Use the Terrain wizard in ArcCatalog to define and build a terrain.

Terrains are located in feature datasets. Terrains, and the feature classes used to build them, must reside at the same location. One benefit of using the feature dataset as a container is that it ensures all the data used to build a terrain has the same spatial reference. In an SDE database, it’s also the organizational level at which data gets registered as versioned for editing.

Since the mass points and breaklines have z-values, which will be used to define the terrain surface, the height source for them is set to Shape. This means that z comes from the shape geometry. The topo_clip_poly feature class contains a 2D polygon. It defines the horizontal extent of the surface and minimizes interpolation artifacts around the surface perimeter. Clip polygons work best when they’re smaller than the extent of the data that is being used to provide z-values. The water polygons are represented by 2D geometry but have the height attribute SPOT. Each polygon can have its own height, but that height is constant. For lakes, this is fine, since they’re flat. With this data, there are no measurements inside the lake boundaries that contradict the lake SPOT values, so you can add the boundaries as breaklines; otherwise, they’d be added as replace polygons to ensure their areas get flattened. Everything but the breaklines are used for the terrain’s overview representation. The overview is a generalized representation of the terrain, similar to a vector-based thumbnail. The breaklines are too detailed for this, but all the other information is needed to produce a reasonable looking overview. On the next panel, you’ll define the pyramid for the terrain. Terrain pyramids are used to create multi-resolution surfaces. They’re similar to raster pyramids in concept (that is, coarser pyramids are used for display at smaller scales to improve display performance), but there are differences. The most significant are that they’re composed of vector-based measurements, their construction is based on some measure of vertical accuracy, and they can be used for analysis as well as display. Terrain pyramids are defined using two factors: z-tolerance and reference scale. The z-tolerance of an individual pyramid level represents its approximate vertical accuracy relative to the full resolution data. The reference scale of a pyramid level defines the display scale at which it becomes active.

Using these settings, the full resolution data is used in map displays up to a scale of 1:2,500. Between display scales of 1:2,500 and 1:10,000, only the data necessary to achieve an approximate vertical tolerance of 1.0, relative to the full-resolution data, is used. Between 1:10,000 and 1:25,000, a tolerance of 2.5 is used. Between 1:25,000 and 1:50,000, a tolerance of 5 is used. For any scale smaller than that, a tolerance of 10 is used. Tolerances and scales used to define a terrain pyramid need to be specified based on application requirements. One approach is to mimic the accuracy requirements of a contour map series. A generally accepted rule is that contours should be accurate to within one-half of their interval. For example, if a 1:24,000 scale map within your study area uses a 5-foot contour interval, then the vertical accuracy should be 2.5 units RMSE. Base the pyramid on the scales and contour intervals appropriate for a map series of the terrain.

While breakline vertices are used where needed through all pyramid levels, the actual line enforcement can be restricted to occur in a subset of these levels. For example, road curbsides need not be enforced as triangle edges in a terrain at scales smaller than 1:24,000. You control enforcement via the Pyramid Bounds dialog box.

In the dialog box, z-tolerances indicate at which pyramid levels the enforcement is to take place. Using the given values, the breaklines will be enforced for pyramid levels with z-tolerances >= 0 and <= 1.0. This translates into the breaklines being enforced only at scales larger than 1:10,000. The water and clip polygons are enforced through all scales. This ensures the data boundary is always correct and water bodies remain flat.