the possibility of landslide on the dam slope.

Images