I've been working on a procedural map generator for a strategy game (if I ever get to gameplay). My general development philosophy is "believable but not necessarily realistic", in other words I don't want things like ridiculous inland seas (Rimworld) or easily identifiable perlin noise maps (Kingdoms and Castles). Countless games use procedural generation but I find they don't go deep enough in the generation process.

With this in mind I want to go somewhat deeper and identify a nice balance between creating a fun, believable world and avoiding a too technical process.

So far my process is:

  1. Identify tectonic plates, their general directions, and their borders
  2. Oceanic tectonic plates get a lower base height than continental ones (randomly chosen)
  3. Apply the tectonic plate border effects to the height map
  4. Apply a thin layer of perlin noise to the whole map for flavor
  5. Identify ocean/land
  6. Determine wind directions <- I am here
  7. Determine moisture, rivers, water cycle
  8. Apply erosion

I've been thinking ahead, and I'd like to implement waterfalls and some minor research led me to rock layers since waterfalls are caused by different terrain hardnesses.

1) Would it be fine to just do a separate layer of rock types from a perlin noise diagram, or should I use the tectonic plates I already have to implement something more realistic?

2) What else are rock layers important for in worldbuilding and map generation? I was considering them as a possible resource for eventual cities and factions to make use of, but right now I'm focused more on natural elements of the world. Erosion comes to mind.

Screenshot of the generated map (height map and relief shading)

Screenshot of the generated map (tectonic plates)

  • $\begingroup$ Shouldn't this go on Earth Sciences SE? You could ask for the general factors determining rock layers, and make your picking which ones are feasible to model. And "what could rock layers be used for in worldbuilding" is a different question. $\endgroup$
    – KeizerHarm
    Jan 13, 2020 at 9:31
  • 3
    $\begingroup$ I think it's appropriate to ask here since the question isn't how to be 100% scientifically accurate, but if rock layers can just be perlin-noised into existence for worldbuilding purposes and what uses other than waterfalls and resources rock layers can have. I think it's valid. $\endgroup$ Jan 13, 2020 at 9:42
  • 4
    $\begingroup$ I think you need to be careful when considering scale. Waterfalls are many orders of magnitude below tectonic plates. If it is world building on the planetary scale then waterfalls are lost in the noise. If its a small scale civilization where large waterfalls matter then plate tectonics are not realy in the picture. $\endgroup$
    – Slarty
    Jan 13, 2020 at 10:17
  • $\begingroup$ Good point. I'm interested in more of a "massive waterfall" situation like Niagara Falls or Angel Falls. Something like this. They'd be far and few between. $\endgroup$ Jan 13, 2020 at 10:32
  • 1
    $\begingroup$ Also, a map of waterfalls in the USA shows a definite correlation between waterfall concentrations and the West/East coasts, especially California's tectonic-caused mountain-heavy terrain. $\endgroup$ Jan 13, 2020 at 11:03

1 Answer 1


One problem with your model is that tectonic upheaval does not precede erosion: it is a simultaneous process. For example the Colorado river flows through a plateau in the form of the Grand Canyon. A river flowed there and the land rose up against it slowly enough that the river just kept on trucking. Something similar happened at the Amazon but the Andes rose quickly enough to reverse the flow direction so now that great river flows east instead of west.

Plate tectonics and erosion go hand in hand and which one dominates depends on the local rock composition, the speed of upheaval, rainfall and temperature.

So this is how you get one kind of waterfall: the upstream land is lifted up and water flows through. It doesn't even matter if they are the same kind of rock: it just needs to rise on one side and not be eroded away by the time you see it. See for example the Yellaya falls

On the other hand you can have something like Iguazu and this is probably the kind of river you are looking for: two layers with varying hardness, the top one eroding away slower than the lower one.

So the scale of your waterfalls will depend entirely on both the scale of the rivers and the erosion resistance of the rocks they are flowing through.

Iguazu is broad and massive but in absolute terms not that high. Some of the highest waterfalls in the world fall from plateaus that are not that large but very different from your neighbors: Tepuis

So what are you looking for? If you want soaring high falls you'll need relatively small features with high variance, if you want massive cataracts you are better off with including an extra step in your tectonic plate movement.

  • $\begingroup$ Second this. Long ago I was playing with world generation--I used a combination of faulting (draw a line across the world, raise one side, lower the other) and erosion. Lots of faulting of decreasing size at first, then cycles of faulting and erosion. $\endgroup$ Jan 14, 2020 at 3:17

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .