Re: [nbos] [FWE][FM8]Geology in the Fractal World"Mike Oliver" Mon Aug 6th, 2007
Amazingly, I understood most of that - maybe you ought to consider
teaching (maybe you already did). Thanks for taking the trouble. I agree
that a smart young programmer should be able to produce an appropriate
algorithm but I guess she/he'd first need to understand the geology in
terms of appreciating what is being replicated and then that programmer
would have to mesh their product with FWE or FT.
So hold yourself ready! (But don't hold your breath)
[mailto:nbossoftware-bounces-at-nbos.com] On Behalf Of T'Star
Sent: 06 August 2007 14:22
Subject: Re: [nbos] [FWE][FM8]Geology in the Fractal World
I will caveat that doing tectonics accurately is VERY computationally
intensive. Faking it should be somewhat easier. After all,
geologists have been taking shortcuts for decades to avoid multiple
month computation times and relatively accurately.
Continental drift is simply the motion of continents due to sea floor
spreading (the Atlantic is getting wider at the mid ocean ridge.),
subduction (the pacific ocean plate is going UNDER the North American
and Asian plates, so is getting smaller. The Mediterranean sea
underwent a similar process it use to be a MUCH larger body called the
Sea of Tethys.) Mountains are built by several processes "mountain
building Events" tend to occure at hot spots (Hawaii, Yellow Stone)
and the boundaries of techtonic plates. When two plates slame into
each other they can either rotate along one another (strike slip fault
like the San Andreas) or one can be 'pushed' under the other.
Continental crust is much lighter than oceanic crust (Basalts, which
make up most ocean floor are denser than granites and Andisites that
tend to make up continental crust. Therefore they tend to go under
these other plates.) As they go down there is folding along the edge
of the contental crust, there tend to be earthquakes (which can cause
uplift of the continental crust) and the subducting oceanic crust
melts and the more bouyant liquid material rises and causes volcanoes.
This is how the Andes mountains were formed and part of how the
rockies were formed (though that is a much more complicated system.)
Now for continent on continent collisions. When continental crust
hits continental crust the density difference is MUCH less. The
subcontinent of India is currently subducting (very slowly) under the
continent of Asia. The result of this is the Himalayan mountains.
This is a much less common type of plate collision, though it has
happened repeatedly through out geologic history.
If you look at most continents, the continental cratons (that is the
central part of the continent that is generally geologically inactive)
is typically quite flat. That is because mountains require tectonic
activity, and once the uplift of the mountains STOPS, erosion and
weathering take over and flatten them down over millions of years.
(That is why the Appalachians are so very much shorter than the
rockies. They are much, much older.)
So when two continents "drift" into one another you get a continent on
continent collision that results in VERY tall mountains, but usually
you get mountain building more coastal regions, especially along
subduction zones. (Note this is a summary of about 4 semesters of
tectonic information. These are the basic, major ways that entire
ranges are built. not all the dozens of different other ways that
rocks get pushed UP.)
On a programing sense, what I think they did was just put a 'gradient'
for elevation that requirement, which set fractal boundaries and
incrementally increased elevation... unfortunately this doesn't allow
for large continental interior planes like the Steppes of central
Asia. Most of Siberia, the north American great planes and Canadian
Shield, Central Africa, and the large planar regions in South America
that make up most of Argentina and Brazil. FWE does put some
mountains near the coast and you can 'tweak' elevations, but there is
a tendency for mountainous interiors, and planar coastal regions. The
main tectonic way to generate that kind of setting is to have ALL your
plate collision boundaries in the central areas of continents, and
that just does not happen that universally. Especially not with the
low proportion of seas in most of the fractally generated worlds.
Some where, sooner or later Erosion and weathering are going to beat
out uplift and those mountains are going to flatten... and it's not
going to be at the same rate all over the world.
I am not a programmer, but it would seem to me, that if some bright
young programmer would be willing to put in slightly more complicated
algorithm. (Note, we're not going for actual geological accuracy in
our fractal stuff... just the appearance of realism.) So have some
'steeper' gradients, and make them more common than continental
mountain ranges. When the weathering algorithms are run, don't
weather things in the same. I know it will take longer to process,
but some parts of the world go UP while other parts of the world go
down. There's already a good bit of randomization with the fractal
coastlines and fractal elevation features and so forth. How much
harder, from a programing stand point, would it be to set 'weathering'
to be dependent on how far from a coast line one is? to have the
program flip a coin periodically to decide "ok this time we're doing
internal mountains". Note: The Water weathering tends to be much
less on continental interiors. The bigger the continent, the dryer
the interior, because the more moisture is likely to be dropped
somewhere else, before it gets to the interior. Which means on places
like the great Planes yo get much more wind erosion to water erosion
than you get in say... Florida. There are exceptions, there are
always exceptions, but rather than trying to program in all the
geologic laws, I would like to see some of these features randomized a
little more. I can figure out a techtonic configuration to make most
things work, but about 99% of what I saw out of Fractal Terrains and
about 90% of what I see out of FWE can only be explained in continent
on continent collisions.
Now that was a longer ramble than I intended, I hope it made sense? I
can elaborate on any given point.
On 8/6/07, Mike Oliver <mike-oliver-at-blueyonder.co.uk> wrote:
> Hi Heather:
> This should probably go on the CC2 group list but I don't believe you
> post there.
> I started using Fractal Terrains to generate all my worlds but have
> almost abandoned it for very much the reasons you outline. The terrain
> generated tended always to be coastal lowlands graduating to a central
> mountain peak - not much in the way of coastal mountain chains and the
> like (although that may be the way I had set up the generation
> I was just about to give FWE the chance to take over that aspect of my
> world building but maybe I shall have the same problem.
> I believe that Joe Slayton, who designed FT, has openly stated that he
> did not attempt to replicate the effects of continental drift or
> tectonic plates in the software. However, you say that "continent on
> continent collision" mountain building is present. It is probably my
> ignorance but what is the difference between this and continental
Nbossoftware mailing list
Nbossoftware mailing list