A high-precision mountainous river roughness extraction method based on multi-source remote sensing images

Abstract

The application discloses a high-precision mountainous river roughness extraction method based on multi-source remote sensing images, and comprises the following steps: S1, multi-source remote sensing image data acquisition and data preprocessing; S2, river center line extraction and river range determination based on remote sensing images; S3, vegetation coverage calculation and river vegetation area identification; 4, river vegetation height calculation; S5, hydraulic radius and water depth calculation; S6, river roughness inversion based on vegetation-hydraulic parameters; the application has the advantages of low cost and strong timeliness; without large-scale field measurement, the application mainly adopts multi-source remote sensing images, so that the data acquisition cost is obviously reduced; the mountainous river parameter extraction can be realized, the cycle is shortened to 1-2 weeks, and the dynamic monitoring demand is met; reliable parameters are provided for high-precision hydrological simulation; the application has wide applicability and strong expansibility; the application is suitable for mountainous rivers with different terrains and different vegetation types (herbaceous, shrub and arbor); InSAR deformation data can be added subsequently, and the river scouring and silting height change monitoring function is expanded.

Description

Technical Field

[0001] This invention belongs to the field of remote sensing technology and hydrology and water resources, specifically involving a high-precision method for extracting roughness of mountain rivers based on multi-source remote sensing images. It is applicable to scenarios such as hydrological simulation of mountain rivers, flood risk assessment, ecological restoration planning and water resources management, and is suitable for mountain watersheds with complex terrain and high difficulty in conventional measurement. Background Technology

[0002] Mountain rivers, as key carriers of the watershed hydrological cycle, have their channel extent, vegetation cover, and hydraulic parameters (such as roughness) directly impacting flow patterns. They serve as core foundational data supporting hydrological model construction, accurate flood forecasting, and scientific ecological protection. In the field of acquiring relevant parameters for mountain rivers, traditional technical approaches mainly fall into two categories:

[0003] 1. Field measurement method: This method uses professional equipment such as total station and RTK to conduct precise field measurements of river width, depth, cross-sectional morphology and riverbed sediment. At the same time, it combines quadrat surveys to obtain vegetation height and coverage information, and then calculates the hydraulic radius and roughness using empirical formulas.

[0004] 2. Hydrological model inversion method: Based on measured data such as flow rate and water level observed by hydrological stations, the flow state of the river is simulated by hydrodynamic models (such as MIKE11 and HEC-RAS) to realize the inversion of key parameters such as river roughness.

[0005] Existing methods for extracting parameters from mountainous river channels have the following key problems in practical applications:

[0006] 1. High cost and limited coverage of field measurements: Mountainous terrain is complex, and field measurements require a lot of manpower and resources. They are also limited by weather and terrain, making it difficult to achieve large-scale periodic dynamic monitoring. In uninhabited areas and areas prone to geological disasters, it is even impossible to carry out field measurements, resulting in data gaps.

[0007] 2. Time lag and limited accuracy: Traditional hydrological model inversion methods rely on real-time observation data such as flow and water level, which usually have a certain time lag; mountain rivers often have large deviations in model calculation results due to large topographic relief and complex watershed changes. Summary of the Invention

[0008] The purpose of this invention is to provide a high-precision method for extracting roughness of mountain river channels based on multi-source remote sensing imagery, aiming to solve the core pain points of existing mountain river parameter extraction methods, namely "high cost, low accuracy, and time lag". Specifically, it includes: 1. reducing the cost of extracting mountain river parameters, eliminating excessive reliance on field measurements, and achieving large-scale, periodic monitoring; 2. adopting a correlation model between vegetation parameters and hydraulic parameters to improve the timeliness and inversion accuracy of roughness extraction.

[0009] To achieve the above objectives, the solution of the present invention is as follows:

[0010] A high-precision method for extracting roughness of mountain river channels based on multi-source remote sensing imagery includes:

[0011] Step S1: Acquisition and preprocessing of multi-source remote sensing image data:

[0012] S1.1 Acquisition of Multi-Source Remote Sensing Image Data:

[0013] S1.1.1 Obtain multispectral / panchromatic image data, UAV digital orthophoto (DOM) data, and auxiliary data of the study area from remote sensing satellite imagery. The auxiliary data includes: riverbed sediment particle size data and field-measured vegetation height data.

[0014] S1.1.2 Collect ground control points for image data registration and accuracy verification, including river channels, riverbank vegetation, and bare land;

[0015] S1.2 Data Preprocessing:

[0016] S1.2.1, Preprocessing of remote sensing satellite image data:

[0017] S1.2.1.1 Radiometric Correction: For imagery, the sensor calibration coefficient of satellite imagery is introduced to convert the original grayscale value output by the sensor into a radiance value;

[0018] S1.2.1.2 Atmospheric Correction: Eliminate the influence of atmospheric scattering and absorption on remote sensing images to obtain surface reflectance;

[0019] S1.2.1.3, Geometric Correction: Eliminates geometric distortions caused by satellite orbit, sensor attitude, and Earth curvature; ensures that the imagery matches the actual position on the Earth's surface in space, providing accurate spatial positioning;

[0020] S1.2.1.4 Orthorectification: Eliminates topographic distortion in images to obtain the correct spatial position in mountainous areas;

[0021] S1.2.1.5 Image Fusion: Fusion of multispectral and panchromatic images to generate high-resolution multispectral images for subsequent river channel and vegetation identification;

[0022] S1.2.1.6 Image mosaicking: The fused images are stitched together to obtain the complete study area.

[0023] S1.2.2, UAV image preprocessing:

[0024] S1.2.2.1 Geometric correction: Match the pixel coordinates of the image with the ground coordinate system based on the ground control points;

[0025] S1.2.2.2 Orthorectification: Removes distortion caused by terrain and outputs a digital orthophoto map;

[0026] S1.2.2.3 Generating DEM and DSM: The coordinate information of ground points and non-ground points is obtained through UAV overlapping image matching, stereo image pair matching and aerial triangulation measurement steps, and finally DEM and DSM are generated;

[0027] Step S2: Extraction of the river centerline and determination of the river range based on remote sensing imagery:

[0028] S2.1. Based on the field survey, visually interpret the remote sensing images and generate the initial river centerline in ArcGIS software;

[0029] S2.2. Based on the initial river centerline, set the river width and generate a buffer layer; then check the topological relationships of the main stream, tributaries and river nodes to ensure the spatial continuity of the river network, and finally output the river buffer vector file.

[0030] S2.3. Based on the river buffer vector map layer in the file, crop the remote sensing satellite imagery in ArcGIS;

[0031] S2.4. Use ArcGIS / ENVI to perform supervised classification on the cropped image, then solve the problem of misclassification of ground features, perform fragmentation processing and river integrity optimization, and extract the river range vector file;

[0032] S2.5. Generate a vector file for the centerline of the river channel based on the river channel extent vector file;

[0033] Step S3: Vegetation Cover Calculation and River Vegetation Area Identification:

[0034] S3.1 Vegetation Cover Calculation: This includes NDVI calculation and vegetation cover calculation;

[0035] The NDVI calculation is based on the fused spectral image from step S1.2.1.5 of step S1, and the Normalized Difference Vegetation Index (NDVI) is calculated using the following formula. ,generate data;

[0036] ;

[0037] Where: NIR represents the surface reflectance in the near-infrared band; Red represents the surface reflectance in the red band;

[0038] The vegetation coverage calculation is based on the generated NDVI and uses a bisection model method to calculate the river channel vegetation coverage. ;

[0039] ;

[0040] in: For the non-vegetation-covered parts of the study area Mean, ranging from 0.05 to 0.1; For the pure vegetation pixels in the study area Mean; for outliers in the calculation results that exceed 0-1, truncate them, setting values ​​less than 0 to 0;

[0041] S3.2 Identification of vegetation areas within the river channel:

[0042] set up The threshold is used to identify vegetation areas within the river channel, combined with the river channel vector file extracted in step S2; a vector file of vegetation areas within the river channel is generated for subsequent parameter calculations.

[0043] Step S4, Calculation of river channel vegetation height:

[0044] S4.1 After spatial registration in DSM and DSM, calculate the difference:

[0045] ;

[0046] In the formula: This is the initial value of vegetation height;

[0047] S4.2. Using field-measured vegetation height data, establish Linear regression model with measured height

[0048] ;

[0049] in: , These are the regression coefficients;

[0050] S4.3, to After correction, combined with the vector file of vegetation area within the river channel in step S3, the final vegetation height distribution map within the river channel is obtained, which is used for hydraulic parameter calculation.

[0051] Step S5, Calculation of hydraulic radius and water depth:

[0052] S5.1 Based on the river centerline and river range vector files generated in step S2, take cross sections every 2m along the river centerline from the river starting point, and extract the topographic profile of each cross section in combination with DEM data.

[0053] S5.2 Based on the extracted river cross-section data, calculate the hydraulic elements of each cross-section: cross-section area A, which is obtained by integrating the cross-section topographic profile; wetted perimeter X, which is the length of the water flow in contact with the riverbank within the cross-section; and water surface width B, which is the planar distance calculated from the cross-section parameters.

[0054] S5.3 Calculate the hydraulic radius of each cross section according to the hydraulic radius formula. ;

[0055] ;

[0056] S5.4. Use linear interpolation to generate a hydraulic radius distribution map of the entire river channel;

[0057] S5.5, average water depth It is expressed as the ratio of the cross-sectional area A to the width of the water surface B. ;

[0058] Step S6: Channel roughness inversion based on vegetation-hydraulic parameters:

[0059] S6.1, combined with the vegetation coverage obtained in step S3 The vegetation height obtained in step S4 The hydraulic radius obtained in step S5 , water depth The roughness is extracted using the following roughness comprehensive model based on Manning's formula:

[0060] ;

[0061] ;

[0062] ;

[0063] In the formula: The combined Chezy coefficient of vegetation and riverbed; The Chezy coefficient of the riverbed; It is the acceleration due to gravity; This represents the vegetation drag coefficient. The density of vegetation in the vertical direction, expressed as vegetation cover. replace; For water depth; The height of the vegetation; is the Kalman coefficient, with a value of 0.4; The hydraulic radius; The equivalent sand grain roughness is determined by the measured particle size data of riverbed sediment. For roughness;

[0064] S6.2 Calculate the comprehensive roughness of the river channel and generate a roughness distribution map with a spatial resolution of 2m;

[0065] S6.3. Based on on-site visual judgment, obtain the rough roughness value of the studied river section from the river roughness lookup table, and compare and verify the obtained roughness value with it.

[0066] The scheme is further described as follows: In step S1.1.1, the spatial resolution of the multispectral image / panchromatic image data is 2.6 meters / 0.65 meters, and the spatial resolution of the UAV digital orthophoto DOM data is 0.2 meters.

[0067] The scheme further includes: in step S1.2.1.5, the spatial resolution of the generated high-resolution multispectral image is 0.65 meters, and the blue, green, red, and near-infrared bands are included.

[0068] The solution further includes: in step S3.2, the identification of vegetation areas within the river channel: the setting... Threshold, as vegetation >0.1, as a non-vegetation feature <0.1, the non-vegetated features include bare rocks and buildings.

[0069] The scheme further includes: in step S6, the vegetation type is determined according to the land use type in the roughness comprehensive model, and the vegetation drag coefficient is set to 1.8 for grassland with lower height and 1.5 for shrubland and woodland with higher height.

[0070] Compared with the prior art, the present invention has the following significant advantages:

[0071] 1. Low cost and high timeliness: No large-scale field measurements are required; multi-source remote sensing images are mainly used, significantly reducing data acquisition costs; parameters of mountain rivers can be extracted, and the cycle is shortened to 1-2 weeks, meeting the needs of dynamic monitoring.

[0072] 2. Close parameter correlation and improved roughness inversion accuracy: The roughness inversion based on the coupled model of vegetation parameters and hydraulic parameters reduces the error and provides reliable parameters for high-precision hydrological simulation;

[0073] 3. Wide applicability and strong scalability: Applicable to mountainous rivers with different terrains and vegetation types (herbaceous, shrub, tree); InSAR deformation data can be added later to expand the function of monitoring changes in river scour and sedimentation height.

[0074] The invention will be further explained in detail below with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0075] Figure 1 This is a schematic diagram of the overall process of the technical solution of the present invention. Detailed Implementation

[0076] A high-precision method for extracting roughness of mountain river channels based on multi-source remote sensing imagery is proposed. This method utilizes multi-source remote sensing imagery to achieve high-precision extraction of the river channel extent, calculation of vegetation parameters, and roughness inversion in mountainous areas. This embodiment uses Gaofen-7 imagery data; see [link to Gaofen-7 imagery]. Figure 1 The high-precision extraction and hydraulic parameter inversion of mountain river channels are achieved through six key steps, the specific process of which is as follows:

[0077] Step S1: Acquisition and preprocessing of multi-source remote sensing image data:

[0078] S1.1 Acquisition of Multi-Source Remote Sensing Image Data:

[0079] S1.1.1 Obtain multispectral / panchromatic image data, UAV digital orthophoto (DOM) data, and auxiliary data of the study area from remote sensing satellite imagery. The auxiliary data includes: riverbed sediment particle size data and field-measured vegetation height data.

[0080] Wherein: the spatial resolution of the multispectral / panchromatic image data is 2.6 meters / 0.65 meters, and the spatial resolution of the UAV digital orthophoto DOM data is 0.2 meters;

[0081] S1.1.2 Collect ground control points (GCPs) for image data registration and accuracy verification, including river channels, riverbank vegetation, and bare land;

[0082] S1.2 Data Preprocessing:

[0083] S1.2.1, Preprocessing of remote sensing satellite image data:

[0084] S1.2.1.1 Radiometric Correction: For imagery, the sensor calibration coefficient of satellite imagery is introduced to convert the original grayscale value output by the sensor into a radiance value;

[0085] S1.2.1.2 Atmospheric Correction: Eliminate the influence of atmospheric scattering and absorption on remote sensing images to obtain surface reflectance;

[0086] S1.2.1.3, Geometric Correction: Eliminates geometric distortions caused by satellite orbit, sensor attitude, and Earth curvature; ensures that the imagery matches the actual position on the Earth's surface in space, providing accurate spatial positioning;

[0087] S1.2.1.4 Orthorectification: Eliminates topographic distortion in images to obtain the correct spatial position in mountainous areas;

[0088] S1.2.1.5 Image Fusion: Fusion of multispectral and panchromatic images to generate high-resolution multispectral images for subsequent river channel and vegetation identification;

[0089] Wherein: the spatial resolution of the generated high-resolution multispectral image is 0.65 meters, and the blue, green, red, and near-infrared bands are included;

[0090] S1.2.1.6. Image mosaicking: stitching together the fused images to obtain the complete study area;

[0091] S1.2.2, UAV image preprocessing:

[0092] S1.2.2.1 Geometric correction: Match the pixel coordinates of the image with the ground coordinate system based on the ground control points;

[0093] S1.2.2.2 Orthorectification: Removes distortion caused by terrain and outputs a digital orthophoto map;

[0094] S1.2.2.3 Generating DEM and DSM: The coordinate information of ground points and non-ground points is obtained through UAV overlapping image matching, stereo image pair matching and aerial triangulation measurement steps, and finally DEM and DSM are generated;

[0095] Step S2: Extraction of the river centerline and determination of the river range based on remote sensing imagery:

[0096] S2.1. Based on the field survey, visually interpret the remote sensing images and generate the initial river centerline in ArcGIS software;

[0097] S2.2. Based on the initial river centerline, set the river width and generate a buffer layer; then check the topological relationships of the main stream, tributaries and river nodes to ensure the spatial continuity of the river network, and finally output the river buffer vector file.

[0098] S2.3. Based on the river buffer vector map layer in the file, crop the remote sensing satellite imagery in ArcGIS;

[0099] S2.4. Use ArcGIS / ENVI to perform supervised classification on the cropped image, then solve the problem of misclassification of ground features, perform fragmentation processing and river integrity optimization, and extract the river range vector file;

[0100] S2.5. Generate a vector file for the centerline of the river channel based on the river channel extent vector file;

[0101] Step S3: Vegetation Cover Calculation and River Vegetation Area Identification:

[0102] S3.1 Vegetation Cover Calculation: This includes NDVI calculation and vegetation cover calculation;

[0103] The NDVI calculation is based on the fused spectral image from step S1.2.1.5 of step S1, and the Normalized Difference Vegetation Index (NDVI) is calculated using the following formula. ,generate data;

[0104] ;

[0105] Where: NIR represents the surface reflectance in the near-infrared band; Red represents the surface reflectance in the red band;

[0106] The vegetation coverage calculation is based on the generated NDVI and uses a bisection model method to calculate the river channel vegetation coverage. ;

[0107] ;

[0108] in: For the non-vegetation-covered parts of the study area Mean, ranging from 0.05 to 0.1; For pixels with pure vegetation (such as dense forests and well-grown farmland) in the study area Mean; for outliers in the calculation results that exceed 0-1, truncate them, setting values ​​less than 0 to 0;

[0109] S3.2 Identification of vegetation areas within the river channel:

[0110] set up The threshold is used to identify vegetation areas within the river channel, combined with the river channel vector file extracted in step S2; a vector file of vegetation areas within the river channel is generated for subsequent parameter calculations.

[0111] Wherein: the setting Threshold, as vegetation >0.1, as a non-vegetation feature <0.1, the non-vegetated features include bare rocks and buildings;

[0112] Step S4, Calculation of river channel vegetation height:

[0113] S4.1 After spatial registration in DSM and DSM, calculate the difference:

[0114] ;

[0115] In the formula: This is the initial value of vegetation height;

[0116] S4.2. Using field-measured vegetation height data, establish Linear regression model with measured height

[0117]

[0118] in: , These are the regression coefficients;

[0119] S4.3, to After correction, combined with the vector file of vegetation area within the river channel in step S3, the final vegetation height distribution map within the river channel is obtained, which is used for hydraulic parameter calculation.

[0120] Step S5, Calculation of hydraulic radius and water depth:

[0121] S5.1 Based on the river centerline and river range vector files generated in step S2, take cross sections every 2m along the river centerline from the river starting point, and extract the topographic profile of each cross section in combination with DEM data.

[0122] S5.2 Based on the extracted river cross-section data, calculate the hydraulic elements of each cross-section: cross-section area A, which is obtained by integrating the cross-section topographic profile; wetted perimeter X, which is the length of the water flow in contact with the riverbank within the cross-section; and water surface width B, which is the planar distance calculated from the cross-section parameters.

[0123] S5.3 Calculate the hydraulic radius of each cross section according to the hydraulic radius formula. ;

[0124] ;

[0125] S5.4. Use linear interpolation to generate a hydraulic radius distribution map of the entire river channel;

[0126] S5.5, average water depth It can be approximated or directly expressed as the ratio of cross-sectional area A to water surface width B;

[0127] ;

[0128] Step S6: Channel roughness inversion based on vegetation-hydraulic parameters:

[0129] S6.1, combined with the vegetation coverage obtained in step S3 The vegetation height obtained in step S4 The hydraulic radius obtained in step S5 , water depth The roughness is extracted using the following roughness comprehensive model based on Manning's formula:

[0130] ;

[0131] ;

[0132] ;

[0133] In the formula: The combined Chezy coefficient of vegetation and riverbed; The Chezy coefficient of the riverbed; It is the acceleration due to gravity; This represents the vegetation drag coefficient. The density of vegetation in the vertical direction, expressed as vegetation cover. replace; For water depth; The height of the vegetation; is the Kalman coefficient, with a value of 0.4; The hydraulic radius; The equivalent sand grain roughness is determined by the measured particle size data of riverbed sediment. For roughness;

[0134] S6.2 Calculate the comprehensive roughness of the river channel and generate a roughness distribution map with a spatial resolution of 2m;

[0135] S6.3. Based on on-site visual judgment, obtain the rough roughness value of the studied river section from the river roughness lookup table, and compare and verify the obtained roughness value with it.

[0136] In the roughness comprehensive model, vegetation type is determined according to land use type. The vegetation drag coefficient is set to 1.8 for grassland with lower height and 1.5 for shrubland and woodland with higher height.

[0137] The core technical points of this embodiment include:

[0138] 1. Image-based collaborative river channel extraction method: River channel extraction technology based on river centerline and pixel classification achieves high-precision extraction of river channel extent under complex terrain;

[0139] 2. Vegetation cover and vegetation height estimation techniques coupled with field measurements: vegetation cover is estimated using a bisection model and vegetation height is calculated using linear regression, combined with field measurement data to ensure inversion accuracy;

[0140] 3. Roughness inversion model coupled with vegetation and hydraulic parameters: A roughness comprehensive model based on Manning's formula is adopted to associate vegetation height, coverage and other factors with hydraulic parameters, which breaks through the limitations of traditional empirical formulas and improves the accuracy of roughness inversion.

[0141] Figure 1 Starting with "Data Acquisition," the course sequentially presents six core steps: "Acquisition and Preprocessing of Multi-Source Remote Sensing Image Data," "Extraction of River Centerline and Determination of River Range Based on Remote Sensing Imagery," "Calculation of Vegetation Coverage and Identification of River Vegetation Areas," "Calculation of River Vegetation Height," "Calculation of Hydraulic Radius and Water Depth," and "Inversion of River Roughness Based on Vegetation-Hydraulic Parameters." The steps are connected by arrows, clearly demonstrating the technical logic.

Claims

1. A high-precision method for extracting roughness of mountain river channels based on multi-source remote sensing imagery, characterized in that, The high-precision method for extracting roughness of mountain river channels includes: Step S1: Acquisition and preprocessing of multi-source remote sensing image data: S1.1 Acquisition of multi-source remote sensing image data: S1.1.1 Obtain multispectral / panchromatic image data, UAV digital orthophoto (DOM) data, and auxiliary data of the study area from remote sensing satellite imagery. The auxiliary data includes: riverbed sediment particle size data and field-measured vegetation height data. S1.1.2 Collect ground control points for image data registration and accuracy verification, including river channels, riverbank vegetation, and bare land; S1.2 Data Preprocessing: S1.2.1, Preprocessing of remote sensing satellite image data: S1.2.1.1 Radiometric Correction: For imagery, the sensor calibration coefficient of satellite imagery is introduced to convert the original grayscale value output by the sensor into a radiometric value; S1.2.1.2 Atmospheric Correction: Eliminate the influence of atmospheric scattering and absorption on remote sensing images to obtain surface reflectance; S1.2.1.3, Geometric Correction: Eliminates geometric distortions caused by satellite orbit, sensor attitude, and Earth curvature; ensures that the imagery matches the actual position on the Earth's surface in space, providing accurate spatial positioning; S1.2.1.4 Orthorectification: Eliminates topographic distortion in images to obtain the correct spatial position in mountainous areas; S1.2.1.5 Image Fusion: Fusion of multispectral and panchromatic images to generate high-resolution multispectral images for subsequent river and vegetation identification; S1.2.1.6 Image mosaicking: The fused images are stitched together to obtain the complete study area. S1.2.2, UAV image preprocessing: S1.2.2.1 Geometric correction: Match the pixel coordinates of the image with the ground coordinate system based on the ground control points; S1.2.2.2 Orthorectification: Removes distortion caused by terrain and outputs a digital orthophoto map; S1.2.2.3 Generating DEM and DSM: The coordinate information of ground points and non-ground points is obtained through UAV overlapping image matching, stereo image pair matching and aerial triangulation measurement steps, and finally DEM and DSM are generated; Step S2: Extraction of the river centerline and determination of the river range based on remote sensing imagery: Step S3: Vegetation Cover Calculation and River Vegetation Area Identification: S3.1 Vegetation Cover Calculation: This includes NDVI calculation and vegetation cover calculation; The NDVI calculation is based on the fused spectral image from step S1.2.1.5 of step S1, and the Normalized Difference Vegetation Index (NDVI) is calculated using the following formula. ,generate data; ； Where: NIR represents the surface reflectance in the near-infrared band; Red represents the surface reflectance in the red band; The vegetation coverage calculation is based on the generated NDVI and uses a bisection model method to calculate the river channel vegetation coverage. ; ； in: For the non-vegetation-covered parts of the study area Mean, ranging from 0.05 to 0.1; For the pure vegetation pixels in the study area Mean; for outliers in the calculation results that exceed 0-1, truncate them, setting values ​​less than 0 to 0; S3.2 Identification of vegetation areas within the river channel: set up The threshold is used to identify vegetation areas within the river channel, combined with the river channel vector file extracted in step S2; a vector file of vegetation areas within the river channel is generated for subsequent parameter calculations. Step S4, Calculation of river channel vegetation height: S4.1 After spatial registration in DSM and DSM, calculate the difference: ； In the formula: This is the initial value of vegetation height; S4.

2. Using field-measured vegetation height data, establish Linear regression model with measured height ； in: , These are the regression coefficients; S4.3, to After correction, combined with the vector file of vegetation area within the river channel in step S3, the final vegetation height distribution map within the river channel is obtained, which is used for hydraulic parameter calculation. Step S5, Calculation of hydraulic radius and water depth: S5.1 Based on the river centerline and river range vector files generated in step S2, take cross sections every 2m along the river centerline from the river starting point, and extract the topographic profile of each cross section in combination with DEM data. S5.2 Based on the extracted river cross-section data, calculate the hydraulic elements of each cross-section: cross-section area A, which is obtained by integrating the cross-section topographic profile; wetted perimeter X, which is the length of the water flow in contact with the riverbank within the cross-section; and water surface width B, which is the planar distance calculated from the cross-section parameters. S5.3 Calculate the hydraulic radius of each cross section according to the hydraulic radius formula. ; ； S5.

4. Use linear interpolation to generate a hydraulic radius distribution map of the entire river channel; S5.5, average water depth It is expressed as the ratio of the cross-sectional area A to the width of the water surface B. ; Step S6: Channel roughness inversion based on vegetation-hydraulic parameters: S6.1, combined with the vegetation coverage obtained in step S3 The vegetation height obtained in step S4 The hydraulic radius obtained in step S5 , water depth The roughness is extracted using the following roughness comprehensive model based on Manning's formula: ； ； ； In the formula: The combined Chezy coefficient of vegetation and riverbed; The Chezy coefficient of the riverbed; It is the acceleration due to gravity; This represents the vegetation drag coefficient. The density of vegetation in the vertical direction, expressed as vegetation cover. replace; For water depth; The height of the vegetation; is the Kalman coefficient, with a value of 0.4; The hydraulic radius; The equivalent sand grain roughness is determined by the measured particle size data of riverbed sediment. For roughness; S6.2 Calculate the comprehensive roughness of the river channel and generate a roughness distribution map with a spatial resolution of 2m; S6.

3. Based on on-site visual judgment, obtain the rough roughness value of the studied river section from the river roughness lookup table, and compare and verify the obtained roughness value with it.

2. The high-precision method for extracting roughness of mountain river channels according to claim 1, characterized in that, In step S1.1.1, the spatial resolution of the multispectral / panchromatic image data is 2.6 meters / 0.65 meters, and the spatial resolution of the UAV digital orthophoto DOM data is 0.2 meters.

3. The high-precision method for extracting roughness of mountain river channels according to claim 1, characterized in that, In step S1.2.1.5, the spatial resolution of the generated high-resolution multispectral image is 0.65 meters, in the blue, green, red, and near-infrared bands.

4. The high-precision method for extracting roughness of mountain river channels according to claim 1, characterized in that, Step S2, which involves extracting the river centerline and determining the river extent based on remote sensing imagery, includes: S2.

1. Based on the field survey, visually interpret the remote sensing images and generate the initial river centerline in ArcGIS software; S2.

2. Based on the initial river centerline, set the river width and generate a buffer layer; then check the topological relationships of the main stream, tributaries and river nodes to ensure the spatial continuity of the river network, and finally output the river buffer vector file. S2.

3. Based on the river buffer vector map layer in the file, crop the remote sensing satellite imagery in ArcGIS; S2.

4. Use ArcGIS / ENVI to perform supervised classification on the cropped image, then solve the problem of misclassification of ground features, perform fragmentation processing and river integrity optimization, and extract the river range vector file; S2.

5. Generate a vector file of the river centerline based on the river range vector file.

5. The high-precision method for extracting roughness of mountain river channels according to claim 1, characterized in that, In step S3.2, the identification of vegetation areas within the river channel: the setting Threshold, as vegetation >0.1, as a non-vegetation feature <0.1, the non-vegetated features include bare rocks and buildings.

6. The high-precision method for extracting roughness of mountain river channels according to claim 1, characterized in that, In step S6, the vegetation type is determined according to the land use type in the roughness comprehensive model. The vegetation drag coefficient is set to 1.8 for grassland with lower height and 1.5 for shrubland and woodland with higher height.

Images