Do we need better predictions to adapt to a changing climate?
April 6th, 2009Posted by: Roger Pielke, Jr.
That is the title of a new EOS Forum piece that I am a co-author on that appeared in the 31 March issue. Here is an excerpt:
Many scientists have called for a substantial new investment in climate modeling to increase the accuracy, precision, and reliability of climate predictions. Such investments are often justified by asserting that failure to improve predictions will prevent society from adapting successfully to changing climate. This Forum questions these claims, suggests limits to predictability, and argues that society can (and indeed must) make effective adaptation decisions in the absence of accurate and precise climate predictions.
The piece can be found at the following location:
Dessai, S., M. Hulme, R. Lempert, and R. Pielke, Jr. 2009. Do We Need Better Predictions to Adapt to a Changing Climate? Eos, Vol 90, No. 13, pp. 111-112. (PDF)
Comments welcomed.
April 6th, 2009 at 6:33 am
According to the IPCC the models don’t make predictions, projections or forecasts but create scenarios. The difference is calculating confidence intervals. The climate modelers don’t like calculating confidence intervals since they reveal the models to be practically worthless. Somebody should inform the EOS Forum about this basic point of science.
April 6th, 2009 at 6:44 am
As a modeller, though not of the climate, my inclination is that should substantially more climate modelling occur, that it better understand and partition the various sources of uncertainty – structural, parametric, etc – as opposed to increasing accuracy and precision.
Another reason for improving the models could be to facilitate greater acceptance, maybe, but I think what perception changes that are left to be made are more a matter of communication of what we already know rather than knowing more.
Other than that, I like what I read.
April 6th, 2009 at 7:32 am
“Many scientists have called for a substantial new investment in climate modeling to increase the accuracy, precision, and reliability of climate predictions.”
Until the numerical solution methods and the coding of these have been Independently Verified there are no reasons to think that the numbers now being calculated in fact represent the fundamental intent of the modeling.
Note that the procedures of Verification must always precede the processes of Validation.
The accuracy, precision, and reliability of climate ‘predictions’ cannot be improved until the globally accepted activities of Verification, Validation, and Software Quality Assurance have been successfully applied to all aspects of all software used in all analyses. In the absence of these activities failure is very likely with high probability.
The specific problem as I see it is summarized here:
http://danhughes.auditblogs.com/2008/12/08/the-fundamental-issue/
Several aspects are discussed in these:
http://danhughes.auditblogs.com/category/verification/
April 6th, 2009 at 8:59 am
“climate models will, as in the past, play an important, and perhaps central, role in guiding the trillion dollar decisions that the peoples, governments and industries of the world will be making to cope with the consequences of changing climate.”
As someone who has been involved with computational physics and CFD all of my professional life, this statement by the participants in the May 2008 World Modelling Summit for Climate Prediction is stunning in both its arrogance and apparent lack of any rational perspective on the limits of non-linear numerical simulations. Of course, I suppose these people have to make these extreme statements in order to keep their government-fund projects coming in…
April 6th, 2009 at 9:12 am
Last year Rind of GISS gave a talk titled “The Consequences of Not Knowing Low- and High-Latitude Climate Sensitivity”
http://pubs.giss.nasa.gov/docs/2008/2008_Rind.pdf
Rind noted that “the current situation is one where we cannot constrain tropical or high-latitude sensitivity to within a factor of 2; this has some very large consequences for our ability to predict future climate change impacts.”
He concluded that “over the past 25 years we have not been able to quantitatively improve our understanding of low- and high-altitude (or even global) climate sensitivity. That does not mean we have not learned many things; we are more knowledgeable about why models are getting different responses in various locations, and, as the preceding discussion has shown, we are in a position to better understand the consequences of not knowing these sensitivities.”
April 6th, 2009 at 1:13 pm
My favorite quote is “In addition, climate is only one of many
important processes that will influence the
success of any future adaptation efforts, and often it is not the most important factor. Our current ability to predict many of these other
processes—such as the future course of globalization, economic priorities, regulation, technology, demographics, cultural preferences,
and so forth—remains much more limited than our ability to predict future
climate. This raises the question of why improved climate predictions ought to be given such a high priority in designing adaptation
policies.”
I would hypothesize that there is strong pressure to put lots of $ into models because the climate modeling community has grown to a point where they can effectively make a large sociopolitical statement.. perhaps the climate equivalent of “making their own weather”.
If we were more disciplined about science investments, we would develop a panel of decisionmakers and derive a set of alternative research strategies with pricetags, and a set of scenarios (of changes in social, economic and other conditions) and have the decisionmakers on the research strategy that would be the best investment.
Our research investment strategy sometimes seems to be the hungry baby bird method- whoever’s beak is highest gets fed. With all the cognitive processes that humans are capable of, I think we would be able to do better.
April 6th, 2009 at 7:15 pm
If one looks back at the Hansen et al. 1988 GRL paper that kicked off the era of global warming alarm, one sees that the temperatures predicted for Scenario A and Scenario B are significantly higher than the present temperatures (i.e. 2007 to present):
http://rankexploits.com/musings/wp-content/uploads/2008/06/hansenlineartrend.jpg
But it wasn’t so much that the model was bad. And it wasn’t even that the CO2 assumption for Scenario A and Scenario B were too high (they were actually too low). It was that the assumptions for methane, CFCs, and other trace gases were too high.
Garbage in, garbage out.
P.S. Note that the temperatures in 2007 and 2008 were very close to Scenario C. So what is Scenario C?
Under Scenario C, anthropogenic global warming was completely solved in the year 2000 (i.e., emissions of CO2 and all other GHGs exactly balance sinks, so atmospheric concentrations are not increasing for any GHG).
April 7th, 2009 at 12:17 am
I think where modelling will get interesting is in exploring the climate space between high- and low-emissions trajectories. Will we be able to tell if emissions-limitations schemes (if any are instituted) are actually making a difference? There has been some discussion and analysis of the costs of anticipated climate scenarios, but how will we evaluate the benefits (or perhaps the presumed averted costs) of emissions reduction? It may be a big ask if the climate science community is asked 20, 30, 50, or 100 years from now – “how are we doing? You said we had to reduce carbon emissions – now how have we benefited?” In other words there may be a demand to distinguish between the climate we get under a Kyoto-type agreement ad what we would have gotten with no emissions limitations.
This will be a massive detection/attribution problem.
I think there needs to be more honesty with society with what is being “sold.”
First we don’t know what constitutes “dangerous” climate change. There’s been a lot of focus on 2°C warming (above what baseline?) as a limit beyond which there are “dangerous” or “catastrophic” climate changes. (By the way I do not think science should use terms like these as they are not scientific terms but rather involve a set of value judgments.) But that’s a somewhat arbritrary threshold, as it’s a global average, with some places warming faster and more than others, and the models don’t do as well with regional variables as they do with global disributions (as has been discussed here).
But even if we were to accept the 2° C threshold, we don’t know exactly what GHG concentrations would result in that much warming. We also have to acknowledge that even if global society signs on, and adheres to, a post-Kyoto emissions agreement, we have limited ability to set the atmospheric GHG concentrations because we cannot control the behavior of the natural sources of sinks of GHGs like CO2 and methane (Mark Bahner alludes this problem).
Finally we should acknowledge that even if we could set the GHG concentration in the atmosphere at a given desired level, we would still be playing a “all-else-being-equal” game. That is, we cannot at present predict, let alone control, the behavior of other drivers of the climate system.
Though I am a proponent of reducing our carbon emissions, for a range of reasons ( including but not limited to risks posed by climate change), I think science needs to be more honest with society about the limitations of climate science.
April 7th, 2009 at 6:09 am
I would suggest we figure out what’s really going on before we start screwing around with the climate.
If hubris could be quantified …..
Hey maybe some of those climate scientists with a penchant for making up new statistics could be put to work on inventing a statistical calculation for hubris levels. It would be a win, win, win.