Indie game storeFree gamesFun gamesHorror games
Game developmentAssetsComics

Less Computational-Intensive Climate Model (?)

A topic by Hirkondev Moydogren created 59 days ago Views: 125 Replies: 5
Viewing posts 1 to 2

One thing that still bugs me about SotE is it's climate model, the results doesn't seem very realistic to me. Now, of course I DO understand how hard is to get realistic results using climate models, even dedicated computers have a hard time calculating them! So my idea was to see if it was possible to implement what I think it would be a much more easier model to calculate, because it really isn't that much of a model, to be fair. I'm talking about Pixie's approach to calculate climates in a fictional planet as described on this thread from Cartographer's Guild:
For one side, I'm sure this result isn't 100% accurate either, but I think that, from an user's point of view, the climates as showed in Pixie's continent are pretty nice-looking. I also believe that, as I said, it might be easier for the computer to calculate something like this. I think the problem would be to ask the computer to approximate how it would draw ocean currents, or the pressssure zones, but once this is done, I'mm fairly certain that the rest would be pretty straightforward.
I also like this approach because it could give us some parameters to mess arounnd with the world's climate, that could change the end results. For example, we could add some sorte o multiplicative factor for temperature calculations that would increase or decrease the average temperature of the planet, and thus change the placement of biomes (fewer, smaller tropical rainforests, for example). Or perharps change the influence of winds, so maybe we could affect how mediterranean climates appear? 

As I said, this is not meant to be a realistic approach to climate. It's just to give a somewhat nice result whithout too much work for the computer to do. But I shoud stress that I'm not any kind of expert in programming or how computers work, how such model would be implemented, or even if is it feasible at all, I'm just giving a suggestion to see if it could work.

I skimmed over it and I think this is actually roughly how climate in sote works currently, but the constants are not tuned.

+ I think we exposed climate defines in the latest release

I'll give a more throughout reply in the coming days, it's 2 am in my timezone at the moment

Thanks for the answer. I understand how hard is to model a climate properly, that's why I thought that using a simpler model with convicing (even if not entirely realistic) results could help. One problem I have with the current model in sote is that climates seems to appear in well defined bands. Sure, I know that they roughly works like this in real life, but Mediterranean climates, for example, in my worlds often ends up like an ellipsoid shape strechting into the continent. In real life, the borders between climates tends to be "noisier", in sote they're pretty much straight lines. I don't know if this could be the reason, but perharps the influence of lattitude is too strong compared to others like topography?

Compare with the map on this link ( climates doesn't seem to follow a very strong banded pattern. Here, AAzélor trriedd to apply his method do draw the climates of a Pangaea-like world (maybe this IS pangaea, but the right half of the continent looks weird to me, I think). He even took into account the fact that the planet was overall warmer during this time, because of a stronger greenhouse effect, so we could add the possibilty to manipulate some constants to this process to allow less Earth-like conditions. I think that, in the end, my point here is a balance between realism and "uniqueness"; most of the worlds I produce in sote are really unique in terms of topography, but they always seems to follow this "striped" pattern too close.

Now, I admit that a huge shortcoming of this system would be to make emergent climate features (like the global weather influence of a major eruption, effects of global warming/cooling, etc.) harder to implement, so this should be considered as well; the downside is that, with this method, once generated, climate would pretty much be done, and could be harder to calculate interesting variables with it given that we brushed-off or over-simplyfied a lot of important aspects.

Yeah, well. Placing whole biomes by hand and placing them by providing monthly figures for temperature and rainfall with code are not exactly the same task.

We still need to retune climate tho.

They are different, but I thought that this method would be, at the cost of being less realistic, easier on the computer than an actual global circulation mode

I think you may be confused. This method is more or less already what SotE is doing. We stopped using a GCM after the Patreon-only release of version 0.1