A method, system, electronic device and storage medium for extracting river channel cross-sections

Abstract

The present invention relates to a method, system, electronic device and storage medium for extracting river cross-sections, belonging to the field of hydrological remote sensing, and includes: obtaining satellite data to obtain the original water level-river width dataset and historical river width dataset of the river cross-section; screening the original water level-river width dataset to obtain the water level-river width dataset; using the historical river width dataset to obtain the median river width, combining the existing median runoff and slope to obtain the median cross-sectional area; using the median river width to obtain the median water level from the water level-river width dataset; obtaining the low water level control point according to the median river width, the median cross-sectional area and the median water level; obtaining the high water level control point according to the water level-river width dataset; generating the water level-river width hydrograph of the river cross-section according to the water level-river width dataset, the low water level control point and the high water level control point, and completing the extraction of the river cross-section.

Description

Technical Field

[0001] The present invention relates to a method, a system, an electronic device and a storage medium for extracting a river cross-section, belonging to the field of hydrological remote sensing. Background Art

[0002] Rivers are an important part of the global water cycle process, and river runoff estimation plays an important role in preventing flood disasters, protecting biodiversity, etc. The scope of runoff monitoring by traditional hydrological stations is limited, while remote sensing technology, thanks to the characteristics of large-scale continuous observation, makes it possible to obtain spatio-temporally continuous runoff estimation at a large scale. Using satellite remote sensing to obtain water levels or river widths and then inversely calculating runoff is one of the widely used runoff estimation methods. The accuracy of this method depends on river cross-section information. However, most existing studies obtain high-precision river cross-section information through local hydrological stations or unmanned aerial vehicles, which limits the application of remote sensing runoff estimation in data-deficient areas such as western China.

[0003] In recent years, with the development of remote sensing technology and the improvement of computing power, researchers have been able to obtain long-term water level and river width data with high spatial resolution at the global scale. Under this background, some researchers have proposed to rely only on multi-source satellite remote sensing, couple observation elements such as water levels and river widths, and construct a water level-river width hydrological curve by solving or interpolating all water level-river width data pairs to achieve river cross-section reconstruction, and estimate runoff based on this. However, remote sensing data has high uncertainty, and the cross-section above the water surface that can be directly observed often has a limited range. Directly using unquality-controlled remote sensing data and extrapolating the underwater cross-section by mathematical methods may produce large deviations.

[0004] Aiming at the problems that remote sensing cannot observe the underwater terrain, the quality of remote sensing data is uneven, and the uncertainty of directly extrapolating the underwater cross-section is relatively large, the present invention proposes a method for extracting a river cross-section by combining remote sensing altimetry and prior knowledge, performs a series of quality controls on remote sensing data, introduces prior knowledge to constrain uncertainty in the process of constructing a water level-river width hydrological curve, which is of great significance for improving the accuracy of remote sensing-estimated runoff in practice.

[0005] Chinese invention patent CN112729258B discloses a method for continuously measuring river flow based on satellite big data:

[0006] The first step is to select a flow measurement cross-section and select a slope cross-section at a certain distance upstream and downstream respectively.

[0007] Second, based on the altimetry satellite or three-dimensional imaging mapping satellite, establish the water level-time process line graph of the flow measurement section, and establish the river width-time process line graph of the flow measurement section according to the ortho-remote sensing image; interpolate the river width at the water level observed by the satellite in the water level-time process line graph, and interpolate the water level at the river width observed by the satellite in the river width-time process line graph; based on all the interpolated water level-river width data points, establish the water level-river width relationship function of the flow measurement section; similarly, obtain the water level-river width relationship function of the slope section.

[0008] Third, through the real-time data of the ortho-remote sensing image satellite, measure the river width data of the flow measurement section and the upper and lower slope sections, and calculate the real-time water level according to the water level-river width relationship; refer to the natural channel roughness coefficient table in the hydraulics textbook to obtain the roughness coefficient.

[0009] Fourth, divide the water level difference between the upper and lower slope sections by the distance to obtain the slope; according to the water level-river width relationship function of the flow measurement section, extrapolate the lowest water level by methods such as triangular similarity (i.e., approximately represent the underwater terrain that cannot be observed with a triangle); according to the complete water level-river width relationship function, obtain the cross-sectional area and hydraulic radius at any water level; calculate the flow corresponding to the real-time water level of the flow measurement section according to the Manning formula of hydraulics.

[0010] However, this Chinese invention patent has the following disadvantages:

[0011] First, there is a lack of necessary quality control for remote sensing data. The water level and river width data obtained from satellite remote sensing may have large measurement errors and need to be screened before use, but this is ignored in the prior art.

[0012] Second, the extrapolation of the underwater terrain that cannot be observed is too simple. The prior art only relies on the lowest two water level-river width observation data points and extrapolates the lowest point of the section by the method of triangular similarity, ignoring the continuity of the above-water and underwater parts of the section.

[0013] Third, the prior art does not explain how to obtain a continuous water level-river width hydrological curve from discrete water level-river width data points. Due to data quality reasons, there are often consistency problems in water level-river width data, that is, abnormal situations such as the river width increasing but the water level decreasing may occur. Therefore, it is not reasonable to directly perform linear interpolation on all data points.

[0014] In view of the above disadvantages, the present invention first proposes a series of quality control methods for remote sensing data, aiming to screen out accurate and reliable data. Secondly, for the underwater terrain that cannot be observed, the present invention introduces prior knowledge obtained from multi-source geographical data to construct two control points, high and low, so as to constrain the uncertainty of section extraction. Finally, by non-linearly regressing and fitting the data points, a continuous water level-river width hydrological curve is obtained. Summary of the Invention

[0015] In view of the limitation that it is difficult to directly observe underwater topography by remote sensing, the present invention proposes a method for extracting river channel cross-sections by combining remote sensing altimetry and prior knowledge, which is used to characterize the shape characteristics of river channel cross-sections. Its feature lies in extrapolating the coupled remote sensing observed water level-river width data points to the underwater area that cannot be directly observed, and by controlling the quality of remote sensing data and introducing prior knowledge to constrain uncertainties, it provides scientific and technological support for further improving the accuracy of remote sensing estimated runoff in data-deficient areas.

[0016] The first aspect of the present invention provides a method for extracting river channel cross-sections, including:

[0017] Obtaining satellite data to obtain the original water level-river width dataset and historical river width dataset of the river channel cross-section;

[0018] Screening the original water level-river width dataset to obtain the water level-river width dataset;

[0019] Using the historical river width dataset to obtain the median river width, and combining the existing median runoff and slope to obtain the median cross-sectional area;

[0020] Using the median river width to obtain the median water level from the water level-river width dataset;

[0021] Obtaining low water level control points according to the median river width, the median cross-sectional area and the median water level;

[0022] Obtaining high water level control points according to the water level-river width dataset;

[0023] Generating the water level-river width hydrograph of the river channel cross-section according to the water level-river width dataset, the low water level control points and the high water level control points to complete the extraction of the river channel cross-section.

[0024] Furthermore, the screening includes screening according to uncertainty, screening according to reverse order relationship and screening according to outliers.

[0025] Furthermore, using the historical river width dataset to determine the extraction range of the river channel cross-section, and further obtaining the median river width.

[0026] Furthermore, calculating the river width corresponding to the high water level control point and the river width corresponding to the low water level control point by using the historical river width dataset, and determining the extraction range of the river channel cross-section according to the two.

[0027] Furthermore, obtaining the median runoff through the existing global runoff simulation dataset, obtaining the slope through the existing global river attribute dataset, and obtaining the median cross-sectional area through the Manning formula.

[0028] Further, the water level of the high water level control point is the highest water level in the water level-river width dataset plus a set height difference.

[0029] Further, the water level-river width hydrograph is fitted by a power function.

[0030] The second aspect of the present invention provides a river cross-section extraction system, including:

[0031] A data acquisition module for acquiring satellite data to obtain the original water level-river width dataset and the historical river width dataset of the river cross-section;

[0032] A data screening module for screening the original water level-river width dataset to obtain a water level-river width dataset;

[0033] A parameter calculation module for obtaining the median river width using the historical river width dataset, combining the existing median runoff and slope to obtain the median cross-sectional area; using the median river width to obtain the median water level from the water level-river width dataset;

[0034] A control point calculation module for obtaining a low water level control point based on the median river width, the median cross-sectional area, and the median water level; obtaining a high water level control point based on the water level-river width dataset;

[0035] A river cross-section extraction module for generating the water level-river width hydrograph of the river cross-section based on the water level-river width dataset, the low water level control point, and the high water level control point to complete the extraction of the river cross-section.

[0036] The third aspect of the present invention provides an electronic device, including a memory and a processor, the memory stores a computer program, the computer program is configured to be executed by the processor, and the computer program includes instructions for executing the above method.

[0037] The fourth aspect of the present invention provides a storage medium storing a computer program, and when the computer program is executed by a computer, the above method is implemented.

[0038] The beneficial effects of the present invention are as follows:

[0039] 1. The prior art lacks necessary quality control for remote sensing data. The present invention proposes a set of quality control methods for water level-river width data to avoid the influence of unreliable data on subsequent cross-section extraction.

[0040] 2. The extrapolation of the unobservable underwater terrain in the prior art is too simple. The present invention uses a variety of prior knowledge to propose a method for calculating high and low water level control points, effectively constraining the uncertainty of the water level-river width hydrograph.

[0041] 3. The prior art does not explain how to obtain a continuous water level-channel width hydrograph from discrete water level-channel width data points. The present invention proposes to use weighted non-linear regression to fit the water level-channel width hydrograph, ensuring the monotonicity of the curve. At the same time, constraints on high and low water level control points are added to ensure that the extrapolation of the hydrograph from water level-channel width sample data is more reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a flowchart of an embodiment of the present invention.

[0043] Figure 2 It is a schematic diagram of the historical channel width distribution histogram calculated in an embodiment of the present invention.

[0044] Figure 3 It is a comparison diagram of the water level-channel width hydrograph calculated in an embodiment of the present invention and the measured results.

[0045] Figure 4 It is a framework diagram of the channel cross-section extraction system in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] The present invention will be further described in detail below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0047] An embodiment of the present invention provides a method for extracting a channel cross-section, and the specific process is as Figure 1 shown.

[0048] Step 1: Obtain the original data for extracting the channel cross-section based on remote sensing satellite observation data.

[0049] 1.1 Select the channel cross-section to be studied. Among the static river nodes predefined by the Surface Water and Ocean Topography Satellite (SWOT), determine the node that is spatially closest to the channel cross-section, and obtain the temporally coupled water level data and channel width data from this node, as well as the corresponding uncertainty estimates, to obtain the original water level-channel width data set of the channel cross-section.

[0050] 1.2 Obtain the ortho-remote sensing images from 1984 to 2023 from the Landsat satellite, and extract the water body pixels through the water body classification algorithm; based on the existing channel center line data, construct a buffer zone with a size of twice the average channel width at the cross-section within the range of 300 meters upstream and downstream of the channel cross-section, and divide the water body area in the buffer zone by the 600-meter buffer zone length to obtain the historical channel width data set of the channel cross-section.

[0051] Step 2: Obtain the prior knowledge of the channel based on multi-source geographical data.

[0052] 2.1 Using the historical river width dataset obtained in step 1.2, construct a histogram of river width distribution with 20 groups, and use the upper boundary of the last group with a frequency greater than 3% as the high water level river width , and use the lower boundary of the first group with a frequency greater than 3% as the low water level river width , where the frequency is the percentage of the number of river width observations in the data to the total number of observations. The high water level river width and the low water level river width The interval between the two is used as the extraction range of the river cross-section. As shown in Figure 2 , use the historical river width dataset to calculate the median river width of the river cross-section . Through this step, the research scope is controlled within the common width of the river.

[0053] 2.2 Using the existing global runoff simulation dataset, calculate the median runoff ; using the existing global river attribute dataset, obtain the slope of the river cross-section ; the Manning roughness coefficient is empirically set to 0.035.

[0054] Step 3 performs quality control on the original water level-river width dataset.

[0055] 3.1 According to the uncertainty estimate obtained in step 1.1, remove the water level-river width data in the original water level-river width dataset obtained in step 1.1 where the relative river width uncertainty is greater than 0.4 or the water level uncertainty is greater than 0.1 m. The purpose of this step is to remove the water level-river width data with excessive measurement uncertainty.

[0056] 3.2 For any two water level-river width data i and j in the water level-river width data screened in step 3.1, if and , or and , then define the relationship between the two data as an "inverse order" relationship. After traversing and calculating the pairwise relationships of all water level-river width data, remove the water level-river width data with the most inverse order relationships. Repeat step 3.2 until the inverse order ratio in the relationships between each water level-river width data and all other data is less than 50%, where the inverse order ratio is the ratio of the number of inverse order relationships between a data and all other data to the total number of relationships between the data and all other data. The purpose of this step is to remove the water level-river width data whose relative size relationship with most data is inconsistent and is more likely to be biased.

[0057] 3.3 Calculate the sample median water level of the water level-river width data screened in step 3.2, remove the outliers where the water level deviates from the sample median water level by more than twice the sample water level standard deviation, and the river width is less than the low water level river width Or greater than the river width at high water level The purpose of this step is to remove outliers in height and width. Through steps 3.1, 3.2 and 3.3, the screened water level-river width data set is obtained, which achieves the purpose of removing data with large deviations and low quality, thereby optimizing the water level-river width data and completing quality control.

[0058] Step 4: Calculate the key control points of the water level-river width hydrological curve.

[0059] 4.1 The hydraulic radius can be approximately replaced by the river width, according to the hydraulic Manning formula. , solve the equation to get the median cross-sectional area .

[0060] 4.2 From the water level-river width dataset filtered in step 3.3, select the river width with the median The 5 water levels with the closest width to the river width data are linearly regressed and the median river width is used Water level prediction in linear regression as median water level .

[0061] 4.3 Water level-river width curve below the median water level using power function Approximately, is the lowest water level, is the coefficient, is an exponent. The power function satisfies ,and ,in Set to 1, representing the triangular cross-section assumption. and By solving the above two equations simultaneously, we get:

[0062] 4.4 Calculation of low water level river width The corresponding water level on this power function , as a low-water level control point to constrain underwater section characteristics.

[0063] 4.5 The river bank is used as the high water level control point, representing the outer boundary of the section, and the river width is the high water level width. , water level Set as the highest water level in the filtered water level-river width dataset Add a specific height difference based on . Set to 0.2 times the water level range in the filtered water level-river width dataset.

[0064] Step 5: Calculate the water level-river width hydrological curve.

[0065] 5.1 Based on the filtered water level-river width dataset obtained in Step 3 and the high and low water level control points obtained in Step 4, using weighted non-linear regression, a power function is used to fit the water level-river width hydrological curve. Among them, all water level-river width data are equally divided into regression weights, and the weights of the high and low water level control points are both . The weight is set to 0.5.

[0066] Taking the Dala Hydrological Station on the Nenjiang River as an example, all the water level-river width measurement data in the measured flow results table of this hydrological station recorded in the hydrological yearbook in 2020 are used as the measured results. Through the method of the present invention, the river width of the high water level control point is calculated to be 424.7 meters, the water level is 126.7 meters, the river width of the low water level control point is 197.6 meters, and the water level is 123.1 meters. The expression of the calculated water level-river width hydrological curve is . Since the elevation benchmarks of the measurement and calculation are different and cannot be directly compared, first, the water level-river width data point with the median water level in the measurement results is determined, and the elevation benchmark of the calculated water level-river width hydrological curve is adjusted to coincide with this point. The expression of the water level-river width hydrological curve after the elevation benchmark adjustment is , and the comparison between the calculation results and the measurement results is as Figure 3 shown. Through Figure 3 it can be seen that for a river about 300 meters wide, the error between the calculated curve and 87.5% of the measured water level-river width data is within ±0.6 meters, which proves the effectiveness of the present method.

[0067] The embodiment of the present invention also provides a river channel cross-section system, the structure of which is as Figure 4 shown, and is used to implement the above method.

[0068] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Those of ordinary skill in the art can modify or equivalently replace the technical solutions of the present invention without departing from the spirit and scope of the present invention. The protection scope of the present invention shall be subject to what is described in the claims.

Claims

1. A method for extracting a river section, comprising: Obtain satellite data to obtain the water level-river width original data set and historical river width data set of the river section; the historical river width data set includes the low water level river width ; Screening the water level-river width original data set to obtain a water level-river width data set; The median river width is obtained using the historical river width dataset , combined with the existing median runoff and gradient , and the median cross-sectional area is obtained ; Using the median river width, obtaining a median water level from the water level-river width dataset; Obtaining a low water level control point according to the median river width, the median cross-sectional area and the median water level; Obtaining a high water level control point according to the water level-river width data set; Generate a water level-river width hydrological curve of the river section according to the water level-river width data set, the low water level control point and the high water level control point, and complete the river section extraction; The low water level control point is obtained by the following method: According to the hydraulic Manning formula , Manning roughness coefficient According to experience, it is set to 0.035, and the median cross-sectional area is obtained by solving the equation ; From the water level-river width dataset, select the river width with the median The 5 water levels with the closest width to the river width data are linearly regressed and the median river width is used Water level prediction in linear regression as median water level ; Median water level The following water level-river width curves are expressed using a power function Approximately, is the lowest water level, is the coefficient, is an exponent; the power function satisfies ,and ,in Set to 1, representing the triangular cross-section assumption; unknown and By solving the above two equations simultaneously, we can get: Calculate low water river width The corresponding water level on this power function , as a low-water level control point to constrain underwater section characteristics.

2. The method according to claim 1, characterized in that The screening includes screening according to uncertainty, screening according to inverse relationship and screening according to outliers.

3. The method according to claim 1, characterized in that: The historical river width dataset is used to determine the river section extraction range, and further obtain the median river width.

4. The method according to claim 3, characterized in that The historical river width data set is used to calculate the river width corresponding to the high water level control point and the river width corresponding to the low water level control point, and the river section extraction range is determined based on the two.

5. The method according to claim 1, characterized in that The median runoff is obtained through the existing global runoff simulation data set, the gradient is obtained through the existing global river attribute data set, and the median cross-sectional area is obtained through the Manning formula.

6. The method according to claim 1, characterized in that The water level of the high water level control point is the highest water level in the water level-river width data set plus a set height difference.

7. The method according to claim 1, characterized in that The water level-river width hydrological curve is fitted by a power function.

8. A river section extraction system, comprising: The data acquisition module is used to acquire satellite data to obtain the water level-river width original data set and the historical river width data set of the river section; the historical river width data set includes the low water level river width ; A data screening module, used for screening the water level-river width original data set to obtain a water level-river width data set; A parameter calculation module is used to obtain the median river width using the historical river width data set , combined with the existing median runoff and gradient , and the median cross-sectional area is obtained ; Using the median river width to obtain the median water level from the water level-river width dataset; A control point calculation module, used to obtain a low water level control point according to the median river width, the median cross-sectional area and the median water level; and obtain a high water level control point according to the water level-river width data set; A river section extraction module, used for generating a water level-river width hydrological curve of the river section according to the water level-river width data set, the low water level control point and the high water level control point, to complete the river section extraction; The low water level control point is obtained by the following method: According to the hydraulic Manning formula , Manning roughness coefficient According to experience, it is set to 0.035, and the median cross-sectional area is obtained by solving the equation ; From the water level-river width dataset, select the river width with the median The 5 water levels with the closest width to the river width data are linearly regressed and the median river width is used Water level prediction in linear regression as median water level ; Median water level The following water level-river width curves are expressed using a power function Approximately, is the lowest water level, is the coefficient, is an exponent; the power function satisfies ,and ,in Set to 1, representing the triangular cross-section assumption; unknown and By solving the above two equations simultaneously, we can get: Calculate low water river width The corresponding water level on this power function , as a low-water level control point to constrain underwater section characteristics.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, the computer program is configured to be executed by the processor, and the computer program includes instructions for executing the method according to any one of claims 1 to 7.

10. A storage medium storing a computer program, wherein when the computer program is executed by a computer, the method according to any one of claims 1 to 7 is implemented.

Images