The Third Spatial Dimension - Introducing Geographic Information Systems with ArcGIS - page 565

FIGURE 9-13
Specifying terrain pyramids can seem a bit confusing but need not be, once you understand
the concept. First off, if you view a terrain in detail—at a large scale (that is, 1 divided by a
small number, e.g. 1:500)—you get all the elevation information that can be wrung out of the
data. Pyramids come into play as you decrease the scale, zooming out, viewing larger areas
with less detail. The idea is that you can accept information from fewer data points and only
look at elevation differences if they are greater than a given threshold. So, perhaps, when you
look at the image at a scale of 1:5000, you need only those points that result in an elevation
difference of 10 feet. If you use a scale of 1:20000, you might be content with elevation
differences of 40 feet. The area we are looking at fills the computer screen, very approximately,
at a scale of about 1:24000, which is the scale used by USGS 7.5 minute topological
quadrangles. To set the pyramid threshold levels and elevation differences do the following.
84. Study Figure 9-14. It implies that if the image is zoomed in to a scale of greater than 1:2500
(e.g., 1:800), you will get the full TIN detail. From 1:2500 to 1:5000 you will see less detail—
FIGURE 9-14