It would seem like it would be a useful function to be able to calculate vegetation distribution across a generated terrain represented by a heightmap. If we could calculate vegetation density, and flora type per cell, we could then seed the terrain appropriately. The benefits are plausible and realistic terrains which are automatically generated. So, given a heightmap representing a terrain, what are the relevant factors which determine vegetation density and type? What comes to mind are soil type, terrain steepness, water content, temperature, and aridity. Without claiming to have a degree in ecology, I think I can still come up with reasonable mechanisms which define the above factors. It would seem that loose grained soil rich in biological content is the most conducive to vegetative growth. Such soils might be found where sediment has deposited over the millenia from higher ground. Sediment essentially follows the same path that water does. See below. The amount of water (and sediment which does not get deposited) which flows through any given area can be calculated with a contributing area function. The contributing area function is essentially recursive. For any cell, to calculate contributing area, merely sum the contributing area of all neighboring cells which flow into the cell in question plus the area of the neighboring cells themselves which flow into the cell in question. It is a ercursive function, but by caching the values of cells already calculated, no duplicate work is done. If, before we calculate contributing area, we have a measure of how much rain falls on each cell, we can then calculate the amount of water which flows through any cell. If you look at any aerial photo, you will notice how rich the vegetation is in river valleys, gorges, and so on. This is what the contributing are function helps to calculate. Temperature has been discussed in the ''Snow on terrain?'' function. Temperature defines separate ecosystems. This essentially defines what type of flora will grow, given the proper soil and water content. Examples include alpine tundra, pine and pinyon (evergreen), deciduous, and palm and cacti. Aridity, which defines rainfall, is necessary also. Not only is it relevant to the water flow calculution above, but it also contributes directly to the soil it rains on. Rain shadows are a good way to calculate rainfall. Once again, see the ''Snow on terrain?'' thread. Well, so there you have it. Probably more fodder for debate. These are all just my preliminary random thoughts, and so of course they are prone to error and require refinement.

quote:Original post by MButchers
1) Animals - these have a very large impact on
the flora of an ecosystem, for example if there are a
lot of grazers the landscape will be limited in the
amount of trees and larger shrubs and instead would
primarily comprise of grasses. Also if there are few
large animals to distribute the seeds of large seeded
species then these will be very localized.

This is a good point. Certainly so if the world was populated with animal herds. The reason being that animal herds would only seem to be plausible in areas where grazing is. But I also think a plausible looking (not necessarily accurate) model cuold be built without taking into account animals.

quote:Original post by MButchers
2) Soil type - some of the poorest soils in the world
are those found in rainforest so the arguments for soil
quality dont realy hold true in all conditions.

Can you explain why? Because it seems (and I may be wrong, I don't really know) that even if many rainforests had poor soil to begin with, that would rapidly change from the decay of organic material which is constantly ocurring on the rainforest floor. In a sense, it seems that rainforests provide their own soil rich environment.


Edited by - bishop_pass on October 18, 2001 10:56:36 AM

quote:Original post by TerranFury
A tangentially related but potentially even more useful question is: How can you generate the plants themselves, and draw them quickly? Of course, everyone knows about pseudo-fractal methods. But those aren't about to be rendered with any kind of speed. And billboarding, though efficient, looks like crap.

Any ideas?

A darn good question! Naturally, the burden of such details falls upon your shoulders.

But seriously, there is value in determining vegetation statistical values even if you aren't able to render lush foliage in real time. Examples include interstellar games where planet and landform views may be mostly in the form of satellite views. Another application might be war strategy games which display the view once again mostly as a satellite type view. A side benefit in war strategy games would be the movement and cover factors with respect to troops if given vegetation density information. Another application is a first person perspective RPG where the database for the land exists, but the view given to the player is limited, as he is right there on the ground.

[To no one in particular] Some more thoughts regarding my initial proposal and my ideas about snow and water: I do not in general seek to do massive ecological or meteorological simulations. People often confuse my ideas with a desire to model in every detail the physical processes in the real world. Nothing could be further from the truth. Instead, what I often propose are semi-detailed models which attempt to capture in relative detail the salient contributing factors of world phenomena, attempting to produce results which are naturally plausible from the standpoint of observance. I strive to include some detail and study to the models without going overboard or reducing the problem to a simple hack which has obvious flaws.



Edited by - bishop_pass on October 18, 2001 2:53:53 PM

Since I like to do visual and written discussions, I have included a photograph which I took:




So, here we are looking east down Darwin Canyon in the western portion of Death Valley, California. Death Valley is a very arid place, because it lies in the rain shadow of the High Sierra. Despite the aridity, foliage prospers in the floor of the canyon because of the presence of water flowing through. The surrounding landscape receives very little rain, but the funneling aspect of the canyon ensures foliage growth.

Edited by - bishop_pass on October 18, 2001 3:14:12 PM

quote:Original post by MButchers
Animals are not determined by vegitation but vice versa,
plants evolve far quicker than animals due to a far easier
cross fertilization between the species! (IMHO)

except for the rare case when humans plant non-idigenous plants somewhere, or a seed is stuck on a bird that flies to a new continent, the plants ARE there first... animals DO have an impact on it, as they can over-graze to kill off a plant species, or allow something else to grow when they eat the plants that are blocking light or whatever; but animals ARE determined by vegetation. think about it: plant seeds just land somewhere (whether by wind or bird or whatever), and try their best to grow. animals, on the other hand, cannot move to an area unless there is a food source waiting for them.
also, plants do not generally "cross fertilize" between species (the term cross-pollination refers to two plants of the same species fertilizing each other''s flowers).

quote:...plants are also very adapt as primary
colonizers and will evolve relativly quickly to fit into a niche
environment.

this is also not quite true. when there is no other life in an area, lichen and other very low plants can take hold, and eventually make the area capable of supporting more evolved plant life... but the plants themselves do not evolve to fit a niche. over millions of years they do, but it is not proper to phrase it as they "will evolve quickly to fit..."; actually the lichens just prepare the area by making the stone into organic soil.

--- krez (krezisback@aol.com)