Bob Tisdale:
The new paper Risbey et al. (2014) will likely be very controversial based solely on the two co-authors identified in the title above (and shown in the photos to the right). As a result, I suspect it will garner a lot of attention…a lot of attention. This post is not about those two controversial authors, though their contributions to the paper are discussed. This post is about the numerous curiosities in the paper. For those new to discussions of global warming, I’ve tried to make this post as non-technical as possible, but these are comments on a scientific paper.
OVERVIEW
The Risbey et al. (2014) Well-estimated global surface warming in climate projections selected for ENSO phase is yet another paper trying to blame the recent dominance of La Niña events for the slowdown in global surface temperature warming, the hiatus. This one, however, states that ENSO contributes to the warming when El Niño events dominate. That occurred from the mid-1970s to the late-1990s. Risbey et al. (2014) also has a number of curiosities that make it stand out from the rest. One of those curiosities is that they claim that 4 specially selected climate models (which they failed to identify) can reproduce the spatial patterns of warming and cooling in the Pacific (and the rest of the ocean basins) during the hiatus period, while the maps they presented of observed versus modeled trends contradict the claims.
IMPORTANT INITIAL NOTE
I’ve read and reread Risbey et al. (2014) a number of times and I can’t find where they identify the “best” 4 and “worst” 4 climate models presented in their Figure 5. I asked Anthony Watts to provide a second set of eyes, and he was also unable to find where they list the models selected for that illustration.
Risbey et al. (2014) identify 18 models, but not the “best” and “worst” of those 18 they used in their Figure 5. Please let me know if I’ve somehow overlooked them. I’ll then strike any related text in this post.
Further to this topic, Anthony Watts sent emails to two of the authors on Friday, July 18, 2014, asking if the models selected for Figure 5 had been named somewhere. Refer to Anthony’s post A courtesy note ahead of publication for Risbey et al. 2014. Anthony has not received replies. While there are numerous other 15-year periods presented in Risbey et al (2014) along with numerous other “best” and “worst” models, our questions pertained solely to Figure 5 and the period of 1998-2012, so it should have been relatively easy to answer the question…and one would have thought the models would have been identified in the Supplementary Information for the paper, but there is no Supplementary Information.
Because Risbey et al. (2014) have not identified the models they’ve selected as “best” and “worst”, their work cannot be verified.
INTRODUCTION
The Risbey et al. (2014) paper Well-estimated global surface warming in climate projections selected for ENSO phase was just published online. Risbey et al. (2014) are claiming that if they cherry-pick a few climate models from the CMIP5 archive (used by the IPCC for their 5th Assessment Report)—that is, if they select specific climate models that best simulate a dominance of La Niña events during the global warming hiatus period of 1998 to 2012—then those models provide a good estimate of warming trends (or lack thereof) and those models also properly simulate the sea surface temperature patterns in the Pacific, and elsewhere.
Those are very odd claims. The spatial patterns of warming and cooling in the Pacific are dictated primarily by ENSO processes and climate models still can’t simulate the most basic of ENSO processes. Even if a few of the models created the warning and cooling spatial patterns by some freak occurrence, the models still do not (cannot) properly simulate ENSO processes. In that respect, the findings of Risbey et al. (2014) are pointless.
Additionally, their claims that the very-small, cherry-picked subset of climate models provides good estimates of the spatial patterns of warming and cooling in the Pacific for the period of 1998-2012 are not supported by the data and model outputs they presented, so Risbey et al. (2014) failed to deliver.
There are a number of other curiosities, too.
ABSTRACT
The Risbey et al. (2014) abstract reads (my boldface):
The question of how climate model projections have tracked the actual evolution of global mean surface air temperature is important in establishing the credibility of their projections. Some studies and the IPCC Fifth Assessment Report suggest that the recent 15-year period (1998–2012) provides evidence that models are overestimating current temperature evolution. Such comparisons are not evidence against model trends because they represent only one realization where the decadal natural variability component of the model climate is generally not in phase with observations. We present a more appropriate test of models where only those models with natural variability (represented by El Niño/Southern Oscillation) largely in phase with observations are selected from multi-model ensembles for comparison with observations. These tests show that climate models have provided good estimates of 15-year trends, including for recent periods and for Pacific spatial trend patterns.
Curiously, in their abstract, Risbey et al. (2014) note a major flaw with the climate models used by the IPCC for their 5th Assessment Report—that they are “generally not in phase with observations”—but they don’t accept that as a flaw. If your stock broker’s models were out of phase with observations, would you continue to invest with that broker based on their out-of-phase models or would you look for another broker whose models were in-phase with observations? Of course, you’d look elsewhere.
Unfortunately, we don’t have any other climate “broker” models to choose from. There are no climate models that can simulate naturally occurring coupled ocean-atmosphere processes that can contribute to global warming and that can stop global warming…or, obviously, simulate those processes in-phase with the real world. Yet governments around the globe continue to invest billions annually in out-of-phase models.
Risbey et al. (2014), like numerous other papers, are basically attempting to blame a shift in ENSO dominance (from a dominance of El Niño events to a dominance of La Niña events) for the recent slowdown in the warming of surface temperatures. Unlike others, they acknowledge that ENSO would also have contributed to the warming from the mid-1970s to the late 1990s, a period when El Niños dominated.
CHANCE VERSUS SKILL
The fifth paragraph of Risbey et al. (2014) begins (my boldface):
In the CMIP5 models run using historical forcing there is no way to ensure that the model has the same sequence of ENSO events as the real world.This will occur only by chance and only for limited periods, because natural variability in the models is not constrained to occur in the same sequence as the real world.
Risbey et al. (2014) admitted that the models they selected for having the proper sequence of ENSO events did so by chance, not out of skill, which undermines the intent of their paper. If the focus of the paper had been need for climate models to be in-phase with obseervations, they would have achieved their goal. But that wasn’t the aim of the paper. The concluding sentence of the abstract claims that “…climate models have provided good estimates of 15-year trends, including for recent periods…” when, in fact, it was by pure chance that the cherry-picked models aligned with the real world. No skill involved. If models had any skill, the outputs of the models would be in-phase with observations.
ENSO CONTRIBUTES TO WARMING
The fifth paragraph of the paper continues:
For any 15-year period the rate of warming in the real world may accelerate or decelerate depending on the phase of ENSO predominant over the period.
Risbey et al. (2014) admitted with that sentence, if a dominance of La Niña events can cause surface warming to slow (“decelerate”), then a dominance of El Niño events can provide a naturally occurring and naturally fueled contribution to global warming (“accelerate” it), above and beyond the forced component of the models. Unfortunately, climate models were tuned to a period when El Niño events dominated (the mid-1970s to the late 1990s), yet climate modelers assumed all of the warming during that period was caused by manmade greenhouse gases. (See the discussion of Figure 9.5 from the IPCC’s 4th Assessment Report here and Chapter 9 from AR4 here.) As a result, the models have grossly overestimated the forced component of the warming and, in turn, climate sensitivity.
Some might believe that Risbey et al (2014) have thrown the IPCC under the bus, so to speak. But I don’t believe so. We’ll have to see how the mainstream media responds to the paper. I don’t think the media will even catch the significance of ENSO contributions to warming since science reporters have not been very forthcoming about the failings of climate science.
Risbey et al (2014) have also overlooked the contribution of the Atlantic Multidecadal Oscillation during the period to which climate models were tuned. From the mid-1970s to the early-2000s, the additional naturally occurring warming of the sea surface temperatures of the North Atlantic contributed considerably to the warming of sea surface temperatures of the Northern Hemisphere (and in turn to land surface air temperatures). This also adds to the overestimation of the forced component of the warming (and climate sensitivity) during the recent warming period. Sea surface temperatures in the North Atlantic have also been flat for the past decade, suggesting that the Atlantic Multidecadal Oscillation has ended its contribution to global warming, and, because by definition the Atlantic Multidecadal Oscillation lasts for multiple decades, the sea surface temperatures of the North Atlantic may continue to remain flat or even cool for another couple of decades. (See the NOAA Frequently Asked Questions About the Atlantic Multidecadal Oscillation (AMO)webpage and the posts An Introduction To ENSO, AMO, and PDO — Part 2 and Multidecadal Variations and Sea Surface Temperature Reconstructions.)
For more than 5 years, I have gone to great lengths to illustrate and explain how El Niño and La Niña processes contributed to the warming of sea surface temperatures and the oceans to depth. If this topic is new to you, see my free illustrated essay “The Manmade Global Warming Challenge” (42mb). Recently Kevin Trenberth acknowledged that strong El Niño events cause upward steps in global surface temperatures. Refer to the post The 2014/15 El Niño – Part 9 – Kevin Trenberth is Looking Forward to Another “Big Jump”. And now the authors of Risbey et al. (2014)—including the two activists Stephan Lewandowsky and Naomi Oreskes—are admitting that ENSO can contribute to global warming. How many more years will pass before mainstream media and politicians acknowledge that nature can and does provide a major contribution to global warming? Or should that be how many more decades will pass?
RISBEY ET AL. (2014) – AN EXERCISE IN FUTILITY
IF (big if) the climate models in the CMIP5 archive were capable of simulating the coupled ocean-atmosphere processes associated with El Niño and La Niña events (collectively called ENSO processes hereafter), Risbey et al (2014) might have value…if the intent of their paper was to point out that models need to be in-phase with nature. Then, even though all of the models do not properly simulate the timing, strength or duration of ENSO events, Risbey et al (2014) could have selected, as they have done, specific models that best simulated ENSO during the hiatus period.
However, climate models cannot properly simulate ENSO processes, even the most basic of processes like Bjerknes feedback. (Bjerknes feedback, basically, is the positive feedback between the trade wind strength and sea surface temperature gradient from east to west in the equatorial Pacific.) These model failings have been known for years. See Guilyardi et al. (2009)and Bellenger et al (2012). It is very difficult to find a portion—any portion—of ENSO processes that climate models simulate properly. Therefore, the fact that Risbey et al (2014) selected models that better simulate the ENSO trends for the period of 1998 to 2012 is pointless, because the models are not correctly simulating ENSO processes. The models are creating variations in the sea surface temperatures of the tropical Pacific but that “noise” has no relationship to El Niño and La Nina processes as they exist in nature.
Oddly, Risbey et al (2014) acknowledge that the models do not properly simulate ENSO processes. The start of the last paragraph under the heading of “Phase-selected projections” reads [Reference 28 is Guilyardi et al. (2009)]:
