Continued Discussion with Andreas Sterl On The Climate Science Weblog Of July 17 2008

Andraes Sterl and I continued our constructive discussions of the weblog of August 15 regarding his published paper that was originally weblogged on July 17 2008, and our-mail exchange (edited for clarity) is presented below.

Comment By Andreas Sterl 14 Aug 2008

How can you validate a forecast? Only by waiting to the date for which the forecast is valid and to measure. Not very practical for 2100. In this sense every forecast/projection/scenario/speculation/call-it-how-you want for more than a few months in advance is non-scientific, as it cannot be validated. Should we therefore stop thinking about what the future might bring? No, we should not, but we should be aware of the limitations, and we should check the results for plausibility, obeying of physical (and other) laws, mechanisms behind, etc.

Reply by Roger A. Pielke Sr. 14 August 2008

The scientific method requires that real data be used to test models (and to seek to refute their forecasts). We can not do that with 2100 as you correctly recognize. Thus, why should policymakers believe the forecasts? What is the basis for claiming that they are skillful on the global and regional scales? Even Kevin Trenberth, one of the IPCC authors, has recognized this limitation [see]

We do not need the models to tell us that CO2 is increasing due to human activity since this is observed. That is where this part of the process should focus; on the unknown consequences of this added CO2, not by providing definitive (but unverifiable) forecasts to policymakers.

For the more general issue of dealing with the future environment, we have advocated a vulnerability bottom-up resource-based perspective (e.g. see my invited IGBP essay and also Chapter E in the book

Kabat, P., Claussen, M., Dirmeyer, P.A., J.H.C. Gash, L. Bravo de Guenni, M. Meybeck, R.A. Pielke Sr., C.J. Vorosmarty, R.W.A. Hutjes, and S. Lutkemeier, Editors, 2004: Vegetation, water, humans and the climate: A new perspective on an interactive system. Springer, Berlin, Global Change – The IGBP Series, 566 pp.

This vulnerability perspective fits quite well with Pavel’s “climate proofing” except the universe of what the future can hold needs to be broader than what is predicted by the multi-decadal global climate models.

We are risking the loss of credibility if the models continue to deviate from the observed trends (such as the absence of recent global warming; e.g. see Figure 7 in}. I agree that humans are altering the climate system, but as you state below, “there are (a lot of) processes that are not included in the models”. The 2005 NRC report (basically ignored in the 2007 IPCC report) indicates that missing human climate forcings are quite important (

Thus, my conclusion is that papers which present predictions are not valid science contributions, unless they are validated by predictions. That the IPCC models are failing so far in this test is ably discussed in the weblog

Comment By Andreas Sterl 14 Aug 2008

Yes, there are (a lot of) processes that are not included in the models. Our current knowledge (belief?) is that those processes are of less importance than the increase of greenhouse gases (this is opposite to your view, I know). We have done our work under certain assumptions. If anybody thinks that these assumptions are insufficient, he (she) should repeat the work with his (her) assumption and report on the results. this will benefit the scientific community as it sheds light on the relative importance on assumptions.

Reply by Roger A. Pielke Sr. 14 August 2008

In my recent testimony [], we show that

“In Matsui and PielkeSr. (2006), it was foundfrom observations of the spatial distribution of aerosols in the atmosphere in the lower latitudes, that the aerosol effect on atmospheric circulations, as a result of their alteration in the heating of regions of the atmosphere, is 60 times greater than due to the heating effect of the human addition of well-mixed greenhouse gases.”

This analysis clearly shows that the other climate forcings are first order, and thus, by not being properly represented in the models, guarantees that the regional prediction forecasts decades into the future will not be skillful.

These other human climate forcings include land cover/land use (e.g see and, and a diversity of aerosol effects such as nitrogen deposition, the thermodynamic effect of aerosols, etc) as I briefly discuss at

Comment By Andreas Sterl 15 Aug 2008


It seems that we are not so far apart after all. I can agree with most of
 your arguments – with two important exceptions:

1. The question whether increasing greenhouse gases are the first order driver of the climate, or whether other human-caused effects are.  This is a purely scientific question – it can be answered by more research. For instance, in the case of the example you gave, aerosols, one could make make sensitivity runs altering several aspects of plausible (i.e., in accordance with present knowledge) aerosols physics and/or aerosol concentrations for several concentrations of CO_2. This would give the relative importance of the two processes.

 Reply by Roger A. Pielke Sr. 15 August 2008

 We, and others, have already done this. The first issue, of course, is what are the societally and environmentally important metrics to use to define what are “first order drivers of the climate”. The IPCC uses the global average radiative forcing, as represented by the globalaverage surface temperature trend. However, the 2005 NRC report concludes that this metric is incomplete. What we have concluded in our research is that the most important climate metrics involve those that alter circulation patterns, such as ENSO, the AO, etc. These are regional scale changes in circulations which have important consequences to the weather experienced throughout the globe. The globalaverage temperature trend is grossly inadequate as a metric for these changes in circulation.

For example, for increased summer heat in the Netherlands, it is not a global average increase in temperature that would dominate, but alterations in the intensity and position of the middle and upper tropospheric large scale troughs and ridges. We showed this in our examination of the 2003 European heat wave; see

Chase, T.N., K. Wolter, R.A. Pielke Sr., and Ichtiaque Rasool, 2006: Was the 2003 European summer heat wave unusual in a global context? Geophys. Res. Lett., 33, L23709, doi:10.1029/2006GL027470

ConnolleyW.M. 2008: Comment on “Was the 2003 European summer heat wave unusual in a global context?” by Thomas N. Chase et al. Geophys. Res. Lett., 35, L02703, doi:10.1029/2007GL031171
<a href=””

Chase, T.N., K. Wolter, R.A. Pielke Sr., and Ichtiaque Rasool, 2008: Reply to comment by W.M. Connolley on ..Was the 2003 European summer heat wave unusual in a global context?..Geophys. Res. Lett., 35, L02704, doi:10.1029/2007GL031574.

The IPCC models have not demonstrated skill in predicting these regional, circulation change events in the last few decades. Thus, no confidence should be placed on their ability to predict these events decades into the future.

Moreover, even if the global average temperature did not increase, humans are still altering the circulation patterns through the diverse range of climate forcings that I overview in

Pielke Sr., Roger A., 2008: A Broader View of the Role of Humans in the Climate System is Required In the Assessment of Costs and Benefits of Effective Climate Policy. Written Testimony for the Subcommittee on Energy and Air Quality of the Committee on Energy and Commerce Hearing .Climate Change: Costs of Inaction. . Honorable Rick Boucher, Chairman. June 26, 2008, Washington, DC., 52 pp. [].

Comment By Andreas Sterl 15 Aug 2008

2. The second exception is more fundamental and either philosophic or semantic. It is “What is a forecast?”, or “How should the result of a climate model run be presented?” You argue that the (scientific?) process should focus on “the unknown consequences of this added CO2”. I completely agree with this statement, and this is what our paper is about: It explores the possible consequences of increasing greenhouse gas concentrations on extreme maximum temperatures. The result is that they     

1. increase faster than average temperatures,    

2. may reach dangerous levels in highly populated areas, and    

3. the value for Phoenix (to repeat myself) is 49.1 deg C.    

Which of these statements do you consider “unscientific”? And why?

Reply by Roger A. Pielke Sr. 15 August 2008

The presentation of model results without validation is not science. What the IPCC multi-decadal global climate models actually accomplish are process studies. This improves our understanding of whether particular climate forcings are important, but they should not be interpreted as predictions (projections) since, not only have they not shown forecast skill, but they do not include all of the known forcings and feedbacks.

 I discussed this topic a few years ago in my Climatic Change article

Pielke Sr., R.A., 2002: Overlooked issues in the U.S. National Climate and IPCC assessments. Climatic Change, 52, 1-11.

in which Mike MacCraken had an informative response

 MacCracken, M., 2002: Do the uncertainty ranges in the IPCC and U.S. National Assessments account adequately for possibly overlooked climatic influences. Climatic Change, 52, 13-23.

In terms of your items #1 to # 3 [with respect to #1]  This has not been confirmed with comparisons to data. This is a hypothesis [which, of course, is what a model is].

Dangerous levels of heat are already reached in many cities around the world with current and past climate. The first task should be to reduce the threat this poses to the urban population irrespective of how climate changes in the future. Our research and that of others, show that it is the urban alteration of the landscape (e.g. reduction in trees and other vegetation; waste heat from industry and residential areas; greater storage of heat in asphalt, concrete, etc) which is the much more important reason for excessive heat in urban areas. This excess heat is documented, for example, in

The value of temperature that you present for Phoenix is at what location and height? Phoenix already reaches this temperature at 2m above the surface. This matters as the urban landscape has spatially varying values that determine the threat that is posed to people. An important issue is also that people have learned to live and thrive in that environment. They have minimized their vulnerability to the excessive heat in Phoenix.

Comment By Andreas Sterl 15 Aug 2008

Furthermore, you ask why policy makers should believe in unverifiable results. The answer is because they are based on sound principles. they should not believe the value of 49.1 for Phoenix, but they should believe that Phoenix may experience very hot (much hotter than today) temperatures, and that these temperatures will cause problems.

Reply by Roger A. Pielke Sr. 15 August 2008

The models are engineering code. Only the pressure gradient force, advection and gravity are from basic physics. All other representations of the physics are parameterizations, which are almost always one-dimensional column models. The boundary layer parameterizations and deep cumulus parameterizations, for example, are generally developed using empirical tuning values from a small set of field data using case study days.

I discuss these major limitations in the models in my book

Pielke, R.A., Sr., 2002: Mesoscale meteorological modeling. 2nd Edition, Academic Press, San Diego, CA, 676 pp.

and in my modeling classes where I have the students dissect the parameterizations (e.g. most recently in

The models are not fundamental physics.

Comment By Andreas Sterl 15 Aug 2008

Finally, you say “that papers which present predictions are not valid  science contributions, unless they are validated by predictions”. I do not understand that. Do you mean that my number of 49.1 for Phoenix would become a valid scientific contribution if one (or more) other researchers, using a different model, came up with the same number? And what if he found 49.2?

Reply by Roger A. Pielke Sr. 15 August 2008

Model to model comparisons are not validation. The only validation is from observations. If the IPCC models were run to predict the seasonal weather for the next several years (as initial value problems), and they could show skill (statistically) in forecasting the seasonal temperature and precipitation anomalies, then they would have passed a necessary condition to seek to predict for longer time periods into the future. However, they have not yet even passed this test.

Comments Off on Continued Discussion with Andreas Sterl On The Climate Science Weblog Of July 17 2008

Filed under Climate Science Misconceptions, Guest Weblogs, Q & A on Climate Science

Comments are closed.