Wednesday, April 8, 2015

The Weather Problem Described

Do your players care if the weather is sunny?

In most worlds, this is a question players never have to consider.  If there are tables for the weather, they are very simple - a few possibilities based on a simple random die roll at best.  Chances are, the table hasn't been memorized by the DM, isn't consulted regularly and may come up only because of the druid's call lightning spell or some other detail that needs addressing.

That is because weather is ineffably hard to manage; there simply are no words.  Everything about weather is ridiculously chaotic and imprecise, while at the same time adhering to rigid principles of geography and physics.  Offering tables for weather is like trying to predict the position of one of two feral cats, fighting violently inside an iron cage, a minute from now.

The only possible solution is to simplify.  It needs to be understood, however, that the more a given system simplifies the weather, the less relevant it becomes.  That is why role-playing games do not include rules for weather - because any weather system that can be detailed in the space of two pages inside a book will be inherently useless where it comes to contributing to the player experience.

Anyone who has set out to make a weather system for their world discovers very quickly how much work it takes to lift that system to the point where it is relevant.  Be it known that it is a massive amount of work.  More than an ordinary DM, anywhere, wants to put into their campaign, just so the party can get wet once in a while.  For the amount of work it requires, even an elaborate weather system's value will defy the DM's persistence.

Yet I and others find ourselves fighting to get such a system built.  Why?

The most obvious reason is that weather is important in all our lives.  In a culture without technology, even a culture of fifty years ago, weather was even more important.  For a pre-modern culture, for a technology that existed four hundred years ago, weather was the greatest single factor in everyone's life.  There's no question about that.  All culture revolved around it.

For some of us, who want to reproduce the experience of living in a world where the elements are fundamental to the outlook of the inhabitants, weather remains the fatal flaw in the design.  Without it, the world is a fake, a sham, artificial . . . an inauthentic, cheesy swindle, like the moment in a film when it's realized the 'metal armor' worn by the actors is made of painted plastic.

It just won't do.

And we would solve it - except that a weather system is such a phenomenal frustration to build, we're literally forced to accept our impotence there.  It isn't just that we haven't got a weather system . . . it is that there seems no possible way to create one.

Those of us who try recognize the wall we're bashing our heads against.  The reality is, we're going to kill ourselves before that wall comes down.

Okay, so there's the motivation.  Let's pause for a moment and talk about what we'd like a weather table to do.

Well, temperature for a start.  Knowing the temperature would allow us to measure how the characters are dressed, how much water they will need to drink, whether the rivers or ponds they mean to cross are frozen over, how important a fire is, whether they're better off being in an inn instead of camping out, the practicality of armor in both very hot and very cold environments, etcetera.  There's a lot of character actions tied to temperature.

Next, knowing whether it is raining or not would be helpful.  Rain makes a big difference in travel and combat - as anyone knows who remembers when professional football had to contend with outdoor stadiums.  Rain increases hypothermia, makes life much more slippery and unpleasant, is a critical factor in whether rivers and streams can be crossed and in general speaks to many details about setting up camp, riding a horse, crossing through wilderness and the onset of disease.  What's more, comparing precipitation to temperature tells us whether we can expect a downpour or a blizzard.

Because it matters, the magnitude of the rain is also something we'd like to know.  Is it going to rain a lot?  Is the snow deep?  There's a wide gap between a few spattering raindrops and a deluge.  We need to know how much as well as if it rains.

Next comes wind.  Wind is where it gets tricky.  We can throw a dart at a board to get temperature; we can make rain/sunny a 50/50 proposition if we want and roll a d4 to know how much rain . . . but wind is a problem.  It isn't enough to simply roll a die to determine which direction the wind is coming from - if wind direction were fully random, how would ships travel from one side of the world to the other?  It isn't enough to roll a die to see how strong the wind is, because a complete calm will devastate ship travel while a little too much wind will destroy whole towns.  Making a wind chart means creating a bell curve of some kind . . . but when three or four random gales turn up in a month, pretty soon that bell gets taller and taller.  Whereupon, well, we get a situation where the wind is the same every day.

Moreover, so many of the other features we want out of wind are both important and hard to make universal.  We want a generation system for storms.  Rain doesn't always fall from a thundercloud, but we want to know when it does.  We want a generation system for specific types of storms: hurricanes, tornadoes, ice storms, hail storms, dust storms, sand storms, freezing rain, fog, mist and all the lesser versions of that list that produce a 'feel' for the weather without actually being the full-on manifestation.

Which brings up another order of magnitude for everything discussed so far: how long?  How long does the temperature last?  How long does the rain fall?  How long does the wind blow?  How long must the characters endure a thunderstorm, a sand storm, a fog?  Are the characters in the direct path of the hurricane, or are they on the fringe?  How close is the tornado?  When it occurs, are the characters even in danger?  Does the freezing rain fall all morning, or just for a minute or so?  Does the hail beat against the ground for a half a minute or does it hammer down for a quarter of an hour, creating a hail fog?

See, the real interesting thing about the weather are all the deviations that turn up.  Temperature, precipitation and wind are not enough; these things combine with the hydrography and topography of an area to produce these strange, fascinating spin-off results that can't be managed with a simple table, for two fundamental reasons:

  • Everywhere in the world is different, and thus requires a unique table.  Fogs turn up in my part of the world rarely.  They usually don't persist.  A table that will produce a 'fog' result that applies where I live would be useless for the Eastern Seaboard or the North Sea or the Bay of Bengal.  What's more, the difference between these places is quintessentially important to the pleasure of one part of the world having different weather than another part; so we are talking hundreds of tables to produce a meaningful variety.
  • Interconnectivity between different elements of the weather is hopelessly imprecise; so we know the temperature, the rainfall and the amount of wind - so what?  Who's to say that these three things, in the magnitude they presently have, will produce a specific, given result?  Does a wind shear always produce a tornado?  Do thunderclouds always produce rain?  Does lightning always strike at the ground?  Is a certain place on the coast always hit by a hurricane in the right season?
There's another consideration that I hesitate to bring up, since it isn't a problem my world has.  Just to put it out there, however, to emphasize the difficulties most worldbuilders have where weather is concerned:  what is the data for temperature and precipitation for a world like, say, Greyhawk?  Where are the hurricane tracks?  Where does 'tornado alley' occur, due to consistent wind shears due to masses of air between the pole and the subtropics?  Where do the doldrums start?  If we were going to map out sea currents, where would we put them?  For sea currents, as we know, are very important to places in the world like Alaska or Norway, places so far north that if it were not for the Japan Current and the Gulf Stream, they would be virtually uninhabitable.

I hope this starts to suggest the enormity of the problem.  If the reader truly wants to know the full scope of what could be realized, spend a few days with wikipedia, following every link on every page like the one I used to give 'types of fog.'  Start here.  Stay in the general subject of meteorology and diligently keep track of every phenomenon that could make a significant change to a normal, everyday event in your present campaign.

You'll run out of energy before you run out of phenomenon.


  1. Would it have to be a matter of random (or rather, procedural?) generation?

    Especially in your campaign, which takes place on a fantastic version of our actual earth, could you not gather some sources that describe actual weather events of the era?

    I know that in modern times there are people who are very dedicated to tracking and charting weather data and patterns. I doubt you can find much data from the 17th century, but would the players really notice if, say, you took the actual weather of a region in your world as charted a for April 8th, 1955, and applied the same weather conditions to your game set in the same region on April 8th 1655?

    Maybe have a 1/100 chance of a weather even being much worse than the data you have (so that the party might run into that "Storm of the century!" even if the data says it's a run of the mill storm.)

    Maybe this is sidestepping the problem rather than solving it though.

  2. You anticipate me, Matt, but that's a fine thing.

    I looked extensively into the suggestion you make, regarding weather on specific days and using those for my world. Here's where it didn't work:

    Most parts of the world don't keep those records or, at least, don't include them on line. In fact, just finding online data for yesterday's weather, here, would be hard. I could comb through online newspapers, but they tend to sketch over places where English readers are scant - and I don't know enough French or Chinese to follow those papers for other parts of the world.

    Right now I have a part about 120 miles northeast of Palmyra in Syria - a warzone where records have ALWAYS been sketchy. The data I use right now (which I'll talk about in the later posts this week) begins in 1960. Syria has been more or less at war since 1960, so day-to-day records are flat out impossible.

    So, sadly, I had to ditch that idea. At any rate, as you say, it doesn't solve the problem for other people who run a world very different than mine.

  3. A party NE of Palmyra, not a 'part.'

  4. I believe has data based on airport records, so if you can find an airport near where you are running you could possibly extrapolate.

    Of course, there are parts of the world far enough away from any airport that the information you gather would be useless. Also, the records there do not necessarily go back very far, so you might have to use more modern weather reports. I agree that it's not the best solution, but if you're running Greyhawk, or the popular places in the Forgotten Realms, or any other pseudo-European temperate every-place style fantasy world, you could do worse than to use actual weather from the US and Europe.

    The only real solution that I can think of would be building a digital model to take into account all of the variables and possibilities. You know, the sort of thing that might be feasible for a university studying meteorological phenomena or a lab alpha testing new weather monitoring equipment. Not really a feasible solution for tabletop hobbyists. Not yet anyway.

  5. Alexis, have you considered trying to find a scientifically-based weather simulator like this -

    Many of the links from the intro page are broken, but if you can input local information and get a result like this - - it might be useful.

  6. Jason Juta,

    That is certainly a very interesting program. I downloaded it, but it appears to need some special additional file which I can't seem to access.

    Also - and this is just from reading some of the site - it seems to be based on the idea of getting your own weather forecast from present, current conditions, based upon existing stations. In other words, I can't input the conditions, the site draws those that exist - which means that if I want to generate five or six days worth of data to cover a party's travel over a single session, it's no good. Also, if I need data for December and it happens to be April, again I'm out of luck.

    I will look harder at it; never know what interesting stuff might be in a document/program of this nature.

  7. Matt, gives climate data for stations worldwide; some of that data comes from airports.

    One thing that we need to remember; something like a meteorological lab is interesting in modelling the weather in order to produce feedback that covers far, far more detail than I would ever need, particularly about things that could never, ever matter for a D&D world. We have very different agendas.

    Truth is, I don't need 'accurate' data; I just need data that is geo-sensitive, moderately diverse, has a reasonable chance of providing the occasional interesting result and is most importantly universal.

    I think I have the basis for that . . . but again, anticipating future posts.

  8. Oh wow. I'm very interested in hearing what basis you have in mind. I'll sit down and wait patiently for the next post.

  9. Weather is extremely important to me, too. I don't like the idea of the party running around in the land-of-always-summer. Changing this one thing means a more immersive and therefore better game.

    I like your solution to this: simplicity is key in this endeavour. As I am working on my world, one main idea which comes to me is connected to the world map. You have your hex-map and I have a map consisting of regions. I want to have as much information as possible on that map, structured in layers. So that I can change the layer to topography or culture or faction or anything.

    This could be done with weather, too, I think. All these huge weather phenomena , like the Gulf Stream, are persistent. So why not mapping those as well? There could be some major 'wind lines' or 'tornado alleys' which can be expressed in numbers, just like the flow of your rivers.

    Temperature could be structured regionally. You would indeed have to implement different tables for every region, but the data will be more precise that way. And if there is a hurricane or blizzard, you could 'move' them on the map like you would move a NPC on it.

    So I'll try to look at weather like I would look at a NPC. This character may move around and change the baseline values of the region (temperature, rainfall ...) and combining this with seasonal changes (rain + freezing temperatures = snow) you will have a good basis for future extensions.

    However, the weather will only matter to the players if it affects everything, all the stats, the equipment etc. In the end, I would really like to have my group think about going outdoors in mid-December and how to get to the next village.

    Tracking this is another problem, but you have suggested a solution there: have the players gain XP if they overcome this challenge. Damage dealt or maybe even damage prevented (by carefully preparing for the task) should have rewards and if they do, the players will indeed be more than happy to track anything. :)

    I'm looking forward to your next posts!


If you wish to leave a comment on this blog, contact with a direct message. Comments, agreed upon by reader and author, are published every Saturday.

Note: Only a member of this blog may post a comment.