Method for forming seabed digital elevation model, including obtaining data on depths and elevation of seabed from optical and hydroacoustic sensor channels

Abstract

FIELD: processing seabed monitoring data.SUBSTANCE: claimed method for generating a digital elevation model of the seabed is distinguished by the fact that at least two data sets are obtained, for each of which a private confidence indicator is calculated, then interchannel and retrospective discrepancies are calculated in terms of depth, object contours, acoustic texture and relief morphology. The cause of the discrepancies is classified, on the basis of which a verification method is assigned: a repeat pass from a different angle, shooting with a different sensor channel, visual verification, high-frequency hydroacoustic refinement, repeated alignment using bottom landmarks or exclusion of an area from automatic confirmation. After completing the check, a passport is generated for each cell and included in the final digital elevation model of the seabed when the specified confidence level is reached, and if the level is not reached, the cell is marked as disputed.EFFECT: reduction in the number of false elements of the digital elevation model.1 cl, 1 dwg

Description

[0001] The invention relates to the processing of hydrographic, bathymetric, hydroacoustic, and optical seabed survey data, the construction of digital elevation models, the monitoring of underwater infrastructure, the detection of bottom objects, and the visual inspection of irregularities. The solution can be used in unmanned aerial vehicles, underwater gliders, autonomous surface vehicles, and towed and stationary water survey systems.

[0002] RU 2 858 003 C1 "Device for mapping seabed relief" has been adopted as the closest equivalent. This equivalent describes a system comprising an optical unit, an acoustic unit, an altimeter, a side-scan sonar, a multibeam echosounder, a profiler, a route computer, a satellite navigation system, and an electronic cartographic navigation information system. The description of this equivalent also covers the construction of a digital seabed relief model, the use of retrospective bathymetric data, the search for systematic discrepancies, and relief visualization.

[0003] World-class technology includes methods for integrating multibeam echosounder surveys, side-scan sonar, optical surveys, laser bathymetry, navigation measurements, and archival bathymetric data. Specific areas of development include neural network and probabilistic bathymetry reconstruction, drift correction for autonomous vehicles, local sonar frame matching, and confidence mapping. However, existing solutions often lack a complete technological chain: identifying discrepancies between sources, classifying the cause of the discrepancy, selecting a verification sensor channel, scheduling a repeat observation, and incorporating the result into the digital model only after confirmation.

[0004] Disadvantages of existing technical solutions:

[0005] When combining optical, hydroacoustic and retrospective data, discrepancies are often eliminated by post-processing without automatically determining the cause of their occurrence;

[0006] The digital elevation model may include areas where data from different channels contradict each other, but no indication of the need for visual verification has been generated;

[0007] Known solutions do not always distinguish between the discrepancy caused by real bottom change, navigation error, acoustic shadow, multi-reflection, water turbidity or interpolation error;

[0008] Visual inspection of bottom objects and inhomogeneities is usually assigned by the operator, which increases labor intensity and depends on the qualifications of the specialist;

[0009] A single digital model cell passport is not generated, containing the depth, measurement source, confidence level, discrepancy type, confirmation history, and re-observation recommendation.

[0010] The technical challenge is to improve the reliability of the digital seabed elevation model by automatically identifying and classifying discrepancies between optical, hydroacoustic and historical data, assigning visual or sensor verification only to technically questionable areas and incorporating data into the model based on the confirmed confidence level.

[0011] The solution to the technical problem is ensured by the fact that the Method for forming a digital model of the seabed relief, including obtaining data on the depths and relief of the seabed from optical and hydroacoustic sensor channels, bringing the said data to a single coordinate system and constructing a digital model of the seabed relief, characterized in that for each cell or local section of the digital model, at least two sets of data are obtained, selected from a group including data from a multibeam echo sounder, a side-scan sonar, a profiler, a laser bathymeter, an optical camera and a retrospective bathymetric model, for each set of data a private confidence indicator is calculated, depending at least on the quality of the navigation reference, the level of interference, the overlap of the survey, the angle of observation or probing, the stability of marks and consistency with neighboring sections, then interchannel and retrospective depth discrepancies are calculated,contours of objects, acoustic texture and relief morphology, classify the cause of the said discrepancies as one of the conditions including a real change in relief, navigational offset, acoustic shadow, re-reflection, optical opacity, insufficient resolution, interpolation error, bottom object or unconfirmed anomaly, based on the classified cause, automatically assign a verification method selected from a group including a repeated pass from a different angle, shooting with a different sensor channel, visual verification, high-frequency hydroacoustic refinement, repeated alignment using bottom landmarks or excluding a section from automatic confirmation, after completing the verification, generate a passport for each cell containing the confirmed depth, the source of confirmation, the confidence interval, the type of the detected discrepancy and the admissibility indicator for using the cell,and include the cell in the final digital seafloor elevation model only when a specified confidence level is reached, and if the specified level is not reached, a task for additional verification is generated or the cell is marked as controversial.

[0012] The technical result consists in reducing the number of false elements of the digital model, reducing systematic errors in combining surveys of different times and different natures, increasing the reliability of detecting bottom objects and changes in relief, reducing the volume of manual expert verification and increasing the reproducibility of constructing a digital model of the seabed relief.

[0013] Essence of the invention.

[0014] The essence of the invention is that a digital seafloor relief model is formed not by simply combining measured depths and interpolating, but by cell-by-cell comparison of multiple data sources with automatic detection of discrepancies, determination of their technical causes, and the assignment of visual or sensory verification. Each cell of the model is considered an object with a set of attributes: coordinates, depth, measurement source, navigational reference accuracy, sensor type, acquisition time, noise level, presence of acoustic shadow, optical visibility indicator, consistency with neighboring cells, and historical data.

[0015] In the first stage, data is obtained from a multibeam echosounder, side-scan sonar, profiler, laser bathymeter, optical camera, or other sensor channels. The data is converted into a unified coordinate system and a unified depth reference system. A unique confidence indicator is calculated for each source, taking into account the survey geometry, carrier height or depth, beam angle, interference level, overlap, navigation quality, water conditions, and the stability of the marks.

[0016] In the second stage, the current data is compared with the historical model and with each other. For each cell, the interchannel depth discrepancy, the discrepancy between bottom object contours, the change in acoustic texture, the discrepancy between the optical and acoustic signatures, the difference from adjacent cells, and the difference from expected relief morphology are calculated. The discrepancies are then classified by type: probable actual bottom change, navigational offset, acoustic shadow, multi-reflection, water optical opacity, insufficient resolution, interpolation error, probable bottom object, or unconfirmed anomaly.

[0017] In the third stage, the verification method is selected based on the type of discrepancy. If the discrepancy is related to a possible bottom object, visual verification is performed using an optical channel or high-frequency sonar from a lower altitude. If the discrepancy is related to navigational drift, a re-alignment is performed using stable bottom landmarks. If the cause is water turbidity, a hydroacoustic channel is selected. If the cause is related to an acoustic shadow, the survey angle or sounding frequency is adjusted. The verification priority is determined based on the discrepancy magnitude, proximity to infrastructure, the hazard of the object, the accuracy class of the area, and the cost of a re-over.

[0018] In the fourth stage, after the verification is completed, the digital model cell's datasheet is updated. The datasheet includes the confirmed depth, confidence interval, confirmation source, type of verification performed, a note indicating a real change or error, and a flag indicating whether the cell is acceptable for navigation, engineering, or monitoring purposes. If the verification fails to confirm one of the options, the cell is marked as questionable and excluded from the automatic generation of the final model until further verification.

[0019] Thus, the method overcomes the drawback of known solutions: that inconsistent data can be smoothed or interpolated without a technical understanding of the cause of the discrepancy. The proposed solution generates a digital model not only as a depth array, but also as a model with controlled reliability and documented data origin.

[0020] The invention is explained by a figure.

[0021] The figure shows an algorithm for generating a digital seabed relief model based on three groups of data: optical data from UAVs, hydroacoustic data from underwater gliders, and retrospective data from archives.

[0022] 1 - Optical data obtained by UAV.

[0023] The UAV acquires data on the topography of shallow waters, coastal areas, and areas with sufficient water transparency. This data can be obtained from a bathymetric lidar, an optical camera, or another remote sensor.

[0024] 2 - Hydroacoustic data obtained by underwater gliders.

[0025] Underwater gliders generate data on depth, bottom profile, bottom objects, and irregularities using a multibeam echosounder, side-scan sonar, profiler, and altimeter. This data is especially important in deep or turbid waters where optical instruments are limited.

[0026] 3 - Retrospective data from archives.

[0027] The system receives previously acquired bathymetric maps, archival surveys, digital seabed models, and other archive materials. These are used as an additional source for comparison and identification of relief changes.

[0028] 4 - Block of combination and analysis of discrepancies.

[0029] Data from various sources is converted to a single coordinate system. The system then compares optical, hydroacoustic, and historical data. This identifies discrepancies: depth changes, contour shifts, the appearance of new objects, data gaps, and areas of varying reliability.

[0030] 5 - Map of discrepancies and priorities.

[0031] Based on the analysis, a map is generated, assigning a verification priority to each area. High priority is assigned to areas with significant discrepancies between sources, low reliability, or signs of new objects. Medium priority is assigned to areas with moderate discrepancies, and low priority is assigned to areas with consistently consistent data.

[0032] 6 - Visual verification planning block.

[0033] This block determines which areas require additional surveys, the best means to do so, and the order in which they are conducted. It takes into account the area's priority, accessibility, depth, water clarity, and UAV and underwater glider limitations.

[0034] 7 - Assigning tasks to UAVs.

[0035] UAV missions are created for areas suitable for aerial or optical surveys. These missions can include the area's coordinates, altitude, route, required resolution, survey mode, and repeat observation conditions.

[0036] 8 - Visual survey of high priority areas.

[0037] The UAV is surveying areas where discrepancies or insufficient reliability were previously identified. The obtained data is used to confirm or clarify disputed terrain features.

[0038] 9 - Update digital elevation model and reliability map.

[0039] After verification, the system updates the seafloor digital elevation model, the confidence map, and the priority map. Areas where discrepancies are resolved are given a higher confidence level. If discrepancies persist, they can be re-sent for survey by another sensor channel.

[0040] The cause-and-effect relationship lies in the fact that calculating partial confidence indicators prevents the equal use of data of different quality. Comparing sources by depth, contours, acoustic texture, and morphology reveals discrepancies that cannot be detected by individual measurements alone. Classifying the cause of the discrepancy allows for the selection of the correct verification method: use a hydroacoustic channel in case of turbid water, change the perspective in case of an acoustic shadow, perform a re-alignment in case of a navigational shift, and assign visual or high-frequency verification in case of a probable object. The cell passport links the final model to the origin and confidence level of the data. This reduces the number of false model elements, increases the reliability of change detection, and reduces the amount of manual verification.

[0041] The method is industrially applicable in hydrographic services, in the design and operation of ports, oil and gas and energy infrastructure, cable lines, pipelines, bridges, offshore facilities, fisheries, and environmental areas. Implementation is possible on a shore-based computing system, aboard a UAV, an autonomous surface vehicle, or an underwater glider, provided standard sensor data and navigational reference are available.

[0042] Implementation example

[0043] During a submarine cable route survey, the system receives data from a multibeam echosounder and side-scan sonar and uses a two-year-old historical depth model. In one section, the multibeam echosounder shows a local depression, the sonar records an acoustic shadow, and the historical model does not contain any depression. The system calculates the discrepancy between depth and morphology, classifies the section as a "bottom object or washout feature requiring verification," and schedules a repeat glider pass with a different perspective and high-frequency sonar. After the repeat pass, the object boundary is confirmed, but the optical survey is given low confidence due to water turbidity. The cell passport is updated with the confirmed source—the high-frequency hydroacoustic verification—a confidence interval, and a flag indicating its suitability for cable monitoring.In another section, the discrepancy turned out to be a navigational shift; after alignment using stable bottom reference points, the data is included in the model without a physical re-pass.

Claims

A method for generating a digital seabed relief model, including obtaining data on the depths and relief of the seabed from optical and hydroacoustic sensor channels, reducing the said data to a single coordinate system and constructing a digital seabed relief model, characterized in that for each cell or local section of the digital model at least two sets of data are obtained, selected from a group including data from a multibeam echo sounder, a side-scan sonar, a profiler, a laser bathymeter, an optical camera and a retrospective bathymetric model, for each set of data a particular confidence indicator is calculated taking into account the quality of the navigation reference, the level of interference, the overlap of the survey, the angle of observation or sounding, the stability of marks and consistency with adjacent sections, then within the corresponding cell the current sets of data are compared with each other and with the retrospective data and the discrepancies in depths, contours of bottom objects are determined,acoustic texture, optical and acoustic features and relief morphology, classify the cause of the detected discrepancy as one of the conditions including a real change in relief, navigational offset, acoustic shadow, multi-reflection, optical opacity, insufficient resolution, interpolation error, bottom object or unconfirmed anomaly, and, based on the classified cause, assign a verification method, wherein in the case of a possible bottom object, visual verification by an optical channel or high-frequency hydroacoustic refinement is assigned, in the case of navigational offset, a repeated alignment using stable bottom landmarks is assigned, in the case of optical opacity, verification by a hydroacoustic channel is assigned, in the case of an acoustic shadow, a repeated pass with a changed angle or sounding frequency is assigned, and for the other specified conditions, the verification method is selected from the group including a repeated pass from a different angle, survey by a different sensor channel, visual verification,high-frequency hydroacoustic refinement, repeated alignment using bottom landmarks, or exclusion of a section from automatic confirmation; after the check is completed, a passport is generated for each cell containing the confirmed depth, the source of confirmation, the confidence interval, the type of the detected discrepancy, and the admissibility indicator for using the cell; the cell is included in the final digital seabed relief model when the data is confirmed by the result of the assigned check and the specified confidence level is reached; if the specified level is not reached, an additional check task is generated or the cell is marked as controversial.

Images