Once in writing a timeline in which Japan was building up its 8-8 Fleet, in a world without a Washington Treaty, I was reminded by a reader about the 1923 Tokyo Earthquake, how it wrenched the hulls of ships and created a huge cost in terms of clean-up and rebuilding. I had forgotten about Natural Phenomena, and had been too focused on ploughing on with the timeline as I wanted to tell it.
In a discussion in another timline, a World War 2 where priorities had gone differently was leading to a US build up in Okinawa for the prospective invasion of Japan in the early Autumn of 1945. It was pointed out that a massive storm had hit that spot about the time in the timeline that the US would have been at the height of its build-up; it would have been a true Kamikaze, a Divine Wind that would have scattered the Allied force as surely as the original had Kublai Khan's fleet.
Earthquakes and storms, thee unknowables of any campaign. Would the writer of a forward-looking timeline in 2000 have built in Hurricaine Katrina and the huge consequences that had? The tempation might be to throw one in, and if you did your readers would shout ASB!! and pillory you.
OTL these events come out of the blue, usually quite literally, and to artificially build them into a timeline can look like the playing of God, or the taking of sides. If instead of a few days of awful weather keeping Allied aircraft grounded in late 1944, you double it and allow the German forces in the Battle of the Bulge to do even better you will be accused of being pro-Nazi and manipulating immutables to their benefit.
As a result, many writers look to build up a list of major events - hurricaines, tsunamis, typhoons and earthquakes and throw these in at their 'correct' OTL date. This has the dual effect both of the author knowing that his protagonist's best-laid plans are going to be obliterated, and for the reader the knowledge that something is about to go very wrong. In terms of veracity, this is unsatisfactory - nothing else is immutable in the timeline, and everything else is open to the influence of the butterfly effect.
One option of course is to simply strip out OTL events like these and pretend they did not happen - not a very satisfactory option since these events are not rare in the scheme of things. They constantly happen - somewhere, and with little warning. An inventive author can then decide that authorial license (on the same level as poetic) allows him to choose, either consciously or randomly, some other storm, earthquake etc. But his detractors will feel righteous in crying that its not how things did happen, and that in changing known Acts of God the author is entering into ASB territory.
But they would not be!
If you have had a big battle in your timeline, one that did not happen in OTL but does in your Alternate History, it would begin to change weather patterns. All that smoke in the air would initially have a small impact but the butterfly effect is not supposed to be restricted to causality, the whole point of the beating of its wings effect is that it begins to destabilise OTL and introduce random chaotic features into what we know.
A major war featuring large artillery battles and bombardments would so disrupt weather patterns that nothing we know that really happened would do so in an ATL. This effect would be both whether it is an OTL war being fought in an alternate manner, or a war that never happened at all in OTL, or not at that time. Some might say that an OTL war fought differently would have analogous battles, smoke, and disturbance and that overall it might even out in the weather patterns. It might well do so, but the exact dates, location and severity of storms would still be changed, even if the frequency and number are not.
Of course in any timeline, hurricaines and storms would still happen. These won't be something you can calculate, or hope to calculate. If modern meteorological experts cannot properly predict more than a couple of weeks ahead, the allohistorian cannot even begin to work out how his alternate battles have changed weather patterns, let alone how an alternate war in its entirety has.
As above, the choices are to choose the time and place yourself, as an author, or to have them chosen randomly. The first choice is not always the playing of God - it can depend on how you are telling the story. It will certainly come as a surprise to the reader, but it need not be playing God with your timeline if you are telling it in a narrative fashion - a rain storm in Hamburg, a hurricaine in Pensacola, these might not be vital to your timeline but could add a lot of colour and atmosphere if used as a backdrop to your characters.
If you do not want to be choosing when and where and how these storms, hurrricaines and typhoons occur, then you need to reintroduce the randomness. I say reintroduce deliberately as in OTL these events were random. Nobody at the time knew that they were coming, and the effect was often devastating. Writing a ttimeline where you adopt this approach can be very complex, or reasonably simple, depending on your level of detail and how important localised storms are to your narrative.
A sensible option would be to do a statistical summary of major, or even just massive, events whose existence either did, or could have had the potential to change the course of events (change history, as we say after the event). If you calculate a couple of devastating hurricaines a year in the Caribbean/Gulf they will occur between certain months, and you can use some random model to decide where and when. A randomiser with a counting-down device built in makes most sense, so you test for it as you go along, and if over 4 months your odds go from 1-4, 1-3, 1-2, 1-1 then if it does not happen on the longer odds you know for a fact that it will by the time it is a certainty. The location can be from a considered list of 50 or so, chosen once the date has come up. It could be as simple as names in hat, or with a dice or something. As long as the macro effect reflects OTL then nobody can argue against its validity.
That said, massive localised disturbances, such as the nuclear obliteration of Cuba, would so severely disturb the regional climate that the frequency and intensity of hurricaines is likely to spike up and would need to be reflected in the model. If there is anyone left alive to care!
Earthquakes, of course, are a different matter but I would be willing to bet that underground nuclear tests have had some effect there, knocking off what would now be an ATL record of earthquakes in a world where there were no such tests, onto the track we have today.
Thus, what at first seems to be immutable is not at all, and the randomness that existed in OTL can be reintroduced. There is no fixed pattern of weather, whatsoever, and even events such as major earthquakes can be thrown off by Man's own activities.
Grey Wolf
Twitter: http://www.twitter.com/ahf_fiction