Caution: only tested with
MSFS 2002 and 2004

Mesh Overview: Introduction
The elevation and slope of the land exert an enormous influence on life on this planet. For example, they affect weather patterns, natural resources, and the distribution of vegetation and animals (including human). They also play an important role in providing recreational opportunities. But perhaps their most important roles for the flight sim community are as aids to VFR (Visual Flight Rules) navigation and enhanced visual interest (realism).

Elevation data for a region is generally measured at fixed horizontal intervals (the Source Resolution), producing a "grid" of elevation data points. The spacing of these original data points ultimately determines the accuracy of any subsequent representation of the original topography of the land.

Elevation Grid

The Source Resolution of the default data used in Flight Simulator varies; some areas are more accurate than others. The most common seems to be over 1000 meters between data points, although some areas are much better. This is a significant improvement over previous versions, and is adequate for many purposes, but is rarely realistic. (Compared with, for example, my high resolution mesh, which has a Source Resolution of 30 or 10 meters.) 

Once the simulator has calculated the elevations for the area on-screen, it overlays this grid with textures, based on vegetation, land use, and population density information for the area. The simulator includes a wide range of such texture images; these are selected, merged, and rotated to provide a seamless representation of the surface of the land.

Mesh with texture applied

Finally, fixed scenery and auto-generated structures and trees are added.

At least in principle. In practice, things are a bit more complex. Edited from various sources:

Terrain visualization is a difficult problem for applications requiring accurate images of large data sets at high frame rates, such as flight simulation. A continuous level of detail (CLOD) terrain engine reduces the amount of polygons that the renderer has to draw.

One way of doing this is to split up the terrain in patches of arbitrary size. This patch is then reduced to two triangles, ignoring any elevation points that are inside the patch. The computer then calculates the eventual triangle position on the screen and compares this to the points on the heightmap that were earlier ignored. If the difference of height in pixels between the triangle and the heightmap is greater than a certain threshold value (e.g. 3 pixels) then the triangle is split in two. After that, the two new triangles are recalculated again, and the process is repeated until the screen error is less than the threshold value.

As a result, CLOD produces more triangles (accuracy, resolution) in the foreground and less near the horizon. It also doesn’t waste polygons on purely flat patches of terrain. This is a dynamic process, and the morphing of terrain can sometimes be seen as you travel over complex areas. And sometimes the Autogen scenery has trouble settling at the correct elevations.