Mapping and visualizing shipwrecks in high resolution

In order to ensure safe navigation, Flemish Hydrography (FH) uses multibeam echosounders to precisely determine the depth and shape of shipwrecks. After a detailed survey and processing of the data, results are presented in an online wreck database. In recent years, the survey methodology has been refined and enhanced to deliver the highest possible level of detail and reliability using the available in-house equipment, keeping pace with technological evolutions in the world of multibeam echosounders. This article describes several acquisition and processing settings specifically for wreck surveying and their impact on the resulting data.

Flemish Hydrography (FH), part of the Coastal Division of the Flemish Agency of Coastal and Maritime Services (MDK), serves as the Belgian Hydrographic Office and is a member of the International Hydrographic Organization (IHO). Among its key responsibilities is the charting of shipwrecks in the Belgian sector of the North Sea and in the tidal part of the Scheldt River.

To support this, FH has established a resurvey strategy that determines the sequence of wreck surveys. Prioritization is based on weighted parameters such as proximity to fairways, the date of the most recent survey, surrounding depths, wreck length and whether the site represents a complete wreck or only scattered remains. Based on these criteria, survey planning is carried out.

During routine multibeam hydrographic surveys, large areas are mapped with sufficient data to populate a 1x1m grid after processing. This approach allows larger objects and shipwrecks to be charted. Additionally, it is essential to use an accurate sound velocity profile during surveying, and to use correct patch values for the multibeam transducer(s). Moreover, to capture the finer details and characteristics of a wreck, the acquisition settings of the multibeam system must be optimized to achieve the highest possible resolution. Several acquisition and processing settings specifically for wreck surveying are outlined below, including details of their impact on the resulting data.

Dimensional control

A dimensional control survey provides highly accurate information on the offsets of all survey equipment and reference points within a dedicated vessel reference frame. This survey is typically conducted using a total station or photogrammetry, and must be complemented by careful calibration of heading and attitude sensors. When performed correctly, this ensures confidence in the setup. Conversely, inaccurate dimensional surveys – for example, those relying on tape or rod measurements – introduce small errors that can result in incorrect seabed positioning. During wreck surveys, where multiple survey lines are sailed back and forth, such errors cause fine structures to misalign between lines of differing headings, leading to blurred details.

Line plan

Establishing an effective survey plan for each wreck is essential. Previous studies (Westley et al., 2019) have examined optimal strategies, but practical experience within Flemish Hydrography shows that the best approach is to sail lines along the central axis of the wreck, covering it with as many back-and-forth passes as possible. Additional parallel lines spaced closely (5-10m) help capture vertical structures that may appear blurred in one line but are digitized in another. Sailing lines perpendicular or at an angle to the wreck’s axis tend to scatter seabed points in a less coherent manner, complicating interpretation. Moreover, it is advised to sail as slow as possible to maximize hit count and to sail straight survey lines. This is easier during good weather conditions.

Beam spacing

Most multibeam manufacturers provide the option to use equiangle or equidistant beam spacing or a high-density mix of both. For wreck surveys, equidistant spacing is clearly preferable, as it ensures evenly distributed points across the seabed. This is crucial since every detail of a wreck is significant. While repeated passes provide sufficient point density, equidistant spacing produces a more consistent and visually coherent seabed pattern compared to equiangle spacing, which concentrates points near nadir and sparsely distributes them toward the outer beams.

Frequency

It is commonly known that a higher frequency gives higher resolution. This improvement is not due to the number of recorded points, but rather to the narrower beam-opening angles at higher frequencies, which enhance seabed detection accuracy within each beam. Several manufacturers now offer very high-resolution multibeam systems that are particularly advantageous for wreck surveys, as can be seen in Figure 1.

Backscatter, water column and extra detections

Both backscatter and water column data have been recorded and evaluated over multiple shipwrecks but did not deliver significant additional extra information. Backscatter is more effective over large areas with varying sediment types, while wrecks are generally too small to reveal additional information. Water column data is promising, as several echoes in the water column can also be seen. This means that vertical structures appear more clearly than in standard bathymetry, but extracting meaningful new detections remains challenging. Advances in processing software may eventually provide effective workflows for these datasets.

On the Kongsberg EM2040 multibeam system, an ‘Extra Detections’ option can be activated which divides the water column into user-defined classes where new detections are digitized. In theory, this could significantly improve the mapping of vertical seabed structures. Tests have shown mixed results; upper classes often contain false echoes, while classes closer to the seabed contain potentially valid detections (see Figure 2). Some of these are clearly linked to real structures, while others are not, making the outcome uncertain. Further research is required, but the option remains promising.

Processing settings

During wreck surveys, most online filters should be disabled to maximize the number of detections, and false echoes may be mixed with real data depending on local conditions. As a result, meticulous manual processing is required (Figure 3), examining the wreck slice by slice. Unlike standard bathymetric data, which is nowadays processed automatically, wreck surveys demand manual interpretation to distinguish wreck structures from algae, nets or debris. Skilled surveyors are essential, although future integration of artificial intelligence into processing software is expected to automate these tasks.

Visualization

Gridding vs. point clouds

Bathymetric datasets can be presented as grids/meshes or point clouds. For detailed objects such as wrecks, point clouds are preferable, as they preserve the true position of points without blurring caused by triangulation or gridding. Shading is also critical to give the results a type of 3D feel.

Several shading techniques were tested in BeamworX AutoClean and CloudCompare (open-source 3D point cloud and mesh processing software). The CloudCompare Portion de Ciel Visible (PCV) ShadeVis plugin produced the best results, while Eye Dome Lighting (EDL) and SSAO also proved useful (see Figure 4).

Colour scale

Choosing an appropriate colour scale is essential to represent wrecks both effectively and naturally. Data must be presented fairly, without distortion, and remain universally interpretable – even in black and white – while being reproducible. Flemish Hydrography therefore adopted the Lapaz colour scale, one of the scientific scales developed by Fabio Crameri.

Foreseeable technological evolutions

Multibeam manufacturers are introducing high-frequency systems with unprecedented point densities, achieved either through larger transducers with more elements, or through dual, quad and even octo-swath configurations. At the same time, AI algorithms for data cleaning are being developed, which will also benefit high-detail wreck surveys. Monitoring and testing these technological advances is crucial for all multibeam users.

Conclusions

Shipwreck surveys require a tailored approach distinct from regular seabed surveys. Key considerations include: using the highest possible frequency, sailing slowly to maximize hit count, maintaining a low opening angle, disabling most online filters, applying accurate sound velocity profiles and patch values, using equidistant mode, processing data carefully, sailing as many lines as possible along the wreck’s longitudinal axis, only survey during good weather conditions, and maintaining straight survey lines. When these criteria are met, wrecks can be surveyed with exceptional detail, as can be seen in Figure 6. Lastly, surveyors should remain attentive to ongoing developments in hydrographic technology.

About the online wreck database

The online database contains all documented shipwrecks in the Belgian part of the North Sea and will soon be expanded to also include those located in the Belgian section of the Scheldt River. The website (see Figure 5) provides both a map view, showing the exact position of each wreck, and a list view with a detailed metadata table. Users can perform a custom search to display only the wrecks relevant to their interests.

Each wreck has its own dedicated page featuring all available metadata, historical records, multibeam images and an interactive 3D visualization module. Information can be filtered according to user preferences, and the complete dataset can be exported as a CSV file, making it easy to use in other applications.

Further reading

Flemish Hydrography: https://www.afdelingkust.be/en/flemish-hydrography

Online wreck database: https://wrakkendatabank.afdelingkust.be/

Crameri, F. (2023). Scientific colour maps (v8.0.1), Zenodo. https://doi.org/10.5281/zenodo.8409685

Westley, K., Plets, R., Quinn, R. et al. Optimising protocols for high-definition imaging of historic shipwrecks using multibeam echosounder. Archaeol Anthropol Sci 11, 3,629-3,645 (2019). https://doi.org/10.1007/s12520-019-00831-6