Toward interoperable bathymetric data in ocean observing systems: integrating hydrographic standards and FAIR data practices

1 Introduction

From a hydrographic point of view bathymetry is the science of measuring and mapping the seafloor topography from a particular vertical datum and is the backbone of all maritime activities and oceanographic observations. Bathymetry is not listed as an Essential Ocean Variable (EOV) in the Global Ocean Observing System framework. It is better understood as the geodetic reference layer against which EOVs are located, interpreted and modelled, and its status as infrastructure rather than as a variable is precisely why its interoperability failures propagate into every EOV that depends upon it. Accurate seafloor-depth data is crucial for determining the physical response of the water column and distribution of marine life (Wölfl et al., 2019). Bathymetric data is essential for numerical modelling and ocean circulation studies since both the roughness of the seabed and the depth play an important role on the heat transport and ocean current paths (Levin et al., 2019). In terms of societal safety, it is the backbone of tsunami and storm surge forecasting, and inundation models are very sensitive to coastal bathymetry, with small topographic features that can affect the dissipation of wave energy (Lecours et al., 2016). Moreover, bathymetry plays a crucial role in coastal monitoring for erosion, flood-risk mapping and the sustainable implementation of Marine Spatial Planning (MSP), which is a tool for balancing competing interests for the exploitation of offshore renewable energy, the planning of submarine infrastructure (e.g. telecommunications cables), and the protection of marine biodiversity (Murawski et al., 2025). It is also important for habitat and ecosystem mapping, and will be used to generate the geomorphometric metrics (e.g. rugosity, slope) used to inform predictions of benthic species occurrence and coral reef health (Hasara et al., 2025). Lastly, the traditional function of bathymetry is still very relevant: to enable safe navigation of the global shipping fleet, which increasingly depends on the availability of high-definition surfaces for fuel-saving and autonomous navigation (Sikirin and Golodov, 2026). Bathymetry is a unique spatial baseline layer in the multi-disciplinary architecture of modern ocean observing systems. It serves as a static geodetic reference frame over which physical, biological, geological and chemical observations are superimposed to obtain a comprehensive “Digital Twin of the Ocean” (DTO) (Butkiewicz et al., 2022). Bathymetry is not only crucial for certain variables, it is also closely linked to other variables in a symbiotic relationship such as the models of chemical flux at the sediment-water interface and those used to estimate biological productivity in the photic zone, which are both explicitly depth dependent (Lee and Kim, 2024). Therefore, bathymetric surfaces are one of the main decision-supporting layers for climate adaptation measures and resource management, but can be severely hindered by data fragmentation. At present, bathymetric data is spread in a silo environment consisting of national hydrographic offices, academic research institutes, defence agencies, and private corporations such as energy and cable companies (Thierry et al., 2019). However, there have been advances in aggregation and coordination in international repositories such as the International Hydrographic Organization (IHO) Data Centre for Digital Bathymetry (DCDB) and remote-sensing platforms like ICESat-2 (Bellabad et al., 2025). Institutional requirements (e.g., preservation of navigational safety for hydrographic offices versus commercial protection by private companies) often lead to “data-sharing policies” that limit access to data and make the development of a truly global, high-resolution seafloor model difficult (Schmitt et al., 2025). This fragmentation is exacerbated by several major sources of technical and semantic inconsistency. Geodetic barriers include differences in horizontal coordinate-reference systems and, more importantly, vertical datum incompatibility. Horizontal incompatibility is often attributed to differences between ITRS and WGS84, but this comparison is not well formed: ITRS is a reference system, realised by successive ITRF solutions, whereas WGS84 is a datum with its own sequence of realisations, G730 through G2296, each aligned to the contemporaneous ITRF at the centimetre level. At the accuracy of hydrographic survey, modern realisations of the two are not meaningfully different. What does produce incompatibility is which frame realisation was used, at which coordinate epoch the coordinates are expressed, and whether coordinates have been propagated for plate motion between the epoch of survey and the epoch of use. On a plate moving 3 to 5 cm per year, a survey reused ten years after acquisition without epoch propagation carries a horizontal offset of 0.3 to 0.5 m: significant for Special Order coastal work and for merging repeat surveys across decades, and irrelevant at the 15 arc-second posting of a global compilation grid, some 450 m at the equator. In practice, hydrographic data is usually referenced to the Lowest Astronomical Tide (LAT) to ensure the highest safety limits for shipping purposes, whereas certain scientific models may need Mean Sea Level (MSL) or geoid-based references (Drennon et al., 2023). These surfaces are not necessarily easily reconcilable, and often they are not reconcilable without high-quality metadata. In addition, the spatial resolution and the encoding structures vary and the data also come in several different formats from NetCDF to XYZ, which results in a “patchwork” of data that demands extensive manual cleaning (Levin et al., 2019). Many datasets are missing key information about uncertainty and provenance, and quality-control procedures are not necessarily equal among institutions, precluding users from determining the suitability of the data for particular uses (Haklay et al., 2018). There are also semantic barriers, stemming from the lack of controlled vocabularies and the use of standardized metadata, such as ISO 19115/19139, to enable automated systems to find and combine datasets from different sources (Lee et al., 2022). Further, there are a number of different licensing and access requirements, offered from open to highly proprietary, which restrict the reusability of high-resolution data (Murawski et al., 2025). Hydrographic standardization and FAIR (Findable, Accessible, Interoperable, Reusable) data management are two closely related areas that are poorly explored in terms of research (Dongli et al., 2025). Although hydrographic standards like IHO S-100 and S-102 greatly enhance the quality of measurements and the structure of the data for navigation, this does not automatically mean they comply with the FAIR principles for wider scientific use (Duan et al., 2021). On the other hand, use of FAIR principles facilitates the discovery and re-use of ocean data, but are not seen to meet the stringent requirements for hydrographic accuracy, sensor calibration and vertical-reference transformations required for safety-critical applications. Current literature focuses on these two aspects of the system, namely on the standardization of the hydrographic data and on the FAIR data management, without offering much guidance about how these can be implemented throughout the bathymetric data’s life cycle (Duan et al., 2021). This review aims to explore the potential for the coupling of hydrographic standards and FAIR data practices to enhance the interoperability, accessibility, and re-use of bathymetric data within the broader ocean-observing context, to meet these challenges. In this review, ocean observing system is used as the broader context within which bathymetric observations and derived data products are generated, integrated, distributed, and used. Consistent with the IOC-UNESCO framework, an ocean observing system encompasses the broader value chain from observing requirements and system design through observations, integration and modelling, data-product generation, distribution, and subsequent use. The present review does not assess this complete value chain; rather, it focuses specifically on the bathymetric observation and data workflow within this broader ocean-observing context. This includes bathymetric acquisition platforms and technologies, together with subsequent processing, quality control, spatial harmonization, standardization, metadata, publication, and reuse of bathymetric data. Thus, observational platforms refers primarily to the means through which bathymetric observations are acquired, whereas ocean observing system denotes the broader context in which these observations and derived products are integrated and used. This study aims at following objectives: review the main sources and characteristics of bathymetric data; identify technical, semantic, geospatial, legal, and institutional interoperability barriers; examine the contribution of hydrographic and geospatial standards to bathymetric interoperability; identify the FAIR deficiencies recurrently reported in bathymetric-data workflows and the mechanisms available to address them; develop an element-level crosswalk between the clauses of hydrographic and geospatial standards and the fifteen FAIR sub-principles, classifying each mapping by its evidential basis; propose an integrated interoperability framework specifying the components, their interfaces, the data flow between them, and a staged implementation pathway. This review aims to answer the following research questions: What are the specific factors that constrain the interoperability of bathymetric data within the broader ocean-observing context? How are existing hydrographic and geospatial standards helping in interoperability data management? What is the current status of FAIRness in bathymetric workflows? What are the gaps in interoperability where standards have been defined? Last but not least, how can hydrographic standards and FAIR practices be combined in an overall operational framework.

2 Review methodology

This article presents a structured scoping review, following PRISMA-ScR conventions, without quantitative synthesis or risk-of-bias appraisal. The review comprises two components: a bibliographic search of peer-reviewed and conference literature, and a normative component covering standards. The bibliographic search was conducted in Scopus, Web of Science, IEEE Xplore, and ScienceDirect, covering publications from January 2000 to July 2026, with no lower bound for foundational works. Searches were executed on 20 July 2026. The search string combined terms for bathymetry/hydrography with interoperability, standards, metadata, and data management concepts. Reference lists and grey literature were also screened. Records were eligible if they addressed acquisition, processing, standardisation, quality, publication, or reuse of bathymetric data, and were in English. Exclusions included studies not addressing data standardisation/reuse, inland surveys, duplicates, or unavailable full texts. Screening was performed by the author alone, acknowledged as a limitation. After duplicate removal, title/abstract screening, and full-text assessment, a final set of records was retained, supplemented by normative documents and repository sources. Evidence was synthesised narratively around the data lifecycle, examining standards, interoperability barriers, and FAIR principles. Normative documents were obtained directly from issuing bodies including IHO, ISO, OGC, and others, consulting editions in force on 20 July 2026 (International Hydrographic Organization, 2024a, 2024b, 2024c, 2024d; International Organization for Standardization, 2014, 2023a, 2023b; Open Geospatial Consortium, 2023, 2024). Revisions during preparation are noted. Statements on standard requirements are referenced normatively, while interpretations from secondary literature are explicitly identified. All mappings in tables are labelled as normative, evidence-based, or interpretive.

Retained sources were evaluated against three criteria rather than by formal risk-of-bias appraisal, which is not applicable to a corpus combining normative documents with heterogeneous empirical and technical literature. First, provenance: normative statements about a standard were admitted only from the standard itself, with secondary literature used for interpretation and implementation evidence and identified as such. Second, currency: where a source described a specification that had since been superseded, its claims were checked against the edition in force and retained only where still applicable. Third, specificity: sources were retained where they addressed bathymetric or hydrographic data directly, or where a general data-management finding was demonstrably transferable to the bathymetric case, with the transfer stated rather than assumed.

3 Bathymetric data in ocean observing systems

In the context of the changing ocean observing system, bathymetric information is the basic geodetic information that underlies all other ocean observations. Technically, it is important to differentiate between bathymetry and hydrography, with the latter being the mapping of all the physical features of the water body and the seabed – navigational hazards, seabed type, and coastal structures (Dongli et al., 2025).

Current hydrographic practice, adhering to the International Hydrographic Organization (IHO) standards S-44 and S-100 universal data model, goes beyond the mere recording of depth and incorporates “hydrospatial” information, including bathymetric data and other layers of spatial and temporal information. This data can be acquired in a variety of forms along the entire processing maturity of the data, from raw acoustic observations (unprocessed sensor pings) to corrected soundings that are adjusted for environmental factors such as sound velocity and tidal offsets, and to gridded surfaces, such as Bathymetric Attributed Grids (BAGs) in IHO S-102 specifications, or to derived terrain products like slope, aspect, rugosity layers for morphological analysis (Nadzir and Munthe, 2025). Typically, these products are classified after the intended use, with the distinction between navigation, research, and model-ready bathymetry. Navigation-grade products are those products that adhere to IHO Hydrographic Surveys standards with a focus on safety of transit (and legal certainty in nautical charting) and features of high-order detection with low level of uncertainty. Research-grade datasets, however, usually focus on high-resolution detail or scientific accuracy for specific environmental studies, and in some cases may be based on opportunistic platforms or unmanned surface vehicles (USVs) or crowd-sourced initiatives that supplement data gaps in waters too deep for survey vessels to access. (Solana Rubio et al., 2023) Model-ready bathymetry is a specific subset that is prepared in a spatial harmonization and interpolation way, so that it can be used for hydrodynamic and climate modelling, which demands consistency and continuity of bathymetry over a larger area than the bathymetric surveys for that area. The products are crucial for planetary-scale models such as GEBCO, where multiple survey sources need to be seamlessly combined (Armstrong and Mayer, 2000). Bathymetry’s contribution to ocean observing workflows is multi-fold: it is used to position the autonomous sensors and benthic platforms that are used for water-column observations to ensure that the observations are correctly placed within the broader topography. It is a key parameter in hydrodynamic models that influence tidal propagation, sediment transport and current flow on the seabed. The physical morphology of the land and sea surface derived from multibeam backscatter and bathymetry is one of the major considerations for biological diversity and is the basis for habitat classification and delineating marine protected areas (Nadzir and Munthe, 2025). In addition, high-accuracy bathymetric data is essential for coastal-risk assessment, to model inundation from tsunamis and storm surge with high accuracy, and to monitor sea-bed changes in vulnerable regions where rapid erosion is likely to occur. These Digital Terrain Models are also essential to ensure sustainable marine resource development and to prevent damage to subsea cables for effective marine resource management and navigation planning in growing maritime congestion (Contarinis et al., 2022). A strong lifecycle strategy is needed for this data to remain accessible to various stakeholders, from the design of the survey to the acquisition and careful calibration of sonar systems. The processing stage includes quality control (QC) and the use of datum transformations to make the tide data consistent vertically between low and high tidal regimes (Figure 1). This lifecycle is followed by spatial harmonization and standardization (by adopting the IHO S-102 or S-101 specifications), and comprehensive metadata generation and open access publication. The important point is the detailed documentation of the entire lifecycle, which is crucial for the integration into global repositories. Survey design, equipment configuration, offsets, and processing parameters are important aspects of the survey record, and are not just administration, but are essential for following FAIR (Findable, Accessible, Interoperable, Reusable) data practices. This transparency facilitates a version history and future reprocessing of raw data as computational techniques develop, maximizing the scientific value and ensuring that bathymetric assets can be reused effectively both within the technical and disciplinary framework (Alexander et al., 2023). The term hydrospatial denotes the extension of hydrographic practice beyond the depiction of depth for safety of navigation to the integrated management of all spatially and temporally referenced marine information, including the water column, the seabed and the dynamic surfaces that reduce them, held to a common reference framework and intended for multiple uses rather than for charting alone. Bathymetry denotes the measurement of water depth and the resulting depiction of seabed relief. Hydrography is the broader discipline concerned with the measurement and description of all physical features of the water body and seabed affecting navigation and marine activity, of which bathymetry is one component. A bathymetric surface is a continuous representation of the seabed derived from soundings, whether gridded or triangulated, as distinct from the soundings themselves and from the physical seabed it represents. A digital bathymetric model is a bathymetric surface held as a regular or variable-resolution grid of depth values referenced to a stated vertical datum. A digital terrain model is the more general case, covering land and seabed elevation without restriction to the submarine domain, and is used here only where topobathymetric continuity across the shoreline is at issue.

4 Sources and characteristics of bathymetric data

4.1 Ship-based hydrographic survey systems

Multibeam echo sounders (MBES) are the foundation of modern shipbased hydrography, and they are capable of generating high resolution, full coverage bathymetric surveys necessary for detailed seafloor characterization (Pasandeh et al., 2025). Unlike older technologies, MBES systems map over a broad swath of acoustic beams and generate large, areal maps of the sea bed in a single pass, with beam angles as narrow as 1° (Murawski et al., 2025). One of the major strengths of MBES is its ability to capture both depth (bathymetry) and backscatter intensity. Backscatter data is useful in determining the composition and/or roughness of the seafloor that is important for benthic habitat characterization and sedimentary processes (Gavande et al., 2015). The use of MBES is however expensive due to the requirement of specialised survey vessels, costly equipment and highly trained staff (Khomsin and Saputro, 2021). Moreover, the accuracy of MBES depends on the accurate integration of auxiliary sensors. This includes high frequency sound-speed sensors for correcting for acoustic refraction through the water column, inertial motion units (IMUs) for correcting for vessel heave, pitch, roll and yaw, and high precision GPS/GNSS for tidal and ellipsoidal corrections (Ai et al., 2016). If appropriate, environmental corrections and rigorous calibration are not performed, the resulting bathymetric surfaces may contain significant artifacts. Single-beam echo sounders (SBES), on the other hand, are much less expensive and easier to use for the measurement of depth. SBES systems are commonly used in small-scale surveys, in shallow inland waters and for providing baseline data in the past (Eke and Havasi, 2022). They are easy to operate with fewer auxiliary sensors needed than MBES. The main drawback of SBES is its limited spatial extent. The system only measures the depth directly below the transducer (the nadir) and that means there are large areas between the survey lines that are not covered. Such “line-based” data requires complex interpolation techniques to produce continuous 3D models, and sometimes uncertainty exists about the small scale seafloor features between the lines (Gavande et al., 2015). Despite these disadvantages, SBES is an important method for calibrating other acoustic sensors and for surveying in sites where the high cost and draft of an MBES-equipped vessel is impractical (Fezzani et al., 2025). Finally, side-scan and interferometric systems provide specialized capabilities for seabed imaging and shallow-water charting. Side-scan sonar (SSS) is the most important tool for high resolution imaging of the seabed and is used to provide a 2D image of the seabed that is best suited for the detection of objects and geological mapping (Akerele et al., 2025). Although SSS can offer excellent detail on seafloor texture and man-made features, it is not as appropriate for direct, quantitative depth measurements as MBES is. On the other hand, interferometric bathymetric sonars (also phase-differencing bathymetric sonars) are becoming more common in shallow water applications (between 1.5 and 50 m). Unlike conventional multibeam systems with a comparatively low swath width, these systems can cover a width extending up to 12 times the sensor altitude and yet deliver high resolution imagery. Phase-differencing bathymetric sonars can meet International Hydrographic Organization (IHO) Special Order requirements under favourable conditions, but this is conditional rather than a property of the sensor class. The constraints are the nadir gap, where near-vertical returns lack the phase differencing on which the measurement depends; degradation of phase coherence at long range and in the presence of multipath or a rough seabed; and consequent dependence on line planning that secures nadir coverage from adjacent lines. Compliance is therefore demonstrated per survey against S-44 Table 1 rather than asserted in general (Ai et al., 2016). These ship-based systems collectively comprise a multi-tiered set of tools to balance the ocean-observing system’s requirements for resolution, coverage, and operational efficiency.

Table 1. Comparison of bathymetric data sources.

4.2 Autonomous and uncrewed platforms

The use of autonomous and uncrewed platforms has revolutionized the way bathymetric data is acquired in modern ocean observing systems. High-resolution mapping can be achieved using Autonomous Underwater Vehicles (AUVs) and Uncrewed Surface Vessels (USVs) which decrease the operational overhead and human risk involved in traditional ship-borne mapping (Bush et al., 2016). USVs are primarily used in shallow water environments like estuaries and areas where large vessels cannot operate, whereas AUVs are specifically designed for deep-sea research. AUV-mounted multibeam echosounders (MBES) can provide centimetre-scale resolution, with a much higher resolution than surface-mounted instruments (Shi and Zhou, 2020). In addition, gliders and Remotely Operated Vehicles (ROVs) are used as complements to these systems; gliders enable long-term, wide area monitoring and ROVs can be used to visually verify and to sample specific seafloor features, such as hydrothermal vents (Thompson et al., 2012). These platforms are especially beneficial when using them in dangerous, off-shore, and deep-water settings. Uncrewed systems can operate in locations that are high in ships’ traffic or hazardous weather conditions and collect data without putting crew at risk (Specht, 2023). Long-range AUVs allow for “ship-free” benthic surveys where the support vessel can move to another location and perform additional tasks while the AUV remains deployed on-site. This modularity can reduce the carbon footprint of marine research, and make bathymetric gaps more economically viable to close worldwide (Bodenmann et al., 2023). Integrated systems, such as USVs with aerial photogrammetry, offer the capability to rapidly model dynamic seabed topography, with critical data for navigation safety and coastal management. Despite these benefits, autonomous bathymetry provides significant technical challenges to navigation accuracy, sensor calibration and data transmission. One of the major challenges is the need for precise positioning on the sea floor, frequently achieved using Inertial Navigation Systems (INS) with Doppler Velocity Logs (DVL) and acoustic positioning networks (LBL or USBL) (Klischies et al., 2018). In order to reduce the accumulation of trajectory errors, more sophisticated methods like Bathymetric Simultaneous Localization and Mapping (BSLAM) and feature matching are being used to align point cloud data (Yang et al., 2025). Moreover, low-bandwidth communications in extreme environments or in deep sea requires advanced onboard autonomy and path planning capabilities that enable autonomous vehicles to make real-time decisions about survey coverage and mission integrity (Macesker et al., 2023). Standardization of data gathered by autonomous platforms is vital for the interoperability and usefulness of its data in wider ocean observing systems. Current efforts focus on aligning autonomous data streams with international standards, such as the International Hydrographic Organization (IHO) S-44 for survey quality and the S-100 Universal Hydrographic Data Model (Fajrin et al., 2025). It is essential to have strong metadata protocols with documentation of calibration records, processing parameters, and datum transformations to follow FAIR (Findable, Accessible, Interoperable, and Reusable) data practices. The development of these technologies will be crucial to the adoption of standardized, and automated processes, which will be essential to the integration of autonomous bathymetric products into global hydrographic and scientific repositories (Ma et al., 2019).

4.3 Remote-sensing methods

Satellite-derived bathymetry (SDB) has become a revolutionary remote-sensing technique for obtaining shallow-water coverage across broad ocean regions where acoustic bathymetry is logistically or economically unfeasible (Bernardis et al., 2023). SDB is a passive sensing method that uses multispectral imagery from platforms like Sentinel-2 and Landsat to determine water depth based on the spectral radiance reflected from the seafloor (Jawak et al., 2015). This method can be especially useful for monitoring areas over large scales, such as coral reefs and coastal ecosystems. But the accuracy of SDB is very sensitive to environmental conditions such as water turbidity and characteristics of the bottom reflectivity. In clear waters, SDB is more likely to reach a maximum depth penetration of around 20 to 30 meters, due to the light being rapidly absorbed and scattered by water molecules and suspended particles in the water. In order to achieve hydrographic reliability, SDB workflows may need to be carefully calibrated, based on in-situ depth measurements or high accuracy spaceborne LiDAR data like the photon-counting laser altimetry from NASA’s ICESat-2 mission. The airborne bathymetric LiDAR (ALB) is the foremost active remote-sensing technology available for high-resolution mapping of the land-water transition zone (Xie et al., 2025). ALB systems employ green-spectrum lasers that are able to penetrate the water column and measure the seafloor directly, with vertical accuracy comparable to that measured by ship-based survey systems in optimal conditions. This technique is particularly effective in the shallow depths of the nearshore environment where safe operation of boat mounted multibeam systems is not possible because of turbulent surf conditions. Although ALB is a very high-resolution sensor, water clarity is a significant limitation to its use, with the maximum measurement depth typically being 2–3 times the Secchi depth (Murawski et al., 2025). Additionally, the operational expenses of ALB (mobilizing aircraft and maintaining specialized sensors) are significantly more costly than those associated with satellite-based techniques, and it may only be used for specific navigational projects or for targeted, high priority coastal hazard assessments. Unlike these coastal methods, satellite altimetry-derived bathymetry is the only viable option for global coverage in the deep ocean. This technique is not direct measurement of depth, but rather relies on the radar altimeter to measure the height of the sea surface, which reflects variations in the gravitational pull of underlying seafloor features such as seamounts and trenches (Wölfl et al., 2019). Altimetry-derived products, including those incorporated into the GEBCO and Seabed 2030 grids, are vital for the regional interpretation and understanding of the global tectonic processes, but can only be used at coarse spatial resolution, typically on a few kilometres scale. A distinction must be drawn here between posting, the sample interval of a grid, and resolution, the shortest wavelength the data can actually resolve. Altimetry-derived bathymetry resolves seafloor features at wavelengths of roughly 12 to 16 km, notwithstanding that the GEBCO grid is posted at 15 arc-seconds, approximately 450 m at the equator; the two figures describe different quantities and should not be used interchangeably. Resolution in this sense is far coarser than the metre-scale detail obtained by acoustic multibeam echo sounding, and maps produced from altimetry are accordingly unsuitable for local navigation or for locating small-scale hazards. However, these models give a crucial overview of the ocean border and can be used as a benchmark to determine focus points for future high-resolution exploration (Taak et al., 2025). SDB, ALB and satellite altimetry constitute a multi-layered remote-sensing hierarchy in global ocean observing systems, which represents a balance between the trade-offs of spatial resolution, environmental sensitivity and geographical scale (Melet et al., 2020).

4.4 Crowdsourced bathymetry

Crowdsourced bathymetry (CSB) is fundamentally a paradigm shift in ocean mapping, in which the bathymetric data are collected from many unspecialized vessels, such as merchant ships, fishing boats, and yachts, during their normal operation as a marine craft (Grinker et al., 2022). These boats typically carry standard navigation echo sounders to meet the International Convention for the Safety of Life at Sea (SOLAS) regulations or as best navigational practice, which results in a large network of potential sensors that are rarely used (Keating, 2023). To get this data, ships are fitted with NMEA0183 and NMEA2000 data loggers, and these are now increasingly being added to low-cost wireless systems, which can combine the depth data for processing in the cloud (Calder, 2023). This method enables taking depth soundings from vessels as they travel along coastal or open waters without the expensive cost of a dedicated hydrographic survey mission. CSB is strategically important because it can help to address some of the gaps in the global seafloor map that could not be filled by other methods. The GEBCO_2026 release (GEBCO Compilation Group, 2026) reports that 28.7% of the ocean floor is represented by measured bathymetry, against 27.3% in the 2025 release, corresponding to approximately 104 million square kilometres; the figure is revised annually and should be read as a dated snapshot rather than a constant. It also requires care in interpretation. Mapped, in the Seabed 2030 sense, is a depth-banded cell-occupancy criterion, a cell counting as mapped if at least one sounding falls within it, at cell sizes ranging from 100 m in water shallower than 1500 m to 800 m below 5750 m. This is a statement about the presence of a measurement, not about its quality. Surveyed to modern hydrographic standards is a different statement, referring to compliance with the total vertical and horizontal uncertainty limits, feature detection and feature search requirements, and bathymetric coverage specifications of IHO S-44. A cell may be mapped in the first sense while containing no sounding that would satisfy any S-44 order. The distance between these two statements, one about coverage and one about fitness for purpose, is the interoperability problem this review addresses (Gilbert et al., 2025). These data are used in international collaborative projects to progress toward a global bathymetric model, specifically the Nippon Foundation-GEBCO Seabed 2030 project and the International Hydrographic Organization (IHO) CSB project. Using the “crowd”, researchers can create more detailed regional maps; for example, 75m grids for the Newfoundland and Labrador shelves offer much finer resolution than the coarse datasets available previously. The usefulness of CSB is frequently limited by the issue of data reliability, due to the lack of control in the collection process, compared with a professional hydrographic survey (Wang et al., 2022). A major technical challenge is the lack of a standard procedure for calibrating the sensors which can cause errors because of unknown draft changes or sensor transducer offset. The data gathered are also frequently missing the required sound velocity profiles and/or the correct vertical datum information, which is essential for correct vertical positioning (Grinker et al., 2022). The opportunistic data collection also results in non-uniform spatial distribution of data, making the detection of outliers and development of repeatable bathymetric grids difficult. New workflows are adopting automated pipelines that ensure the data is processed to meet the FAIR (findable, accessible, interoperable and reusable) principles to transform the raw (Debroisse and Miles, 2024), disparate CSB data into high quality products. In recent years, advanced. computational methods have been developed to identify outliers and combine the repetition soundings, such as N-order interpolated variance estimators and intelligently optimized 3D-DBSCAN algorithms (Wang et al., 2022). Moreover, tools such as “VBI Compare” simplify the process of comparing volunteer data with trustworthy sources to ensure information quality. These methods, which involve calculating vertical uncertainty per sounding, and cross-platform validation, enable the maritime industry to increase confidence in CSB for use in nautical charting, survey planning and environmental modelling (Muthusamy et al., 2025).

4.5 Legacy and historical datasets

Legacy and historical bathymetric datasets are an important legacy that document the seafloor condition before the advent of modern digital hydrography. Most of these datasets consist of analog records like paper nautical charts, historical survey sheets (or “smooth sheets”), and primary lead line soundings, and are the only quantitative data on seafloor topography in many coastal areas in the past two centuries (Lira and Taborda, 2024). These records are normally represented digitally by a careful digitization of the legacy contours or by the collection of a series of points of soundings from early single-beam echo sounders and mechanical measurements. In some cases, prior acoustic surveys, such as some of the earlier multibeam surveys from the late 20th century, are also built into these historical archives. These early electronic data sets are more technologically advanced than the manual soundings they replace, but they often need extensive reprocessing to meet current technology standards for precision and spatial density (Zimmermann and Prescott, 2015). These legacy datasets are important because they can serve as longterm baseline environmental data that could not be gathered with modern surveys. Historical bathymetry can offer a centennial-scale context for analysis of long-term change in the seabed, and the morphodynamics of the coast. The data are critical for quantifying the cumulative impacts of human activities such as dredging, land reclamation and large-scale construction of coastal infrastructure on the estuarine volume and tidal dynamics (Li et al., 2023). Furthermore, they offer the potential to predict the impact of climate-driven sea-level rise and altered sediment fluxes on sensitive systems, on decadal to centennial time scales, such as ebb-tidal deltas and coral reefs. If there are no historical benchmarks, the results of hydrodynamic models and decision-support systems are not placed in the appropriate context and cannot distinguish between natural topographic variability and human-made changes in morphology (Yates et al., 2017). However, technical problems and inconsistencies exist in the assimilation of legacy bathymetry into contemporary ocean observing systems. A major issue is the lack or the absence of full or missing metadata, which may be incompatible with current standards such as ISO 19115. This lack of documentation can obscure important details regarding the methodology used in the survey, instrument calibration, and original policies of data sharing. Furthermore, the accuracy of the legacy data is highly variable because of the accuracy of the original lead-line measurements, and the potential for inaccurate manual recording and transcribing. A further difficulty with data reuse is that previous surveys were occasionally quoted in an unknown horizontal datum, also that vertical surveys were computed from local tidal observations rather than from some common global geoid or Mean Sea Level (MSL) (Zimmermann and Prescott, 2015). These data must be georeferenced precisely and then undergo complex datum transformations in order to be successfully incorporated into modern Digital Bathymetric Models (DBMs) based on recent surveying using GNSS (global navigation satellite system). This typically takes place by aligning manually different datasets and eliminating outdated or incorrect soundings by hand-checking original analog documents. The preservation and digitisation of legacy bathymetric data, however, remain an important priority for ocean observing programmes, despite these intensive processing needs. They offer the essential minimum standard for enhancing the predictive power of flood-risk assessments, and for embedding contemporary coastal management in a sound knowledge of how the seafloor has evolved over time (U.S. Geological Survey, 2014).

4.6 Data-source selection and fitness for purpose

The choice of bathymetric data sources is determined by a “fitness for purpose” assessment, in which spatial resolution, vertical uncertainty, geographic coverage, and operational cost are considered in a trade-off. The requirements are strictest for safety-critical applications, e.g. navigation, generally following IHO S-44 standards. High-accuracy, full-bottom search is required for Special Order and Order 1a surveys, and this is primarily requires the use of ship-based multibeam echo sounders (MBES) (Mavraeidopoulos et al., 2017). Satellite derived bathymetry (SDB) is becoming more and more considered as an alternate, low-cost method for shallow-water bathymetry in remote locations, but it is still dependent on strict validation against these acoustic standards to ensure that against-ship safety margins are met (Bernardis et al., 2023). In contrast, regional oceanographic and hydrodynamic modelling often prioritizes broad spatial coverage over centimetre-scale precision. Global gridded products such as GEBCO or EMODnet are often used to define the bathymetric “envelope” for large-scale simulations (Tükenmez et al., 2023). These products can, however, have systematic error or “random bumpiness” on scales under 10 km, affecting predictions of tidal energy flux and bottom friction (Marchesiello et al., 2024). Thus, in complex coastal areas, a multisource approach is usually needed to combine these regional grids with higher-resolution nautical charts or SDB to create resolution in the nearshore area (Andréfouët et al., 2025). Studies of coastal hazards are another middle ground, in which high-resolution nearshore data is critical for resolving the topographic features that control wave dissipation, and include tsunami inundation and storm-surge forecasting. Airborne LiDAR and SDB are especially well suited for this purpose because they offer both rapid and wide area coverage of the land-water transition zone which is often logistically impractical or dangerous for traditional ship-based surveys. Likewise, mapping of benthic habitat using MBES is dependent on the high-resolution geomorphometry and backscatter intensity to accurately classify seafloor substrates and coral reef structures (Murawski et al., 2025). Derivation of secondary metrics such as rugosity and slope play a significant role in source selection for these applications. Finally, the choice of marine spatial planning and scientific reconstruction includes resolving different, multi-platform datasets to manage resources and set historical baseline. In such scenarios, it is crucial to explicitly quantify the uncertainty associated with each data source for sound decision making. In emergency response, like post-disaster assessment or search and recovery, the speed and repeatability of remote-sensing technologies such as SDB or the nature of crowd-sourced bathymetry (CSB) can surpass the precision of slower and more costly acoustic campaigns (Yunus et al., 2019). Finally, an interoperable ocean observing system must facilitate dynamic discovery of these diverse sources, enabling users to choose the data most useful to their particular scientific or operational goals (Vrdoljak, 2021). The comparison of bathymetric data sources is summarized in (Table 1).

5 Spatial processing and harmonization of bathymetric data

5.1 Georeferencing and coordinate-reference systems

Georeferencing provides an essential spatial basis for linking bathymetric datasets within the broader ocean-observing context. The fundamental concept behind this is the difference between geographic and projected coordinate reference systems (CRS). Geographic coordinate systems (GCS) are used to represent the Earth’s surface using angular units such as latitude and longitude and a reference ellipsoid (Wölfl et al., 2019). A reference ellipsoid is a geometric figure of revolution fitted to the Earth by best global or regional fit; WGS84 and GRS80 are such ellipsoids. The geoid is a surface of a different kind altogether: an equipotential surface of the Earth’s gravity field, closely approximating undisturbed mean sea level, and physically rather than geometrically defined. Geoid-ellipsoid separation reaches approximately -106 m and +85 m globally, so neither surface is an approximation of the other and the two cannot be substituted. The distinction matters operationally because the two ellipsoids in common hydrographic use are not themselves a source of incompatibility: GRS80 and WGS84 differ only in flattening, amounting to some 0.1 mm in the semi-minor axis, which is negligible at any survey order. Horizontal incompatibility arises instead from datum realisation and coordinate epoch, and vertical incompatibility from the choice of reduction surface (Li et al., 2023). Projected coordinate systems (PCS) on the other hand, convert these angular coordinates to a two dimensional surface in a manner that is suitable for the calculation of distances or areas which are vital in navigation and engineering planning (Kastrisios and Masetti, 2022). Also, the accuracy of coordinate transformation is a key factor in ensuring the interoperability of bathymetric data, particularly when integrating data from multiple sources and/or time periods. These variations are responsible for the changes in the horizontal and vertical reference datums, the ellipsoidal parameters, and the velocities of the tectonic plates (Hoy et al., 2025). Conversion of ellipsoidal heights (derived from GNSS) to local vertical datums (Mean Lower Low Water) remains a complex task requiring accurate geoid models and a transformation tool like VDatum or PROJ library (Vrdoljak, 2021). If these transformation parameters are not carefully documented, large spatial offsets result. The consequences of unregistered or poorly defined coordinate systems are far reaching, including misaligned data sets, “ghost” features on the seafloor and an increased risk to maritime navigation and subsea infrastructure (Nadzir and Munthe, 2025). Undocumented CRS also make datasets ‘dark’ in terms of data management, as they can not be linked to other geospatial data layers confidently, directly contradicting the ‘Interoperable’ and ‘Reusable’ principles of ‘FAIR’ data management practices. To counter these risks, the hydrographic community is increasingly asking for the use of a standardised coordinate-reference identifier, namely a EPSG (European Petroleum Survey Group) code. These codes enable software systems to automatically and accurately interpret the spatial context of a bathymetric file by creating a unique machine-readable registry of the parameters associated with the CRS (International Organization for Standardization, 2019). In the emerging IHO S-100 Universal Hydrographic Data Model, specifying CRS explicitly using standard identifiers is a pre-requisite for the distribution of “navigation-grade” products such as S-102 (Bathymetric Surface). Moreover, the development of a common global geodetic reference frame, such as the International Terrestrial Reference Frame (ITRF), has become critical for many activities that aim to provide a comprehensive global digital elevation model of the ocean floor, such as Seabed 2030, which relies on data from thousands of different sources to be harmonized and assembled (Alexander et al., 2023).

5.2 Vertical reference systems

The interoperability and safety of bathymetric data are critical aspects of ocean observing systems, thus the development of a strong vertical reference system is essential. The Chart Datum (CD) is the fundamental datum in hydrographic practice and it is the most important reference surface for determining water depths on nautical charts. The most commonly used standard for CD is the Lowest Astronomical Tide (LAT), which IHO S-32 defines as the lowest tide level that can be predicted to occur under average meteorological conditions and under any combination of astronomical conditions, the prediction being made over one 18.6-year nodal cycle; LAT is therefore a predicted datum rather than an observed extreme low water (El-Diasty et al., 2022). These measurements are taken into account by a conservative “safe depth” issued by hydrographic offices to avoid the risk of vessels grounding in periods of extreme low tide. Tidal datums, like LAT or Mean Lower Low Water (MLLW) are required for navigation reasons and are naturally location and time dependent, and therefore discontinuous in the navigable data coverage along the coast (Abd Rahman and Md Din, 2020).

In contrast to these navigational datums, Mean Sea Level (MSL) is the fundamental basis for the vast ocean observation and geodetic use. MSL is the average level of the ocean surface over a long time, excluding seasonal and tidal fluctuations (International Hydrographic Organization, 2019). It provides valuable linkages between land and sea vertical systems, has applications for coastal-risk mapping, sea-level rise observation and offshore infrastructure positioning. However, MSL is not a flat surface, but is influenced by ocean circulation and atmospheric pressure and must be corrected for sea surface topography when integrated with global geodetic models; the abbreviation SST is avoided here because in a marine-science context it is read as sea surface temperature (International Hydrographic Organization, 2019). A significant shift in bathymetry paradigms occurred in the move from tide gauge-referenced bathymetric surveys to Ellipsoidally Referenced Surveying (ERS). This method introduces a geodetic surface model that is mathematically the same around the world, like the WGS84 or GRS80 ellipsoid, as a coordinate baseline. The ellipsoidal heights are computed from the Global Navigation Satellite Systems (GNSS) and are relatively continuous vertical reference from land to sea, while the tidal datums are anchored in the sea surface. These ellipsoidal heights are used to give navigable depths by using geoid models and vertical separation surfaces, such as HyVSEPs or VDatum (Robin et al., 2016). Two distinct transformation chains are involved here and they must not be elided. The first is geodetic: an orthometric height H is obtained from an ellipsoidal height h through the geoid undulation N, as h = H + N, where N is supplied by a gravimetric geoid model and H is referred to the geoid. The second is hydrographic: a depth referred to chart datum is obtained through a modelled separation between the ellipsoid and the chart datum, SEP = h - CD, where the chart datum is realised from water-level observations and tidal or hydrodynamic modelling rather than from the gravity field, and where the geoid-to-chart-datum offset is itself a modelled quantity carrying its own uncertainty. An orthometric height and a tidal height therefore refer to different surfaces, reached by different means, and cannot be joined by a slash or treated as alternative expressions of one reduction. The instruments that implement these chains also differ in kind: NOAA VDatum is a transformation framework composing geodetic, tidal and hydrodynamic model components, not itself a separation surface, whereas the Canadian Hydrographic Service HyVSEPs are separation surfaces, that is, gridded realisations of the ellipsoid-to-chart-datum offset for a defined region. Technical problems arise when datasets refer to different vertical datums are integrated. Discontinuities are common at the national vertical-reference system boundaries across regions and at the coastline boundaries with the ocean and the ocean vertical-reference system (Williams and Esteves, 2017). Proper adjustments, based on consistent transformation parameters, are required to avoid these “datum jumps” that can cause high uncertainty in hydrodynamic models and habitat classification. In addition, there is limited documentation on datum transformations and original reference frames which can limit reusability of legacy bathymetric products (Biscotti et al., 2018). The solution to these interoperability issues lies in the use of the IHO S-100 Universal Hydrographic Data Model, and in particular the IHO water level product specification S-104 (International Hydrographic Organization, 2023a). These standards will enable the exchange of dynamic vertical references digitally, which are required to keep bathymetric data discoverable, accessible and properly harmonized between different observing platforms (Liebsch et al., 2023).

5.3 Vertical datum transformation

Vertical datum transformation is an essential task for interoperability of bathymetric data from different ocean observing systems. Historically, the hydrographic surveys were predominantly based on local tide-gauge relationships to determine a water-level reduction, known as a Chart Datum (CD) (Hwang et al., 2025). In a complex coastal or offshore tidal area, however, there are not enough physical tide gauges to use for interpolation, which is not linear, and may introduce errors (Hamden et al., 2023). To address these limitations, modern hydrography is moving towards Ellipsoidally Referenced Surveying (ERS), which relies on Global Navigation Satellite Systems (GNSS) for direct reference of soundings to a stable, global geodetic ellipsoid, instead of the unstable water level (Hamden et al., 2023). Geoid models are used as the basic connection between ellipsoidal heights and the vertical datums based on the gravity field on land or sea. Three surfaces are commonly conflated at this point and should be separated. EGM2008 is a global geopotential model complete to degree and order 2190, not a regional geoid model, although regional gravimetric geoid models are frequently computed using it as a reference field together with local gravity data. A geoid model does not yield a mean sea surface: mean sea surface is an altimetric product, such as DTU21MSS or CNES-CLS22, derived from multi-mission satellite altimetry, and it relates to the geoid through mean dynamic topography as MSS = N + MDT. Mean dynamic topography is of order plus or minus 1 m globally and exceeds 1.5 m in western boundary current regions, so the geoid and the mean sea surface are not interchangeable as offshore reduction datums. Where no tide gauge is available to realise a chart datum offshore, the reduction datum must therefore be named explicitly as the geoid, the mean sea surface, or a tidal datum realised through a separation model; metadata recording only mean sea level does not distinguish between these and is insufficient for reuse, since the three differ by amounts exceeding the total vertical uncertainty allowance of Special Order and Order 1a surveys (Ke et al., 2020). In more active regions, hydrodynamic models are coupled with the other models to simulate the propagation of tidal waves and estimate the tidal constituents over a continuous grid. These models can be used to calculate the difference between the MSS and the low or high-water levels to determine critical tidal surfaces like Lowest Astronomical Tide (LAT) or Highest Astronomical Tide (HAT) (Hamden et al., 2023). Combining these data sources creates a separation surface, a mathematical model that shows the distance separating the ellipsoid from any of the tidal datums at any point of the coordinate. Important applications are the NOAA VDatum and the Canadian Hydrographic Service (CHS) Hydrographic Vertical Separation Surfaces (HyVSEPs), which provide seamless vertical transitions over large maritime jurisdictions (Robin et al., 2016). They are usually distributed in the form of transformation grids, which give a standardized way to convert bathymetric data between different vertical reference systems that is compatible with modern hydrographic software. The application of these grids is significant for topobathymetric elevation models, which are required to address coastal risks, by combining data from the land and sea (Thatcher et al., 2016). Despite these developments, there is an inherent uncertainty in the conversion process which needs to be specified for navigational safety and for good data integrity. These errors mainly come from the precision of GNSS positioning, the spatial resolution and accuracy of the geoid model, and the accuracy of the hydrodynamic tidal simulations (Tian et al., 2022). Published RMSE values for separation models validated against independent tide-gauge benchmarks vary widely with tide-gauge density, tidal regime, and the quality of the underlying geoid and hydrodynamic models. Quoting a single best-case value without region, sample size, metric definition and confidence level is not informative, and validation results should be reported with that context. It is crucial to understand how these uncertainties can be propagated to meet the requirements of IHO S-100. Each link in the chain carries uncertainty, and these must be propagated explicitly rather than assumed away. The relevant terms are the uncertainty of GNSS ellipsoidal positioning, the commission and omission error of the geoid model, the uncertainty of the separation or tidal model realising chart datum, typically the dominant term offshore and in regions of sparse tide-gauge control, and the uncertainty of the sounding itself. Where these are independent they combine in quadrature into the total vertical uncertainty; correlated terms, such as a systematic bias in a regional separation model across an entire survey block, must be carried as a bias rather than folded into a random term, since they do not reduce with averaging and propagate directly into any grid built from the survey. Tidal datums are additionally realised over a defined epoch, conventionally the 18.6-year nodal cycle for LAT or a national tidal datum epoch, and secular sea-level change means that a datum realised on one epoch and applied to a survey conducted on another carries an offset that grows with the separation between them; the datum epoch should therefore be recorded in metadata alongside the datum name, since the name alone does not identify the surface. Reporting a bathymetric grid with a per-node uncertainty layer that omits the transformation terms understates its true uncertainty, and does so by an amount that is invisible to the user and unrecoverable downstream. The S-100 framework imposes stringent quality requirements and FAIR (Findable, Accessible, Interoperable, and Reusable) documentation for the integration of multi-source marine geospatial data (Hamden et al., 2023).

5.4 Environmental and instrumental corrections

Systematic mitigation of the environmental and instrumental errors at harmonization and spatial processing stages are crucial to the integrity of high resolution bathymetric datasets (International Hydrographic Organization, 2024a). The interoperability of the many ocean observing systems requires careful correction of these variables to ensure the accuracy and comparability of the seafloor models. In order to deal with refraction effects resulting from the passage of acoustic signals through the layers of water with different physical characteristics, fundamental corrections for sound-speed must be made. The temporal and spatial variation in the water column can cause a considerable beam bending due to the high dependence of the sound velocity to temperature, salinity and pressure. Correct Sound Velocity Profiles (SVPs) or Conductivity, Temperature, and Depth (CTD) measurements are needed to prevent systematic artifacts like the “smile” or “frown” curvature in multibeam swaths from degrading the metric reliability of the data (International Hydrographic Organization, 2024a).

Vertical harmonization is also very important for tidal corrections and for the overall management of water-level variability. Conventional hydrography was based on local tide gauges, but current workflows are increasingly moving towards adopted Global Navigation Satellite System (GNSS) methods for continuous vertical referencing (Hoffmann et al., 2022). This transition to ERS helps to mitigate the uncertainty of interpolation between tide gauges at widely spaced locations, especially in regions with complicated coastal topography where the relationship between sea-level changes and atmospheric pressure and wind-driven surge is complex. These geodetic techniques enable the connection of bathymetry with other Essential Ocean Variables (EOVs) in a common reference frame (Su, 2022). Instrumental corrections focus on the physical geometry and dynamics of the survey platform. Geometric distortions due to the motion of the vessels (heave, pitch, roll) require compensation, which can be achieved with the aid of high-precision Inertial Measurement Units (IMUs) or Inertial Navigation Systems (INS) combined with GNSS. Acoustic footprint stabilisation is provided either by compensation in real-time or by post-processing. Moreover, the distances and angles between the transducer, the motion sensor and the positioning antenna have to be determined using a “patch test” calibration (International Hydrographic Organization, 2024a). This process is used to isolate angular misalignments from the overall system to avoid “wobbles” or systematic shifts in the point clouds obtained. Draft corrections are used to correct for the vertical distance between the water surface and the transducer. This includes static draft (the ships displacement at rest) and dynamic draft (changes in the ships displacement due to fuel or cargo load or squat effects, which are caused by ships speed through the water) (Rattray et al., 2015). Advanced processing methods like Post-Processed Kinematic (PPK) positioning, which enhance horizontal and vertical accuracy by combating multi-pathing and satellite geometry errors, help reduce positioning errors caused by GNSS multi-pathing or satellite geometry. These corrections can then be synchronized across resolutions and hardware configurations to produce high resolution digital bathymetric models (DBMs) appropriate for geomorphological analysis (Wolfl et al., 2019). In summary, standardized correction procedures are essential to ensure accuracy for ocean mapping, monitoring infrastructure, and long-term monitoring changes in the ocean in an interoperable ocean observing system (Kint et al., 2023).

5.5 Data cleaning and quality control

Bathymetric data cleaning and quality control (QC) are essential in achieving interoperability and reliability of marine geospatial products. The workflow has changed from the labour-intensive manual editing to algorithmic methods that detect and throw out gross errors, systematic noise, and random outliers (Figure 2). These methods are based on statistical filtering or machine learning (ML) architectures (El-Diasty et al., 2022). Advanced supervised ML models provide a strong alternative, and are able to produce high accuracy depth estimates while rejecting false seafloor data with high accuracy; if these outliers are not effectively rejected, depth estimation accuracy errors can exceed international standards (Zhao et al., 2021). Systematic validation through cross-line analysis is imperative for consistency of internal data and is used to detect inconsistencies between overlapping passes of a survey and to provide a measure of how repeatable measurements are (International Hydrographic Organization, 2024a). This is supported by surface inspection and visual validation, through the use of digital bottom models (DBMs) or attributed grids, enabling the hydrographer to compare the interpolated results with the raw sounding data. In addition, multi-source bathymetry needed to be normalized to fill data gaps without losing fine scale topographic information (Wolfl et al., 2019). One of the main purposes of such QC processes is the compliance with the International Hydrographic Organization (IHO) S-44 and S-102 standards. The Combined Uncertainty and Bathymetry Estimator (CUBE) is commonly used to satisfy these needs. CUBE is a per-node multiple-hypothesis depth estimator: total propagated uncertainty (TPU) is computed per sounding upstream of CUBE from the vessel uncertainty model, and is then used to weight each sounding’s contribution to nodes within its radius of influence and to disambiguate competing depth hypotheses. CUBE does not itself compute TPU and is not a multi-sensor fusion algorithm. The output is usually in the form of Bathymetric Attributed Grids (BAGs), which serve as a common baseline for the development of high-resolution navigation products (Ai et al., 2016). This is particularly important to maintain adequate sounding density, and errors in data cleaning can seriously affect the quality of these grids, especially in areas of complexity or deep water. Modern multibeam, echosounders produce a large amount of data, and the transition from manual to automated QC is essential to handling that data. This automation, however, requires the retention of rejected-data records in order to meet FAIR (Findable, Accessible, Interoperable, and Reusable) data practices (Lecours et al., 2016). By documenting information about the data’s origin and rejection criteria, transparency and reproducibility are ensured, enabling future users to comprehend how the data was processed and reconsider and verify original observations as necessary (Lecours et al., 2016).

5.6 Interpolation and surface generation

Estimation of bathymetric surfaces from discrete-depth observations, such as acoustic or optical measurements, requires spatial interpolation. There are two general types of methods deterministic and geostatistical, which have different implications for the accuracy of the data and for interoperability. Deterministic methods are based on pre-established mathematical functions, like Inverse Distance Weighting (IDW) or Nearest-neighbour interpolation. The influence of the sampled point decreases with the distance from the target cell, therefore IDW is commonly used when the density of the dataset is relatively high (Alam and Nuruzzaman, 2025). Although nearest-neighbour interpolation is a simple and fast method, the resulting surfaces are discontinuous and may not accurately represent the underlying seafloor topography if it is complex (Li et al., 2023). In more complicated topography, Triangulated Irregular Networks (TIN) are used to generate triangles from the point data, which is a flexible model suitable for different observation densities. Geostatistical interpolation, which is mainly Kriging, provides a more robust framework by taking into account the spatial autocorrelation of seafloor depths. Unlike deterministic approaches, Kriging offers a statistically optimum estimate of the parameter and an estimate of the uncertainty of the prediction, which is essential for assessing the quality of the bathymetric model (Li et al., 2023). Kriging considers the spatial structure of the seabed when analysing the variogram of the depth measurement, and can provide reliable interpolation in areas of sparse data (Parente and Vallario, 2019). Spline interpolation can be used for applications that demand high smoothness, e.g., hydrodynamic modelling. This approach uses a mathematical polynomial to pass through data points with the minimum possible curvature on the surface but has to be carefully parameterised to prevent artificial oscillations in areas of steep terrain. Recently, many modern hydrographic workflows have shifted to automatic surface generation with the Combined Uncertainty and Bathymetry Estimator (CUBE) algorithm. CUBE is an evidence-based depth and total propagated uncertainty (TPU) calculation program that performs the depth calculation and TPU simultaneously at each grid cell (Ai et al., 2016).

CUBE uses multiple hypotheses for the depth value to automatically filter out the outliers, and to select the most probable seafloor surface, which reduces the amount of manual processing time that is required for navigational safety, while not sacrificing that safety. The choice of resolution in a grid and the way the aggregation of grid cells is done are key factors during the data lifecycle. Resolution optimization is a trade-off between capturing the fine scale morphology and the associated data volume and computational burden (Zhao et al., 2021). Grid-cell aggregation is the summarizing of a number of soundings into one of the cells, which should be done in accordance with international standards, ensuring interoperability. The use of the International Hydrographic Organization (IHO) standard S-102 for gridded bathymetry data makes these products compatible with modern navigation systems and other ocean observing platforms. This standardization is crucial in the development of FAIR (Findable, Accessible, Interoperable, and Reusable) datasets, which will help the world’s oceans be mapped to the greatest extent possible, such as with the Seabed 2030 project (Wölfl et al., 2019).

5.7 Resolution harmonization

Spatial resolution is key to the interoperability of ocean observing systems, because bathymetric data are acquired with a wide range of sensors, and the sampling density varies significantly. High-resolution multibeam echosounder (MBES) data is often re-sampled to coarser regional resolution for integration with other data, but these re-sampled data sets must avoid loss of topographic integrity through systematic workflows (Eakins et al., 2015). These datasets include data from different grid sizes (e.g., meter local data with kilometre scale SB derived from satellites), which require complicated fusion techniques (Fanti et al., 2024). Many methods for fusing point clouds have emerged to combine overlapping point clouds into one “best-available” surface, with a preference for higher quality data and lower quality data being supplemented by lower resolution “proxies”. One of the main technical issues is avoiding the creation of artificial precision—high-frequency artifacts in the process, which do not fit the data from the observations (Calder and Mayer, 2003). These are addressed by modern deep learning frameworks that incorporate convolutional neural networks with dropout methods, which attempt to supply uncertainty measurements that are spatially explicit, pinpointing areas where the reconstructed model might possibly be over-interpolated and consequently lack empirical support (Gülher and Alganci, 2025). Moreover, the choice of resolution is scale dependent, with the identification of seafloor geomorphological features, such as reef morphologies, sand waves, and volcanic edifices, being strictly constrained by the size of the grid cell in the delivery of the final product. Research in marine geomorphometry has shown that the derivatives of the original data, such as slope, aspect, and curvature, can vary significantly as the resolution of the raw data is altered, and therefore the resolution of the analysis must be carefully chosen to reflect the physical processes being investigated (Lecours et al., 2016). To overcome such restrictions modern hydrographic standards, in particular IHO S-102, have adapted to multi-resolution data structures. These structures, such as Variable Resolution (VR) grids and quadtree architectures, can keep high-resolution soundings in the more difficult or shallow areas (e.g., navigation channels and rocky outcrops), and switch to lower resolution in the deeper, more uniform areas (Kazimierski, 2023). This can enhance storage and computational efficiency for “big data” applications, while also following the FAIR data practices principle that the resolution of the data product provided matches the density and precision of the acquisition phase (Danielson et al., 2016). In coastal areas, the creation of a topo-bathymetric continuum requires the ability to merge such multi-resolution structures with land-based elevation models, usually through iterative multi-band ratio methods and the creation of hydrodynamic equilibrium profiles that will create a morphological connection across the land-sea interface (Andréfouët et al., 2025). The advanced multi-resolution frameworks can be incorporated into ocean observing systems to deliver a continuous, comprehensive, and scientifically sound representation of the seafloor at global and local scales (Danielson et al., 2016).

5.8 Edge matching and survey merging

A proper integration of the different bathymetric datasets to produce seamless regional or global models demands a careful edge matching and merging of the bathymetric surveys in order to maintain spatial continuity (Wölfl et al., 2019). Overlap analysis serves as the foundational step in this process, where redundant soundings from adjacent or overlapping survey swaths are compared to assess consistency and quantify systematic biases (Li et al., 2023). Boundary artifacts like “cliffs” or artificial ridges are often seen where the data from various sensors and platforms overlap, and usually result from variations in vertical datum, sound speed corrections, or sensor calibrations (Contarinis et al., 2022). To harmonize the datasets, it is important to detect these discontinuities and include localized adjustments, such as tapering or weighted averaging, to ensure a smooth transition between the datasets (Parente and Vallario, 2019). Large temporal gap between surveys adds a considerable amount of complexity to the merging process, especially in dynamic coastal regions where the dynamics of seabed morphology are fast. Priority of newer and better data is crucial when combining historical data with current data (Hoffmann et al., 2022). There are also workflows that use a hierarchical weighting scheme or “conflict-resolution rules” that give preference to data sets with lower overall propagated uncertainty (TPU) or more recent acquisition dates. For example in projects such as Seabed 2030 or EMODnet, metadata-driven prioritisation will provide for the replacement of older, less accurate measurements obtained by lead line or single-beam with high-resolution multibeam measurements. When there are irreconcilable differences, cross validation using a third independent source or median based filtering can be used to resolve the difference while maintaining geomorphological integrity of the seafloor. Documenting such converging decisions is at the heart of FAIR data practices (Thierry et al., 2019). During the edge-matching phase, it is important to explicitly record the prioritization logic used, and the version history and smoothing parameters used for the edge-matching (Mitchell, 2016). The transparency will enable future users to be able to trace down the steps in the data integration and to understand the origins of the final gridded surface, essential for scientific modelling and safe navigation. With the switch to IHO S-102, automated merging tools for surveys are growing in use, minimising human error and ensuring multi-source bathymetric products are interoperable at an international level (Contarinis et al., 2022).

5.9 Uncertainty propagation

The quantification and propagation of uncertainty is a key to the interoperability and reliability of bathymetric data sets in an ocean observing system (El-Diasty et al., 2022). The concept of Total Propagated Uncertainty (TPU) is defined by modern hydrography as a total estimation of all individual error components of a sounding (Parente and Vallario, 2019).

The main sources of measurement uncertainty are sensor-related parameters, such as the acoustic characteristics of the echosounder, the accuracy of the sound velocity profiles, and the accuracy of the heave, pitch and roll motion compensation (Mitchell, 2016). Positional uncertainty, also known as horizontal error, is affected by the precision of GNSS and INS and can differ greatly depending on the geometry of the satellites and the availability of the signals in remote and deep-water areas (Yang et al., 2025). The vertical transformation uncertainty is one of the most important layers of the TPU, especially when harmonizing datasets over different datums. When soundings are transformed from ellipsoidal to tidal height (like LAT), errors are introduced, typically caused by inaccuracies in the regional geoid model and/or spatial variability in the hydrodynamic tidal model (Liebsch et al., 2023). Moreover, uncertainties occur when creating gridded surfaces and Digital Bathymetric Models (DBMs) by interpolation. Geostatistical methods such as Kriging and the Combined Uncertainty and Bathymetry Estimator (CUBE) algorithm explicitly compute the standard deviation or confidence interval for each cell of the grid from the density and quality of the observations in its neighbouring cells (Ai et al., 2016). Temporal uncertainty is a concern about the stability of the seabed between surveys. Bathymetric observations may become out of date in dynamic coastal environments or where there is active sediment transport. This uncertainty can be quantified through detailed knowledge of the seafloor geomorphology, and the frequency of updates (Kint et al., 2023). The complex nature of these uncertainties are being more clearly captured in the final products, such as the International Hydrographic Organization (IHO) S-102 and S-104 standards. The Bathymetric Attributed Grid (BAG) format is used to represent the depth and uncertainty layers in a single machine-readable file for automated decision-support systems and safety-of-navigation applications to interpret the reliability of the underlying terrain data (Alexander et al., 2023). Spatial processing methods for bathymetry are summarized in (Table 2).

Table 2. Spatial processing methods for bathymetry.

6 Hydrographic, geospatial and data-exchange Standards

6.1 Role of standards

The adoption of international standards is essential to the consistency, quality and interoperability of bathymetric data in global ocean observing systems (Kastrisios and Masetti, 2022). The International Hydrographic Organization (IHO) provides standards, especially the S-44 standard for hydrographic surveys, that set high standards for data consistency and survey quality. These specifications establish maximum allowable uncertainties for horizontal positioning and vertical depth measurements for various orders of survey, from Exclusive Order up to Order 2, and afford a quantitative basis for assessment of the reliability of bathymetric observations (International Hydrographic Organization, 2024a). Following these benchmarks, hydrographic offices and research institutions can demonstrate that the data gathered by various types of platforms such as surface vessels and autonomous systems, are of sufficient quality to satisfy the high safety-of-navigation and scientific accuracy requirements (Li et al., 2023). In addition to measurement accuracy, geospatial standards outline the structure of the data and metadata that are required to manage the data successfully. The IHO S-100 Universal Hydrographic Data Model provides an updated model for consolidating multiple marine data products, including gridded bathymetry (S-102) and surface currents (S-111) (International Hydrographic Organization, 2023b), into a single, extensible environment (Alexander et al., 2023). This modular approach enables the creation of “next-generation” electronic navigational charts (ENCs) and provides a link between the traditional hydrographic domain and the broader marine spatial data infrastructures (MSDIs). In addition, ISO 19115 and 19139 standards for geographic metadata ensure that datasets are well documented and can be used to evaluate the origin, lineage and quality of the data (Thierry et al., 2019). Semantic consistency and validation are also important aspects of the standardization process. Standardized vocabularies and ontologies, typically operating via NERC Vocabulary Server and/or the IHO Registry, enable data terms to be interpreted in the same way between different technical systems and disciplines. Modern hydrographic processing workflows include validation protocols that can be used to automatically verify the data against a set of logical and topological rules, helping to minimise potential human error and to provide high quality information for the data lifecycle (Ai et al., 2016). This rigor is necessary for the integrity of such products as the Bathymetric Attributed Grid (BAG) that have embedded uncertainty layers and metadata layers to aid in evidence-based decision-making. Lastly, the general intent of these standards is to enable seamless exchange and reuse of bathymetric data based on the FAIR (Findable, Accessible, Interoperable, and Reusable) principles. Frictionless transfer of large datasets between producers and consumers is possible with standardized data exchange formats, like GML or HDF5, enabling global projects, such as Nippon Foundation-GEBCO Seabed 2030 (Murawski et al., 2025). Hydrographic standards promote interoperability and allow the maritime sector to capitalize on historical and current data for various applications ranging from hydrodynamic modelling to coastal risk assessment and marine resource management, all of which drive sustainable blue economy development (Kastrisios and Masetti, 2022).

6.2 IHO standard

The Standards established by the International Hydrographic Organization (IHO) support the quality and interoperability of bathymetric data in the modern ocean observing system. IHO S-44 standard, especially the 6th edition, provides the basic requirements for hydrographic surveys, including the requirement for specific Survey Orders from Exclusive to Special up to Order 1a, 1b and 2 surveys that ensure that the datasets produced are “fit for purpose” (Hasara et al., 2025). These orders are imposing stringent requirements for Total Vertical Uncertainty (TVU) and Total Horizontal Uncertainty (THU) parameters essential to measuring the confidence of soundings (International Hydrographic Organization, 2024a). Feature detection under S-44 Edition 6.2.0 is specified through the specification matrix, by survey order and depth band, alongside feature search and bathymetric coverage requirements, rather than as a single cubic-feature criterion. The 6th Edition also replaced the 5th Edition concept of full seafloor coverage with the concepts of bathymetric coverage and feature search, a substantive change with direct consequences for how survey completeness is described in metadata. In addition, the density and completeness of the survey, in turn, affects the metadata and quality assessments that are needed for integration downstream and are determined by bathymetric coverage specifications ranging from complete area coverage to systematic sampling (International Hydrographic Organization, 2024a). The IHO S-100 Universal Hydrographic Data Model is the basis for machine-readable interoperability across the marine domain, replacing traditional survey-based methods, and enabling data management. S-100 is an extensible product specification framework that uses a registry-based method to specify feature and portrayal catalogues unlike the previous S-57 (Contarinis et al., 2020).

Feature catalogues are generally XML-based and give a machine-to-machine definition of marine entities, whereas portrayal catalogues use Lua scripting or XSLT to standardise the way that data is displayed in different systems (International Hydrographic Organization, 2024c). This architecture includes strong coordination reference data and standardised metadata to make datasets discoverable and automatically interpretable by autonomous systems and marine spatial data infrastructures (MSDI). The multi-dimensionality of S-100 can also be exploited to enable the development of a digital government and highly integrated e-navigation environment (International Hydrographic Organization, 2024c). In this context, the IHO S-102 Bathymetric Surface product specification is dedicated to gridding-type representations of depth which go beyond the traditional vector format associated with nautical charting. S-102 offers a high resolution grid structure, which not only contains depth values, but also contains required uncertainty information, for dynamic safety contouring and precise navigation. Beyond just navigation, the metadata needs in S-102 make possible the preservation of data provenance and quality, critical for the use of this data for other ocean applications, including habitat mapping, coastal engineering, and climate modelling in the wider ocean-observing-system context (Lecours et al., 2016). S-102 is designed to ensure that high volume bathymetric data can be easily shared seamlessly with other hydrographic offices and scientific research institutions through giving a gridded format. Lastly, the IHO S-98 standard (International Hydrographic Organization, 2024d) defines the rules for layer interaction and system-level integration, in essence as the “glue” to ensure interoperability between different S-100 product specifications. S-98 defines interoperability levels that describe the interaction and coexistence of different data layers, including Electronic Navigational Charts (S-101) and bathymetry (S-102) within a display system. This is especially important for integrating with other marine layers, such as surface currents (S-111) or water levels (S-104), to achieve a full picture of the real-time marine environment (Sikirin and Golodov, 2026). Through the control of these interactions the integration of multi-source bathymetric data does not trigger conflicts and harmonised marine data products are reliable and safe in an interoperable hydrographic ecosystem (Ai et al., 2016).

6.3 Validation specifications

Strong validation of bathymetric data is essential to the integrity of ocean observing systems and to progressing towards fully interoperable, FAIR bathymetric data. In the changing world of hydrographic standards, validation has been replaced by more complex and machine-readable validation specifications, which allow for automated quality control (Kastrisios and Masetti, 2022). These specifications have been developed to ensure rigorous testing of the structural, semantic and qualitative characteristics of data products to satisfy the strict requirements of the International Hydrographic Organization (IHO) S-100 framework and related product specifications, such as S-102. At the core of modern validation workflows is the implementation of machine-readable validation checks. Hydrographic offices can take traditionally text-based hydrographic specifications and office specifications and convert them to executable code and then standardize data review with the use of automated tools like the QC Tools and CA Tools suites. These automated processes rely on rules, models, and logic derived from XML and similar data structures that closely resemble the decisions that would be made by expert analysts, thereby greatly minimizing the effects of human error in the processing and minimizing the variability in large volumes of multibeam and lidar data. In order to achieve machine-to-machine interoperability, information must be machine-readable and validated automatically without human interaction between automated systems in an Ocean Observing System (OOS). The structural validation is a second layer critical element, related to the integrity of the data container itself. The underlying HDF5 (Hierarchical Data Format version 5) architecture of bathymetric products (e.g., S-102: Bathymetric Surface) and surface currents (e.g., S-111: Surface Currents) should be validated to guarantee that complex multi-dimensional arrays and group hierarchies are properly structured (Kastrisios and Masetti, 2022). In the same manner, products using Geography Markup Language (GML) for encoding features, such as S-100 or S-122, have to be schema validated to check for their syntactic correctness. Structural validation ensures the integrity of the Marine Spatial Data Infrastructure (MSDI) from corrupt or non-standard files that could cause crash of downstream navigation and/or modelling software. In addition to structural integrity, metadata validation provides the complement by ensuring bathymetric datasets are well documented and follow a standard format. To meet the FAIR principles, and specifically the Findability and Reusability principles, the geographic metadata must be compliant with ISO 19115; IHO specific extensions are required. Validation checks in this domain are for the existence and correctness of provenance information, coordinate reference system (CRS) information, and lineage records. The “fitness for purpose” of a dataset cannot be confidently evaluated without verified metadata, especially when combining data from different origins such as crowdsourced bathymetry and satellite-derived bathymetry (SDB) (Chénier et al., 2019). More specific validation includes value range and consistency validation. The algorithms are used to compare individual data elements (including depth soundings, horizontal/vertical uncertainty values, and coordinate bounds) with physically or statistically based threshold limits (International Organization for Standardization, 2023b). For example, if the data is used for navigation, the total vertical uncertainty (TVU) has to be within the limits defined by IHO S-44 survey orders (Nadzir and Munthe, 2025). Spikes, outliers or systematic offsets due to environmental conditions and instrumentation drift can be effectively detected by value-range checks. These are intended to identify data points with unusual values, which would otherwise affect the final harmonised bathymetric surface and would therefore not reflect the true seabed morphology (Calder and Mayer, 2003). Last but not least product-compliance testing is the last “gatekeeper” prior to data publication. Analogous to the S-58 standard for Electronic Navigational Charts (ENCs), automated compliance testing for the S-100 series verifies that a data product adheres to its specific product specification in its entirety. This involves checking attribute values required by encoding rules, topological relationships, and mandatory attribute values. The strict adherence to compliance testing is crucial for the smooth integration of bathymetric information into global observing systems, where datasets from multiple jurisdictions can be combined into a harmonised, high resolution bathymetric model without conflict. Together, these validation specifications create an essential infrastructure that helps to convert raw observations into trusted, interoperable, FAIR marine geospatial assets (International Hydrographic Organization, 2024a).

6.4 ISO geospatial standards

International Organization for Standardization (ISO) geospatial standards give the basis for modern hydrographic data structures, such as the International Hydrographic Organization (IHO) S-100 Universal Hydrographic Data Model (Duan et al., 2021). The hydrographic community has aligned with the ISO 19100 series, which provides the mechanisms through which bathymetric data can be made technically valid and compatible with other marine data infrastructures (MSDI) and global ocean observing systems (Butkiewicz et al., 2022).

Coordinate Reference System (CRS) is the fundamental maths required to support the spatial integrity of bathymetric datasets, and ISO 19111 (Referencing by coordinates) is a vital standard for this. In hydrography, it is key to address the complexity of horizontal datums and different vertical datums (e.g. Mean Sea Level, Chart Datum) that are applied when acquiring data and harmonizing data. The of ISO 19111 compliant metadata allows researchers to establish and verify that the soundings are correctly located in 3D use space, thus enabling seamless integration of multi-source data collected at different times or on different data collection platforms (Ferreira et al., 2017). The findability and reusability of bathymetric products are primarily ensured by metadata management, as described in ISO 19115 (Geographic metadata) and its extension ISO 19115-2 (Extensions for imagery and gridded data). For bathymetry, the schemas outlined in ISO 19115–2 specifically address bathymetry surfaces and remotely sensed data (e.g., multibeam echosounder MBES point cloud and satellite-derived bathymetry (Kastrisios and Masetti, 2022). These metadata records are usually represented in the ISO 19139 (Metadata — XML schema implementation) standard, which enables automated extraction of the records by web-based catalogs and discovery services. The standardized encoding is essential to achieve FAIR (Findable, Accessible, Interoperable, and Reusable) data practices, enabling different maritime systems to automatically organise and understand the characteristics of the datasets without human intervention (Buck et al., 2019). The quantification and reporting of bathymetric data reliability are dealt with by ISO 19157 (Geographic data quality) which sets out a structured approach to describing data quality elements, including positional, temporal, and thematic consistency. For the IHO S-100 series, ISO 19157 is used to populate the quality-of-bathymetry (M_QUAL) layers, which are used to help users determine if a particular dataset is “fit for purpose” for applications from deep-sea scientific research to high precision coastal engineering. This minimizes consumer uncertainty by giving a clear indication of the estimated total propagated uncertainty (TPU) of the data and the compliance with the international survey orders (Ziaimatin et al., 2020). Lastly, ISO 19110 (Methodology for feature cataloguing) provides the means by which the features that are described in bathymetric products (e.g., wrecks, shoals, seafloor types) are defined by a consistent and standardized logic. The feature catalogs developed under this standard provide a dictionary for maritime information systems, supporting consistent interpretation of an object identified on an electronic navigational chart (ENC) or bathymetric attributed grid (BAG) is understood in the same way on all software platforms. These ISO standards can, along with other factors, create the “system of systems” required to integrate hydrographic information into the digital twin of the ocean, offer a collaborative and interoperable space for marine geospatial practitioners (Snowden et al., 2019).

6.5 Bathymetric attributed grid

The Bathymetric Attributed Grid (BAG) is a non-proprietary, open source file format for the storage and exchange of multi-dimensional bathymetric data, developed by the Open Navigation Surface Working Group (ONSWG) (Open Navigation Surface Working Group, 2023). The underlying principle of BAG is the use of a grid-based representation which differs from the conventional point-cloud and vector-based soundings and aims to be more efficient in the surface-model approach. This representation is usually done with the Hierarchical Data Format version 5 (HDF5), a file format that can store a large-scale seafloor elevation dataset and detailed metadata and several attribute layers in one file (UNH/NOAA Joint Hydrographic Center 2024 Performance and Progress Report Executive Summary, 2024). The data is stored in a regular or variable-resolution grid, allowing hydrographic offices to process the growing amount of high-density multibeam echosounder (MBES) data without compromising spatial integrity for safe maritime navigation (Bathymetric Attributed Grid (BAG) 2.0.1 Community Standard, 2026).

An important technical feature of the BAG format is the encoding of uncertainty in addition to depth at each grid node (Oh et al., 2019). In contrast to simpler Digital Elevation Models (DEMs), every BAG file will have a required “Elevation” layer and a corresponding “Uncertainty” layer. The values of uncertainty are normally derived from the statistical processing algorithms, which provide a degree of confidence in each depth estimate based on the precision of sensor and environmental factors (Calder and Mayer, 2003).Recent studies have focused on the need to develop these uncertainty models with greater accuracy, particularly by incorporating corrections provided by the Global Navigation Satellite System (GNSS) and the proper water-level adjustments, to ensure that the BAG surface reflects the “total propagated uncertainty” (TPU) required by the International Hydrographic Organization (IHO) S-44 standards. This comprehensive uncertainty storage capability is vital for today’s risk-based navigation, enabling Mariners and automated systems to determine, in real time, the level of confidence in the charted depth (International Hydrographic Organization, 2024b).

BAG has become the common intermediate and final product for numerous national hydrographic agencies including the Canadian Hydrographic Service and NOAA in the context of hydrographic production. It reduces ‘ping-to-chart’ latency, because the surfaces can be reviewed and edited by the hydrographer instead of individual soundings. In addition, the format includes “Tracking Lists” that can be used to document the original source data for any nodes that are manually mapped and edited by a cartographer, which helps to maintain the source data’s provenance across the production workflow. It is essential for the modern management of hydrographic data to be transparent, and to have a history of the various decisions that have been made in regard to it, which can be tracked (Open Navigation Surface Working Group, 2023). BAG and S-102 are related in origin but are distinct artefacts with distinct roles, and they are not interchangeable. The first edition of S-102 drew directly on the ONSWG BAG work, and that lineage is why the two are often conflated, but the specification in force is Edition 3.0.0 of December 2024, which is not a BAG container (International Hydrographic Organization, 2024b). S-102 Edition 3.0.0 is encoded as S-100 HDF5 under S-100 Part 10c, carries discovery metadata conforming to S-100 Part 17, defines quality through the QualityOfBathymetryCoverage feature, and requires a Digital Certification Block under the S-100 Data Protection Scheme for datasets produced for navigational purposes; Editions 2.1.0 and 2.2.0 are superseded. The practical consequence for interoperability is that the two sit at different points in the lifecycle. BAG is a processing and interchange format carrying per-node uncertainty and an edit tracking list, used within and between hydrographic offices. S-102 is a regulated navigational product specification, with certification, portrayal and interoperability obligations that BAG neither imposes nor claims. Treating a BAG surface as though it carried the regulatory status of S-102 overstates it; treating S-102 as merely a container format understates its specification content. Both errors are common in the secondary literature. The S-102 metadata structure is designed to match that of the ISO 19115/19139 standard adopted by the S-100 framework, ensuring that bathymetric surfaces can be cleanly interleaved with other marine layers such as S-111 (surface currents) and S-104 (water levels) (Sikirin and Golodov, 2026).

The benefits of BAG for wider interoperability stem from the principles of “Findable, Accessible, Interoperable, Reusable” (FAIR) data practices. It is open format, which delivers the benefit of eliminating vendor lock-in and enables data sharing between hydrographic offices, research institutions and commercial GIS platforms. Yet there are some constraints such as the amount of computational resources needed to process and display a high-resolution or variable-resolution grid in a limited resource ship borne system (International Hydrographic Organization, 2024b). Furthermore, although the format is technically interoperable, a substantial modernization of the national data infrastructure and regulatory practices will be necessary to make the transition to fully dynamic surface-based navigation from traditional paper and vector charts necessitated by the adoption of S-102 at global level (Sikirin and Golodov, 2026).

6.6 NetCDF and CF conventions

The Network Common Data Form (NetCDF) is a basic format for managing and sharing scientific data with multidimensional characteristics in oceanography and climate studies. NetCDF is a self-describing, machine-independent format that is designed for handling complex arrays with data values that are directly tied to their spatial and temporal coordinates. This multidimensionality enables the simultaneous storage of longitudinal (X), latitudinal (Y), vertical (depth/elevation) and temporal (T) dimensions in one single binary file and provides a comprehensive overview of the marine environment. This structure allows for the storage of not only the bathymetric data but also the metadata and auxiliary variables that are associated with the survey, such as sound speed profiles, or uncertainty in the survey, directly within the same data container (Open Navigation Surface Working Group, 2023). The wide-spread use of NetCDF in ocean observing systems is strongly influenced by the use of this format as the major delivery mechanism for major global and regional bathymetric data compilations, such as GEBCO and EMODnet. These efforts depend on the use of NetCDF to disseminate Digital Terrain Models that are essential boundaries for numerical simulations (Schmitt et al., 2025). Moreover, NetCDF is the standard way to provide bathymetric and ice-shelf topography input to high-resolution modelling frameworks such as the Regional Ocean Modelling System (ROMS) and the High-resolution Ice-Ocean Modelling and Assimilation System (HIOMAS), which helps ensure the accuracy of the physical features of the seafloor in different simulation environments (Wolfl et al., 2019). The Climate and Forecast (CF) conventions offer a common set of metadata protocols which improve the interoperability of NetCDF significantly. The CF conventions establish strict conventions for variable naming (Standard Name table), coordinate reference system and units specification (UDUNITS) so that data can be made intelligible to a wide range of software packages without requiring special parsing. These conventions allow vertical datums and horizontal projections to be clearly identified for bathymetric data, which is crucial for the correct spatial registration of multiple data sources. The CF conventions also make NetCDF data products FAIR, meaning that they are Findable, Accessible, Interoperable, and Reusable, which facilitates automated discovery and integration into marine spatial data infrastructures (MSDI). While NetCDF and the CF conventions have many advantages in scientific modelling, they have limitations when compared to formal hydrographic and navigational standards, such as the IHO S-100 Universal Hydrographic Data Model (Wang et al., 2022). The data format NetCDF is primarily intended for data analysis and scientific visualization, and does not support feature-attribute relationships and portrayal catalogues like those specified in IHO S-102 (Bathymetric Surface) necessary for safety-of-navigation applications. Compared with the S-100 series, which is designed for secure and regulatory compliant exchange using protocols such as SECOM, NetCDF does not natively support the integrity and authenticity requirements of official electronic navigational charts (ENCs). Therefore, NetCDF is currently the preferred format for research and ocean modelling, while being a complementary format to the formal product specification for hydrographic products used in maritime safety and e-navigation purposes (Sikirin and Golodov, 2026).

6.7 OGC standards and APIs

Interoperability of bathymetric data in ocean observing systems is grounded in the standards and Application Programming Interfaces (APIs) of the Open Geospatial Consortium (OGC). The standards serve as a common language and protocol for discovering, visualizing and disseminating marine geospatial data across a variety of platforms and stakeholder groups (Snowden et al., 2019). As the marine domain moves to embrace FAIR (Findable, Accessible, Interoperable, and Reusable) data practices, web services, while still significant in their own right, have become increasingly outmoded and replaced by new, more resourceful APIs for high performance machine access and cloud-based workflows. Historically, the OGC Catalogue Service for the Web (CSW) has been essential for metadata discovery as a key initial stage of the data lifecycle. The CSW standard allows for the publication and discovery of collection-level metadata, enabling researchers and automated systems to find bathymetric data in a distributed infrastructure through query using standardised attributes like spatial extent, temporal range, provenance etc (Martinez et al., 2023). CSW implementations are typically based on ISO 19115 Bathymetric metadata profiles, which allow bathymetric data to be described in enough detail to be harvested by global portals such as EMODnet or the GEBCO (General Bathymetric Chart of the Oceans) grid (International Organization for Standardization, 2014). CSW has served as the workhorse for catalogue services, but it is also seeing more and more augmentation by the OGC API - Records standard (Open Geospatial Consortium, 2024), which is a more developer-friendly, JSON-based approach to metadata discovery, more in line with a modern web architecture and less complex for integration into mobile and browser applications. The Web Map Service (WMS) and Web Coverage Service (WCS) are still ubiquitous for visualization and delivery of bathymetric surfaces. WMS can also be used to render complex bathymetric data as map images for easy use in web portals and Geographic Information Systems (GIS) without transferring the high resolution raw data (Oh et al., 2019). This is particularly useful for rapid visualization in maritime navigation portals. WCS, however, offers the raw coverage data (e.g., depth grids) so that the user can do client-side processing, such as building terrain models or mapping habitats. These services are currently being modernised with the OGC API - Tiles and OGC API - Maps, where the idea is to optimise data delivery by tiling mechanisms that minimise latency and enhance the user experience of web-based bathymetric viewers, enabling the loading of high-resolution seafloor details asynchronously (Lahtinen et al., 2025). The OGC SensorThings API (STA) and the legacy Sensor Observation Service (SOS) are used to manage real-time and near realtime sensor data, including tide gauge data, bottom pressure sensors, or autonomous underwater vehicles (AUVs). The SensorThings API is especially suited for the “sensor to decision” segment of the chain, because it provides a common machine-readable format for both observations and metadata, enabling the semantic enrichment of oceanographic data (Open Geospatial Consortium, 2024).

STA enables ocean observing systems to ingest time-series data from sensors as they are taken into the water into digital systems, thereby enabling real-time quality control and multi-sensor fusion from multi-distributed research centres. One major change in the way hydrographic data is managed is the use of cloud-based geospatial access. Cloud Optimized GeoTIFF (COG) and Zarr are formats that can provide efficient, random access to large-scale bathymetric datasets without the need for full file downloads, which are suitable for cloud-based storage. These cloud-oriented architectures, together with OGC APIs, facilitate the design of “Digital Twins” of the ocean to which large quantities of bathymetric and environmental data can be queried and analysed at scale (UNH/NOAA Joint Hydrographic Center 2024 Performance and Progress Report Executive Summary, 2024). The ecosystem directly contributes to the FAIR principles, helping to make the data findable in standardized catalogues, accessible via open APIs and interoperable through automated discovery and machine-to-machine exchange throughout the global research community. This transition from human-centric to machine-centric access is crucial for the future of operational oceanography and the UN Decade of Ocean Science for Sustainable Development’s success (Lahtinen et al., 2025).

6.8 Ocean best practices system

The Ocean Best Practices System (OBPS) is now a key component of the global ocean observing infrastructure, offering a standardized approach to documentation, archiving and sharing of ocean observing methods and workflows (Pearlman et al., 2019). The OBPS is managed under the auspices of the Intergovernmental Oceanographic Commission (UNESCO-IOC) and aims to overcome the long-standing problem of methodological fragmentation, in which critical protocols are not readily accessible or well documented (Palevsky et al., 2024). It provides a community-driven repository, which helps move from “expert knowledge” to standardized procedures that are accessible worldwide (Buttigieg et al., 2019). The OBPS offers the required structural support to ensure that bathymetric data processing chains, from the output of the sensors to harmonized surfaces, can be made transparent, machine readable, and discoverable. One of the key characteristics of the OBPS is the ongoing identification of best-practice documents using Digital Object Identifiers (DOIs). This practice gives methodologies a permanent “cite-ability” and a permanent availability, giving researchers the ability to refer to the exact version of a protocol used in the acquisition of data from or in the processing of data (Horstmann et al., 2021). The adoption of DOIs helps to maintain the long-term validity of the methodological archive, reducing the problem of “link rot” typical in the documentation of projects. For bathymetric processing, this version control is especially important because the evolution of algorithms (such as variants of the CUBE) and software can have a major influence on the bathymetric product produced. The OBPS has introduced formal citation of methods, giving it a new status as a primary scientific contribution. Rigorous community review process underpins the authority and reliability of the practices within the OBPS. This process will include the consensus and peer review of a set of subject-matter experts to ensure that a method is deemed to be “best practice” (Pearlman et al., 2019). The reviews assess the progress of the practice, distinguishing between new methods and approaches that have been widely tested and accepted. This is a joint vetting process and the documentation is not only accurate but also easy to use and suitable in various organizational contexts. For hydrographic organisations, this community-based approach can be considered as an addition to formally established standards such as those of the IHO as a practical ‘how to’ guide that can support the effective translation of high level standards into day to day operations. Additionally, the OBPS plays a key role in facilitating research transparency and reproducibility (Palevsky et al., 2024). The system incorporates the features of semantic search and fine-scale indexing, as well as community-based ontologies, enabling practitioners to search and compare relevant protocols in specific environments and/or instruments. Structural transparency is a prerequisite of scientific reproducibility, because it allows to explicitly define steps in data collection and to verify them by external users. This transparency enables direct comparisons of the datasets produced by various entities in the context of large-scale bathymetric projects such as EMODnet or GEBCO, where the processing steps that have been used are clearly documented and can be accessed (Samuel et al., 2021). Lastly, the OBPS can be a key element in linking observations to hydrographic standards and the implementation of the FAIR (Findable, Accessible, Interoperable, and Reusable) data principles. The OBPS enables data products to contain their full provenance by metadata-tagging protocols and federated systems such as the Ocean Data and Information System (ODIS). This connectivity is important for interoperability of bathymetric data where users can assess the “fitness for purpose” of a dataset according to the methods used to create it. Overall, the OBPS functions as a catalyst for the harmonisation of ocean observations, enabling sustainable ocean management and equitable access to information for the global ocean science community (Horstmann et al., 2021).

6.9 Standards classification

A coherent framework for bathymetric data needs to clearly differentiate the different types of standards that apply to the bathymetric data life cycle from acquisition to dissemination. Survey standards are at the bottom of this pyramid, in which standards prescribe the technical requirements and the accuracy criteria for the collection of the raw data. The International Hydrographic Organization (IHO) S-44, Edition 6.2.0, adopted in October 2024, is foremost, setting out Orders of Survey that specify horizontal and vertical uncertainty limits, bathymetric coverage, feature detection and feature search requirements according to the navigational use of the area (International Hydrographic Organization, 2024a). Multibeam echosounders or satellite-derived bathymetry (SDB) data must be processed in a way that ensures the resulting data is suitable for the intended use before further processing is done (Hell and Jakobsson, 2011). These requirements form the basis for data models, which offer an abstract structure for organizing geospatial information. The IHO S-100 Universal Hydrographic Data Model is the “master” standard that, through the ISO 19100 series, defines a General Feature Model (GFM) (Sikirin and Golodov, 2026). This conceptual model enables the definition of features, attributes, and relationships in a machine-readable registry that can aid the migration from the 2D paper chart to a multidimensional marine spatial data infrastructure (MSDI) (Contarinis et al., 2020). These products are stored and shared using file formats and encodings. In the legacy formats, both depth and uncertainty were stored in a fixed binary format, while modern formats tended to support interoperable formats such as the Bathymetric Attributed Grid (BAG) (structure based on HDF5 used to store depth and uncertainty) (Oh et al., 2019). Other common formats include the Geography Markup Language (GML) for vectorized features and GeoTIFF for coverage data, which can be read by many Geographic Information Systems (GIS) applications. To support discoverability and understanding of the lineage of these files, metadata standards such as ISO 19115 are applied, including some context information about the “who, what, where, and when” of the dataset (Contarinis et al., 2020). These are frequently accompanied by quality criteria such as ISO 19157, which defines the quality of the data in terms of data integrity, logical consistency and thematic accuracy, providing a means by which the suitability of a bathymetric layer for a particular application can be judged. Moreover, standardizing the terminology used across different systems is vital to the achievement of “true” interoperability through semantic vocabularies and registries. Examples of this include the NERC Vocabulary Server (NVS) and the use of Simple Knowledge Organization System (SKOS), which provide consistent standard definitions for terms such as “mean sea level” or “depth area” across the world (Contarinis et al., 2020). These protocols are then used to retrieve this information from the web. The Web Map Service (WMS) and the Web Coverage Service (WCS) are OGC services used for visualization and access to raw data for machine-to-machine exchange respectively. Lastly, best practice documents, such as those compiled by the Ocean Best Practices System (OBPS), offer the procedures and working instructions that will help move from technical standards to field action. These categories create a nested architecture that ensures bathymetric data is FAIR (Findable, Accessible, Interoperable and Reusable) to the entire global ocean observing community (Pearlman et al., 2019). Standards and specifications relevant to bathymetric interoperability are summarized in (Table 3). The crosswalk was constructed in four steps: fixing the normative corpus; decomposing each standard into obligatory elements; mapping each element against FAIR sub-principles as normative, evidence-based or interpretive; and recording residual gaps.

Table 3. Standards and specifications relevant to bathymetric interoperability.

7 FAIR data practices in bathymetric workflows

The addition of the FAIR principles (Findable, Accessible, Interoperable, Reusable) to bathymetric workflows (Figure 3) goes beyond data storage to a machine-actionable ecosystem (Wilkinson et al., 2016). The term FAIRness is often used interchangeably with “open data”, but the term has a specific definition and centres on the technical ability of data and metadata to be processed by autonomous systems without any human interaction (Tanhua et al., 2019). Machine-actionability is essential in ocean observing systems, where autonomous sensors produce enormous amounts of data, to scale data discovery and integration in global networks. This is true for both metadata and underlying data, and high-quality metadata is the digital “handshake” that enables algorithms to determine suitability prior to time-consuming data transfer. In hydrography the concept of findability is rooted in the use of persistent identifiers (PIDs) for data sets and researchers, including Digital Object Identifiers (DOIs) for data sets and ORCIDs for researchers, which make data citable and traceable across its lifecycle. These identifiers should be accompanied by rich, searchable metadata that includes spatial and temporal coverage, instrument specifications such as the models of multibeam echosounder, and platform information (International Organization for Standardization, 2014). In order to build up global visibility, metadata should be indexed into the appropriate metadata repository and connected with other publications, bathymetric surfaces should be discoverable by the domain-specific portals and geospatial search engines. Accessibility means that both data and metadata can be retrieved through a standardized communication protocol, such as Open-source Project for a Network Data Access Protocol (OPeNDAP) or OGC web services (Larkin et al., 2025). Modern infrastructures are based on Application Programming Interfaces (APIs) to offer machine-readable access for automatic ingestion into “Digital Twins” of the ocean. Most importantly, FAIR practice does not imply that there is no restriction on access; it is the documentation of the access that is required. For limited datasets, metadata should still be publicly available to indicate that the data exists and describe the steps necessary for its secure retrieval (Martínez et al., 2023). Inter-operability is enabled by the use of common data models and standardized file formats that enable integration of diverse data sets at the same fidelity level. Consistency is crucial and relies heavily on the IHO S-100 Universal Hydrographic Data Model and the Bathymetric Attributed Grid (BAG) format. In addition to file structure, interoperability requires machine-readable coordinate reference systems (CRS), vertical datums (VD) that are explicitly defined and units of measurement that are common to both parties. Controlled vocabulary and ontologies, e.g. NERC Vocabulary Server, ensure that variables and units are defined unambiguously, further enhancing semantic interoperability. The ultimate goal of the FAIR is reusability, which involves complete documentation of the provenance and quality of the data (De Martino et al., 2019). The processing history of a reusable bathymetric dataset should include details of how the raw bathymetric soundings were created into gridded products, instrument calibration information, and standardized uncertainty representations. Clear licenses information, preferably in machine-readable Creative Commons templates, describes the legal conditions for secondary use (RDA FAIR Data Maturity Model Working Group, 2020). Moreover, version control and connection with obtained software versions made for processing allows for reproducing the results and adapting them to different scientific or engineering applications. Assessment indicators are indicators that can be used to monitor the implementation of FAIR, for example when a valid PID, a declared vertical datum, or machine-readable licenses are present. There are, however, constraints to implementation, such as a dataset that may be technically FAIR, but is not scientifically accurate due to the absence of environmental artifacts or calibration drifts in the metadata (Aguilar Gómez et al., 2025). Furthermore, information in a data set can be technically available but not suitable for integration due to missing or insufficient metadata provenance or semantic alignment. So FAIRness should be considered as a technical standard to complement but not to substitute for careful scientific validation and domain-specific quality control (Tanhua et al., 2019). Recurrent FAIR deficiencies in bathymetric data workflows and the standards-based mechanisms available to address them for bathymetric datasets is summarized in (Table 4).

Table 4. FAIR assessment framework for bathymetric datasets.

Table 4 sets out the FAIR deficiencies most consistently reported across bathymetric data workflows, together with the standards-based mechanism available to address each and the residual gap. It is a synthesis of deficiencies described in the literature and of the mechanisms specified in the standards reviewed in Section 6; it is not a measurement of the FAIRness of any defined population of datasets. Quantifying that population, by scoring a sample of holdings from the IHO DCDB, EMODnet Bathymetry, GEBCO, NCEI and PANGAEA against the RDA maturity indicators, is identified as the necessary next step and is not attempted here.

The indicators are not proposed as a new assessment framework. They are the subset of the RDA FAIR Data Maturity Model indicators (RDA FAIR Data Maturity Model Working Group, 2020) for which a bathymetry-specific implementation mechanism exists and restricted further to those on which the reviewed literature reports a recurrent deficiency.

8 Multidimensional interoperability challenges

The multidimensional interoperability of bathymetric data is a great challenge, demanding technical, semantic and institutional solutions to make ocean observing systems truly global. At the syntactic level, the persistence of incompatible file formats and proprietary software dependencies frequently restricts data portability (Buck et al., 2019). Although some standards, like the Bathymetric Attributed Grid (BAG) and NetCDF-CF (Open Navigation Surface Working Group, 2023; CF Conventions Committee, 2024) have become popular, there is still a significant amount of legacy data that is only partially machine-readable, because the encoding documentation is either missing or varies in its approach to naming variables (Owens et al., 2022).

Interoperability is not a property conferred by a file format, and the adoption of an open encoding is a necessary but far from sufficient condition. Six levels can usefully be distinguished, and they are ordered in the sense that failure at a lower level cannot be repaired at a higher one. Syntactic interoperability concerns whether a receiving system can parse the bytes at all, and is the level at which BAG, S-102, NetCDF and HDF5 operate (Open Navigation Surface Working Group, 2023; International Hydrographic Organization, 2024b); it is where the bathymetric community has made most progress and, for that reason, the level most often mistaken for the whole problem. Semantic interoperability concerns whether the receiving system can determine what the parsed values mean - which quantity a depth attribute reports, against which surface, in which units, and what a quality attribute asserts - and depends on controlled vocabularies published as resolvable identifiers rather than on the encoding. Geospatial interoperability concerns whether coordinates and depths can be placed in a common frame, and requires the horizontal realisation, coordinate epoch, vertical datum and separation model to be declared and machine-resolvable; this is where multi-source bathymetric integration most often fails, and no encoding choice repairs it. Temporal interoperability concerns whether observations separated in time can be compared, and requires the epoch of both the coordinate frame and the tidal datum realisation, since a datum name alone does not identify a surface. Organisational interoperability concerns whether the institutions holding the data have compatible mandates, release procedures and service commitments. Legal interoperability concerns whether the licences attached to two datasets permit their combination and redistribution, a question no technical measure can settle. A dataset may be fully interoperable syntactically and unusable in practice through failure at any of the other five levels, and the barriers reviewed below are grouped accordingly.

Besides these technical issues, data size and complexity of Application Programming Interface (API) support are also an issue for cloud data processing and data exchange in real time. Another challenge is semantic interoperability, where the different communities of research and hydrographic communities give different definitions for the different terms such as depth, elevation, seabed features, etc (Thierry et al., 2019). Lack of controlled vocabularies leads to inconsistencies in data sources which lead to different units of measurement and different terminology (Hoeberechts et al., 2017). Metrological information and uncertainty parameters are often embedded in free text descriptions and, without machine-actionable ontologies would not be correctly understood by automated systems (Hippolyte et al., 2023). Another factor which can impede geospatial interoperability, although less so, is the mismatch of horizontal coordinate reference systems (CRS) and vertical datums. Data synthesis involves complex and sometimes imprecise datum transformations due to inconsistent grid structures and resolutions, and unaligned coastal boundaries (International Hydrographic Organization, 2024c). A number of standards for maritime surveys are used to define the quality of bathymetric products, including IHO S-44, but the process of validation varies, so the bathymetric products lack common quality flags and uncertainty documentation provided by the various providers. This variation introduces a problem for end-users to assess the fitness-for-purpose of aggregated datasets. The temporal interoperability problem arises because of the changes in the sea bottom surface, which are due to sedimentation and coastal erosion, and also due to manmade activities like dredging, which results in the lack of updated versions or acquisition dates for the older surveys (Hasara et al., 2025). The management of such temporal changes is an essential component of the sustainability of reliable Marine Spatial Data Infrastructures (MSDI) and is often not part of the data archiving process (Weatherall et al., 2015). The most frequent obstacles to the movement of data are legal and organizational obstacles. Often, redistribution of high resolution bathymetry is limited, and there are complications regarding the compatibility of open and restricted licenses that can lead to legal silos. Reducing the size of agencies as a result of prioritizing and distributing technical expertise among several units results in poor coordination and clear data stewardship (Stewart et al., 2019). Even with clear need for FAIR practices, limited funding and a lack of a willingness to adopt new standards, such as IHO S-100, still hinder FAIR adoption within the global ocean observing community. This involves a multifaceted strategy of technical solutions (e.g., semantic web technologies) and a fundamental paradigm shift in international data governance and sharing of data among institutions (Pearlman et al., 2019).

9 Toward integration frameworks

Implementing hydrographic standards and FAIR data principles is laying the groundwork for a global marine data ecosystem (Figure 4). The International Hydrographic Organization (IHO) S-100 framework is developing and being linked to the FAIR principles as a vehicle for transitioning from legacy data formats to dynamic and interoperable digital ocean worlds (Alexander et al., 2023). Besides direct applications such as maritime navigation, data products such as the S-102 bathymetric surface can be mapped to FAIR-compliant metadata schemas to be integrated into other scientific platforms. This integration makes the hydrographic data not only meet the requirements of safe navigation but also being discoverable and machine-actionable for various oceanographic research uses (Sikirin and Golodov, 2026). This transition is centred on the development of an interoperability model based on workflows with an approach of “FAIR by design”. These models have been embedded in the data processing chain ranging from data acquisition to the production of gridded products of bathymetric data. These systems enable high data provenance and data traceability by automating the metadata and encoding processing steps in the workflow. These are systematic pipelines that play an important role in merging multi-source bathymetry data, thereby merging data from different platforms without error, ensuring that the data is consistently consistent in different data types (Wilkinson et al., 2016). An important part of this picture of standardization is the Ocean Best Practices System (OBPS), a machine-readable database of marine methodologies. The OBPS shall be built to capture best practices in structured machine-actionable formats so that automated systems can discover and automatically implement standard practices for collection and processing of hydrographic data. The OBPS can enhance the quality and reliability of ocean information and enables the best practices use by researchers and autonomous agents ensuring the consistency of ocean data gathered from the ocean observation systems worldwide (Tanhua et al., 2019). Furthermore, artificial intelligence (AI) and machine learning (ML) are increasingly applied to the harmonisation process of heterogeneous bathymetric data sets. Harmonization models based on AI are now employed to carry out automatic cross-calibration of datasets with a high degree of spatial and temporal resolution and different coordinate systems, and with varying quality (Aguilar Gómez et al., 2025). These algorithms can detect systematic errors and correlate multi-source data to a common vertical datum, reducing the need for much manual effort to quickly produce consistent high-resolution bathymetric models. These AI powered workflows, when included in FAIR-compliant workflows, enable the harmonization itself to be documented as part of the data provenance, which means that the integrated products are transparent, reproducible and can be used in complex ocean observing systems (Gaignard et al., 2023).

These developments warrant a more guarded assessment than they usually receive. Eight considerations bear on whether machine-learning methods are admissible in a hydrographic workflow, and they are not equally satisfied. Uncertainty propagation is the most consequential: a learned transformation generally does not expose a calibrated uncertainty on its output, so a correction applied to soundings by such a model breaks the propagation chain described in Section 5.3 at precisely the point where it must remain intact, and a total vertical uncertainty computed downstream of an uncharacterised learned step understates the true figure by an unknown amount. Validation against independent observations is therefore not optional but a precondition, and validation against held-out data from the same survey is insufficient, since it shares the systematic errors the model may have learned to reproduce; independent tide-gauge benchmarks, crossline comparisons or contemporaneous surveys of known order are required. Generalisation is a related difficulty: models trained in one tidal regime, sediment environment or sensor configuration frequently degrade when transferred, and degradation in this setting is often silent, producing plausible surfaces rather than obvious failures. Training-data dependence compounds this, because the archives available for training are precisely the legacy holdings whose datum documentation and uncertainty characterisation this review identifies as deficient, so a model may inherit and then propagate the errors it was intended to detect.

The remaining four considerations concern accountability rather than accuracy. Explainability matters here in a specific and limited sense: a hydrographic office must be able to state why a particular sounding was rejected or adjusted, and a model that cannot support that statement is unsuitable for chart production regardless of its aggregate performance. Reproducibility requires that model weights, training data, preprocessing and random seeds be archived alongside the product, since a result that cannot be regenerated cannot be audited after an incident. Provenance requires that any learned transformation be recorded as a distinct processing step in the lineage, identifying the model and version applied, which the metadata standards reviewed in Section 6 can accommodate through LI_ProcessStep but rarely do in practice (International Organization for Standardization, 2014). Suitability for safety-critical use follows from the preceding: for navigational products subject to S-44 order compliance and the certification obligations of S-102 Edition 3.0.0 (International Hydrographic Organization, 2024a, 2024b), machine-learning methods are at present appropriate as decision support flagging candidate outliers for human adjudication, rather than as autonomous processing stages. For research-grade and model-ready products the tolerance is wider, but the uncertainty and provenance requirements are not thereby waived.

The framework is defined by its interfaces rather than by its components, since it is at the interfaces that interoperability either holds or fails. Five are specified. The acquisition-to-processing interface (I1) carries raw soundings together with the vessel uncertainty model and the sensor configuration record, and its contract is that no sounding crosses it without the parameters needed to compute total propagated uncertainty. The processing-to-harmonisation interface (I2) carries per-node depth and uncertainty together with an explicit declaration of the horizontal reference frame, its realisation and coordinate epoch, and the vertical datum with the separation model and datum epoch used to realise it; this is the interface at which the vertical-datum heterogeneity identified is either resolved or propagated silently. The harmonisation-to-encoding interface (I3) binds the harmonised surface to a product specification, S-102 Edition 3.0.0 for navigational products or NetCDF-CF for scientific reuse (International Hydrographic Organization, 2024b; CF Conventions Committee, 2024), and its contract is that the uncertainty layer and the lineage record survive encoding. The encoding-to-publication interface (I4) attaches a persistent identifier, an ISO 19115–3 metadata record with vocabulary-bound keywords, and a machine-readable licence (International Organization for Standardization, 2023a). The publication-to-reuse interface (I5) exposes the product through an open retrieval protocol, OGC API - EDR or OGC API - Records (Open Geospatial Consortium, 2023, 2024), and its contract is that metadata remain resolvable under A2 even where the data themselves are access-controlled.

Data flow through these interfaces is unidirectional in the production sense but carries two return paths that are essential and usually absent. The forward path runs from raw acquisition through processing, harmonisation, encoding and publication to reuse. The first return path carries reprocessing: because raw observations are retained with their calibration record at I1, a product can be regenerated when sound-speed modelling, geoid models or separation surfaces improve, and the framework requires that regeneration produce a new version under the same collection identifier rather than silently overwriting its predecessor. The second return path carries reuse feedback: discrepancies detected downstream, at crossings between surveys or against independent tide-gauge benchmarks, are written back as quality assertions against the published product rather than remaining in the downstream user’s workspace. Without these two paths the lifecycle in Figure 1 is a pipeline rather than a cycle, and the provenance that R1.2 requires is recorded only once, at publication, when in practice it accumulates throughout the product’s life.

Implementation is staged rather than wholesale, and the stages are usefully described as maturity levels against which an institution can locate itself. At Level 0 products are published as depth grids with narrative metadata, no uncertainty layer and no declared vertical datum realisation; this remains common for legacy and regional holdings. At Level 1 an uncertainty layer accompanies every depth value, the horizontal and vertical reference systems are declared by EPSG identifier where one exists, and a persistent identifier is minted at collection level; this is achievable without changing acquisition or processing practice and addresses F1, R1.1 and R1.3. At Level 2 metadata conform to the ISO 19115–3 encoding with vocabulary-bound keywords and a structured lineage record, and are registered in a catalogue reachable through OGC API - Records (Open Geospatial Consortium, 2024), addressing F2, F3, F4 and R1.2; this is the level at which holdings become discoverable outside their institution. At Level 3 products are encoded to S-102 Edition 3.0.0 or NetCDF-CF, exposed through OGC API - EDR, and carry qualified references to the water-level and current products that reduce them, addressing I1, I2, I3 and A1. At Level 4 the reprocessing and feedback return paths are operational, versioning is automated, and quality assertions from downstream reuse are ingested. The ordering matters: Level 2 without Level 1 produces discoverable products of undeclared quality, which is a worse outcome for the mariner and the modeller than undiscoverable products of known quality.

This framework is not proposed in a vacuum, and its relation to three existing efforts should be stated plainly. The EMODnet Bathymetry ingestion workflow already implements much of Levels 1 and 2 for its contributing partners, harmonising heterogeneous national submissions against controlled vocabularies and INSPIRE obligations, but it operates as a centralised aggregation in which harmonisation decisions are taken by the portal rather than declared by the producer, so the provenance of a datum transformation is often recoverable only as far as the point of ingestion. The IODE Ocean Data and Information System (ODIS) and its ODISCat catalogue address the discovery layer across the whole ocean domain through federated metadata brokering, and are stronger than anything hydrography has built at F4, but ODIS is deliberately agnostic about product-level quality semantics and therefore does not reach R1.2 or R1.3 for bathymetry. IHO C-17 provides the institutional guidance (International Hydrographic Organization, 2017) for publishing hydrographic holdings into spatial data infrastructures but is guidance rather than a conformance target, and offers no mechanism for asserting conformance. The contribution of the framework proposed here is to join what these three treat separately: it carries producer-declared quality and datum provenance forward through the same path that carries discovery metadata, so that a dataset arriving at ODIS or EMODnet already asserts, in machine-readable form, the survey order it meets, the datum realisation it uses and the uncertainty it carries. It is a specification of interfaces, not a new infrastructure, and is intended to be implementable within the workflows these three already operate.

10 Applying the framework: an illustrative walkthrough

The framework set out in Section 9 is specified rather than demonstrated, and this section traces a bathymetric dataset through it to show how the interfaces operate and which of the barriers identified in Sections 5 and 8 each one addresses. The walkthrough is illustrative: it follows a coastal survey of a type routinely undertaken in the Red Sea, chosen because the region combines a narrow shelf, rapid offshore deepening, sparse tide-gauge control and a well-documented history of heterogeneous vertical datums, which together make it an unusually clear case of the problems this review addresses. It is not a report of a particular survey, and no measured values are presented; parameters that would be observed in practice are shown in brackets.

At acquisition, a multibeam survey is conducted over a nearshore reef-fringed area intended to support both navigational charting and habitat mapping. The requirement at interface I1 is that no sounding leaves acquisition without the parameters needed to compute total propagated uncertainty: GNSS positioning mode and estimated positional uncertainty, vessel attitude and its uncertainty, transducer offsets from the survey control point, and the sound-speed profiles used, each with its time and position of measurement. In the illustrative case the survey is positioned by PPP or RTK, with stated horizontal and vertical uncertainty and referenced to ITRF realisation and coordinate epoch. The barrier addressed is the one identified previously: a survey whose frame realisation and epoch are not recorded cannot later be combined with repeat surveys without an unquantified horizontal offset, which on a plate moving 3 to 5 cm per year reaches 0.3 to 0.5 m over a decade and is material at Special Order.

At processing and quality control, soundings are cleaned and per-sounding total propagated uncertainty is computed from the vessel uncertainty model. CUBE is applied per node, using TPU to weight contributions and disambiguate competing hypotheses, as described in Section 5.5. Cross-line comparison provides the internal consistency check, and the retained record of rejected soundings and rejection criteria satisfies the provenance condition R1.2 rather than being discarded as processing residue. The contract at interface I2 is that the surface passing to harmonisation carries per-node depth and uncertainty together with explicit declarations of the horizontal frame, its realisation and epoch, and the vertical datum with the separation model used to realise it.

Vertical reduction is the stage at which the Red Sea setting is most instructive, and where the framework does the most work. Soundings acquired ellipsoidally must be reduced to chart datum for the charting product and are best retained on the ellipsoid for the scientific product. Two chains are involved and are kept separate, as Section 5.2 sets out: the orthometric height is obtained through a gravimetric geoid model, while the depth referred to chart datum is obtained through a modelled ellipsoid-to-chart-datum separation, SEP = h - CD, realised from tidal modelling constrained by such tide-gauge observations as exist. Where tide-gauge control is sparse the uncertainty of the separation model is typically the dominant term in the vertical budget, exceeding the sounding uncertainty itself, and the framework requires it to be carried explicitly rather than absorbed. Because a regional separation bias is systematic across a survey block, it is propagated as a bias term rather than combined in quadrature, since it does not reduce with averaging. The datum epoch is recorded alongside the datum name, without which the surface is not identified. This is the point at which the framework prevents the failure, where a per-node uncertainty layer that omits the transformation terms understates the true figure invisibly and unrecoverably.

At gridding and encoding, the surface is produced at a posting appropriate to the sounding density, and two products diverge at interface I3. For navigation, an S-102 Edition 3.0.0 dataset is encoded as S-100 HDF5 under Part 10c (International Hydrographic Organization, 2024c), with the QualityOfBathymetryCoverage feature populated to carry the uncertainty, data assessment and feature detection attributes, discovery metadata conforming to Part 17, and a Digital Certification Block under the S-100 Data Protection Scheme. For scientific reuse, a BAG surface retains the elevation and uncertainty layers and the tracking list of cartographic edits, or a NetCDF-CF encoding is produced with CF standard names and grid mapping attributes (CF Conventions Committee, 2024). The distinction drawn in Section 6.5 is operative here: these are not alternative containers for one thing but different artefacts with different obligations, and the framework requires both to be derivable from the same harmonised surface rather than one being converted into the other after the fact.

At metadata generation and publication, interface I4 requires a persistent identifier minted at both collection and granule level, an ISO 19115–3 metadata record with keywords bound to a named thesaurus (International Organization for Standardization, 2023a), a structured LI_Lineage record carrying the processing history including the separation model and its version, quality reported against ISO 19157–1 measures (International Organization for Standardization, 2023b), and a licence expressed as a URI to a Creative Commons instrument rather than as prose. The survey order achieved is stated as a machine-readable claim rather than asserted in an accompanying report. At interface I5 the product is exposed through OGC API - EDR for retrieval (Open Geospatial Consortium, 2023) and registered in a catalogue reachable through OGC API - Records, with the metadata remaining resolvable under A2 even if the data are subsequently access-controlled.

At discovery and reuse, a third party locates the dataset through the catalogue, retrieves it by identifier, and determines from its metadata alone whether it is fit for an intended purpose: a modeller establishes the vertical datum, its epoch and the separation uncertainty, and can therefore decide whether the product can be merged with an adjacent survey; a habitat analyst establishes the posting, the per-node uncertainty and the feature detection capability. Discrepancies found at the boundary with a neighbouring survey are returned as quality assertions against the published product along the reuse-feedback path described in Section 9, and improvements to the regional geoid or separation model trigger regeneration along the reprocessing path, producing a new version under the same collection identifier. It is at this last stage that the difference between the framework and conventional practice is clearest: the questions the reuser must answer are settled from machine-readable metadata rather than by correspondence with the producing office, which is what interoperability means in operational terms, and each of the six levels distinguished had to hold for it to be possible.

Interoperability of bathymetric data in ocean observing systems also improves the ocean and coastal modelling capabilities (Figure 5). High resolution, standardized topo-bathymetric digital terrain models allow researchers to get better model initialization and greatly minimize depth-related numerical errors that can cascade through hydrodynamic simulations (Seyfried et al., 2024). These interoperable datasets help to create more accurate computational grids and to give reliable boundary conditions necessary for comparison of results between different modelling platforms (Eakins et al., 2015). Interoperability plays a pivotal role in the accurate simulation of coastal inundation, storm surge, and tsunamis for coastal-hazard forecasting (Drennon et al., 2023). Wave transformation and sediment-transport modelling require a seamless connection between land and sea elevation data, which is only possible by combining harmonised hydrographic standards and FAIR data practices. Integrated FAIR compliant seabed information is also beneficial to marine spatial planning (MSP) for data-informed decisions on site suitability for offshore-energy development and strategic positioning of submarine cables (Vallius and Kaskela, 2024). By providing a consistent evidence-based argument across a range of administrative jurisdictions and levels, interoperability enables planners to strike a balance between route-of-shipping requirements and the need for habitat protection and the designation of marine protected areas (MPAs) (Vallius and Kaskela, 2024). Moreover, the shift to the IHO S-100 system will change the way sailors navigate and stay safe, as it allows high-definition bathymetric-surface products to be integrated with other associated layers in the marine environment. This structural uniformity enables navigation products to be consistently furnished with quality and uncertainty information, enabling faster updates and enhanced situational awareness for the mariner.

Standardized bathymetry is the key for classification of seafloor and benthic habitat modelling in ecosystem and habitat studies. The integration of bathymetric data with biological and chemical observations is essential for high-quality coral reef mapping and assessment of biodiversity, which is useful for effective fisheries management (Wolfl et al., 2019). Shared national and regional infrastructures, such as EMODnet, enable the development of interoperability among organizations, which supports cross-agency data sharing and dramatically decreases the need for redundant duplication of costly hydrographic survey work. This synergy eases the incorporation of research and operational information, especially in case of emergencies in maritime and coastal areas (Stewart et al., 2019). Lastly, these practices enable the empowerment of global ocean-mapping efforts, such as Seabed 2030, to collate standardized datasets from all over the world. Better documentation of data gaps and systematic reuse of legacy datasets via interoperable portals enable the global community to come closer to a complete and transparent map of the world’s oceans (Drennon et al., 2023).

11 Actionable recommendations

The adoption of interoperable bathymetric data needs a coordinated and multi-stakeholder effort to connect the dots between the traditional hydrographic production and the modern ocean observing systems where FAIR principles are applied. The main goal for the use of the machine-actionable output is to help the transition from human-centric to machine-actionable outputs, which is the main mission of the hydrographic offices to adopt standardized, nonambiguous formats for publishing horizontal and vertical reference information. Important elements of this transition include the provision of uncertainty metrics with all depth measurements, compliant with conventions like IHO S-44 or S-102, and the use of persistent identifiers (PIDs) such as DOIs, which will ensure that datasets are both discoverable and uniquely citable by time (International Hydrographic Organization, 2024a). In addition, these offices should enforce strict versioning and ensure there is detailed documentation of processing and transformation techniques to guarantee data integrity when reusing the data downstream (RDA FAIR Data Maturity Model Working Group, 2020). Bathymetric needs must be built into the system design for ocean observing programs. This includes the use of common metadata profiles and the use of standards to connect bathymetry to sensor and platform observations. This integration enables the assessment of the fitness of the data for use in applications that are sensitive to it, such as hydrodynamic modelling and coastal hazard assessment. As primary custodians of this data, repositories must adapt to accommodate contemporary web-based discovery and access protocols, including OGC API-EDR, and keep metadata persistent when access to the data is limited (Li et al., 2023). They also are required to create machine-readable licenses and to connect datasets to software and methods documenting their provenance (Wilkinson et al., 2016). The task of improving the semantic consistency of the relationship between hydrographic models and FAIR assessment indicators is the responsibility of standards organizations. Machine readable bridges between different standards are recommended as is the provision of more support for controlled vocabularies such as NERC Vocabulary Server (NVS) (Bloch Haimson et al., 2025). Individual researchers should refrain from publishing “orphaned” depth grids without documentation of sources, should record all the procedures involved in interpolation and transformation, and maintain uncertainty over the whole data life cycle (Bloch Haimson et al., 2025). Lastly, funding agencies are leading this cultural change, making it necessary to require strong FAIR data-management plans and to allocate resources for legacy data rescue and training for low-resource institutions, to enable a fair transition to interoperable ocean data (Wilkinson et al., 2016).

12 Limitations

Several limitations qualify this review. The crosswalk combines normative mappings traceable to specific clauses with mappings resting on documented practice or on the interpretation of the present work; although the evidential basis of each is stated, the crosswalk remains a proposal and carries no endorsement by any standards body. It is also unvalidated by implementation, no dataset having been encoded against it. The FAIR deficiencies identified here are those reported in the literature rather than measured; their distribution and severity would require a quantitative assessment of a defined sample of holdings against established maturity indicators. The proposed framework is likewise illustrative rather than validated, and no claim is made that its adoption improves interoperability outcomes at scale. Coverage of a fragmented and partly non-bibliographic literature cannot be assumed complete, and material held in hydrographic office reports, standards-committee papers and unindexed technical documentation may be under-represented. FAIRness and hydrographic fitness for purpose remain distinct properties, and a dataset may satisfy one while failing the other. Because hydrographic specifications are under active revision, statements concerning particular editions have limited currency. Finally, interoperability is constrained by institutional mandate, licensing, national data policy and funding as much as by technical specification, and these constraints are not soluble by standards alone.

13 Conclusion

The persistent fragmentation of bathymetric data across national hydrographic offices, research institutions and private organisations remains a significant constraint on the effective integration and use of bathymetric information within global ocean observing systems, generating redundant survey effort and data products that cannot readily be combined. This review set out to answer five questions, and the evidence assembled supports the following answers. First, the factors that constrain interoperability are not principally the absence of standards: they are the heterogeneity of vertical reference realisation, the loss of uncertainty and provenance information at the boundaries between processing stages, the absence of machine-resolvable identifiers for datasets and for the vocabulary of the domain, and institutional and licensing constraints that technical specification cannot reach. Second, existing hydrographic and geospatial standards contribute unevenly: the element-level mapping in Table 3 indicates that the reusability conditions are well served, through mandatory paired depth and uncertainty encoding and quantified survey-order thresholds, while findability is addressed only incidentally, since identifiers specified within the S-100 framework are unique to an exchange set rather than globally resolvable. Third, the FAIRness of bathymetric workflows cannot be stated with confidence from the present evidence base: the literature reports recurrent deficiencies in persistent identification, licensing and provenance, but no systematic assessment of bathymetric holdings against maturity indicators has been published, and this review does not supply one. Fourth, the principal gap where standards exist is semantic rather than syntactic: encoding is well specified, whereas the authoritative terminology of hydrography is not published in dereferenceable form, so machine agents cannot resolve what a declared datum or quality attribute means. Fifth, the framework proposed in Section 9 sets out one way in which hydrographic standards and FAIR practice may be combined operationally, by specifying the interfaces at which quality and datum provenance must be carried forward rather than by adding new infrastructure.

A distinction should be drawn between what these answers establish and what follows from them only as expectation. The reviewed literature supports the characterisation of the barriers and the mapping of standard elements onto FAIR sub-principles; it does not establish that adopting the proposed framework improves interoperability outcomes, because no such adoption has been evaluated. The claim advanced here is therefore that integration of hydrographic standardisation with FAIR practice provides the mechanisms through which structural soundness, semantic richness and machine-actionability may be achieved and, critically, tested - not that it delivers them. Whether it does is an empirical question, to be settled by assessment against indicators rather than inferred from the encoding a dataset uses. The next steps follow directly: a quantitative FAIR assessment of a defined sample of bathymetric holdings, publication of hydrographic vocabularies as dereferenceable identifiers, and implementation of the crosswalk in at least one production workflow. As the community moves toward the Seabed 2030 objective, the value of interoperable bathymetry for coastal-hazard forecasting, marine spatial planning and navigational safety is considerable, and is offered here as a well-founded expectation rather than a demonstrated outcome.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. Grammarly were used to improve the grammar, clarity, and readability of the manuscript. BioRender (https://www.biorender.com/) was used for figure preparation. The authors reviewed, edited, and approved the final manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: B. Praveen Kumar, Indian National Centre for Ocean Information Services, India

Reviewed by: Ahmed Zaki, Delta University for Science and Technology, Egypt

Alina Corina Bala, Politehnica University of Timişoara, Romania