Wednesday, March 28, 2012

Wilderness Terrain Generation Project

One of my current projects involves rules for generation of wilderness terrain on both a continental and regional level for all game systems.  My idea is in the testing stages, but I've noticed some interesting phenomena in some of my mechanics and assumptions.

Starting Assumptions
  • Some people will want to generate Big Picture aspects of the game world (world size, axial tilt, number of moons, etc.) and others will either pick or assume these aspects, or not care in the first place.
  • The aspects of a terrain we need to know, on a continental basis, are the presence of: salt water, fresh water, elevation, predominant vegetation, and intelligent-life settlement.
  • Terrain comes in bunches -- it's not "choppy," though there are pockets of varying terrain.  I have come up with a simple mechanic to determine "sameness" from hex to hex.
  • Hills are near mountains -- whether due to tectonics, glaciation, or volcanic activity
  • Forests are more likely near fresh water sources -- I'm frankly not sure if this is actually true, but it feels right
What's interesting here, beyond the assumptions themselves, is the effects of combination of these assumptions and the inclusion or removal of the "sameness" mechanic.

Order and Method of Generation

The assumptions dictate a certain order of terrain generation; you have to know whether mountains and/or fresh water are present before you can generate elevation and vegetation.

The question then becomes how to generate these baseline elements (coastline/mountains/fresh water) with a simple mechanic and optional user choice.

Weather and Rainfall

One of the harder elements to simulate is that of weather.  There are deserts, for instance, in every latitude of the Earth (some of which are fairly close to fresh water features), and areas of forest without a lot of water on the ground.  Put another way, the mere presence or absence of lakes and rivers is insufficient to determine the prevailing vegetation.  Simulating this in a simple way is tricky.


On the continental level, a larger scale is necessary so that you can create a map of a sufficiently large area within a manageable mapspace.  The scale also has to be small enough to depict fresh water sources (which determine vegetation) in a meaningful way.  I presently believe about 50 miles to the inch is an appropriate continental/campaign map scale, and 2 miles to the inch on a regional/local level.


Generating a world or campaign map should be a fun experience.  I'm experimenting with many different ideas to inject ease of use, creativity, and fantasy into the process.  I want to make sure this isn't completely antiseptic and scientific for those who want funkiness and ways to generate campaign ideas.


  1. One of the guidelines I use for estimating rainfall and forest placement is that if you have coastline, wind off the sea tends to carry water inland until it hits mountains. On the seaward side of mountains, you get a lot of rain, and on the landward side (if there's more land thataways) you get a rain shadow.

  2. That's part of the weather issue -- you have to know the prevailing wind pattern. This adds another layer or two to the process. What I'm trying to come up with is an elegant, simplified system that requires as few chart references as possible.

    In fact, I'm considering trying to infer weather and rainfall from the placement of fresh water resources. In other words, if you have a blank continent with mountains, rivers, and lakes, you have in a sense already dictated a significant portion of the weather and precipitation patterns. Thus, if I have a vegetation rule that states "forests are more likely if within X miles of fresh water," I have subsumed precipitation in the conversation.