Thinking about improvements to hydrology.
Rainfall per-tile should be measured in mm.
Multiply by area to get volume of water. Measure this in m3.
This is river flow at that tile, unless tile is a water tile.
How much moisture evaporates from an ocean tile? How much atmospheric moisture falls as rain?
Quite hard to find data for this.
A couple of approachs at redblob. One just uses distance from water to determine moisture. The other simulates the water cycle.
Approach 1:
- https://simblob.blogspot.com/2010/09/polygon-map-generation-part-2.html
Approach 2:
- https://simblob.blogspot.com/2018/09/mapgen4-rainfall.html
My approach was based on approach 2. But I don't do any sorting, I traverse the board following the wind direction. I didn't really follow the precipitation rules to the letter either. Simply making my approach line up more with the approach given there would probably improve my results a bit.
To be honest though what I have right now looks OK. The main worry is about physical units, and about how i'm conflating rainfall and river flow in terms of moisture at a tile.
In terms of physical units, I can probably stick with some variation of the current approach, but convert to physical units where appropriate.