GNSS at Sea, and Where It Stops

GNSS covers the surface of the ocean completely and the inside of it not at all. Every subsea positioning system in existence is a response to that second half.

What you get on the surface

A basic receiver gives a few metres. Fine for passage planning. Useless for survey.

Real-time kinematic and precise point positioning close the gap by correcting for the errors a single receiver cannot see: ionospheric delay, satellite clock and orbit error, tropospheric effects. RTK uses a base station at a known location, so it needs to be within range of shore infrastructure. PPP uses a global correction stream delivered by satellite, which works in the middle of an ocean where no base station could reach.

Offshore survey vessels run PPP for that reason. Centimetre-level positioning, hundreds of miles from land, without a shore reference.

Heading is a separate problem

A single antenna tells you where the vessel is. It does not tell you which way it is pointing.

That matters more than it sounds. Every sensor aboard sits at an offset from the antenna, and every measurement has to be referred back to a common reference frame. Get heading wrong and the entire survey rotates about the antenna.

Dual-antenna GNSS measures the vector between two receivers. A gyrocompass gets there independently. Most survey spreads run both and compare them, because a heading error nobody notices corrupts every product downstream of it.

Below the waterline

Seawater absorbs the GNSS frequencies within centimetres, so a submerged vehicle needs something else entirely. Three approaches, usually combined.

USBL puts a transducer array on the vessel hull and a transponder on the vehicle, then works out the vehicle's position from the timing and phase of the acoustic reply. Accuracy runs from roughly 1 to 2 percent of slant range on standard equipment down to about 0.06 percent on the best systems. Since accuracy scales with distance, a vehicle working deep is positioned less precisely than a shallow one by the same hardware.

LBL lays an array of transponders on the seabed, surveys their positions carefully, then trilaterates the vehicle against them. It delivers better accuracy than USBL over a wider working area. The cost is mobilisation: someone has to deploy the array, calibrate it, and recover it afterwards.

Inertial navigation carries no external reference at all. Gyros and accelerometers track motion from a known starting point, usually aided by a Doppler velocity log measuring speed over the seabed. It drifts, and the drift grows with time.

How they get combined

Inertial systems drift, slowly, while giving continuous high-rate position. Acoustic systems hold true over time but arrive intermittently and carry noise on every fix.

Tightly coupled navigation takes both. The inertial solution runs continuously and each acoustic fix pulls it back toward truth, so the drift never accumulates far. Sparse LBL takes this further, cutting the number of seabed transponders by leaning on inertial, velocity and depth data to fill the gaps. Fewer transponders means less mobilisation time, which on a vessel day rate is the number that matters.

Depth is the easy one

A pressure sensor gives depth directly, accurately and cheaply, with no drift and no external reference of any kind. The vertical axis is solved.

All the difficulty in subsea positioning lives in the horizontal plane, which is worth remembering when reading a specification. A system quoting impressive overall accuracy may be leaning on the one axis that was never hard.