Multibeam Sonar 101: From Raw Pings to Accurate 3D Seafloor Models
Multibeam echosounders have become the standard tool for mapping harbours, channels and coastal structures in New Zealand. They fire a fan of acoustic beams that sweep a wide swath of seabed on each pass. With careful planning, disciplined fieldwork and a deliberate processing workflow, those raw pings become clean digital elevation models and meshes that engineers, councils and scientists can actually use. This article walks through the full journey: how multibeam works, how to design a survey, what happens in processing, where uncertainty creeps in, and which deliverables make decisions easier for your stakeholders.
How multibeam works
An MBES transducer sends out many narrow beams arranged in a cross-track fan. Each beam returns a two-way travel time that converts to range using the speed of sound in water. Combine that range with the beam angle and the vessel’s motion and position from an integrated GNSS and inertial system, and you can compute the three-dimensional point on the seabed where that beam hit. Thousands of these points arrive every second. When the system is correctly aligned and the speed of sound structure is well modelled, the fan builds a dense, geometrically consistent point cloud.
Multibeam is worth the effort when you need true full-coverage mapping rather than spot depths, when deliverables must support decisions such as berth clearance or dredge volumes, and when you’ll return to the same site to track change. It also plays well with backscatter processing so you can infer relative substrate types and focus divers or ROVs where they matter most.
Designing a survey that processes cleanly
Good outputs start long before you push off the dock. Decide the objective and uncertainty target up front, because that decision drives line spacing, overlaps, and grid resolution later. If you’re aiming for reliable detection of small changes around piles or an outfall, you will run tighter lines and accept slower production. Choose a vertical strategy early and document it: in many New Zealand projects that means ellipsoid heights tied to NZVD2016, or chart datum via a separation model. Consistency here saves hours in post-processing.
Every project should include a patch test before production lines. The patch estimates latency and tiny angular misalignments between sensors (roll, pitch and yaw). It is common to re-run the patch after any mechanical change or if results suggest an alignment drifted. Once you lock those values, your production lines can focus on coverage and data density rather than fighting geometry.
Field acquisition in practice
On the water, the routine is simple but strict. Use a hull-mounted sound-velocity sensor to stabilise near-surface steering and cast a profiler whenever the water mass changes. Keep speed and heading steady through the line; re-run anything with dropouts or erratic motion. Record event marks as you pass structures, suspected outfalls or anomalies so they are easy to relocate in post.
From pings to product
Processing starts with ingestion and synchronisation. Bring in the sonar, GNSS/INS and sound-velocity data, then confirm timestamps, lever-arm offsets and the patch test values you intend to use. Apply sound-speed corrections at two levels: the at-head sensor that stabilises beamforming, and ray-tracing through your most recent sound-velocity profile to correct refraction through the water column.
With geometry under control, move to cleaning. Automated spike, angle and range filters are excellent at removing the worst outliers, but manual passes are still important around structures, steep slopes and in areas with suspended matter. Choose a cell size tied to your depth and beam footprint—harbour projects often land around 0.25 to 0.5 metres—and grid a surface that reflects your uncertainty target. If backscatter is part of the brief, process a mosaic in parallel so you can correlate shape with relative hardness and roughness.
Quality assurance threads through the whole pipeline. Crosslines provide an independent check on consistency. A total propagated uncertainty (TPU) surface should rise gradually toward swath edges and sit lowest under the nadir. Visual artefacts such as roll stripes, smiles and ripples are clues that something upstream needs attention.
Meshed bathymetry makes gradients and channel shapes easy to read for non-GIS users.
Resolution, cell size and the real trade-offs
Grid resolution is always a negotiation between purpose, physics and file size. In shallow harbours between two and fifteen metres, a quarter-metre grid commonly captures the detail users expect while keeping files manageable. Channel approaches down to thirty metres often sit at half-metre to one-metre cells. Offshore work can be coarser again. If your question involves scour or clearance near structures, err on the finer side; if the task is reconnaissance or change-detection across a wide area, accept a slightly larger cell size and rely on consistent line planning to deliver confidence.
Accuracy and uncertainty without the jargon
Every survey is a balance between random noise and systematic error. Sea state, outer-beam signal-to-noise and GNSS variance add randomness you can manage by choosing weather windows and trimming swath angles. Systematic problems come from mis-aligned sensors, incorrect sound-velocity models, timing errors or inconsistent vertical control. A good result shows small crossline differences, a smooth surface free from striping, and uncertainty that behaves as expected from the centre of the swath to the edges. Keep a concise QA note that lists patch results, GNSS/INS metrics, the timing and position of each SVP cast, and summary crossline statistics.
Deliverables people actually use
Different teams read the data in different ways. Engineers and councils appreciate a GeoTIFF bathymetry grid with a matching hillshade, slope map and contours in common GIS formats. Project archives often include a point cloud in LAZ or XYZ for anyone who needs to interrogate raw returns. Meshes are powerful for communication because gradients and channel shapes appear immediately without specialist GIS software. When you revisit a site, difference-of-DEM maps make change obvious. A short PDF quicklook with a legend, coordinate system and vertical datum on the first page saves back-and-forth and helps non-technical readers understand what they are seeing.
Common pitfalls and how to avoid them
The issues we see most often are sound-speed drift through a changing water column, vertical datum confusion when ellipsoid heights and chart datum are mixed without a separation model, and over-wide swaths that chase coverage at the expense of edge noise. Projects also suffer when teams skip crosslines or keep using old patch values after mechanical changes. None of these are fatal if you plan for them. Cast when the water mass changes, choose the vertical strategy on day one, trim outer beams when noise rises, and make crosslines part of your standard line plan.
Biosecurity, structures and why backscatter helps
For biosecurity patrols such as Caulerpa surveillance, bathymetry shows the shape of the habitat while backscatter hints at the substrate. Hard structures, sheltered pockets and low-energy zones stand out. Combining the two lets you focus divers or ROVs on targets that are most likely to harbour growth, and it also gives you a baseline for later change-detection surveys.
A lightweight toolchain
Acquisition is handled by the MBES vendor suite working alongside the navigation system. Processing tools range from CARIS, Qimera and Hypack/Hysweep to open options such as MB-System. For GIS and 3D, QGIS, ArcGIS Pro, CloudCompare and PDAL cover most needs, and a simple Cesium or Three-based viewer helps with web delivery. For quick communication and QA, we like Ākau3D because depth bands, lighting and simple profiles tell the story without overwhelming stakeholders.
A worked example from end to end
Imagine a small harbor inner channel. Import the sonar, navigation and sound-velocity data, confirm time sync and offsets, and verify the patch. Choose NZVD2016 for your vertical control or set up a separation to chart datum. Apply sound-speed corrections and run automated cleaning, then make a careful manual pass near piles, slopes and any obvious artefacts. Grid at a quarter-meter if you’re interested in local scour and clearance; otherwise choose a half-meter to one meter cell for a broader reconnaissance picture. Produce a hill shade and slope to complement the DEM, and if the scope includes substrate context, build a backscatter mosaic. Validate with crosslines and a quick TPU review. Export a GeoTIFF, contours, a compressed LAZ point cloud and, if communication is the priority, a lightweight mesh. Finish with a one-page QA summary that explains methods, uncertainties, datum and any limitations.
Equipment spotlight: Surveyor 240 MBES
The Surveyor 240 MBES is a compact, survey-grade multibeam echosounder designed for fast, full-coverage mapping in harbours, rivers, marinas and coastal projects. Its wide swath and tight beam geometry generate dense, clean soundings on every pass, turning short line plans into reliable bathymetry, contours and 3D meshes. Mounted on a small vessel or USV, it is an efficient way to certify berth pockets, quantify dredge volumes, locate outfalls and scour, and build baselines for repeat change-detection surveys.
Built for modern workflows, Surveyor 240 integrates with RTK GNSS and INS, supports sound-velocity profiling, and outputs standard formats for downstream processing including DEM/GeoTIFF, point clouds, contours and meshes. Oceanova can supply complete kits—head, topside, cabling and pole or hull mounts—and provide setup, patch-test support and post-processing guidance. Request an RFQ to confirm lead times, recommended accessories and an installation tailored to NZVD2016 or local chart datum.
If you want to explore the visual side and show results to non-specialists, our lightweight viewer is available here:
Ākau3D on GitHub - https://github.com/oceanovanz/akau
Surveyor 240 MBES- https://oceanova.nz/products/surveyor-240-mbes