[Suggestion] Add Rain Shadows

[SnakeAssassins]

https://en.wikipedia.org/wiki/Rain_shadow
https://en.wikipedia.org/wiki/Orographic_precipitation

I think that comparing ridges and continentalness would be a good way to make a lot of educated guesses about how humid an area should be in a way that would give terrain gen more variety, and would be a lot more scalable than looking for rivers when used with mods that have big continents

[cosmicdan]

I did consider this some time ago, but I ultimately decided against it, at least for now, because of how costly it would be. So, yes, I may implement this one day because it's cool, but... if you'll indulge me for a minute, I will explain what would need to be done for this. At least, as far as I could reason with it - maybe I am mistaken, please do correct me if so.

Your suggestion implies that this could be done as an.... alternative to looking for rivers? If so, then.... how? Because, ultimately, oceans and rivers are still the source of humidity, we still need to search for them since they are the source of humidity, the "creator" of it. Searching for oceans is simple enough, continentalness is cheap noise. Rivers, though, has to calculate "ridges" (weirdness) already, which is probably the most expensive climate noise (and is also why river search can be disabled for a major performance boost). Looking for mountains also requires sampling for weirdness too, except we'd need to do it more times. We'd also need to determine the direction of the prevailing wind for a given "latitude" - that itself is simple enough and easy to model. Then we'd also need to determine the direction of that wind in relation to the nearest bodies of water, and apply some "weight" to that direction - e.g. if a body of water is in the opposite direction of a prevailing wind, then the distance "costs twice as much" kinda thing. Not all that different to AI pathfinding.

So that's actually a bunch of extra sampling and calculations that are separate (and in-addition) to finding nearest oceans/rivers. Right now, we just search in a radius (an outward archimedean spiral) for the first river and/or ocean (i.e. the closest, at least probably) and calculate humidity based on the distance to them. Relatively simple, but still measurably slower than vanilla - mostly because we need to generate weirdness (ridges) noise for every point on the spiral to determine if a river exists. But with rain shadows, we'd need to search for all bodies of water within the configured radius, measure the distance and direction of them all, and then "trace" the weirdness values between the origin (climate point we're currently generating) to each found water body (because, remember, mountains are based on weirdness), compare those directions and distances to the prevailing wind direction (which itself gradually changes direction as the latitude changes), and calculate a final "compound" humidity. Well, we would cull the points that are roughly the same direction as another and just use whatever is closest, I suppose. And I'm sure there's other "optimization" or "shortcuts" that could be optionally done, though I can't fathom any right now.

So yeah, it'd be a cool feature, for sure, and hopefully I can do it one day. For realism, and for variety - as you said. Though I don't see how it allows educated guesses, nor how it's more scalable than searching for rivers, simply because rivers are still where humidity comes from in the first place; a river or ocean "creates" humidity, and mountains just modify it (amplify it on the approach and diminish it on the decline, in relation to prevailing wind direction). But again, if you believe I am mistaken or have missed something, please do share your thoughts!