Automatic correction method for river channel width based on BG segmentation and MK trend test
By combining BG segmentation and MK trend verification with a digital elevation model, the width of the river channel cross-section is automatically corrected, solving the problem of remote sensing image interpretation error and achieving accuracy and efficiency in river engineering design and flood simulation.
Patent Information
- Application Number
- CN202411607512.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-12
- Publication Date
- 2025-10-28
- Estimated Expiration
- 2044-11-12
AI Technical Summary
Current technologies for correcting river channel width rely on measured flood data and remote sensing image interpretation, which are prone to errors and make it difficult to accurately describe the width of narrow, meandering, and continuous river channels. Furthermore, they consume a significant amount of manpower.
A method based on Bernaola-Galvan (BG) segmentation and Maan-Kendall (MK) trend test was adopted, combined with digital elevation model data, to automatically correct the width of the river channel cross-section through slope and aspect analysis, including slope MK trend test and aspect BG abrupt change test, and adjust the cross-section width.
It enables automatic correction of river channel cross-sectional width, improves the accuracy of river engineering design and flood simulation, reduces the consumption of human resources, and is applicable to various geographical environments.
Smart Images

Figure CN119475787B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of river channel cross-section width technology, and in particular to an automatic correction method for river channel cross-section width based on BG segmentation and MK trend test. Background Technology
[0002] River topographic data is an important foundation for constructing hydrological and hydrodynamic numerical models, and it is also an important reference for studying river evolution and river engineering design. The width of a river is closely related to its flow, and accurate river cross-section data can improve the accuracy of engineering design and flood simulation.
[0003] Traditional ground-based mapping methods can accurately reflect the actual situation of rivers. However, due to the difficulty in obtaining river cross-section data or the high cost of measurement, the current extraction of river topographic data mainly relies on digital elevation models combined with visual interpretation of remote sensing images. With the development of computer and remote sensing technologies, the temporal and spatial resolution of remote sensing images has been significantly improved. Deep learning-based remote sensing image interpretation provides a new method for river cross-section extraction, making it possible to quickly identify complex water boundaries. However, the morphology of rivers in images is related to changes in water resources, such as upstream reservoir construction and periods of floods and droughts, which can affect the results of water body identification in remote sensing images, and thus affect the width of the cross-section. Furthermore, water body identification based on remote sensing interpretation is more suitable for rivers with regular or wide boundaries flowing into the sea and mountainous rivers. For narrow, meandering, and continuous rivers, the identification error is relatively large. Therefore, the width of river cross-sections extracted from remote sensing image interpretation results has a large error and cannot completely and accurately describe the river width. The correction of the cross-section width still requires further visual interpretation or hydrological and topographic data query methods, which requires a lot of manpower. Summary of the Invention
[0004] The purpose of this invention is to overcome the shortcomings and defects of existing river cross-section width correction techniques, such as over-reliance on measured flood data and errors in remote sensing image interpretation. Instead, it provides an automatic river cross-section width correction method based on BG (Bernaola-Galvan) segmentation and MK (Maan-Kendall) trend verification. This method achieves automatic river cross-section width correction entirely based on digital elevation model data, and can provide technical support for river cross-section processing in hydrological and hydrodynamic numerical model construction and river engineering design.
[0005] An automatic correction method for river channel cross-section width based on BG segmentation and MK trend test includes the following steps:
[0006] S1. Calculation of slope and aspect of the study area;
[0007] S2. Generate the initial river channel cross-section;
[0008] S3. Cross-sectional information extraction;
[0009] S4. Cross-sectional slope MK trend test;
[0010] S5. Cross-sectional aspect BG abrupt change test;
[0011] S6. Adjust the cross-sectional width based on the slope MK trend test and the aspect BG abrupt change test results.
[0012] In step S1, the calculation of the slope and aspect of the study area includes calculating the slope and aspect of the study area based on DEM data, using the fitted surface method, traversing the entire DEM data using a 3x3 window, and for each central grid cell, calculating the horizontal distance d between the current grid cell and its eight adjacent grid cells. x and elevation difference d y Then based on the horizontal distance d x and elevation difference d y The slope and aspect are calculated, and the slope is divided into eight aspects.
[0013] Wherein, the horizontal distance d of the grid e x and elevation difference d y The calculation is as follows:
[0014]
[0015] In step S1, the slope is obtained by calculating the ratio of the elevation difference to the horizontal distance between each grid cell and its adjacent grid cells.
[0016]
[0017] In the formula, S P Slope percentage;
[0018] The slope aspect is obtained by calculating the direction of the maximum rate of change of elevation for each grid cell:
[0019]
[0020] In the formula, A s Slope direction;
[0021] The eight slope directions include due north (0-22.5)(337.5-360), due south (157-202.5), due west (247.5-292.5), due east (67.5-112.5), northeast (22.5-67.5), southeast (112.5-157.5), southwest (202.5-247.5), and northwest (292.5-337.5).
[0022] In step S2, the river centerline is drawn based on the water body identification results interpreted by remote sensing. The river centerline is segmented according to the cross-sectional interval requirements. At the breakpoint, a river cross-section perpendicular to the river centerline is generated. The cross-section width is consistent with the water body boundary, and cross-section shapefile data is obtained.
[0023] In step S3, the section information extraction includes: reading section data with the help of the open-source Geospatial Data Abstraction Library (GDAL) and overlaying each section line with Dem, aspect and slope raster data to extract the elevation, aspect and slope profile points of each raster cell on the section.
[0024] In step S4, the cross-section is divided into left slope and right slope according to the slope aspect / lowest point of the river channel. The slope of the left slope and right slope of each cross-section is subjected to MK trend analysis to determine whether the cross-section passes the MK trend test. If yes, the shapefile file and elevation point matrix results of the cross-section are output. If no, proceed to step S5.
[0025] In step S4, the cross-section is divided into left and right slopes based on the slope aspect / lowest point of the river channel. MK trend analysis is then performed on the slopes of the left and right slopes of each cross-section, including:
[0026] For the distance sequence x of the left or right slope i Construct the ordered sequence S of slope. k :
[0027]
[0028] r i x represents i >x j Given the cumulative number of samples (1≤j≤i), assuming the distance sequences are random and independent, calculate the standard normal distribution statistical variable UF. k :
[0029]
[0030]
[0031] In the formula, E(S) k S is the slope order sequence. k The mean, For S k The variance, if |UF k |≥U α This indicates that the slope exhibits a predetermined trend of change, U αThis represents the critical value used to determine whether a change in the trend of a time series is significant at a given significance level α; UB is calculated by reversing the order of the distance series. k sequence:
[0032] UB k =-UF k (k = n, n-1, ..., 1);
[0033] Determine UF k and UB k Do the sequences have intersections?
[0034] If the slope of the left and right slopes is UF k and UB k Each sequence has only one intersection point. The coordinates before the intersection point on the left bank and the coordinates after the intersection point on the right bank are defaulted to the left and right endpoints of the river channel cross-section. The cross-section width is reduced and adjusted, and the shapefile file and elevation point matrix results of the cross-section are output. If there are multiple intersection points or no intersection points, proceed to step 5.
[0035] In step S5, the section aspect BG abrupt change test includes:
[0036] Obtain the slope aspect of the left and right slopes of the river section that failed the MK trend test. If there are no consecutive pixels with opposite directions in the slope aspect of the left and right slopes, move the section within a predetermined length range upstream and downstream along the river flow direction with the resolution of the raster data as the step size, and make the section length consistent with the water body range. Then repeat step S4 to extract the elevation, slope, and slope aspect matrix of the moved section. Re-perform the MK trend test on the left and right slopes until there are consecutive pixels with opposite directions in the left and right slopes. Then perform BG segmentation on the slope aspect matrix of the left and right slopes of each section until it cannot be segmented.
[0037] The step of performing BG segmentation on the slope aspect lattice of the left and right slopes of each cross section includes:
[0038] Calculate the average slope aspect U of the distance before and after each point i on the cross section for both the left and right slopes. L U R and the standard deviation of the slope aspect S of the front and rear distances. L S R , i = 2, 3, ..., n-1; Statistical analysis of the average slope aspect U of the left slope. k and the average slope aspect U of the right slope R The difference between them, T(i);
[0039]
[0040]
[0041] In the formula, S D To combine the deviations, N L and N R Let T(i) be the number of samples representing the distance before and after point i. The larger T(i) is, the greater the difference in slope aspect between the distances before and after point i. Calculate the maximum value T(i) among the samples. m And statistical significance P(T) m ):
[0042]
[0043] In the formula, γ = 4.19lnN - 11.54, δ = 0.4, v = N - 2, and N is the distance between the left and right slopes. In time series analysis, the Beta distribution is calculated by taking the square of the maximum value of the test statistic, where I is the function operator and the parameters are δv and δ. The complementary cumulative distribution function value at point; let a critical value P0 be given, and the confidence level be 0.95. If the significance of a point P(T) is... m )≥P k If the distance sequence is true, then it is divided into two segments at that point; otherwise, it is not divided. The calculation is repeated for the two resulting slope subsequences, and the statistical significance P(T) is evaluated. m And determine until all subsequences can no longer be divided.
[0044] In step S6, adjusting the cross-sectional width based on the slope MK trend test and aspect BG abrupt change test results includes:
[0045] If the slope of the left or right slope is UF k and UB k If the sequence has no intersection and the slope aspect BG segmentation result has one or more abrupt change points, then shorten the length of the left or right slope; where, shorten the left or right slope to the position of the next river channel cross-section mileage point of the farthest slope aspect BG segmentation abrupt change point. The cross-section width is adjusted by default, and the adjusted cross-section shapefile and elevation point matrix are output.
[0046] When shortening the left or right slope to the next channel cross-section mileage point after the furthest slope aspect BG abrupt change point, for example, if the mileage of the left slope is [0, 11, 24, 32, 43, 55], where 11, 32, and 43 are slope aspect abrupt change points, shortening the left slope to the next channel cross-section mileage point after the furthest slope aspect BG abrupt change point means shortening the left slope to position 55. Similarly, if the mileage of the right slope is [156, 172, 188, 197, 202, 223], where 188, 197, and 202 are slope aspect abrupt change points, shortening the right slope to the next channel cross-section mileage point after the furthest slope aspect BG abrupt change point means shortening the left slope to position 172.
[0047] If the slope of the left or right slope is UF k and UB k If the sequences have no intersections and the slope aspect BG segmentation results have no abrupt changes, then the elevation, slope aspect, and slope of the cross-section are re-extracted after extending the left or right slope by one pixel. MK trend analysis is then performed on the slope of the adjusted left or right slope. When the extended left or right slope is greater than the width of the cross-section itself, the cross-section is moved again within a predetermined length range upstream and downstream along the river flow direction, with the resolution of the raster data as the step size, and the cross-section length is consistent with the water body range. Step S4 is then repeated to extract the elevation, slope, and slope aspect lattice of the moved cross-section. MK trend testing is then performed on the left and right slopes again until there are corresponding and continuous pixels in the left and right slope directions. Then, BG segmentation is performed on the slope aspect lattice of the left and right slopes of each cross-section until segmentation is no longer possible.
[0048] If the UF of the left or right slope k and UB k If the sequence has multiple intersections and the slope aspect BG segmentation result of the left or right slope has one or more abrupt change points, then the length of the left or right slope is shortened; in particular, the left or right slope is shortened to the position of the nearest slope aspect BG segmentation abrupt change point, the elevation, slope aspect, and slope of the cross section are re-extracted, and MK trend analysis is performed on the slope of the left or right slope of the adjusted cross section.
[0049] When shortening the left or right slope to the nearest aspect BG abrupt change point, for example, if the mileage of the left slope is [0, 11, 24, 32, 43, 55], where 11, 32, and 43 are aspect abrupt change points, shortening the left slope to the nearest aspect BG abrupt change point means shortening the left slope to position 11. Similarly, if the mileage of the right slope is [156, 172, 188, 197, 202, 223], where 188, 197, and 202 are aspect abrupt change points, shortening the right slope to the nearest aspect BG abrupt change point means shortening the left slope to position 202; when UF k and UB k When the sequence has only one intersection point, the cross-section width is considered to be adjusted, and the adjusted cross-section shapefile and elevation point matrix are output.
[0050] In a large number of cross-sectional tests, no UF was found on the left or right slope. k and UB k The sequence has multiple intersections, and the BG segmentation results of the left or right slope have no abrupt change points.
[0051] Because the changes in river slope aspect are periodic and nonlinear, the Bernaola-Galvan mutation detection method used in this invention can more accurately capture the existence of mutation points. Since slope has a trend, the Maan-Kendall test used in this invention can effectively detect the trend characteristics of slope.
[0052] This invention adaptively generates cross-sections of different widths based on water body identification results from remote sensing interpretation. By comprehensively analyzing the non-stationarity of slope aspect and the monotonic trend of slope, it obtains the abrupt change information of the cross-section and then corrects the width of the river channel cross-section.
[0053] This invention comprehensively considers the periodic changes in river slope aspect and the trend changes in slope gradient, enabling more accurate analysis of the morphological characteristics of river cross-sections and achieving automatic correction of river cross-section width. This method has the advantages of being objective, easy to implement, and widely applicable. Attached Figure Description
[0054] Figure 1 This is a flowchart of an automatic river channel cross-section width correction method based on BG segmentation and MK trend test according to an embodiment of the present invention.
[0055] Figure 2 This is a 3x3 window diagram used in this embodiment of the invention to calculate the horizontal distance and elevation difference of any grid.
[0056] Figure 3 These are the river centerline and initial river cross-section generated based on the water body range in this embodiment of the invention;
[0057] Figure 4-1 This is an initial cross-sectional view of the initial river channel extracted from DEM data according to an embodiment of the present invention (the horizontal axis represents the mileage of the river channel section, and the vertical axis represents the elevation of the river channel section).
[0058] Figure 4-2 This is a graph showing the MK trend test results of the left slope of the initial river section in an embodiment of the present invention (the horizontal axis represents the river section mileage).
[0059] Figure 4-3 This is a graph showing the MK trend test results of the right slope of the initial river section in an embodiment of the present invention (the horizontal axis represents the river section mileage); wherein the right slope passes the MK trend test.
[0060] Figure 5 This is a diagram showing the adjustment results of the right slope of the cross section according to the slope MK trend test in an embodiment of the present invention.
[0061] Figure 6-1 This is a graph showing the trend test results of the right slope MK of the cross section in an embodiment of the present invention (the horizontal axis represents the mileage of the river section).
[0062] Figure 6-2This is a diagram showing the slope aspect BG segmentation result of the cross section in an embodiment of the present invention (the horizontal axis represents the river section mileage, the vertical axis represents the test statistical value T, and the vertical height represents the magnitude of the statistical value T).
[0063] Figure 7-1 This is a river cross-section diagram after the slope MK trend test and slope aspect BG segmentation results have been adjusted according to the embodiments of the present invention (the horizontal axis represents the river cross-section mileage, and the vertical axis represents the river cross-section elevation).
[0064] Figure 7-2 , Figure 7-3 These are the MK trend test results of the left and right slopes of the river channel after the adjustment according to the embodiments of the present invention. Detailed Implementation
[0065] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
[0066] The automatic correction method for river channel cross-section width based on BG segmentation and MK trend test according to embodiments of the present invention includes the following implementation steps:
[0067] S1. Calculation of slope and aspect of the study area;
[0068] Based on DEM (Digital Elevation Model) data, the slope and aspect of the study area were calculated, and the aspects were classified to obtain slope and aspect raster data respectively; the slope was divided into eight aspects.
[0069] The eight slope directions include due north (0-22.5)(337.5-360), due south (157-202.5), due west (247.5-292.5), due east (67.5-112.5), northeast (22.5-67.5), southeast (112.5-157.5), southwest (202.5-247.5), and northwest (292.5-337.5).
[0070] Specifically, slope and aspect raster data are calculated based on Digital Elevation Model (DEM) data: the slope and aspect of the study area based on the DEM data are calculated using the fitted surface method, and a 3x3 window is used to traverse the DEM data of the study area; for example... Figure 2As shown, each 3x3 grid cell in the 3x3 window consists of a central grid cell (with pixel value e) located in the center of a symmetrically arranged square with equal side lengths. The surrounding grid cells are: the left grid cell (with pixel value e1), the top grid cell (with pixel value e2), the right grid cell (with pixel value e3), the bottom grid cell (with pixel value e4), the top-left grid cell (with pixel value e5), the top-right grid cell (with pixel value e6), the bottom-right grid cell (with pixel value e7), and the bottom-left grid cell (with pixel value e8). Adjacent grid cells with overlapping sides form a grid. The horizontal distance d between the central grid cell and each of the following grid cells in the DEM data is calculated by sliding the 3x3 window: [Diagram showing the horizontal distance d between the central grid cell and each of the following grid cells: left, top, right, bottom, top-left, top-right, bottom-right, and bottom-left grid cells]. x and elevation difference d y The slope and aspect raster data are generated for subsequent extraction of river cross-section information.
[0071] S2. Generate the initial river channel cross-section;
[0072] Based on the water body identification results from remote sensing interpretation, the river centerline is drawn. The river centerline is then segmented according to the required cross-sectional intervals. At the breakpoints, river cross-sections perpendicular to the centerline are generated. The left and right bank positions of the cross-sections coincide with the water body boundaries, resulting in cross-sectional shapefile data, such as... Figure 3 As shown; drawn according to the water body area as follows. Figure 3 The length shown is the centerline of the river channel in the study area. According to the engineering requirement of 1000-meter intervals for the cross-sections of the river channel in the study area, the centerline of the river channel in the study area is divided into segments to obtain multiple breakpoints of the centerline of the river channel in the study area. At the multiple breakpoints of the centerline of the river channel in the study area, cross-sections of the river channel in the study area are made perpendicular to the centerline of the river segment where the breakpoint is located. The width of each cross-section of the river channel in the study area is aligned with the water body boundary of its river mileage, thus obtaining the shapefile data of each cross-section of the river channel in the study area.
[0073] S3. Cross-sectional information extraction;
[0074] The cross-sectional data is read through the entire process and each cross-sectional line is overlaid with Dem, aspect and slope raster data to extract the elevation, aspect and slope profile points of each raster cell on the cross-section.
[0075] Specifically, the shapefile data of each river cross section in multiple study areas were read using the open-source raster spatial data conversion library GDAL. The cross section line of each river in multiple study areas was overlaid with the DEM data, slope aspect and slope raster data of the study areas, and the elevation, slope aspect and slope profile points of each raster cell on each river cross section in multiple study areas were extracted respectively.
[0076] S4. Cross-sectional slope MK trend test;
[0077] Each cross-section is divided into left and right slopes, with the pixel representing the slope direction (slope direction value of -1) as the boundary. If there is no slope direction value of -1, the pixel with the lowest elevation of the cross-section is used as the boundary. MK trend analysis is performed on the slope of the left and right slopes of each cross-section.
[0078] Calculate the standard normal distribution statistic UF of the cross-sectional slope profile lattice of each river channel in multiple study areas. k With the inversion statistic UB k Sequence; based on slope aspect, each river channel cross section in multiple study areas is divided into left and right slopes; the standard normal distribution statistic UF of the slope profile lattice of the left and right slopes of each river channel cross section in multiple study areas is calculated respectively. k With the inversion statistic UB k sequence;
[0079] Determine UF k and UB k Does the sequence have an intersection? If the slopes of the left and right slopes are UF k and UB k Each sequence has only one intersection point. The coordinates before the intersection point on the left bank and after the intersection point on the right bank are defaulted to the left and right endpoints of the river channel cross-section, and the cross-section width is adjusted by reducing its size. For example... Figure 4-1 , Figure 4-2 as well as Figure 4-3 As shown, the right slope of this section passed the consistency test of MK, and the adjusted right slope is as follows. Figure 5 As shown; the output includes shapefiles and elevation lattice results for each river channel cross-section in the study area that pass the MK trend test.
[0080] S5. Cross-sectional aspect BG abrupt change test;
[0081] For cross-sections that fail the MK trend test, the slope aspect of the left and right slopes is first determined based on the flow direction of the river segment. If there are no consecutive pixels with opposing slope aspects, the cross-section is moved within a 100m range upstream and downstream along the river flow direction, with the resolution of the raster data as the step size, and the cross-section length is consistent with the water body range. Then, the elevation and slope lattice of the moved cross-section is extracted, and the MK trend test is re-performed on the left and right slopes until there are consecutive pixels with opposing slope aspects. If there are consecutive pixels with opposing slope aspects, the slope aspect lattice of the left and right slopes of each cross-section is segmented using BG.
[0082] S6. Adjust the cross-sectional width based on the slope MK trend test and the aspect BG abrupt change test results.
[0083] like Figure 6-1 , Figure 6-2 As shown in this application, based on the MK trend analysis results and BG segmentation results of the left slope of the cross section, the UF of the left slope is... k and UB k If the sequence shows multiple intersection points, the cross-sectional length of the left slope needs to be shortened. The left bank endpoint should be shortened to the location of the nearest aspect BG segmentation abrupt change point. The slope MK trend test should then be performed on the left slope again. Based on the test results, it should be determined whether aspect BG segmentation is necessary. Figure 7-1 , Figure 7-2 as well as Figure 7-3 The image shows the final cross-sectional result after shortening the left slope.
[0084] Depend on Figure 7-1 , Figure 7-2 as well as Figure 7-3 It can be seen that the standard normal distribution statistical variable UF of the slope profile lattice of the left slope of a certain river channel cross section is... k And the inversion statistic UB k The standard normal distribution statistical variable UF of the sequence and right slope profile lattice k And the inversion statistic UB k Each sequence has only one intersection point, assuming that the cross-sectional width of the river channel has been adjusted.
[0085] Comparing the river channel cross-section before and after the adjustment, it can be seen that the width correction effect on the left and right banks of the river channel cross-section is good.
[0086] The embodiments of the present invention illustrate that the automatic correction method for river channel cross-section width based on BG segmentation and MK trend test is entirely based on digital elevation model data, and combines the MK trend test method and the BG segmentation method to correct the river channel cross-section width.
[0087] Compared to other methods for extracting and correcting river width, this method eliminates the need to search for river hydrological data. It identifies the changing trends and anomalies in the river cross-section elevation based entirely on the trend of the slope and the periodicity and nonlinearity of the slope aspect. The method achieves rapid calculation by directly using DEM data.
[0088] Therefore, the automatic correction method for river channel cross-section width based on BG segmentation and MK trend test in this invention has the characteristics of not relying on historical hydrological data, rich physical meaning, fast calculation, and wide applicability.
[0089] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. It will be apparent to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention.
[0090] Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, it is intended that all variations falling within the meaning and scope of the equivalents of the claims be included within the invention.
[0091] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.
Claims
1. An automatic correction method for river channel cross-section width based on BG segmentation and MK trend test, characterized in that, Including the following steps: S1. Calculation of slope and aspect of the study area; S2. Generate the initial river channel cross-section; S3. Cross-sectional information extraction; S4. Cross-sectional slope MK trend test; S5. Cross-sectional aspect BG abrupt change test; S6. Adjust the cross-sectional width based on the slope MK trend test and the aspect BG abrupt change test results; In step S4, the cross-section is divided into left and right slopes based on the slope aspect / lowest point of the river channel. MK trend analysis is then performed on the slopes of the left and right slopes of each cross-section, including: For distance sequences on the left or right slope Constructing an ordered sequence of slopes : ; express The cumulative number of samples, Assuming the distance sequences are random and independent, calculate the standard normal distribution statistical variable. : ; ; ; In the formula, For slope order sequence The mean, for The variance, if This indicates that the slope exhibits a predetermined trend of change. Indicates at a given significance level Below is a critical value used to determine whether a change in the trend of a time series is significant; the distance series is then arranged in reverse order, and the following calculations are performed. sequence: ; judge and Do the sequences have intersections? If the slopes of the left and right slopes and Each sequence has only one intersection point. The coordinates before the intersection point on the left bank and the coordinates after the intersection point on the right bank are taken as the left and right endpoints of the river channel cross-section by default. The width of the cross-section is reduced and adjusted. The shapefile file and elevation point matrix results of the cross-section are output. If there are multiple intersection points or no intersection points, proceed to step 5. In step S6, adjusting the cross-sectional width based on the slope MK trend test and aspect BG abrupt change test results includes: If the slope of the left or right slope and If the sequence has no intersection and the slope aspect BG segmentation result has one or more abrupt change points, then shorten the length of the left or right slope; where, shorten the left or right slope to the position of the next river channel cross-section mileage point of the farthest slope aspect BG segmentation abrupt change point. The cross-section width is adjusted by default, and the adjusted cross-section shapefile and elevation point matrix are output. If the slope of the left or right slope and If the sequences have no intersections and the slope aspect BG segmentation results have no abrupt changes, then extend the left or right slope by one pixel and re-extract the elevation, slope aspect, and slope of the cross-section. Perform MK trend analysis on the slope of the adjusted left or right slope. When the extended left or right slope is greater than the width of the cross-section itself, move the cross-section upstream and downstream within a predetermined length range along the river flow direction with the resolution of the raster data as the step size, and make the cross-section length consistent with the water body range. Then repeat step S4 to extract the elevation, slope, and slope aspect matrix of the moved cross-section. Perform MK trend test on the left and right slopes again until there are corresponding and continuous pixels in the left and right slope directions. Then perform BG segmentation on the slope aspect matrix of the left and right slopes of each cross-section until it cannot be segmented. If it's a left slope or a right slope and If the sequence has multiple intersections and the slope aspect BG segmentation result of the left or right slope has one or more abrupt change points, then the length of the left or right slope is shortened; in particular, the left or right slope is shortened to the position of the nearest slope aspect BG segmentation abrupt change point, the elevation, slope aspect, and slope of the cross section are re-extracted, and MK trend analysis is performed on the slope of the left or right slope of the adjusted cross section. when and When the sequence has only one intersection point, the cross-section width is considered to be adjusted, and the adjusted cross-section shapefile and elevation point matrix are output.
2. The automatic correction method for river channel cross-section width based on BG segmentation and MK trend test according to claim 1, characterized in that, In step S1, the calculation of the slope and aspect of the study area includes: calculating the slope and aspect of the study area based on DEM data, using the fitted surface method, traversing the entire DEM data using a 3x3 window, and for each central grid cell, calculating the horizontal distance between the current grid cell and its eight adjacent grid cells. and elevation difference Then based on horizontal distance and elevation difference The slope and aspect are calculated, and the slope is divided into eight aspects.
3. The automatic correction method for river channel cross-section width based on BG segmentation and MK trend test according to claim 2, characterized in that, In step S1, the grid horizontal distance and elevation difference The calculation is as follows: ; ; The slope is obtained by calculating the ratio of the elevation difference to the horizontal distance between each grid cell and its adjacent grid cells; ; In the formula, Slope percentage; The slope aspect is obtained by calculating the direction of the maximum rate of change of elevation for each grid cell; ; In the formula, Slope direction; The eight slope directions include due north (0-22.5)(337.5-360), due south (157-202.5), due west (247.5-292.5), due east (67.5-112.5), northeast (22.5-67.5), southeast (112.5-157.5), southwest (202.5-247.5), and northwest (292.5-337.5).
4. The automatic correction method for river channel cross-section width based on BG segmentation and MK trend test according to claim 3, characterized in that, In step S2, the river centerline is drawn based on the water body identification results based on remote sensing interpretation. The river centerline is segmented according to the cross-sectional interval requirements. At the breakpoint, a river cross-section perpendicular to the river centerline is generated. The cross-section width is consistent with the water body boundary, and cross-section shapefile data is obtained.
5. The automatic correction method for river channel cross-section width based on BG segmentation and MK trend test according to claim 4, characterized in that, In step S3, the section information extraction includes: reading section data with the help of an open-source raster spatial data conversion library and overlaying each section line with Dem, aspect and slope raster data to extract the elevation, aspect and slope profile points of each raster pixel on the section.
6. The automatic correction method for river channel cross-section width based on BG segmentation and MK trend test according to claim 5, characterized in that, In step S4, the cross-sectional slope MK trend test includes: dividing the cross-section into left and right slopes according to the slope aspect / lowest point of the river channel cross-section, performing MK trend analysis on the slope of the left and right slopes of each cross-section, and determining whether the cross-section passes the MK trend test. If yes, the shapefile file and elevation point matrix results of the cross-section are output; otherwise, proceed to step S5.
7. The automatic correction method for river channel cross-section width based on BG segmentation and MK trend test according to claim 1, characterized in that, In step S5, the section aspect BG abrupt change test includes: The left and right slope aspects of the river cross-section that failed the MK test are obtained. If there are no consecutive pixels with opposite directions in the slope aspects of the left and right slopes, the cross-section is moved within a predetermined length range upstream and downstream along the river flow direction with the resolution of the raster data as the step size, and the length of the cross-section is consistent with the water body range. Then, step S4 is repeated to extract the elevation, slope, and slope aspect matrix of the moved cross-section. The MK trend test is re-performed on the left and right slopes until there are consecutive pixels with opposite directions in the left and right slopes. Then, the slope aspect matrix of the left and right slopes of each cross-section is segmented by BG until it cannot be segmented.
8. The automatic correction method for river channel cross-section width based on BG segmentation and MK trend test according to claim 7, characterized in that, The BG segmentation of the slope aspect lattice for the left and right slopes of each cross section includes: Calculate each point on the cross-section Average slope aspect of the front and rear distances on the left and right slopes , and the standard deviation of the slope aspect of the front and rear distances. , , ; Statistical analysis of the average slope aspect of the left slope and the average slope aspect of the right slope Differences between ; ; ; In the formula, To merge the deviations, and For point The number of samples for the front and back distances. The larger the value, the more likely it is to indicate a point. The greater the difference in slope aspect between the front and rear distances, the more accurate the calculation... The largest maximum value in and statistical significance : ; In the formula, N represents the distance between the left and right slopes; This is the square of the maximum value of the test statistic in time series analysis, where I is the function operator, and it calculates the value with parameters... and Beta distribution in The complementary cumulative distribution function value at the location; let a critical value be defined. With a confidence level of 0.95, if the significance of a point is... If the distance sequence is true at that point, it is divided into two segments; otherwise, it is not divided. The calculation is repeated for the two resulting slope subsequences, and the statistical significance is evaluated. And determine until all subsequences can no longer be divided.
Citation Information
Patent Citations
Long runoff sequence internal trend analysis method based on M-K trend test
CN112115179A
End-to-end scene classification method based on circular memory attention network
CN112686137A