Do models accurately predict climate change?

Summary: Climate models are important for several reasons. Large flows of tax dollars go to their construction and operation. Their predictions dominate the public policy debate about climate change (to the exclusion of other tools, such as predictability studies). In this post eminent climate scientist Roger Pielke Sr. explains that long-term model forecasts have shown little skill at forecasting. Post your questions in the comments; he’ll answer as time permits.  {1st of 2 posts today).

“I offer a toast to the future, the undiscovered country.”
— Klingon Chancellor Gorkon in “Star Trek VI: The Undiscovered Country“.

The undiscovered country from whose bourn no traveler returns, puzzles the will and makes us rather bear those ills we have than fly to others that we know not of? …And thus the native hue of resolution is sicklied o’er with the pale cast of thought, and enterprises of great pith and moment with this regard their currents turn awry, and lose the name of action.  {Hamlet}

Temperature change in NOAA's GFDL CM2.1 model.
Projected change in annual mean surface air temperature from the late 20th century (1971-2000 average) to the middle 21st century (2051-2060 average). This is based on a “middle of the road” estimate of future emissions ( IPCC SRES A1B). These results are from the GFDL CM2.1 model, but are consistent with a broad consensus of modeling results. From NOAA.

How accurately do the global climate models simulate the real climate?

Guest post by Roger A. Pielke Sr.

The climate models are useful as sensitivity experiments but using them to claim an ability to skillfully project climate, even on the global scale, in the coming decades has not been shown.

The new seminal Stephens et al paper provides a clear documentation of the level of model skill: “The albedo of Earth” in Reviews of Geophysics, March 2015. There is also a power point talk on this: “Is the Earth’s climatesystem constrained?” Among their conclusions is that …

“Climate models fail to reproduce the observed annual cycle in all components of the albedo with any realism, although they broadly capture the correct proportions of surface and atmospheric contributions to the TOA {top of atmosphere} albedo. A high model bias of albedo has also persisted since the time of CMIP3, mostly during the boreal summer season. Perhaps more importantly, models fail to produce the same degree of interannual constraint on the albedo variability nor do they reproduce the same degree of hemispheric symmetry.”

The technical term albedo “is the fraction of solar energy (shortwave radiation) reflected from the Earth back into space. It is a measure of the reflectivity of the earth’s surface. Ice, especially with snow on top of it, has a high albedo: most sunlight hitting the surface bounces back towards space” (From the Earth & Space Research website). CMIP3 is phase 3 of the Coupled Model Intercomparison Project (CMIP) of the World Climate Research Programme (WCRP). They collect the output of global climate models (i.e., coupled atmosphere-ocean general circulation models).

Stephens et al further bolsters the conclusions we summarized in the preface to Climate Vulnerability: Understanding and Addressing Threats to Essential Resources (5 volumes, 2013)…

“…for decadal and multidecadal predictions little, if any, predictive skill has been shown in hindcast climate model predictions of changes in regional weather statistics beyond what is available to the impacts community via the historical, recent paleorecord and a worst case sequence of weather events.”

This conclusion is similar to that in “Regional climate downscaling – what’s the point?” by R. A. Pielke Sr. and R.L. Wilby in Eos Forum, 31 January 2012.

IPCC

What does the IPCC say?

In AR5, there are two chapters about models’ skill. Chapter 11 discusses “decadal predictability”:  “Near-term Climate Change: Projections and Predictability“. Chapter 12 examines “Long-term Climate Change: Projections, Commitments and Irreversibility“.

I view the distinction as key, as the “projection” chapter assumes that just the forcings (CO2, etc.) dominate on time periods longer than a decade. They base this on comparisons of model runs with and without those forcings; not with tests on their skill and replicating observed real-world multi-decadal global and regional climate patterns.

Question World

Post your questions and comments

Please post your questions in the comments. Comments with evidence that refutes this conclusion are invited. Please keep them brief (i.e., no essays), and respond with a direct quote where possible to keep the conversation clear. This is a technical thread; please avoid comments about life, politics, and personalities.

See the follow-up post: Thomas Kuhn & Twitter tell us what we need to know about climate science.

Roger Pielke Sr.

About the author

Roger Pielke Sr. is currently a Senior Research Scientist in Cooperative Institute for Research in Environmental Science. He is also an Emeritus Professor of Atmospheric Science at Colorado State University, and now serves there as a Senior Research Associate.

His list of accomplishments, honors, and publications is too long to list here. See his bio for details. See his website for commentary on climate science issues. Also see his presentations, especially these…

Editor’s suggestions: articles for laypeople about this vital subject

These provide an introduction to the subject, and a deeper review of this frontier in climate science.

Judith Curry (Prof Atmospheric Science, GA Inst Tech) reviews the literature about the uses and limitation of climate models…

  1. What can we learn from climate models?
  2. Philosophical reflections on climate model projections.
  3. Spinning the climate model – observation comparison — Part I.
  4. Spinning the climate model – observation comparison: Part II.

Truth Will Make You Free

For More Information

To learn more about the state of climate change see The Rightful Place of Science: Disasters and Climate Change by Roger Pielke Jr. (Prof of Environmental Studies at U of CO-Boulder, and Director of their Center for Science and Technology Policy Research).

If you liked this post, like us on Facebook and follow us on Twitter. For more information see The keys to understanding climate change and My posts about climate change. Also, see these posts about computer models, especially these..

  1. About models, increasingly often the lens through which we see the world.
  2. Will a return of rising temperatures validate the IPCC’s climate models?
  3. We must rely on forecasts by computer models. Are they reliable?
  4. A frontier of climate science: the model-temperature divergence.

A sample of the large body of research about this topic

  1. The Robustness of the Climate Modelling Paradigm“, A. M. R. Bakker, Jan 2015.
  2. Overestimated global warming over the past 20 years” by John C. Fyfe et al, Nature Climate Change, Sept 2013.
  3. Preface to Climate Vulnerability, Understanding and Addressing Threats to Essential Resources, R. A. Pielke Sr., Editor in Chief, (2013).
  4. Comments on “The North American Regional Climate Change Assessment Program: Overview of Phase I Results” by R. A. Pielke Sr. in Bulletin of the. American Meteorological Society, July 2013.
  5. Regional climate downscaling – what’s the point?” by R. A. Pielke Sr. & R.L. Wilby in Eos Forum, 31 Jan 2012.
  6. Dealing with complexity and extreme events using a bottom-up, resource-based vulnerability perspective. Extreme Events and Natural Hazards: The Complexity Perspective“, Pielke Sr. et al, in the Geophysical Monograph Series 196 of the American Geophysical Union (2012).
  7. Should we assess climate model predictions in light of severe tests?” by Joel Katzav in Eros, 7 June 2011.
  8. Should we believe model predictions of future climate change?” by Reto Knutti in Philosophical Transactions A, Dec 2008.
  9. Contrast between predictive and vulnerability approaches” by R. A. Pielke Sr. & T.J. Stohlgren in Chapter E.3 of Vegetation, Water, Humans and the Climate: A New Perspective on an Interactive Systeme, P. Kabat et al., Editors, (2004).
  10. Conclusions” by R. A. Pielke Sr. & L. Bravo de Guenni in Chapter E.7 in Vegetation, Water, Humans and the Climate: A New Perspective on an Interactive System, P. Kabat et al., Editors, (2004).

21 thoughts on “Do models accurately predict climate change?

  1. If I may ask the first question — I believe these discussions should start by benchmarking views to those in the IPCC’s reports. Their most recent, Assessment Report #5, gave a detailed analysis in Working Group I’s Chapter 9: “Evaluation of Climate Models”. The Executive Summary said:

    Most simulations of the historical period do not reproduce the observed reduction in global mean surface warming trend over the last 10 to 15 years. There is medium confidence that the trend difference between models and observations during 1998–2012 is to a substantial degree caused by internal variability, with possible contributions from forcing error and some models overestimating the response to increasing greenhouse gas (GHG) forcing.

    Most, though not all, models overestimate the observed warming trend in the tropical troposphere over the last 30 years, and tend to underestimate the long-term lower stratospheric cooling trend.

    How does this compare with your views? Also, the news media seldom mention this finding. Is it significant — and if so, why?

    Note for readers — According to the IPCC…

    climate variability refers to variations in the mean state and other statistics (such as standard deviations, the occurrence of extremes, etc.) of the climate on all spatial and temporal scales beyond that of individual weather events. Variability may be due to natural internal processes within the climate system (internal variability), or to variations in natural or anthropogenic external forcing (external variability).”

    1. The IPCC distinguishes between decadal time periods, where variations from internal dynamics are very important, to long multi-decadal time periods where they assume that CO2 and the other greenhouse gases input by human activity dominate. They use models to make this claim.

      However, the real world observations indicate there are longer term natural variations as well; e.g. see our paper: Nonlinearities, feedbacks and critical thresholds within the Earth’s climate system. Climatic Change, 2004.

      We concluded that “The Earth’s climate system is highly nonlinear: inputs and outputs are not proportional, change is often episodic and abrupt, rather than slow and gradual, and multiple equilibria are the norm.”

      The IPCC has not completely accepted this reality, unfortunately.

      See also my paper: Climate prediction as an initial value problem. Bulletin American Meteorological Society, 1998.

  2. Fun but sad to read the discussion on Twitter about this post. Some very smart people are debating not the post, but the 140 characters in the Tweet about it. They’re pretending to expect a degree of precision in the Tweet that Einstein couldn’t achieve (and probably would consider quite mad to attempt).

    I say they’re “pretending” because titles of press releases and promotional tweets are usually written by PR people — not the authors of papers. They’re about as accurate as should be expected for headlines by PR people seeking to get attention for papers. If these people really had such standards, they’d be objecting to large numbers of Tweets and press release tittles every day.

    But of course they wouldn’t do that. People would laugh at them. It’s the equivalent of refusing to read the book because you disagree with its contents based on the title. Mothers teach children not to do that.

    The larger strategy seen is attempt to hide from public view disagreements among climate scientists. Whether its exaggeration of the consensus (e.g,, the “97% of scientists believe”) or smearing as “deniers” scientists whose peer-reviewed research disagrees with activists — it’s a failing strategy. Irrespective of the public climate debate, it deserves to fail — otherwise public policy debates revolving around science would become even more polarized and dysfunctional than they are today.

  3. Dear Dr. Pielke,

    Can Contrails that form into clouds {not chemtrails} cause regional climate change?

    Sincerely

    Mike

    1. Mike – They do play a role in local and regional climate. See, for example, the paper “Clouds, contrails and climate“.

      They behave similar to cirrus clouds that reflect a portion of the sunlight back into Space. This can reduce maximum temperatures slightly. The contrails do not have much of an effect when they are narrow, but often they spread horizontally, This is when they have a larger effect, particularly in well traveled air corridors.

      Roger Sr

  4. What in your opinion is most needed currently. Better data? Better models (possibly incorporating new feedback mechanisms)? Betteróways of deciding which models are better (Some method of honest reproduction?)

    1. Social Bill,

      Emergency medical teams have more than one person because often many things must be done simultaneously for the patient to survive. Ditto for serious public policy challenges. Fortunately just as a person can walk and chew gun, a competently run nation can pursue multiple lines of public policy action to address climate change.

      See this page for my core beliefs about climate change. For the past five years my recommendations have been the same:

      1. More funding for climate sciences. Many key aspects (e.g., global temperature data collection and analysis) are grossly underfunded.
      2. Wider involvement of relevant experts in this debate. For example, geologists, statisticians and software engineers have been largely excluded — although their fields of knowledge are deeply involved.
      3. Run government-funded climate research with tighter standards (e.g., posting of data and methods, review by unaffiliated experts), as we do for biomedical research.
      4. We should begin a well-funded conversion to non-carbon-based energy sources, for completion by the second half of the 21st century — justified by both environmental and economic reasons (see these posts for details).
      5. Begin more aggressive efforts to prepare for extreme climate. We’re not prepared for repeat of past extreme weather (e.g., a real hurricane hitting NYC), let alone predictable climate change (e.g., sea levels climbing, as they have for thousands of years).
    2. socialbill – What is most needed are robust statistical assessments of the skill of the models at predicting climate statistics and changes in climate statistics on local, regional and global scales over multi-decadal time periods. These tests can be done in hindcast runs where we know reasonably well the human radiative forcings from added CO2 and other greenhouse gases.

      That they have inadequately done this in an objective, quantitative way is a serious omission of a critical part of the scientific method.

      Roger Sr.

  5. Reblogged this on Centinel2012 and commented:
    The current models assume a stay state prior to man and his carbon based fuels. This assumption that current global temperature are going up “only” because of CO2 are false. Going back almost 3000 years there is a clear cycle of war and cold period of around 1,000 year and since the last cold period ended between 1600 and 1650 we are in a long cycle of increasing temperatures for at least another 100 years. The current pause is caused by a shorter cycle related to El Nino La Nina and the Atlantic Multidecadal Oscillation (AMO) which is in a 30 year cold cycle right now. When that ends temperatures will go back up.

    1. centinel,

      Another assumption, and more clearly questionable, is that we will burn off much of the Earth’s fossil fuel reserves during the next 85 years — much of which will be coal — with no carbon capture. That strikes me as possible but unlikely.

  6. Important background information about use of climate models

    From “SCMIP5 Global Climate Change Viewer CMIP5 Global Climate Change Viewer (GCCV)” at climate.gov —

    Important Notice for Using Climate Projections

    Climate projections can be useful for making decisions about the future, but the limitations of climate models make it easy to misinterpret or misuse their results. Be aware that:

    • Climate projections are not predictions, as they are based on assumptions about future human emissions of greenhouse gases and other policy choices.
    • Climate projections do not attempt to predict the timing of meteorological events like storms, droughts, or El Niños. The location and timing of future extreme weather events cannot be deduced from climate model projections.
    • Projections vary from model to model: the best projection dataset for one location and purpose may not be the best for other situations. Considering a range of projections may help you gain a more complete picture of potential future risks.
    • While you can obtain some sense of uncertainty in the projections by comparing results from multiple models, this approach may still systematically underestimate uncertainty.
    • Some climate projection datasets contain daily-average projections; other datasets contain only monthly averages.
    • The increased spatial resolution of statistically downscaled projections available for temperature and precipitation may not be available for all parameters. In addition, increased resolution does not necessarily equate to greater fidelity or reliability.
  7. On what is said from “SCMIP5 Global Climate Change Viewer CMIP5 Global Climate Change Viewer (GCCV)” at climate.gov — Important Notice for Using Climate Projections, here are my responses.

    “Climate projections are not predictions, as they are based on assumptions about future human emissions of greenhouse gases and other policy choices.”

    In hindcast runs they are predictions since we reasonably well know the human emissions of greenhouse gases and other policy choices. The future projections still are predictions (“what if” predictions) and their trajectories can be tracked.

    “Climate projections do not attempt to predict the timing of meteorological events like storms, droughts, or El Niños. The location and timing of future extreme weather events cannot be deduced from climate model projections.”

    This correct. However, the projections do predict changes in the statistics of these weather events. We can examine how well the models do at predicting these changes using hindcast runs where we find they do not yet have that skill. Thus why should the impacts community accept their future projections as robust?

    “Projections vary from model to model: the best projection dataset for one location and purpose may not be the best for other situations. Considering a range of projections may help you gain a more complete picture of potential future risks.”

    Comparison between models without a real world comparison is not robust science

    “While you can obtain some sense of uncertainty in the projections by comparing results from multiple models, this approach may still systematically underestimate uncertainty.”

    Very true. However, they do not accept what this means. The uncertainty must be larger than they claim since they have no regional predictive skill on multi-decadal time scales when run in hindcasts.

    “Some climate projection datasets contain daily-average projections; other datasets contain only monthly averages.”

    I am not sure why they even included this here.

    “The increased spatial resolution of statistically downscaled projections available for temperature and precipitation may not be available for all parameters. In addition, increased resolution does not necessarily equate to greater fidelity or reliability.”

    They wrote “increased resolution does not necessarily equate to greater fidelity or reliability”. As we show in our paper — Pielke Sr., R.A., and R.L. Wilby, 2012: Regional climate downscaling – what’s the point? Eos Forum — they are slaves of the parent global model. They can, at most, be used for model sensitivity studies. Downscaling of multi-decadal climate models presents only a false illusion of more skill.

    Clearly, the IPCC has not properly vetted the model skill.

Leave a Reply