ON THIS PAGE
“We need 3D data” is not yet a technical requirement. It is the beginning of a conversation.
Elevation and 3D products can represent bare-earth terrain, the visible surface including buildings and vegetation, photorealistic textured models, point clouds or gridded elevation surfaces. The right product depends on what the project needs to measure, visualize or simulate.
Start by asking what surface you need
Terrain-focused workflows
Flood modelling, slope analysis, drainage, line-of-sight and some engineering tasks may require the ground surface with buildings and vegetation removed or minimized.
Surface-focused workflows
Urban visualization, communications analysis and some planning workflows may need the top of buildings, trees and other visible objects.
Visualization-focused workflows
Simulation and digital-twin applications may prioritize realistic appearance, texture and interactive performance rather than a bare-earth analytical surface.
DEM, DTM and DSM: understand what the terms mean
DEM is often used as a general term for gridded elevation data. DTM usually refers to a terrain-oriented representation emphasizing bare earth. DSM usually represents the visible top surface, including buildings and vegetation.
Do not assume every supplier uses these terms identically. Ask what features are included and how the surface was produced.
Resolution and accuracy are not the same thing
A dataset can have a small grid spacing but still have larger vertical uncertainty. Resolution describes sampling/detail; accuracy describes how close values are to their true positions.
- What is the horizontal grid spacing?
- What horizontal accuracy is expected?
- What vertical accuracy is expected?
- How is accuracy measured?
- Which vertical datum is used?
- Are buildings and vegetation included?
Vertical datum is critical
Elevation without a defined vertical reference can create serious downstream errors. A dataset may use ellipsoidal height or an orthometric height referenced to a geoid or national datum.
Stereo satellite imagery can produce elevation data
Stereo imaging observes the same ground from different viewing geometries. Photogrammetric processing uses parallax between the images to estimate surface height.
Satellite stereo is useful when large-area elevation data is needed without airborne mobilisation. Performance depends on image geometry, terrain, land cover, cloud and processing quality.
WorldView 3D expands satellite-based terrain options
Vantor’s current WorldView 3D product line provides a rapid 50 cm-class 3D option and a higher-fidelity 15 cm HD 3D option. Vantor states that Rapid 3D can be delivered within 24 hours of image availability and that the HD 3D option provides 15 cm resolution with 3 m accuracy in all dimensions.
When satellite-derived 3D is attractive
- large or remote AOIs;
- areas where airborne mobilisation is difficult;
- projects needing refreshed 3D context;
- simulation and planning applications;
- broad terrain intelligence;
- change-based 3D refresh programmes.
When aerial photogrammetry may be better
Aerial survey can provide customized acquisition geometry, very fine GSD and strong control over project-specific mapping parameters. It is often attractive for city-scale engineering and high-detail urban modelling.
When LiDAR may be better
LiDAR directly measures range and can produce dense 3D point clouds. It is valuable for corridor mapping, detailed terrain, vegetation structure and projects requiring point-level 3D observations.
Do not select technology before defining the output
| Need | Potential route |
|---|---|
| Broad terrain over a large remote area | Satellite-derived elevation/3D |
| High-detail urban mapping | Aerial photogrammetry or detailed satellite 3D |
| Bare-earth terrain beneath vegetation | Often LiDAR |
| Rapid current 3D context | WorldView 3D Rapid may be relevant |
| Photorealistic city visualization | Aerial oblique/photogrammetric 3D or textured model |
Terrain products need QA beyond visual inspection
- horizontal accuracy;
- vertical accuracy;
- voids;
- spikes and pits;
- water-surface treatment;
- building/vegetation artifacts;
- edge matching;
- datum correctness;
- consistency across tiles.
Contour generation introduces another set of choices
Contours are derived from an elevation surface, so contour quality cannot exceed the underlying terrain quality. Contour interval should reflect both map scale and vertical accuracy.
Hydrological applications may require terrain conditioning
Raw elevation surfaces can contain pits, barriers or artifacts that interfere with drainage modelling. Hydrological workflows may require sink filling, stream enforcement or structure removal depending on the model.
Urban terrain creates special challenges
Dense high-rise environments create occlusion for optical stereo and complex surfaces for photogrammetry. If the project requires accurate ground elevation between tall structures, acquisition method and processing strategy should be evaluated carefully.
Water surfaces need defined treatment
Photogrammetric surfaces can behave unpredictably over uniform or reflective water. Specifications should state how oceans, lakes and reservoirs are represented—flattened, interpolated, masked or left as no-data.
Tile edges can create artifacts
Large elevation programmes are frequently delivered in tiles. QA should confirm that adjacent tiles align consistently and do not contain edge discontinuities.
Update frequency can be as important as resolution
For a rapidly developing city, a very detailed model that is several years old may be less useful than a slightly coarser surface that can be refreshed more frequently.
Consider downstream software early
- file format;
- tiling;
- coordinate system;
- vertical datum;
- texture requirements;
- level-of-detail strategy;
- streaming architecture;
- storage volume.
3D becomes more useful when connected to enterprise data
A terrain model becomes operationally richer when connected to parcels, assets, buildings, infrastructure, sensor feeds or planning information inside GIS.
Build a specification around the decision
- AOI;
- surface type required;
- horizontal resolution;
- horizontal and vertical accuracy;
- vertical datum;
- required file format;
- recency;
- planned analysis or visualization workflow;
- whether contours, point clouds or textures are needed.
SIME’s role
SIME supports elevation and 3D requirements through satellite stereo, DEM/DTM workflows, current WorldView 3D options and integration with broader GIS and visualization environments.
Further reading
Surface smoothing can improve appearance but alter analysis
Elevation products may be filtered or smoothed to remove noise. Excessive smoothing can remove real small-scale terrain features. The processing approach should reflect whether the surface is primarily for visualization or quantitative analysis.
Breaklines can improve engineered and hydrological surfaces
Breaklines represent important linear features such as ridges, stream channels, road edges or shorelines. In some terrain models they are used to enforce known discontinuities or drainage structure.
Vegetation treatment should be explicit
A DSM may include tree canopies, while a DTM attempts to represent the ground. In densely vegetated areas, deriving bare earth can be difficult from optical photogrammetry. LiDAR may provide a better route where ground returns beneath vegetation are important.
Accuracy should be validated using independent checkpoints
Where possible, compare the elevation product with independent surveyed points not used during production. Report the accuracy method and sample distribution rather than providing a single unexplained number.
3D visualization introduces level-of-detail requirements
A city-scale viewer cannot always load the highest-detail geometry everywhere at once. Tiled 3D formats and level-of-detail strategies allow the application to load more detail near the viewer and less detail farther away.
Simulation and autonomy workflows may have different needs
Simulation can require consistent terrain, semantic objects and predictable coordinate frameworks. Autonomy workflows may prioritize current, machine-readable 3D structure rather than photorealistic texture. State the intended downstream system when procuring data.
Plan for future refresh
If the project area changes frequently, define how updates will be incorporated. Will the entire dataset be replaced or only changed zones? Will versions be preserved? Can users compare terrain epochs?
Source-date and acquisition metadata should stay with the terrain product
A terrain model represents conditions at a particular time. In rapidly developing areas, users need to know when the source imagery or LiDAR was acquired. Include acquisition dates and production dates in metadata so teams can judge whether the surface is still current enough.
Point clouds and gridded surfaces are complementary
Point clouds preserve individual 3D observations and can support detailed classification and measurement. Gridded DEM/DSM products are easier for many GIS analyses. Some projects should retain both rather than converting everything into one representation.
Coordinate precision and units should be checked during integration
Large 3D datasets can expose coordinate and unit mistakes quickly. Confirm whether horizontal units are metres or degrees, vertical units are metres or feet, and whether the receiving application interprets the vertical reference correctly.
Visual textures should not be confused with geometry accuracy
A highly realistic textured model can appear more accurate than it is. Texture quality affects appearance; geometry accuracy affects measurement. Specify and validate them separately.
Procurement should distinguish base product from derived analytics
Viewshed, slope, contour, cut/fill or flood layers are derived analyses, not automatically part of an elevation-data delivery. Define whether the supplier is expected to provide only the terrain surface or also analytical derivatives.
