The landscape
beneath
the water.

Blue is a beginning, not a description.

On a small world map, the ocean can look like the space between places. Give that blue a height field and the reading changes. Continents have submerged margins. Islands belong to wider foundations. A basin contains rises, depressions and routes through an uneven landscape.

The opening plate is a rendering of NOAA’s ETOPO 2022 relief model. Its colors and shadows make a surface legible; they are not a photograph taken after the water was removed. A bright slope faces the chosen light. A pale blue area lies higher in our ocean palette. Those are two different visual signals. Inspect source

The distinction matters because a persuasive image can make its evidence disappear. We see a continuous surface and easily imagine continuous observation. Instead, begin with a question we can carry from Earth’s interior to the Moon: what was measured, and what had to be inferred?

Three places. One depth palette.

Start in the Atlantic. Trace the raised spine that runs through the basin, then follow A–B across it. The plan view tells you how a feature continues sideways; the section tells you how model height varies along one selected line. Neither view contains the whole explanation.

Now choose Hawaiʻi. Land occupies a small part of the scene, but the underwater slopes continue well beyond it. The coastline is an intersection with a reference level, not the edge of the mountain. Switch to depth colors to read height without the added directional shadows.

In the Mariana region, look for the long dark depression beside the arc. Move the section marker through it. This latitude gives one cut through the model, not a measurement of the deepest point in the ocean. A point elsewhere along the trench can have a different value.

Each move changes the question. A region gives context. A section gives a sequence. A cursor gives one model sample. The smaller the window, the more carefully we need to remember what sits outside it.

The ship listens. The satellite reads the surface.

A multibeam sonar measures returning sound. Its fan of beams samples a swath below a moving ship. Turning travel times into depths requires the speed of sound in the water and a survey process that accounts for the instrument’s position and orientation. The resulting map is already a processed representation of observations. Inspect source

Satellite radar altimetry follows another path. It measures the sea surface. Broad variations in that surface reflect, among other influences, the gravity field associated with mass below. Those observations help constrain estimates of seafloor shape. The radar is not photographing the deep bottom through the water. Inspect source

These methods complement one another. A broad view helps us locate patterns; a survey can resolve detail within its coverage. Asking which method is “better” without asking about the place, scale and purpose loses the most useful part of the comparison.

We have not put a “measured” switch on these regional maps. Their model values alone do not establish which cells have direct sounding support. That would require additional source and survey-coverage records. The two appearance buttons change how the same data are drawn, not what evidence lies beneath them.

A grid is a set of decisions.

ETOPO assembles multiple source datasets into a continuous elevation grid. Its published heights refer to the EGM2008 geoid, with horizontal positions in WGS84. Source-ID grids identify contributing datasets; a dataset identifier is not, by itself, a cell-by-cell certificate of direct observation. Inspect source

Our regional illustrations sample the published 60-arcsecond derivative at wider intervals: five arcminutes in the Atlantic, three around Hawaiʻi and three in the Mariana region. The spacing belongs to this display. It does not claim the same density of measurements, the same horizontal distance in every direction, or the same accuracy everywhere. Inspect source

Sampling also changes what can be seen. A narrow peak may fall between retained cells. A regional section may bypass a canyon. Enlarging the image makes the existing samples larger; it does not recover details we did not keep. The numeric table preserves the downloaded profile values, but extra decimal places do not establish extra accuracy.

Some striking marks on ocean maps record the history of mapping itself. NOAA explains that apparent grid-like formations can arise where more detailed ship surveys sit against broader background maps. A sharp line deserves a source question before it becomes a story about the seabed. Inspect source

That habit connects the ocean to the rest of the atlas. An earthquake signal can support a model of an unseen interior. An orbital image and an elevation map can give different readings of the Moon. Here, the blue becomes a landscape—and the landscape becomes an invitation to inspect how it was made visible.