There are numerous programs and resources that can be used to ascertain pretty exactly the average temperature and general climate of a region.

But is there any way to get a quantitative estimation of rainfall or record high and low temperatures in a region?

I know general trends like, since my region is situated on the edge of the polar front, with an ocean to its east, high mountains to the west, and strong persistent westerly winds, it probably gets a considerable amount of rain. Or, I know that due to the lack of large landmasses to the north of my region, it probably will not have the comparatively cold record lows found in places like the NW US or Eastern Europe.

Finding an Earth-proxy isn't exactly correct either, since my reliable modelling for average temperatures suggest that the climate regime for my region is not completely equivalent to South Chile (my closest proxy). While these little differences may not mean too much for average temperature prediction, they can introduce unintended consequences to your world.

For example, Puerto Montt has the closest climate to my region's captial. They are remarkably similar in terms of year-round average temperatures and in rainfall (though mine shows less seasonality). However, in the winter months, Puerto Montt has an average low of roughly 3.5 C. My region's capital has average lows of 1.9 C in winter. This difference is negligible for getting a sense of the average climate, but it has a massive impact on other variables. With its milder winters, Puerto Montt experiences snowfall rather infrequently and sporadically, mostly in July. However, my region's capital gets 7-8 reliably snowy days a year that can happen as late as October, an event which rarely if ever happens in Puerto Montt. So, while proxies may be useful for parameters which are easily defined or that you can already predict, they are not useful for estimating other variables. I would expect record temperatures to be especially susceptible to this.

In short, qualitative analysis and proxies are insufficent. Are there any ways to get real quantitative numbers for things like monthly record temperatures?

  • $\begingroup$ Are you asking about real world locations? Or are you asking for a fictional world you are building? I personally don't think you can ever accurately predict the temperature of any location because there are far too many factors to take into account to ever get a realistic simulation (of global weather events). This really doesn't feel like a worldbuilding question, because this feels like one of the aspects that is just too specific to be accurately represented in a imaginary world... $\endgroup$ – Shadowzee Sep 6 '18 at 4:07
  • $\begingroup$ Fictional locations. And there are modelling software such as the ones used for climate change forecasting that can do all of this stuff really well, assuming an Earth-like regime of conditions. I'm looking for specific quantitative methods though, not proprietary research models. $\endgroup$ – Antarctica07 Sep 6 '18 at 4:30
  • $\begingroup$ How do you know that this software is pretty exact or can do this "really well"? I think anything that pretends to be accurate to 1/10 of 1°C like your 1.9°C there, well, the good thing is that you can't test it and find out that it's wrong I guess $\endgroup$ – Raditz_35 Sep 6 '18 at 15:14
  • $\begingroup$ I was simply listing the value that I had; I don't need 1/10 of 1 C precision to know that 1.9 is lower than 3.5. Plus, I never said that the software I was using worked "really well", only that there are in fact, climate models that can do this sort of thing with great accuracy. Finally, I really only have what is available to me, so I will assume that I have lower winter average than my nearest proxies and that this must have some unknown effect on my record temperature values that I should at least attempt to consider. $\endgroup$ – Antarctica07 Sep 6 '18 at 15:26

Use Statistics

Here are the steps

  1. Using historical data from your reference climate, determine the average temperature, rainfall amount, or any other quantifiable parameter you are interested in.
  2. Using the same historical data determine the variability of your climate and calculate the standard deviations for the parameters that you're looking at. Note how far away the maximums are away from the average in terms of how many multiples of standard deviations it is away.
  3. Change the averages to get exactly the type of average climate that you would want in your world. Call your new average "m"
  4. Decide if there are any reasons why your new climate should be more variable or less variable than your reference climate (it probably shouldn't be, that's why you're using the reference), and change the standard deviation accordingly. Call this standard devaation "s"
  5. Plot a normal distribution curve centered at your new average and with the standard deviation that you have chosen.
  6. Decide how long the historical climate record goes back in time ("n" years) in your story. For your reference it looks like Puerto Montt Began taking temperature data around 1970, so almost 50 years. Calculate 1/(n) and lookup how many standard deviations "z" away from the mean this is using the Complementary Cumulative Standard normal table
  7. The estimate for your new record high (or low) will be m + zs (m - zs for low)

There are certainly some problems with this method, but any result should certainly pass any sort of casual scrutiny test that anyone would come up with. Which, I think, is really at the heart of any type of world building question.


A good source for climate data

When I try to model climate data, I use the site en.climate-data.org. This has a wide variety of sites with relevant climate data.

I use experience and maps to find locations that suit my needs. Once you have looked at enough, you can get a feel for what changes in latitude, proximity to the coast, and altitude will do.

For example, if you want Puerto Montt, except a bit colder in the winter, you could try Bergen, Norway, which is about the same temps in summer, but around freezing in winter time. Consequentially, it sees a lot more snow in winter, and regularly sees freezing temps from October to April.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.