I actually still see this thread as an overall confusing/vague answer (talk about arithmetic overload), so let me attempt to fix it once and for all, for everyone else.
A little primer:
An Unreal terrain is made up of 2 measurements, quads and components (the third one, sections, are essentially the same as components, except they are made specifically for LOD calculation afaik). The quad is the very basic measurement at 4 bytes aka 4 vertices. The component is made up of some amount of quads, which you specify in the landscape creation menu. So you can have like 63x63, 127x127 quads… etc, but this will be 1 component. A different amount of quads means a different map size.
According to the unreal documentation, every component you create has it’s own “render thread processing cost”, and is going to cost 1 or 4 draw calls, depending on the amount of sections you have. So, the less components you have, the better the performance there will be. Also, it says that you get faster LOD transitions (meaning smaller map but better looking if your character is moving fast) and occlusion of more terrain (terrain will be more responsive/less “glitchy” i suppose) if you have smaller components, meaning smaller maps. Basically, smaller map = better response…
Measuring all of this:
Well, you have to take two other values when you want to measure your landscape. These are resolution and scale. Your scale, by default, is measured in centimeters (every distance is, unless you changed it in the project settings, which you should take note of here and adjust accordingly) and is by default set to 100(x) x 100(y). Also, the side length of 1 quad is the same as your scale (area of a quad = scale x scale, in cm). That means that 1 quad = 1x1 meter, or 100x100 centimeters.
Your resolution is the amount of quads in the entire landscape. This means that you can skip the quad multiplication and multiply the scale by the resolution to get the distance in cm of one dimension of the landscape. So if you make a map and the final resolution is 1009x1009 (keeping in mind what I said above about performance) and the scale is 100x100, that means that it is 100k x 100k cm, or 1k x 1k meters, which would be quite small for an open world game imo.
Heightmapping all of this:
When it comes to heightmapping, there is always a bit of trial and error when it comes to the application, but a standard rule is 1 pixel = 1 vertex. Now since, as I said above, a quad is 4 vertices, that means that the amount of vertices is the amount of quads in one dimension*4 + 1, squared. You add 1 because there is always another row of vertices that attach to the last set of quads. Note that if you changed the scale, you would also need to change the “scale” of your heightmap. So, if your scale is at 1000, aka 10 meters, you would want your heightmap to represent 10 meters^2 per pixel if you want an accurate representation…
tldr if you want to measure your map multiply your scale x by your resolution x, your scale y by your resolution y, then multiply those two products to get the area of your map in centimeters.