Here is where I started on my self-directed course in "Climate Models 101". This is highly recommended. Here's a sample:
My M. Sc. in math allows me to at least read these equations, if not understand them. I note the caveat at the bottom, which admits right off the top that these equations assume a world that is not quite ours. This theme is repeated over and over in the IPCC reports but pushed aside in the "executive summaries." I do happen to know that these equations describe a chaotic physical situation, even in the ideal case, which means they cannot be used to predict anything with confidence. In fact, this discovery was what launched the whole field of Chaos Theory. The clue is that some entities appear on both sides of the equations above, a sign that the system has feedback. Feedback is the fingerprint of chaos. We also see the IPCC models struggling with feedback.
With this background, I move on to the "meat and potatoes" of Climate Science:
"Demystifying Climate Models" is thorough, readable, and completely free from Amazon (and perhaps other sources). It's essential reading for anyone trying to decipher IPCC reports. The book gets down and dirty in the details. There is no math in the book, but those who understand the math (such as from "Physics of Climate Change" will get a fairly good idea of how that math figures into climate modeling.
Obviously, coupling is a major source of "model uncertainty."
"Demystifying Climate Models" is thorough, readable, and completely free from Amazon (and perhaps other sources). It's essential reading for anyone trying to decipher IPCC reports. The book gets down and dirty in the details. There is no math in the book, but those who understand the math (such as from "Physics of Climate Change" will get a fairly good idea of how that math figures into climate modeling.
One subtle point we should notice. All diagrams are mini-models, whether they are "marching bars" or the more sophisticated conceptual model below. Virtually all of the economics you learn in the "101" course is in terms of conceptual models with graphical expression. They do not reflect the real world. That is an entirely different matter.
A key takeaway from the "Demystifying" book is illustrated by the graph below:
Total Uncertainty
Several factors affect how much we can trust a model:
- Initial condition uncertainty (green) reflects uncertainty in our knowledge of the present situation or (if the model is run on the past, say 1000 years ago) how uncertain we are about the conditions then. For me, the key here is that the models are not great at "predicting the past" since our detailed knowledge of the past is sketchy.
- Scenario uncertainty reflects our uncertainty about some future scenarios. What kind of a word will we be talking about? In the IPCC, there are standardized scenarios that reflect C02 emissions according to human decision-making. This goes a long way to nailing down exactly the world we are talking about, but the scenarios may not describe that future world well.
- Model uncertainty comes from all kinds of potential errors in the models, which are discussed in detail in the book. The most interesting one relates to the difficulty models have in representing small-scale events such as hurricanes. In general, handling water is tricky, and water is by far the most important greenhouse gas.
I was surprised to find that there are models for specific systems, such as the ocean, land, atmosphere, and ice. Each model type is developed separately according to the most relevant science. These are then "bolted together" by a process called "coupling," which attempts to make things consistent where the different systems overlap or come in contact - where the atmosphere touches the Earth, for example.
 |
| IPCC Model Interaction |
The book gets into how each model divides the earth into grids or chunks. The chunks need not be all the same size. Some attempt is made to make the grid finer where expected effects will vary at a small scale, such as over cities or cropland. Different models (say ocean and atmosphere) need not use the same grid, nor will different versions of the same model type use the same grid, since you can "plug in" different models, say, for atmospheric aerosols. That depends on what you want to get out of the model, which is perhaps the most important point in the entire book. You configure your model for a specific purpose - to support some decision process or answer some specific question.
It is left to the reader to decide if the crazy zoo of models produces information to support decision-makers, as the IPCC desires. While they don't quite come out and say so, very few people have the background to understand what the models are doing, to explain the results, or to develop policies in response. In other words, the usefulness of climate science is in the eye of the beholder.
I was impressed by one remark: that uncertainty should not be used as a reason for a policy decision, such as inaction. In such a situation, more certainty is unlikely to change the decision. That's something to wrap your head around. Call it the "uncertainty fallacy."
A minor surprise was how the model can be pretty good for one measure, say surface temperature, but no help for another, say ocean level rise. It turns out that the more we know about the ocean's surface level, the crazier it gets. For example, if you go from the East Coast of the US toward Africa, you are going uphill! The ocean is actually a bit tilted. Also, the land is sinking in some places (Virginia coast) and rising in others (Greenland). Flooding is a major source of climate panic (including in "Physics of Climate Change") but climate models don't do a good job of dealing with such small variations and uncertain data. Another is with hurricanes, which are "small" by model grid standards. Models do a bad job of simulating hurricanes, yet the panic doctrine always claims frequency and severity of hurricanes is getting worse and worse.
I get the impression that the model's general prediction of temperature changes is pretty trustworthy, for example, in the Arctic. On the other hand, all the other disastrous implications are less certain or not in the models at all. For example, we can be pretty sure the Arctic will get warm but far less certain if or when the ice cap will melt. That's because "ice" is a separate block in the model and subject to the poorly understood nature of water vapor, which, I remind you, is the main greenhouse gas.
I hope I have encouraged the reader to pick up this book. After all, I am giving a summary of a summary here. It's particularly valuable if you intend to go ahead and actually read the IPCC report instead of assuming what's in it or assuming it's bunk.
===========
P.S. It has occurred to me that the shotgun marriage (coupling) between independent models in the climate field is rarely attempted between models in other fields, which, in theory, should describe the same world:
- Ecology - specific effects on living systems;
- Materials requirements, such as supply and demand for lithium, copper, etc.;
- Demographic models - changing effects of an aging population;
- Projections of global wealth;
- Remediation programs associated with IPCC Climate Models;
- Geopolitical models that consider the impact of new alignments and trade blocks;
- Fine-tuned consideration of the actual standard of living improvements in the developing world;
- Impact of AI and robotics (another apocalyptic scenario).
- Insect Apocalypse;
- Fresh Water use/shortage - the "Global Hydrological System";
- The general collapse of the ecosystem;
For example, in the box marked "Humans" above, all human factors are collapsed into a handful of scenarios. In the worst case (business as usual), the world population is assumed to hit 14 billion by the end of the century. This is not what population models predict (9 billion peak seems to be the consensus number). The same models assume growing GDP and associated consumption, ignoring the effect and spiraling prices for materials such as copper. Proposed "Net-Zero" utopias ignore the need for copper in all forms of generators, including wind and solar. Fifth-generation nuclear gets rosy treatment, ignoring the geopolitical problems in acquiring and sharing technology. Huge costs associated with rebuilding our supply chains in response to geopolitical changes (such as China vs. the US) are seldom considered. Our current energy-intensive global supply chains (and global peace) are assumed to be available to make everything from nuclear reactors to windmills.
