A terrain dataset is a multiresolution, TIN-based surface built from measurements stored as features in a geodatabase. They're typically made from lidar, sonar, and photogrammetric sources. Terrains reside in the geodatabase, inside feature datasets with the features used to construct them.
Terrains have participating feature classes and rules, similar to topologies. Common feature classes that act as data sources for terrains include the following:
- Multipoint feature classes of 3D mass points created from a data source such as lidar or sonar
- 3D point and line feature classes created on photogrammetric workstations using stereo imagery
- Study area boundaries used to define the bounds of the terrain dataset
The terrain dataset's rules control how features are used to define a surface. For example, a feature class containing edge of pavement lines for roads could participate with the rule that its features be used as hard breaklines. This will have the desired effect of creating linear discontinuities in the surface.
Rules also indicate how a feature class participates through a range of scales. Edge of pavement features might only be needed for medium- to large-scale surface representations. Rules could be used to exclude them from use at small scales, which would improve performance.
A terrain dataset in the geodatabase references the original feature classes. It doesn't actually store a surface as a raster or TIN. Rather, it organizes the data for fast retrieval and derives a TIN surface on the fly. This organization involves the creation of terrain "pyramids" that are used to quickly retrieve only the data necessary to construct a surface of the required level of detail (LOD) for a given area of interest (AOI) from the database. The appropriate pyramid level is used relative to the current display scale or can be chosen by the user in analysis functions, so the appropriate level of resolution is used to satisfy accuracy requirements.