[crossfire] weather, lattitude, town location, and the world

[Brendan Lally]

On 11/11/05, Anton Oussik <
     antonoussik at gmail.com
     > wrote:
>
      On 11/11/05, Lalo Martins <
      lalo.martins at gmail.com
      > wrote:
     >
      > Hmm.  Maybe "bigworld" is not big at all :-P Brendan's calculations
     >
      > still make sense to me generally, except that now I'm thinking about
     >
      > one-chain-wide mountains and finding them a bit silly.  But that can
     >
      > pass, since those are relatively rare.
     >
     
     >
      Such rock formations may actually be possible, wih a hard surface and
     >
      wind and such. Since the planet is not Earth and since we are willing
     >
      to accept teleportation and beds to reality, then why not small
     >
      naturally occuring rock formations?
     
These /do/ exist in the real world, a quick google turned up
     http://home.bawue.de/~jjk/travel/Urlaub%202002/Canyonlands/Needles%202.png
     

They don't look like they occupy much more than a tenth of an acre.
Of course, arguably they should be called rock formations, and not
mountains, but that is a different point.

Personally I would say that if bigworld were not already established,
it would be nicer to use a height map, so that each square would have
a height associated with it, and then whether they are hill, mountain,
plain, river or desert could be inferred from the height values (the
areas that cities are currently on would need to be flat, with a
depression to the side with between them and the nearest sea facing
mountains, to redirect the resulting river).

This would have the added advantage of reducing the size of the
bigworld maps download, although at a cost of slower startup time.

Note that I don't actually suggest this is done, but merely seek to
describe how it might work.

Effectively to go from a height map to terrain, you would need to
determine a coastline (being areas that have height <0) determine the
water absorption, (inversely proportional to depth and distance from
shore should be a reasonable approximation) and then propagate water
along a path roughly parrallel to the coastline, losing water with a
rate given by grad(height),

low flat areas inland would be deserts under that model, unless there
was a small lake in which case an oasis would form,

if there were a mountain range then, there would be a watershed given
by the peak, so that a river would form along the path that involves
the quickest drop to sea level.

if the mountains are not very high, then water would flow down the
other side too, creating savannah or temperate forests (based on
temperature)

the rivers being created there would create areas of river delta, with
either swamps/rainforest or bogs/farmland (depending on temperature,
or local rainfall?)

If this were done in a way that avoided the use of random numbers,
this would still be a fixed value, and it might even be possible to
set exits to go for 'nearest accessible square of arch'  - for example
a mountain cave might be set to be placed on the nearest mountain
square to its current location (probably only 4-5 squares away, unless
the heightmap is altered substantially).