Method for automatically measuring water flow velocity of complex mountainous river channel

Abstract

This invention discloses an automated method for measuring flow velocity in complex mountainous rivers, comprising: acquiring river topographic data and establishing a spatial measurement reference surface; acquiring the real-time water surface distance and determining the real-time water depth based on the spatial measurement reference surface; acquiring vertical velocity data at measuring points within the real-time water depth range, determining whether to add new measuring points based on the differences in velocity data at adjacent measuring points, and obtaining a vertical adaptive densification sequence; determining the depth-average velocity and its first uncertainty based on the sequence; adjusting the trigger threshold along the transverse section using the first uncertainty to determine whether to add new vertical lines, and obtaining a transverse adaptive densification sequence; determining the cross-sectional average velocity and its second uncertainty based on the transverse sequence; determining whether to add new cross sections along the longitudinal direction, and outputting three-dimensional velocity field data. This invention achieves dynamic optimization of the spatial measuring point layout, significantly improving the noise resistance and spatial representativeness of complex flow field detection.

Description

Technical Field

[0001] This invention relates to contact velocity measurement technology for hydraulic engineering physical models of water conservancy projects, and more particularly to an automated method for measuring the flow velocity of water in complex mountainous rivers. Background Technology

[0002] Physical model testing is a core research method for reproducing complex hydraulic phenomena in real hydraulic engineering projects and for engineering design and evaluation. In mountainous river models, flow velocity, as the most fundamental dynamic parameter characterizing water flow, directly determines the accuracy of designs for flood discharge from weirs and dams, the layout of diversion structures, and scour protection. Obtaining high-fidelity, highly representative three-dimensional velocity field data is of irreplaceable technical value for revealing the evolution mechanism of highly complex, non-steady three-dimensional flow regimes.

[0003] Current methods for acquiring flow velocities in hydraulic physics models often rely on statically deploying contact current meters on pre-defined fixed grid sections or finite vertical lines. These methods typically involve point-by-point mechanical scanning with a fixed spatial step size. When dealing with complex boundary conditions, due to the strong nonlinear spatial heterogeneity of the flow field, the fixed-step sampling mode easily generates a large amount of redundant data in calm flow regions, while in areas with dramatic flow changes such as deep pools and reef wakes, it often misses crucial local velocity variation characteristics. Furthermore, contact probes inevitably introduce significant pulsation variance in strongly turbulent flow, leading to objective data fluctuations in single-point measurements.

[0004] In summary, existing contact-based velocity acquisition methods struggle to balance high spatial representativeness and global detection efficiency when dealing with complex flow fields exhibiting strong three-dimensional heterogeneity and severe turbulence. Furthermore, uncontrollable data fluctuations in the underlying hardware can easily cause systematic interference to the reconstruction of the overall flow field morphology. Therefore, it is necessary to investigate a method that can improve the adaptive capability of data acquisition and the overall robustness of the system under complex and variable flow conditions. Summary of the Invention

[0005] Purpose of the invention: To provide an automated method for measuring the flow velocity of water in complex mountainous rivers, in order to solve the aforementioned problems existing in the prior art.

[0006] Technical solution: An automated method for measuring the flow velocity of rivers in complex mountainous areas, comprising:

[0007] Acquire river topographic data and establish a spatial measurement reference surface based on the river topographic data;

[0008] Obtain the real-time water surface distance to the water surface, and determine the real-time water depth based on the real-time water surface distance and the spatial measurement reference plane;

[0009] Within the range of the real-time water depth, obtain the flow velocity data of vertical measuring points, determine whether to encrypt new measuring points based on the difference in flow velocity data of adjacent measuring points, and obtain a vertical adaptive encryption sequence;

[0010] The depth-average flow velocity and its first uncertainty for the corresponding vertical line are determined based on the vertical adaptive encryption sequence.

[0011] The depth-average flow velocity and its first uncertainty of adjacent vertical lines are obtained along the transverse cross section. The trigger threshold of transverse encryption is adjusted using the first uncertainty to determine whether to encrypt a new vertical line, thus obtaining a transverse adaptive encryption sequence.

[0012] The cross-sectional average flow velocity and its second uncertainty are determined based on the aforementioned transverse adaptive encryption sequence.

[0013] Along the longitudinal channel, based on the average flow velocity of the adjacent cross sections and its second uncertainty, it is determined whether to densify a new cross section, thus obtaining a longitudinal adaptive densification sequence;

[0014] The three-dimensional flow field data is output based on the acquired adaptive encrypted sequences of each dimension.

[0015] Beneficial effects: To address the problem that fixed step size sampling is difficult to balance the efficiency of global detection with the representativeness of local features, this invention adopts a spatial multi-level adaptive encryption mechanism based on interpolation error estimation. It transforms the traditional first-order gradient criterion into second-order curvature tracking, enabling the spatial measurement point layout to accurately lock the flow regime change boundary such as the backflow separation zone. At the same time, it maintains sparse sampling in the uniform shear flow region, effectively avoiding data redundancy caused by invalid detection.

[0016] To address the issue of local velocity misjudgment easily caused by strong turbulent flow impulse noise, this invention introduces a dual-threshold decision mechanism that integrates measurement variance and statistical significance testing. This mechanism completely separates the true flow field gradient from random turbulent fluctuation artifacts, significantly reducing the probability of invalid mesh subdivision driven by environmental white noise. Simultaneously, to address the macroscopic decision distortion caused by the upward propagation of lower-level measurement errors, this invention constructs a cross-level uncertainty propagation feedback mechanism. This mechanism aggregates the lower-level impulse noise and interpolation residuals step by step, dynamically adjusting the encryption trigger threshold of the upper-level space. This makes the decision sensitivity in the spatial dimension adaptively constrained by the physical reliability of the lower-level data, achieving system-level closed-loop fault tolerance.

[0017] Furthermore, to address the issue of coordinate accuracy being constrained by complex media and mechanical deformation, this invention effectively eliminates translational, tilting, and torsional distortions of the orbital system, as well as acoustic attenuation interference at the air-water interface, through spatial trend surface fitting correction under anhydrous conditions and a two-step wet-dry method for water depth measurement based on temperature compensation. This reduces the absolute error of global spatial positioning and dynamic water depth tracking from the centimeter level to the millimeter level, significantly improving the objectivity and realism of the flow field reconstruction. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating an automated method for measuring the flow velocity of water in complex mountainous rivers, as provided in an embodiment of this application.

[0019] Figure 2 This is a schematic diagram of the process for calculating the real-time water depth provided in the embodiments of this application.

[0020] Figure 3 This is a schematic diagram of the process for establishing a spatial measurement reference surface based on the river topographic data provided in this application embodiment.

[0021] Figure 4 This is a schematic diagram of the process for determining whether to encrypt a new measuring point based on the difference in flow velocity data between adjacent measuring points, provided in an embodiment of this application. Detailed Implementation

[0022] Example 1: An automated method for measuring the flow velocity of rivers in complex mountainous areas, mainly including the following steps:

[0023] Step 101: Obtain river topographic data and establish a spatial measurement reference surface based on the river topographic data.

[0024] In this embodiment, the acquired river topographic data typically originates from the design parameters of a hydraulic engineering physical model of a water conservancy project or from preliminary survey data. Specifically, the physical model is a simulation device constructed in a laboratory at a scaled-down ratio of a real water conservancy project, based on similarity theory. The river topographic data includes digital information such as the underwater topography, cross-sectional morphology, and boundary features of the model. The cross-sectional morphology includes a cross section perpendicular to the water flow direction and a longitudinal section along the water flow direction. Key topographic data covers spatial geometric parameters such as elevation, river width, and curvature.

[0025] To achieve automated measurement of flow velocity across the entire watershed, this embodiment constructs a measurement hardware architecture based on a three-axis platform system. This platform specifically includes a horizontal axis, a vertical axis, and a telescopic rod carrying the measurement equipment. To ensure the accuracy of spatial measurements, the horizontal axis must be perpendicular to the river's central axis during installation, and the horizontal and vertical axes must be located on the same horizontal plane, with the horizontal height difference between their two ends strictly controlled within 1 millimeter. Establishing a spatial measurement reference surface based on the riverbed topographic data provides a unified coordinate reference system for subsequent dynamic measurements, ensuring precise alignment between the spatial positioning of the flow velocity sensor and the actual terrain of the physical model.

[0026] Step 102: Obtain the real-time water surface distance (to the water surface), and determine the real-time water depth based on the real-time water surface distance and the spatial measurement reference plane.

[0027] This step is used to obtain the hydrodynamic boundary conditions on the measurement vertical line in a wet environment where the physical model is in operation with water. Specifically, the system uses a ranging sensor mounted above the measurement platform to emit a detection signal downwards, obtaining the absolute distance from the sensor's transmitter to the physical interface of the water surface, i.e., the real-time water surface distance. Since a spatial measurement reference surface containing riverbed elevation information has been established in previous steps, the river channel water depth at this spatial point can be obtained in real time by geometrically calculating the elevation of the measurement reference surface at the current sensor's coordinates and the real-time water surface distance. This water depth data provides physical constraints for the subsequent vertical movement boundary of the flow velocity sensor; for example, the system can calculate the effective detection distance of the flow velocity sensor underwater based on this real-time water depth.

[0028] Step 103: Obtain flow velocity data of vertical measuring points within the range of the real-time water depth, determine whether to encrypt new measuring points based on the difference in flow velocity data of adjacent measuring points, and obtain a vertical adaptive encryption sequence.

[0029] After acquiring real-time water depth, the propeller velocity sensor mounted at the bottom of the telescopic mast will be driven by a stepper motor to enter the water body. Initially, the system can deploy initial measuring points at preset uniform intervals within an effective water depth range of 0.02 meters below the water surface to 0.02 meters above the riverbed, for example, arranging 5 initial measuring points at equal intervals. The system will stop at each initial measuring point and collect flow velocity data.

[0030] After acquiring flow velocity data, the system performs a difference analysis on the flow velocity distribution characteristics of adjacent measuring points. In mountainous rivers, influenced by complex riverbed boundaries, the water flow often exhibits strong three-dimensional turbulent characteristics, and the velocity distribution along the water depth is usually nonlinear. If the system determines that the velocity difference between two adjacent measuring points meets the preset nonlinear abrupt change characteristics, it considers that there may be a backflow separation point or shear force inflection point in that area. At this time, the system automatically generates an encryption command, driving a stepper motor to transport the flow velocity sensor to the depth position between these two adjacent measuring points for supplementary measurement. This evaluation and encryption insertion process iterates vertically until the difference between adjacent measuring points no longer meets the triggering condition or reaches the physical constraint boundary, ultimately forming a set of measuring points unevenly distributed along the vertical line, i.e., the vertical adaptive encryption sequence.

[0031] Step 104: Determine the depth-average flow velocity and its first uncertainty for the corresponding vertical line based on the vertical adaptive encryption sequence.

[0032] After completing the adaptive measurement of a single vertical line, the system uses the velocity data and depth coordinates of all measuring points on the vertical line to calculate the cross-sectional average velocity of the vertical line, i.e., the depth-average velocity, using conventional numerical integration methods.

[0033] Furthermore, this invention not only outputs the mean value but also simultaneously quantifies the reliability of this mean value. The first uncertainty is used to characterize the error level of the calculated depth-average flow velocity. Due to the presence of turbulent fluctuations in the physical model and the interpolation residual error in the velocity distribution between discrete measuring points, the system will deduce the first uncertainty based on the sampling variance of each measuring point and the spatial span parameter between adjacent measuring points. The larger the value of this first uncertainty, the lower the measurement quality of the vertical line or the more severe the water flow turbulence, and its influence weight on subsequent cross-level decisions will change accordingly.

[0034] Step 105: Obtain the depth-average flow velocity and its first uncertainty along the transverse cross-section. Use the first uncertainty to adjust the trigger threshold for transverse encryption to determine whether to encrypt a new vertical line, thus obtaining a transverse adaptive encryption sequence. In some embodiments, after performing the above vertical measurement and encryption on each vertical line sequentially along the transverse cross-section, based on the depth-average flow velocity and its first uncertainty of the adjacent vertical lines, use the first uncertainty to adjust the trigger threshold for transverse encryption to determine whether to encrypt a new vertical line, thus obtaining a transverse adaptive encryption sequence;

[0035] After completing the measurement of one vertical line, the horizontal axis stepper motor translates the measuring body to the next initial vertical position of that cross section. After obtaining the depth-averaged flow velocity of two adjacent vertical lines, the system determines their lateral differences.

[0036] In this embodiment, the system incorporates the first uncertainty passed from the lower layer into the lateral decision-making logic. Specifically, the trigger threshold for determining whether the lateral differences are significant is dynamically adjusted using the first uncertainty of adjacent vertical lines. If the first uncertainty of adjacent vertical lines is generally high, the system will adaptively increase the trigger threshold for lateral encryption, making it more conservative in its judgment, thereby avoiding misjudging the turbulent noise at the lower level as a real abrupt change in the lateral flow field. Conversely, if the first uncertainty is small, the trigger threshold is lowered, and the system can sensitively capture the lateral velocity gradient and issue encryption commands, inserting new measurement vertical lines between adjacent vertical lines. Through this adaptive judgment and insertion process, a well-spaced vertical line distribution is finally formed on the cross-section, i.e., the lateral adaptive encryption sequence.

[0037] Step 106: Determine the cross-sectional average flow velocity and its second uncertainty based on the transverse adaptive encryption sequence.

[0038] After completing the adaptive traversal of the entire cross section, the system integrates the depth-average flow velocity and spatial position of all vertical lines in the cross section to calculate the overall flow velocity index representing the cross section, namely the cross section average flow velocity.

[0039] Similarly, based on the first uncertainty of each vertical line and the interpolation residuals between adjacent horizontal vertical lines, the system further derives a composite error index characterizing the reliability of the average flow velocity of the cross section, namely the second uncertainty. This uncertainty achieves cross-dimensional error accumulation and quantification from "point -> line -> surface".

[0040] Step 107: Determine whether to densify a new cross section based on the average flow velocity of adjacent cross sections and its second uncertainty along the longitudinal channel, thus obtaining a longitudinal adaptive densification sequence. In some embodiments, after performing the above-described lateral measurement and densification on each cross section sequentially along the longitudinal channel, determine whether to densify a new cross section based on the average flow velocity of adjacent cross sections and its second uncertainty, thus obtaining a longitudinal adaptive densification sequence.

[0041] After completing the measurement of the current cross-section, the longitudinal axis stepper motor transports the entire transverse axis platform along the river channel to the next initial cross-section downstream. The system compares the cross-sectional average flow velocity of adjacent cross-sections and combines it with the transmitted second uncertainty. At river bends or abrupt topographic changes, when the longitudinal flow velocity changes significantly and the second uncertainty is within the confidence interval, the system determines to insert a new cross-section between two adjacent cross-sections for scanning, thereby generating the longitudinally adaptive encryption sequence distributed along the river channel.

[0042] The decision logic for longitudinal densification is similar to that for the vertical and horizontal dimensions. The system calculates the longitudinal midpoint deviation based on the average flow velocity of three adjacent cross-sections using the linear interpolation deviation method, and synthesizes the uncertainty of the longitudinal midpoint deviation based on the second uncertainty and the error propagation law. The system simultaneously tests both the actual significance condition and the statistical significance condition; only when both threshold conditions are simultaneously met does it determine whether to densify a new cross-section between two adjacent cross-sections.

[0043] Step 108: Output three-dimensional flow field data based on the acquired adaptive encrypted sequences of each dimension.

[0044] Through the aforementioned layer-by-layer adaptive encryption and uncertainty control transmission across the vertical, horizontal, and longitudinal dimensions, the system ultimately acquires a discrete set of measuring points covering the entire hydraulic physics model. The system packages the three-dimensional spatial coordinates of these non-uniformly distributed measuring points with their corresponding time-averaged flow velocity values, outputting high-fidelity three-dimensional velocity field data. This data accurately presents the complex hydraulic phenomena in deep pools, shallow shoals, and the wake regions of obstacles, providing reliable data support for engineering scheme evaluation.

[0045] Example 2: Calibrate the waterless terrain and correct the spatial trend surface. That is, before performing wet hydrodynamic measurements, construct a high-precision spatial measurement reference surface through multi-point field measurements and calculations under waterless conditions.

[0046] Step 201: Obtain the measured elevations of the points distributed at the preset calibration points under anhydrous dry conditions.

[0047] Before the specific measurement task is initiated, the system controls the measurement track platform to operate in a dry environment without water. The system schedules the measurement subject to move to multiple preset calibration points located at different positions on the river physical model. At each preset calibration point, ultrasonic sensors emit detection signals downwards and receive reflected echoes to obtain the physical distance from the sensor to the riverbed surface. This distance is then combined with the sensor's own installation elevation to calculate the measured riverbed elevation data at that spatial coordinate. In some implementations, the preset calibration points can be selected using a regular gridded distribution strategy, or they can be locally densified in characteristic areas with drastic topographic curvature changes, such as river bends and deep pools, to ensure that the sampling point set can comprehensively reflect the bottom geometry of the entire model area.

[0048] Step 202: Extract the corresponding design elevation from the river topographic data and calculate the elevation deviation between the measured elevation and the design elevation.

[0049] The system reads pre-stored river topographic data from a hydraulic physics model and indexes the design elevation data for the corresponding locations based on the plane coordinates of preset calibration points. This design elevation data is typically a theoretical benchmark value rigorously calculated based on the actual prototype topography and the model's scale. Subsequently, the system performs deviation calculations at each calibration point to extract the spatial difference between the actual physical environment and the theoretical design environment.

[0050] e_i = z_meas(x_i, y_i) - z_design(x_i, y_i);

[0051] Where e_i is the elevation deviation at the i-th preset calibration point, z_meas(x_i, y_i) is the measured elevation obtained by the sensor at the plane coordinates (x_i, y_i), z_design(x_i, y_i) is the design elevation corresponding to the coordinates (x_i, y_i) in the river topographic data, x_i is the longitudinal coordinate of the i-th preset calibration point, and y_i is the lateral coordinate of the i-th preset calibration point.

[0052] Step 203: Fit a spatial trend surface based on the elevation deviation of each calibration point. The spatial trend surface characterizes the combined translational, tilting and torsional deviations of the measuring equipment.

[0053] Specifically, fitting a spatial trend surface based on the elevation deviation of each of the calibration points includes:

[0054] The spatial distribution of the elevation deviation is fitted using a bilinear trend surface to obtain the bilinear trend surface:

[0055] The error coefficients of the bilinear trend surface are determined by the least squares method. Each error coefficient corresponds to the overall translational deviation coefficient, longitudinal and lateral tilt deviation coefficients, and torsional deviation coefficient characterizing the plane twisting feature of the mapping measurement track platform.

[0056] Due to the unavoidable installation deviation between the track platform and the hydraulic model, this deviation typically exhibits a continuous spatial distribution. The system utilizes the elevation deviations of all preset calibration points to construct a global or local error distribution surface. Specifically, in this embodiment, the system uses a bilinear trend surface to fit the spatial distribution of the elevation deviation. By introducing the least squares algorithm to perform regression analysis on the deviation data of all calibration points, the error coefficients of the bilinear trend surface are solved and determined. The specific spatial trend surface fitting formula is as follows:

[0057] e(x, y) = a_0 + a_1 * x + a_2 * y + a_3 * x * y;

[0058] Where e(x, y) is the spatial trend surface estimation deviation calculated at any planar coordinate (x, y), a_0 is the overall translational deviation coefficient, a_1 is the longitudinal tilt deviation coefficient, a_2 is the lateral tilt deviation coefficient, and a_3 is the torsional deviation coefficient characterizing the planar twisting characteristics. x is the longitudinal coordinate of the target point, and y is the lateral coordinate of the target point. In this formula, each error coefficient corresponds to the physical deformation characteristics of the mapping measurement track platform. For example, a_0 reflects the fixed offset of the overall track height, a_1 and a_2 reflect the installation slope of the guide rail in the longitudinal and lateral directions, while a_3 characterizes the planar torsional characteristics of the track caused by uneven settlement of the support structure.

[0059] In some optional implementations, after solving the bilinear trend surface coefficients, the system can also perform anomaly detection. By calculating the absolute value of the residuals at each preset calibration point, if the absolute value of the residual at a calibration point is determined to be greater than 2 mm, the system marks that coordinate point as an anomaly and outputs a prompt message for manual review. This 2 mm threshold is usually set as the upper limit of the allowable error under the hydraulic model manufacturing process. By eliminating gross errors exceeding this upper limit, the robustness of the spatial trend surface fitting can be further improved.

[0060] Step 204: The spatial trend surface is used to compensate for errors in the measured terrain data, generating the spatial measurement reference surface. Specifically, the spatial trend surface is used to compensate for errors in the measured elevations obtained at each measurement location, generating the spatial measurement reference surface.

[0061] After obtaining the bilinear trend surface model containing various error coefficients, the system applies it to all subsequent measurement locations. For any measurement node within the river channel model, the system uses a pre-set compensation algorithm to convert the measured physical elevation into an absolute elevation aligned to a unified design coordinate system, completing the calibration phase. The specific error compensation calculation formula is as follows:

[0062] z_corrected(x, y) = z_meas(x, y) - e(x, y);

[0063] Wherein, z_corrected(x, y) is the spatial measurement reference surface elevation generated at coordinates (x, y) after error compensation, z_meas(x, y) is the original measured elevation acquired by the sensor, and e(x, y) is the estimated deviation calculated by substituting coordinates (x, y) into the aforementioned spatial trend surface equation. After global gridding error compensation processing, the system finally generates a smooth and continuous spatial measurement reference surface database. This reference surface not only eliminates the mechanical errors of the track system but also fully preserves the local micro-topographic features of the actual riverbed bottom, providing solid coordinate data support for accurate calculation of water depth and location of flow velocity sensors under subsequent water-filled conditions.

[0064] Example 3 describes the process of obtaining water depth boundary conditions at the gas-water dual-medium interface in a hydraulic physics model, including a preferred implementation based on dry and wet two-step measurement and an alternative implementation based on single-pulse dual-echo extraction.

[0065] Step 301, obtaining the real-time water surface distance and determining the real-time water depth based on the real-time water surface distance and the spatial measurement reference plane, specifically involves emitting a sound wave pulse and receiving the reflected echo from the corresponding interface, extracting the corresponding distance data based on the flight time of the reflected echo, and calculating the real-time water depth by combining it with the spatial measurement reference plane.

[0066] In this embodiment, acoustic ranging technology is applied to a non-contact distance detection process. A sensor mounted on a measuring track platform emits acoustic pulses of a specific frequency towards the physical interface below, simultaneously activating an internal high-precision timing unit. When the acoustic pulse encounters the interface between air and water, or the interface between water and the riverbed, due to the difference in acoustic impedance between the different media, some or all of the acoustic energy will be reflected, forming an echo. The system records the total time from transmission to reception by capturing the reflected echo returning to the sensor, i.e., the flight time. Furthermore, the system combines the propagation speed of sound in the propagation medium to convert this flight time into the one-way spatial distance from the sensor to the reflecting interface, i.e., the distance data. This non-contact detection process avoids the problem of traditional contact-type water gauges being easily disturbed or having their local flow patterns interfered with in complex water flows, thus providing a physical reference for the subsequent control of the descent depth of the current meter.

[0067] Step 302, the calculation of the real-time water depth includes: acquiring the dry flight time from the measuring probe to the riverbed under dry conditions, simultaneously acquiring the current ambient temperature as a calibration temperature, and converting the dry flight time into a dry physical distance using the calibration temperature as the spatial measurement reference surface; acquiring the wet flight time from the measuring probe to the water surface under wet conditions during model operation, and converting the wet flight time into a wet physical distance using the real-time operating temperature as the real-time water surface distance; and subtracting the dry physical distance from the wet physical distance to obtain the real-time water depth. The dry physical distance represents the reference distance from the measuring probe to the riverbed.

[0068] To address the physical limitation of energy attenuation when sound waves penetrate water from air to reach the riverbed, this embodiment provides a preferred implementation that reduces dual-medium ranging to two independent single-medium ranging measurements. Specifically, in the case of an unfilled hydraulic model, the sound wave pulse emitted by the sensor propagates only in the single air medium and is reflected by the riverbed surface. The system records the round-trip time as the dry-state flight time and simultaneously collects the ambient air temperature as the calibration temperature. Based on the calibration temperature, the system calculates the actual sound speed at that time using a conventional empirical formula for the change of air speed with temperature, and multiplies this sound speed by half of the dry-state flight time to obtain the absolute spatial span from the sensor installation location to the riverbed, i.e., the dry-state physical distance. This dry-state physical distance is stored in the control center, constituting the spatial measurement reference surface in this embodiment.

[0069] During the wet-state phase of the hydraulic model's operation while it is filled with water, the sensor emits another sound wave pulse downwards. This pulse is reflected upon encountering the water surface. The system records the wet-state flight time of the sound wave during its round trip in the air segment and simultaneously acquires the air temperature at that moment as the real-time operating temperature. The system calculates the current speed of sound corresponding to the real-time operating temperature and multiplies it by half of the wet-state flight time to determine the absolute distance from the sensor to the real-time water surface, i.e., the wet-state physical distance. This design, which extracts the ambient temperature and performs independent sound speed compensation under both dry and wet conditions, eliminates system errors introduced by air temperature fluctuations at different measurement stages.

[0070] After obtaining the two physical distances mentioned above, the system performs a geometric difference calculation, using the following formula:

[0071] H_water = D_dryphysical - D_wetphysical;

[0072] Where H_water is the real-time water depth obtained by solution, D_dryphysical is the dry physical distance calculated after calibration temperature compensation, and D_wetphysical is the wet physical distance calculated after real-time operating temperature compensation under wet conditions.

[0073] The difference between the dry and wet physical distances is essentially calculated by subtracting a fixed riverbed elevation boundary from a dynamic water surface elevation boundary. Since both distance extraction processes are performed in a single air medium, the system can acquire water depth using a conventional ultrasonic module, avoiding the signal processing bottleneck caused by the air-water interface's acoustic transmittance of only about 0.1%, thus improving the data acquisition success rate of the physical model under real-world testing conditions.

[0074] Step 303, the real-time water depth is obtained by acquiring the first reflection echo time of the water surface and the second reflection echo time of the riverbed generated by a single sound wave pulse under wet conditions during model operation. The real-time water surface distance is determined based on the first reflection echo time of the water surface and the real-time air speed of sound calculated based on the current ambient temperature. The real-time water depth is determined based on the time difference between the first reflection echo time of the water surface and the second reflection echo time of the riverbed and the preset sound speed in the water.

[0075] As an alternative and extension to the above scheme, this embodiment also provides an implementation method for synchronously acquiring dual-boundary data within a single measurement cycle. In this working mode, under wet conditions during model operation, a single acoustic pulse emitted by the sensor first undergoes a reflection at the air-water interface. The system records the flight time corresponding to the peak value of this echo extracted by the receiver, i.e., the first reflection echo time at the water surface. Simultaneously, the acoustic wave that penetrates into the water body continues to propagate downwards and undergoes a secondary reflection at the riverbed surface, penetrating the water surface and returning to the air, where it is captured again by the receiver. The system uses a time gain compensation mechanism to extract the peak value of this signal and records it as the second reflection echo time at the riverbed.

[0076] During the calculation phase, the system first uses half of the first reflection echo time from the water surface multiplied by the real-time air speed of sound calculated based on the current air temperature to extract the distance from the sensor to the water surface, i.e., the real-time water surface distance. For the calculation of water depth parameters, the system performs time-difference operations:

[0077] H_water = c_water * (t_2 - t_1) / 2;

[0078] Where H_water is the target output parameter, i.e., the real-time water depth, c_water is the preset underwater sound speed, t_2 is the time of the second reflection echo from the riverbed, and t_1 is the time of the first reflection echo from the water surface.

[0079] In this embodiment, the preset underwater sound velocity can be approximated as a constant of 1480 meters per second. For higher precision flow field observation, the system can also be equipped with a water temperature acquisition module and dynamically updated using empirical underwater sound velocity models such as the Mackenzie formula. By calculating the difference between the second and first reflection echo times, the system extracts the net flight time of the ultrasonic pulse propagating back and forth within the water body, and then directly calculates the local water depth. This scheme avoids the pre-calibration process and is suitable for physical model test scenarios where riverbed map data exhibits dynamic scouring and deposition evolution.

[0080] Example 4: A detailed explanation of how, in complex flow field measurements, a dual-threshold mechanism of curvature tracking and statistical noise reduction can be used to accurately locate the recirculation separation zone and filter out turbulent pulsation noise.

[0081] Step 401: Obtain the measured velocity data of the current measuring point and its adjacent measuring points; deduce the estimated velocity data at the current measuring point based on the measured velocity data of the adjacent measuring points; extract the local deviation between the measured velocity data of the current measuring point and the estimated velocity data, the local deviation being used to characterize the degree of nonlinear change in the local flow field profile; when the local deviation meets the preset encryption triggering condition, determine to encrypt a new measuring point between the current measuring point and the adjacent measuring points.

[0082] In this step, the system abandons the conventional method of using absolute velocity gradient as the basis for judgment. Due to the complex morphology of mountainous rivers, including numerous uniform shear flows and backflow separation zones, the degree of nonlinear variation in the flow field profile is the core factor determining the magnitude of spatial interpolation error. After acquiring data sequences from three consecutive spatial measurement points, the system uses the data distribution patterns of adjacent measurement points at both ends to mathematically deduce the theoretical expected value of the intermediate target position, i.e., the estimated velocity data. By calculating the deviation between the measured velocity data and the estimated velocity data, the system extracts the local deviation. This index directly reflects the curvature characteristics of the flow profile within that local space.

[0083] Step 402: Perform linear interpolation calculation using the measured flow velocity data of the adjacent measuring points to obtain the linear interpolated estimated flow velocity at the current measuring point location; calculate the difference between the measured flow velocity data of the current measuring point and the linear interpolated estimated flow velocity, and use the difference as the midpoint deviation; wherein, when the local deviation meets the preset encryption triggering condition, it includes: comparing the midpoint deviation with the preset reference flow velocity scale, and if the comparison result exceeds the preset allowable interpolation error coefficient, it is determined that the actual significance condition is met.

[0084] Specifically, the system performs equal-interval first-order linear interpolation using measured flow velocity data from two adjacent measuring points above and below the current measuring point to obtain the estimated flow velocity. The system then subtracts this estimated flow velocity from the measured flow velocity to obtain the difference.

[0085] After obtaining the above parameters, the system performs a comparison operation. The system extracts the absolute value of the midpoint deviation and calculates its ratio to a preset reference velocity scale. When this ratio is strictly greater than the preset allowable interpolation error coefficient, the system updates the judgment state, confirming that the current local flow regime meets the actual significance condition. This logic based on interpolation error definition ensures that even in the downstream of deep pools or the backflow separation zone of boulder wakes, the system can still keenly detect the strong curvature change from positive to negative velocity, even if the absolute value of the velocity is small, thereby triggering the densification of measurement points at key flow regime inflection points.

[0086] Step 403: Extract the measured flow velocity data of all measuring points that have been measured in the current vertical section; calculate the root mean square value of the measured flow velocity data of all the measuring points that have been measured, and dynamically assign the root mean square value to the reference flow velocity scale corresponding to the current measuring point.

[0087] In the comparison calculation, the reference velocity scale is not a static constant, but a statistical benchmark with adaptive dynamic updating characteristics. The system traverses all measured points already collected and stored on the current measurement vertical line, extracting all their measured velocity data. The system calculates the sum of squares of all extracted velocity data, calculates the average value, and then performs a square root operation to obtain the root mean square value. The system dynamically assigns this root mean square value as the reference velocity scale corresponding to the current measured point and substitutes it into the aforementioned comparison calculation, eliminating the influence of absolute velocity magnitude differences on the decision threshold.

[0088] Step 404, obtaining the measured flow velocity data of the current measuring point and its adjacent measuring points, specifically involves repeatedly sampling at each measuring point to obtain the mean of the single-point flow velocity as the measured flow velocity data, and simultaneously obtaining the standard deviation of the single-point flow velocity characterizing the intensity of turbulence fluctuations; the method further includes: performing error propagation calculation based on the standard deviation of the single-point flow velocity of the current measuring point and its adjacent measuring points to obtain the uncertainty of the midpoint deviation; when the local deviation meets the preset encryption triggering condition, it further includes: comparing the midpoint deviation with the uncertainty of the midpoint deviation, and if the comparison result exceeds the preset statistical test threshold, it is determined that the statistical significance condition is met; only when both the actual significance condition and the statistical significance condition are met simultaneously, it is determined that the operation of encrypting new measuring points is triggered.

[0089] For operating conditions where the physical model of water flow is predominantly turbulent, this step introduces a statistical verification mechanism to filter out white noise interference during instrument measurement. Specifically, the system performs continuous sampling for at least 30 seconds at each discrete measurement point to fully cover several turbulent integral time scales of the local water flow. Based on this, the system repeats the measurement at this point at least three times. The system calculates the arithmetic mean of the multiple measurements as the time mean of the single-point flow velocity and calculates its dispersion to obtain the standard deviation of the single-point flow velocity.

[0090] After reading the standard deviation of the single-point flow velocity at three consecutive measuring points, the system synthesizes the uncertainty of the midpoint deviation in that region according to the law of error propagation. This synthesis process is achieved in the following way:

[0091] σ_Δ_i = sqrt(σ_i^2 + σ_(i-1)^2 + σ_(i+1)^2) / 2; or σ_Δ_i = sqrt(σ_i² + σ_{i-1}² / 4 + σ_{i+1}² / 4);

[0092] Where σ_∆_i is the uncertainty of the midpoint deviation after synthesis, sqrt represents the mathematical operation of calculating the square root, σ_i is the standard deviation of the single-point velocity of the current measuring point, σ_(i-1) is the standard deviation of the single-point velocity of the previous adjacent measuring point, and σ_(i+1) is the standard deviation of the single-point velocity of the next adjacent measuring point.

[0093] The system further divides the absolute value of the midpoint deviation by the uncertainty of the synthesized midpoint deviation to generate a statistical test index. If this index is greater than the preset statistical test threshold, the system determines that the statistical significance condition is met. Finally, the system performs a logical AND operation, and only when both the actual significance condition and the statistical significance condition are true, the system controls the stepper motor to insert a new detection node between the corresponding measurement points.

[0094] In another embodiment, after the system reads the standard deviation of the single-point flow velocity at three consecutive measuring points, it sums the square of the standard deviation of the current measuring point with one-quarter of the square of the standard deviation of the adjacent measuring points according to the error propagation law, and then takes the square root.

[0095] Step 405: Obtain the permissible false trigger probability of the measurement system, and determine the target confidence level based on the false trigger probability; set approximately 95% confidence level as a benchmark, and extract the standard deviation multiple corresponding to the benchmark as the statistical test critical value to filter spurious curvature signals caused by turbulent pulsations. In other words, based on the target confidence level, extract the standard deviation multiple corresponding to the target confidence level as the statistical test critical value to filter spurious curvature signals caused by turbulent pulsations.

[0096] When configuring system parameters, the preset allowable interpolation error coefficient is set within the range of 0.03 to 0.05. Based on this, the system maps the 95% target confidence level to the value 2.0 by querying the standard normal distribution table. The system then embeds 2.0 as the statistical test threshold into the decision logic.

[0097] To further illustrate the technical effectiveness of the aforementioned dual-threshold mechanism, let's take a straight river boundary layer with uniform shear flow characteristics as an example: Because the velocity profile of uniform shear flow is approximately linear, although there may be significant differences in absolute velocity gradients between adjacent measuring points, the midpoint deviation (reflecting second-order curvature) remains very small. Its ratio to the reference velocity scale is far below the allowable interpolation error coefficient. The system directly determines that the actual significance condition is not met and rejects the issuance of invalid densification commands, effectively saving physical measurement resources. In contrast, if the system degenerates and uses a first-order gradient-based determination method, it will induce a large number of invalid mesh subdivision operations in the aforementioned linear velocity region.

[0098] Example 5 mainly describes how the turbulent fluctuation variance and spatial interpolation residual error of the bottom measurement points are propagated upward along the spatial dimension in the three-dimensional flow field measurement of a complex mountainous river channel.

[0099] Step 501: Extract the velocity variance and depth integral weight of each measuring point in the vertical adaptive encryption sequence, and calculate the measurement impulse noise propagation distribution along the vertical line; extract the estimated local deviation value of each interpolation sub-interval in the vertical adaptive encryption sequence, and calculate the vertical spatial interpolation residual error by combining the depth span of each interpolation sub-interval; superimpose and merge the measurement impulse noise propagation distribution and the vertical spatial interpolation residual error to generate the first uncertainty.

[0100] In this step, the system quantifies the reliability of the vertical depth-averaged flow velocity. Because the data from the bottom measuring points is affected not only by the random variance of turbulent fluctuations acquired during repeated sampling at single points, but also by the truncation error left by linear interpolation between discrete measuring points when aggregating upwards, the system calculates these two errors independently before synthesizing them. Specifically, the system extracts the depth position information of each measured node in the vertical direction, calculates the geometric distance between adjacent nodes, and assigns trapezoidal integral weights accordingly, i.e., the depth integral weights. The system multiplies the single-point flow velocity variance of each measuring point by the square of its corresponding depth integral weight and then sums the results to generate the impulse noise propagation distribution parameters.

[0101] Further, the system extracts the spatial span of each interpolation sub-interval, i.e., the depth span. To ensure the consistency of the uncertainty in physical dimensions, the system extracts the error integral variance contribution parameter to calculate the vertical spatial interpolation residual error. The system multiplies 1 / 8 of the cube of the depth span of the interpolation sub-interval with the square of the estimated value of its internal local deviation, and sums the products of all sub-intervals. Subsequently, the system superimposes and merges the above-mentioned measurement impulse noise propagation distribution with the vertical spatial interpolation residual error, and divides it by the square of the total water depth along the vertical line to generate the first uncertainty. This first uncertainty is essentially represented by the variance of the depth-average flow velocity. Its calculation formula is as follows:

[0102] σ_Vj^2 =a (1 / H_j^2) * ∑(w_k^2 * σ_vjk^2) + b (1 / H_j^2) * ∑((h_k^3 / 8) * |f_k''|^2);

[0103] Wherein, σ_Vj^2 is the first uncertainty of the j-th vertical line, H_j is the total water depth of the j-th vertical line, ∑ represents the traversal summation operation performed on all measured nodes or interpolation sub-intervals on the vertical line, w_k is the depth integral weight corresponding to the k-th measuring point, σ_vjk^2 is the velocity variance of the k-th measuring point on the j-th vertical line, h_k is the depth span of the k-th interpolation sub-interval, and |f_k''| is the estimated value of the local deviation of the k-th interpolation sub-interval. a and b are coefficients that can absorb dimensions.

[0104] In some scenarios, σ²_Vj can also be calculated using the following formula: σ²_Vj = (1 / H_j²) × Σ(w_k² × σ²_vjk) + (1 / H_j²) × Σ(c_interp × h_k² × Δ_k²). c_interp is the scaling factor related to the integration method, h_k is the depth span of the k-th interpolation sub-interval, and Δ_k is the estimated local deviation of the k-th interpolation sub-interval.

[0105] Step 502: Obtain the depth-averaged flow velocity of the current vertical line and its adjacent vertical lines, and calculate the lateral midpoint deviation based on the depth-averaged flow velocity; extract the first uncertainty corresponding to the current vertical line and its adjacent vertical lines, and calculate the uncertainty of the lateral midpoint deviation through lateral error propagation; compare the lateral midpoint deviation with the uncertainty of the lateral midpoint deviation, and when the comparison result exceeds the preset lateral significance threshold, determine to densify new vertical lines between the adjacent vertical lines, so that the trigger threshold for lateral densification is adaptively constrained by the measurement quality of the underlying vertical data.

[0106] In the lateral measurement dimension, the system calculates and extracts the lateral midpoint deviation based on the average flow velocity at depth along three adjacent perpendicular lines using a linear interpolation deviation method. The system does not directly issue encryption commands based on this lateral midpoint deviation; instead, it incorporates error information transmitted from the lower layer. The system obtains the first uncertainty of the aforementioned generated adjacent perpendicular lines, and according to the error propagation law, sums the first uncertainties of the current perpendicular line, the previous adjacent perpendicular line, and the next adjacent perpendicular line, then divides the sum by 4 to calculate the uncertainty of the lateral midpoint deviation. Alternatively, it obtains the first uncertainties of the aforementioned generated adjacent perpendicular lines, and according to the error propagation law, weights the first uncertainty of the current perpendicular line by a coefficient of 1, and weights the first uncertainties of the previous and next adjacent perpendicular lines by coefficients of 1 / 4 respectively, then synthesizes the results to calculate the uncertainty of the lateral midpoint deviation.

[0107] A dynamic linkage judgment mechanism is implemented between execution levels. The absolute value of the lateral midpoint deviation is extracted, and its ratio to the uncertainty of the lateral midpoint deviation is calculated. When this ratio is strictly greater than a preset lateral significance threshold, the system determines that the statistical test condition is met and outputs a control command to densify new vertical lines between adjacent vertical lines. For example, in a certain local shallow water area, due to the extremely small number of effective measurement points that the vertical probe can collect and the severe water turbulence, the first uncertainty of the vertical line calculated by the system will increase exponentially. This high error state, when propagated to the upper level, will cause the uncertainty of the lateral midpoint deviation to be amplified synchronously. In the comparison calculation, because the denominator becomes larger, the system's comparison result is unlikely to exceed the lateral significance threshold, and the lateral densification criterion automatically becomes conservative. In this way, the system avoids making erroneous lateral densification decisions based on unreliable lower-level data, realizing closed-loop fault tolerance and decision-making linkage across levels.

[0108] In some embodiments, determining the depth-average flow velocity and its first uncertainty for the corresponding vertical line based on the vertical adaptive encryption sequence includes:

[0109] Extract the flow velocity data and depth coordinates of each measuring point in the vertical adaptive encryption sequence, calculate the depth integral weight corresponding to each measuring point, and obtain the depth average flow velocity by numerical integration based on the flow velocity data and the depth integral weight.

[0110] Based on the velocity variance obtained from multiple repeated samplings at each measuring point and the depth integral weight, the propagation distribution of the measurement pulse noise along the vertical line is calculated.

[0111] The midpoint deviation of each interpolation sub-interval in the vertical adaptive encryption sequence is extracted as the local deviation estimate, and the vertical spatial interpolation residual error is calculated by combining the depth span of each interpolation sub-interval.

[0112] The first uncertainty is generated by superimposing and merging the measured impulse noise propagation distribution with the vertical spatial interpolation residual error.

[0113] Step 503: Aggregate the first uncertainty of each vertical line in the horizontal adaptive encryption sequence to obtain the horizontal measurement impulse noise propagation distribution; calculate the horizontal spatial interpolation residual error based on the span parameter of each horizontal interpolation sub-interval; superimpose and merge the horizontal measurement impulse noise propagation distribution and the horizontal spatial interpolation residual error to generate the second uncertainty.

[0114] After completing the adaptive traversal of the transverse section and acquiring a sufficient number of vertical line data, the system needs to further propagate the error information to the vertical dimension. Similar to the vertical-to-transverse derivation logic, the system aggregates the first uncertainty of all discrete vertical lines on the current section. The system assigns transverse integration weights based on the geometric spacing of each vertical line in the transverse section, multiplies the first uncertainty of each vertical line by the square of its corresponding weight, and sums the results to calculate and extract the transverse measurement impulse noise propagation distribution.

[0115] Simultaneously, the system extracts the horizontal distance between adjacent transverse perpendiculars as the span parameter. Combining this with the distribution characteristics of the transverse velocity gradient, it calculates the transverse spatial interpolation residual error caused by the discretization of the continuous cross-section using a finite number of perpendiculars. The system numerically superimposes and merges the transverse measurement impulse noise propagation distribution with the transverse spatial interpolation residual error, and performs total river width scaling to generate the second uncertainty representing the overall reliability of the cross-section's average flow velocity. This second uncertainty is then output and provided to the vertically layered encryption judgment module, forming a complete three-level linked uncertainty transmission chain, providing a progressively feedback benchmark for the acquisition of hydrodynamic data across the entire watershed.

[0116] In some embodiments, the cross-sectional average flow velocity and its second non-overlapping velocity are determined based on the transverse adaptive encryption sequence.

[0117] Certainty includes:

[0118] Extract the depth-average flow velocity and lateral position coordinates of each vertical line in the lateral adaptive encryption sequence, calculate the lateral integral weight corresponding to each vertical line, and obtain the cross-sectional average flow velocity by numerical integration based on the depth-average flow velocity and the lateral integral weight.

[0119] The first uncertainty of each vertical line in the horizontal adaptive encryption sequence is aggregated, and the horizontal measurement impulse noise propagation distribution is obtained by combining the horizontal integral weight;

[0120] The residual error of horizontal spatial interpolation is calculated based on the distance between adjacent vertical lines of each horizontal interpolation sub-interval as the span parameter.

[0121] The second uncertainty is generated by superimposing and merging the transverse measurement impulse noise propagation distribution with the transverse spatial interpolation residual error.

[0122] Example 6: How to intercept invalid or dangerous encryption commands by using dual boundary conditions of physical size and iteration number when the system performs three-dimensional adaptive mesh encryption.

[0123] Step 601: Obtain the preset minimum spatial spacing and the preset maximum number of iterations, and initialize the adaptive encryption loop count for the current dimension to zero.

[0124] In adaptive densification algorithms, without physical boundary constraints, theoretical mesh subdivision could theoretically proceed indefinitely. To prevent this, the system reads pre-configured boundary constraint parameters during the initialization phase. Specifically, the system obtains a preset minimum spatial spacing, which is primarily limited by the physical dimensions of the flow velocity sensors. For example, for propeller-type flow meters, the system sets the minimum spatial spacing to twice the diameter of the sensor impeller. This ensures that when flow velocity sensors are densely deployed, the flow fields between probes will not interfere with each other due to excessively small spacing between adjacent measuring points, while also preventing physical collisions between the probes and the riverbed bottom. Furthermore, the system simultaneously obtains a preset maximum number of iterations. In complex mountainous river channels, local areas may exhibit extreme flow conditions such as strong eddies, causing continuous and drastic changes in flow velocity curvature. To prevent the control algorithm from getting stuck in an infinite densification loop in such areas, the system limits the maximum number of iterations to between 3 and 4.

[0125] Step 602: Verify the space span to be inserted corresponding to the position to be encrypted. If the space span to be inserted is less than the minimum space spacing, or the number of adaptive encryption loops executed in the current dimension has reached the maximum number of iterations, then intercept the encryption command and terminate the adaptive encryption iteration of the current dimension.

[0126] Before issuing an encryption command, a boundary security check must be performed if both threshold verification conditions are met. The system calculates the absolute geometric distance between the two measuring points at the ends of the current interval to be encrypted, and extracts the spatial span to be inserted. The system then compares this spatial span to be inserted with the aforementioned minimum spatial spacing. If the comparison result shows that the spatial span to be inserted is less than the minimum spatial spacing, it indicates that further subdivision will exceed the hardware's physical detection limit, and the system immediately intercepts the encryption command.

[0127] Simultaneously, the system's internal counter records the grid subdivision level of the current dimension in real time. The system compares the number of adaptive encryption loops executed in the current dimension with the maximum number of iterations. If the comparison shows that the current number of loops >= the maximum number of iterations, the system also intercepts the encryption command. By setting this hard boundary, even under extremely complex local turbulence conditions, the maximum number of measurement points on a single measurement vertical line is strictly constrained within a reasonable range. For example, based on an initial of 5 measurement points and a maximum of 4 iterations, the final number of measurement points on a single vertical line is constrained to a maximum of approximately 80 points. Based on this, the system terminates the adaptive encryption iteration of the current dimension and forces the measurement equipment to enter the detection process of the next spatial dimension, thereby ensuring that the total measurement time of the entire hydraulic physics model measurement task remains within a controllable range while maintaining the sampling density of local key flow states.

[0128] Example 7: In a certain scenario, the implementation process is briefly as follows:

[0129] Step 1: Obtain river topographic data and build an experimental measurement track platform based on the river topographic data from the hydraulic physics model.

[0130] In this step, the hydraulic model is a physical simulation device based on similarity theory, constructed in a laboratory at a scale to a real hydraulic engineering project. It visually presents complex hydraulic phenomena by reproducing processes such as water flow dynamics, sediment movement, and structural stress, providing scientific support for the design, operation, and management of hydraulic engineering projects. River topographic data includes digital information such as underwater topography, cross-sectional morphology, and boundary features. Cross-sectional morphology includes cross-sections and longitudinal sections; the cross-section refers to the profile perpendicular to the water flow direction, while the longitudinal section is the profile along the water flow direction. Key topographic data includes elevation, river width, curvature (radius of curvature), and river bend morphology.

[0131] The experimental measurement track platform includes an ultrasonic sensor 1, stepper motors 2 and 3, a telescopic rod 4, a flow velocity sensor 5, a horizontal axis 6, stepper motors 7 and 8, and a vertical axis 9. The horizontal axis 6 is perpendicular to the river channel's central axis, and the horizontal axis 6 and vertical axis 9 are located on the same horizontal plane, with an error of less than 1 mm between the two ends of the horizontal and vertical axes.

[0132] Step Two: The ultrasonic sensor 1, located at the top of the current meter, emits ultrasonic waves of a specific intensity. These waves are reflected by the water body being measured and the riverbed. The distance between the current sensor and the water body and the riverbed is determined by calculating the time difference of the reflected ultrasonic waves. After the operation is completed, the horizontal stepper motor 3, carrying sensor 1, completes the cross-sectional measurement and transmits the measured cross-sectional distances to the water surface, riverbed, and the location information of the measurement points to the control center. The control center then compares and corrects the measurement information with the acquired topographic data.

[0133] In this implementation scheme, data collected by ultrasonic ranging sensors is first acquired synchronously, and then integrated into a complete dataset according to a unified format of "time-section number-water surface distance-riverbed distance". Ultrasonic ranging sensor 1 is installed on top of the current meter (calculating water surface and riverbed distances based on the time difference between ultrasonic wave transmission and reception and reflectivity), and propeller current meter is used as current sensor 5. All sensors operate at a frequency of acquiring data once every 15 seconds, repeated three times to ensure the accuracy of each set of data. A threshold comparison method is used to verify the measurement data. The specific operation is as follows: the control center compares the water surface width and valley topography of the transverse measurement section obtained by calculation module 1 with the river channel topography data from the acquisition module (the water surface width of the hydraulic model is calculated based on the scaling ratio between the model and the prototype), and corrects the speed control signal of the transverse stepper motor 3.

[0134] Step 3: Based on the water surface distance and riverbed distance received by the control center, calculate the optimal spacing of the water flow velocity measuring points suitable for the hydraulic model according to the test requirements; through the control center, generate adjustment commands for adjusting the speed and time of the stepper motor 2. The stepper motor located on top of the current meter transports the current sensor 5 to different underwater depths according to the control signal. The current sensor measures the vertical water flow velocity of the river channel and transmits the measurement data to the storage system.

[0135] In this step, the optimal spacing refers to obtaining the velocity characteristic values ​​(surface velocity, bottom velocity, velocity inflection point, etc.) of the vertical measurement section. Surface velocity refers to the flow velocity at the free surface of the fluid (i.e., the interface in contact with air) in a river, channel, or pipe; in hydraulic models, it often refers to the velocity at a distance of 0.02m below the water surface. Bottom velocity refers to the flow velocity at a specific small distance (often 0.02m from the bottom of the riverbed in hydraulic models) near the riverbed. It is a key indicator characterizing the intensity of bottom flow, directly related to whether sediment can be initiated and transported, and the survival environment of benthic organisms. The velocity inflection point refers to the inflection point on the velocity profile (velocity distribution) curve. Mountainous river channels have complex topography, which can lead to complex three-dimensional flow states such as separation, eddies, and secondary flows. To accurately measure the location and velocity of the velocity inflection point, the vertical velocity measurement points need to be densified. The specific operation is as follows: Initially, it is assumed that 5 flow velocity points are measured at uniform intervals along the vertical cross section. The control center analyzes the adjacent measuring points based on the measurement data of the flow velocity sensor 5. When the deviation between adjacent flow velocity measuring points is greater than 10% (the specific difference is determined according to the test requirements), a measuring point is added between the two adjacent measuring points according to linear interpolation. The control center generates adjustment commands to adjust the running speed, time, and flow velocity measurement of the stepper motor 2. The above operation is repeated until the flow velocity measurement of the vertical cross section is completed.

[0136] Step 4: After steps 2 and 3 are completed, stepper motor 2 returns to its initial position. Stepper motor 3, located on the horizontal axis, moves the measuring body to the next vertical measurement section according to the set step length. Step 3 is repeated until the measurement of the entire cross-section is completed, and the position information and flow velocity of the measurement points are transmitted to the control center. The control center corrects based on the measurement information from adjacent vertical sections.

[0137] The stepper motor 2 returning to its initial position refers to its position before the control center generates the adjustment command. The measuring body refers to the integrated unit of the ultrasonic ranging sensor 1, the telescopic rod 4, and the flow velocity sensor 5.

[0138] In this step, the optimal spacing refers to obtaining the flow velocity characteristic values ​​(bank velocity, maximum velocity, velocity inflection point, etc.) of the vertical measurement section. Bank velocity typically refers to the water flow velocity near the bank or boundary of a river, canal, or coastline. In hydraulic models, to prevent the propeller of the velocity sensor 5 from touching the bank slope, the bank velocity measurement point is often set at a position 2cm at the water-solid (bank slope) interface. Mountainous river channels have complex topography, which can lead to complex three-dimensional flow patterns such as flow separation, eddies, and secondary flows. To accurately measure the location and velocity of the velocity inflection point, the vertical velocity measurement sections need to be densified.

[0139] The specific operation is as follows: Considering that the flow pattern of mountainous rivers is mainly turbulent, and the overall flow velocity within the river channel exhibits an exponential or logarithmic distribution due to the influence of the boundary layer (referring to the riverbed), according to Newton's law of internal friction, the fluid internal friction force T is related to the contact area A between the flow layers and the velocity gradient. Proportional This is a proportionality coefficient. It indicates that the flow velocity is higher in the middle of the river and lower near the bank.

[0140]

[0141] Initially, it is assumed that five vertical velocity measurement sections are measured at uniform intervals in the transverse section. The control center analyzes the average velocity of adjacent vertical velocity measurement sections. When the average velocity deviation is greater than 10% (the specific difference is determined according to the test requirements), the control center adds measurement sections by linear interpolation for the adjacent vertical measurement sections. The control center generates adjustment commands for adjusting the operating speed, time, and velocity measurement of the stepper motor 3. The above operation is repeated until the velocity measurement of the transverse section is completed.

[0142] Step 5: Stepper motor 7, located on the longitudinal axis, moves the transverse axis to the next cross section according to the step size set according to the test requirements. The next cross section refers to the cross section measurement interval determined according to the test requirements. In the hydraulic model, the velocity measurement cross section is about 1m. In the river bend section and the section with abrupt topography, the measurement needs to be densified according to the measurement results. The control center is corrected according to the measurement information of the adjacent cross sections.

[0143] The specific operation is as follows: Initially, it is assumed that the transverse cross-sections are measured at 1m intervals. The control center analyzes the average flow velocity of adjacent transverse flow velocity measurement cross-sections. When the deviation of the average flow velocity or the maximum flow velocity is greater than 10% (the specific difference is determined according to the test requirements), the control center adds a measurement cross-section by linear interpolation of the adjacent vertical measurement cross-section. The control center generates adjustment commands for adjusting the operating speed and time of stepper motors 7 and 8 as well as the flow velocity measurement, and measures the newly added flow velocity measurement cross-section. Repeat the above operation until the river flow velocity measurement is completed.

[0144] According to one aspect of this application, the encryption criterion process can also be as follows:

[0145] In numerical analysis, linear interpolation in the interval The maximum upper bound of the error is:

[0146] in The distance between measuring points, Let be the second derivative (curvature) of the velocity profile. Therefore, curvature, rather than gradient, should be used as the encryption criterion.

[0147] For the measurements already taken on the vertical profile flow rate point For each internal measurement point ( ), calculate the midpoint deviation:

[0148] in The estimated value is obtained by linear interpolation between two adjacent measurement points:

[0149] For the case of equal spacing, it simplifies to: ;

[0150] Midpoint deviation The relationship with local curvature is This directly reflects the magnitude of the linear interpolation error.

[0151] The encryption trigger condition is: ;

[0152] in For reference velocity scale (take the root mean square value of the velocity at each measured point on this vertical cross section). The allowable interpolation error coefficient (determined based on experimental accuracy requirements, it is recommended to take...) (This corresponds to a relative interpolation error tolerance of 3% to 5%).

[0153] When this condition is triggered, in the interval and After adding measuring points to the midpoints to form a denser measuring point distribution, a reassessment was performed.

[0154] According to one aspect of this application, in order to solve the problem that measurement noise caused by turbulent pulsation may generate spurious curvature signals, the following solution is provided:

[0155] In physical models of mountainous river channels, water flow is mostly in a turbulent state. A single reading from a flow velocity sensor (propeller current meter) at a given measuring point is actually the time-averaged flow velocity. Superimposed turbulence deviation : ;

[0156] Assume repeated measurements at each measuring point Next (take) (times), to obtain the time average and standard deviation The standard error of the velocity estimation at this measuring point is: ;

[0157] Midpoint deviation The uncertainty (derived from the uncertainty propagated from the three measurement points):

[0158] ;

[0159] In the case of equal spacing, The error propagation formula;

[0160] Note that, based on the curvature criterion, a statistical significance test is added to form a dual-threshold decision mechanism: encryption is only triggered when the midpoint deviation simultaneously meets both the conditions of "actual significance" and "statistical significance".

[0161] Condition 1 (Actual significance, from Improvement 1): ;

[0162] Condition 2 (Statistical Significance): ;

[0163] in To determine the critical value for the statistical test, take... This corresponds to a confidence level of approximately 95%.

[0164] Only when both conditions one and two are met is it determined that the region contains real flow velocity curvature features that require encryption.

[0165] According to one aspect of this application, the first step is completed at the vertical level. After adaptive measurement of a vertical line, calculate the depth-averaged flow velocity and its uncertainty along that vertical line:

[0166] ;

[0167] in The water depth at this vertical line, This represents the actual number of measurement points after adaptive encryption. The weights are for trapezoidal integrals.

[0168] The uncertainty of the average flow velocity consists of two parts—the measurement noise propagation term and the interpolation error term:

[0169] ;

[0170] The first term represents the result of measurement noise propagating through the integral weight at each measuring point; the second term represents the estimate of the linear interpolation residual error for each sub-interval. Estimated from the midpoint deviation in Improvement 1, For the first The spacing between sub-intervals.

[0171] In the horizontal encryption decision-making process, the midpoint deviation is changed to: ;

[0172] The uncertainty is:

[0173] The horizontal encryption triggering condition also adopts a dual threshold mechanism (actual significance + statistical significance).

[0174] If the vertical measurement quality of a certain vertical line is poor (few measuring points, shallow water, strong turbulence), its Large, leading to Larger data sets make it more difficult to meet statistical significance conditions, automatically leading to a more conservative horizontal encryption criterion—avoiding decisions made on higher levels based on unreliable lower-level data. Conversely, if vertical measurements are sufficient (resulting in denser measurement points after adaptive encryption), the horizontal encryption criteria become more conservative. The smaller size and improved sensitivity of the lateral criterion enable the capture of more subtle changes in lateral flow velocity. This adaptive sensitivity adjustment is an essential manifestation of hierarchical coupling, which cannot be achieved by a single-level encryption method.

[0175] According to one aspect of this application, the process of vertical adaptive measurement is as follows:

[0176] Read water surface distance Distance from the riverbed Tolerance coefficient Statistical critical value Minimum measuring point spacing (Take twice the diameter of the sensor impeller), maximum number of iterations .

[0177] Divide the area into equal parts between 0.02m below the water surface and 0.02m above the riverbed. A set of measurement points is formed, creating an initial set of measurement points. Repeat the measurement at each measuring point. Second-rate( Record the average flow rate. and standard deviation .

[0178] Iterative Encryption (the first) iteration ):

[0179] (a) For all internal points in the current measurement point set Calculate the midpoint deviation. and its uncertainty ;

[0180] (b) Test the double threshold condition: if and , will the interval and Marked as "to be encrypted";

[0181] (c) For all marked intervals, if the interval length Insert a new measuring point at the midpoint of the interval and perform the measurement;

[0182] (d) If no new measurement points are generated in this iteration, or if the target has been reached... ,termination.

[0183] Output: Locations of all measuring points along the vertical line, flow velocity values, and uncertainties; depth-averaged flow velocity. and its uncertainty .

[0184] According to one aspect of this application, the correction algorithm is further as follows:

[0185] There may be installation deviations (translation, rotation) between the track platform and the hydraulic model, which require systematic calibration and correction by comparing the measured terrain with the known terrain.

[0186] Calibration phase: Under anhydrous conditions, using an acoustic sensor... Riverbed elevation was measured at several calibration points located at different locations along the river channel. The design elevation calculated based on the model's scale ratio. By comparison, the elevation deviations of each calibration point are obtained:

[0187] The spatial distribution of the deviation is fitted using a bilinear trend surface:

[0188] coefficient Determined by the least squares method:

[0189] in Reflecting the overall translational deviation, , Reflects longitudinal and lateral tilt deviations. It reflects the torsional deviation.

[0190] Correction Phase: For all subsequent measurement locations, the measured elevations will be corrected to:

[0191] ;

[0192] If the residual at a certain calibration point Exceeding the tolerance of model manufacturing (usually the model scale) If the value is less than or equal to 100 mm, mark the point to indicate that there may be a local manufacturing defect or measurement abnormality in the model, which requires manual verification.

[0193] According to one aspect of this application, the dual-echo detection and dual-medium sound velocity calculation process specifically includes:

[0194] The sensor is mounted at a fixed height above the river channel and emits ultrasonic pulses downwards. The pulses pass through the air layer and the water layer in sequence, generating reflected echoes at the air-water interface and the water-bottom interface, respectively.

[0195] Step 1: Launch and Data Acquisition

[0196] The sensor emits an ultrasonic pulse and simultaneously starts a high-precision timer. The receiver records the complete echo time-domain waveform, which shows two distinct echo peaks: the first echo originates from reflection from the water surface (time of flight denoted as ). The second echo comes from reflection from the riverbed (recorded as flight time ). ).

[0197] Step 2: Dual-echo identification and extraction

[0198] Since the water surface reflection coefficient depends on the air-water acoustic impedance mismatch ratio, the water surface echo has strong energy. However, the riverbed echo undergoes two crossings of the air-water interface (one transmission into the water and one transmission back into the air) and round-trip attenuation in the water, resulting in a significant reduction in signal strength. Data processing requires the following tasks:

[0199] The original echo signal is bandpass filtered for noise reduction, and then threshold detection or matched filtering is used to pinpoint the peak positions of the two echo envelopes in the time domain. The first peak corresponds to... This refers to the round-trip time of the pulse from the sensor, through the air to the water surface, and back to the sensor. The second peak corresponds to... This refers to the total round-trip time of the pulse as it travels through the air to the water surface, penetrates the water surface, propagates in the water to the riverbed, is reflected, and then penetrates the water surface again before returning to the sensor via the air.

[0200] In practice, the second echo is easily obscured by the trailing edge of the surface echo, multiple reflections, and interference from underwater scatterers. Time gain compensation (TGC) is needed to amplify the long-distance echo, and an adaptive threshold is required to eliminate the "blind zone" after the first echo in order to reliably extract the second echo.

[0201] Step 3: Determining the speed of sound

[0202] Speed ​​of sound in air It is sensitive to temperature and requires a temperature sensor installed next to the main sensor to measure the air temperature in real time. (Unit: °C), calculated using empirical formula:

[0203] ;

[0204] Speed ​​of sound in water It depends on water temperature, salinity, and pressure. For shallow freshwater channels, this can be simplified to depend solely on water temperature. Empirical formulas, or approximations of constants. If high accuracy is required, a water temperature sensor should be installed, and corrections should be made using the Mackenzie formula or a similar model.

[0205] Step 4: Geometric Solution

[0206] The distance from the sensor to the air segment above the water surface is: ;

[0207] The round-trip time of the pulse in the water is the total flight time minus the round-trip time in the air segment: ;

[0208] The water depth is: ;

[0209] According to one aspect of this application, the specific process of the two-step measurement method is as follows:

[0210] Phase 1: Dry calibration (anhydrous conditions)

[0211] When the river is dry (or drained), the ultrasonic sensor measures the distance from its installation location to the riverbed surface. In this mode, the pulse propagates only through the air, which is the standard ultrasonic ranging operating mode.

[0212] For each measurement point (or the location corresponding to each sensor) within the sensor coverage area, record the dry distance. ,in This indicates the plane coordinates of the measurement point. If the riverbed is uneven, multiple locations need to be used for calibration. The air temperature at each calibration point should also be recorded. This is used for subsequent temperature compensation.

[0213] The calibration values ​​are stored in the controller as a "riverbed reference level". Specifically: ;

[0214] If the absolute elevation of the sensor installation location is known... Then the riverbed elevation is:

[0215]

[0216] This step is essentially a complete spatial mapping of the riverbed topography, which is completely consistent with the dry calibration in "Improvement 5: Topographic Correction" in the supplementary plan. The idea behind topographic correction is to acknowledge that the riverbed is not flat, not to assume an ideal plane, but to record the actual topography point by point as a benchmark.

[0217] Phase Two: Wet Operation (After Water Filling)

[0218] After the river channel is filled with water, the sensor measures distance downwards. The pulse is reflected upon encountering the water surface (the water surface has an extremely high reflection coefficient, resulting in almost total reflection) and returns to the sensor. At this point, the measured distance is the air segment distance from the sensor to the water surface. ;

[0219] Note that the temperature used here is the real-time temperature at the time of operation. To calculate the speed of sound in air and perform temperature compensation.

[0220] Phase 3: Water Depth Calculation

[0221] Water depth equals the dry reference distance minus the wet measurement distance:

[0222] If expressed in terms of elevation, then the water surface elevation is... The water depth is:

[0223] ;

[0224] Phase 4: Error Correction

[0225] Temperature compensation is a crucial step. If the calibration temperature is 20°C while the operating temperature is 5°C, the air speed of sound differs by approximately 2.7%, which will introduce an error of about 27 mm over a distance of 1 m. The solution is to avoid directly storing the distance value. Instead, it stores the original flight time. During runtime, the speed of sound is recalculated using a uniform real-time velocity. Alternatively, more simply, since the water depth is determined by the difference, and both measurements use the same velocity of sound formula (taking the runtime temperature), the calibrated flight time needs to be converted to the equivalent distance at the current temperature. ;

[0226] The flight time was actually measured at the calibration temperature. The physical distance is fixed (the riverbed is stationary), so the correct approach is to first reconstruct the physical distance from the calibration data, and then directly use the difference between the physical distances.

[0227] ; ; ;

[0228] Temperature values ​​need to be recorded synchronously during calibration.

[0229] In addition, if the installation height of the sensor changes slightly after long-term operation (due to thermal expansion and contraction of the bracket, settlement, etc.), it needs to be recalibrated periodically under waterless conditions, or a reference reflective surface at a known distance should be set up for online verification.

[0230] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details in the above embodiments. Within the scope of the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all fall within the protection scope of the present invention.

Claims

1. A method for automatically measuring flow velocity of a river in a complex mountainous area, characterized in that, include: Acquire river topographic data and establish a spatial measurement reference surface based on the river topographic data; Obtain the real-time water surface distance to the water surface, and determine the real-time water depth based on the real-time water surface distance and the spatial measurement reference plane; Within the range of the real-time water depth, obtain the flow velocity data of vertical measuring points, determine whether to encrypt new measuring points based on the difference in flow velocity data of adjacent measuring points, and obtain a vertical adaptive encryption sequence; The depth-average flow velocity and its first uncertainty for the corresponding vertical line are determined based on the vertical adaptive encryption sequence. The depth-average flow velocity and its first uncertainty of adjacent vertical lines are obtained along the transverse cross section. The trigger threshold of transverse encryption is adjusted using the first uncertainty to determine whether to encrypt a new vertical line, thus obtaining a transverse adaptive encryption sequence. The cross-sectional average flow velocity and its second uncertainty are determined based on the aforementioned transverse adaptive encryption sequence. Along the longitudinal channel, based on the average flow velocity of the adjacent cross sections and its second uncertainty, it is determined whether to densify a new cross section, thus obtaining a longitudinal adaptive densification sequence; The three-dimensional flow field data is output based on the acquired adaptive encrypted sequences of each dimension.

2. The method of claim 1, wherein, The real-time water surface distance to the water surface is obtained, and the real-time water depth is determined based on the real-time water surface distance and the spatial measurement reference plane. Specifically, this is done by emitting a sound wave pulse and receiving the reflected echo from the corresponding interface, extracting the corresponding distance data based on the flight time of the reflected echo, and calculating the real-time water depth.

3. The method of claim 2, wherein, The calculation of the real-time water depth includes: The dry flight time from the measurement probe to the riverbed is obtained under waterless and dry conditions. The dry flight time is converted into a dry physical distance by combining the calibration temperature, which is used as the spatial measurement reference surface. The wet flight time from the measurement probe to the water surface is obtained under wet conditions during model operation. The wet flight time is converted into wet physical distance by combining the real-time operating temperature, and used as the real-time water surface distance. The real-time water depth is obtained by subtracting the dry physical distance from the wet physical distance.

4. The method of claim 2, wherein, The real-time water depth is obtained by acquiring the first reflection echo time of the water surface and the second reflection echo time of the riverbed generated by a single sound wave pulse under wet conditions during model operation. The real-time water surface distance is determined based on the first reflection echo time of the water surface and the real-time air speed, and the real-time water depth is determined based on the time difference between the first reflection echo time of the water surface and the second reflection echo time of the riverbed and the preset underwater speed.

5. The method according to claim 1, characterized in that, The establishment of a spatial measurement reference surface based on the river channel topographic data includes: Obtain the measured elevations distributed at preset calibration points under anhydrous dry conditions; Extract the corresponding design elevation from the river channel topographic data, and calculate the elevation deviation between the measured elevation and the design elevation; A spatial trend surface is fitted based on the elevation deviation of each calibration point, and the spatial trend surface characterizes the combined translational, tilting and torsional deviations of the measuring equipment. The spatial trend surface is used to compensate for errors in the measured terrain data to generate the spatial measurement reference surface.

6. The method according to claim 1, characterized in that, The method of determining whether to add new measuring points based on the difference in flow velocity data from adjacent measuring points includes: Acquire the measured flow velocity data of the current measuring point and its adjacent measuring points; The estimated flow velocity data at the current measuring point is derived from the measured flow velocity data of the adjacent measuring points. The local deviation between the measured velocity data and the estimated velocity data at the current measuring point is extracted. The local deviation is used to characterize the degree of nonlinear change in the local flow field profile. When the local deviation meets the preset encryption trigger condition, it is determined that a new measurement point is encrypted between the current measurement point and the adjacent measurement point.

7. The method according to claim 6, characterized in that, The estimated flow velocity data at the current measuring point is derived from the measured flow velocity data of the adjacent measuring points, and the local deviation between the measured flow velocity data and the estimated flow velocity data at the current measuring point is extracted, specifically including: Linear interpolation calculations are performed using the measured flow velocity data from the adjacent measuring points to obtain the linear interpolation estimated flow velocity at the current measuring point location; Calculate the difference between the measured flow velocity data at the current measuring point and the estimated flow velocity by linear interpolation, and use the difference as the midpoint deviation. The step of meeting the preset encryption triggering condition when the local deviation amount meets the preset encryption triggering condition includes: comparing the midpoint deviation amount with the preset reference flow velocity scale; if the comparison result exceeds the preset allowable interpolation error coefficient, it is determined that the actual significance condition is met.

8. The method according to claim 7, characterized in that, The process of obtaining the measured flow velocity data of the current measuring point and its adjacent measuring points involves repeatedly sampling at each measuring point, obtaining the mean value of the single-point flow velocity as the measured flow velocity data, and simultaneously obtaining the standard deviation of the single-point flow velocity that characterizes the intensity of turbulent fluctuations. The method further includes: performing error propagation calculation based on the single-point flow velocity standard deviation of the current measuring point and its adjacent measuring points to obtain the uncertainty of the midpoint deviation; The step of the local deviation amount meeting the preset encryption triggering condition further includes: comparing the midpoint deviation amount with the uncertainty of the midpoint deviation amount; if the comparison result exceeds the preset statistical test threshold, it is determined that the statistical significance condition is met. The operation of encrypting new measurement points is triggered only when both the actual significance condition and the statistical significance condition are met simultaneously.

9. The method according to claim 1, characterized in that, The determination of the depth-averaged flow velocity and its first uncertainty for the corresponding vertical line based on the vertical adaptive encryption sequence includes: Extract the velocity variance and depth integral weight of each measurement point in the vertical adaptive encryption sequence, and calculate the measurement impulse noise propagation distribution along the vertical line; Extract the estimated local deviation values ​​of each interpolation sub-interval in the vertical adaptive encryption sequence, and calculate the vertical spatial interpolation residual error by combining the depth span of each interpolation sub-interval. The first uncertainty is generated by superimposing and merging the measured impulse noise propagation distribution with the vertical spatial interpolation residual error.

10. The method according to claim 1, characterized in that, The step of adjusting the trigger threshold for lateral encryption using the first uncertainty to determine whether to encrypt a new vertical line includes: Obtain the depth-average flow velocity of adjacent vertical lines, and calculate the lateral midpoint deviation based on the depth-average flow velocity; Extract the first uncertainty corresponding to the adjacent vertical lines, and calculate the uncertainty of the lateral midpoint deviation through lateral error propagation; The lateral midpoint deviation is compared with the uncertainty of the lateral midpoint deviation. When the comparison result exceeds the preset lateral significance threshold, it is determined that a new vertical line is to be densified between the adjacent vertical lines, so that the trigger threshold for lateral densification is adaptively constrained by the measurement quality of the underlying vertical data.

Images