DEM

A digital terrain model that represents ground surface elevation as a numerical grid.

DEM

Overview

DEM (Digital Elevation Model, 수치표고모델) is a digital terrain model that represents elevation information of the ground surface in numerical form. It is generally stored as a regular grid (raster), triangulated irregular network (TIN), or point cloud, and is used as foundational data in various fields such as terrain analysis, hydrological modeling, urban planning, and disaster management. A DEM is not a simple map image but a dataset in which an elevation value is assigned to each location, so it plays a key role in calculating slope, aspect, watersheds, viewsheds, flood inundation, and more.

Key Details

Definition and Composition

A DEM is a model that approximates the continuous elevation of the Earth's surface as discrete grid cells or surfaces. Grid-based DEMs store an average or representative elevation value in each cell, and higher resolution allows finer terrain representation. TIN represents terrain undulation with triangular facets, allowing adjustment of storage efficiency and accuracy, while point clouds are based on many 3D points acquired by LiDAR or photogrammetry. DEM quality is determined by spatial resolution, vertical accuracy, datum, update time, and methods for handling missing data and noise.

Distinction between DEM, DSM, and DTM

In practice, DEM is used as a broad term for digital elevation models, but in detail DSM (Digital Surface Model) and DTM (Digital Terrain Model) are distinguished. A DSM includes the surfaces of objects above the ground such as buildings, trees, and bridges, whereas a DTM represents the bare terrain surface with such objects removed. Therefore, DTM is suitable for flood, drainage, and earthwork analysis, while DSM can be useful for urban landscape, communications viewshed, and solar analysis. Canopy height models (CHM) are sometimes derived as the difference between DSM and DTM.

Generation Methods

DEMs are generated by various methods such as aerial photography, satellite stereo imagery, LiDAR, InSAR, drone photogrammetry, GNSS surveying, and conversion from existing digital maps. LiDAR is widely used to produce high-quality DTMs owing to its high vertical accuracy and vegetation penetration characteristics. Satellite stereo imagery can update large areas relatively quickly but may be affected by clouds, shadows, and building distortion. Recently, there has been an increasing number of cases producing local high-resolution DEMs at low cost by combining drones with structured light and multi-view photogrammetry.

Applications

DEMs are used in hydrology to extract watershed boundaries, stream networks, and flow paths, and form the basis for flood inundation simulation and flood risk mapping. In civil engineering and construction, they are used for cut-and-fill volume calculation, road alignment design, and viewshed analysis for communications and wind power generation. In agriculture, slope and drainage conditions are analyzed for precision agriculture, and in urban planning they are used for landscape analysis, sunlight and shadow analysis, and digital twin construction. They are also used as background data to reproduce real terrain in games, films, and virtual reality.

Quality and Standards

DEM quality is assessed by vertical RMSE, resolution, continuity of terrain representation, and presence of artifacts. Interpolation errors and missing data are likely to occur at building edges, water bodies, steep slopes, and forest areas. Because datum, coordinate system, grid spacing, and accuracy standards for elevation data differ by country, metadata must be checked when performing fused analysis. Representative datasets include SRTM, ASTER GDEM, ALOS World 3D, Copernicus DEM, and USGS 3DEP; in Korea, elevation data provided by organizations such as the National Geographic Information Institute are used.

Recent Trends

In 2024–2025, the DEM field is rapidly moving toward ultra-high resolution, high-frequency updates, and cloud-based analysis. Hybrid production methods combining commercial satellites, drones, and LiDAR point clouds are spreading, and deep learning-based interpolation and super-resolution techniques are being applied to missing-data reconstruction and terrain refinement. Satellite observations such as ICESat-2 and GEDI are used for large-scale vertical datum correction and forest structure analysis. In addition, large-scale DEM processing using COG (Cloud Optimized GeoTIFF), STAC, and open-source GIS, along with digital twins and climate-crisis-responsive flood analysis, are establishing themselves as major trends.

DEM