Terrain scalability

For large projects, the ability to scale is of utmost importance. Terrain datasets are designed to do just this. They can handle projects involving hundreds of millions, even billions, of points. Terrain tools facilitate the use of large point collections, such as lidar, that would normally pose a problem to databases. Scalability is achieved primarily through two means: terrain pyramids and the multipoint shape type.

Terrain pyramiding is used to improve performance. It does so by providing a scale-dependent means of data reduction. Pyramids reference only the data needed to construct a surface of an approximate accuracy. On-the-fly surface construction, display, and analysis are faster for smaller-scale applications because only a thinned subset of the data is required. The original data is not moved or averaged in any way. The exact positional information of the measurements is maintained. Two types of pyramids can be used to build a terrain dataset: z-tolerance and window size.

Using the z-tolerance pyramid type, pyramiding is accomplished through the application of a z-tolerance-based filter that is used to thin points. You eliminate noncritical points to produce derivative surfaces that are within an approximate vertical accuracy relative to the full-resolution data.

With the window size pyramid type, pyramiding is carried out through the designation of a window size filter. It thins points for each pyramid level by partitioning the data into equal areas (windows) and selecting just one or two points from each area as representatives. It essentially controls horizontal sample density with a controllable bias towards high points, low points, or average height points.

Additionally, the enforcement of lines and polygons is controlled on a per-pyramid-level basis. For example, breakline enforcement can be restricted to the highest one or two resolution pyramid levels. Some features, such as study area boundaries and lake shorelines, might need representation through all scales but not at the same detail. Generalized representations can be used at coarse scales, while the full detail is only applied at larger scales.

In the graphic below, the left-hand terrain is made from a coarse-resolution pyramid level that would typically be used at small scales. The right-hand representation is from a higher-resolution pyramid level. Note how the coarse version on the left has only the low-resolution shoreline information enforced. Full-detail shorelines plus all other breaklines are used on the right.

pyramid pair

Each pyramid level has an assigned vertical tolerance or window size and a scale threshold. This is used to control the scale range associated with each level when the terrain dataset is displayed in a map. The number of pyramid levels, their tolerances, and their thresholds are all user definable.

The following table provides an example of a z-tolerance terrain pyramid definition. There are five levels plus the implicit full-resolution level. The full-resolution terrain will be used at scales larger than 1:5,000. The level using a 1.0 unit z-tolerance filter is used at scales between 1:5,000 and 1:10,000 and so on.

Terrain pyramid definition

Scale threshold

Z-tolerance (feet)

1:5,000

1.0

1:10,000

2.5

1:20,000

5.0

1:50,000

10.0

1:100,000

25.0

Pyramid table

Terrain pyramids are cumulative. Each level within a pyramid does not contain a separate and independent set of all the measurements it needs. Rather, to go from a coarse-level pyramid to a more refined level involves adding measurements to those belonging to the coarse level. The full-resolution level is really the sum of all the lower-level measurements plus a few more. This helps to improve performance when using a terrain and reduces storage overhead.

Related Topics


8/18/2010