Method and system for monitoring lake water level based on remote sensing image

Abstract

The invention discloses a method and system for monitoring the lake water level based on remote sensing images, and particularly relates to the technical field of remote sensing monitoring, and the method comprises the steps: constructing a complete three-dimensional terrain model of a lake through historical remote sensing images and data of underwater detection equipment, including water surface and underwater shapes; a quantitative relation model of the terrain and the water level is established by utilizing a digital twinborn technology and actually measured water level data, the model is repeatedly tested and optimized, the precision is ensured, and the model is finally applied to real-time water level monitoring, so that the water level information of a target lake is automatically acquired and updated; the problem that a terrain-water level relation model is difficult to accurately monitor lake water level changes in the prior art is solved. The system further comprises a water body boundary correction module and a boundary contour processing module, the boundary of the lake is recognized and extracted more accurately through a remote sensing image and a machine learning model, the boundary is optimized, and accuracy and completeness of water body information are guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of remote sensing monitoring technology, and more specifically, to a method and system for monitoring lake water level based on remote sensing images. Background Art

[0002] Lake water level, as the height of the lake surface, is an important indicator for measuring the amount of lake water and the health of the lake ecosystem. This indicator is usually expressed relative to a fixed datum, which can be sea level, the lowest point of the lake, or other selected reference points. Monitoring lake water levels is extremely important for water resource management, flood control and drought relief, and ecological protection.

[0003] As a non-contact monitoring method, remote sensing technology has the advantages of wide monitoring range, fast monitoring speed and high monitoring accuracy, and has been widely used in lake water level monitoring. Remote sensing technology uses sensors carried by platforms such as satellites or drones to monitor lakes over a long distance and over a large area. By processing and analyzing the surface features of water bodies in remote sensing images (such as color, texture, etc.), we can identify the spatial distribution and changing trends of lake water levels. In addition, radar remote sensing technology can also accurately measure the height changes of lake water surfaces by emitting radar waves and receiving their echo signals, thereby realizing real-time monitoring of lake water levels.

[0004] However, the existing method of monitoring lake water levels using remote sensing images still has the problem that the existing terrain-water level relationship model is too simplified, making it difficult to accurately monitor lake water level changes. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method and system for monitoring lake water levels based on remote sensing images. By utilizing terrain data and water body boundary information in remote sensing images, a terrain-water level relationship model is established to achieve high-precision and real-time lake water level monitoring to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: A method for monitoring lake water level based on remote sensing images, comprising the following steps: Step 1: Construct the first lake terrain model: select the remote sensing image A of the target lake at the lowest water level from the historical data, and construct the above-water 3D model A; use underwater equipment to obtain the water depth and terrain information when the water level is not the lowest, and construct the underwater 3D model B; fuse the above-water 3D model A and the underwater 3D model B, and output the first terrain model of the target monitoring lake; Step 2: Construct a terrain-water level relationship model: Select a reference area from several areas of the target monitored lake, and use the digital twin technology to establish a terrain-water level relationship model based on the first terrain model that has been constructed, combined with the remote sensing image of the reference area and the corresponding water level measurement data; Step 3: Continuously adjust and optimize the parameters of the terrain-water level relationship model through multiple iterations, so that the terrain-water level relationship model can accurately reflect the relationship between the lake terrain and the water level, and verify the credibility of the terrain-water level relationship model; Step 4: When the credibility meets the requirements, the acquired terrain-water level relationship model is applied to the water level monitoring of the target monitoring lake, and the water level information of the target monitoring lake is output.

[0007] Preferably, the process of acquiring the first terrain model includes: In data fusion, the target monitoring lake is divided into several areas according to logical areas and numbered; Obtain the accuracy of remote sensing image data, terrain data and underwater equipment data for each area, perform weighted summation of the accuracy, and output the fused intersection terrain; Based on the fused intersection terrain, a calibration matrix of the above-water 3D model A and a calibration matrix of the underwater 3D model B are established respectively, and the non-intersection part is adjusted based on the calibration matrix; The fused intersection part and the calibrated non-intersection part constitute the first terrain model of the target monitored lake.

[0008] Preferably, the method for obtaining the fused intersection terrain is: Convert the remote sensing image data, terrain data and underwater equipment data of each area into key point coordinates, and set the number of key points based on the accuracy requirements; Obtain the key point coordinates of remote sensing image data, terrain data, and underwater equipment data, recorded as dz_a, dz_b, dz_c respectively; The accuracy of remote sensing image data, terrain data, and data obtained by underwater equipment are respectively recorded as zl_a, zl_b, and zl_c; The fused key point coordinates R_d are calculated by the following formula;

[0009] Among them, δ represents the common accuracy correction term of remote sensing image data, terrain data and data obtained by underwater equipment. It is obtained based on the test, and δ is used to correct the impact of possible correlation between data sources on the results.

[0010] Preferably, the process of acquiring the terrain-water level relationship model comprises the following steps: Step 101, select a reference area: select a reference area from among several areas of the target monitoring lake, the reference area having the characteristics of stable terrain and easy observation of water level changes, so as to facilitate subsequent water level measurement and model verification; Step 102: Collect water level measurement data of the reference area, using water level gauges and water level gauge tools to ensure the accuracy and reliability of the water level measurement data; Step 103: Using digital twin technology, the physical lake in the reference area is combined with the corresponding virtual model, the remote sensing image of the reference area is input, the simulation result of the water level in the reference area is output, and the mapping relationship between the remote sensing image and the water level measurement data is established; Step 104: using the overall water level distribution information of the lake provided by the remote sensing image, combined with the water level data of the reference area, a terrain-water level relationship model is established based on the mapping relationship between the remote sensing image and the water level measurement data.

[0011] Preferably, the number of the reference areas is greater than or equal to 1, and is based on user settings.

[0012] Preferably, the parameters of the terrain-water level relationship model are adjusted, and the parameter space is searched by using an optimization algorithm to find a parameter combination that minimizes the prediction error of the terrain-water level relationship model; the parameters include but are not limited to: terrain elevation, terrain slope, terrain roughness, and a mapping relationship between remote sensing images of the reference area and water level measurement data.

[0013] Preferably, the accuracy and stability of the terrain-water level relationship model are verified by a field verification method, and the credibility of the terrain-water level relationship model is evaluated according to the verification result, wherein the credibility includes a first credibility and a second credibility, and the verification method of the terrain-water level relationship model includes the following steps: Step 201, obtaining the first credibility of the water level prediction result of the reference area: after several tests using the field verification method, the mean square error and the coefficient of variation of the error between the predicted value and the field measurement value are obtained, which are recorded as k1 and k2 respectively; k1 and k2 are weighted and summed, and normalized to the range of 0 to 1, and the weight is determined according to actual needs to obtain the first credibility Ka; Step 202: If the first credibility meets the requirement, proceed to the next step; otherwise, issue an early warning, indicating that the terrain-water level relationship model is abnormal; Step 203, obtaining the second credibility of the prediction results of other regions except the reference region; After several tests using the field verification method, the mean square error and coefficient of variation of the mapping relationship between the output remote sensing image and the water level measurement data are denoted as k3 and k4 respectively. The weighted sum of k3 and k4 is performed and normalized to the range of 0 to 1. The weight is determined according to actual needs to obtain the second credibility Kb.

[0014] Preferably, after obtaining the target monitored lake water level information, a water body boundary correction step is also included: Used to correct the first water body boundary information corresponding to the first terrain model, and obtain the undetermined water body boundary area based on the first water body boundary information and remote sensing image information; transmit the remote sensing data corresponding to the undetermined water body boundary area to the trained water body boundary recognition model to realize the correction of the water body boundary and output the second water body boundary information; The water body boundary recognition mode reliability index Ls is obtained as follows: Record the marked water body boundary as the first vector data; Marking the water body boundary output by the water body boundary recognition model as second vector data; The first vector data and the second vector data are both sets containing coordinate points; Convert the vector data into raster data so that each raster cell has a corresponding value (for example, 0 represents non-water body and 1 represents water body), compare the first vector data with the second vector data, Assume that the first vector data after rasterization (real water body boundary) is R1, the second vector data after rasterization is R2, and the size of the grid is m×n; The reliability index Ls of the water body boundary recognition model is calculated by the following model

[0015]

[0016] in, represents the mean square error loss of the first vector data and the second vector data; i and j represent the row index and column index in the raster data respectively, i represents the row number in the raster data, counting from 1 to the total number of rows m in the raster; j represents the column number in the raster data, counting from 1 to the total number of columns n in the raster; and are the true value and model predicted value at the i-th row and j-th column of the grid respectively; The water body boundary correction step is connected with the boundary contour processing step: The boundary contour processing step refers to the post-processing of the segmentation results output by the water body boundary recognition model, including morphological operations to clean up noise and irregular edges; and using contour extraction algorithms to display the boundary contours of the water body.

[0017] Preferably, the method further includes: a lake water level monitoring quality early warning step, combining the first credibility Ka, the second credibility Kb and the water body boundary recognition mode reliability index Ls to obtain a lake water level monitoring quality coefficient, and judging whether to issue an early warning to the user based on the relationship between the lake water level monitoring quality coefficient and a preset threshold, including: Combining the first credibility Ka, the second credibility Kb and the water body boundary recognition mode reliability index Ls, the lake water level monitoring quality coefficient SJ is calculated by the following formula;

[0018] When the lake water level monitoring quality coefficient exceeds the threshold, an early warning message is sent to the user, indicating that there may be inaccurate monitoring data; When the lake water level monitoring quality coefficient does not exceed the threshold, it indicates that the lake water level monitoring is normal and no measures are taken.

[0019] To achieve the purpose of the present invention, a system for monitoring lake water level based on remote sensing images is provided, comprising: The first lake terrain model construction module selects the remote sensing image A of the target lake at the lowest water level from historical data to construct the above-water 3D model A; uses underwater equipment to obtain the water depth and terrain information when the water level is not the lowest, and constructs the underwater 3D model B; integrates the above-water 3D model A and the underwater 3D model B to output the first terrain model of the target monitoring lake; The terrain-water level relationship model construction module selects a reference area from several areas of the target monitored lake, and uses the digital twin technology to establish a terrain-water level relationship model based on the first terrain model that has been constructed, combined with the remote sensing image of the reference area and the corresponding water level measurement data; The terrain-water level relationship model credibility verification module continuously adjusts and optimizes the terrain-water level relationship model parameters through multiple iterations, so that the terrain-water level relationship model can accurately reflect the relationship between lake terrain and water level, and verify the credibility of the terrain-water level relationship model; The terrain-water level relationship model credibility application module applies the acquired terrain-water level relationship model to the water level monitoring of the target monitoring lake when the credibility meets the requirements, and outputs the water level information of the target monitoring lake.

[0020] Technical effects and advantages of the present invention: (1) The method for monitoring lake water level based on remote sensing images provided by the present invention constructs a first terrain model of the target monitored lake and establishes a terrain-water level relationship model in combination with digital twin technology, thereby effectively realizing accurate simulation and monitoring of the target lake water level; constructs a three-dimensional model of the target lake through historical data and measured data, and then uses digital twin technology to establish a mapping relationship between the remote sensing image and the water level measurement data based on the data of the reference area, and infers the terrain-water level relationship model; finally, optimizes the model parameters through multiple iterations and verifies its credibility, thereby realizing real-time monitoring of the target monitored lake water level, thereby solving the problem in the prior art that the terrain-water level relationship model is difficult to accurately monitor lake water level changes.

[0021] (2) The method for monitoring lake water level based on remote sensing images provided by the present invention further enhances the credibility and practicality of the monitoring results by introducing water body boundary correction and monitoring quality early warning links. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a flow chart of the method for monitoring lake water level based on remote sensing images of the present invention.

[0023] Figure 2 This is a flowchart of the water body boundary processing of the present invention.

[0024] Figure 3 This is a structural block diagram of the lake water level monitoring system based on remote sensing images of the present invention. DETAILED DESCRIPTION

[0025] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0026] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the present application, its application, or uses.

[0027] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered as part of the specification.

[0028] Example 1, see Figure 1 A flow chart of a method for monitoring lake water level based on remote sensing images, the present invention provides Figure 1 A method for monitoring lake water level based on remote sensing images is shown, comprising the following steps: Step 1: Construct the first lake terrain model: select the remote sensing image A of the target lake at the lowest water level from the historical data, and construct the above-water 3D model A by combining the terrain data and image A; use underwater equipment to obtain the water depth and terrain information when the water level is not the lowest, and construct the underwater 3D model B; by comparing the elevation and position, fuse the above-water 3D model A and the underwater 3D model B, and output the first terrain model of the target monitoring lake; Explanation: Remote sensing images at the lowest water level are crucial for understanding the topographic features of lakes at extremely low water levels, and provide basic data for subsequent reconstruction of the surface topography; Explanation: The above-water 3D model A and the underwater 3D model B have an intersection. Based on the intersection, the underwater terrain 3D simulation model is scaled, converted and corrected to achieve registration of the above-water terrain 3D simulation model and the underwater terrain 3D simulation model, and the lake terrain model is obtained, including: Identify the intersection area: Determine the intersection area by comparing the elevation information and spatial location of the two models; Scaling and transforming underwater terrain models: Scaling, transforming and correcting underwater terrain models based on intersection areas; Fusion model: After scaling, conversion and correction, the surface terrain model and the underwater terrain model are fused to obtain a complete and continuous lake terrain model, which can more accurately reflect the actual terrain characteristics and water level changes of the lake; Step 2: Construct a terrain-water level relationship model: Combine the physical lake in the reference area with the corresponding virtual model, establish a mapping relationship between the remote sensing image of the reference area and the water level measurement data by adjusting the parameters, input the remote sensing image of the reference area, and output the simulation result of the water level in the reference area; use the overall water level distribution information of the lake provided by the remote sensing image, combined with the water level data of the reference area, and establish a terrain-water level relationship model through the calculation function of the digital twin technology; Explanation: The embodiment of the present invention does not specifically limit the selection of the reference area, and the reference area can be flexibly selected according to user settings or according to the following principles: Select areas with representative topographic features as reference areas. Consider the impact of water flow on lake water levels, select areas with relatively stable or representative water flow, and avoid areas with complex water flow and susceptible to external interference. Consider the convenience and safety of field measurement and data collection, and select areas that are easy to reach and measure. The number of reference areas should be flexibly set according to actual needs, and at least one area should be selected as a reference. In order to improve the accuracy and comprehensiveness of monitoring, multiple areas can be selected as references based on the size, shape and complexity of the lake. Step 3: Continuously adjust and optimize the parameters of the terrain-water level relationship model through multiple iterations, so that the terrain-water level relationship model can accurately reflect the relationship between the lake terrain and the water level, and verify the credibility of the terrain-water level relationship model; Step 4: When the credibility meets the requirements, the acquired terrain-water level relationship model is applied to the water level monitoring of the target monitoring lake, and the water level information of the target monitoring lake is output.

[0029] It needs to be further explained in the embodiment of the present invention that the process of obtaining the first terrain model includes: In data fusion, the target monitoring lakes are divided into several logical areas, such as according to the shape, size or water flow characteristics of the lake, and numbered; Obtain the accuracy of remote sensing image data, terrain data and underwater equipment data for each area, and output the fused intersection terrain through weighted summation of the accuracy, comprehensively consider the reliability and accuracy of different data sources to improve the accuracy of the fusion result; Based on the fused intersection terrain, a calibration matrix of the above-water 3D model A and a calibration matrix of the underwater 3D model B are established respectively, and the non-intersection part is adjusted based on the calibration matrix; Explanation: The calibration matrix refers to the transformation parameters that adjust the non-intersecting part data to be consistent with the intersecting part data; it is used to adjust the non-intersecting part data to be consistent with the intersecting part data; The fused intersection part and the calibrated non-intersection part constitute the first terrain model of the target monitored lake.

[0030] It needs to be further explained in the embodiment of the present invention that the method of obtaining the fused intersection part terrain is: Convert the remote sensing image data, terrain data and underwater equipment data of each area into key point coordinates, and set the number of key points based on the accuracy requirements; Obtain the key point coordinates of remote sensing image data, terrain data, and underwater equipment data, recorded as dz_a, dz_b, dz_c respectively; The accuracy of remote sensing image data, terrain data, and data obtained by underwater equipment are respectively recorded as zl_a, zl_b, and zl_c; The fused key point coordinates R_d are calculated by the following formula;

[0031] Among them, δ represents the common accuracy correction term of remote sensing image data, terrain data and data obtained by underwater equipment. It is obtained based on the test, and δ is used to correct the impact of possible correlation between data sources on the results.

[0032] It needs to be further explained in the embodiment of the present invention that the process of acquiring the terrain-water level relationship model includes the following steps: Step 101, select a reference area: select a reference area from among several areas of the target monitoring lake, the reference area having the characteristics of stable terrain and easy observation of water level changes, so as to facilitate subsequent water level measurement and model verification; Step 102: Collect water level measurement data of the reference area, using water level gauges and water level gauge tools to ensure the accuracy and reliability of the water level measurement data; Step 103: Using digital twin technology, the physical lake in the reference area is combined with the corresponding virtual model, the remote sensing image of the reference area is input, the simulation result of the water level in the reference area is output, and the mapping relationship between the remote sensing image and the water level measurement data is established; Step 104: using the overall water level distribution information of the lake provided by the remote sensing image, combined with the water level data of the reference area, a terrain-water level relationship model is established based on the mapping relationship between the remote sensing image and the water level measurement data.

[0033] What needs to be further explained in the embodiments of the present invention is that the parameters of the terrain-water level relationship model are adjusted by using an optimization algorithm (such as gradient descent method, genetic algorithm, etc.) to search the parameter space to find the parameter combination that minimizes the prediction error of the terrain-water level relationship model; the parameters include but are not limited to: terrain elevation, terrain slope, terrain roughness, and the mapping relationship between the remote sensing image of the reference area and the water level measurement data.

[0034] It is necessary to further explain in the embodiment of the present invention that the accuracy and stability of the terrain-water level relationship model are verified by a field verification method, and the credibility of the terrain-water level relationship model is evaluated according to the verification result. The credibility includes a first credibility and a second credibility. The verification method of the terrain-water level relationship model includes the following steps: Step 201, obtaining the first credibility of the water level prediction result of the reference area: after several tests using the field verification method, the mean square error and the coefficient of variation of the error between the predicted value and the field measurement value are obtained, which are recorded as k1 and k2 respectively; k1 and k2 are weighted and summed, and normalized to the range of 0 to 1, and the weight is determined according to actual needs to obtain the first credibility Ka; Step 202: If the first credibility meets the requirement, proceed to the next step; otherwise, issue an early warning, indicating that the terrain-water level relationship model is abnormal; Step 203, obtaining the second credibility of the prediction results of other regions except the reference region; After several tests using the field verification method, the mean square error and coefficient of variation of the mapping relationship between the output remote sensing image and the water level measurement data are denoted as k3 and k4 respectively. The weighted sum of k3 and k4 is performed and normalized to the range of 0 to 1. The weight is determined according to actual needs to obtain the second credibility Kb.

[0035] Summary: The embodiment of the present invention constructs a first terrain model of the target monitored lake, and establishes a terrain-water level relationship model in combination with the digital twin technology, thereby effectively realizing the accurate simulation and monitoring of the water level of the target lake; constructs a three-dimensional model of the target lake through historical data and measured data, and then uses the digital twin technology to establish a mapping relationship between the remote sensing image and the water level measurement data based on the data of the reference area, and infers the terrain-water level relationship model; finally, optimizes the model parameters through multiple iterations to verify its credibility, thereby realizing real-time monitoring of the water level of the target monitored lake, and solving the problem in the prior art that the terrain-water level relationship model is difficult to accurately monitor the changes in the lake water level.

[0036] Embodiment 2: In the method for monitoring lake water level using remote sensing images, there may still be a problem of low water body boundary extraction accuracy. Since the color, texture and other characteristics of the lake water body and the surrounding environment are not obviously different, or are affected by interference factors such as clouds, fog and shadows, the extraction accuracy of the water body boundary is low, and it is difficult to accurately reflect the change of the lake water level. Based on this, the content of Embodiment 2 of the present invention is provided. See also Figure 2 The water body boundary processing flow chart of the embodiment of the present invention is different from the embodiment 1 in that it also includes a water body boundary correction step, which specifically includes the following steps: Step 301: Obtain first water body boundary information based on a first terrain model; Explanation: The first terrain model provides the topographic features of the lake, and the boundary position of the water body is inferred based on the topographic features. Therefore, the first water body boundary information is usually a relatively rough outline, but it provides a basis for subsequent corrections; Step 302: obtaining a water body boundary pending area based on the first water body boundary information and the remote sensing image information; Explanation: The first water body boundary information is superimposed and compared with the remote sensing image information to find the difference area between the two; the difference area is usually caused by aquatic plants, algae coverage, water level changes, and terrain changes. What needs to be further explained in the embodiment of the present invention is to identify and correct, and output a water body boundary pending area; Step 303: transmit the remote sensing data corresponding to the area to be determined of the water body boundary to the trained water body boundary recognition model to correct the water body boundary and output the second water body boundary information.

[0037] It needs to be further explained in the embodiment of the present invention that the process of acquiring the water body boundary recognition model satisfies the following steps: Data collection: Collect remote sensing images containing aquatic plants and algae coverage in different remote sensing bands to ensure that the remote sensing images have diverse lighting conditions, aquatic plants and algae distribution, and water depth; Data annotation: annotate the collected remote sensing images, mark the boundaries of the water bodies, convert the annotated data into the format required for model training; divide the annotated data into training sets and validation sets; Build and optimize the water body boundary recognition model: extract features related to water body boundaries from remote sensing images, such as grayscale value, texture features, and shape features, use feature selection methods to reduce feature dimensions and improve model training efficiency; select adaptive thresholds to segment water bodies based on the complexity of water body boundaries; use deep learning technology, such as convolutional neural networks (CNNs), to build initial recognition models; Data training: Use the training set data to train the model and verify the performance on the validation set. Set the loss function based on the difference in the water body boundary coordinates, such as the mean square error loss, and minimize the loss function value by adjusting the model parameters. Adjust and optimize the initial recognition model according to the test results, train the loss function value to meet the preset requirements, and obtain a trained water body boundary recognition model. Practical application: Apply the trained water body boundary recognition model to extract the actual lake water body boundary and output the segmentation result; It needs to be further explained in the embodiment of the present invention that the water body boundary correction step is connected with the boundary contour processing step: The boundary contour processing step refers to the post-processing of the segmentation results output by the water body boundary recognition model, including morphological operations (such as dilation, erosion, opening operation, closing operation, etc.) to clean up noise and irregular edges; using contour extraction algorithms (such as Canny edge detection, Hough transform, etc.) to display the boundary contour of the water body.

[0038] In a possible embodiment, the method further includes: a lake water level monitoring quality warning step, combining the first credibility Ka, the second credibility Kb and the water body boundary recognition mode reliability index Ls to obtain a lake water level monitoring quality coefficient, and judging whether to warn the user based on the relationship between the lake water level monitoring quality coefficient and a preset threshold, including: It needs to be further explained in the embodiment of the present invention that the water body boundary recognition mode reliability index Ls is obtained in the following manner: Record the marked water body boundary as the first vector data; Marking the water body boundary output by the water body boundary recognition model as second vector data; The first vector data and the second vector data are both sets containing coordinate points; Convert the vector data into raster data so that each raster cell has a corresponding value (for example, 0 represents non-water body and 1 represents water body), compare the first vector data with the second vector data, Assume that the first vector data after rasterization (real water body boundary) is R1, the second vector data after rasterization is R2, and the size of the grid is m×n; The reliability index Ls of the water body boundary recognition model is calculated by the following model

[0039]

[0040] in, represents the mean square error loss of the first vector data and the second vector data; i and j represent the row index and column index in the raster data respectively, i represents the row number in the raster data, counting from 1 to the total number of rows m in the raster; j represents the column number in the raster data, counting from 1 to the total number of columns n in the raster; and are the true value and model predicted value at the i-th row and j-th column of the grid respectively.

[0041] Combining the first credibility Ka, the second credibility Kb and the water body boundary recognition mode reliability index Ls, the lake water level monitoring quality coefficient SJ is calculated by the following formula;

[0042] When the lake water level monitoring quality coefficient exceeds the threshold, an early warning message is sent to the user, indicating that there may be inaccurate monitoring data; When the lake water level monitoring quality coefficient does not exceed the threshold, it indicates that the lake water level monitoring is normal and no measures are taken.

[0043] What needs to be further explained in the embodiments of the present invention is that an early warning is issued based on the changing trend of the lake water level monitoring quality coefficient. If the lake water level monitoring quality coefficient shows a significant fluctuation or a continuous upward / downward trend in a short period of time, an early warning message is issued to indicate that there is a problem with the monitoring data or that attention is required.

[0044] Explanation: The warning information should include the name of the lake, monitoring time, quality coefficient value, degree of exceeding the threshold, and possible recommended measures; the warning information is sent to users via SMS, email, and APP push to ensure that users can receive it in time and take appropriate measures; after receiving the warning information, users should perform inspection and maintenance work according to the prompts, and feedback the processing results to the system administrator.

[0045] Summary: Although the presence of aquatic plants and algae increases the difficulty of extracting the boundaries of lake water bodies, accurate identification of water body boundaries can be achieved by combining feature extraction, multi-threshold segmentation, post-processing, and possible machine learning and deep learning techniques. Through remote sensing images and machine learning models, the boundaries of lakes can be more accurately identified and extracted, and the boundaries can be optimized to ensure the accuracy and completeness of water body information.

[0046] Example 3, see Figure 3 The structural block diagram of the system for monitoring lake water level based on remote sensing images is provided. The embodiment of the present invention provides a system for monitoring lake water level based on remote sensing images, including: The first lake terrain model construction module selects the remote sensing image A of the target lake at the lowest water level from historical data to construct the above-water 3D model A; uses underwater equipment to obtain the water depth and terrain information when the water level is not the lowest, and constructs the underwater 3D model B; integrates the above-water 3D model A and the underwater 3D model B to output the first terrain model of the target monitoring lake; The terrain-water level relationship model construction module selects a reference area from several areas of the target monitored lake, and uses the digital twin technology to establish a terrain-water level relationship model based on the first terrain model that has been constructed, combined with the remote sensing image of the reference area and the corresponding water level measurement data; The terrain-water level relationship model credibility verification module continuously adjusts and optimizes the terrain-water level relationship model parameters through multiple iterations, so that the terrain-water level relationship model can accurately reflect the relationship between lake terrain and water level, and verify the credibility of the terrain-water level relationship model; The terrain-water level relationship model credibility application module applies the acquired terrain-water level relationship model to the water level monitoring of the target monitoring lake when the credibility meets the requirements, and outputs the water level information of the target monitoring lake.

[0047] It needs to be further explained in the embodiment of the present invention that the system further comprises: a water body boundary correction module and a boundary contour processing module. The water body boundary correction module is used to correct the first water body boundary information corresponding to the first terrain model, obtain the water body boundary pending area based on the first water body boundary information and the remote sensing image information; transmit the remote sensing data corresponding to the water body boundary pending area to the trained water body boundary recognition model to realize the correction of the water body boundary and output the second water body boundary information; The boundary contour processing module is used to post-process the segmentation results output by the water body boundary recognition model, including morphological operations to clean up noise and irregular edges; and use a contour extraction algorithm to display the boundary contour of the water body.

[0048] Finally: 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 in the protection scope of the present invention.

Claims

1. A method for monitoring lake water level based on remote sensing images, characterized in that: The following steps are involved: Step 1: Construct the first lake terrain model: select the remote sensing image A of the target lake at the lowest water level from the historical data, and construct the above-water 3D model A; use underwater equipment to obtain the water depth and terrain information when the water level is not the lowest, and construct the underwater 3D model B; fuse the above-water 3D model A and the underwater 3D model B, and output the first terrain model of the target monitoring lake; Step 2: Construct a terrain-water level relationship model: Select a reference area from several areas of the target monitored lake, and use the digital twin technology to establish a terrain-water level relationship model based on the first terrain model that has been constructed, combined with the remote sensing image of the reference area and the corresponding water level measurement data; Step 3: Continuously adjust and optimize the parameters of the terrain-water level relationship model through multiple iterations, so that the terrain-water level relationship model can accurately reflect the relationship between the lake terrain and the water level, and verify the credibility of the terrain-water level relationship model; Step 4: When the credibility meets the requirements, the acquired terrain-water level relationship model is applied to the water level monitoring of the target monitoring lake, and the water level information of the target monitoring lake is output.

2. The method for monitoring lake water level based on remote sensing images according to claim 1, characterized in that: The process of acquiring the first terrain model includes: In data fusion, the target monitoring lake is divided into several areas according to logical areas and numbered; Obtain the accuracy of remote sensing image data, terrain data and underwater equipment data for each area, perform weighted summation of the accuracy, and output the fused intersection terrain; Based on the fused intersection terrain, a calibration matrix of the above-water 3D model A and a calibration matrix of the underwater 3D model B are established respectively, and the non-intersection part is adjusted based on the calibration matrix; The fused intersection part and the calibrated non-intersection part constitute the first terrain model of the target monitored lake.

3. The method for monitoring lake water level based on remote sensing images according to claim 2, characterized in that: The method for obtaining the fused intersection terrain is as follows: Convert the remote sensing image data, terrain data and underwater equipment data of each area into key point coordinates, and set the number of key points based on the accuracy requirements; Obtain the key point coordinates of remote sensing image data, terrain data, and underwater equipment data, recorded as dz_a, dz_b, dz_c respectively; The accuracy of remote sensing image data, terrain data, and data obtained by underwater equipment are respectively recorded as zl_a, zl_b, and zl_c; The fused key point coordinates R_d are calculated by the following formula; ; Among them, δ represents the common accuracy correction term of remote sensing image data, terrain data and data obtained by underwater equipment. Based on the test, δ is used to correct the impact of possible correlation between data sources on the results.

4. The method for monitoring lake water level based on remote sensing images according to claim 2 is characterized in that: The process of acquiring the terrain-water level relationship model includes the following steps: Step 101, selecting a reference area: selecting a reference area among several areas of the target monitoring lake; Step 102, collecting water level measurement data of the reference area; Step 103: Using digital twin technology, the physical lake in the reference area is combined with the corresponding virtual model, the remote sensing image of the reference area is input, the simulation result of the water level in the reference area is output, and the mapping relationship between the remote sensing image and the water level measurement data is established; Step 104: using the overall water level distribution information of the lake provided by the remote sensing image, combined with the water level data of the reference area, a terrain-water level relationship model is established based on the mapping relationship between the remote sensing image and the water level measurement data.

5. The method for monitoring lake water level based on remote sensing images according to claim 1, characterized in that: The number of the reference areas is greater than or equal to 1 and is based on user settings.

6. The method for monitoring lake water level based on remote sensing images according to claim 1, characterized in that: Adjust the parameters of the terrain-water level relationship model, and use the optimization algorithm to search the parameter space to find the parameter combination that minimizes the prediction error of the terrain-water level relationship model.

7. The method for monitoring lake water level based on remote sensing images according to claim 1, characterized in that: The verification method of the terrain-water level relationship model includes the following steps: Step 201, obtaining the first credibility of the water level prediction result of the reference area: after several tests using the field verification method, the mean square error and coefficient of variation of the error between the predicted value and the field measurement value are obtained, which are recorded as k1 and k2 respectively; Perform a weighted summation on k1 and k2 and normalize them to the range of 0 to 1. The weight is determined according to actual needs to obtain the first credibility Ka; Step 202: If the first credibility meets the requirement, proceed to the next step; otherwise, issue an early warning, indicating that the terrain-water level relationship model is abnormal; Step 203, obtaining the second credibility of the prediction results of other regions except the reference region; After several tests using the field verification method, the mean square error and coefficient of variation of the mapping relationship between the output remote sensing image and the water level measurement data are denoted as k3 and k4 respectively. The weighted sum of k3 and k4 is performed and normalized to the range of 0 to 1. The weight is determined according to actual needs to obtain the second credibility Kb.

8. The method for monitoring lake water level based on remote sensing images according to claim 1, characterized in that: After obtaining the water level information of the target monitored lake, the step of correcting the water body boundary is also included: Used to correct the first water body boundary information corresponding to the first terrain model, and obtain the undetermined water body boundary area based on the first water body boundary information and remote sensing image information; transmit the remote sensing data corresponding to the undetermined water body boundary area to the trained water body boundary recognition model to realize the correction of the water body boundary and output the second water body boundary information; The water body boundary recognition mode reliability index Ls is obtained as follows: Record the marked water body boundary as the first vector data; Marking the water body boundary output by the water body boundary recognition model as second vector data; The first vector data and the second vector data are both sets containing coordinate points; Convert the vector data into raster data so that each raster cell has a corresponding value, and compare the first vector data with the second vector data. Assume that the first vector data after rasterization is R1, the second vector data after rasterization is R2, and the size of the raster is m×n; The reliability index Ls of the water body boundary recognition model is calculated by the following model ; ;in, represents the mean square error loss of the first vector data and the second vector data; i and j represent the row index and column index in the raster data respectively, i represents the row number in the raster data, counting from 1 to the total number of rows m in the raster; j represents the column number in the raster data, counting from 1 to the total number of columns n in the raster; and are the true value and model predicted value at the i-th row and j-th column of the grid respectively; The water body boundary correction step is connected with the boundary contour processing step: The boundary contour processing step refers to the post-processing of the segmentation results output by the water body boundary recognition model, including morphological operations to clean up noise and irregular edges; and using contour extraction algorithms to display the boundary contours of the water body.

9. The method for monitoring lake water level based on remote sensing images according to claim 8, characterized in that: The method further includes: a lake water level monitoring quality early warning step, combining the first credibility Ka, the second credibility Kb and the water body boundary recognition mode reliability index Ls to obtain a lake water level monitoring quality coefficient, and judging whether to issue an early warning to the user based on the relationship between the lake water level monitoring quality coefficient and a preset threshold, including: Combining the first credibility Ka, the second credibility Kb and the water body boundary recognition mode reliability index Ls, the lake water level monitoring quality coefficient SJ is calculated by the following formula; When the lake water level monitoring quality coefficient exceeds the threshold, an early warning message is sent to the user, indicating that there may be inaccurate monitoring data; When the lake water level monitoring quality coefficient does not exceed the threshold, it indicates that the lake water level monitoring is normal and no measures are taken.

10. A system for monitoring lake water levels based on remote sensing images, characterized in that: include: The first lake terrain model construction module selects the remote sensing image A of the lowest water level of the target lake from historical data and constructs a three-dimensional water model A; Use underwater equipment to obtain water depth and terrain information when the water level is not the lowest, and construct underwater three-dimensional model B; integrate the above-water three-dimensional model A and the underwater three-dimensional model B to output the first terrain model of the target monitored lake; The terrain-water level relationship model construction module selects a reference area from several areas of the target monitored lake, and uses the digital twin technology to establish a terrain-water level relationship model based on the first terrain model that has been constructed, combined with the remote sensing image of the reference area and the corresponding water level measurement data; The terrain-water level relationship model credibility verification module continuously adjusts and optimizes the terrain-water level relationship model parameters through multiple iterations, so that the terrain-water level relationship model can accurately reflect the relationship between lake terrain and water level, and verify the credibility of the terrain-water level relationship model; The terrain-water level relationship model credibility application module applies the acquired terrain-water level relationship model to the water level monitoring of the target monitoring lake when the credibility meets the requirements, and outputs the water level information of the target monitoring lake.

Images