A Digital Surface Model (DSM) is a gridded raster of elevation values that captures the highest reflective surface at each location, including the bare ground plus everything on top of it: buildings, tree canopy, bridges, and other above-ground features. It contrasts with a Digital Terrain Model, which represents the bare earth.
Why it matters
The DSM is the surface a sensor actually "sees." Photogrammetric and LiDAR first-return processing naturally produce a DSM. It is the right input when you care about what occupies the surface: line-of-sight and viewshed studies, solar potential, canopy height, flight-obstruction analysis, and 3D city models. For hydrology, slope, and geomorphology you usually want bare-earth terrain instead.
DSM vs DTM vs DEM
- DSM — top surface including objects (first return / photogrammetric surface).
- DTM (Digital Terrain Model) — bare-earth elevation after objects are filtered out.
- DEM (Digital Elevation Model) — a general umbrella term; depending on the source it may mean either a DSM or a DTM, so always check the metadata.
A canopy height model is derived simply as DSM minus DTM.
Concrete example
Global products such as Copernicus GLO-30 (about 30 m) and the older SRTM (1 arc-second, roughly 30 m) are effectively surface models that include vegetation and built structures, not true bare earth. LiDAR can yield a DSM and a DTM from the same point cloud by separating first and ground-classified returns.
Common pitfall
Running slope or watershed analysis on a DSM produces spurious terraces and dams wherever buildings and forest edges sit on the grid. Use a DTM for hydrological and geological terrain work, and confirm the vertical datum and units before combining elevation sources.