There is a better elevation map of the Moon than of the floor of the Pacific. On 20 April 2026, at the International Hydrographic Organization’s triennial Assembly in Monaco, the Nippon Foundation-GEBCO Seabed 2030 Project put the number at 28.7 per cent. That is the proportion of the world’s ocean floor now charted to modern standards, about 104 million square kilometres, with close to five million square kilometres added over the past year. Mitsuyuki Unno, executive director of the Nippon Foundation, delivered the figure to the assembled delegations. GEBCO published the corresponding grid three days later.

Held against the planetary record, that percentage is striking for an unobvious reason. Mars, the Moon and Venus have each been surveyed globally, by single coordinated spacecraft campaigns, at levels of detail nobody has achieved across the whole floor of Earth’s ocean, and the shortfall has little to do with funding or with how much the seabed matters.

The obstacle is water.

What counts as measured

Mars Global Surveyor carried the Mars Orbiter Laser Altimeter, which fired infrared pulses at the surface ten times a second and timed the return. Between 1999 and 2001 it gathered more than 600 million elevation points. The USGS global digital elevation model built from that data runs at 463 metres per pixel, with points accurate to about one metre in radius and roughly 100 metres horizontally.

The Moon is better again. Its laser altimeter, aboard the Lunar Reconnaissance Orbiter, has logged over 6.5 billion height measurements since July 2009, with vertical precision near ten centimetres. Merged with stereo imagery from Japan’s SELENE Terrain Camera, the model resolves the lunar surface at about 60 metres at the equator, and the standard global product is distributed at 118 metres per pixel.

Venus is the more qualified case. Magellan’s synthetic aperture radar imaged 98 per cent of the planet at 100 to 150 metres, completing that coverage between September 1990 and September 1992, which is why Venus renders so sharply in the archive. The altimetry behind those images was coarser, with a topographic objective set at 50 kilometres horizontally and 100 metres vertically. So the comparison holds most cleanly for Mars and the Moon, where laser altimetry did the work directly.

Seabed 2030 works to targets that are modest by contrast, deliberately so. The project sets a depth-dependent grid: 100 by 100 metre cells in water shallower than 1,500 metres, 200 metre cells to 3,000 metres, 400 metre cells across the 3,000 to 5,750 metre band that covers 72.6 per cent of the seafloor, and 800 metre cells in the deepest trenches. Founding documents framed the ambition in terms of features larger than 100 metres, but the tiers are what the regional centres deliver against, and in the abyssal bands a single depth value stands in for a square 400 or 800 metres across.

Why seawater is the obstacle

Radar and laser altimetry work above an airless or thin-atmosphere body because nothing absorbs the pulse on the way down. Seawater absorbs both. A satellite over the Pacific cannot see through four kilometres of it, which leaves the multibeam echo sounder, mounted on a hull and dragged across the water directly above its target, as the only instrument that reliably records deep seafloor shape.

Ship-based sonar resolves the bottom at 200 to 400 metres. Its swath widens with depth but stays narrow relative to an ocean basin. Sean Mullan, an ocean mapping instructor at Memorial University of Newfoundland’s School of Ocean Technology, set out the scale of the job in The Conversation in July 2021, citing published estimates: one survey vessel would need around 350 years to adequately cover the seabed below 200 metres, and another 620 years for the shallow margins.

MOLA covered Mars in two years. The 28.7 per cent represents decades of accumulated ship time from many nations.

The gravity workaround

A global map of the seafloor does exist, and anyone can download it. The GEBCO_2026 Grid gives elevation values for ocean and land on a 15 arc-second interval, 43,200 rows by 86,400 columns, 3.73 billion data points with no gaps. GEBCO is explicit about what that involves: the grid is fully populated, with areas not supported by direct measurement based on predicted bathymetry.

Predicted bathymetry is inferred rather than sounded.

Seamounts and ridges carry more mass than the water they displace, and that extra mass pulls the sea surface up by a few centimetres directly above them. Satellite altimeters detect those bumps, and a gravity model converts them into an estimate of the shape below. As a description of basin-scale structure it is useful and widely relied upon. It is also a calculation about a place no instrument has visited.

What SWOT changed, and what it did not

Recent movement on this came from an instrument built for something else. Yao Yu, David Sandwell and Gérald Dibarboure published Abyssal marine tectonics from the SWOT mission in Science on 13 December 2024, using a single year of data from the NASA-CNES Surface Water and Ocean Topography satellite, built to measure rivers and ocean eddies. One year of SWOT’s wide-swath radar altimetry carried more gravity detail than thirty years of conventional nadir altimeters, resolving seafloor structure at eight kilometres.

Eight kilometres is a real gain on what came before and remains far coarser than a ship. NASA’s summary of the work notes it could lift the count of known seamounts from around 44,000 towards 100,000, picking up features under 500 metres tall. Abyssal hills, the parallel ridges left at spreading centres, are generally put at about 70 per cent of the ocean floor, and the SWOT gravity field picks out individual hills where earlier altimetry showed a blur.

The arithmetic on the deadline

Roughly 258 million square kilometres of seabed sit outside the 28.7 per cent, and the past year added about five million. Closing that gap on schedule would require the remainder to arrive by a route other than the one that produced this year’s increment: crowdsourced bathymetry logged by transiting commercial vessels, uncrewed surface craft running survey lines at lower cost, and the release of proprietary data already on industry servers.

The GEBCO_2026 grid landed on 23 April, the eighth produced under Seabed 2030, and its SRTM15+ base layer draws on the SWOT gravity field from Yu’s paper and machine learning methods. A ninth is roughly a year out. GEBCO’s rhythm is annual though the month drifts, with recent releases landing between April and July. The figure to hold is not the percentage but the size of the yearly addition, and whether it begins to compound.