DTM vs DSM: two ways to describe a surface

Two questions

A DTM asks for the ground. A DSM asks for the top surface.

In common British usage, a digital terrain model represents bare earth. Buildings, trees, vehicles and other raised objects are removed or excluded. A digital surface model retains upper surfaces, so its heights can describe rooftops and vegetation as well as exposed ground.

The distinction is about what the elevation values mean, not how attractive the finished shading looks. Two rasters can share the same grid spacing, coordinate system and file type while one describes a forest canopy and the other describes the land beneath it.

From observations to a product

Removing an object is a classification and modelling decision.

An airborne LiDAR survey can record returns from canopy, branches, roofs and ground. Processing classifies those points by likely surface. A terrain product uses accepted ground points and interpolates through places where non-ground objects have been removed. A surface product selects returns intended to describe the top of what the sensor encountered.

That clean description hides real judgement. Dense vegetation may leave few reliable ground returns. A building removed from a DTM leaves an area that must be filled from surrounding evidence. Bridges can be treated as structures above the terrain rather than as ground. Water often needs its own surface treatment.

Not every DTM begins with LiDAR. Mapping authorities can use photogrammetry, breaklines, survey points and automated methods. Likewise, a DSM can be produced from imagery or radar. The delivered specification should describe the source and processing.

What a route samples

The right surface depends on the journey being described.

• Woodland trailA DSM may follow canopy height rather than the ground travelled beneath it. A suitable DTM aims to remove that canopy.
• Urban routeA DSM can place the line on roofs if the route is slightly misaligned. A DTM better describes general ground, but may simplify kerbs and structures.
• BridgeA bare-ground model may describe the river or valley below after the deck is removed. Neither that value nor a roof-like surface value automatically equals deck height.
• TunnelA single surface model normally describes terrain above the route, not the travelled level underground.

The result can still look convincing. Shading makes any continuous surface easy to read, so visual plausibility is a poor test of whether the chosen surface matches the route.

Ignore the acronym race

Choose the named product whose specification answers the task.

USGS documentation records more than one convention for DTM. Ordnance Survey defines its terrain products as ground surfaces with protruding features removed. The Environment Agency publishes separate DTM and DSM products with explicit processing descriptions. These examples show why an assumed universal dictionary is unsafe.

1. Surface

   Read the provider's own definition.

   Look for bare earth, top surface, first return, last return or another precise description.

2. Method

   Find out how objects and gaps were handled.

   Classification, interpolation, breaklines and manual editing can materially affect the result.

3. Quality

   Check resolution, accuracy, datum and date separately.

   A DTM is not automatically finer, newer or more accurate than a DSM.

4. Purpose

   Match the surface to the question.

   Ground travel, canopy analysis and building modelling require different evidence.

The DEM overview explains the wider family of height models, while the beacen Library shows terrain in route context.

Read the provider's surface definition before relying on either acronym. DTM and DSM name intended representations; neither guarantees finer, newer or more accurate data.