Buy the best multibeam on the market, bolt it to the wrong boat, and you will collect expensive noise. The uncomfortable truth of hydrography and ocean survey is that data quality is decided by the platform as much as by the sensor. What follows is a tour of what the modern kit can really do, how deep, how accurately, and the concrete physical problems a survey platform has to solve to let it. It is written to be readable without a hydrography degree, but the numbers are real.
- Keeping the hull acoustically quiet
A multibeam echo sounder works by sending a fan of sound down at the seabed and timing the echoes across a wide arc. Its enemy is air. Cavitation bubbles and the sweep-down of aerated water from the bow get entrained under the hull, right across the transducer face, and because the sonar runs at high frequency (a modern shallow-water head sweeps roughly 150–700 kHz), even a thin curtain of microbubbles will scatter the beam and blank part of a swath.
There is a physical trade-off underneath this: higher frequencies give finer resolution but are absorbed faster, so they reach less deep and are more easily disrupted by bubbles; lower frequencies reach further but resolve less. The faster and harder the boat works, the more air it drags under itself, so the problem is worst exactly when you want to survey efficiently.
The solution is a hull decision, not a sensor decision: shape the flow under the transducer with computational fluid dynamics before the boat is built, recess the transducer into a faired sea-chest or gondola, and keep bow-generated bubbles from sweeping aft over it. Flow noise across the transducer face matters too, not just bubbles. Get this wrong and no amount of processing recovers the data, which is why it is the single most under-appreciated factor in survey-boat design.
- A structure stiff and silent enough to trust
A survey system is calibrated so it knows the exact offsets and angles between the sonar, the motion sensor and the GNSS antennas, down to millimetres and hundredths of a degree. If the hull flexes as fuel burns off or the sun heats one side, those offsets drift and the seabed smears. A carbon-fibre sandwich composite hull, built in one shot by vacuum infusion, is exceptionally stiff and dimensionally stable, thanks to carbon’s high stiffness-to-weight and very low thermal expansion, it holds its geometry better than a heavier steel or aluminium equivalent. It is also light, which buys two things at once: a shallow draft to work the shallows, and enough weight margin that a ~15.6 m platform can still be lifted onto a mother ship’s davit at around 15 tonnes and yet transit at close to 39 knots on twin 600 hp drives.
Noise and vibration matter too, but not in the way people assume. A multibeam works hundreds of kilohertz above engine noise, so the two rarely compete directly in the same acoustic band. What vibration really threatens is the calibrated sensor mount and the acquisition electronics, and a noisy cabin wears down a crew working a full survey shift. So the platform is engineered to decouple both: a floating floor separates the cabin from the hull to cut structure-borne noise, ProZero Research & Survey Workboats, SoMe Series and the transducer mount and electronics are isolated from machinery vibration. A gyro stabiliser adds to this by physically damping roll, which, as the next section explains, directly widens the part of each swath you can actually use.
- Knowing exactly where the sonar was, and how it was pointing, to the microsecond
A sounding is only as good as your knowledge of where the transducer was and how it was tilted at the instant of the ping. That job falls to a GNSS-aided inertial navigation system, the class of kit exemplified by an Applanix POS MV OceanMaster. It fuses two GNSS antennas with an inertial measurement unit so that heading is held to about 0.01° (given a GNSS antenna separation of roughly 4 m) and, crucially, is immune to the magnetic errors that plague a compass; heave is measured with no settling time; position is resolved to a few centimetres using RTK corrections in real time or post-processed kinematic (PPK) afterwards; and every value is time-tagged to the microsecond so there is zero latency between the navigation data and the acoustic data.
Why it matters is geometry. At the edge of a wide swath (±60° and beyond) a tiny attitude error is multiplied by the slant range, so a few hundredths of a degree can become decimetres of vertical error in deeper water. That is why the whole system has to be calibrated with a “patch test” before real work begins, running reciprocal lines over a slope, a flat and a discrete object to solve, in order, the timing latency and then the roll, pitch and yaw misalignment (the boresight) between the motion sensor and the sonar head. Latency alone, uncorrected, throws soundings along-track; even tens of milliseconds are visible in the data. The motion system and its calibration are what let a surveyor trust the outer beams instead of throwing them away.
- The sonar itself, how deep, how wide, how fine, and what else it sees
Now the sensor can do its job. A modern dual-head shallow-water multibeam, the Kongsberg EM-series is the reference point, maps from about 0.5 m under the keel to several hundred metres at 300 kHz, extending to around 600 m in a lower-frequency (200 kHz) mode (in shallow coastal work the useful band is the top 100 m or so). It is frequency-agile across ~150–700 kHz: around 300 kHz for general mapping, 400 kHz for structure inspection, and 600–700 kHz for small-object detection and fine 3D work, with beamwidths as tight as 0.4° by 0.7°. It fires up to 50 pings a second and returns up to around 1,024 soundings per ping in standard mode, rising to some 4,000 in quad-swath modes, with depth resolution of only a few millimetres. To place each of those soundings it detects the seabed differently across the fan, by echo amplitude near vertical, where the return is strong, and by phase (interferometry) out at the grazing angles where amplitude alone is unreliable.
The dual head is the point in shallow water. When the water is only a few metres deep, a single head’s fan touches only a narrow strip of seabed, so coverage collapses and you need many more passes; two heads angled outboard keep the swath very wide, up to around 220° with dual receivers, and the survey productive. Smart beam-steering (steering the beams toward the platform’s predicted pitch) keeps detections clean even on small, fast or uncrewed platforms that pitch sharply. And bathymetry is not all a multibeam records: the strength of each return (backscatter) is used to classify the seabed into sand, gravel, rock or reef, the raw material of habitat mapping, and the full water-column echo can reveal fish, gas seeps, wrecks standing proud or a mid-water target, turning a depth sensor into a broader observation tool.
- The sound-speed trap
Here is the problem that catches out the unwary: sound does not travel through seawater at a constant speed. It varies from roughly 1,450 to 1,550 m/s with temperature, salinity and pressure, and a beam launched at an angle bends as it crosses layers of different sound speed, temperature inversions, haloclines and freshwater lenses all create them. Assume the wrong profile and a flat seabed is recorded as a smile or a frown; the outer beams can be metres out while the centre still looks perfect.
The fix is two-fold. A sound-velocity probe at the transducer measures the surface sound speed continuously, which the sonar needs to form its beams at the correct angles in the first place. And sound-velocity casts through the water column let the processing software ray-trace each beam back to its true position. Because sound speed changes through the day and across a survey area, the casts have to be repeated often, which is why underway systems such as a moving-vessel profiler, dropping and recovering a probe while the boat keeps running, are used to avoid stopping the survey every time. It is unglamorous work, but it is the difference between a chart you can dredge to and one you cannot.
- Turning raw pings into a chart, datum and processing
Two more problems stand between the sonar and a usable chart. The first is vertical reference: a depth only means something relative to a datum. Traditional surveys tie soundings to tide gauges; modern practice increasingly uses ellipsoidally referenced survey, where the GNSS-INS gives height above the ellipsoid directly and a separation model converts that to chart datum, removing much of the uncertainty tide models introduce, especially offshore and in complex estuaries. The second is sheer volume: a day’s survey can produce hundreds of millions of soundings, far too many to check by hand. The industry standard, CUBE, Combined Uncertainty and Bathymetry Estimator, propagates the uncertainty of every sounding and uses it to estimate the most probable seabed, and it is increasingly paired with machine-learning cleaners that flag outliers a rule-based filter misses. The direction of travel is clear: more data, collected faster, cleaned more automatically, with the surveyor supervising rather than hand-editing every ping.
- Getting it in the water, protected, and increasingly, autonomous
The rest is integration discipline. Over-side universal sonar mounts on a rigid pole (a Z-pole of around 7.5 ft) deploy additional sensors clear of hull turbulence and retract cleanly for transit. The acquisition electronics live in a climate-controlled, dehumidified, shock-mounted 19-inch rack, because survey computers fail from heat, damp and vibration long before they fail from age; every cable run is sealed through certified penetrations; and the electrical system is engineered to stay quiet, because electrical noise ends up in the data just as surely as acoustic noise does.
And the platform is increasingly its own operator. An autonomy package, the Sea Machines class of system, with panoramic and bullet cameras lets the same boat run optionally-crewed or fully uncrewed along long, pre-planned survey lines, obeying the collision-avoidance rules of the road and holding an anti-grounding envelope, controlled over the horizon via satellite links. That is exactly the repetitive, tightly-planned work machines do better than tired people. In research programmes such as MERLIN (led by the University of Birmingham) the next step is being built now: platforms that don’t just map autonomously but grasp and sample the seabed themselves, turning a survey vessel from an observer into a tool that reaches in and acts.
The point
Every one of these choices is made before the sonar is switched on: hydrodynamics, stiffness, silence, motion reference and its calibration, sound-speed correction, vertical datum, protected and quiet electronics, autonomy. Get them right and an ordinary sensor delivers excellent data; get them wrong and the best sensor in the world cannot save the survey. That is the whole case for treating a survey vessel as an instrument that happens to float, whether it is mapping a harbour approach, monitoring a reef, or working the edge of an abyssal plain.
