Why Climate Modeling is Still Just ‘Garbage in, Garbage Out’
Dr. Javier Vinós, who recently joined us for our third live webinar whose recorded version you can watch here.
Has written a compelling book Solving the Climate Puzzle that offers both a convincing overview of the complexity of climate today and an intriguing alternative explanation of both short- and long-term warming and cooling to the “Enhanced CO2 Hypothesis”.
It does take some cheek to challenge orthodoxy in any field, but his justification for putting forward his own theory is that the conventional one actually performs very badly despite all the sound and fury of its advocates.
And while we strongly recommend the book and webinar on general grounds, we want to focus here on one particular aspect of his argument: that the computer models used to simulate climate are so feeble that to present them as reliable borders on imposture.
To some extent it’s one of those things hidden in plain sight: out of the hundreds of millions of years of climate vaguely like our own, with complex multi-celled life in an oxygen-rich environment, the models basically only get about 35 years in the late 20th century right.
When atmospheric CO2 and temperature rise together, because their creators were dogmatically certain the former drove the latter. Give them any other period of time, long or short, and they muff it because they’re not accurate representations of real-world relationships and data, they’re a projection of their creators’ largely evidence-free convictions.
For instance “known warm-pole paleoclimates are notoriously difficult to reproduce with climate models.” Or “models do not accurately reproduce this small interannual variability in albedo” despite it being “a critical aspect of the planet’s climate and energy balance.”
Or “Climate models have been unable to explain why wind speed shows multidecadal trends” or “climate models don’t show the observed [Atlantic Multidecadal Oscillation] oscillatory modes and are unable to project the expected behavior”. The more we read, the more we noticed this kind of comment over and over again whether discussing evidence or processes, and whether discussing past or present.
The book is, as we say, a compelling read. Though at times you may well, as we did, find yourself fearing that you’ll definitely fail the exam, unable even to remember the difference between the Holton-Tan effect and the Dobson-Brewer circulation.
Which is, again, a demonstration that even in this reasonably high-level overview, climate is enormously more complicated than anyone parroting the “simple physics” or “settled science” slogans begins to suspect.
Whether it’s a good Christmas gift for the alarmist on your list depends primarily on whether they have a reasonably open mind. For instance Vinós explains that:
“Climate models are incredibly complex and fragile. While there are different types of models, a state-of-the-art general circulation model (such as those participating in the 6th Intercomparison Project) with a grid resolution of 1×1° (about 100 x 100 km, 60 x 60 miles) and 30 layers consists of a staggering 2 million cells.
Typically, these models track seven variables – wind speed in 3 dimensions, pressure, temperature, density, and water vapor content – for each grid cell.”
And anyone who made it through the Holton-Tan and Quasi-Biennial Oscillation and so forth will instantly recognize that a model that claims to encompass all this complexity in just seven variables is already somewhere between delusion and insolence.
Especially when it works at a resolution of 10,000 sq. kilometres when, as he also notes, crucial climate processes including cloud formation take place at a molecular level.
There are other internal issues that he highlights:
“these models are iterative programs, where the output of one time step serves as the input for the next, making them inherently more prone to instability than other types of computer models…. In addition, certain processes lack equations that can describe them accurately.
In such cases, these processes are approximated by numerical parameters. If a model doesn’t perform as expected, these parameters are adjusted until the desired result is achieved.”
So it’s not so much garbage in, garbage out as it is dogma in, dogma out. Even the fragility is no joke:
“A climate model may consist of 2 million lines of code. Because of the interconnectedness and iterative nature of their components, these models are exceptionally fragile, unlike the real climate.
A recent example of this fragility is when scientists discovered a flaw in the CESM2 model in how it simulated the interaction between moisture and condensation nuclei, which affects cloud formation.
It took a team of 10 scientists five months to identify the problem and correct the error in the data. Any change in a model can easily derail it.”
That they so readily produce preposterous results with a small tweak to a function or parameter again suggests that they are not remotely reasonable representations of reality.
Indeed they don’t even appear to be, which again anyone coming to the field from the outside would consider compelling even though, or especially because, the insiders seem to have excessive tunnel vision on this point:
“A telling example of how climate models represent a model world rather than the real world is their inability to provide an accurate surface temperature for the planet….
Scientists aren’t concerned that climate models differ by 3°C (5.4 °F) in simulating the planet’s temperature, even though that difference is half the temperature that separates our interglacial period from the Last Glacial Maximum.
What matters to them is the consistency of the changes over time within the simulations and the fact that the ‘cold world’ models predict similar changes as the ‘warm world’ models when subjected to the same forcings.
It is surprising, to say the least, that a 3°C difference doesn’t significantly affect the performance of these models.”
Except all that complex code is devoted to one thing and one thing only: making them repeat ad nauseam that CO2 drives temperature.
It’s noteworthy that Vinós himself came to the topic with a background in “molecular biology, neuroscience, and cancer research”.
So he brought a finely-honed scientific mind but was not inculcated by the gatekeepers as a young aspiring researcher into the conventional pieties and the dominant “paradigm,” a term he often uses.
Which is one reason he noticed so many holes in the orthodoxy and in particular in those famous models.
After many criticisms of them by-the-way in dealing with specific aspects of climate, he finally provides a roundup based on just four months of reading papers on their failings that surely ought to make any alarmist blanch.
Especially as Vinós calls it “only a minimal fraction of what is published on the subject” and which we reproduce a great length because it’s a classic case where quantity has a quality all its own:
“* Models tend to overestimate the cooling and show slower recovery after volcanic eruptions. * Models fail to accurately represent the stratospheric response to solar changes. * Models fail to predict severe winter weather resulting from Arctic amplification. *
Models struggle to reproduce the cooling period observed between 1945 and 1975. * Models fail to simulate climate shifts, such as the one in 1976. * Models fail to reproduce historical patterns of ocean warming. *
Models fail to capture temperature trends in the tropical troposphere and stratosphere. * Models failed to predict the divergence of Arctic and midlatitude winter temperature trends. * Models fail to reproduce Northern Hemisphere snow cover trends. *
Most models underestimate warming in the early 20th century and overestimate warming after 1998. * Models tend to overestimate atmospheric warming. * Models predict midlatitude ozone trends in the lower stratosphere that are inconsistent with observations since 1998. *
Model predictions are inconsistent with observed changes in the sea surface temperature gradient in the equatorial Pacific Ocean. * None of the models accurately reproduce the increase in summer high-pressure blocking over Greenland. *
Models lack global-scale multidecadal variability. * All models show warming in the tropical upper troposphere that is absent from observations. * Models exhibit a ‘signal-to-noise paradox’ where they predict observed climate variability better than their own variability, indicating an underestimated signal-to-noise ratio. *
Models show a cold bias in the equatorial cold tongue. * Models do not realistically reproduce the observed annual cycle of albedo. * Models do not accurately capture the small interannual variability in albedo. *
Models generate a spurious double Intertropical Convergence Zone in the tropical Pacific. * Models fail to reproduce the interhemispheric albedo symmetry. * Heat transport is climate-state invariant in models despite large changes in the temperature gradient. * Models poorly simulate temperature trends in the lower stratosphere and inconsistently reproduce tropical tropopause temperatures and water vapor changes. *
Models hindcast a strengthening of the Brewer-Dobson circulation in the lower stratosphere during the second half of the 20th century, in contrast to observations. * Models underestimate the Holton-Tan effect, and each model represents the relationship between the Quasi-Biennial Oscillation and the polar vortex differently. *
Models produce ten times smaller interannual changes in ocean latent heat flux than observed. * Models simulate enhanced warming over Antarctica – Antarctic amplification – whereas no warming has been observed for this continent.”
Aren’t they ashamed? They should be. Especially as they themselves would never set foot in an airplane designed using models that defective nor, we trust, invest in a stock based on such manifestly feeble simulations.
See more here Climate Discussion
Please Donate Below To Support Our Ongoing Work To Defend The Scientific Method
PRINCIPIA SCIENTIFIC INTERNATIONAL, legally registered in the UK as a company incorporated for charitable purposes. Head Office: 27 Old Gloucester Street, London WC1N 3AX.
Trackback from your site.
Tom
| #
Global warming GIGO is competing with the medical establishment’s GIGO.
Reply
VOWG
| #
Computer modeling is always GIGO. It gave us covid. Now that should be enough to tell anyone to stop paying attention to computer models.
Reply