A Digital Elevation Model (DEM) is a raster dataset in which each cell stores an elevation value, producing a continuous numerical surface of terrain height. It is the foundational input for nearly all terrain analysis — slope, aspect, hillshade, curvature, watershed delineation, and viewsheds are all computed from a DEM.
DEM, DTM, and DSM
"DEM" is often used generically, but precision matters. A DTM (Digital Terrain Model) represents the bare-earth surface with vegetation and buildings removed. A DSM (Digital Surface Model) includes everything the sensor saw — tree canopy, rooftops, and other features. LiDAR can produce both, because its returns can be classified into ground and non-ground points. Using a DSM where a DTM is needed will inflate slope and obscure the actual ground.
Why it matters
Resolution and vertical accuracy drive the quality of every derivative. Global DEMs such as SRTM (~30 m) or Copernicus GLO-30 (30 m) are fine for regional screening; engineering and landslide work usually needs LiDAR-derived DEMs at 1–2 m. Cell size directly affects slope: coarser grids smooth out steep local features and underestimate slope.
Common pitfall
Computing slope or hillshade on a DEM stored in a geographic CRS (degrees). Horizontal units (degrees) and vertical units (metres) are then mismatched, so slope values are wrong unless a z-factor correction is applied or the DEM is reprojected to a projected CRS in metres. Also watch for NoData voids and vertical-datum mismatches, which create artefacts and offsets.