A watershed extraction method based on outlet relocation

Abstract

The present application provides a watershed extraction method based on outlet relocation, including: determining a buffer zone based on a given outlet, iteratively calculating the local cumulative flow in the buffer zone according to the flow direction data and outputting the global cumulative flow; determining the main stream network based on the global cumulative flow and flow direction data, relocating the given outlet coordinates to the main stream network; relocating the outlet coordinates to the tributaries of the main stream; determining the reference outlet according to the main stream relocation results and the tributary relocation results, and extracting the watershed according to the reference outlet. The outlet relocation algorithm in the present application is insensitive to preset parameters, suitable for mass extraction of watersheds of different areas, and only uses the flow direction and given outlet coordinates as input. The algorithm provides a tool for automatically extracting a large number of watersheds for hydrological simulation in the big data era.

Description

Technical Field

[0001] The invention relates to the technical field of watershed extraction, and in particular to a watershed extraction method based on water outlet relocation. Background Art

[0002] A watershed is an area where precipitation is concentrated and drains into a common outlet. In the era of big data, where a large amount of Earth system data is available, the rapid extraction of numerous watershed boundaries has become a basic requirement for hydrological analysis. In the past, watershed boundaries were generally extracted based on topographic contours. With the development of remote sensing and geographic information systems, many techniques for extracting watersheds based on digital elevation models have emerged.

[0003] However, existing watershed extraction technologies (such as ArcGIS) involve two time-consuming manual processes: (i) determining the cumulative flow threshold to obtain the digital river network by trial and error; (ii) manually relocating the given outlet coordinates to the correct digital river network. Based on current watershed extraction technologies, it is necessary to relocate the outlet to the river network based on the user's knowledge of the outlet, which may be impractical for extracting a large number of watersheds.

[0004] Therefore, how to solve the above problems has become a technical problem that needs to be solved urgently by those skilled in the art. Summary of the invention

[0005] The purpose of the embodiments of the present application is to provide a watershed extraction method based on outlet relocation, which can at least solve the above-mentioned problem.

[0006] A first aspect of an embodiment of the present application provides a watershed extraction method based on water outlet relocation, comprising: determining a buffer zone based on a given water outlet, iteratively calculating a local cumulative flow in the buffer zone according to flow direction data and outputting a global cumulative flow;

[0007] Determine the mainstream network based on global cumulative flow and flow direction data, and relocate the given outlet to the mainstream network;

[0008] Relocate the outlet coordinates to the tributary of the main stream;

[0009] A reference water outlet is determined at least according to the main stream relocation result and the tributary relocation result, and a watershed is extracted according to the reference water outlet.

[0010] In some embodiments, the buffer zone includes a plurality of cumulative flow regions; wherein the local cumulative flows in the plurality of cumulative flow regions are calculated according to the flow direction data to output the global cumulative flows respectively.

[0011] In some embodiments, the buffer includes a plurality of cumulative flow regions; wherein the local cumulative flows in the plurality of cumulative flow regions are calculated according to the flow direction data to output the global cumulative flows respectively, including:

[0012] The multiple cumulative flow areas include at least a first cumulative flow area and a second cumulative flow area; wherein the highest local cumulative flow value of the buffer area is used as the first outlet of the first cumulative flow area, and the global cumulative flow corresponding to all local cumulative flows in the first cumulative flow area that flow into the first outlet is calculated;

[0013] The maximum local cumulative flow value outside the first area of ​​the buffer zone is taken as the second water outlet, and the global cumulative flow corresponding to all local cumulative flows in the second cumulative flow area that flow into the second water outlet is calculated.

[0014] In some embodiments, the largest first cumulative flow among the global cumulative flows on the main flow is used as the relocation result of the main flow; wherein the main flow relocation result is used to extract the main flow, and the first cumulative flow has the characteristics of being stable and error-free.

[0015] In some embodiments, the first cumulative flow has stable characteristics, including:

[0016] The first cumulative flow is closer to the given water outlet than the global cumulative flow on the main stream; and / or

[0017] The first cumulative flow has an adjacent upstream river section that is farther from the given water outlet; and / or

[0018] The length of the adjacent upstream river section of the first cumulative flow is not less than the side length of the buffer zone.

[0019] In some embodiments, the buffer zone is rectangular;

[0020] The first cumulative flow with stable characteristics has a length of an adjacent upstream river section that is not less than 1 / 10-1 / 2 of the side length of the buffer zone.

[0021] In some embodiments, the first accumulated stream has an error-free feature, including:

[0022] The first cumulative flow is closer to the given water outlet; and / or

[0023] There is no other global cumulative flow with stable characteristics upstream of the first cumulative flow.

[0024] In some embodiments, the 80-90 quantile of the first cumulative flow is used as a threshold to output the main flow relocation result.

[0025] In some embodiments, each global cumulative flow value on the main stream is subtracted from its adjacent global cumulative flow value, and the confluent cumulative flow is determined according to the size of the difference; wherein the confluent cumulative flow is used to extract tributaries of the main stream.

[0026] In some embodiments, the accumulated flow closest to the given outlet in the global accumulated flow on the tributary is used as the relocation result of the tributary.

[0027] In some embodiments, the step of determining the reference water outlet at least according to the main stream relocation result and the tributary relocation result includes:

[0028] Based on at least the relocation results of the tributaries and the relocation results of the main stream, the cumulative flow closest to the given outlet coordinates is selected as the reference outlet.

[0029] The above technical solution of the present application has the following beneficial technical effects:

[0030] (1) The iterative calculation in this application can greatly reduce the calculation cost. Compared with the traditional method of calculating the global cumulative flow based on the flow direction in a large range at one time, the iterative calculation method of this application greatly reduces the repeated calculation amount in the traditional method. In addition, the traditional method calculates the global cumulative flow of all grids in a large range, while this application uses an iterative calculation method to only calculate the global cumulative flow in the buffer zone at a given outlet, which greatly reduces the amount of calculation and fully meets the needs of relocating the outlet and extracting the basin boundary;

[0031] (2) In this application, the correct relocation of the water outlet of the basin is determined based on the relocation results of the main stream and tributaries, which can greatly improve the positioning accuracy of the water outlet of the basin. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 is a flow chart of a watershed extraction method based on water outlet relocation provided by an embodiment of the present application;

[0033] Figure 2 It is a flowchart of a global cumulative flow iterative calculation provided by an embodiment of the present application;

[0034] Figure 3 is a schematic diagram of a mainstream relocation algorithm provided in an embodiment of the present application;

[0035] Figure 4 is a schematic diagram of a tributary relocation algorithm provided in an embodiment of the present application;

[0036] Figure 5 It is an optimized cumulative flow structure diagram provided by an embodiment of the present application;

[0037] Figure 6 This is a comparison chart of algorithm accuracy provided by an embodiment of the present application. DETAILED DESCRIPTION

[0038] In order to make the purpose, technical scheme and advantages of the present application clearer, the present invention is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present invention. In addition, in the following description, the description of known structures and technologies is omitted to avoid unnecessary confusion of the concept of the present invention.

[0039] The accompanying drawings show schematic diagrams of the structures of the embodiments of the present application. These figures are not drawn to scale, and some details may be omitted for the purpose of clarity. The various regions, shapes, and relative sizes and positional relationships shown in the figures are only exemplary, and may deviate in practice due to manufacturing tolerances or technical limitations, and those skilled in the art may additionally design regions with different shapes, sizes, and relative positions according to actual needs.

[0040] Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0041] In addition, the technical features involved in the different embodiments of the present application described below can be combined with each other as long as they do not conflict with each other.

[0042] The present application will be described in more detail below with reference to the accompanying drawings. In each of the accompanying drawings, the same elements are represented by similar reference numerals. For the sake of clarity, the various parts in the accompanying drawings are not drawn to scale.

[0043] In one embodiment of the present application, reference Figure 1 , a watershed extraction method based on outlet relocation is provided, comprising: S101, determining a buffer zone based on a given outlet, iteratively calculating a local cumulative flow in the buffer zone according to flow direction data and outputting a global cumulative flow;

[0044] S102, determining a mainstream network based on global cumulative flow and flow direction data, and relocating a given outlet to the mainstream network;

[0045] S103, relocating the coordinates of the water outlet to a tributary of the main stream;

[0046] S104. Determine a reference outlet at least based on the main stream relocation results and the tributary relocation results, and extract the watershed based on the reference outlet.

[0047] It should be understood that the present application is based on the processing of the digital elevation model; wherein the cumulative flow is a grid of the number of pixels flowing to each grid of the digital elevation model, obtained by the following method: there is a unit of water at each point in the digital elevation model, and according to the natural law that water flows from high places to low places, the water volume flowing at each point is calculated based on the input flow direction data.

[0048] In the embodiments of the present application, on the one hand, the iterative calculation in the present application can greatly reduce the calculation cost. Compared with the traditional method of calculating the global cumulative flow based on the flow direction in a large range at one time, the iterative calculation method of the present application greatly reduces the amount of repeated calculations in the traditional method. In addition, the traditional method calculates the global cumulative flow of all grids in a large range, while the present application uses an iterative calculation method to only calculate the global cumulative flow in the buffer zone at a given outlet, which greatly reduces the amount of calculation and fully meets the needs of relocating the outlet and extracting the basin boundary; on the other hand, the relocation results of the main stream and tributaries in the present application are used to determine the correct relocation of the basin outlet, which can greatly improve the positioning accuracy of the basin outlet.

[0049] In some embodiments, the buffer includes a plurality of cumulative flow areas; wherein the local cumulative flows in the plurality of cumulative flow areas are calculated according to the flow direction data to output the global cumulative flows respectively.

[0050] In some embodiments, the buffer includes a plurality of cumulative flow regions; wherein the local cumulative flows in the plurality of cumulative flow regions are calculated according to the flow direction data to output the global cumulative flows respectively, including:

[0051] The multiple cumulative flow areas include at least a first cumulative flow area and a second cumulative flow area; wherein the highest local cumulative flow value of the buffer area is used as the first outlet of the first cumulative flow area, and the global cumulative flow corresponding to all local cumulative flows in the first cumulative flow area that flow into the first outlet is calculated;

[0052] The maximum local cumulative flow value outside the first area of ​​the buffer zone is taken as the second outlet, and the global cumulative flow corresponding to all local cumulative flows in the second cumulative flow area that flow into the second outlet is calculated.

[0053] In a specific embodiment, it should be noted again that the cumulative flow is a grid of the number of pixels in each grid of the flow direction digital elevation model, and the cumulative flow can be obtained based on the input flow direction data. The calculation of the cumulative flow is one of the main factors limiting the speed of watershed extraction. Based on the above problems, a specific embodiment of the present application discloses an automatic outlet relocation (AOR) algorithm, which uses an iterative algorithm to quickly calculate the cumulative flow of a buffer zone at a given outlet, including:

[0054] First, a square buffer zone is cut out from the given outlet of the basin (the given outlet is determined by the given longitude and latitude of the hydrological station) in the direction of the flow of the given outlet; it should be noted that the shape of the buffer zone includes but is not limited to regular shapes and irregular shapes. Preferably, a regular shape is selected as the buffer zone. In the embodiment of the present application, a square can be selected as the buffer zone.

[0055] Furthermore, the side length of the buffer zone is 20 km, e.g. Figure 2 ( Figure 2 Including five figures a, b, c, d and e). In some embodiments, the specific shape of the buffer zone includes but is not limited to a square, preferably a regular rectangle or a circle, and the side length can be selected as required. Figure 2 Figure a is a schematic diagram of flow direction. Figure a is a 7×7 regular matrix, which includes 49 local cumulative flows. The direction of the arrow corresponding to each local cumulative flow is used for its flow direction.

[0056] Furthermore, the local cumulative flow of the buffer is calculated, where "local" means ignoring the grids outside the buffer and finding the grid with the largest local cumulative flow within the matrix. Figure 2 Attached figure b is the local cumulative flow grid diagram corresponding to Figure a, and the largest local cumulative flow is the grid P in Figure b (the value of P is 22, which indicates that 22 grids converge to grid P); among them, the values ​​of each grid in Figure b correspond to the cumulative flow values ​​of the flow direction arrows in Figure a.

[0057] Further, the buffer zone includes a plurality of cumulative flow areas, and the plurality of cumulative flow areas at least include a first cumulative flow area and a second cumulative flow area. It is understandable that the number of cumulative flow areas in the buffer zone includes but is not limited to two.

[0058] In the present application examples, reference Figure 2Figure ce of Figure ce (in Figure ce, the grids filled with the same section line are the same cumulative flow areas), Figure c is the first global cumulative flow diagram, Figure d is the second global cumulative flow diagram, and Figure c is the final global cumulative flow diagram. Figure e of the buffer zone includes three cumulative flow areas, namely the first cumulative flow area, the second cumulative flow area and the third cumulative flow area. Specifically, the highest cumulative flow value (grid P) in the buffer zone of Figure b is selected as the first outlet of the first cumulative flow area, and all grids corresponding to the first cumulative flow area of ​​the buffer zone are converted to global cumulative flow values. Referring to Figure ce, the first cumulative flow area includes 22 grids including the first outlet P, and the flow direction of all grids in the first cumulative flow area eventually points to the first outlet P, that is, the first cumulative flow area includes 22 first global cumulative flows. It should be understood that "global" here means considering all grids, that is, including grids outside the buffer zone. For example, referring to Figure bc, all grids flowing into P in the first cumulative flow area are converted to global cumulative flow values, specifically, the local cumulative flow value of the first outlet P is converted from 22 to a global cumulative flow value of 579.

[0059] Further, the local cumulative flow maximum value outside the first area of ​​the buffer zone in Figure c is selected as the second water outlet (such as the area outside the first cumulative flow area of ​​the buffer zone in Figure c), and the second water outlet is shown as grid Q in Figure c. The global cumulative flow value is calculated based on the second water outlet to obtain the global cumulative flow value of the second cumulative flow area. With specific reference to Figure de, the second cumulative flow area includes 19 grids, and the flow directions of all grids in the second cumulative flow area ultimately point to the second water outlet Q, that is, the second cumulative flow area includes 19 second global cumulative flows; wherein the local cumulative flow value of the second water outlet Q is converted from 19 to a global cumulative flow value of 436.

[0060] Further, the local cumulative flow maximum value outside the first cumulative flow area and the second cumulative flow area in the buffer zone in Figure d is selected as the third water outlet, and the third water outlet is the grid with a value of 8 in Figure d. The global cumulative flow value is calculated based on the third water outlet to obtain the global cumulative flow value of the third cumulative flow area. With specific reference to Figure e, the third cumulative flow area includes 8 grids, and the flow directions of all grids in the third cumulative flow area ultimately point to the third water outlet, that is, the third cumulative flow area includes 8 third global cumulative flows (it should be noted that the remaining global cumulative flows in Figure e are characterized as the third global cumulative flows of this embodiment); wherein, the local cumulative flow value of the third water outlet is converted from 8 to a global cumulative flow value of 18.

[0061] In the present application, the first cumulative flow area, the second cumulative flow area and the third cumulative flow area are iteratively calculated in turn until all grids are calculated as global cumulative flows, that is, all local cumulative flow grids in the buffer zone are converted into global cumulative flows, and the final calculation result of the global cumulative flow is reflected in Figure e.

[0062] In this application, on the one hand, the iterative calculation in this application can greatly reduce the calculation cost. Compared with the traditional method of calculating the global cumulative flow based on the flow direction in a large range at one time, the iterative calculation method of this application greatly reduces the amount of repeated calculations in the traditional method. In addition, the traditional method calculates the global cumulative flow of all grids in a large range, while this application uses an iterative calculation method to only calculate the global cumulative flow in the buffer zone at a given outlet, which greatly reduces the amount of calculation and fully meets the needs of relocating the outlet and extracting the basin boundary.

[0063] In some embodiments, the largest first accumulated flow in the global accumulated flows on the main flow is used as the relocation result of the main flow; wherein the main flow relocation result is used to extract the main flow, and the first accumulated flow has the characteristics of being stable and error-free.

[0064] In some embodiments, the first cumulative flow has stable characteristics, including:

[0065] The first cumulative flow is closer to the given outlet than the global cumulative flow on the main stream; and / or

[0066] The first cumulative flow has an adjacent upstream river section that is farther from the given outlet; and / or

[0067] The length of the adjacent upstream river section of the first cumulative flow shall not be less than the side length of the buffer zone.

[0068] In some embodiments, the buffer zone is rectangular;

[0069] The first cumulative flow with stable characteristics has a length of adjacent upstream river section not less than 1 / 10-1 / 2 of the side length of the buffer zone.

[0070] In some embodiments, the first accumulated stream has error-free characteristics, including:

[0071] The first cumulative flow is closer to the given water outlet; and / or

[0072] There is no other global cumulative flow with stable characteristics upstream of the first cumulative flow.

[0073] In some embodiments, the 80-90 quantile of the first cumulative flow is used as a threshold to output the main flow relocation result.

[0074] In a specific example provided in this application, after calculating the global cumulative flow of the buffer zone, the main stream relocation and tributary relocation are performed, and one is selected as the final relocation result. The key step of the main stream relocation is to find the potential outlet relocation location, which is referred to as a "stable grid" in this application. Figure 3As shown, this application takes four stable grids of the main flow network determined based on global cumulative flow and flow direction data as an example, namely grids A, R, X and Y. It can be understood that Figure 3 Only four stable grids are shown. In actual applications, there may be multiple stable grids, among which the four stable grids are distributed on the main stream of the main stream network.

[0075] Furthermore, based on the global cumulative flow and flow direction data obtained, the four stable grids of the main stream network are analyzed to determine the stable grids, including:

[0076] By comparing with other global cumulative flow grids on the mainstream network, the stable global cumulative flow grid (stable grid) is close to a given outlet of the buffer zone (i.e. Figure 3 The center grid of the given outlet that needs to be relocated).

[0077] In this application, by comparing with other global cumulative flow grids on the mainstream network, the purpose of this step is to ensure that the downstream river section of the stable grid (determined by the flow direction, the global cumulative flow grid of the downstream river section is larger than the stable grid) is farther away from the center than the stable grid.

[0078] Furthermore, the stable grid has an adjacent upstream river section that is farther from the given outlet. The adjacent upstream river section is calculated based on the following method, which includes: starting from the stable grid, searching for the upstream grid with the smallest global cumulative flow gradient. The specific implementation process is as follows Figure 3 Take the X grid in as an example, find the grid flowing to grid X near X, that is, the "upstream grid", calculate the cumulative flow difference between grid X and its upstream grid, that is, the "cumulative flow gradient", and select the upstream grid with the smallest cumulative flow gradient as its upstream river section. Repeat this operation until the searched upstream grid is closer to the given outlet. For example, in the main stream with grid X, the upstream river section of grid X includes 5 upstream grids close to the given outlet; in the main stream with grid Y, the upstream river section of grid Y includes 6 upstream grids close to the given outlet. In other words, the 5 upstream grids close to the given outlet upstream of grid X constitute the adjacent upstream river section of grid X, and the 6 upstream grids close to the given outlet upstream of grid Y constitute the adjacent upstream river section of grid Y. Similarly, the above-mentioned method for judging the upstream river section is applicable to the judgment of grid A and grid R, so it will not be repeated.

[0079] Further, the length of the adjacent upstream river section of the stable grid is not less than the side length of the buffer zone. Specifically, the length of the adjacent upstream river section of the stable grid is not less than 1 / 4 of the side length of the buffer zone; it can be understood that, except for 1 / 4 of the side length of the buffer zone, the length of the adjacent upstream river section of the stable grid is not less than 1 / 10-1 / 2 of the side length of the buffer zone. Specifically, refer to Figure 3 , Figure 3is a 17×17 grid matrix, each grid is a unit grid with a side length of 1; Figure 3 In , it is represented by a square buffer with a side length of 17. Figure 3 , the adjacent upstream river section of the stable grid X includes 5 upstream grids, and the adjacent upstream river section of the stable grid Y includes 6 upstream grids; the lengths of the adjacent upstream river section of the stable grid Y and the adjacent upstream river section of the stable grid X are both greater than or equal to 4, therefore, it can be determined that both the grid X and the grid Y are stable grids. Similarly, the above determination method is applicable to the determination of the A grid and the R grid, so it will not be repeated.

[0080] In the present application, the above three methods for determining stable grids have the same purpose, namely, to ensure that the stable grid is closer to a given outlet than its adjacent upstream and downstream river sections, which is a characteristic that a potential correct outlet should have.

[0081] Furthermore, there may be errors in the initial selection of a stable grid. In this application, it is necessary to make an error judgment on the above-mentioned stable grid. The judgment method includes:

[0082] Determine whether the given outlet grid and the stable grid are in the same watershed. For example, in this application, the judgment is based on the stable grids A, R, X and Y. The stable grid with error 1 feature is generally located at the edge of the buffer zone of the given outlet, that is, far away from the given outlet. Figure 3 , it can be seen that relative to the stable grids A, R, and Y, the stable grid X has this error ( Figure 3 , the X grid is a stable grid with an error of 1), therefore, the stable grid X is excluded as a potential outlet relocation location.

[0083] Determine whether there is an error 2 feature of the stable grid upstream of each stable grid. For example, in this application, the judgment is based on the stable grids A, R, X and Y. In this application, the upstream is defined as the reverse flow direction with the smallest cumulative flow gradient. Figure 3 , since there is a stable grid A upstream of the stable grid Y, the stable grid Y has this error ( Figure 3 In the figure, the Y grid is a stable grid with error II, so the stable grid Y is excluded as a potential outlet relocation location. Furthermore, since the stable grid R is not upstream of the stable grid A with the smallest gradient, the stable grid A does not have this error and can be used as a potential outlet relocation location. Furthermore, the situation with the above error usually occurs on a curved river like the stable grid Y.

[0084] In this application, the above error judgment method is used to exclude the stable grid X that is too far from the given outlet and the redundant stable grid Y on the same main stream. Finally, the error-free stable grid A and the stable grid R are selected as potential outlet relocation positions. Further, the stable grid A with the largest cumulative flow and error-free and stable characteristics is used as the main stream relocation result, that is, the stable grid A is used as the first cumulative flow. The first cumulative flow is used to extract the main stream.

[0085] In some embodiments, the present application uses the 80-90 percentile of the cumulative flow corresponding to the stable grid (80%-90% of the cumulative flow value of the stable grid A) as the cumulative flow threshold, and returns the cumulative flow threshold to the stable grid A as the main stream relocation result. It is worth noting that in the present application, the cumulative flow threshold is an input parameter for identifying a grid with a cumulative flow greater than the threshold as a digital river network. The present application uses the 80-90 percentile of the cumulative flow corresponding to the stable grid (80%-90% of the cumulative flow value of the stable grid A) as the cumulative flow threshold, so that the main stream relocation is insensitive to the cumulative flow threshold, which can effectively improve the main stream positioning efficiency.

[0086] This application uses the 80-90% quantile of the cumulative flow of all grids in the buffer as the cumulative flow threshold. Any cumulative flow threshold within the 80-90% quantile interval will return a stable grid A as the main stream relocation result, that is, it is insensitive to the preset parameter of the cumulative flow threshold. On the other hand, compared with the traditional method that is sensitive to the search radius, an inappropriate search radius may cause a huge deviation in the extracted basin boundary. This application determines the outlet relocation search radius based on the side length of the buffer as a preset parameter, making the main stream relocation algorithm in this application insensitive to the preset parameter of the search radius.

[0087] In some embodiments, each global cumulative flow value on the main stream is subtracted from its adjacent global cumulative flow value, and the confluent cumulative flow is determined based on the size of the difference; wherein the main stream relocation result (stable grid A, i.e., the first cumulative flow) is used to extract the main stream, and the confluent cumulative flow is used to extract the tributaries of the main stream.

[0088] In some embodiments, the accumulated flow closest to a given outlet in the global accumulated flow on the tributary is used as the relocation result of the tributary.

[0089] In a specific embodiment, since the coordinates of a given outlet provided by a hydrological station are usually close to the confluence of a river network, it is necessary to determine the tributaries near the main stream after relocating the main stream. Figure 4 , Figure 4The process of tributary relocation is explained, including: based on the stable grid A (first cumulative flow) of the main stream relocation result, tracking the upstream and downstream of the stable grid A to draw the main stream, the upstream is the upstream grid with the smallest global cumulative flow gradient as described above, and the downstream is determined by the flow direction. The confluence cumulative flow is determined by comparing the difference between adjacent global cumulative flows on the main stream with a preset threshold; wherein the confluence cumulative flow is used to draw the tributaries of the main stream. Specifically, the global cumulative flow value of each grid can be subtracted from the global cumulative flow value of its adjacent grid, and the difference is 10% of the change in the cumulative flow value of each grid as the extraction standard for the confluence cumulative flow; it can be understood that the change in the cumulative flow value of each grid includes but is not limited to 10%, and the percentage of the change can be set as needed.

[0090] Specifically, in one embodiment of the present application, the global cumulative flow value of each grid on the main flow is subtracted from the cumulative flow value of its adjacent grids to find the grid whose cumulative flow value of each grid increases by at least 10% to determine the confluence cumulative flow, that is, Figure 4 Grid C in . Further, the upstream tributaries of the corresponding grid of the convergent cumulative flow (grid C) are tracked to extract the tributaries; wherein, the upstream tributaries are the reverse flow direction with the smallest cumulative flow gradient except for the main flow direction. Further, the reverse flow direction means that for the convergent cumulative flow grid (grid C), the cumulative flow grids of the nearby tributaries are located upstream of it, and the flow direction is opposite to that of the convergent cumulative flow grid. The cumulative flow gradient refers to the cumulative flow difference between the cumulative flow grid of the tributary and the convergent cumulative flow grid.

[0091] Furthermore, the cumulative flow closest to the given outlet in the global cumulative flow of the tributary is taken as the relocation result of the tributary. Figure 4 , the distance between grid B on the tributary and the given outlet is the shortest, so the tributary relocation result is returned to grid B, that is, grid B is a potential outlet relocation location. Compared with the traditional algorithm that only performs relocation once, the tributary relocation algorithm in this application considers multiple grids as the final outlet relocation result on the basis of the main stream relocation algorithm, which improves the accuracy of outlet relocation and thus ensures the accuracy of watershed extraction.

[0092] In some embodiments, a reference water outlet is determined at least based on the relocation results of the main stream and the tributaries, including: based at least on the relocation results of the tributaries and the main stream, selecting the cumulative flow closest to the given outlet coordinates as the reference water outlet.

[0093] In a specific embodiment of the present application, grid A is the main stream relocation result, grid B is the tributary relocation result, and grid C is the confluence accumulation flow. The grid closest to the given outlet coordinates is selected from the three grids as the final outlet relocation result, that is, the reference outlet. Figure 4In the figure, grid B is the final outlet relocation result.

[0094] In this application, this application only needs to refer to grid A and grid B, or grid A, grid B and grid C to determine the reference water outlet, without the need for other auxiliary information. It has the characteristics of simple operation and fast positioning speed, which can significantly improve work efficiency.

[0095] In addition, we design an optimized accumulated flow structure (OFAS) to speed up the extraction of watersheds. Figure 5 As shown, OFAS uses grid T as the outlet to calculate the cumulative flow, and grid T is determined by comparing with the cumulative flow threshold. In this application, the cumulative flow threshold can be selected as 6. Row representation Figure 5 Figure AB of the row, Col characterization Figure 5 In the column of Figure ab, FA represents the cumulative flow value corresponding to the grid determined by Row and Col. Figure 5 Attachment c of the specification describes OFAS in detail, which is a matrix of 3 times FA, where 3 represents the number of rows. Further, in the figure ab provided in the embodiment of the present application, 7 rows and 7 columns are included; that is, figure ab is a 7×7 matrix, including 49 cumulative flow grids. In the attached figure ab, the grids filled with the same section lines are the same cumulative flow areas. Specifically, the cumulative flow is calculated with grid T (watershed outlet) as the first reference outlet, and the watershed flowing into grid T is calculated. The watershed includes 43 cumulative flow grids (including 34 watershed grids with cumulative flows less than 6 and 9 grids with cumulative flows not less than 6). The sub-basin cumulative flow is calculated with grid S (sub-basin outlet) as the second reference outlet, and the sub-basin area flowing into grid S is calculated. The sub-basin area includes 11 cumulative flow grids (including 8 sub-basin grids less than 6 and three grids with cumulative flows not less than 6, and the three grids with cumulative flows not less than 6 are 11 (S grid), 9 and 6 respectively). Based on the above calculation, in the statistical diagram ab, 43 grids flow into the grid S or the grid T, that is, the FA value is 43.

[0096] Reference Accumulated Flow Structure (OFAS) Figure 5 In the attached figure c, OFAS cumulative flow structure consists of three rows, Row, Col and FA, which is equal to 43 here. OFAS has 43 columns, corresponding to the number of grids in the basin and subbasin in Figure a. The first two rows of OFAS are the row and column positions of the corresponding grid in Figure a, and the third row is the cumulative flow value corresponding to the grid determined by Row and Col in Figure ab.

[0097] Furthermore, the ordering of OFAS conforms to a basic property: for any given column with a cumulative flow value equal to FA, its subsequent FA-1 column will flow into the given column. Take grid T as an example. T is located in the 7th row and the 1st column, and its cumulative flow value is equal to 43. Therefore, its subsequent 42 columns flow to grid T, corresponding to the 42 grids in Figure a (including 34 watershed grids with cumulative flows less than 6 and 8 grids with cumulative flows not less than 6).

[0098] In this application, the ordering of OFAS brings advantages to the extraction of watersheds. For example, in order to extract the sub-watershed that flows into grid S, we first find the row and column position of grid S, that is, the 5th row and the 4th column. Then we search for the matching column in OFAS, that is, the first sub-watershed grid column in Figure c. Since the third row value of the first sub-watershed grid column is 11, the cumulative flow value of the second reference outlet grid S is determined to be 11. Further, according to the FA-1 column, the next 10 columns of the sub-watershed grid column (the column corresponding to grid S) are determined to flow into the given column. It should be noted that Figure 5 The cumulative flow value of the grid S in the attached figure c represents the number of its own columns and the number of columns behind it; thus, it can be determined that the number of columns behind the grid S in Figure c is 10.

[0099] Furthermore, based on the number of columns of the grid S itself and the number of columns behind it determined in Figure c, the sub-basin in Figure b can be quickly extracted. In the present application, the rapid extraction of sub-basins based on the cumulative flow structure (OFAS) of the present application can improve the efficiency of relocating the main stream outlet and the tributary outlet. In addition, unlike the traditional grid structure that only stores the cumulative flow, OFAS implicitly stores the flow direction information through sorting; therefore, a simple matching search can be used to extract the sub-basin with OFAS, thereby achieving the purpose of rapid extraction of the sub-basin, which can effectively reduce the computational cost.

[0100] Specifically, the automatic outlet relocation (AOR) algorithm based on this application can extract a large number of watersheds of different sizes. In order to evaluate the accuracy of the AOR algorithm, we designed a comparative experiment with the ArcGIS Snap PourPoint (SPP) algorithm in 1398 watersheds in the United States.

[0101] The experimental data includes watershed boundaries, watershed outlet coordinates (longitude and latitude), and flow directions. The given watershed boundaries are considered as the benchmark for evaluating the accuracy of AOR. The watershed boundaries come from the Geospatial Attributes of Gages for Evaluating Streamflow-II (GAGES-II) dataset. The GAGES-II dataset extracts 9332 watershed boundaries from the flow directions at a resolution of 30m, supplemented by manual inspection to ensure the accuracy of watershed extraction. We further performed two quality assurance steps and selected 1398 boundaries from the 9332 boundaries as the evaluation benchmark:

[0102] 1. Ensure that the absolute deviation of area between the watershed boundaries of the GAGES-II dataset and the watershed boundaries of the United States Geological Survey (USGS) National Water Information System is less than 5%.

[0103] 2. Select GAGES-II boundaries that have “good” visual consistency with the 10-digit Hydrological Unit Code (HUC10) boundaries in the National Watershed Boundary Dataset (USGS, 2013), where GAGES-II records visual consistency qualitatively using “good”, “fair”, or “poor”.

[0104] The outlet coordinates and flow direction of the watershed are the inputs of the AOR algorithm. The outlet coordinates of 1398 watersheds are obtained from the USGS National Water Information System. The flow direction used in the experiment is MERIT Hydro (Yamazaki et al., 2019). MERIT Hydro is a global flow direction dataset with a resolution of 90m and has been proven to be effective in extracting digital river networks and watershed boundaries. With the outlet coordinates and MERIT Hydro as input, we use the AOR algorithm to extract 1398 watershed boundaries. The boundaries extracted by AOR are compared with those of GAGES-II to evaluate the accuracy of the AOR algorithm.

[0105] We used an indicator to estimate the accuracy of the extraction. If the intersection area between the extracted boundary and the GAGES-II boundary is greater than 90%, the extraction is considered correct. For example, suppose the watershed extracted by an AOR algorithm and the GAGES-II watershed are D and G, respectively. Their intersection area is D∩G. If (D∩G) / D and (D∩G) / G are both greater than 90%, the watershed boundary extracted by the AOR algorithm is considered correct. Compared with the traditional method of checking whether the areas of two watersheds are similar, the advantage of this indicator is that it can avoid misidentifying two watersheds with similar areas but non-intersecting boundaries.

[0106] The test environment is Python 3.7.7 on Windows 10. The hardware includes an Intel i9-9900X CPU and 64G memory, where the memory is mainly used to store flow directions and cumulative flows. We use C language to implement the global cumulative flow calculation algorithm in the buffer as a C extension of Python. Both the main stream relocation algorithm and the tributary relocation algorithm are written in pure Python.

[0107] The Snap Pour Point (SPP) algorithm of ArcGIS was used as an alternative method for comparison with the AOR algorithm. The SPP algorithm relocates the outlet to the grid with the highest cumulative flow within a search radius. According to Jenson (1991), the search radius of the SPP algorithm was set to 2 km in this application.

[0108] The AOR algorithm in this application only takes the flow direction and outlet coordinates as input. Furthermore, the AOR algorithm was compared and analyzed with the ArcGIS SPP algorithm using 1398 watershed boundaries with manual quality assurance as the evaluation benchmark. The results show that the AOR algorithm correctly extracted 1316 watersheds with an accuracy of 94.1%, and the SPP algorithm correctly extracted 1137 with an accuracy of 81.3%. The AOR algorithm performed well in large watersheds and reduced the error of the SPP algorithm by 70.6% in small watersheds to prove that the AOR algorithm of this application is more suitable for delineating watersheds of different sizes. In addition, the AOR algorithm extracted 1398 watersheds in 58 minutes using a single CPU core on a flow direction with a resolution of 90m through the optimized cumulative flow structure (OFAS). Therefore, the AOR algorithm provides a large number of opportunities for watershed extraction for hydrological simulation in the era of big data.

[0109] In one example, we ran the proposed AOR algorithm and ArcGIS SPP algorithm on the same flow direction dataset (MERIT Hydro). The AOR algorithm delineated 1398 watersheds at 90m resolution in 58 minutes. Figure 6The results of 1398 watershed delineations of the two algorithms are shown. The AOR algorithm correctly delineated 1316 watersheds and the SPP algorithm correctly delineated 1137 watersheds, using the watershed boundaries provided by the Geospatial Attributes of Gages for Evaluating Streamflow-II (GAGES-II) dataset as an evaluation reference. The AOR algorithm has 197 correct watershed delineations (large dots), while the SPP algorithm fails in these watersheds. There are 64 watersheds (triangular boxes) that are incorrectly delineated by both algorithms. The AOR algorithm has defects in 18 watersheds (squares), which is contrary to the correct description of the SPP algorithm in these watersheds. In summary, the accuracy of the AOR algorithm is 94.1%, while the accuracy of the SPP algorithm is 81.3%.

[0110] We divided the 1398 watersheds into four equal parts using the quartiles of the watershed area, namely small watersheds, medium watersheds, large watersheds, and extra-large watersheds. The SPP algorithm had 136, 55, 40, and 30 incorrect extractions for the small, medium, large, and extra-large watershed areas, respectively. This result indicates that the SPP algorithm performs better in large watersheds than in small watersheds. This watershed area-dependent performance is expected because the SPP algorithm relocates the outlet to the grid with the highest cumulative flow within the search radius. Correspondingly, the AOR algorithm had 40, 6, 20, and 16 incorrect extractions. The significant improvement of the AOR algorithm over the SPP algorithm occurred in small watersheds, indicating that the AOR algorithm is more suitable for the extraction of watersheds of different areas.

[0111] It should be noted that Figure 6 Results of the proposed AOR algorithm and ArcGIS SPP algorithm for 1398 watershed delineation. The manually quality-assured watershed boundaries were used as reference boundaries to evaluate the accuracy of the two algorithms. √ indicates that the boundary derived by the algorithm matches the reference boundary, and × indicates that they do not match. The number of squares, triangles, large dots, and small dots is 18, 64, 197, and 1119, respectively.

[0112] The present invention has been described above with reference to the embodiments of the present invention. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of the present invention. The scope of the present invention is defined by the appended claims and their equivalents. Without departing from the scope of the present invention, a person skilled in the art may make various substitutions and modifications, which should all fall within the scope of the present invention.

Claims

1. A watershed extraction method based on outlet relocation, characterized in that: include: Determine the buffer zone based on the given outlet, iteratively calculate the local cumulative flow in the buffer zone according to the flow direction data and output the global cumulative flow; Determine the mainstream network based on global cumulative flow and flow direction data, and relocate the given outlet to the mainstream network; Relocate the outlet coordinates to the tributary of the main stream; Determine a reference outlet at least according to the main stream relocation result and the tributary relocation result, and extract the watershed according to the reference outlet; The buffer zone includes a plurality of accumulated flow areas; The local cumulative flows in multiple cumulative flow areas are calculated according to the flow direction data to output the global cumulative flows respectively, including: The multiple cumulative flow areas include at least a first cumulative flow area and a second cumulative flow area; the highest local cumulative flow value of the buffer zone is used as the first outlet of the first cumulative flow area, and the global cumulative flow corresponding to all local cumulative flows in the first cumulative flow area that flow into the first outlet is calculated; Taking the largest local cumulative flow outside the first cumulative flow area in the buffer zone as the second water outlet, calculating the global cumulative flow corresponding to all local cumulative flows in the second cumulative flow area flowing into the second water outlet; Subtract each global cumulative flow value on the main stream from its adjacent global cumulative flow value, and determine the confluence cumulative flow according to the difference; the confluence cumulative flow is used to extract the tributaries of the main stream; The accumulated flow closest to the given outlet in the global accumulated flow on the tributary is used as the relocation result of the tributary; The step of determining the reference water outlet at least according to the main stream relocation result and the tributary relocation result comprises: Based on at least the relocation results of the tributaries and the relocation results of the main stream, the cumulative flow closest to the given outlet coordinates is selected as the reference outlet.

2. The method according to claim 1, characterized in that The largest first accumulated flow in the global accumulated flows on the main flow is used as the relocation result of the main flow; wherein the main flow relocation result is used to extract the main flow, and the first accumulated flow has the characteristics of stability and no errors.

3. The method according to claim 2, characterized in that The first cumulative flow has stable characteristics, including: The first cumulative flow is closer to the given water outlet than the global cumulative flow on the main stream; and / or The first cumulative flow has an adjacent upstream river section that is farther from the given water outlet; and / or The length of the adjacent upstream river section of the first cumulative flow is not less than the side length of the buffer zone.

4. The method according to claim 3, characterized in that The buffer zone is rectangular; The first cumulative flow with stable characteristics has a length of an adjacent upstream river section that is not less than 1 / 10-1 / 2 of the side length of the buffer zone.

5. The method according to claim 2, characterized in that: The first accumulated stream has error-free characteristics, including: The first cumulative flow is closer to the given water outlet; and / or There is no other global cumulative flow with stable characteristics upstream of the first cumulative flow.

6. The method according to claim 2, characterized in that The 80-90 quantile of the first cumulative flow is used as the threshold to output the main stream relocation result.

Images