IMO, weather, asteroids, volcanoes and earthquakes are ASB. If you are doing a major TL, ask a moderator in a PM.
If man can change something, it is not ASB. If only god can change something, it is ASB. So unless you TL involves using nuclear weapons to trigger a volcano, I don't see how to avoid the ASB section. Disease are easy to change because it only takes person X dying before he spreads the disease to the second victim to stop a pandemic. We don't know who, but some had the first case of the Spanish Flu from an animal, and if he had died before he became infectious then the epidemic would have been avoided. Likewise, there are probably many epidemics that did not happen because trapper X who was bitten by a flea with the plague died before he could make it back to his village.
It must be noted that human actions can have serious effects on earthquakes- beyond direct actions (as I recall, several fellows had a plan to, essentially, lubricate the San Andreas fault). Building a city takes a lot of material from one area and moves it to another area- it's certainly reasonable that, in a TL where Los Angeles remained a relatively small (by modern standards) city, the earthquake patterns would have been different. Random chance could hypothetically move a major geologic event several years one way or the other, especially because we know so little about the actual triggers of the event- it's entirely possible that Mt St Helens could have gone in 1978, but some chance events within the volcano itself postponed it to 1980. Or, perhaps chance events could have caused a more moderate eruption, where a lot of steam and ash was released but the whole mountain didn't explode. We simply don't know enough to say how chancy those events were, and if we're truly rerolling the dice, it may well be that some event happened at a different time. (If a major event, like a large earthquake or a volcanic eruption occurs in OTL, it seems particularly likely that the cause of it would persist through ATLs, and thus
something would happen at St Helens in the late-70s/earfly-80s timeframe, but the exact details are perhaps
slightly mroe fluid- if the POD is 1970, odds are it'll be May 18 1980, but if the POD is 1920 or earlier, there might be a bit of variance.) Now, if you were to try a TL where it erupted in 1950 rather than 1980, that would be less likely with a POD later than, say 1400AD- but still possible, because again, we don't know how much randomness and chance is involved in geologic events.
And as regards astronomy- a meteoroid becoming a meteor(ite) is, perhaps, more likely than you credit- since
so many rocks hurl around the earth
so often, the tiniest of orbital wobbles could change the planet's exact position enough that a near-miss becomes a hit. Hundreds, perhaps thousands, of rocks of a few dozen feet across pass an astronomically negligible distance from the Earth every year, and dozens of rocks much larger pass incredibly close all the time. And since gravity is a universally-acting force, every change from OTL is a change in Earth's gravitational impact, which changes the gravitational forces in every object in the universe, even if the change is only slight. It would likely take little more than an altered global settlement pattern to create an orbital wobble big enough to cause
this guy to hit Denver in 1972 rather than passing by entirely. And the farther back the POD, the more chance of some slight change- if Genghis Khan's forces killed a few hundred thousand more people, there would likely be millions of tons more biomass in areas where there weren't in OTL, which could shift the orbit of the earth ever so slightly such that, centuries later, the Tunguska object misses Earth entirely or, alternatively, hits in central China or western Europe. (Of course, these places would be almost unrecognisably different from their OTL counterparts that far after the POD, but big shifts do take time.)
Seismic and astronomical PODs in and of themselves are unlikely (until and unless we find out how random 'near misses' are), but after a given POD, such events will be similar (because there's not much humans can do about the general way of things) but they certainly needn't be
exactly the same. (That said, I wouldn't fault anyone with a timeline where major events like these play out exactly as OTL. It's virtually impossible to predict what exact changes might take place, so unless it's something obvious, there's no real reason to bother shifting the events unless you really want/need to.)
Weather, on the other hand, is so variable you could easily make it a POD (provided it's not something too far out there- it's highly unlikely you'll find a typhoon hitting Sacramento, for example).
Climate, on the other hand, while seriously affected by humans, tends to be affected in a
moderately predictable way, and tends to be somewhat resistant to change by virtue of geography- however, if that geography changes...