Terrain dataset considerations

2D versus 3D

While it's generally advisable to have 3D sources of information used to define terrains, the input measurements aren’t all required to have elevation values. Two-dimensional measurements can have their place when defining a surface. For example, a study area boundary might be needed to properly delineate a surface's interpolation zone. Often, these are derived from 2D cartographic sources such as a political boundary layer. Another example is the use of 2D breaklines. These can be useful when the terrain surface is processed using a smooth interpolator. While they don’t add height information along their length, they indicate to the interpolator that a break in slope occurs across them. Examples include water boundaries and pavement edges. Areas obscured by dense vegetation that prevent accurate sampling can be collected as 2D polygons and added using the softerase SFType.

Choosing a pyramid type

The z-tolerance pyramid filter is most effective with bare-earth lidar, while the window size pyramid filter is most efficient with all or 1st return lidar points.

The z-tolerance pyramid filter is slower but better for thinning data that is to be used for analysis where having control of vertical accuracy is important. The window size filter is faster, but is for more general purpose use as it is based on horizontal sample density.

The z-tolerance filtering employed by the pyramiding process does not work as well when tree canopy is included. This is because points within the canopy extent may be very close to one another in x,y but differ a great deal in z; some hit the ground while others are in the branches. The points do not thin out well because the filter considers them significant.

The window size filtering employed by the pyramiding process can be used with any type of point data. This is because pyramids are determined based on a window size and a user-specified window statistic. The specified window statistic can either be the minimum, maximum, mean, or both minimum and maximum point height. The algorithm selects only one point for each window based on the chosen window statistic. Therefore, the delineation of tree canopy, vegetation, and buildings is more apparent. Further thinning capabilities are available using the window size algorithm, allowing points to be thinned based on similar environmental characteristics in neighboring windows.

Example terrain applications

Terrains can be used in a variety of ways, from small projects to large. They offer benefits for data storage and management, surface analysis, mapping, and visualization. Here are a few examples:

Working with rasters, TINs, and contours


As a general rule, terrains should be made from vector-based source measurements rather than rasters. Terrains are best used to make raster surfaces rather than be made from them. When there is no other choice but to use a raster, it needs to be converted to a point feature class. The resulting points can then be used to build a terrain dataset. The Raster to Multipoint geoprocessing tool can facilitate this process.


A terrain should be made from the original features used to build a TIN rather than the TIN itself. This is particularly true in the presence of breaklines. If the original data is not available, you can decompose a TIN into features using the TIN Node, TIN Line, and TIN Domain geoprocessing tools. Use the resulting feature classes to construct the terrain.


Contours, as with rasters, are not the best source of data from which to build a terrain. Rather, terrains should be used to make contours. If there's no other source of information, contours can be used. It's recommended they be stored in a 2D polyline feature class with an attribute for height, considering that for each feature the height of every vertex is the same. They should be included in the terrain using an SFType of masspoint. Softline is also a possibility, but it is less efficient.

Related Topics