Bathymetric data were acquired during the Red Sea Decade Expedition (RSDE) from February to June 2022 aboard the R/V OceanXplorer (Fig. 1), with additional shallow-water coverage collected using the support vessel Metalshark38. The Multibeam echosounder system (MBES) survey was conducted opportunistically, primarily during night operations and while transiting between sites. As the expedition's primary objectives focused on sediment sampling, biological surveys, and environmental characterisation using ROV and crewed submersible dives, MBES surveys were designed to provide local- to regional-scale seafloor context to inform dive planning and sampling locations. Accordingly, survey line placement, orientation, and spacing were adjusted dynamically to accommodate operational priorities, dive schedules, and environmental conditions. This mission-driven acquisition strategy produced spatially heterogeneous bathymetric coverage, with higher-density mapping concentrated around high-activity dive and sampling areas and comparatively limited coverage in other regions (Fig. 1).

Deep-water bathymetry was acquired using a hull-mounted Kongsberg EM 304 multibeam echosounder (20–32 kHz) installed on the R/V OceanXplorer. Shallow-water bathymetry (generally < ∼ 100 m) was collected using a Teledyne RESON SeaBat T50-P multibeam echosounder (190–420 kHz) mounted on Metalshark38, an 11.6 m support vessel deployed from the main ship (Fig. 2a). For both platforms, vessel positioning and attitude were provided by integrated GNSS and motion reference systems.

Sound-velocity profiles used for multibeam correction were collected from 115 expendable bathythermograph (XBT) deployments conducted throughout the expedition. Profiles were collected opportunistically along survey lines to capture spatial and temporal variability in water-column structure, then stored and processed in Sound Speed Manager (Masetti et al., 2020) before being imported into Qimera during bathymetric processing.

MBES data were processed post-expedition at Habitat and Benthic Biodiversity Laboratory (HaBB), King Abdullah University of Science and Technology (KAUST) using Qimera v2.4.1 (QPS, Netherlands). Raw MBES files from all survey platforms were processed with a standardised workflow to ensure internal consistency across Legs, Phases, and acquisition systems, and to generate bathymetric surfaces suitable for regional- to local-scale analyses (Fig. 2).

Given the overall volume of the dataset and the spatial extent of the survey, MBES processing was distributed across multiple operators. Four dedicated processors independently handled the data from individual survey Legs for both the OceanXplorer deep-water acquisition and the Metalshark38 shallow-water surveys. A fifth processor coordinated the overall workflow, integrating and quality-checking the individual projects to ensure consistency across the full dataset.

This distributed processing strategy was implemented using the project-import functionality of Qimera, which allows projects to be exchanged as QPD files. Individual processors worked on separate project instances while periodically exporting updated QPD files that were imported into a central merged project. This configuration enabled near-real-time oversight of processing progress, cross-checking of cleaning decisions, and harmonisation of processing parameters across all Legs.

Sound-velocity corrections were applied using profiles processed with Sound Speed Manager (SSM), and navigation and altitude data were imported and verified. Tide corrections were applied using Red Sea tide stations available from the General Authority for Survey and Geospatial Information of Saudi Arabia (GASGI) database. Survey lines were then systematically inspected for timing offsets, navigation artefacts, and other acquisition issues prior to cleaning. Processing steps, including sound-velocity correction, manual and automated filtering, and adaptive cleaning based on depth and slope, were applied to minimise artefacts commonly associated with MBES data, such as outer-beam noise and slope-related distortions.

Prior to data acquisition, patch tests were performed for MBES systems to determine the angular mounting offsets of the transducer arrays relative to the vessel reference frame. The resulting roll, pitch, and heading offsets, as well as time latency values, were applied to all raw data during processing and are stored in the respective vessel configuration files in QPS Qimera. Table 1 summarises the calibration offsets for each system.

After filtering processes, the accepted soundings were gridded into seven Digital Bathymetric Models (DBMs). Two fixed grid resolutions were applied: 5 m for shallow-water data acquired by the T50-P aboard Metalshark38, and 40 m for deep-water data acquired by the EM 304 aboard the R/V OceanXplorer, reflecting differences in depth, survey design, and line spacing among Legs and Phases (Table 2). Across the 2022 RSDE survey, a total of 2863 MBES survey lines were processed, producing bathymetric coverage of 49 418.7 km² (Table 2). High-resolution shallow-water mapping collected from Metalshark38 contributed 133.8 km² of gridded bathymetry, while deep-water mapping from the R/V OceanXplorer contributed 49 284.9 km². Figure 3 shows the full extent of the gridded bathymetry across the Red Sea (panel a), which integrates all seven DBMs, alongside eight representative subsets (panels b–i), two per expedition Leg. These enlarged panels highlight the diversity of geomorphological settings captured by the survey. Erosive and gravity-driven processes are widespread along the margin, from gully fields incising the upper slope (panel b) to canyon flanks affected by slope instability (panel d) and a mass-wasting deposit related to a submarine landslide (panel g). Salt-related morphologies record the mobility of the widespread Miocene evaporites, expressed as a salt-flow lobate front (panel c) and as compressional ridges and flow lineaments related to salt movement (panel f) (Feldens and Mitchell, 2015; Mitchell et al., 2010). Volcanic and tectonic features characterise the deep central axis, including volcanic edifices and extensional lineaments (panel h) (Augustin et al., 2016a) and hummocky deposits related to lava flows in the Mabahiss Deep (panel e) (Fittipaldi et al., 2025b). Finally, panel (i) shows a pockmark field formed by fluids expelled through the underlying evaporite sequence (Feldens et al., 2016).

In the absence of pre-existing processed bathymetric datasets in the surveyed areas, all cross-check analyses are self-consistency assessments in which the survey data are compared against reference grids derived from the same acquisition. Bathymetric studies in unmapped or poorly constrained areas commonly assess data quality through internal self-consistency metrics such as repeated surveys, overlapping swaths, and inter-pass disparity, rather than through comparison against independent grids (Calder and Mayer, 2003; Mohammadloo et al., 2020; Sauter et al., 2025).

Internal consistency of the processed bathymetric data were assessed using the cross-check tool in QPS Qimera, which computes the depth bias between individual soundings and a reference grid constructed from overlapping survey lines. Specifically, the analysis quantifies inter-pass disparity, defined as the vertical variability between overlapping soundings acquired during different survey passes or under different swath geometries (Sauter et al., 2025). By comparing these internal repetitions, the dispersion of the data is used as a proxy for total measurement uncertainty, ensuring that the final DBMs maintain high spatial coherence even in the absence of independent ground-truth datasets. Cross-check analyses were performed in one representative sub-area per MBES for both the first and last legs of the campaign (Leg 1 and Leg 4), providing a systematic assessment of data quality at the beginning and end of the five-month expedition. Validation was restricted to areas where survey line geometry allowed robust statistical evaluation, specifically where parallel survey lines were intersected by perpendicular or near-perpendicular line. The use of cross-lines maximises the detectability of inter-pass disparity. The selected sub-areas encompass the range of depth intervals and acquisition conditions encountered during the campaign and provide a consistent basis for evaluating internal data quality and system stability.

For the deep-water EM 304 dataset, cross-checks were conducted in areas L1P1 (Leg 1, February 2022) and L4P1 (Leg 4, May–June 2022), covering depth ranges of 569–1682 and 617–1280 m, respectively. Results were compared against the depth-uncertainty limits (THU and TVU) of the IHO S-44 Exclusive Order, used here as a stringent numerical benchmark rather than as an applicable survey order, since these limits are the tightest defined in the standard. For the shallow-water T50-P dataset, cross-checks were conducted in areas L1P2 (Leg 1, February 2022) and L4P2 (Leg 4, June 2022), covering depth ranges of 44–136 and 32–186 m, respectively, and evaluated against IHO S-44 Order 1, applicable to coastal and shelf environments. For all datasets a consistent geometry of survey lines was selected for the cross-check, a set of parallel lines crossed by a perpendicular one. The comparison grid has the same resolution of the final processed bathymetry, 40 m for EM304 and 5 m for T50P.

In all four areas, the cross-check tests were accepted under the respective IHO standards (Table 3). For the EM 304 system, depth bias means were 0.055 m (L1P1) and 0.077 m (L4P1), indicating negligible and mutually consistent systematic offset across the full duration of the campaign. For the T50-P system, depth bias means were 0.000 m (L1P2) and -0.008 m (L4P2), with standard deviations of 0.337 and 0.246 m, the lowest values across all four validation areas, consistent with the higher sounding density and finer grid resolution achievable in shallow water. The close agreement between Leg 1 and Leg 4 results for both systems, separated by approximately five months of continuous operations, demonstrates the stability of the system calibrations over time.

All bathymetric grids are projected in WGS 84/UTM Zone 37N (EPSG:32637). The vertical datum is the Lowest Astronomical Tide (LAT). Tidal corrections were applied using predicted tide heights from the Saudi Tide Tables 2022 (Volume I: Red Sea and Gulf of Aqaba), published by the General Authority for Survey and Geospatial Information of Saudi Arabia (GASGI), in which all tidal heights are referenced to LAT.

Coverage is spatially heterogeneous across the dataset, reflecting the mission-driven acquisition strategy of the expedition: survey lines were planned dynamically to support dive operations and sampling activities rather than to achieve uniform areal coverage. Higher sounding density is therefore concentrated around dive and sampling sites, while other regions have comparatively sparse or absent coverage. In addition, typical sources of uncertainty inherent to multibeam echosounder data should be considered when using this dataset. These include depth-dependent uncertainty scaling, associated with increasing beam footprint size and decreasing signal-to-noise ratio with depth, as well as reduced accuracy at outer swath angles. Furthermore, steep seafloor slopes may introduce local artefacts due to beam geometry and interpolation effects during gridding. The increase in standard deviation with depth observed in the cross-check analysis is consistent with these well-established characteristics of MBES systems.