Stand a thousand measuring rods across a valley, record the height at the top of each one, and you technically have enough information to rebuild that valley's entire shape in 3D. That's the whole idea behind a Digital Elevation Model, just with satellites, lasers, and stereo cameras doing the standing instead of you.

Explore a Sample Terrain

Tap each pin on the terrain below to see its elevation reading, exactly how a DEM is built from scattered sample points before the surface gets interpolated between them. This is a generic terrain used purely to illustrate the method.

Five points obviously isn't enough to describe a whole landscape precisely, real DEMs use thousands or millions of points. But the underlying method never changes: sample the height at known locations, then mathematically interpolate everything in between to build a continuous surface.

Four Ways to Collect the Sample Points

How those elevation points get collected depends on the terrain and the tools available. Switch through the four common sampling structures below.

Random points paired with a Triangulated Irregular Network, or TIN, tend to work best in rugged terrain, since sample density can increase exactly where the ground changes fastest, ridgelines, cliff edges, riverbanks, while staying sparse on flat stretches where extra points add nothing. This is the same principle behind a good DTM built from drone or LiDAR survey data.

Quick Check

Why DEMs Matter Beyond Pretty 3D Maps

A DEM isn't decoration sitting on top of a GIS, it's a foundational layer other analyses depend on. Drape a satellite image over a DEM and you get a geo-corrected, terrain-accurate picture. Feed a DEM into a flood model and it tells you exactly where water will pool. Feed it into a route planning tool and it flags where a road would need excessive cutting and filling.

Every contour line you've ever seen started life as a scattered set of elevation dots.

On real infrastructure projects, DEM accuracy directly decides how reliable everything built on top of it turns out to be, from drainage design to route alignment. That's why GIS solutions built on carefully surveyed elevation data hold up under scrutiny, while ones built on generic global datasets, often only accurate to 100 metres vertically, tend not to.

Next time you see a 3D terrain view on any mapping app, remember it started as nothing more than a scattered set of height readings, connected mathematically into a surface you can actually walk through on screen.