Guidelines

• By checking this box, I am saying that I have read the guidelines and believe to be following them to the best of my ability

Description of feature request

With there being plenty of suggestions out there for variation, I wanted to quickly address predictability. In the mod's current state, storms will spawn in 100% randomly and there's no way for the player to know when the next one will happen. I'll admit it isn't a huge issue right now since the storms are mostly visual, but I know there are weather-dependent mob farms out there, and it might be a good idea to cue players in on approaching weather especially if you plan to make an addon with tornadoes and hurricanes that could potentially destroy their builds.

Concept

Fronts immediately jumped to my mind thinking about this problem. In real life, fronts are parts of larger low-pressure systems and they constantly change shape. However, I think adding entire lows to the mod wouldn't be feasible due to the following:

1. The cloud system uses a static Voronoi grid which just slides in some direction over time, making it impossible for a front to change direction/orientation relative to the whole system.
2. Lows are often on the scale of hundreds to thousands of kilometers. It wouldn't be fun to walk over a million (1000km * 1000m) blocks just to dodge a cold front. Scaling the low down wouldn't get rid of the other issues, either.
3. Simulating the complex interactions and life cycle of a low pressure system would be costly for performance and not worth the effort. edit: I explain why I think that's the case in the reply with the low pressure system diagram

Instead, I think it would be better to keep fronts as their own entities if something like this were to be added. They could be at any orientation relative to the player and would basically just be an organized grouping of cloud types.

Generation

When I talk about the order of cloud types, I'm referring to how they would appear as the front went over the player's position.

In code, a frontal Voronoi cell would be able to be chosen as the next cloud type if at least one of its sides is bordering one of the preceding cloud type. The image below should hopefully illustrate that a bit better.

I don't have many suggestions about how the fronts would actually be placed, but they would be mainly linear and probably shouldn't extend more than 8-10 cells long.

Types

Here are the two major types of fronts I propose. What I describe here is pretty accurate to what's seen in real life.

• Warm Fronts would consist of milder, showery weather. Players would know of an approaching warm front well in advance due to the presence of high-level clouds (up to your discretion, but I put the names of the clouds that appear in real life in the diagram) transitioning into the currently existing nimbostratus clouds.
• Cold Fronts would be more intense, and contain cumulonimbus clouds and thunderstorms. Note that this doesn't have to be a complete line; it can be made up of isolated cells too. If you ever choose to put squall lines or derechos in your disaster addon, they could be a special variant of cold front with some sort of defining feature like an arcus (shelf) cloud.

There are other types in real life, but most of them are either redundant contentwise or impractical:

• Occluded Fronts are pretty much just a mix between warm and cold fronts, and wouldn't make much sense given they form from a cold front overtaking a warm front, which would be impossible with the current system.
• Stationary Fronts wouldn't make much sense either because cloud types can't dissipate which would leave some parts of the world permanently rainy.
• Dry Lines are too intricate, and aren't as well-known as any of these other fronts. Supercells sometimes form along them, but they could just be part of discrete-cell cold fronts instead.

Final Notes

(edited for clarity on non-frontal storms)
I'd like to clarify that these fronts would NOT fully replace the random single-celled systems being used right now. My main goal here is to allow players to predict potentially high-impact weather - within a reasonable extent - if they know what they're looking for. The diagram visualizes my approach to separating out the "organized" and "random" weather, where the individual cells would 1) be smaller and 2) less likely to contain severe weather with a few exceptions.

This is only one solution to the predictability issue and I'm sure there are others that could work too. I just saw this as a good excuse to shoehorn in fronts as a suggestion. ;)

The 2nd issue can be solved by scaling the fronts down. Crackers mentioned that he will scale down the hurricanes.

The 2st issue can be solved by scaling the fronts down. Crackers mentioned that he will scale down the hurricanes.

That's something I didn't think of, just scaling the entire low pressure system down.

The other issue still remains, though: relative movement of fronts compared to the rest of the system would be impossible because the cells themselves are static. Having fronts move if entire lows were added is important 1) so the player experiences both the cold front and warm front as the system passes and 2) to simulate the real lifecycle of a low pressure system.

To clarify what I mean by "lifecycle", the surface map below has a low in an earlier stage of development on the left, and the one close to the center has reached the occlusion phase (i.e. the cold front has overtaken part of the warm front and formed an occluded front).

But allowing cells to move would be extremely computationally expensive. The Voronoi diagram around the changed areas would have to be recalculated every time a seed point is altered (which would be the mechanism in which the frontal clouds themselves would be able to move), likely once every few tick(s) with how fast the weather updates.

For those reasons, I still think the simplified approach with the fronts on their own is better. I just don't think it's worth overengineering the fronts that much; the player likely wouldn't notice they were experiencing the effects of a low pressure system because the weather patterns would be so spread out even at a "smaller" scale (on the tens of thousands of blocks rather than millions).

What matters more to me is the player being able to know to some extent what sort of weather may be coming if they've played long enough to figure it out, or just looked it up on a wiki.

This is very detailed, thank you for taking the time to come up with this! I do like this idea and will plan to try to incorporate this somehow into the voronoi diagram system.

I'm planning on moving the voronoi diagram generation to the GPU, and send cloud type data to the GPU as the center of voronoi cells. This will combine the fast speed of the GPU to compute the texture while still allowing more complex and slower cloud type calculations to be left for the CPU . This should allow for cells to shift and move around (though this is still in the though stage so no promises here).