# What computational resources would work for a 550-year climate forecast of our earth?

Folks, I'm in a bit of a jam.

I am writing a fictional future history to support a series of post-collapse science fiction stories set ca. 2560 Gregorian. The first story, The Borax Road Affair, has been accepted for publication in the forthcoming anthology After Oil 3: The Years of Rebirth from Founders House Publishing.

This world, which I'm calling the "Circumpolar", is a set of civilizations that have grown up around the "Polar Sea" (their name for the now ice-free Arctic Ocean, which has become the maritime crossroads of the world.) Now that I'm writing the history, mainly in order to provide migratory and linguistic backgrounds for the populations of the Circumpolar, I need to track some kind of believable climate development over the centuries.

What I'm looking for is a whole-earth, system-based climate modelling program. It needs to work with our planet, not with an invented planetary surface.

I'm having rotten luck finding useful resources - mostly it's just well-intentioned (but useless to me) sites that will allow me to Calculate My Carbon Footprint. My carbon footprint is irrelevant to this project. I need the planetary picture for centuries out.

Another very real problem: the underlying climate models on the few climate modelers/calculators that I've been able to find do not allow midstream parametric insertions that can modify the run.

## What I need, particularly

I'd prefer an online modeler if possible. If I need to run it offline, preferably something written in Python or Java.

I do not need predictive confidence and power. I just need scientific plausibility.

Specifically, I need to be able to provide:

• in which population normally acts as a dependent variable but can be externally modified at will, at a particular time (e.g. "OK, I can do a plague here and a famine there...")

• in which carbon dioxide and methane emissions due to human activity are treated as dependencies of population, but, again, can be tweaked by introducing time (and possibly geospatial) parametric coefficients based on economy and technology

I realize it's not too realistic, but I'd ideally also like a model which can handle such difficult topics as:

• Seasons.

• Icecap changes, especially Greenland, but also Antarctica.

• Sea level changes.

• Ocean anoxic event potential as atmospheric CO2 content increases.

I am working within very explicit constraints here, and I would deeply appreciate it if answers and comments respect those constraints.

## Constraint #1: Certain ships have already sailed/Toothpaste can't go back into the tube/I'm already on record.

Because the first story is in the publication process, I can't walk back any of the assumptions I made when writing that story. In particular:

• I've already got my technological profiles sorted out. The energy budget of the Circumpolar civilizations does not include fossil fuels, nuclear, or exotic energy sources. In other words, all of the people in existence have access to energy drawn from the Sun - or, occasionally, from geothermal sources. I make a point of this because many people might otherwise want to discuss this. It's off the table.

• A less-than-worst-case global warming has peaked, due to rapid collapse-based deindustrialization in the 21st and 22nd centuries, and is starting to subside. I will need to be able to curve-fit my climate model outputs to fit that.

• Western North America has become uninhabitably arid for points from modern-day Denver south. I had to wing it for the Borax Road story, but the existence of the "Dead Dry" region is a pivotal plot point. Desert conditions reach as far north as the plains of Alberta (when the people in the story, who live in the foothills not far from modern-day Banff, speak of salvaging resources from "The City", they are talking about Edmonton.) This is another thing I will need to tweak my climate models to accommodate.

• The Polar Sea is largely still ecologically alive. Any ecological collapses, especially anoxic events, have so far not destroyed the living species of the Arctic. The biome has changed, of course, but it isn't quite pining for the Fjords just yet.

## Constraint #2: Ye cannae change the laws of physics. Also, magic can't change the climate.

The After Oil series editor, John Michael Greer, proprietor of The Archdruid Report, put out his call for submissions in contest form. Here's the official statement of the contest rules:

The requirements are the same as before:

• Stories should be between 2500 and 7500 words in length;
• They should be entirely the work of their author or authors, and should not borrow characters or setting from someone else’s work;
• They should be in English, with correct spelling, grammar and punctuation;
• They should be stories—narratives with a plot and characters—and not simply a guided tour of some corner of the future as the author imagines it;
• They should be set in our future, not in an alternate history or on some other planet;
• They should be works of realistic fiction or science fiction, not magical or supernatural fantasy—that is, the setting and story should follow the laws of nature as those are presently understood; [emphasis mine]
• They should deal directly with the impact of peak oil, and the limits to growth in general, on the future; and as before,
• They must not rely on “alien space bats”—that is, dei ex machina inserted to allow humanity to dodge the consequences of the limits to growth. (Aspiring authors might want to read the whole “Alien Space Bats” post for a more detailed explanation of what I mean here.)
That is to say, the stories that will find a place in the second anthology, like those that populated the first, will feature human beings like you and me, coping with the aftermath of the industrial age in a world that could reasonably be our future, and living lives that are challenging, interesting, and maybe even appealing in that setting. I’d like to make an additional suggestion this time around: don’t settle for your ordinary, common or garden variety postpetroleum future. Make it plausible, make it logical, but make it different.

So, there you have it. I stated my constraints at length; thank you for reading them; please treat them as they are. I'm working within them, not building a world from scratch. :-)

• Beware of chaos, as in the science-y chaos, because sometimes you'll get a different result if you advance decade-by-decade than if you just run the simulation for several. Feb 15 '15 at 20:47
• I think the difficulty here is that you are not so much looking for a simulation (in the science sense) as any reasonable model. As a result you may actually do better with games tools (although I can't think of any specific tools to suggest off hand) than simulation ones. It's an area to look into as well anyway. Feb 15 '15 at 21:24
• Have you looked at EdGCM? edgcm.columbia.edu Feb 15 '15 at 22:08
• Problem is, we still do not know Earth's climate enough to be sure. We know about huge risks, which can include positive feedback, like methane from melting permafrost, or methane clathrates on ocean floors, which can melt if ocean warms. We have no idea how bad the additional damage could be, but because of positive feedback, it could be really bad (so bad that your deindustrialization might not be enough to revert the damage). Feb 16 '15 at 16:56
• it is trivial to make a climate forecast for any period in the future. Making it accurate is the problem. May 20 '15 at 23:39

There are only two models that I'm aware of that could scale out to the timeline you've listed...

CMIP5 - http://cmip-pcmdi.llnl.gov/cmip5/ attempts to tackle climate beyond 2100.

and the other is CM2.x done through NOMADS (NOAA National Operational Model Archive & Distribution System) ... http://nomads.ncdc.noaa.gov/thredds/catalog/gfdlcm2x/catalog.html is their older site, I haven't updated my links recently.

Most of the other models I'm aware of won't even try to do anything more than 1 month out.

Both of these don't really progress past 2100 as we really don't have much for information on the topic...Our ever changing date of the ice free polar region is a good example of that. One element that we are truely in the dark on is the sun (pun intended) and it's various phases are heavily unknown. Sunspotting appears heavily on the decline and if that pattern continues, we may see a minimum sun cycle that could see the globe cool a bit once again. Unfortunately there is so much variation (including human behavior) that needs to be included that you probably could create a climate simulation, run it 5000 times, and not have one of those simulations end up with the same result.

The processing power required to run these models is pretty intensive...that and the best forecast model we've found is to run the model thousands of times and work with the averages of the many runs.

Honestly, at that date range...you can be pretty creative. If you want a model for it, you are simply depending on someone else to be that creative for you and program it in modelling form.

Complete side note... Banff is a considerable distance from Edmonton...Go with 'Jasper' about 3 hours drive north of Banff if you want mountain living near Edmonton.

Since the climate is a vast, chaotic system where outputs are not linearly related to inputs either spatially or temporally, what you are asking for is essentially impossible. Just looking at how difficult it is to get an accurate 7 day forecast from the Weather Network or any other weather prediction service should give you some idea of what you are dealing with.

Since you are writing a story (or series of stories), you can plug in some events which have historically changed the climate wherever it will fit. Supervolcano explosions like Tambora are known to have cooling effects on the climate (the 1815 "year without sumer" is a clear example). Changes in the output of the sun have a definite effect; the Little Ice Age seems clearly linked to the Maunder minimum. What warms the planet seems a little less clear; the good weather that the Roman Empire seems to have enjoyed or the European Warm Period are obviously not a result of any technological activity by human beings, but the ability to do croft farming in Greenland in the 1100's or grow grapes in Scotland and make wine in the 1300's are pretty clear evidence that things really were different then.

And of course since your story is set 500 years in the future, there is no reason that human beings of that time could not simply adjust the climate to their own liking, either through releasing gasses into the atmosphere in controlled amounts, or using technological and biological manipulation to subtract gasses. Orbiting platoons of solar mirrors to illuminate or shade parts of the Earth would also have global effects. You can imagine other technological "fixes" that human beings 500 years from now might be doing.

• Upvoted for mentioning what everyone else is avoiding: climate is a chaotic system, essentially impossible to predict over long time scales (and not too easy over a week), no matter what politicians tell us. May 16 '16 at 3:57
• @frank your statement is conflating weather and climate. Jul 7 '16 at 21:57
• @frank To clarify: "weather" is "Will it rain tomorrow in NYC?" but "climate" is "will it rain between 25-30% of the time on the East Coast of Eurasia?". Climate change is predictable; not very precisely but enough to show a general trend. Most climate change predictions are simplistic, joining up the dots of "what has happened" with a line, then extrapolating that line and reading off what it will say in future. Actually modelling the weather for hundreds of years is much more computationally expensive, and whilst it's less precise it's more accurate most of the time. Nov 28 '17 at 19:47

If you are aware that solid prediction is not possible, and you find predestination distasteful, you can put biases into whichever direction you want to push the balance.

Do you want dystopia? Unleash the methane.

Want Earth all cleaned back? Let humans invent/design efficient bacteria which converts CO$_2$ from atmosphere back to oil, and pump it back to underground.

Both outcomes are plausible. BTW such bacteria was already found, it is just hard to grow on industrial scale (and Yale bio-magicians are working on it).

• These models aren't accurate over the short time frames they are designed to model, using them for predictions 500 years away would be worse than throwing darts. Putting it another way, you would be just as wrong but with extreme confidence, lol and some people would assume that you were correct based upon the use of a computer model. Apr 9 '15 at 3:12
• @Jim2B - throwing darts is a valid method which works better than human intuition - you might heard about en.wikipedia.org/wiki/Monte_Carlo_method - it all depends how honest are your inputs. Facts are stubborn thing. Jan 23 '18 at 4:20

I don't know much about weather, but I do know Universe Sandbox$^2$ offers a basic climate simulation where you can get the world to snowball from changes in CO2, throw big rocks at it, move it closer to the sun etc. This might not be scientifically accurate, but it would give you very satisfying pictures to look at.

• This is a good answer. And its available on steam! Jan 23 '18 at 1:24

I would love to put something brilliant, witty, helpful and profound here, as this question deserves it.

Sadly I don't have these at hand. I can tell you that there is software (opensource) called "Earth System Modeling Framework". It is not an off-the-shelf ready solution, and is known to require substantial computing resources but might help point you in the right direction.

To your research thus far, have you considered that non-linear events would be a near certainty? Methane Clathrate releases, depopulations' effect on reforestation, ocean levels and salinity, changing ocean currents' effect on coastal temperatures, etc.

I look forward to your publications, one of the many things that set apart authors like Michael Crichton was the research behind the pages of words. Yours is a similar course, I wish you success !!

I know a lot about coarse paleoclimate models. I can help a little bit here.

In brief, I can guarantee you that no desktop model exists to do what you want; this is firmly supercomputer territory. At best you'll need to sneak a peak at other people's simulations. Even this will be hard, hopefully you can find reports from a CMIP5 model that break out the decade by decade temperature and precipitation map. WRI's Cait tool or the https://climexp.knmi.nl/start.cgi might be of help here. The Potsdam climate inst. might also have nice pictures.

Overall, you can sketch out the plausible from a suite of simulation results and write a story around that.

The biggest stories to play out here will be human and political -- when does a particular region break down in response to a few years of less-than-sustenance harvests? How does that snowball geopolitically? -- that are more in the control of the author than they are slaves to a computer prediction.

I suggest that the movement of the ITCZ and the consequent precipitation, alongside a general warming that enhances storm events and lethal heatwaves will be the main global stressors. Flashpoints could be a drying of Africa (precip limited crops), flooding of bangladesh as sea levels rise, a food crisis in Indonesia if the wet season no longer allows three rice harvests, or Pakistan/India given the Bangladesh migration plus their own monsoon concerns.

EDIT: I did just look at edGCM, which purports to do something close to what you desire. Great, well done those guys and I recommend taking a look. Word of warning: They must use a lot of parameterisations to simplify their model, which will almost certainly mean that they have linear extrapolations around current behaviour instead of working out all the physics from the ground up. This will be good enough to point you in the right direction, but I suspect that it does not treat oceans properly, nor have a real precipitation scheme that will demonstrate effects like the moving ITCZ. Which means it will be kind of correct, but rapidly lose accuracy if pushed into extreme behaviour like a fully-liquid arctic (because that would fundamentally alter ocean dynamics, and nothing short of a supercomputer will be able to solve for new dynamics and predict all of the knock-on climate effects).

We currently can't even predict 550 hours in the future with take-it-to-the-bank certainty.

Make the global climate whatever it needs to be for the purposes of your narrative. Do keep it plausible, as the marching orders dictate, which means that:

• Summer is warmer/hotter/not as cold as winter;
• The equator is warmer/hotter/not as cold as the poles;
• Interior continental areas have greater temperature extremes than coastal areas;
• The water that evaporates in one place eventually has to precipitate somewhere;
• Rapid changes in local weather are rare.
• 550 hours: irrelevant . The short term scale weather doesn't matter, only averages and trends. The chaotic nature of short term means that you can't predict the rainfall on a specific day, but on different scales it becomes tractable again. Jul 7 '16 at 22:00
• I don't think he's writing a simulation, and such models are well beyond your checklist. Jul 7 '16 at 22:01