A large-scale watershed hydrological modeling method based on variable grids

Abstract

The present invention provides a large-scale watershed hydrological modeling method based on a variable grid, which belongs to the field of large-scale distributed hydrological modeling. The large-scale watershed hydrological modeling method is as follows: first, the size of the basin basic grid is determined based on the spatial resolution of the available basic data and weighing the calculation time of the distributed hydrological model at different resolutions; second, the grid level is determined according to the rainfall level of a flood event and the flash flood prevention level; finally, according to the grid level of different partitions, the basic grid is aggregated into coarse grids of different levels to complete the variable grid hydrological modeling of the watershed. The present invention is applicable to large-scale watershed flood forecasting systems, and can meet the timeliness requirements of flood control decisions while ensuring the refined modeling of key areas.

Description

Technical Field

[0001] The present invention belongs to the field of large-scale distributed hydrological modeling, and relates to a large-scale watershed distributed hydrological modeling method based on a variable grid. Background Art

[0002] River basin hydrological models are the core of flood forecasting systems. With increasing demands for flood forecast accuracy and spatiotemporal refinement, lumped hydrological models are evolving into distributed models capable of precisely describing the spatiotemporal characteristics of rainfall and underlying surfaces. However, large-scale river basins cover vast areas and have complex and diverse underlying surface conditions. Traditional fixed-grid modeling requires basin-wide distributed flood forecasting schemes based on 1km grids, requiring hundreds of thousands or even millions of grid divisions. This requires excessively long calculation times for a single flood, failing to meet the timeliness requirements of flood forecasting. Consequently, distributed hydrological refinement modeling has largely remained at the theoretical level or applied to small and medium-sized river basins with relatively good data.

[0003] A rainstorm flood often occurs in a localized area of a large watershed. The runoff generation and convergence process in the center of the rainstorm is crucial to flood formation. Using a fine grid can accurately describe the runoff generation and convergence process in the center of the rainstorm, while the non-major runoff-generating areas that occupy a large proportion of the watershed have less impact on flood formation. Using a coarse grid does not reduce the overall flood forecast accuracy of the large watershed, but it can significantly improve computational efficiency and meet the computational time requirements of flood forecasting. Previous research has explored distributed hydrological refinement modeling methods. For example, Chinese invention patent CN115130396A proposes a distributed hydrological refinement modeling method for river-type reservoir areas. By dividing sub-basins and slope watersheds and constructing refined hydrological response units, it fully considers the spatial heterogeneity of the underlying surface of the slope units. However, this method uses high-resolution grids throughout the entire watershed, resulting in a significant increase in overall computational overhead and reduced model efficiency. Chinese invention patent CN105160121A proposes a finite element-controlled distributed hydrological modeling method. This method divides finite element units based on the distribution of rain gauges and underlying surface characteristics, which can improve modeling efficiency and simulation accuracy to a certain extent. However, this method is not sufficiently refined in the center of the rainstorm area, making it difficult to accurately depict the local hydrological response process caused by heavy rainfall, thereby limiting the model's application performance in extreme event forecasting. The above method has shortcomings in terms of accuracy in the center of the rainstorm area and efficiency in non-major runoff-producing areas.

[0004] Therefore, in order to improve the modeling precision of key flood control areas in large river basins while taking into account the overall calculation time of the model, a large-scale river basin distributed hydrological modeling method with a variable grid is proposed. Fine grid modeling is used for key flood control areas such as rainstorm centers and densely populated areas, and coarse grid modeling is used for other non-key areas. Summary of the Invention

[0005] In response to the problems existing in the existing technology, the present invention provides a large-scale watershed distributed hydrological modeling method based on variable grids, which realizes fine grid modeling of key areas in the large watershed and coarse grid modeling of non-key areas, ensuring model accuracy while meeting the requirements of computational timeliness.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] A large-scale watershed distributed hydrological modeling method based on variable grids. The research idea of the large-scale watershed distributed hydrological modeling method is to adopt the idea of grid "aggregation" to aggregate fine grids into coarse grids. It mainly includes three parts: determining the size of the basin's basic grid, determining the grid level of different areas in the basin, and aggregating the basin's basic grid into coarse grids of different levels. First, the size of the basin's basic grid is determined based on the spatial resolution of the available basic data (underlying surface data and meteorological and hydrological data) and weighing the calculation time of the distributed hydrological model at different resolutions. Secondly, on this basis, the grid level is determined according to the rainfall level of a flood event and the flash flood prevention and control level. Finally, according to the grid levels of different partitions, the basic grids are aggregated into coarse grids of different levels to complete the variable grid hydrological modeling of the watershed. Specifically, the following steps are included:

[0008] Step 1: Determine the basic grid size of the watershed.

[0009] The basin basic grid is the smallest grid obtained by analyzing the collected basin basic data required for distributed hydrological modeling. The specific process is as follows:

[0010] Step 1.1: Collect the basic watershed data required for distributed hydrological modeling. The basic watershed data includes meteorological and hydrological data and underlying surface data. Meteorological and hydrological data include rainfall, evaporation, temperature, wind speed data, and their station locations and spatial distribution. Underlying surface data includes remote sensing image data including topography, soil, vegetation, and land use.

[0011] Step 1.2: Use the basin basic data collected in step 1.1 as input data for the distributed hydrological model. Classify the impact of the above input data on the distributed hydrological model and determine the requirements of various types of meteorological and hydrological data and underlying surface data on the grid size of the distributed hydrological model. Specific requirements include the following:

[0012] Step 1.2.1: Meteorological and hydrological data have a significant impact on the modeling of distributed hydrological models. For this type of data, the basin base grid size should not be lower than the spatial resolution of the meteorological and hydrological data.

[0013] (1)

[0014] Among them, Grid 降雨 Grid 蒸发 Grid 气温 Grid 风速 are the spatial resolutions of rainfall, evaporation, temperature, and wind speed data, respectively.

[0015] In step 1.2.2, the underlying surface data is obtained through remote sensing images. The spatial resolution of some underlying surface remote sensing images is much higher than that of basic meteorological and hydrological data (for example, the resolution of terrain DEM images can be better than 30m), which can provide more detailed surface cover information. Based on the spatial resolution of the underlying surface data, the grid size requirements are as follows:

[0016] (2)

[0017] Among them, Grid 地形 Grid 土壤 Grid 植被 Grid 土地利用 Represents the spatial resolution of terrain, soil, vegetation and land use respectively.

[0018] However, when the underlying surface data is relatively uniform in space or is generally consistent, even if a very fine grid is used for division, the impact on the hydrological modeling results is very small, but it will significantly increase the time required for model calculation. Therefore, although the underlying surface information can be obtained through remote sensing images with a spatial resolution better than 30m, it is not necessary to use such ultra-high resolution data when constructing the basin basic grid. The spatial distribution characteristics of the underlying surface and the time required for hydrological model calculation should be weighed, and the grid size should be reasonably controlled. 下垫面 , in order to control the calculation time while ensuring the simulation accuracy.

[0019] Step 1.3: Based on the requirements of the above various input data for the spatial resolution of distributed hydrological modeling, determine the reasonable basic grid size of the basin. 基础网格 , the specific formula is:

[0020] (3)

[0021] In summary, the basic grid size of the watershed is determined.

[0022] Step 2: Determine the grid levels in different areas of the basin.

[0023] First, a typical flood process in the basin was selected, and according to the spatial distribution of rainfall in the flood, the maximum rainfall of different areas in the basin for 24 hours was calculated. max ,24h Then, the spatial distribution and level of flash flood control areas in the basin were collected, and the spatial distribution of rainfall was spatially superimposed with the flash flood control areas. max ,24hThe key areas of concern within the basin are determined based on the level of flood control and prevention. The rules for dividing grid levels in different areas are as follows:

[0024] (1) For areas with high risk of flash floods:

[0025] (4)

[0026] Among them, Grid represents the spatial resolution of the ith partition of the watershed.

[0027] P max ,24h The central area of the rainstorm exceeding 100mm is a key flood control area and needs to be modeled using the basic grid. 基础网格 ; For P max ,24h In the non-rainstorm center area with a precipitation level less than 100 mm, coarse grids with multiples of 3, 5, and 10 are used for modeling.

[0028] (2) For areas with medium and low risk of flash floods:

[0029] (5)

[0030] When P max ,24h Less than or equal to 50mm, use 10 times the grid 基础网格 Size modeling; when P max ,24h Over 50mm, use 5 times the grid 基础网格 Size modeling.

[0031] In summary, the grid level division results of different watershed partitions are obtained.

[0032] Step 3: Aggregate the basin basic grid into coarse grids of different levels.

[0033] Based on the results of the grid level division of different watershed partitions obtained in step 2, variable grid aggregation is performed to aggregate the basin basic grids into coarse grids of different levels. For high-risk areas of flash floods located in the center of heavy rain, the basin basic grid size is retained for modeling to finely describe the runoff generation and convergence process in the area. For medium- and low-risk areas of flash floods or non-storm center areas, the basin basic grids are aggregated into coarse grids of different multiples. Variable grid aggregation includes two parts: grid runoff area calculation and grid convergence flow direction extraction. The specific contents are as follows:

[0034] Step 3.1, calculation of grid runoff area. Variable grid aggregation causes the boundaries of basin base grids and coarse grids to overlap, resulting in errors in the calculation of coarse grid runoff area. Therefore, by establishing a mapping relationship between coarse grids and basin base grids, the number of basin base grids contained in each coarse grid is counted, and the runoff area A of the base grids contained in the coarse grids is accumulated. 基础网格,i Get the runoff area A of the coarse grid 粗网格 .

[0035] (6)

[0036] Where n represents the number of base grids contained in the coarse grid.

[0037] Step 3.2, grid confluence flow direction extraction. Variable grid aggregation changes the cell size of different grids, resulting in the traditional flow direction calculation method of fixing one grid confluence to no longer be applicable. Therefore, it is necessary to perform multiple rounds of iterative search and correction between multiple grid sizes with incorrect flow directions to reconstruct the variable grid river network topology. The specific approach is: first identify the grid cells with interrupted or abnormal flow directions, and then find the target cells with lower terrain and reasonable connection relationships in the neighboring areas, and adjust their flow directions to point to the correct downstream cells. Through multiple iterations to continuously correct the abnormal points, the complete and connected river network topology is finally reconstructed to ensure that the water can converge smoothly in the entire area, especially to maintain the continuity and accuracy of the confluence path at the grid junction.

[0038] At this point, the variable grid division work of the large river basin has been completed, and a variable grid distributed hydrological modeling scheme for the large river basin has been constructed.

[0039] The beneficial effects of the present invention are:

[0040] This paper provides a large-scale distributed hydrological modeling method based on a variable grid. While ensuring modeling accuracy in key areas, including the center of a rainstorm, it effectively reduces model computation time in non-key areas by using a coarse grid, achieving a balance between accuracy and efficiency in distributed hydrological models. This method features a flexible grid-level division mechanism that adapts to different basin characteristics and flood forecasting requirements, meeting the timeliness requirements of flood forecasting and providing an efficient and reliable modeling solution for large-basin flood forecasting systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flow chart of the present invention.

[0042] Figure 2 This is the result of grid aggregation of the Jialing River variable grid distributed hydrological modeling.

[0043] Figure 3 This is the result of the river network flow direction calculation of the Jialing River variable grid distributed hydrological modeling.

[0044] Figure 4 This is a distribution map of stations in the Jialing River Basin (specifically, the stations are A, B, C, D, E, F, G, H, I, and J).

[0045] Figure 5 This is the flood simulation result of the hydrological model at site A.

[0046] Figure 6 This is the flood simulation result of the hydrological model at site B.

[0047] Figure 7 This is the flood simulation result of the hydrological model at site C.

[0048] Figure 8 This is the flood simulation result of the hydrological model at site D.

[0049] Figure 9 This is the flood simulation result of the hydrological model at Site E.

[0050] Figure 10 This is the flood simulation result of the hydrological model at site F.

[0051] Figure 11 This is the flood simulation result of the hydrological model at site G.

[0052] Figure 12 This is the flood simulation result of the hydrological model at site H.

[0053] Figure 13 This is the flood simulation result of the hydrological model at site I.

[0054] Figure 14 This is the flood simulation result of the hydrological model at site J. DETAILED DESCRIPTION

[0055] Based on the field of large-scale watershed distributed hydrological modeling, the present invention proposes a large-scale watershed distributed hydrological modeling method based on a variable grid.

[0056] The present invention will be further described below through examples.

[0057] The Jialing River, a typical large river basin, was selected as the research area. The Jialing River basin covers an area of 157,000 km 2 Located in the Daba Mountains, a rainstorm zone, the basin experiences frequent rainstorms. The upper reaches of the Jialing River contain numerous flash flood control areas, making localized torrential rains highly susceptible to flash floods within a short period of time. Therefore, the numerous and extensive flash flood control areas along the Jialing River place high demands on the accuracy and timeliness of hydrological modeling.

[0058] Step 1: Determine the basic grid size of the watershed.

[0059] The basin base grid is the smallest grid derived from analyzing the collected basic basin data required for distributed hydrological modeling. The specific process is as follows: Step 1.1: Collect the basic basin data required for distributed hydrological modeling of the Jialing River Basin. This basic data includes meteorological and hydrological data and underlying surface data for the Jialing River Basin. Meteorological and hydrological data include rainfall, evaporation, temperature, wind speed data, and their station locations and spatial distribution. The underlying surface data includes remote sensing imagery data on topography, soils, vegetation, and land use.

[0060] Step 1.2: Use the basin basic data collected in step 1.1 as input data for the distributed hydrological model. Classify the impact of the above input data on the distributed hydrological model and determine the requirements of various types of meteorological and hydrological data and underlying surface data on the grid size of the distributed hydrological model. Specific requirements include the following:

[0061] Step 1.2.1: Meteorological and hydrological data have a significant impact on the modeling of distributed hydrological models. For this type of data, the basin base grid size should not be lower than the spatial resolution of meteorological and hydrological data. The density of rainfall stations in the Jialing River Basin is 442km. 2 / station, evaporation, temperature and wind speed data are based on China's monthly potential evapotranspiration data with a spatial resolution of 1km, so the Grid 水文气象 is 1km.

[0062] (1)

[0063] Among them, Grid 降雨 Grid 蒸发 Grid 气温 Grid 风速 are the spatial resolutions of rainfall, evaporation, temperature, and wind speed data, respectively.

[0064] In step 1.2.2, the underlying surface data is obtained through remote sensing images. The spatial resolution of the underlying surface data such as soil and vegetation is 1km, and the spatial resolution of the terrain data is 30m. The spatial resolution of the terrain data is much higher than that of the basic meteorological and hydrological data, and can provide more detailed surface cover information. According to the spatial resolution of the underlying surface data, the requirements for the grid size are as follows:

[0065] =30m (2)

[0066] Among them, Grid 地形 Grid 土壤 Grid 植被 Grid 土地利用 Represents the spatial resolution of terrain, soil, vegetation and land use respectively.

[0067] However, since the elevation change of 30m precision DEM data has little effect on the hydrological modeling results, it will significantly increase the calculation time. Therefore, the spatial distribution characteristics of the underlying surface and the calculation time of the hydrological model should be weighed and the grid 下垫面 The distance between the nodes is 1 km to improve the calculation efficiency.

[0068] Step 1.3: Based on the requirements of the above various input data for the spatial resolution of distributed hydrological modeling, determine the reasonable basic grid size of the basin. 基础网格 For 1km, the specific formula is:

[0069] =1km (3)

[0070] In summary, determine the basic grid size of the basin 基础网格 The area of 1km is 1km, and the number of 1km basic grids in the Jialing River Basin is 152,243.

[0071] Step 2: Determine the grid levels in different areas of the basin.

[0072] The typical flood process from August 9 to 29, 2020 in the Jialing River Basin was selected. First, the basin was divided into 10 sub-areas based on the spatial distribution of the flash flood control areas in the Jialing River Basin. Then, the maximum rainfall of different areas in the basin for 24 consecutive hours was calculated. max ,24h Finally, the grid sizes of different areas are determined according to the grid level division rules. The grid level division rules of different areas are as follows:

[0073] (1) For areas with high risk of flash floods:

[0074] (4)

[0075] Where Gridi represents the spatial resolution of the i-th partition of the watershed.

[0076] P max ,24h The central area of the rainstorm exceeding 100mm is a key flood control area and needs to be modeled using the basic grid. 基础网格 ; For P max ,24h In the non-rainstorm center area with a precipitation level less than 100 mm, coarse grids with multiples of 3, 5, and 10 are used for modeling.

[0077] (2) For areas with medium and low risk of flash floods:

[0078] (5)

[0079] When P max ,24h Less than or equal to 50mm, use 10 times the grid 基础网格 Size modeling; when P max ,24h Over 50mm, use 5 times the grid基础网格 Size modeling.

[0080] In summary, the grid level division results of different watershed partitions are obtained, see Table 1.

[0081] Table 1 Grid level division results for different zones in the Jialing River Basin

[0082]

[0083] Step 3: Aggregate the basin basic grid into coarse grids of different levels.

[0084] Based on the results of the grid level division of the different watershed partitions obtained in the second step, variable grid aggregation is performed. For high-risk areas of flash floods located in the center of heavy rain, the basin basic grid size is retained for modeling to finely describe the runoff generation and convergence process in the area. For medium- and low-risk areas of flash floods or areas outside the center of heavy rain, the basin basic grid is aggregated into coarse grids of different multiples. Variable grid aggregation includes two parts: grid runoff area calculation and grid convergence flow direction extraction. The specific contents are as follows:

[0085] Step 3.1, calculation of grid runoff area. According to the grid level division results of different partitions, aggregate the basic grids of different partitions into coarse grids of different levels and calculate the grid runoff area, as shown in the attached figure. Figure 2 The total number of variable-grid hydrological modeling grids is 8141, of which 1km, 3km, 5km, and 10km grids are 3329, 189, 3935, and 688, respectively.

[0086] Step 3.2: Extract the flow direction of the grid confluence. Calculate the flow direction of the river network of the variable grid and perform multiple rounds of iterative search and correction between multiple grid sizes with incorrect flow directions, reconstruct the topological relationship of the variable grid river network, and the flow direction extraction results are shown in the attached figure. Figure 3 shown.

[0087] At this point, the variable grid division work of the Jialing River Basin has been completed, and a variable grid distributed hydrological modeling scheme for the Jialing River Basin has been constructed.

[0088] The variable-grid distributed hydrological model was run on a laptop computer equipped with an 11th Gen Intel(R) Core(TM) i7-11800H processor with a base frequency of 2.30 GHz and 8 cores and 16 threads. Compared to the 5 minutes and 12 seconds required to calculate runoff and flow for a single flood using a 1km grid model, the variable-grid distributed hydrological modeling method calculated runoff and flow for all subsections of the Jialing River in just 29 seconds. The variable-grid distributed hydrological modeling method simulated floods with accuracy as follows: Figures 5 to 14 As shown in the figure, while ensuring the refinement of the model, the model calculation time is greatly reduced.

[0089] The above-described embodiments merely express the implementation methods of the present invention, but should not be understood as limiting the scope of the patent of the present invention. It should be pointed out that for those skilled in the art, several variations and improvements can be made without departing from the concept of the present invention, and these all fall within the scope of protection of the present invention.

Claims

1. A large-scale watershed hydrological modeling method based on a variable grid, characterized by: The large-scale watershed hydrological modeling method is as follows: first, the size of the basin's basic grid is determined based on the spatial resolution of available basic data and the computational time required for distributed hydrological models at different resolutions; second, the grid level is determined based on the rainfall level of a flood event and the level of flash flood control; finally, based on the grid levels of different partitions, the basic grid is aggregated into coarse grids of different levels to complete the watershed variable grid hydrological modeling, including the following steps: Step 1: Determine the basic grid size of the watershed; The basin basic grid is the smallest grid obtained by analyzing the collected basin basic data required for modeling the distributed hydrological model; Step 2: determine the grid levels of different areas in the basin; Step 3: Aggregate the basin basic grid into coarse grids of different levels; Based on the grid-level division results of different watershed regions obtained in step 2, variable grid aggregation is performed to aggregate the basin's basic grids into coarse grids of different levels. For high-risk areas for flash floods located in the center of a rainstorm, the basin's basic grid size is retained for modeling to accurately describe the runoff generation and convergence processes within the area. For medium- and low-risk areas for flash floods or areas outside the center of a rainstorm, the basin's basic grids are aggregated into coarse grids of different multiples. Variable grid aggregation includes two parts: grid runoff area calculation and grid convergence flow direction extraction. The step 1 is specifically as follows: Step 1.1: Collect the basic watershed data required for distributed hydrological modeling; the basic watershed data includes meteorological and hydrological data and underlying surface data. Step 1.2: Use the basic watershed data collected in step 1.1 as input data for the distributed hydrological model. Classify the impact of the above input data on the distributed hydrological model and determine the requirements of various types of meteorological and hydrological data and underlying surface data on the grid size of the distributed hydrological model. Step 1.3: Based on the requirements of the above various input data for the spatial resolution of distributed hydrological modeling, determine the reasonable basic grid size of the basin. 基础网格 , the specific formula is: Grid 基础网格 =min(Grid 水文气象 ,Grid 下垫面 ) In summary, the basic grid size of the watershed is determined.

2. The large-scale watershed hydrological modeling method based on a variable grid according to claim 1 is characterized in that: In step 1.1, the meteorological and hydrological data include rainfall, evaporation, temperature, wind speed data and their site locations and spatial distribution, and the underlying surface data include remote sensing image data including topography, soil, vegetation, and land use.

3. The large-scale watershed hydrological modeling method based on a variable grid according to claim 1 is characterized in that: The step 1.2 is specifically as follows: Step 1.2.1: For meteorological and hydrological data, the basin base grid size should not be lower than the spatial resolution of the meteorological and hydrological data: Grid 水文气象 =min(Grid 降雨 ,Grid 蒸发 ,Grid 气温 ,Grid 风速 ) Among them, Grid 降雨 Grid 蒸发 Grid 气温 Grid 风速 are the spatial resolutions of rainfall, evaporation, temperature, and wind speed data, respectively; Step 1.2.2: The underlying surface data is obtained through remote sensing images. According to the spatial resolution of the underlying surface data, the grid size requirements are as follows: Grid 下垫面 =min(Grid 地形 ,Grid 土壤 ,Grid 植被 ,Grid 土地利用 ) Among them, Grid 地形 Grid 土壤 Grid 植被 Grid 土地利用 Represents the spatial resolution of terrain, soil, vegetation and land use respectively.

4. The large-scale watershed hydrological modeling method based on a variable grid according to claim 1 is characterized in that: The step 2 is specifically as follows: First, a typical flood process in the basin was selected, and according to the spatial distribution of rainfall in the flood, the maximum rainfall of different areas in the basin for 24 hours was calculated. max,24h Then, the spatial distribution and level of flash flood control areas in the basin were collected, and the spatial distribution of rainfall was spatially superimposed with the flash flood control areas. max,24h and mountain torrent control levels to determine key areas of concern within the basin; and determine the grid level division rules for different areas.

5. The large-scale watershed hydrological modeling method based on a variable grid according to claim 4 is characterized in that: In step 2, the grid level division rules for different areas are as follows: (1) For areas with high risk of flash floods: Among them, Grid i represents the spatial resolution of the ith partition of the watershed; P max,24h The central area of the rainstorm exceeding 100mm is a key flood control area and needs to be modeled using the basic grid. 基础网格 ; For P max,24h In the non-rainstorm center area with a rainfall of less than 100 mm, coarse grids with multiples of 3, 5, and 10 are used for modeling; (2) For areas with medium and low risk of flash floods: When P max,24h Less than or equal to 50mm, use 10 times the grid 基础网格 Size modeling; when P max,24h Over 50mm, use 5 times the grid 基础网格 size modeling; In summary, the grid level division results of different watershed partitions are obtained.

6. The large-scale watershed hydrological modeling method based on a variable grid according to claim 4 is characterized in that: The step 3 is specifically as follows: Step 3.1, calculation of grid runoff area; By establishing a mapping relationship between the coarse grid and the basin basic grid, the number of basin basic grids contained in each coarse grid is counted, and the runoff area A of the basic grids contained in the coarse grid is accumulated. 基础网格,i Get the runoff area A of the coarse grid 粗网格 ; Where n represents the number of basic grids contained in the coarse grid; Step 3.2, grid confluence flow direction extraction; Multiple rounds of iterative search and correction are performed among various grid sizes with incorrect flow directions to reconstruct the topological relationship of the variable grid river network.

7. The large-scale watershed hydrological modeling method based on a variable grid according to claim 6 is characterized in that: Specifically, step 3.2 is as follows: first, identify grid cells with interrupted or abnormal flow directions, then search for target cells with lower terrain and reasonable connectivity in their adjacent areas, and adjust their flow directions to point to the correct downstream cells; continuously correct the abnormal points through multiple iterations, and finally rebuild a complete and connected river network topology to ensure that water can converge smoothly throughout the entire area.

Images