It is absolutely absurd and dishonest to run a computerized climate model 480 years out into the future. These models are not that good and not that accurate. We can't even predict weather very well beyond 2 weeks.
The climate is subject to chaos theory.
Chaos: When the present determines the future, but the approximate present does not approximately determine the future.
Uncertainty is an exponential function of time in chaotic, non-linear dynamic systems like the climate. The year 2500 could just as easily be a deep glacial period.
I'm pretty sure that the Earth going round the Sun and the Earth's axis being slightly tilted counts as a model using which you can predict the seasons.
There is nothing wrong with running model in any way a researcher wish. It would be wrong to make predictions about a real system, without studying validity of that prediction.
Authors clearly state "an earth system model shows ..." not the "Earth is going to melt". They say "we encourage other model builders to explore our discovery in their (bigger) models, and report on their findings."
Authors work with their model, use it to calculate some results, they do not pretend, that the results are the Truth. Authors know what validity is, they ask others to take a look at it.
Abbreviated original title is better than the editorialized one: "An earth system model shows self-sustained melting of permafrost even if all man-made GHG emissions stop in 2020"
Or shortened to "Earth model shows self-sustained melting of permafrost if emissions stop now"
Can you please explain why they are not other than just proclaiming they are not? I have never understood this reasoning other than proclamations that they are not.
If I pour water into a bucket at a set rate, I could not with all the computing power in the world tell you where a particular drop of water will end up in that bucket an hour from now, but I can tell you to the millimeter what the depth will be at any point in the future. Hopefully the analogy is clear.
I'll make it a little bit more accurate. If I poor water into a bucket at a set rate, it takes incredible computing power to simulate the hydrodynamics and understand how the water is splashing, the patterns of ripples, and the turbulence in the water, but I can tell you that when I've poured in 2 gallon, the bucket will be full.
I'll make it even better.
I'm spraying water at my dog. Weather is predicting where the hose spray will fly, and how much make it on the dog at any moment. Climate is predicting how wet the dog will be in the end and what will get wet if dog goes in the house.
My (blue) wife claims she can predict the exact amount of water on my dog, and says if I let the mutt in, 1.4 cups of mud will get on the stairs, and I shouldn't do it.
I (red) correctly say that's nonsense, and that she can't predict that with any accuracy. I keep spraying the dog, and let the muddy dog in the house.
Dog doesn't go for the stairs, but heads for the $5000 sofa, staining it to where it's a $100 Goodwill sofa, and shaking mud onto the AV system. Stairs are clean, though.
TL;DR: We can't predict shit will get f-ed up to nearly the degree Democrats say they can. We know shit will get f-ed up, though, with pretty good certainty.
(My own opinion is we've badly underestimated the impact of chemical effects, like ocean acidification, and overestimated the effects of temperature change; but the dog might go for the pantry instead)
It's a matter of scale. If I go outside and turn on a leaf-blower I can tell you with relatively high confidence the speed and direction of general air movement five feet out from the nose. On the other hand, if I tried to tell you the speed and direction of a dozen particles at X, Y, Z location in coordinate space, the effects of chaos (particles individually colliding, etc) would make this infeasible with current technology. Similarly, describing climate is like describing the general direction and speed of air moving out from the leaf blower, while weather is like trying to make the same prediction for small groups of particles within the system. They might sound like similar tasks, but due to chaos one is doable and the other is nearly impossible.
All that being said, with regards to this particular study at the scale of 500 years it's likely our climate models nearly completely break down, as we've yet to successfully create one without significant divergence even a few decades out (although maybe that has changed as by definition the best ones we can test against only include those made before the year ~2000)
Right, I totally get the leaf blower and thank you for writing that. But the scientific boundary between climate and weather cannot be a analogy. It has to be more rigorous than that.
It's not a question of reasoning. It's a question of definitions. The most rigorous definition I've seen is that climate is a probability distribution over things like weather. Climate over a year in San Francisco will get you a distribution of temperatures, humidities, smoke, etc. Climate on Nov 13, spread out over many years is another distribution over weather.
It's easier to deal with these distributions because you don't need to predict the next year of weather to predict what global temperature properties keep the earth in equilibrium with incoming solar energy, for example.
Fantastic! A rigorous explanation (thank you). So you are saying that the mixtures of weather distributions are stable but the individual distributions can shift. I gotta think about that and dont know enough about weather distributions or probability to understand why climate (the mixtures) can be stable but individually weather cannot.
Weather is unstable because you need to know many factors (temperature, humidity, cloud cover, etc) over every single piece of land and water (over many different terrains). And those are unstable minute by minute. To report the weather, you need a moving weather map.
By contrast climate is "simple" in that it can have fewer variables. The most basic climate model is literally zero-dimensional: it treats the earth as a homogeneous gas mixture and that's it. It can be reported in a single variable, the temperature.
The zero-dimensional model won't tell you all of the things that happen: some places get more rain, some less, some even get cooler. There are more complicated models that do that, and to make specific policy recommendations you need those. But there is a crucial discussion that comes from that single number: yes, the world is getting warmer, because of humans turning carbon in the ground into CO2 in the atmosphere, and it's bad enough that action needs to be taken.
The model is still unstable, but much more tightly constrained, to within a fraction of a degree C per year. That's because the atmosphere is so large, and the sheer mass of it means it has to change slowly, but predictably. Over very long scales (tens of thousands of years) additional factors create more instability, but they're not pressing problems the way highly predictable century-scale changes are.
Consider the distribution of places you spend your time during peak shelter in place. It was probably mostly at home. Your bedroom, the kitchen, the bathroom, and occasionally going out to the store.
Take the math out of it for a second. I’m not nearly smart enough to predict which room you’ll be in during a specific minute of a specific day. Predicting that you’re going to be hungry at exactly 12:42 and you’ll go to the store this Sunday at 9:21 is well beyond me.
But I could be much more accurate if I abstract it a bit to a few important properties. During shelter in place, you probably spend 25-35% of the time asleep, maybe 0.1-0.5% at the store (once every 1-3 weeks maybe), etc. I can even confidently predict that as shelter in place relaxes, you’ll likely go out more often, but probably still very rarely to the grocery store more than twice per week.
Your precise movements are impossible to accurately predict more than a few moments out, but their distribution from one day to the next during shelter in place is pretty stable. On a longer timescale, that distribution will shift once shelter in place lifts, and it’s even reasonable to predict how it’ll change.
Another timely analogy might be modeling specifically who has COVID-19 at a point in time versus modeling the distribution. The percent of people in a place who have it is the distribution in question here: the probability that any given person there has it is a simple distribution over true or false. There’s no question of stability over time here because it’s a distribution over people instead of time. You can model how that spreads over time and location so much more easily than predicting the specific individuals who will get it and transmit it.
Disclaimer: I'm not knowledgeable in that topic. But from a general physics PoV:
- weather is mostly about complex local interactions between air pressure / surface temperature / air flow (convection) / humidity etc. these are described by differential equations with very chaotic behaviour (small changes in inputs may have extreme effects on the outputs the further you run the model)
- climate is more about long-time effects of the energy input/output of the whole planetary surface and atmosphere. Mostly solar radiation (input) vs. black-body radiation (output). There are also some complex interactions and feedback cycles e.g. change of albedo (cloud cover), and changes of atmospheric composition (i.e. methane due to permafrost melting), but local weather patterns may cancel out when looking at the sum-total of surface thermal energy.
Climate is also a more macro level view of a system. Think Entropy.
Imagine Oxygen molecules in a box. The individual movements are complex and unpredictable but at a very high level macro view we can see all of these movements in aggregate as temperature.
Temperature is not only much more stable then the chaotic movements of individual atoms but it is predictable and controllable. I can heat a kettle of water to a very specific temperature and I can accurately predict the rate in which it will cool down when I stop heating it. I cannot however control or predict the actions of specific molecules within the kettle.
The lower level you go the more chaotic things become. The higher you go the more deterministic things appear to be.
Weather is the day to day fluctuations in atmospheric conditions. They are extremely sensitive to minor changes making it hard to predict. It’s even possible to actively change the weather by seeding clouds to start rain for example.
Climate involves long term trends which are quite stable and predictable.
It’s kinda like the stock market. There may be lots of fluctuations in a given day, but most traders are interested in long term trends which tend to be more stable and predictable.
I really like this definition because it is understandable. However we do not know long term where the stock market will be. There are countries that have been wiped out and their stock markets decimated (e.g. country that had a revolution). The US stock market is an exception in that distribution.
Perhaps you mean macro vs. micro as in economics. Even there micro is far more predictable than macro. Macro is almost voodoo when it comes to prediction of any sort on any timescale.
I think you are mixing weather and climate. Climate is long term average of weather (directly from definition of word "climate"). Weather is very hard to predict 2 weeks ahead. Climate is stuff like "average rainfall in march in past 50 years". Some processes affecting climate may also be chaotic, but they are slow (by definition, otherwise they would be part of weather), and so I don't think 500 years is beyond speculation at all. Glacial periods have lately been around 100000 years, so 500 years to future we are not going to be much more or less glacial than we are now.
Rapid rises of temperature have coincident with mass extinctions before. Climate for past 10000 years have been remarkably stable, but is dependent on stable ecological processes (see Attenborough latest documentary on Netflix).
From what I've head: these models are very accurate in estimating the mean global temperature difference even for long periods of time.
While the global trend is clear, they cannot estimate local developments very accurately - these are subject to chaos, and local changes.
Note that the authors do not try to predict any local effects.
Well, i think the (honestly quite basic) climate model that came into being in 1974 is 95% accurate then, since it predicted the rise in average temperature until 2020.
What climate modelists do is take the data from 1950, put that in their models, and see how well ir predict the temperature and hydrometry of different areas. I think what they were working on in the 2010s was how much the artic polar circle was at risk of breaking more often, and what that would mean for the average temperature (not much) and precipitations (probably drier winters/springs in europe). Sadly they did not amke prediction about how much the polar winds descending south would kill buds and take a toll on orchards.
However low-confidence those climate projections to 2500 may be, anyone should by know better than to commit the tired old fallacy of equating weather and climate. No matter how chaotic the system, there will not be ”a deep glacial period” in mere 500 years.
There are plenty of places in earth's history where you could apply a predictor of future temperatures based on current conditions and get a better-than-random output 500 years ahead.
Do you have a good reason to believe that there will be a reversal of recent warming trends given the concentration of carbon dioxide and accumulation of other pro-warming gasses in the atmosphere?
I don't have a good reason but I wouldn't discount the possibility. As life reacts to the higher carbon dioxide levels we will enter into uncharted territory.
The biggest possible source would be human climate engineering.
That would be a huge feat to pull off but if we desperate enough who knows what we could do in 500 years.
Don't weaponize your poor understanding of chaos theory into climate change denial.
One of the defining properties of a chaotic system is that phase-space is bounded. Meaning while it's fundamentally difficult to make predictions about a specific trajectory beyond a certain point, one can still say things about trajectories on average.
Giving a year 2500 forecast warning is as silly as using any forecasts in in 1500 to address the situation of today, especially given that change is accelerating.
> approximate present does not approximately determine the future
Well that's an absolutely absurd and dishonest take on chaos theory.
I can tell you that the five year average of all stocks will probably go up by a predictable amount in the next ten years even if I can't tell you what an individual stock will do two weeks from now. The climate of the earth is an even more stable system that doesn't have to worry about predicting when the next political instability will occur. These models are good and accurate and can be used for predictions beyond 2 weeks.
A greenhouse is a very simple thing. What's absurd is to forget to mention than a 3 degrees average rise means a six degrees rise over landmasses. At the deepest of the last glaciation with oceans 120 meters down and ice covering half of Europe and all of Canada, we only had an average of 5 degrees colder than 1900. Holocaust 2.0, your descendants are all invited. (It's Greek for everything burning)
wouldn’t people just migrate away from the equator and habitate places like Northern Canada, Siberia, etc? Yeah it’d be terrible to abandon our cities, but we’d have a long runway to prepare and migrate. Life always finds a way...
The problem with that is that agricultural output in a 3 degree higher world may not be sufficient to sustain a civilization of almost 8 billion people. "Life always finds a way" may be a little sarcastic once people start starving.
One or two million people fleeing from the middle east to Europe caused large political trouble. Imagine what happens when hundreds of millions are displaced.
Let me propose the following thought experiment: for all the folks who claim they can predict climate 400 years (or 100 or 30 years) out, please take the climate models and run them for only 1 year out. This should give you much much lower error bars ergo higher confidence.
Use it to predict winter in NE 1 year out? I would posit that if they can predict average temps 400 years out then average temps 1 year out should have higher confidence? These predictions can be used to trade heating oil futures and make tens of billions. I do know that energy firms employ many climate modelers but unsure of the accuracy of their forecast s even 1 year out.
Imagine you have a gun mounted in a bench vice, pointing down a shooting rage towards a target.
You can map out 50 shots from that gun, and pinpoint the average middle, and how much individual shots diverge from that middle. Then you take the average of another 50 shots and notice the recoil has pushed the average middle slightly upwards. So you make a model predicting the average middle in another 50 shots, accounting for the degree change from the recoil meaning a further distance than the previous to average middles.
Then comes along a guy asking you to predict just 1 shot, the next one, not an average. Which of course is much less accurate. That is what you're suggesting.
Heating oil futures and natural gas futures for the various regions are tied to the average temperature since oil/gas heating usage is tied to temperature across days and not just the min or max.
So to go back to your excellent analogy, I am certainly asking for the average middle (the first moment) and not a point prediction.
ps: you should write more, we need better scientific explanations!
The climate is subject to chaos theory.
Chaos: When the present determines the future, but the approximate present does not approximately determine the future.
Uncertainty is an exponential function of time in chaotic, non-linear dynamic systems like the climate. The year 2500 could just as easily be a deep glacial period.