Tag: CMIP models

Climate Models’ Fatal Flaws Exposed: Major Open Issues in Detection, Attribution, and Projections

Italian physicist Nicola Scafetta has long argued that IPCC climate models (the CMIP ensembles used in AR6 and updates) underestimate solar influences and natural variability while over-relying on greenhouse gas (GHG) forcing. His latest analysis (a 2026 update or related paper) claims the climate is 4–6 times more sensitive to solar forcing (including indirect effects like cloud modulation via cosmic rays) than the IPCC assumes.

Italian physicist Nicola Scafetta has long argued that IPCC climate models (the CMIP ensembles used in AR6 and updates) underestimate solar influences and natural variability while over-relying on greenhouse gas (GHG) forcing.

His latest analysis (a 2026 update or related paper) claims the climate is 4–6 times more sensitive to solar forcing (including indirect effects like cloud modulation via cosmic rays) than the IPCC assumes.

It attributes roughly 50% of warming since 1850 to natural solar-driven variability, ~30% to humans, and ~20% to warm biases in land-surface records, with equilibrium climate sensitivity (ECS) to doubled CO₂ around 1°C (range 0.7–1.3°C).

He also argues models fail to reproduce natural variability or the assumed positive water-vapor/cloud feedbacks that amplify CO₂ warming.

The IPCC does not say “humans are exactly 100% responsible” in a simplistic way, but its assessments come very close for the modern era. From AR6 (2021, with no major reversal in subsequent updates):

  • Best estimate of human-induced warming (1850–1900 to 2010–2019): 1.07°C.
  • Observed warming: ~1.1°C (range 0.9–1.2°C for GMST).
  • Natural forcings (solar + volcanic): –0.1°C to +0.1°C (near zero net).
  • Greenhouse gases: +1.0°C to +2.0°C (partly offset by aerosol cooling of 0.0°C to –0.8°C).

It is “unequivocal” that human influence has warmed the atmosphere, ocean, and land, and “virtually certain” that internal variability alone cannot explain post-1950 warming. Attribution studies (detection-and-attribution, “optimal fingerprinting”) find human activity explains all or virtually all observed warming since the mid-20th century, with natural factors too small to account for the trend.

This is not based solely on models. It integrates:

  • Direct observations and energy-budget constraints.
  • Spatial “fingerprints” (e.g., tropospheric warming + stratospheric cooling matches GHG, not solar).
  • Paleoclimate records showing current rate is unprecedented over millennia.
  • Multiple independent lines (satellite, ocean heat content, etc.).

Scafetta’s work is published and part of legitimate scientific debate (he has papers in peer-reviewed journals critiquing CMIP performance on decadal/multidecadal variability, solar reconstructions, and ECS). Key points in his favor that align with broader uncertainties:

  • Some CMIP6 models run “too hot” (high ECS >4–5°C); IPCC already down-weights them for projections.
  • Cloud feedbacks and aerosol effects remain the largest uncertainties (± large ranges).
  • Solar indirect effects (e.g., via cosmic rays influencing clouds) are debated; some empirical studies suggest amplification beyond direct total solar irradiance (TSI) changes, though IPCC rates the evidence as low confidence and small in magnitude since 1850.
  • Surface temperature records have known issues (urban heat islands, station siting), though adjustments and multiple datasets (satellites, reanalyses) largely converge on the observed warming.

However, several issues prevent this from “fundamentally” invalidating IPCC attribution:

Solar forcing is small post-1850. Direct TSI change is 0.1 W/m² or less; even with 5× amplification, it struggles to explain the observed ~1.1°C trend against the much larger, well-measured GHG radiative forcing (2–3 W/m² net anthropogenic). Long-term solar trends have been flat or slightly declining since the 1950s peak.

Models do reproduce key aspects of variability when all forcings (natural + anthropogenic) are included. They match the overall historical temperature trajectory and many patterns. Persistent discrepancies exist (e.g., tropical troposphere, some regional extremes), but these do not erase the GHG fingerprint.

Attribution is multi-method. Even without models, energy-budget studies, paleoclimate, and observed ocean heat uptake point to dominant anthropogenic forcing. Scafetta’s alternative (strong solar + low ECS) requires rejecting much of this convergent evidence.

Minority view. Scafetta’s conclusions (low ECS ~1°C, solar ~50% of warming) are outliers. The broader literature (including recent CMIP analyses) places likely ECS at 2.5–4°C (AR6 assessed range), and natural variability alone cannot explain post-1950 warming.

Scafetta’s work (and similar critiques from researchers like McKitrick, Soon, or Nikolov/Zeller) usefully highlights real uncertainties—clouds, solar indirect effects, data biases, and model tuning.

Climate science is not “settled” on every detail, and over-reliance on high-ECS models has been criticized even within the mainstream.

However, the claim that IPCC models are fundamentally flawed in a way that overturns ~100% anthropogenic attribution since 1850 does not hold up against the full body of evidence. No paradigm shift has occurred; detection-and-attribution studies remain robust.

Policy implications hinge on risk tolerance: even a lower-end ECS + modest natural variability still projects significant warming under high-emissions paths, with costs from extremes, sea-level rise, and ecosystems.

Ongoing research (better solar data, cloud-resolving models, improved observations) will narrow uncertainties.

Dismissing the human role entirely based on one researcher’s solar-centric analysis would ignore the weight of convergent evidence from physics, observations, and multiple model generations.

The debate is healthy—science advances by testing assumptions, not by declaring victory on either side.

_____________________________________________________________________________________

Detection, attribution, and modeling of climate change: Key open issues

This is the peer-reviewed scientific paper by Italian physicist Nicola Scafetta, published in Gondwana Research (April 2026 issue).

DOI: doi.org/10.1016/j.gr.2025.05.001

The Coupled Model Intercomparison Project (CMIP) global climate models (GCMs) assess that nearly 100% of global surface warming observed between 1850–1900 and 2011–2020 should be attributed to anthropogenic drivers like greenhouse gas emissions. These models also generate future climate projections based on shared socioeconomic pathways (SSPs), aiding in risk assessment and the development of costly “Net-Zero” climate mitigation strategies. Yet, as this study discusses, the CMIP GCMs face significant scientific challenges in attributing and modeling climate change, particularly in capturing natural climate variability over multiple timescales throughout the Holocene. Other key concerns include the reliability of global surface temperature records, the accuracy of solar irradiance models, and the robustness of climate sensitivity estimates. Global warming estimates may be overstated due to uncorrected non-climatic biases, and the GCMs may significantly underestimate solar and astronomical influences on climate variations. The equilibrium climate sensitivity (ECS) to radiative forcing could be lower than commonly assumed; empirical findings suggest ECS values lower than 3 °C and possibly even closer to 1.1 ± 0.4 °C. Empirical models incorporating natural variability suggest that the 21st-century global warming may remain moderate, even under SSP scenarios that do not necessitate Net-Zero emission policies. These findings raise important questions regarding the necessity and urgency of implementing aggressive climate mitigation strategies. While GCMs remain essential tools for climate research and policymaking, their scientific limitations underscore the need for more refined modeling approaches to ensure accurate future climate assessments. Addressing uncertainties related to climate change detection, natural variability, solar influences, and climate sensitivity to radiative forcing will enhance predictions and better inform sustainable climate strategies.

Scafetta argues these issues mean CMIP models are not yet reliable for precise attribution or high-ECS projections, favoring empirical approaches that incorporate stronger natural/solar roles and lower sensitivity. This aligns with a minority of researchers emphasizing under-modeled variability.

Mainstream assessments (IPCC and most CMIP analyses) view the core attribution as robust: human influence (primarily GHGs) is the dominant driver of observed warming since ~1950, with high confidence. Uncertainties are largest for regional extremes, decadal predictions, cloud/aerosol processes, and high-warming tail risks. Progress includes better large ensembles, machine-learning attribution, and process-based constraints (e.g., on clouds). No paradigm shift has overturned the fundamental understanding, but ongoing work (toward CMIP7, improved solar data, cloud-resolving models) continues to refine it.

Blockbuster honesty: Expert modeler admits they can’t predict extreme events, El Nino, tipping points, rain or river flows

Australia’s leading climate modeler wants a big new Climate Agency, and to make the case he admits the current models really can’t predict if rivers will rise or fall, if Antarctica will get bigger or smaller, if sea levels will rise much, or if El Ninos or La Nina will be more common or if the floods of Lismore will occur more often.

From JoNova

By Jo Nova

The science is settled, except when they need more money

Australia’s leading climate modeler wants a big new Climate Agency, and to make the case he admits the current models really can’t predict if rivers will rise or fall, if Antarctica will get bigger or smaller, if sea levels will rise much, or if El Ninos or La Nina will be more common or if the floods of Lismore will occur more often.

To give us some idea of how bad the current models are, he’s recommending we shift from models with 100 kilometer blocks to high resolution models with 1 km cells. These new models will be at least 10,000 times bigger than current ones, and if they increase the vertical slices, they could easily be one hundred thousand or even a million times bigger.

And if they get this super model, they’ll need 10,000 to one million times the energy. But now that we’ve wrecked the grid, good luck running those monster data centers off sunlight and breezes.

Full credit to Tony Thomas for digging through pages of turgid text and webinars to uncover the truth.

Andy Pitman, November 2024

Oh Boy it’s an eye-popping list.

In the past, Pitman has admitted climate change doesn’t necessarily cause more droughts. He’s also said he wouldn’t bet his superannuation (pension fund) on the climate models. This time Pitman admits the models are low resolution, have a lot of “critical gaps”, don’t resolve the oceans or clouds well, and that using the best CMIP6 models (the same ones the expert UN has beaten us over the head with) may risk “fundamentally wrong projections of future climate and its variability.”

Shouldn’t every Australian know this? I mean we’re spending half a trillion dollars to solve “climate change” but the models might be fundamentally wrong?  Doesn’t that matter. The seas might not swamp us, the rivers might keep flowing. Antarctica might not melt?

Where was Andy Pitman when Australia was wrecking its grid, destroying jobs and our lifestyles? Where were any of our climate academics when leaders of our political parties were telling us that every drought, fire and storm was caused by climate change and would only get worse?

The Ivory Tower elitists sat silently by when the government forces bricklayers, farmers and children to solve a climate problem that might be just a modeling error?

And now they want more money — and I would say we desperately need models that work, but what’s the point, if we can’t trust the modelers to be absolutely, scrupulously honest with us?

Climate Science You Can Believe

By Tony Thomas, Quadrant

Pitman even concedes that current climate models can’t predict whether natural disasters will become more or less common in the warming era. Remember his words when you next hear the ABC or Climate Council claiming that such-and-such storms and floods are “climate-fuelled”.

Using current CMIP models, or indeed the regional models that rely on them, therefore risks fundamentally wrong projections of future climate and its variability. — P17 of “The Decadal Plan“

Now they tell us? Modelers don’t know if we’ll get more wind droughts or cloudy days?

Right after we built 12,000 megawatts of weather dependent generators, we find out that the modelers have no idea whether we will get more long spells of cloudy windless days which cripple our grid. This new vulnerability is buried under the label “High impact weather events”. As if windless nights belong in the same category as storms and floods.

It’s hidden in one of the Five Key Questions of National Significance:

5. Where will changes in high impact weather events support and/or undermine net zero ambition and where can associated risks be managed effectively?

Like the rest of the document the language is obtuse, convoluted, and speaks with a forked tongue. They don’t want to come out and just say Australia badly needs models a million times bigger so we can predict the climate.

Tony Thomas sums it up:

He [Pitman et al] also admits that he and fellow climate alarmists have no idea:

          • when and where so-called “tipping points” might arise (wow, so honest!)
          • whether climate change will increase or decrease the Murray Darling water flows
          • whether an increase in CO2 will cause more or less rain for a given location
          • how climate change will impact cities and urban landscapes (Andy, stop upsetting the Melbourne and Sydney city councils’ climate crusaders)
          • how wind droughts and heavy clouding might undermine renewables and net-zero targeting (via week-long blackouts).

In a repudiation of the “settled science” notion the climate crowd has pushed for 25 years, Pitman now acknowledges that despite decades of study, the catastrophists still have no idea if Australia will see more El Nino, rather than La Nina, climate events, or even whether more vegetation will reduce or increase greenhouse emissions (so much for tree plantings offsetting emissions). “These are not easily solvable but offer profoundly different futures for Australia,” he admits (p13). Odd that we are to invest trillions in net zero when we have no idea what’s what.

We need models thousands of times better!

Pitman lays out many shortcomings of current “expert” climate models because he’d really like a much finer resolution model of just 1km2.

Ocean processes operate across many scales, and eddies in the Southern Ocean transfer considerable heat and nutrients14. These eddies are also crucial to the uptake, transport and storage of carbon15. Operating at scales of order 10 km, they are too fine to be resolved in ocean models used for climate projections. This means the role that oceans play in influencing climate are poorly resolved. – p17

As a result of their poor resolution, current climate models do not faithfully represent critical weather systems, and it is the amplification of extremes by weather system processes that cause the extreme events and consequential disasters we observe. For example, the Lismore (NSW) floods in 2022 were associated with multiple weather processes, initiated over the Southern Ocean and interacting with synoptic-scale processes and moist tropical air that led to a sequence of extreme weather events and catastrophic flooding. Global climate models cannot resolve these processes, and therefore cannot tell us if such events will become more common in the future. As the weather that produces extreme events is  connected globally, downscaling using high-resolution regional climate models cannot overcome the limitations introduced in the global models, as downscaling relies on the global models for information at its boundaries.-p17

These transcripts come from Andy Pitmans speech in the launch webinar 

The speech and documents make tough reading. You get the feeling they just didn’t want to tell us straight how bad the models are. I wanted to capture his exact words, for the record.

All those environmental sinks might becomes sources (so much for carbon farming, eh?) Shame about that business you set up…

7:30 [Andy Pitman] Some of these questions, you might think we have answers to….

For example, many people would recognize both terrestrial and marine as providing a critical ecosystem service, it takes up human emissions of CO2, and provides enormous support for Australians Net Zero ambitions … but there’s a little problem with this, we don’t actually know to what degree our terrestrial and marine systems will continue to support our Net Zero ambitions and positive environmental outcomes. They may turn into sources of CO2 and methane, in ways that really undermine our Net Zero ambitions.

We will definitely get more water or less water, more river, or less river, more plants, or less...

9:00 In addition Australia is demonstrably at risk of abrupt changes in weather and climate. Many of you would be familiar with tipping points… It matters a great deal if we could say things about when and where these things might be realized. At the moment we really can’t.

9:30  Water is obviously fundamental to the most arid inhabited continent on Earth. But we don’t actually know whether Climate Change will increase or decrease flows of water through systems like the Murray Darling. We don’t yet know whether changes in rainfall will be helped or hindered by the way ecosystems response under higher elevated CO2 concentration. Whether the higher water use efficiency of vegetation will help the flow of water through the Murray Darling or the vegetation will suppress the flow of water in the Murray Darling….

No we don’t know what will happen to the cities:

Urban Areas, it’ It might surprise a lot of you to know that the climate modeling systems we use internationally do not represent our urban landscapes.

From the PDF Launch document

Page 13: Plants, rain, El nino, who knows?

Some major challenges have been explored for decades — whether we will see a more El Nino or La Nina state in the future, or whether vegetation will help or hinder net zero ambition.

These are not easily solvable…

Page 17 We can’t predict extreme events, and they may not be getting worse, we don’t know:

The Lismore (NSW) Floods in 2022 were associated with multiple weather process, initiated over the Southern Ocean and interacting with synoptic scale processes and moist tropical air that led to a sequence of extreme weather events and catastrophic flooding. Global climate models cannot resolve these processes and therefore cannot tell us if such events will become more common in the future.

Further there is evidence the high-resolution coupled models simulate fundamentally different historical trends in tropical and Southern Ocean sea surface temperatures, reproducing recent observed changews which courses models cannot. They also exhibit greater low-frequency variability in midlatitude regions. Compare with Coupled Model Intercomparison Project (CMIP)-class models. Using current CMIP models, or indeed the regional models that rely on them, therefore risks fundamentally wrong projections of future climate and its variability.

Page 22: The current models are low resolution and can’t do sea ice, clouds, storms, ice sheets, plants, cities and farms.

Projections on timescales of decades to centuries at low spatial detail using Earth System models. The low spatial detail is balanced by large numbers of simulations. The lower computational cost means the more components can be included (e.g. Chemistry, fire, nutrients,) but some processes which may be extremely important are difficult to resolve (eh, sea ice, topographic forcing of clouds and storms, cloud processes, some ice sheet dynamics, vegetation demography, urban landscapes, agricultural areas).

Page 25 contains Antarctic surprises like how 70 billion tonnes of snowfall accumulated across East Antarctica contributing to a net gain of mass in Antarctica — which was a reversal in the mass loss trend over the preceding 20 years..

A future with more ARs [atmospheric rivers] leading to a larger accumulation of snow on the Antarctic continent would be experienced very differently in Australia to a future with fewer ARS. Specifically projections o

The modelers didn’t see that coming either.

REFERENCES

A Decadal Plan for Australian Earth System Science 2024–2033, released November 25th, 2024

The launch webinar 

— This one is for Julian. —

Turtle Model by SAIF 4 from Pixabay

Climate Model Bias 5: Storminess

It is uncontroversial that temperature gradients, at least in part, power our weather. Heat wants to flow from warmer areas to colder areas as it seeks an equilibrium temperature. On a larger scale, meridional transport is also, in part, a function of the temperature difference between the tropics and the poles.

By Andy May

In part 4 the impact of convection and atmospheric circulation on climate was discussed. When circulation patterns change, they change the speed and efficiency of the transport of excess heat from the tropics to the poles. When this heat transport is faster, the world becomes cooler since the poles have a weak greenhouse effect and are more efficient at radiating heat to space. The tropics are humid and have a very strong greenhouse effect because water vapor is the most powerful greenhouse gas.[1] Thus, the thermal energy (heat) transported to the low humidity poles is more easily sent to space than it would be in the tropics.

The temperature gradient formed between the tropics and the poles helps drive this heat transfer, but it is modulated by atmospheric and oceanic circulation and convection patterns. A very steep gradient, such as we have today, increases storminess.

The temperature gradient powers meridional transport and weather

It is uncontroversial that temperature gradients, at least in part, power our weather. Heat wants to flow from warmer areas to colder areas as it seeks an equilibrium temperature. On a larger scale, meridional transport is also, in part, a function of the temperature difference between the tropics and the poles. Leon Barry and colleagues write:

“The atmospheric heat transport on Earth from the Equator to the poles is largely carried out by mid-latitude storms.”[2]Leon Barry, et al., Nature, 2002

Strong mid-latitude storms, cyclones, and other storms carry most excess tropical heat poleward, and the more they carry the colder Earth becomes. The temperature gradient is not the only factor controlling meridional transport, there are many other modulators,[3] but it is an important factor, perhaps the most important one.

The atmosphere is a huge heat engine[4] powered from the surface (except some limited areas in the polar regions) that expels unused heat to outer space.[5] Just like in a train engine, as the heat moves through the atmosphere, it can produce work, the work is our weather, mainly storms. As discussed in part 4, tropical temperatures do not vary much, so the critical factor is the temperature at the poles. The necessary corollary to this concept is that weather is more extreme when the planet is colder, meridional transport is strong, and the temperature gradient is steep. It is less extreme when the planet is warmer, meridional transport is weak, and the temperature gradient is smaller. This is the opposite of the conclusions reached by the IPCC, who believe warmer weather brings more storms based on the CMIP models (AR6 WGI, page 1526-1527). But that is not what is observed. Each hemisphere has more severe weather during its winter months than during its summer months.[6]

Zhongwei Yan and colleagues, which include Philip Jones and Anders Moberg, found that extreme weather has decreased, especially in the winter months since the 19th century, as the world has warmed.[7] They note that the relationship between warmth and weaker weather is most pronounced in Europe and China in the critical winter months. In contrast to Yan, et al., AR6 reports:

“… both thermodynamic and dynamic processes are involved in the changes of extremes in response to warming. Anthropogenic forcing (e.g., increases in greenhouse gas concentrations) directly affects thermodynamic variables, including overall increases in high temperatures and atmospheric evaporative demand, and regional changes in atmospheric moisture, which intensify heatwaves, droughts and heavy precipitation events when they occur (high confidence). Dynamic processes are often indirect responses to thermodynamic changes, are strongly affected by internal climate variability, and are also less well understood. As such, there is low confidence in how dynamic changes affect the location and magnitude of extreme events in a warming climate.”AR6 WGI, page 1527

Thus, in spite of the evidence reported by Yan, et al. and many others (see Chapter 11 in The Frozen Climate Views of the IPCC), AR6 has concluded that anthropogenic global warming is causing and has caused increases in extreme weather. This is another example of bias.

On longer time scales storminess also increases during historically cold periods. It reached a peak in Europe during the very cold Little Ice Age period, labeled “LIA” in figure 1.

Figure 1. Periods of storminess over the past 6,500 years in Europe. After (Vinós, 2022).

Figure 1 shows periods of storminess identified from sedimentological features in Europe from 18 studies. The most concentrated periods of extreme storminess occurred during the cold Little Ice Age,[8] which the IPCC calls the “pre-industrial” period.[9] It is logical that extreme weather and storminess increases in cold times, and that is what is observed. Tropical temperatures do not change much, global temperature change happens mostly in the higher latitudes. As higher latitude temperatures grow colder, the temperature gradient steepens, powering stronger winds, and strong winds are the hallmark of extreme weather.

IPCC AR6 WG1 Model Bias, Summary

The weakest IPCC assumption is that the Sun has remained constant in its effect on climate since 1750. As we showed in part 3, there is a lot of evidence that the Sun has varied in that timeframe and contributed to global warming.[10] The IPCC’s implicit assumption, as shown in figure 2 in part 3, that a change in solar forcing is equivalent to the same change (in W/m2) in greenhouse forcing is simply not true, basic physics tells us that. Solar forcing has a longer residence time, at least in the oceans, and causes more warming Watt-for-Watt due to the higher frequencies in sunlight. The higher frequencies penetrate the ocean surface and the lower frequency greenhouse gas emissions do not.

The CMIP6 models do try and account for ocean heat content, ocean heat buffering, and inertia, but as far as I can see they do not account for the difference in the climatic effects of sunlight and greenhouse gas emissions. If they do, it is buried deeply in AR6 WG1.[11] While they admit there is a difference in ocean uptake in the model documentation,[12] they seem to just ignore the fact. They treat them the same or nearly the same and they are not.

This is important because, according to AR6, increasing ocean heat content is 91% of the total global energy inventory change from 1971-2006 to 2006-2018, two periods of fast warming.[13] Coincidentally or not, this is also the last half of the Modern Solar Maximum,[14] when the oceans would be expected to be expelling the excess heat they accumulated during the first half of the maximum. Oceans expel heat primarily with El Ninos, tropical cyclones (hurricanes), monsoons, and mid-latitude storms. How long it takes for the world’s atmosphere to begin cooling after a solar maximum is unknown, maybe we will find out soon.

While we have many measurements of heat content to 2000 meters ocean depth now, prior to 2004 these measurements are quite rare, thus the specific model used to compute heat content in the ocean, prior to 2000, is very important. It is unfortunate that our measurements of deep ocean heat content improved so dramatically just as the Modern Solar Maximum was ending, we need this data as the heat content was accumulating as well. Satellite measurements of outgoing radiation from Earth’s surface are very sensitive to surface temperature, but not necessarily to below the ocean surface.

There are three processes that work together to keep Earth’s climate stable, the greenhouse effect, clouds, and meridional heat transport. These factors also help determine the residence time of energy received from the Sun. The IPCC has focused exclusively only on the first by developing their CO2 “control knob” climate hypothesis.[15]

The effect of clouds and their variability on climate change is still largely unknown and remains the largest source of uncertainty in calculating the effect of greenhouse gases on our climate.[16] The IPCC cannot tell if the net effect of increasing cloud cover warms or cools the Earth. Because cloud formation and destruction cannot be modeled,[17] their effect is ignored or simply “parameterized”[18] by the IPCC.

With respect to meridional heat transport, energy is only exchanged between the climate system and outer space through the top of the atmosphere, this results in global meridional heat transport necessarily having a net zero value when integrated instantaneously over the entire global climate system. Moving energy from one region to another does not alter the total amount of energy within the system, although it can, and does change the energy residence time. By ignoring the impact of changing convection patterns on energy residence time, the IPCC has concluded that meridional heat transport cannot constitute a significant cause for climate change, their most fundamental mistake.[19]

From both a thermodynamic and dynamic point of view, the defining feature of Earth’s climate system is the transport of energy. Energy transport is the cause of all weather, especially all extreme weather. Most of the solar energy that is not reflected by the planet is stored in the oceans, where 99.9%[20] of the climate system “heat” or thermal energy resides. However, the oceans do not transport energy very well. Differences in water temperature tend to cause vertical movements through altered buoyancy, not lateral movements, so the oceans are temperature stratified. Most of the energy in the climate system is transported by the atmosphere, and what energy is transported laterally by surface ocean currents is driven by winds.[21] The flow of non-solar energy at the atmospheric-ocean boundary (including across sea-ice) is almost always, almost everywhere, from the ocean to the atmosphere.[22]

These first five parts point out the most serious systematic biases in the IPCC AR1 report, Climate Change 2021, The Physical Science Basis. The bias is introduced by ignoring the valid peer-reviewed articles cited in this series of posts and numerous other similar articles, thus it is mostly “reporting” or “confirmation” bias. The evidence that solar variability affects Earth’s climate is huge and well established in the literature as reviewed by Douglas Hoyt and Kenneth Schatten in their excellent book The Role of the Sun in Climate Change,[23] and by Joanna Haigh in her report, Solar Influences on Climate.[24]

Yet, these monumental works, as well as the many papers in the bibliography, are ignored by the IPCC, or brushed off as unimportant. In addition, many of the hundreds of papers cited in Connolly, et al.,[25] and in Nicola Scafetta and Fritz Vahrenholt’s[26] chapter in The Frozen Climate Views of the IPCC are ignored in AR6.

It is well documented that the IPCC/CMIP6 climate models run much too hot relative to observations.[27] They overestimate tropical mid-tropospheric temperature,[28] sea surface temperature,[29] and global warming in general.[30] This bias is acknowledged in AR6 itself, where they write:

“Hence, we assess with medium confidence that CMIP5 and CMIP6 models continue to overestimate observed warming in the upper tropical troposphere over the 1979–2014 period by at least 0.1°C per decade, in part because of an overestimate of the tropical SST trend pattern over this period.“[31]AR6 WGI, page 444

The IPCC/CMIP6 climate models not only predict more warming than we observe, they also “drift” in their calculations with time. In most models the drift is toward more warming, adding to the natural tendency of models to run hot. In addition, the energy and mass budgets do not close properly in individual model runs, violating the physical laws of conservation of mass and energy.[32] In some respects the CMIP6 models do better than CMIP5, but in others they do worse, suggesting that the models are not getting any better with time and may have reached the end of their useful lives.

Model unforced drift contaminates the forced (that is changing CO2 or volcanism) trends. In some models, the drift is a significant proportion of the warming computed. It results from the parameterization of quantities that cannot be modeled, like cloud cover, and from an imbalance between the prescribed initial model state from observations and the physics programmed into the model.

Another culprit can be the “coupling shock” when the various model components, like the ocean and atmospheric components are merged, sometimes called “boundary shock.” Suffice it to say, the fact that mass and energy are not conserved in these models, and they drift, is a real and significant problem. The model drift is simply the model trying to reach a stable state after the start up (or spin-up) shocks occur. Equilibrium, after spin-up, is not reached for over a thousand years, so letting the model run that long just to get rid of the drift is not practical. Usually when the models are corrected for drift, their mass/energy conservation problems are reduced.[33] However, the mismatch between the model physics and the initial model state, which is from observations, is worrying all by itself.

From this modeler’s point of view, the conceptual model or hypothesis that the CMIP6, and its predecessors, were designed to investigate has been shown to be incorrect. Only by introducing increasing amounts of bias can this observation be covered up. But the more bias introduced into the models and in the interpretation of the output, the more obvious it becomes. It is time to quit and develop a new hypothesis and conceptual model to investigate, the “CO2 control knob” idea is dead.[34]

In part 6, I look at model bias in WGII.

Download the bibliography here.


  1. (Pierrehumbert, 2011) and (Wijngaarden & Happer, 2020) ↑

  2. (Barry, Craig, & Thuburn, 2002) ↑

  3. For a discussion of how meridional transport is modulated see: (Vinós, Climate of the Past, Present and Future, A Scientific Debate, 2nd Edition, 2022, pp. 157-174) ↑

  4. (Barry, Craig, & Thuburn, 2002) and (Huang, 2004) ↑

  5. (Barry, Craig, & Thuburn, 2002) ↑

  6. (Yan, et al., 2001) ↑

  7. (Yan, et al., 2001) ↑

  8. (Costas, Naughton, Goble, & Renssen, 2016) and (Sorrel, et al., 2012) ↑

  9. The IPCC does not use the “Little Ice Age” label, they consider it to be poorly defined (IPCC, 2021, p. 295). They define pre-industrial as either before 1850 or before 1750 depending upon context. When referring to the temperature record, 1850-1900 is the reference temperature period for the “pre-industrial.” When referring to human emissions of greenhouse gases, the boundary is 1750. See footnote 13, page 43 of WGI or page 2244 in the AR6 glossary. ↑

  10. (Crok & May, 2023, Ch 6, by Scafetta and Vahrenholt) and (Vinós & May, 2022b) ↑

  11. (Griffies, Harrison, Pacanowski, & Rosati, 2004). The radiation-ocean surface model is parameterized but seems to be limited to the top 10-30 meters of the ocean. The model documentation is unclear. Important point: the average thickness of the ocean mixed layer is more than 50 meters, see here, figure 5 for details. ↑

  12. (Griffies, Harrison, Pacanowski, & Rosati, 2004, p. 137) ↑

  13. (IPCC, 2021, p. 91) ↑

  14. (Usoskin, 2017) ↑

  15. (Lacis, Hansen, Russell, Oinas, & Jonas, 2013), (Lacis, Schmidt, Rind, & Ruedy, 2010), and (IPCC, 2021, p. 179) ↑

  16. (IPCC, 2021, pp. 978-979) ↑

  17. (Nakamura, 2018), or see (Taveira, 2019) and (IPCC, 2021, pp. 978-979) ↑

  18. Parametrization in climate modeling is replacing a process in a model with a simplified set of model parameters. See: (Lu, Liu, Niu, Krueger, & Wagner, 2013) ↑

  19. (May, Meridional Transport, the most fundamental climate variable, 2022g), (Vinós & May, The Sun-Climate Effect: The Winter Gatekeeper Hypothesis (III). Meridional transport, the most fundamental climate variable, 2022c), and (Vinós, Climate of the Past, Present and Future, A Scientific Debate, 2nd Edition, 2022, p. 157) ↑

  20. (May, Ocean Temperature Update, 2020e) ↑

  21. (Huang, 2004) and (Yang, Li, & Wang, 2015) ↑

  22. (Yu & Weller, 2007), (Schmitt, 2018)), and (Vinós & May, The Sun-Climate Effect: The Winter Gatekeeper Hypothesis (VI). Meridional transport is the main climate change driver, 2022e) ↑

  23. (Hoyt & Schatten, 1997) ↑

  24. (Haigh, 2011) ↑

  25. (Connolly et al., 2021) ↑

  26. (Crok & May, 2023, Ch. 6) ↑

  27. (McKitrick & Christy, 2020), (McKitrick R. , Biases in climate fingerprinting methods, 2022), (McKitrick & Christy, A Test of the Tropical 200- to 300-hPa Warming Rate in Climate Models, Earth and Space Science, 2018), (McKitrick R. , Checking for model consistency in optimal ingerprinting: a comment, 2021), (Scafetta N. , CMIP6 GCM ensemble members versus global surface temperatures, 2022b), (IPCC, 2021, pp. 443-444), (Scafetta N. , Impacts and risks of “realistic” global warming projections for the 21st century, 2024), (Mauritsen & Stevens, 2015), and (Spencer, 2024). ↑

  28. (McKitrick & Christy, A Test of the Tropical 200- to 300-hPa Warming Rate in Climate Models, Earth and Space Science, 2018), (IPCC, 2021, p. 444) ↑

  29. (IPCC, 2021, p. 443) ↑

  30. (McKitrick & Christy, 2020) ↑