Friday, March 16, 2018

"Weather in the World of Greyhawk" from Dragon #68

As I might have mentioned before, the idea for designing what would eventually become THE ALMANAC of FANTASY WEATHER came to me when I was re-imagining druid spells for the Book of Spells supplement to SEVEN VOYAGES of ZYLARTHEN. Many of the spells were directly or indirectly linked with weather, which naturally brought up the question - how do you know what the weather is?

As far as I know, none of the original rulebooks for old-school D&D - OD&D, Holmes, B/X or AD&D offered a weather generation mechanism (I didn't go back and check this, so it's possible I'm wrong), which is sort of odd considering that some of the books, such as the DMG, went into such great detail on other things.

However, an article by David Axler, "Weather in the World of Greyhawk" in The Dragon #68, offered a detailed and robust system. As I understand it, that system had been or would be also incorporated into the Greyhawk boxed set. This was where I started.

Readers of FANTASY WEATHER may note that I took many of the same names for precipitation events from that article - Light Rainstorm, Heavy Rainstorm, Light Blizzard, Heavy Blizzard and so on. Since my original intention was not to design a commercial product, I didn't initially worry about copyright, but when I did start to think about that, the naming conventions seemed generic enough to be fair game, and I did notice that much of the same list was used in some of the other newer non-TSR weather products.

I soon altered the list however. Some of the names seemed too American or Western - Hurricane, Tropical Storm, Tornado, etc. - or too obviously culturally specific - Monsoon - for the tone I was trying to set. I didn't want FANTASY WEATHER to sound like the weather report from a North Carolina CBS affiliate, or an AP report from Bangalore. And I folded the article's "special weather" into the more general list. If you used the calculation scheme of the original article, a desert would only get a sandstorm once every eight years, or so I figured it. Where's the fun in that?

I used the article's durations, rainfall/snowfall amounts, wind speed ranges and the like for each precipitation event as a jumping off point, but soon started to fiddle with them for various reasons. For example, I soon discovered that while the rainfall amounts initially seemed realistic, if you did the math, the amount of yearly rainfall for pretty much anywhere in Greyhawk would be off virtually any terrestrial scale - often more rain (in inches or millimeters) than even the wettest place on earth. One of the first things I designed was a method to quickly alter rainfall and snowfall amounts to fit the data for the climate type being simulated. For example, if the weather results for a London-type climate yielded 400 inches of rain on average (roughly ten times the actual total) you could simply divide the rainfall for each event by ten. Of course, you could also just make precipitation events less likely (by a factor of ten) but that possibility wasn't attractive. London is pretty rainy. That doesn't mean you get a huge amount of annual rainfall, measured in inches or millimeters (it's less than New York), but it does mean that you have 150-200 at least slightly wet days per year. Reducing that to 15-20, which would have amounted to desert conditions, didn't seem right.

I realize that the last consideration is more for weather nerds or "realism" nerds. It's unlikely that many readers, referees or players would know or care that the sum of the daily rainfall totals wasn't "realistic." But I figured if I was going to do it, I might as well get it right.

"Weather in the World of Greyhawk" also gave me the idea for those funky rainbows (including that 1 in a 1000 chance of a Bifröst bridge), which I thought, if nothing else, added a fun flavor to things. So I left the idea intact, though I did twiddle the percentages. Later I would add in "supernatural causes", again, suggested by the article, although I substantially modified their chances and content. And I significantly changed how lightning worked. I wanted there to be at least a chance of it in most rain-event types, not just in thunderstorms. Again, this seemed to better match the real-world data, at least as I understood it.

And, of course, I found the article's mechanism for generating each day's weather with dice to be utterly impractical. It just took way too long. That's not the article's fault per se, of course, simply an unavoidable factor of real-world physical realty. That you could do it virtually through an algorithm coded into an Excel Spreadsheet, and then present it clearly in "almanac" form, was the whole idea behind FANTASY WEATHER.

That's the sum of it. I don't mean to sound critical of what in my view was the best treatment of weather and climate to ever appear in a TSR product. Along with a later article, which I'll talk about in the next post, it gave me the framework for how I would look at the problem of fantasy weather. The central insight of the article for me was framing the thing around "precipitation events" and their properties. But there's a lot more to the article than that that either I didn't talk about here or didn't find relevant for the FANTASY WEATHER project but that is still quite interesting and useful in its own right.

If you have The Dragon #68 in hard copy or have access to the PDF, I urge you to read or re-read it.

Tomorrow: "Weathering the Storms" and Köppen climate types.


  1. Decades ago, we didn’t have access to easy info and great analytical tools like we do now. What looked reasonable 30 years ago is possible to refute now with a reasonable amount of work.

    You are very modest. This is an excellent document and you made it look easy.

  2. I know the Almanac says that, due to its high page-count, a print version may not be offered...but I truly hope that one will be. It's such a great book.

  3. Thanks again to both of you! Okay, here's a confession. My published printed products up to now have been thin enough that I didn't have to worry about the copy or content of spines. But 1,130 pages will have a spine. I don't want to deal with spines.

  4. Hey, just found this! Thanks for the observations on Dragon #68 (which is currently the top hit on Google for me on the subject). One extra tidbit: There is very curt weather generation, as an option, in the pre-D&D Chainmail Miniatures rules (p. 22). There's three broad categories (Clear, Cloudy, Rainy), and interestingly, every other turn (1 min/turn), you roll to see if the weather changes or not. The only in-game effects come from the most extreme results (either "excess heat" or "hard rain").

    1. Interesting. I've only read Chainmail closely for the OD&D links, so I missed that.

      I'm assuming the numbers on top (1-3 / 4-5 / 6) apply only to the first roll. Even so, with one-minute turns and, thus, weather rolls made every two minutes, it would seem that even initially Clear days have a close to 100% chance of turning rainy at some point. How likely is it that you can avoid rolling a 6 and then another 6 (before rolling a 4) over a sequence of 360 (or whatever) rolls?