Reservoir capacity curve rechecking method based on satellite height measurement and remote sensing imaging technology

Abstract

The invention discloses a reservoir capacity curve rechecking method based on satellite altimetry and remote sensing imaging technology, which belongs to the field of water conservancy and hydropower engineering and hydrology and water resource management, and comprises the following steps of: inverting and correcting the water level elevation of a reservoir by adopting altimetry data monitored by satellite remote sensing; recognizing a reservoir water body range based on a remote sensing image, and counting an area sequence; constructing a reservoir capacity calculation model based on the synchronous remote sensing water level and area sequence of the reservoir; remote sensing water level-area and water level-storage capacity curves are fitted, the precision is evaluated, and the process that the same satellite task is adopted to invert the same-period water level and area to reconstruct the storage capacity curve is achieved. By adopting the method, the defect that the water level and the area of the reservoir characteristic curve reconstructed by using satellite remote sensing data in the prior art are not synchronous can be avoided, the defects that the traditional height measurement satellite based on laser or synthetic aperture radar can only obtain the sub-satellite point cloud elevation and the step length is relatively long are overcome, and the precision of the reconstructed characteristic curve can be remarkably improved; the method has remarkable advantages in reconstructing characteristic curves of small reservoirs in remote areas and large ranges.

Description

Technical Field

[0001] The invention relates to the technical field of water conservancy and hydropower engineering and hydrology and water resources management, and in particular to a reservoir capacity curve verification method based on satellite altimetry and remote sensing imaging technology. Background Art

[0002] As an important water resource management tool, reservoirs play an important role in flood control, water supply, irrigation and power generation. The effective operation of reservoirs depends on a highly representative storage capacity curve, that is, an accurate quantitative relationship between water level and storage capacity. However, with the operation of completed reservoirs, the representativeness of the storage capacity curves of many reservoirs has been significantly reduced due to natural factors (such as siltation and river scouring) and human factors (such as artificial excavation of river sand and diversion of rivers). This may make it difficult to ensure the benefits of reservoir flood control, power generation and irrigation, and bring risks to the scheduling of reservoir groups. How to achieve convenient batch review of storage capacity characteristic curves is an important issue that needs to be urgently solved in the current project operation management.

[0003] At present, the traditional method of reconstructing reservoir capacity curves mainly relies on field measurements. This usually involves the use of underwater acoustic equipment, laser rangefinders or other similar tools to measure the water area at various elevations of the reservoir. The characteristic curve reconstructed based on field measurement data is highly accurate and reliable, but this type of method also has many limitations. First, most reservoirs are located in remote mountainous areas (even uninhabited areas), and field surveys are time-consuming and labor-intensive, with huge investment costs; secondly, the field measurement method may not be implemented in some special circumstances (such as in bad weather or areas with safety hazards around the reservoir), and the field measurement method is also difficult to conveniently and quickly batch review the reservoir characteristic curve. Therefore, it is necessary to seek a more efficient and safe method to update the reservoir capacity curve.

[0004] In recent years, the rapid development of satellite remote sensing technology has provided new ideas for updating reservoir capacity curves. Satellite remote sensing technology can provide a wide range of surface coverage data, including the area and elevation information of water bodies. Combining satellite altimetry data with high-resolution remote sensing images, the water level and area of ​​reservoirs can be remotely monitored without direct contact with the water body. This method can not only improve the efficiency and safety of data collection, but also significantly reduce costs, especially for areas (reservoirs) that are difficult to measure using traditional methods. Summary of the invention

[0005] The purpose of the present invention is to provide a reservoir capacity curve verification method based on satellite altimetry and remote sensing imaging technology to solve the problems existing in the background technology.

[0006] To achieve the above object, the present invention provides a reservoir capacity curve verification method based on satellite altimetry and remote sensing imaging technology, comprising the following steps: S1, obtain the remote sensing data set above the target reservoir, preprocess the pixel-level data, and construct the reservoir water mask; S2. Obtain high-precision DEM, JRC Monthly Water History (JMWH), and high-resolution satellite altimetry data of the target reservoir area, and use reservoir characteristic water levels (such as dead water level and design flood level) and filtering methods such as straight-through filtering, radius filtering, and statistical filtering to denoise the inverted water level point cloud data to obtain accurate and reliable water level data; S3. Combine the water level, water index, DEM and JMWH data of the reservoir in the same period to build a machine learning model, identify the range of water bodies in the image, extract the water body boundary of the reservoir surface, further estimate the reservoir area, fit the reservoir water level-area relationship and evaluate the goodness of fit; S4. Obtain the water level sequence and reservoir surface area sequence of the same period through high-resolution satellite altimetry and remote sensing images, build a reservoir capacity calculation model based on the two, and calculate the reservoir water storage data; S5. Based on the reservoir storage data, polynomial, power function and exponential function are used to fit the water level-area curve and water level-storage capacity curve, and the goodness of fit is evaluated to optimize the fitting relationship curve.

[0007] Preferably, satellite remote sensing data is obtained, and high-resolution satellite point cloud altimetry data is preprocessed, including statistical filtering, straight-through filtering and DBSCAN, etc., and image raster data is orthorectified, radiometrically calibrated, atmospheric corrected and hillshade processed, etc. The preprocessing also includes system bias correction and aggregation denoising. This process aims to ensure that the point cloud data within the reservoir mask is less affected by noise and has higher accuracy and consistency through some common quality control methods.

[0008] Preferably, when extracting the surface area of ​​the reservoir, the reservoir mask boundary is obtained based on the global dam database, high-precision DEM of the reservoir area and mask technology, SWOT point cloud data is obtained, and water level data (including water level data uncertainty, water level quality indicators and black water range, etc.) is obtained based on the basic information of the reservoir, the water surface range of the reservoir in the same period and its mask boundary.

[0009] Preferably, the pixel information within the reservoir boundary mask is obtained, and the pixel area measurement quality index is used for screening and correction. The index set is constructed based on the high-precision DEM data of the reservoir area, the height measurement point cloud data obtained after denoising and various correction preprocessing, the dark water mark of the mask area, the JMWH of the reservoir area, the dark water ratio, etc., and the training set and validation set are selected to train and verify the classification accuracy of the machine learning model. The machine learning method is used twice. The first time is to combine the basic point cloud denoising algorithm, the basic characteristic water level of the reservoir, JMWH and the high-precision DEM to remove the noise in the point cloud height measurement data, obtain the accurate reservoir water level and assist in obtaining the reservoir surface area; the second time is used to accurately obtain the reservoir surface area.

[0010] Preferably, after the remote sensing reservoir surface area and high-resolution satellite altimetry water level data are extracted and preprocessed, a remote sensing inversion area-altimetry water level sequence is constructed, and the digital elevation model DEM is used to extend the high water level area data, making the altimetry water level-remote sensing area sequence more complete and reliable.

[0011] Preferably, the machine learning model is a supervised learning model, an unsupervised learning model or a reinforcement learning model, etc., the purpose of which is to integrate multiple indicators to distinguish whether it belongs to water area or non-water area (distinguish whether the height measurement point cloud data is noise). Fully consider the information of the given classification indicator set, reduce the noise interference that is not eliminated by preprocessing, and make up for the defect of easy misjudgment of single indicator classification, so as to effectively identify the water level and reservoir surface area. The feature vector contains the image itself information (each band and water body index), reservoir terrain information (DEM), reservoir area inversion water level and time period image information (JMWH), etc.

[0012] Preferably, the reservoir capacity calculation model determines whether the target reservoir is a trapezoidal body or a prism body, and then divides the water body into n Layer, reservoir capacity data is n It is calculated by accumulating the volumes of layered trapezoids or prisms.

[0013] Preferably, the trapezoidal formula of the reservoir capacity calculation model is: ; The prism formula is: ; The cumulative storage capacity formula is: ; In the formula, is the reservoir tolerance between two adjacent water levels; is the water level difference between two adjacent water levels; , are the water surface areas corresponding to two adjacent water levels; is an ordinal number; is the cumulative number; is the initial storage capacity; is the cumulative reservoir volume.

[0014] The latest operating Surface Water and Ocean Topography (SWOT) satellite mission monitoring data can obtain the water level and area of ​​reservoirs over a large range during the same period, overcoming the shortcomings of traditional imaging satellites that can only obtain reservoir surface area and radar or laser satellites that have a large footprint and can only obtain the water level of reservoirs under the satellite. It is highly applicable to reservoirs with no measured data or missing or erroneous measured data.

[0015] The synthetic aperture radar (KaRIn radar interferometer) carried by the SWOT satellite can acquire data at a high-speed mode of 20 m along the track and 10-60 m across the track. For the possible data loss or misjudgment in the narrow and long section of the reservoir, the recognition capability can be enhanced based on DEM and JMWH data to obtain a more accurate reservoir area. The satellite remote sensing images and altimetry data for classification and identification mainly use the observation data of the wide-swath radar interferometer (KaRIn), and other image data (Landsat 8 / 9, Sentinel-1 / 2) are important supplements for reservoir area inversion.

[0016] Therefore, the present invention adopts the above-mentioned reservoir capacity curve verification method based on satellite altimetry and remote sensing imaging technology, which has the following beneficial effects: (1) Realize the process of reconstructing reservoir capacity curve by inverting water level and area in the same period using the same satellite mission (multiple-source satellite image data can be used as a supplement); (2) The monitoring range is wide and the verification cost is low. The data is obtained from satellite observation data, and no longer relies on time-consuming and labor-intensive ground hydrological monitoring and engineering measurements. It is particularly suitable for reservoirs where traditional measurement methods are difficult to apply. (3) It is of great significance to improve the efficiency and accuracy of reservoir management, especially in flood control, power generation and water supply.

[0017] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is an overall flow chart of an embodiment of the present invention; Figure 2 This is a schematic diagram of a reservoir water level data sequence according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the calculation results of water level-reservoir capacity data according to an embodiment of the present invention; Figure 4 This is a schematic diagram of a water level-area curve after reconstruction according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the water level-reservoir capacity curve after reconstruction according to an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. 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.

[0020] See also Figure 1A reservoir capacity curve verification method based on satellite altimetry and remote sensing imaging technology includes the following steps: S1. Obtain the remote sensing data set above the target reservoir, preprocess the pixel-level data, and construct a reservoir water mask. The water level data and water surface area data of the same period are obtained through high-precision satellite altimetry and remote sensing imaging technology. Due to its high temporal and spatial resolution, hundreds of thousands of pixel-level observations of the same water body in the same period, and complete time series characteristics, the observation data is more reliable and the fitted reservoir capacity curve is more representative. This fitting process uses the data points of continuous observations, thereby improving the accuracy and reliability of the model.

[0021] S2. Obtain high-precision DEM of the target reservoir area, JRC Monthly Water History (JMWH), and high-resolution satellite altimetry data of the same period of remote sensing images, and extract water level data and reservoir surface area data. Combine the reservoir characteristic water level (such as dead water level and design flood level, etc.) and filtering methods such as straight-through filtering, radius filtering and statistical filtering to denoise the inverted water point cloud.

[0022] Satellite altimetry data is obtained, high-resolution satellite altimetry data is preprocessed, and the point cloud data is subjected to multiple denoising processes such as pass filtering, statistical filtering, and radius filtering. The preprocessing also includes system bias correction and aggregation denoising. This process aims to ensure that the point cloud data within the reservoir mask is less affected by noise, has high accuracy and consistency, and can build a machine learning model to coordinate various features.

[0023] Remote sensing images and satellite altimetry data are mainly derived from synchronous observation data from the same satellite. When the synchronous observation data, especially the reservoir area, is of poor quality, it is interpolated from the inverted area of ​​other satellites that are closest in time, ensuring the high consistency of the extracted water area and water level data in the temporal and spatial dimensions. By using data from the same observation source and its quality indicators, not only can the data be corrected more accurately and effectively, but also the temporal and spatial deviations that may be introduced due to differences in data sources can be effectively avoided, enhancing the accuracy and credibility of the research results.

[0024] S3. Combine the water level, water body index, DEM and JMWH data of the reservoir in the same period to build a machine learning model, identify the range of water bodies in the image, extract the water body boundary of the reservoir surface, use multiple types of functions to fit the reservoir water level-area relationship and evaluate the goodness of fit, and select the best water level-area relationship.

[0025] When extracting the surface area of ​​the reservoir, the reservoir mask boundary is obtained based on the global dam database, high-precision DEM of the reservoir area and mask technology, and the SWOT point cloud data is obtained. The water level is inverted based on the reservoir characteristic water level, high-precision DEM and JMWH.

[0026] Obtain pixel information within the reservoir boundary mask, screen and correct according to pixel area measurement quality indicators, construct feature vectors based on various indicators and high-precision DEM data, and then build a machine learning model to identify and extract the water body range of the reservoir. Construct an indicator set based on high-precision DEM raster images of the reservoir area, mask area height measurement point cloud data that has been denoised and pre-processed with various corrections, mask area dark water markers, dark water ratios, etc., select training sets and validation sets to train and verify the classification accuracy of the machine learning model. Use machine learning methods twice. The first time is to combine the basic point cloud denoising algorithm to remove noise points in the point cloud height measurement data, obtain accurate reservoir water levels and assist in obtaining the reservoir surface area; the second time is used to accurately obtain the reservoir surface area.

[0027] The machine learning model is a supervised learning model, an unsupervised learning model, or a reinforcement learning model, etc. Its purpose is to integrate multiple indicators to distinguish whether it belongs to water areas or non-water areas (whether the height measurement point cloud is a noise point). It fully considers the information of the given classification indicator set, reduces the noise interference that is not eliminated by preprocessing, and makes up for the defect of easy misjudgment of single indicator classification, thereby effectively identifying the water level and reservoir surface area.

[0028] S4. Use high-resolution satellite altimetry and remote sensing images to obtain the water level series and reservoir surface area series for the same period respectively, and build a reservoir capacity calculation model based on the two to calculate the reservoir water storage data.

[0029] S5. Based on the reservoir storage data, polynomial, power function and exponential function are used to fit the water level-area curve and water level-storage capacity curve, and the goodness of fit is evaluated to optimize the fitting relationship curve.

[0030] After the remote sensing reservoir surface area and high-resolution satellite altimetry water level data are extracted, the altimetry water level-remote sensing area relationship is fitted and the optimal relationship is evaluated and selected. The digital elevation model DEM is used to extend the high water level area data, making the altimetry water level-remote sensing area sequence more complete and reliable.

[0031] The reservoir capacity calculation model determines whether the target reservoir is a trapezoidal body or a prism body, and then divides the water body into n Layer, reservoir capacity data is n The volume of the trapezoidal body or prism body is accumulated. The trapezoidal formula of the reservoir capacity calculation model is: ; The prism formula is: ; The cumulative storage capacity formula is: ; In the formula, is the reservoir tolerance between two adjacent water levels; is the water level difference between two adjacent water levels; , are the water surface areas corresponding to two adjacent water levels respectively; is the ordinal number; is the cumulative number; is the initial reservoir capacity; is the cumulative reservoir volume.

[0032] The monitoring data of the latest operating Surface Water and Ocean Topography (SWOT) satellite mission can obtain the water levels and areas of reservoirs over a large range, overcoming the shortcomings of traditional imaging satellites that can only obtain the reservoir surface area, and the footprints of radar or lidar satellites are too large, that is, they can only obtain the reservoir water level at the sub-satellite point. It has high applicability to reservoirs without measured data or with missing or incorrect measured data.

[0033] The synthetic aperture radar (KaRIn radar interferometer) carried by the SWOT satellite can obtain data in a high-speed mode with a cross-track resolution of 10 - 60 m and an along-track resolution of 20 m. For possible data missing or misjudgment in the narrow section of the reservoir, the DEM data can be used to enhance the recognition ability and obtain a more accurate reservoir area. The satellite remote sensing images and altimetry data for classification and recognition mainly use the simulated observation data of the wide-swath radar interferometer (KaRIn) (SWOT-like data, SWORD).

[0034] Embodiment: Taking the Three Gorges Reservoir in the Yangtze River Basin as an example, the reservoir capacity curve review method is applied in practice. Specifically, machine learning algorithms represented by random forest, extraction of the surface area sequence of the Three Gorges Reservoir based on SWOT-like satellite remote sensing images, extraction of the water level sequence of the Three Gorges Reservoir based on SWOT-like satellite altimetry data, acquisition of altimetry water level - remote sensing area data points based on DEM extended area data at high water levels, fitting of the water level - area function relationship, fitting and evaluation of the water level - reservoir capacity function relationship based on the reservoir capacity model, etc. are introduced. The Three Gorges Reservoir began to impound water in 2003, and the water level - reservoir capacity relationship has used the design curve so far and has been in operation for many years. Due to environmental changes such as sediment deposition, the current reservoir capacity is likely to have changed.

[0035] First, a machine learning classification algorithm represented by random forest is constructed: 1) Extract a sample with a sample size of N, and draw N times with replacement, each time drawing 1, and finally form N samples. These N samples are used to train a decision tree as the sample at the root node of the decision tree.

[0036] 2) When each sample has M attributes, when each node of the decision tree needs to be split, randomly select m attributes from these M attributes, satisfying the condition m << M. Then, select 1 attribute from these m attributes using a certain strategy (such as information gain) as the splitting attribute of this node.

[0037] 3) During the decision tree formation process, each node must be split according to step 2) until it can no longer be split.

[0038] 4) According to the first three steps, a large number of decision trees are established to form a random forest.

[0039] The second step is to obtain reservoir surface area data. This step involves selecting appropriate remote sensing image data and determining the time range of the image data to be 10 years from 2009 to 2018 according to research needs. Subsequently, remote sensing images provided by SWOT-like satellite data are obtained. Next, the remote sensing images are subjected to necessary image correction and denoising, and the target water body range is obtained using machine learning methods based on the elimination of low-quality pixels according to quality indicators. Thereafter, the reservoir area is divided into cells, and the standard size of each cell is determined, and the cells at the boundary are appropriately adjusted to ensure that the entire reservoir area is accurately covered. After the area of ​​each cell is calculated, the total number of cells is counted and the total area is calculated. Finally, the extracted area data are arranged in chronological order to form a reservoir surface area sequence, and the extracted data are filtered and verified to ensure the accuracy and reliability of the data.

[0040] This example obtains SWOT-like remote sensing image data of the Three Gorges Reservoir in the Yangtze River Basin. The selected images basically cover the regular water level range of the Three Gorges Reservoir from the dead water level to the normal water level. The random forest classification algorithm is used to extract and count the water area of ​​the reservoir from the selected images. Some data such as Figure 2 shown.

[0041] The next step is to obtain the reservoir water level data. From the SWOT-like data set provided by the SWOT simulator, extract the height measurement data at the same time as the remote sensing image, and denoise and correct the obtained height measurement data to convert it into a usable water level data series, such as Figure 2 shown.

[0042] Then, the elevation-area relationship is further constructed and optimized. The work in this stage involves collecting reservoir elevation and area data at the same time point, and integrating these data into a scattered point set to form an initial elevation-area data set. After comparing and analyzing the scattered point data, remove the erroneous or unreasonable scattered points in the data set. Then, basic elementary functions such as power functions, exponential functions, and constant functions are selected as the basis of the fitting function, and parameter functions are constructed through rational operations and function compound operations. The least squares method is used to fit the constructed parameter function to determine the values ​​of each parameter, thereby obtaining an accurate area-elevation relationship. These steps ensure the mathematical accuracy and scientificity of the area-elevation relationship and are the basis for constructing the reservoir capacity curve.

[0043] Calculation of the Three Gorges water level-area curve: Data fitting was performed by using the acquired water level and area data. Based on the data fitting effect and fitting parameter values, the exponential function, power function, and second and third degree polynomial functions with higher precision were selected to obtain the relationship and fitting curve. The goodness of fit values ​​are shown in Table 1. By comparing the precision of the fitting functions, the third degree polynomial function with the highest precision index was selected.

[0044] Table 1 Goodness of fit values ​​of exponential function, power function and quadratic and cubic polynomial functions ;

[0045] The water level-area curve of the Three Gorges Reservoir is obtained as follows: Figure 3 As shown, the fitting function is: ; Where: A is the reservoir area, km 2 ; Z is the reservoir water level, m.

[0046] In order to be consistent with the water level range of the original water level-reservoir capacity curve of the Three Gorges Reservoir, the curve was extended to 130 meters according to the fitting function to obtain the complete water level-area curve.

[0047] Finally, constructing the reservoir capacity curve is the ultimate goal of the entire process. Since the Three Gorges Reservoir is a narrow and long river reservoir, the prism formula is selected as the reservoir capacity calculation model. The change in water storage at the interval of slightly increased water level (1 m) is calculated with the help of the trapezoidal formula, and then accumulated into the reservoir capacity. After that, a suitable mathematical function is selected to describe the change of reservoir capacity with water level, and a function describing the change of reservoir capacity is constructed. The constructed reservoir capacity curve is verified and adjusted if necessary to improve its accuracy and reliability, and compared with the design curve.

[0048] Calculation of Three Gorges Reservoir Capacity: The Three Gorges Reservoir is a long and narrow reservoir. The prism method and prism formula are used to calculate the reservoir interval capacity of the conventional water level interval of 145-175 m in the actual operation of the Three Gorges Reservoir. Then the water level-capacity data are accumulated at different water levels. The calculation results are as follows: Figure 4 shown.

[0049] The water level-reservoir capacity data to be fitted were obtained by calculating the water level-area data using the prism method. The water level-reservoir capacity data of the Three Gorges Reservoir were fitted using the same fitting process as the water level-area data. The exponential function, power function, and quadratic and cubic polynomial functions were selected as the fitting functions for the reservoir capacity data. By comparing the accuracy of each fitting function and taking into account the complexity and high fitting risk of the cubic polynomial function, the quadratic polynomial function whose accuracy index is second only to the cubic polynomial function was selected for the reservoir capacity relationship curve, as shown in Table 2.

[0050] Table 2 The fitting accuracy of exponential function, power function and quadratic and cubic polynomial functions ;

[0051] The final water level-storage capacity curve of the Three Gorges Reservoir is as follows: Figure 5 As shown, the fitting function relationship is: ; Where: V is the reservoir capacity, billion m³; Z is the reservoir water level, m.

[0052] In order to ensure the completeness of the water level-reservoir capacity curve, the curve was extended to the 130 m water level using the fitting function.

[0053] Through the above steps, the reservoir capacity curve of the Three Gorges Hydropower Station was reconstructed and verified, which provided strong support for the effective management of the reservoir. The proposed verification method improved the dispatching benefits of flood control, power generation and water supply, and enhanced the efficiency of reservoir operation and management.

[0054] Therefore, the present invention adopts the above-mentioned reservoir capacity curve verification method based on satellite altimetry and remote sensing imaging technology, which can avoid the defect of asynchronous water level and area in reconstructing the reservoir characteristic curve using satellite remote sensing data in the past. In addition, the water level is inverted by using wide-swath satellite-borne radar interferometric altimetry (quasi-) SWOT satellite data, which overcomes the defects of traditional laser-based synthetic aperture radar (SAR) altimetry satellites that can only obtain the elevation of the sub-satellite point and have a long step length. It can not only significantly improve the accuracy of the reconstructed characteristic curve, but also has great significance for the reconstruction of the characteristic curve of small reservoirs in remote areas and over a large range.

[0055] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A reservoir capacity curve verification method based on satellite altimetry and remote sensing imaging technology, characterized in that: The following steps are involved: S1, obtain the remote sensing data set above the target reservoir, preprocess the pixel-level data, and construct the reservoir water mask; S2. Obtain high-precision DEM, JRC Monthly Water History, and high-resolution satellite altimetry data of the target reservoir area, and extract water level data and reservoir surface area data; S3, based on the data in S2, build a machine learning model to identify the range of water bodies, extract the boundaries of reservoir water bodies, and further estimate the surface area of ​​the reservoir. Use a multi-class point cloud denoising algorithm and the results obtained in S2 to screen and denoise the height measurement point cloud data to obtain accurate reservoir surface water levels. Use a variety of fitting function models to fit the relationship between reservoir water level and area and evaluate the goodness of fit. S4. Obtain the water level sequence and reservoir surface area sequence of the same period through high-resolution satellite altimetry and remote sensing images respectively. Based on the reservoir surface area sequence and water level sequence of the same period, construct a reservoir capacity calculation model and calculate the reservoir water storage data; S5. Based on the reservoir storage data, a variety of fitting function models are used to fit the water level-storage capacity curve and evaluate the goodness of fit.

2. The reservoir capacity curve verification method based on satellite altimetry and remote sensing imaging technology according to claim 1 is characterized by: Acquire satellite remote sensing data, preprocess high-resolution satellite point cloud altimetry data, including statistical filtering, pass filtering and DBSCAN, perform orthorectification, radiation calibration, atmospheric correction and mountain shadow processing on image raster data, and preprocessing also includes system bias correction and aggregation denoising.

3. The reservoir capacity curve verification method based on satellite altimetry and remote sensing imaging technology according to claim 2 is characterized by: When extracting the surface area of ​​the reservoir, the reservoir mask boundary is obtained based on the global dam database, high-precision DEM of the reservoir area and mask technology, and the SWOT point cloud data is obtained. The water level data is obtained based on the basic information of the reservoir, the water surface range of the reservoir in the same period and its mask boundary.

4. The reservoir capacity curve verification method based on satellite altimetry and remote sensing imaging technology according to claim 3 is characterized by: The pixel information within the reservoir boundary mask is obtained, and the pixel area measurement quality index is used for screening and correction. The feature vector is constructed based on the indexes, reservoir area JMWH and high-precision DEM data, and then a machine learning model is constructed to identify and extract the water body range of the reservoir.

5. The reservoir capacity curve verification method based on satellite altimetry and remote sensing imaging technology according to claim 4 is characterized in that: After the remote sensing reservoir surface area and high-resolution satellite altimetry water level data were extracted and preprocessed, a remote sensing inversion area-altimetry water level sequence was constructed, and the digital elevation model DEM was used to extend the high water level area data.

6. The reservoir capacity curve verification method based on satellite altimetry and remote sensing imaging technology according to claim 5 is characterized by: The machine learning model is a supervised learning model, an unsupervised learning model or a reinforcement learning model, and the feature vector includes the image information itself, the reservoir topography information, the inverted water level of the reservoir area and the time period image information.

7. The reservoir capacity curve verification method based on satellite altimetry and remote sensing imaging technology according to claim 1 is characterized by: The reservoir capacity calculation model determines whether the target reservoir is a trapezoidal body or a prism body according to the situation, and then divides the water body into n layers according to different water levels. The reservoir capacity data is obtained by accumulating the volume of n layers of trapezoidal bodies or prism bodies.

8. The reservoir capacity curve verification method based on satellite altimetry and remote sensing imaging technology according to claim 7 is characterized by: The trapezoidal formula of the reservoir capacity calculation model is: ； The prism formula is: ； The cumulative storage capacity formula is: ； In the formula, is the reservoir tolerance between two adjacent water levels; is the water level difference between two adjacent water levels; , are the reservoir areas corresponding to two adjacent water levels; is an ordinal number; is the cumulative number; is the initial storage capacity; is the cumulative reservoir volume.

Images