I'm using multiple different systems both in Maya and in Post processing. This topic is only about the flat swirling clouds.
yes but see you don't actually need to make use of fields to achieve the movement what you do is you make several curve flows that accomplish the one tendril (say 200 or so for the full disk). This way they will not merge and you have more positional control on them. Its not real but neither is your question. Oh by the way vortex fields font exist in space.
The other solution is to calculate the needed speed for the exact curve with one central gravity well* and NO OTHER FORCES (possibly a drag field in the middle to make things easier). then send packets of slightly spread speed values from several seed points. The speed value difference smears them out towards the center.
If you were a physicist then you'd use simulation. The reason why i suggest NOT to do so is that you are an artist in a artistic capacity so cheating is the key for that. Almost always the answer for the question dont know how to simulate X how to do it questions when it comes to visualization is: don't simulate it. Use the other facilies you have to do an appearance of what you need. Most likely the real deal does not have those individual lines or if it does much fever of them because the mater comes form different sources.
- Or test it with one point emitter that emits the particles sightly off target. When you get the emitter sorted just radially duplicate it a bit around (or if you want them to start at different distances ask for velocity of your initial tendrill) However in practice you will not find a good value for speed /gravity magnitude thisway as its just too damn sesnitive. The solution can easily jump to solutions that you don't want. One way is to ask a mathematical application to graph the curves for you.
PS: Yes I do real, accurate, engineering simulations for a living. You dont have to go ito the goory detail of testing verifying and getting every parameter just right. Use it to your advantage.
PPS: this might also be viable for a particle expression sampled form a curve but rotated thsiway you get vest of method 1 but less stuff t dealö with.