Geological disaster deformation monitoring method based on remote sensing data

Abstract

The invention discloses a geological disaster deformation monitoring method based on remote sensing data, and relates to the technical field of environment monitoring, and the method comprises the following steps: S1, data source selection and preprocessing; s2, region feature extraction; s3, differentiating the multi-temporal remote sensing images; s4, analyzing a time sequence; s5, performing three-dimensional reconstruction on the deformation area; s6, settlement and landslide monitoring; s7, performing fusion analysis on environment and meteorological data; s8, performing data anomaly detection and automatic early warning; and S9, risk assessment and influence analysis. According to the geological disaster deformation monitoring method based on the remote sensing data, SAR, optical images and LiDAR data are combined, the defect of dependence of a single data source in the prior art is overcome, dynamic data fusion and time sequence analysis are used, multi-dimensional real-time monitoring of earth surface deformation is achieved, the monitoring precision and timeliness are improved, and the geological disaster deformation monitoring method based on the remote sensing data has good application prospects. More comprehensive geological disaster early warning can be provided, and the effectiveness of disaster prevention and control is greatly enhanced.

Description

Technical Field

[0001] The invention relates to the technical field of environmental monitoring, and in particular to a geological disaster deformation monitoring method based on remote sensing data. Background Art

[0002] Geological disaster deformation monitoring technology based on remote sensing data has been widely used in recent years. Remote sensing technologies such as synthetic aperture radar (SAR), optical imaging and laser radar (LiDAR) can obtain surface deformation information over a large area and in a non-contact manner. These technologies have shown significant advantages in monitoring geological disasters such as landslides, surface subsidence, and debris flows, especially in inaccessible mountainous areas and dangerous areas. However, most existing technologies rely on a single data source and cannot fully integrate the advantages of multiple remote sensing data. For example, SAR technology has high accuracy in monitoring large-scale deformation, but its temporal resolution is low; optical imaging provides rich surface detail information, but is sensitive to weather conditions. The application of a single technology has limited monitoring accuracy and timeliness, especially under changing climate conditions and complex terrain, it is difficult to capture the dynamic changes of surface deformation in a timely and accurate manner.

[0003] In order to solve the above problems, we proposed a geological disaster deformation monitoring method based on remote sensing data. Summary of the invention

[0004] In view of the deficiencies of the prior art, the present invention provides a geological disaster deformation monitoring method based on remote sensing data to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solution: a method for monitoring geological disaster deformation based on remote sensing data, comprising the following steps: S1. Data source selection and preprocessing Select appropriate remote sensing data sources, including synthetic aperture radar (SAR), optical satellite imagery, and laser radar (LiDAR); The preprocessing steps include radiometric correction, geometric correction, and atmospheric correction to ensure spatial alignment and consistency between different data sources; S2. Regional feature extraction Use digital elevation models (DEM) or terrain data to extract topographic features of geohazard risk areas, including slope, rainfall intensity, and surface deformation rate, to identify high-risk areas where deformation may occur; After feature extraction, the risk areas are classified and the key areas for subsequent monitoring are determined; S3, Multi-temporal remote sensing image difference Extracting surface deformation information from multi-temporal remote sensing images of the same area; Using the Differential Interferometric Synthetic Aperture Radar (DInSAR) technique, the surface displacement changes between different periods were calculated; Use optical images to monitor large-scale surface changes and identify precursors of large-scale disasters such as landslides and mud-rock flows; S4. Time Series Analysis Time series analysis based on dynamic model-driven methods, combining machine learning and physical models, updating and adjusting models in real time to reflect the impact of environmental factors; Capture sudden changes in geological hazards, such as changes in surface stress after abnormal rainfall or earthquakes, by dynamically adjusting trend analysis of time series; S5. 3D reconstruction of the deformed area Based on multi-source data, three-dimensional reconstruction is performed to build an accurate model of the geological disaster deformation area; through the fusion of LiDAR, optical imaging and high-resolution SAR data, a three-dimensional model of the surface is generated to accurately show the spatial distribution of surface deformation; S6. Subsidence and landslide monitoring By combining surface settlement monitoring technology and remote sensing data, deformation monitoring of settlement and landslide is carried out; S7. Fusion analysis of environmental and meteorological data Fuse remote sensing data with meteorological data to analyze the impact of meteorological changes on surface deformation; Meteorological data is obtained through satellite observation or ground stations, and the relationship between data and surface deformation is analyzed to predict the possibility of future disasters; S8. Data anomaly detection and automated warning Use machine learning algorithms to analyze multi-temporal remote sensing data to identify potential abnormal changes and issue early warnings; S9. Risk Assessment and Impact Analysis A comprehensive risk assessment is conducted through deformation monitoring results combined with data on regional population density and infrastructure distribution.

[0006] To further optimize the technical solution, in step S2, a comprehensive feature scoring function is constructed to integrate three key feature parameters: slope, rainfall intensity and surface deformation rate. The function is as follows;

[0007] in, : Geographic location The comprehensive risk score represents the potential risk of geological hazards in the area; : The slope of the location, extracted from the DEM data, reflects the terrain undulation and landslide possibility; : The rainfall intensity in the area refers to the cumulative rainfall in a certain period of time obtained through meteorological data. Rainfall is an important factor in inducing disasters such as landslides; : Surface deformation rate, the surface displacement rate calculated based on DInSAR remote sensing technology, reflects the dynamic changes of the surface; : Weight coefficient, which reflects the contribution of each characteristic parameter to the risk of geological hazards and is determined through historical data training or expert knowledge; : Offset, used to adjust the final score value; An adjustment function is set to deal with disaster characteristics in different regions. It is assumed that different regional characteristics are weighted and optimized, and nonlinear terms are introduced to make the model more adaptable:

[0008] in, : A nonlinear adjustment factor for slope, used to enhance the contribution of steep terrain in disaster risk; : Use logarithmic function to smooth the impact of rainfall intensity to avoid excessive impact of extreme rainfall values ​​on the model; : Square root processing of deformation rate is used to smooth the influence of small-scale high deformation areas on the overall model.

[0009] To further optimize the technical solution, in step S4, the dynamic model driving method includes the following process: Integration of physical models and machine learning: Combining physical models, such as surface stress models and settlement models, with machine learning techniques. Physical models provide basic mechanism understanding, while machine learning handles complex, nonlinear environmental changes. Dynamic updates in time series: The model is continuously updated based on new data. When the system environment changes, such as a heavy rain or earthquake, the model automatically adjusts to accurately predict potential geological disasters; Multi-source data fusion: The dynamic model uses data from multiple data sources simultaneously, such as surface deformation obtained by synthetic aperture radar (SAR), rainfall intensity in meteorological data, and earthquake data. Multi-dimensional data is fused and weighted by the model, and weights are dynamically assigned according to the characteristics of each data source, optimizing the prediction at each moment. Adaptive adjustment mechanism: The dynamic model handles anomalies and mutations in the data through an adaptive adjustment mechanism. When the system detects sudden changes such as those caused by earthquakes, the model automatically adjusts its parameters to enhance sensitivity to sudden events.

[0010] To further optimize the technical solution, in step S5, it is assumed that the geographic coordinates of a point in the three-dimensional space are , using deformation parameters that fuse multi-source data Representing the time series deformation of this point, the exact model is as follows:

[0011] in, The initial three-dimensional coordinates of a point on the surface are high-precision three-dimensional point cloud data obtained based on LiDAR data; is the three-dimensional coordinate of the point after deformation, that is, the exact position after deformation; For time The surface deformation data of the point is obtained, and the surface displacement information of different time periods is obtained through interference analysis; It is the pixel value of the point in the optical image, reflecting the color and texture information of the surface; For elevation model data, it is based on the fusion of LiDAR and digital elevation model (DEM); It is a function used to fuse multi-source data, supplement and optimize the 3D model details.

[0012] To further optimize the technical solution, the precise model, when used, includes the following process: Data fusion and processing: Get each surface point from LiDAR data The initial coordinates of the point at different times are calculated by combining SAR data The deformation under At the same time, optical imaging Provide surface detail information; Deformation compensation: based on surface deformation data , calculate the displacement of the point in the time series, for each moment , the deformation and the original coordinates Superimpose to obtain the deformed coordinates ; 3D reconstruction: based on the 3D coordinates of all surface points , three-dimensional reconstruction of the entire deformed area is performed through interpolation algorithm or surface fitting algorithm; Fusion of multi-source data: introduction of optical imaging and elevation model , showing the true characteristics of the surface; AI algorithm optimization: Use deep learning models such as CNN or generative adversarial networks to supplement and correct model details.

[0013] To further optimize the technical solution, in step S6, during the monitoring process, the minute settlement and landslide precursors are combined with the remote sensing observation results through algorithm deduction and data fusion for in-depth analysis; It also identifies landslide precursors based on machine learning algorithms, combines surface deformation information with meteorological data and geological structure data, and predicts the possibility of future landslides.

[0014] To further optimize the technical solution, in step S7, the possibility of geological disasters is assumed to be It means that depending on multi-dimensional factors in different time periods, the prediction model is as follows:

[0015] in, For geological disasters in time The risk value of occurrence under For time The accumulated rainfall under is the surface deformation rate; is a geological structural parameter, depending on the coordinates The geological conditions under For time Temperature data below; is the impact factor of historical disaster data; To adjust the parameters.

[0016] To further optimize the technical solution, the prediction model, when used, includes the following process: Multi-source data input and processing: Collect required input data from different sensors and data sources; Risk calculation: The model combines all input data and calculates the The risk of geological disasters ; Adjustment parameter optimization: Use machine learning algorithms to adjust the parameters Optimize, train the model through historical data, and optimize the weights so that the model can accurately predict the time and scale of disasters; Output and interpretation of prediction results: Output the risk value of geological hazards at a certain moment ,if Exceeding the set risk threshold indicates that the area faces a high risk of geological disasters in the future, issuing a warning signal.

[0017] To further optimize the technical solution, in step S8, a convolutional neural network (CNN) or a recurrent neural network (RNN) is used to perform automatic detection in spatial and temporal dimensions; Identify sudden changes and accelerations in surface deformation and generate automated early warning signals to indicate when a disaster may occur.

[0018] To further optimize the technical solution, the method also dynamically updates the deformation data of the monitored area, combines the automated data update and model retraining mechanism, ensures adaptation to environmental changes, and provides continuous disaster monitoring services.

[0019] Compared with the prior art, the present invention provides a method for monitoring geological disaster deformation based on remote sensing data, which has the following beneficial effects: This geological disaster deformation monitoring method based on remote sensing data overcomes the shortcomings of existing technologies that rely on a single data source by combining SAR, optical imaging and LiDAR data. It uses dynamic data fusion and time series analysis to achieve multi-dimensional real-time monitoring of surface deformation. This method not only improves the accuracy and timeliness of monitoring, but also provides more comprehensive geological disaster warnings, greatly enhancing the effectiveness of disaster prevention and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A schematic diagram of a flow chart of a geological disaster deformation monitoring method based on remote sensing data proposed by the present invention; Figure 2 A schematic flow chart of step S1 in a geological disaster deformation monitoring method based on remote sensing data proposed by the present invention; Figure 3 A schematic diagram of a flow chart of a comprehensive feature scoring function in a geological disaster deformation monitoring method based on remote sensing data proposed by the present invention; Figure 4 A schematic flow chart of a dynamic model driving method in a geological disaster deformation monitoring method based on remote sensing data proposed by the present invention; Figure 5 A schematic flow chart of an accurate model in a geological disaster deformation monitoring method based on remote sensing data proposed by the present invention; Figure 6 The present invention provides a flow chart of a prediction model in a method for monitoring deformation of geological disasters based on remote sensing data. DETAILED DESCRIPTION

[0021] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Example

[0022] See also Figure 1 , a geological disaster deformation monitoring method based on remote sensing data, comprising the following steps: S1. Data source selection and preprocessing like Figure 2 As shown, select appropriate remote sensing data sources, including synthetic aperture radar (SAR), optical satellite imagery, and laser radar (LiDAR).

[0023] Preprocessing steps include radiometric correction, geometric correction, and atmospheric correction to ensure spatial alignment and consistency between different data sources.

[0024] In this embodiment, based on the preprocessing, the accuracy of the subsequent processing can be guaranteed and the interference of environmental noise on deformation monitoring can be reduced.

[0025] S2. Regional feature extraction Digital elevation models (DEMs) or terrain data are used to extract terrain characteristics of geohazard risk areas, including slope, rainfall intensity, and surface deformation rate, to identify high-risk areas where deformation may occur.

[0026] After feature extraction, the risk areas are classified and the key areas for subsequent monitoring are determined.

[0027] like Figure 3 As shown, in this embodiment, a comprehensive feature scoring function is constructed to integrate three key feature parameters: slope, rainfall intensity and surface deformation rate. The function is as follows;

[0028] in, : Geographic location The comprehensive risk score represents the potential risk of geological hazards in the area.

[0029] : The slope at the location, which can usually be extracted from DEM data, reflects the terrain relief and landslide potential.

[0030] : The rainfall intensity in the area refers to the cumulative rainfall in a certain period of time obtained through meteorological data. Rainfall is an important factor in inducing disasters such as landslides.

[0031] : Surface deformation rate, the surface displacement rate calculated based on remote sensing technologies such as DInSAR, reflects the dynamic changes of the surface.

[0032] : Weight coefficient, reflecting the contribution of each characteristic parameter to the risk of geological hazards. Determined by historical data training or expert knowledge.

[0033] : Offset, used to adjust the final score value to ensure that the value range is reasonable.

[0034] An adjustment function is set to deal with disaster characteristics in different regions. It is assumed that different regional characteristics are weighted and optimized, and nonlinear terms are introduced to make the model more adaptable:

[0035] in, : A nonlinear adjustment factor for slope that is used to enhance the contribution of steep terrain in disaster risk.

[0036] : Use a logarithmic function to smooth the impact of rainfall intensity to avoid excessive impact of extreme rainfall values ​​on the model.

[0037] : Square root processing of deformation rate is used to smooth the influence of small-scale high deformation areas on the overall model.

[0038] Based on the above model, the following functions are included: slope( ) extraction and application: Slope is calculated from DEM data, which is the ratio of elevation change to horizontal distance. Areas with large slopes, especially those exceeding 30°, are usually high-risk areas for geological disasters. Due to the nonlinear processing, the risk score of the steep slope areas will be amplified, focusing on marking these high slope areas.

[0039] Rainfall intensity Analysis: Rainfall is an important factor inducing geological disasters, especially under continuous rainfall conditions, the risk of geological disasters increases sharply. The use of logarithmic smoothing can prevent extreme rainfall events from dominating the model, thereby avoiding misjudgment of low-risk areas.

[0040] Surface deformation rate Role: Based on remote sensing data, it measures small deformations of the ground surface, such as ground subsidence or displacement. , appropriate adjustments are made to the high-frequency small-range displacements so that the deformation rate does not excessively interfere with the overall model.

[0041] In summary, the above model, when used, includes: Model initialization: Determine the weight parameters by reviewing and analyzing historical geological disaster data And nonlinear adjustment parameters , and adjusted according to different geological conditions. These parameters can be tuned by machine learning methods such as multivariate regression or Bayesian optimization.

[0042] Regional high risk detection: Apply the model to specific locations in an area Scoring is performed to generate a geological hazard risk map for the region. Areas with higher scoring values ​​can be used as key monitoring targets, and further monitoring can be carried out in combination with time series analysis and 3D modeling.

[0043] Dynamic update and optimization: With the real-time update of data such as rainfall and surface deformation, model parameters and outputs can also be adjusted dynamically. The introduction of adaptive algorithms enables the model to adjust weights according to new data, maintaining high accuracy and real-time performance.

[0044] Through this model, geological disaster monitoring can better evaluate the deformation characteristics under complex terrain conditions and flexibly adapt to different disaster scenarios, providing efficient and accurate disaster warning and assessment tools.

[0045] S3, Multi-temporal remote sensing image difference Extracting surface deformation information from multi-temporal remote sensing images of the same area.

[0046] Using the Differential Interferometric Synthetic Aperture Radar (DInSAR) technology, the surface displacement changes between different periods are calculated. It can accurately capture the millimeter-level surface displacement and provide early warning for potential geological disasters.

[0047] Optical images are used to monitor large-scale surface changes and identify precursors of large-scale disasters such as landslides and mud-rock flows.

[0048] S4. Time Series Analysis Time series analysis is performed based on a dynamic model-driven approach, combining machine learning and physical models to update and adjust the model in real time to reflect the impact of environmental factors.

[0049] By dynamically adjusting the trend analysis of time series, sudden changes in geological hazards, such as changes in surface stress after abnormal rainfall or earthquakes, can be captured.

[0050] like Figure 4 As shown, in this embodiment, the dynamic model driving method includes the following process: Integration of physical models and machine learning: Combining physical models, such as surface stress models and settlement models, with machine learning techniques. Physical models provide basic mechanism understanding, while machine learning handles complex, nonlinear environmental changes. Dynamic updates in time series: The model is continuously updated based on new data. When the system environment changes, such as a heavy rain or earthquake, the model automatically adjusts to accurately predict potential geological disasters; Multi-source data fusion: The dynamic model uses data from multiple data sources simultaneously, such as surface deformation obtained by synthetic aperture radar (SAR), rainfall intensity in meteorological data, and earthquake data. Multi-dimensional data is fused and weighted by the model, and weights are dynamically assigned according to the characteristics of each data source, optimizing the prediction at each moment. Adaptive adjustment mechanism: The dynamic model handles anomalies and mutations in the data through an adaptive adjustment mechanism. When the system detects sudden changes such as those caused by earthquakes, the model automatically adjusts its parameters to enhance sensitivity to sudden events.

[0051] Application scenarios of dynamic model-driven methods include: Geological disaster deformation monitoring: In geological disaster monitoring, multi-source data can be combined in real time to predict surface deformation and geological stress changes. For example, in landslide monitoring, the prediction of the time and location of landslides can be dynamically updated through real-time rainfall and terrain data, thereby providing more timely warnings.

[0052] Flood and debris flow prediction: In flood and debris flow monitoring, water level changes and flood expansion paths can be predicted in the next few hours based on real-time data such as rainfall and river water levels. The prediction of water flow can be adjusted in real time according to current environmental conditions (such as rainfall and geological morphology), thereby improving the accuracy of the prediction.

[0053] Surface subsidence monitoring: In urban or industrial areas, dynamic model-driven methods can be used to monitor changes in surface subsidence. Combining LiDAR data, meteorological data, groundwater level and other information, it is possible to predict the compaction and deformation trends of underground soil, helping decision makers take timely measures to prevent damage to buildings or infrastructure.

[0054] The dynamic model-driven approach can handle complex environmental changes more flexibly and accurately, providing real-time prediction and early warning. This approach is particularly suitable for application scenarios that require real-time data feedback and respond to sudden events, such as monitoring of disasters such as landslides and floods.

[0055] S5. 3D reconstruction of the deformed area Based on multi-source data, 3D reconstruction is performed to build an accurate model of the geological disaster deformation area. Through the fusion of LiDAR, optical imaging and high-resolution SAR data, a 3D model of the surface is generated to accurately show the spatial distribution of surface deformation.

[0056] like Figure 5 As shown, in this embodiment, it is assumed that the geographic coordinates of a point in the three-dimensional space are , using deformation parameters that fuse multi-source data Representing the time series deformation of this point, the exact model is as follows:

[0057] in, : The initial three-dimensional coordinates of a point on the surface, based on high-precision three-dimensional point cloud data obtained from LiDAR data.

[0058] : The three-dimensional coordinates of the point after deformation, that is, the exact position after deformation, taking into account the time changes.

[0059] :time The surface deformation data of this point mainly comes from SAR technology, which obtains surface displacement information in different time periods through interference analysis.

[0060] : The pixel value of the point in the optical image reflects the color, texture and other information of the surface and is used to improve the visual accuracy of the three-dimensional model.

[0061] : Elevation model data, based on the fusion of LiDAR and digital elevation model (DEM), is used to accurately represent the surface height and morphology.

[0062] : A function used to fuse multi-source data and use machine learning or deep learning algorithms (such as convolutional neural networks (CNN) and random forests (RF)) to supplement and optimize three-dimensional model details.

[0063] The precise model, when used, includes the following process: Data fusion and processing: Get each surface point from LiDAR data The initial coordinates of the point at different times are calculated by combining SAR data The deformation under At the same time, optical imaging Provides surface detail information, such as texture and color, which can be used to enhance visual effects.

[0064] Deformation compensation: Based on the surface deformation data , we can calculate the displacement of the point in the time series. , we compare the deformation with the original coordinates Superimpose to obtain the deformed coordinates This process ensures that the model can reflect the dynamic changes of the surface, especially for problems such as landslides and surface subsidence.

[0065] 3D Reconstruction: Based on the 3D coordinates of all surface points , the entire deformed area is reconstructed in three dimensions through interpolation algorithms or surface fitting algorithms (such as the triangular mesh Delaunay algorithm). This step can generate a continuous surface model that reflects the surface changes in the entire disaster area.

[0066] Fusion of multi-source data: In the process of building a 3D model, by introducing optical imaging and elevation model , which can more accurately represent the true features of the surface. Optical images provide visual details for the model, while elevation data corrects the three-dimensional shape deviation caused by data errors.

[0067] AI algorithm optimization: Use deep learning models such as CNN or generative adversarial networks (GAN) to supplement and correct model details. For example, when regional data is sparse, the model can infer and complete the missing surface data through AI algorithms to ensure the integrity and accuracy of the model.

[0068] By combining LiDAR, optical imaging and SAR data, a more accurate and complete three-dimensional surface model is provided. Unlike a single data source, the fused model not only has high spatial resolution, but also has detailed information such as surface texture and deformation trend.

[0069] AI algorithms are used to supplement missing data and enhance model details. By learning historical data, the model can adaptively correct errors in the area, ensuring high-precision three-dimensional reconstruction under limited data conditions.

[0070] S6. Subsidence and landslide monitoring By combining surface settlement monitoring technology and remote sensing data, deformation monitoring of settlement and landslide is carried out.

[0071] In this embodiment, during the monitoring process, the minute settlement and landslide precursors are combined with the remote sensing observation results through algorithm deduction and data fusion for in-depth analysis; It also identifies landslide precursors based on machine learning algorithms, combines surface deformation information with meteorological data and geological structure data, and predicts the possibility of future landslides.

[0072] S7. Fusion analysis of environmental and meteorological data The remote sensing data is integrated with the meteorological data to analyze the impact of meteorological changes on surface deformation.

[0073] Meteorological data is obtained through satellite observations or ground stations, and the possibility of future disasters is predicted by analyzing the relationship between the data and surface deformation.

[0074] like Figure 6 As shown, in this embodiment, in step S7, it is assumed that the possibility of geological disasters is It means that depending on multi-dimensional factors in different time periods, the prediction model is as follows:

[0075] in, :Geological disasters in time The closer to 1, the greater the possibility of occurrence.

[0076] :time The cumulative rainfall under the Earth's surface, in millimeters, is used to reflect the impact of rainfall on disasters.

[0077] : The rate of surface deformation, in millimeters per year, is measured based on SAR data or surface sensors to capture the trend of surface changes.

[0078] : geological structural parameters, depending on the coordinates ( ) including rock density, soil composition, etc.

[0079] :time The temperature data below, in degrees Celsius, is used to consider the impact of temperature on surface stability.

[0080] :The impact factor of historical disaster data is based on the time series analysis results of similar disasters that have occurred in history and is used to represent the impact of historical events.

[0081] : Adjustment parameters are used to adjust the weights of different factors to ensure that the model can more accurately reflect the actual situation. Training and optimization can be performed based on specific data.

[0082] When the prediction model is used, it includes the following process: Multi-source data input and processing: The required input data will be collected from different sensors and data sources.

[0083] Risk Calculation: The model combines all input data and calculates The risk of geological disasters The core idea of ​​this formula is to evaluate the probability of disaster occurrence by weighted sum of different factors. The contribution of each factor to disaster risk is determined by the adjustment parameter Assume that a certain area has had a lot of rainfall recently and the surface deformation is more significant (i.e. and If the risk is large, the model will increase the weights of these parameters, thereby increasing the overall risk value.

[0084] Adjustment parameter optimization: Use machine learning algorithms to Optimize. By training the model with historical data, optimizing these weights allows the model to accurately predict the time and scale of disasters. For example, by inputting historical landslide events into the model, adjusting the parameters can determine which factors have the greatest impact on disasters in a specific area.

[0085] Forecast result output and interpretation: The model will output the risk value of geological hazards at a certain moment .if If the risk exceeds the set threshold (for example, 0.7), it indicates that the area has a high risk of geological disasters in the future, and the system will issue an early warning signal. In addition, the model can generate a risk change trend chart for a period of time in the future based on input data at different times, helping decision makers understand the evolution of risks.

[0086] S8. Data anomaly detection and automated warning Use machine learning algorithms to analyze multi-temporal remote sensing data to identify potential abnormal changes and issue early warnings.

[0087] In this embodiment, a convolutional neural network (CNN) or a recurrent neural network (RNN) is used to perform automatic detection in spatial and temporal dimensions; Identify sudden changes and accelerations in surface deformation and generate automated early warning signals to indicate when a disaster may occur.

[0088] S9. Risk Assessment and Impact Analysis A comprehensive risk assessment is conducted through deformation monitoring results combined with data on regional population density and infrastructure distribution.

[0089] In this embodiment, the method also ensures adaptation to environmental changes and provides continuous disaster monitoring services by dynamically updating deformation data of the monitored area and combining automated data updating and model retraining mechanisms.

[0090] The beneficial effects of the present invention are: This geological disaster deformation monitoring method based on remote sensing data overcomes the shortcomings of existing technologies that rely on a single data source by combining SAR, optical imaging and LiDAR data. It uses dynamic data fusion and time series analysis to achieve multi-dimensional real-time monitoring of surface deformation. This method not only improves the accuracy and timeliness of monitoring, but also provides more comprehensive geological disaster warnings, greatly enhancing the effectiveness of disaster prevention and control.

[0091] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0092] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for monitoring geological disaster deformation based on remote sensing data, characterized in that: The following steps are involved: S1. Data source selection and preprocessing Select remote sensing data sources, including synthetic aperture radar, optical satellite imagery, and lidar; The preprocessing steps include radiation correction, geometric correction and atmospheric correction; S2. Regional feature extraction Use digital elevation models or terrain data to extract topographic characteristics of geohazard risk areas, including slope, rainfall intensity, and surface deformation rate; After feature extraction, the risk areas are classified and the key areas for subsequent monitoring are determined; S3, Multi-temporal remote sensing image difference Extracting surface deformation information from multi-temporal remote sensing images of the same area; Using differential interferometric synthetic aperture radar technology, the surface displacement changes between different periods are calculated; Use optical images to monitor large-scale surface changes and identify precursors to large-scale disasters such as landslides and mud-rock flows; S4. Time Series Analysis Time series analysis based on dynamic model-driven methods, combining machine learning and physical models, updating and adjusting models in real time to reflect the impact of environmental factors; Capture sudden changes in geological hazards, including changes in surface stress after abnormal rainfall or earthquakes, through trend analysis of dynamically adjusted time series; S5. 3D reconstruction of the deformed area Based on multi-source data, three-dimensional reconstruction is performed to build an accurate model of the geological disaster deformation area; through the fusion of lidar, optical imaging and high-resolution synthetic aperture radar data, a three-dimensional model of the surface is generated; S6. Subsidence and landslide monitoring By combining surface settlement monitoring technology and remote sensing data, deformation monitoring of settlement and landslide is carried out; S7. Fusion analysis of environmental and meteorological data Fuse remote sensing data with meteorological data to analyze the impact of meteorological changes on surface deformation; Meteorological data is obtained through satellite observation or ground stations, and the relationship between data and surface deformation is analyzed to predict the possibility of future disasters; S8. Data anomaly detection and automated warning Use machine learning algorithms to analyze multi-temporal remote sensing data to identify potential abnormal changes and issue early warnings; S9. Risk Assessment and Impact Analysis A comprehensive risk assessment is conducted through deformation monitoring results combined with data on regional population density and infrastructure distribution.

2. A geological disaster deformation monitoring method based on remote sensing data according to claim 1, characterized in that: In step S2, a comprehensive feature scoring function is constructed, integrating three key feature parameters: slope, rainfall intensity and surface deformation rate. The function is as follows; ； in, :Geographic location( ) represents the potential risk of geological hazards in the area; : The slope of the location, extracted from the digital elevation model data, reflects the terrain undulation and landslide possibility; : The rainfall intensity in the area, which indicates the cumulative rainfall in a certain period of time obtained through meteorological data; : Surface deformation rate, the surface displacement rate calculated based on differential interferometric synthetic aperture radar remote sensing technology, reflects the dynamic changes of the surface; : Weight coefficient, which reflects the contribution of each characteristic parameter to the risk of geological hazards and is determined through historical data training or expert knowledge; : Offset, used to adjust the final score value; Set the adjustment function to deal with the disaster characteristics of different regions. Assume that weighted optimization is performed on different regional characteristics and introduce nonlinear terms: ； in, : Nonlinear adjustment factor for slope; : The impact of smoothing rainfall intensity based on logarithmic function; : Square root processing of deformation rate.

3. The method for monitoring geological disaster deformation based on remote sensing data according to claim 1, characterized in that: In step S4, the dynamic model driving method includes the following process: Integration of physical models and machine learning: Combining physical models, including surface stress models, settlement models and machine learning techniques, the physical models provide mechanism understanding, and machine learning handles complex, nonlinear environmental changes; Dynamic updates in time series: The model is continuously updated based on new data. When the system environment changes, such as when there is a heavy rain or earthquake, the model automatically adjusts; Multi-source data fusion: The dynamic model uses data from multiple data sources simultaneously, including surface deformation obtained by synthetic aperture radar, rainfall intensity in meteorological data, and earthquake data. Multi-dimensional data is fused and weighted by the model, and weights are dynamically assigned according to the characteristics of each data source, optimizing the prediction at each moment. Adaptive adjustment mechanism: The dynamic model handles anomalies and mutations in the data through an adaptive adjustment mechanism. When the system detects sudden changes caused by an earthquake, the model automatically adjusts its parameters.

4. The method for monitoring geological disaster deformation based on remote sensing data according to claim 1, characterized in that: In step S5, it is assumed that the geographic coordinates of a point in the three-dimensional space are , using deformation parameters that fuse multi-source data Representing the time series deformation of this point, the exact model is as follows: ； in, The initial three-dimensional coordinates of a point on the surface are high-precision three-dimensional point cloud data obtained based on lidar data; is the three-dimensional coordinate of the point after deformation, that is, the exact position after deformation; For time The surface deformation data of the point is obtained, and the surface displacement information of different time periods is obtained through interference analysis; It is the pixel value of the point in the optical image, reflecting the color and texture information of the surface; For elevation model data, it is based on the fusion of LiDAR and digital elevation model; It is a function used to fuse multi-source data, supplement and optimize the 3D model details.

5. A method for monitoring geological disaster deformation based on remote sensing data according to claim 4, characterized in that: The precise model, when used, includes the following process: Data fusion and processing: Get each surface point from LiDAR data The initial coordinates of the point at different times are calculated by combining the synthetic aperture radar data The deformation under At the same time, optical imaging Provide surface detail information; Deformation compensation: based on surface deformation data , calculate the displacement of the point in the time series, for each moment , the deformation and the original coordinates Superimpose to obtain the deformed coordinates ; 3D reconstruction: based on the 3D coordinates of all surface points , three-dimensional reconstruction of the entire deformed area is performed through interpolation algorithm or surface fitting algorithm; Fusion of multi-source data: introduction of optical imaging and elevation model , showing the true characteristics of the surface; AI algorithm optimization: Use deep learning models including convolutional neural networks or generative adversarial networks to supplement and correct model details.

6. The method for monitoring geological disaster deformation based on remote sensing data according to claim 1, characterized in that: In step S6, during the monitoring process, the minute settlement and landslide precursors are combined with the remote sensing observation results through algorithm deduction and data fusion for in-depth analysis; It also identifies landslide precursors based on machine learning algorithms, combines surface deformation information with meteorological data and geological structure data, and predicts the possibility of future landslides.

7. The method for monitoring geological disaster deformation based on remote sensing data according to claim 1, characterized in that: In step S7, it is assumed that the possibility of geological disasters occurring is It means that depending on multi-dimensional factors in different time periods, the prediction model is as follows: ； in, For geological disasters in time The risk value of occurrence under For time The accumulated rainfall under is the surface deformation rate; is a geological structural parameter, depending on the coordinates ( ) under geological conditions; For time Temperature data below; is the impact factor of historical disaster data; To adjust the parameters.

8. A method for monitoring geological disaster deformation based on remote sensing data according to claim 7, characterized in that: When the prediction model is used, it includes the following process: Multi-source data input and processing: Collect required input data from different sensors and data sources; Risk calculation: The model combines all input data and calculates the The risk of geological disasters ; Adjustment parameter optimization: Use machine learning algorithms to adjust the parameters Optimize, train the model through historical data, and optimize the weights so that the model can accurately predict the time and scale of disasters; Output and interpretation of prediction results: Output the risk value of geological hazards at a certain moment ,when Exceeding the set risk threshold indicates that the area faces a high risk of geological disasters in the future, issuing a warning signal.

9. The method for monitoring geological disaster deformation based on remote sensing data according to claim 1, characterized in that: In step S8, a convolutional neural network or a recursive neural network is used to perform automatic detection in spatial and temporal dimensions; Identify sudden changes and accelerations in surface deformation and generate automated early warning signals to indicate when a disaster may occur.

10. The method for monitoring geological disaster deformation based on remote sensing data according to claim 1, characterized in that: This method also provides continuous disaster monitoring services by dynamically updating the deformation data of the monitored area and combining automated data update and model retraining mechanisms to adapt to environmental changes.

Images