by Judith Curry
There is growing evidence to support the hypothesis that the pause cause is tied to a change in tropical Pacific Ocean circulations. What are the implications of this for climate sensitivity and attribution of warming in the latter part of the 20th century?
The paper by Kosalka Xie (discussed on the previous thread Pause tied to equatorial Pacific cooling) is generating substantial discussion in the blogosphere and twitosphere. I focus in this post not so much on the pause, but on the warming in the last quarter of the 20th century. I reproduce here the following figure from the paper:
In my previous post, I argued that POGA-C (fixed external forcing) showed substantial warming since 1975 (through 1998), and a substantial fraction of the observed warming (possibly 50% or more, based on eyeball estimate). The significance of this is in context of the IPCC AR4 attribution statement, whereby most (>50%) of the warming in the latter half of the 20th century is anthropogenic.
I’ve had email and twitter discussions on this with Ed Hawkins and John Nielsen-Gammon. One major issue is exactly how Kosalka and Xie conducted their simulations in terms of forcing the equatorial Pacific temperatures; another is how to interpret the implicit external forcing that exists in the POGA-C simulation.
I have been emailing with John Nielsen-Gammon this past weekend, who has two posts on this paper:
Nielsen-Gammon has digitized the results from the paper. He computes trends for 1975-2002, and 2002-2012. He comes up with a trend of 0.19C for POGA-C, and 0.51C for observations. Which might lead you to infer ~37% of the observed trend can be explained by Pacific variability for this period (note, if the end point is 1998 as in my initial analysis, the trend is higher). He then goes on to infer that:
So according to the model, the tropical Pacific by itself was responsible for 0.19 C of warming, and all of that was due to the response of the tropical Pacific to radiative forcing. The effect of natural variability in the tropical Pacific on the linear trend over that period was very small and negative, a mere -0.04 C. So contrary to Curry’s mind-blowing first impression, the results of Kosaka and Xie imply that natural variability in the tropical Pacific did not contribute at all to the rapid warming from 1975 to 2002.
Nielsen-Gammon justifies the bolded statement as TPNV= POGAH – HIST, which he interprets as natural-only changes in the central and eastern tropical Pacific. Which doesn’t make sense to me, since it looks the 1998 El Nino effect had a cooling effect on the climate. I find the bolded statement to be unjustified.
For clarification, I include an excerpt from our email exchange:
JNG: POGA-C shows directly how much specifying 8% of the earth’s surface can influence by itself a global trend. TPNV is consistent with that amount of influence. (See figure comparing them in my blog.)
Here’s a link to the Compo & Sardeshmukh paper: http://www.springerlink.com/content/au9x40l201105273/fulltext.pdf
JC: Yes, the way that fixing that 8% influences the global trend is through setting the global atmosphere/ocean circulation, i.e. natural internal variability.
JNG: Also it’s not just 8% of the ocean that’s affected. I expect the temperature signal is advected to the west tropical Pacific by ocean currents and from there spreads north and south. The strong atmosphere-ocean feedback in the area will also cause the winds to be altered and in turn affect ocean temperatures upstream.
JC: agreed that’s how it works, through circulations. but to have the global temperature raised in a substantial way, something is going on other than specifying the temperature in this region. if you specified 8% of land temperature, it wouldn’t make much difference at all to global temperature. If you specified 8% in the indian ocean, not much would happen either. The point is that specifying the temp in this particular region, this synchronizes the global network of ocean/atmospheric circulations in a realistic way, with the natural internal variability of the coupled ocean/atmosphere system providing the major signal to the global climate.
So, we agree on the basics of the the equatorial Pacific can influence global climate. Where we disagree is to what extent the specification of surface temperature for 8% of the global surface area influences the global trend through radiative forcing. A simple analysis might say less than 10%, simply by virtue of the small area, and by virtue of understanding that say the 1998 El Nino event was not directly triggered by radiative forcing. But JNG seems to think otherwise.
He makes the following point that I agree with:
- the GFDL model is too sensitive to external forcing (the period 1970-2010 warms by 0.2C too much).
Tamino also has a post on this entitled el Nino and the Non-Spherical Cow. He also includes the same figure as JNG, where he looks at POGAH – HIST:
If we take the difference between the POGA-H models (with ENSO constrained to follow historical data) and the HIST models, we see the estimated influence of ENSO on global temperature history:
This is, according to the new research, how ENSO has modified global temperature since 1950. The influence is clear: a pronounced recent ENSO-induced cooling which has cancelled the continued global warming due to man-made CO2, leading to the “hiatus” in the increase of global temperature.
I assume that Tamino also got his numbers by eyeballing Kosalka and Xie’s graphs. At least in Tamino’s version of the diagram, the 1998 shows a small warming anomaly, although the magnitude seems unrealistically small.
Tamino then criticizes my analysis:
Her first mistake — quite an embarrassing one really — was to assume that this [POGA C] was the influence of ENSO on global temperature history. This quite misses the point, that one of the strengths of the new approach is that it allows climate forcing and ENSO to interact in a nonlinear manner. The actual estimate of the influence of ENSO, according to the new research, is shown in the graph labelled “POGA-H minus HIST.”
Oooooh, I’m just blushing with embarrassment. The interesting thing about POGA C is that it uses FIXED external forcing. External forcing has some small influence via the specification of the 8% surface temperatures in the central Pacific, which includes the effects of PDO/ENSO as well as external forcing.
So the question du jour is: Does POGA C or POGA-H minus HIST provide a better estimate of the impacts of ENSO/PDO on the global climate?
I say it is POGA C. There are nonlinear interactions between the forced and unforced variability, and we have seen that GFDL model is too sensitive to external forcing. So I don’t think much of Tamino’s and JNG’s interpretation of POGA-H minus HIST. But that said, all this is not easily untangled.
Tamino closes with this howler:
As the graph labelled “POGA-H minus HIST” shows, the influence of natural variation, at least that part of it from ENSO, has been cooling, not warming, and if we want to assign a percentage we should say that natural variation has been responsible for about negative 25% of global warming. Not only did Judith Curry execute one of the most blatant, most obvious, and most ludicrous examples of cherry-picking, she couldn’t even get the sign of the influence right. That’s what I’ve come to expect from her.
Pay attention, Tamino. The PDO/ENSO had a warming effect during the period 1976 to circa 2000, then a cooling effect since about 2002. The question du siecle is How much of the warming in the last quarter of the 20th century was caused by natural internal variability? Looking at unforced simulations such as POGA-C provides important clues. The conclusion that Tamino draws, that natural variability has a uniformly cooling effect, and JNG’s analysis that it has no effect (or maybe up to 30% with the uncertainty analysis in his second post) are not convincing, particularly in context of their allowing for the PDO to be the pause cause since 2002.
The bottom line is that natural internal variability and forced variability are very difficult to disentangle. IMO the natural internal variability is of intrinsic importance to global climate on multidecadal time scales, and this needs to be considered in an integrated way; not just a forced signal with natural variability noise.
Tsonis et al. provide some insights regarding how to think about this. Below is an additional paper of relevance.
Michael Ghil has a very important new paper [link]
A Mathematical Theory of Climate Sensitivity or, How to Deal With Both Anthropogenic Forcing and Natural Variability?
Abstract. Recent estimates of climate evolution over the coming century still dier by several degrees. This uncertainty motivates the work presented here. There are two basic approaches to apprehend the complexity of climate change: deterministically nonlinear and stochastically linear, i.e. the Lorenz and the Hasselmann approach. The grand unification of these two approaches relies on the theory of random dynamical systems. We apply this theory to study the random attractors of nonlinear, stochastically perturbed climate models. Doing so allows one to examine the interaction of internal climate variability with the forcing, whether natural or anthropogenic, and to take into account the climate system’s non-equilibrium behavior in determining climate sensitivity. This non-equilibrium behavior is due to a combination of nonlinear and random effects. We give here a unified treatment of such effects from the point of view of the theory of dynamical systems and of their bifurcations. Energy balance models are used to illustrate multiple equilibria, while multi-decadal oscillations in the thermohaline circulation illustrate the transition from steady states to periodic behavior. Random effects are introduced in the setting of random dynamical systems, which permit a unified treatment of both nonlinearity and stochasticity. The combined treatment of nonlinear and random effects is applied to a stochastically perturbed version of the classical Lorenz convection model. Climate sensitivity is then defined mathematically as the derivative of an appropriate functional or other function of the systems state with respect to the bifurcation parameter. This definition is illustrated by using numerical results for a model of the El Ni~noSouthern Oscillation.
I have seen mention of this paper pop up several times on different threads, here is an opportunity to discuss this paper in context of this specific issue.
The results of the Kosalka and Xie simulations can be interpreted in numerous ways. Trying to filter out the ENSO from the PDO signal seems to me to be an erroneous thing to do, given their intrinsic relationship. Using these simulations to attribute the pause (since 2002) to the cooling effect of ENSO/PDO has a corollary that the warm phase of the PDO in the last quarter of the 20th century also contributed to this warming.
The focus for the last two decades has been on the forced climate response. Natural internal variability has been regarded as noise. The pause has stimulated research into the contribution from natural internal variability, which is a very welcome development. How can we proceed to better understand the role of natural internal variability on climate change? More climate model simulations are needed along these lines, with different experimental designs and using different climate models. More insights are needed from observational analyses. And better theoretical frameworks are needed for understanding climate sensitivity to external forcing in a system with substantial natural internal variability.