Since 1998 the global temperature has risen more slowly than before. Given the many explanations for colder temperatures discussed in the media and scientific literature (La Niña, heat uptake of the oceans, arctic data gap, etc.) one could jokingly ask why no new ice age is here yet. This fails to recognize, however, that the various ingredients are small and not simply additive. Here is a small overview and attempt to explain how the different pieces of the puzzle fit together.
Figure 1 The global near-surface temperatures (annual values at the top, decadal means at the bottom) in the three standard data sets HadCRUT4 (black), NOAA (orange) and NASA GISS (light blue). Graph: IPCC 2013.
First an important point: the global temperature trend over only 15 years is neither robust nor predictive of longer-term climate trends. I’ve repeated this now for six years in various articles, as this is often misunderstood. The IPCC has again made this clear (Summary for Policy Makers p. 3):
Due to natural variability, trends based on short records are very sensitive to the beginning and end dates and do not in general reflect long-term climate trends.
You can see this for yourself by comparing the trend from mid-1997 to the trend from 1999 : the latter is more than twice as large: 0.07 instead of 0.03 degrees per decade (HadCRUT4 data).
Likewise for data uncertainty: the trends of HadCRUT and NASA data hardly differ in the long term, but they do over the last 15 years. And the small correction proposed recently by Cowtan & Way to compensate for the data gap in the Arctic almost does not change the HadCRUT4 long-term trend, but it changes that over the last 15 years by a factor of 2.5.
Therefore, it is a (by some deliberately promoted) misunderstanding to draw conclusions from such a short trend about future global warming, let alone climate policy. To illustrate this point, the following graph shows one simulation from the CMIP3 model ensemble:
Figure 2 Temperature evolution in a model simulation with the MRI model. Other models also show comparable “hiatuses” due to natural climate variability. This is one of the standard simulations carried out within the framework of CMIP3 for the IPCC 2007 report. Graph: Roger Jones.
In this model calculation, there is a “warming pause” in the last 15 years, but in no way does this imply that further global warming is any less. The long-term warming and the short-term “pause” have nothing to do with each other, since they have very different causes. By the way this example refutes the popular “climate skeptics” claim that climate models cannot explain such a “hiatus” – more on that later.
Now for the causes of the lesser trend of the last 15 years. Climate change can have two types of causes: external forcing or internal variability in the climate system.
External forcing: the sun, volcanoes & co.
The possible external drivers include the shading of the sun by aerosol pollution of the atmosphere by volcanoes (Neely et al., 2013) or Chinese power plants (Kaufmann et al. 2011). Second, a reduction of the greenhouse effect of CFCs because these gases have been largely banned in the Montreal Protocol (Estrada et al., 2013). And third, the transition from solar maximum in the first half to a particularly deep and long solar minimum in the second half of the period – this is evidenced by measurements of solar activity, but can explain only part of the slowdown (about one third according to our correlation analysis).
It is likely that all these factors indeed contributed to a slowing of the warming, and they are also additive – according to the IPCC report (Section 9.4) about half of the slowdown can be explained by a slower increase in radiative forcing. A problem is that the data on the net radiative forcing are too imprecise to better quantify its contribution. Which in turn is due to the short period considered, in which the changes are so small that data uncertainties play a big role, unlike for the long-term climate trends.
The latest data and findings on climate forcings are not included in the climate model runs because of the long lead time for planning and executing such supercomputer simulations. Therefore, the current CMIP5 simulations run from 2005 in scenario mode (see Figure 6) rather than being driven by observed forcings. They are therefore driven e.g. with an average solar cycle and know nothing of the particularly deep and prolonged solar minimum 2005-2010.
Internal variability: El Niño, PDO & co.
The strongest internal variability in the climate system on this time scale is the change from El Niño to La Niña – a natural, stochastic “seesaw” in the tropical Pacific called ENSO (El Niño Southern Oscillation).
The fact that El Niño is important for our purposes can already be seen by how much the trend changes if you leave out 1998 (see above): El Niño years are particularly warm (see chart), and 1998 was the strongest El Niño year since records began. Further evidence of the crucial importance of El Niño is that after correcting the global temperature data for the effect of ENSO and solar cycles by a simple correlation analysis, you get a steady warming trend without any recent slowdown (see next graph and Foster and Rahmstorf 2011). ENSO is responsible for two thirds of the correction. And if you nudge a climate model in the tropical Pacific to follow the observed sequence of El Niño and La Niña (rather than generating such events itself in random order), then the model reproduces the observed global temperature evolution including the “hiatus” (Kosaka and Xie 2013) .
One can also ask how the observed warming fits the earlier predictions of the IPCC . The result looks like this (Rahmstorf et al 2011):
Figure 3 Comparison of global temperature (average over 5 data sets, including 2 satellite series) with the projections from the 3rd and 4 IPCC reports. Pink: the measured values. Red: data after adjusting for ENSO, volcanoes and solar activity by a multivariate correlation analysis. The data are shown as a moving average over 12 months. From Rahmstorf et al. 2012.
And what about the ocean heat storage ? That is no additional effect, but part of the mechanism by which El Niño years are warm and La Niña years are cold at the Earth’s surface. During El Niño the ocean releases heat, during La Niña it stores more heat. The often-cited data on the heat storage in the ocean are therefore just further evidence that El Niño plays a crucial role for the “pause”.
Leading U.S. climatologist Kevin Trenberth has studied this for twenty years and has just published a detailed explanatory article. Trenberth emphasizes the role of long-term variations of ENSO, called pacific-decadal oscillation (PDO). Put simply: phases with more El Niño and phases with predominant La Niña conditions (as we’ve had recently) may persist for up to two decades in the tropical Pacific. The latter brings a somewhat slower warming at the surface of our planet, because more heat is stored deeper in the ocean. A central point here: even if the surface temperature stagnates our planet continues to take up heat. The increasing greenhouse effect leads to a radiation imbalance: we absorb more heat from the sun than we emit back into space. 90% of this heat ends up in the ocean due to the high heat capacity of water. The fact that the ocean continues to heat up, without pause, demonstrates that the greenhouse effect has not subsided, as we have discussed here.
How important the effect of El Niño is will be revealed at the next decent El Niño event. I have already predicted last year that after the next El Niño a new record in global temperature will be reached again – a forecast that probably will be confirmed or falsified soon.
The Arctic data gap
Recently, Cowtan & Way have shown that recent warming was underestimated in the HadCRUT data. After using satellite data and a smart statistical method to fill gaps in the network of weather stations, the global warming trend since 1998 is 0.12 degrees per decade – that is only a quarter less than the long-term trend of 0.16 degrees per decade measured since 1980. Awareness of this data gap is not new – Simmons et al. have shown already in 2010 that global warming is underestimated in the HadCRUT data, and we have discussed the Arctic data hole repeatedly since 2008 at RealClimate. NASA GISS has always filled the data gaps by interpolation, albeit with a simpler method, and accordingly the GISTEMP data show hardly a slowdown of warming.
The spatial pattern
Cohen et al. have shown two years ago that it is mainly the recent cold winters in Eurasia that have contributed to the flattening of the global warming curve (see figure).
Figure 4 Observed temperature trends in the winter months. Despite the significant global warming in the annual mean, there was a winter cooling in Eurasia. CRUTem3 data (land only!), from Cohen et al. 2012.
They argue that an explanation for the “pause” in global warming would have to explain this particular pattern. But this is not compelling: there could be two independent mechanisms superimposed. One that dampens global warming – which would have to be explained by the global energy balance. And a second one that explains the cold Eurasian winters, but without affecting the global mean temperature. I think the latter is likely – these recent cold winters are part of the much-discussed “warm Arctic – cold continents” pattern (see, eg, Overland et al 2011) and could be related to the dwindling ice cover on the Arctic Ocean, as we explained here. Since the heat is just moved around, with Eurasian cold linked to a correspondingly warmer Arctic, this hardly affects the global mean temperature – unless you’re looking at a data set with a large data gap in the Arctic …
What does it add up to?
How does all that fit together now? As described above, I think (just like Trenberth) that natural variability, in particular ENSO and PDO, is the main reason the recent slower warming. From the perspective of the planetary energy balance heat storage in the ocean is the key mechanism.
If the warming is steady after adjusting for ENSO, volcanoes and solar cycles, does the additional correction for the Arctic data gap by Cowtan & Way mean that the warming after these adjustments has even accelerated? That could be, but only by a small amount. As you can see in Figure 6 of our paper (Foster and Rahmstorf), the slowdown is gone after said adjustment in the GISS data and the two satellites series, but there still is some slowdown in the two data sets with the Arctic gap, ie HadCRUT and NCDC. Adding the trend correction by 0.08 degrees per decade from Cowtan & Way to our ENSO-adjusted HadCRUT trend from 1998, you end up at about 0.2 degrees per decade, practically the same value as we got for the GISS data. If one further adds the effect of the above forcings (without the solar activity already accounted for) this would add a few hundredths of a degree. The result would be a bit faster warming than over the entire period since 1980, but probably less than the 0.29 °C per decade measured over 1992-2006. Nothing to get excited about. Especially since based on the model calculations you’d expect anyway trends around 0.2 degrees per decade, because models predict not a constant but a gradually accelerating warming. Which brings us to the comparison with models.
Comparison with models
Figure 5 Comparison of the three measured data sets shown at the outset with earlier IPCC projections from the first (FAR), 2nd (SAR) 3rd (TAR) and 4th (AR4) IPCC report, as well as with the CMIP3 model ensemble. As you can see the data move within the projected ranges. Source: IPCC AR5, Figure 1.4. (Small note: “climate skeptics” brought an earlier, erroneous draft version of this graphic to the public, although it was marked in block letters as a temporary placeholder by IPCC.)
When comparing data with models, one needs to understand a key point: the models also produce internal variations, including ENSO, but as this (similar to the weather in the models) is a stochastic process, the El Niños and La Niñas are distributed randomly over the years. Therefore, only in rare cases a model will randomly produce a sequence that is similar to the observed sequence with reduced warming from 1998 to 2012. There are such models – see the first image above – but most show such phases of slow warming or “hiatus” at other times.
The IPCC has therefore never tried to predict the climate evolution over 15 years, because that’s just too much influenced by random internal variability (such as ENSO), which we cannot predict (at least as yet).
However, all models show such variability – no one who understands this issue could have been surprised that there can be such hiatus phases. They’ve also occurred in the past, for example from 1982, as Trenberth shows in his Figure 4.
The following graph shows a comparison of observational data with the CMIP5 ensemble of model experiments that have been made for the current IPCC report. The graph shows that the El Niño year 1998 is at the top and the last two cool La Niña years are at the bottom of the model projection range (for the various reasons explained above). However, the temperatures (at least according to the data of Cowtan & Way) are within the range which is spanned by 90% of the models.
Figure 6 Comparison of 42 CMIP5 simulations with the observational data. The HadCRUT4 value for 2013 is provisional of course, still without November and December. (Source: DeepClimate.org)
So there is no evidence for model errors here (for more on this see this article) . This is also no evidence for a lower climate sensitivity, even if this was proposed some time ago by Otto et al. (2013). Trenberth et al. suggest that even the choice of a different data set of ocean heat content would have increased the climate sensitivity estimate of Otto et al. by 0.5 degrees. In addition, Otto et al. used the HadCRUT4 temperature data with its particularly low recent warming. With an honest appraisal of the full uncertainty, also in the forcing, one must come to the conclusion that such a short period is not sufficient to draw conclusions about the climate sensitivity.
Conclusion
Global temperature has in recent years increased more slowly than before, but this is within the normal natural variability that always exists, and also within the range of predictions by climate models – even despite some cool forcing factors such as the deep solar minimum not included in the models. There is therefore no reason to find the models faulty. There is also no reason to expect less warming in the future – in fact, perhaps rather the opposite as the climate system will catch up again due its natural oscillations, e.g. when the Pacific decadal oscillation swings back to its warm phase. Even now global temperatures are very high again – in the GISS data, with an anomaly of + 0.77 °C November was warmer than the previous record year of 2010 (+ 0.67 °), and it was the warmest November on record since 1880.
PS: This article was translated from the German original at RC’s sister blog KlimaLounge. KlimaLounge has been nominated as one of 20 blogs for the award of German science blog of the year 2013. If you’d like to vote for us: simply go to this link, select KlimaLounge in the list and press the “vote” button.
References
- K. Cowtan, and RG Way, “Coverage bias in the HadCRUT4 temperature series and its impact on recent temperature trends”, Quarterly Journal of the Royal Meteorological Society, pp. n / an / a, 2013. http://dx.doi.org/10.1002/qj.2297
- RR Neely, OB Toon, S. Solomon, J. Vernier, C. Alvarez, JM English, KH Rosenlof, MJ Mills, CG Bardeen, JS Daniel, and JP Thayer, ”
Recent Increases in anthropogenic SO2from Asia have minimal impact on stratospheric aerosol
“Geophysical Research Letters, vol. 40, pp. 999-1004, 2013. http://dx.doi.org/10.1002/grl.50263 - RK Kaufmann, H. Kauppi, ML man, and JH Stock, “Reconciling anthropogenic climate change with observed-temperature 1998-2008”, Proceedings of the National Academy of Sciences, vol. 108 pp. 11790-11793, 2011. http://dx.doi.org/10.1073/pnas.1102467108
- F. Estrada, P. Perron, and B. Martínez-López, “Statistically derived Contributions of diverse human Influences to twentieth-century temperature changes”, Nature Geoscience, vol. 6, pp. 1050-1055, 2013. http://dx.doi.org/10.1038/ngeo1999
- G. Foster, and S. Rahmstorf, “Global temperature evolution 1979-2010”, Environmental Research Letters, vol. 6, pp. 044 022, 2011. http://dx.doi.org/10.1088/1748-9326/6/4/044022
- Y. Kosaka, and S. Xie, “Recent global-warming hiatus tied to equatorial Pacific surface cooling”, Nature, vol. 501 pp. 403-407, 2013. http://dx.doi.org/10.1038/nature12534
- S. Rahmstorf, G. Foster, and A. Cazenave, “Comparing Projections to climate observations up to 2011,” Environmental Research Letters, vol. 7, pp. 044 035, in 2012. http://dx.doi.org/10.1088/1748-9326/7/4/044035
- AJ Simmons, KM Willett, PD Jones, PW Thorne, and DP Dee, “Low-frequency variations in surface atmospheric humidity, temperature, and precipitation: Inferences from reanalyses and monthly gridded observational data sets”, Journal of Geophysical Research, vol. 115, 2010. http://dx.doi.org/10.1029/2009JD012442
- JL Cohen, JC Furtado, MA Barlow, VA Alexeev, and JE Cherry, “Arctic warming, Increasing snow cover and wide spread boreal winter cooling”, Environmental Research Letters, vol. 7, pp. 014 007, in 2012. http://dx.doi.org/10.1088/1748-9326/7/1/014007
- JE Overland, KR Wood, and M. Wang, “Warm Arctic-cold continents: climate impacts of the newly open Arctic Sea”, Polar Research, vol. 30, 2011. http://dx.doi.org/10.3402/polar.v30i0.15787
- A. Otto, Otto FEL, O. Boucher, J. Church, G. Hegerl, PM Forster, NP Gillett, J. Gregory, GC Johnson, R. Knutti, N. Lewis, U. Lohmann, J. Marotzke, G. Myhre, D. Shindell, B. Stevens, and MR Allen, “Energy budget constraints on climate response”, Nature Geoscience, vol. 6, pp. 415-416, 2013. http://dx.doi.org/10.1038/ngeo1836
A central point here: even if the surface temperature stagnates our planet continues to take up heat. The increasing greenhouse effect leads to a radiation imbalance: we absorb more heat from the sun than we emit back into space.
There’s one piece to that jigsaw which is not often discussed: the primary feedbacks (water vapour, clouds), which ultimately determine the magnitude of the imbalance, are mainly dependent on surface temperature change rather than the mere presence of GHGs or related energy fluxes. That means a period of stagnant surface temperatures, due to ENSO/PDO/IPO, would presumably be expected to correspond with a small reduction in the rate of energy accumulation.
Regarding the Otto et al. 2013 paper, IIRC their results were largely insensitive to the “hiatus” in the HadCRUT4 data because their method used decadal averages, much like the lower panel of Figure 1 above.
“The IPCC has therefore never tried to predict the climate evolution over 15 years, because that’s just too much influenced by random internal variability (such as ENSO), which we cannot predict (at least as yet).”
[edit]
As we all remember, the IPCC said in AR4 explicitly: “For the next two decades, a warming of about 0.2°C per decade is projected for a range of SRES emission scenarios. Even if the concentrations of all greenhouse gases and aerosols had been kept constant at year 2000 levels, a further warming of about 0.1°C per decade would be expected. {10.3, 10.7}. – Since 2000, the concentration of CO2 has increased at BAU speed, but warming has been insignificant. RSS now shows a 207 month period with 0.0 trend. Sadly, Rahmstorf does not offer any falsification criteria for his promise/prediction/projection/guess that warming will resume.
[Response: That AR4 quote does not make a specific forecast for a 15-year period. It is from one of those yellow boxes in the Summary for Policy Makers which highlight key findings in extremely condensed form, as can be seen by the imprecise phrasing (e.g. “about 0.2°C”, without specific error bars) – its intention is to give you a rough indication of how much warming the models project on average in the early part of the 21st Century. But the SPM also very clearly shows the error bars: in Fig. SPM.6 probability distributions are shown for the warming by the year 2020-2029, which even include a small probability of cooling. Unless you quote-mine out of context, you have to say that the IPCC AR4 already very clearly communicated the fact that short-term trends can vary widely due to natural variability. -Stefan]
Re the Arctic data gap
Does the interpolation of the data gap account for the underrepresented initial baseline, according to Beerling et al 2011?
Judith Lean’s Bjerknes Lecture at the AGU last week gave a very nice perspective on these issues as well.
Is anybody aware of any papers that demonstrate that there is statistically significant evidence for the existence of a change in the underlying rate of warming?
It seems to me that the onus should be on those who are confidently claiming that there has been a genuine hiatus in surface temperatures to demonstrate that the apparent flattening is not explainable by the natural variability in the data. I’ve not seen this done, and my initial experiments on breakpoint detection seem to suggest that there is insufficient evidence to suggest that there has been a change in the underlying rate of warming.
Paul S (#1) – Since the Planck Response dominates over positive feedback responses to temperature, wouldn’t a La Nina-like failure of surface temperature to rise lead to an increase rather than a reduction in energy accumulation compared with accumulation during a surface warming – presumably a small increase, so that the observed rise in ocean heat content would still be substantial?
To clarify my above comment, I was suggesting that the observed rise in ocean heat content would be substantial with or without the La Nina effect, representing primarily the persistence of a long term warming trend.
The Otto study’s TCR estimates for the 2000s would be unaffected by whether Levitus or ORAS-4 is used for OHC estimates, is it not?
But regarding ECS, what are your thoughts on this post implying that satellites don’t confirm the change in OHC circa 2005 (thus arguing that ORAS-4 pre-2005 estimates are high due to data errors)?
Dikran Marsupial
A couple of months ago Tamino said “By at least one calculation, the difference is “statistically significant”, but doesn’t cite the source. As always, his piece is well worth reading anyway.
See Tamino’s http://tamino.wordpress.com/2013/09/21/double-standard/
Why hasn’t the ‘pause’ been debunked just by noting that it starts at a point nearly a decade’s worth of warming above the trend in 1997? And if we include the ‘pause’ in the long term trend, the trend actually goes up. These seem like powerful points, but I’ve only seen Gavin note them, not the IPCC, or Met Office.
The major physics at play here is that as long as there is less outgoing radiated energy than incoming, temperatures must increase until we are back at equilibrium. ENSO and deep ocean heating are blips in the bigger scheme of things. If the oceans release energy to the atmosphere, that doesn’t fix the long-term imbalance. If we are in a lower than normal solar cycle, that is also a blip. Increasing GHG concentrations slows outgoing radiation.
This stuff is all interesting and useful but we should not forget the big picture when looking at what influences the blips.
see also:
http://tamino.wordpress.com/2011/04/11/co2-shame/Stefan: the intro says “various ingredients are small and not simply additive” while further down you say “ all these factors indeed contributed to a slowing of the warming, and they are also additive” – but nothing explicitly links these two statements together. Maybe you could clarify?
Matti Virtanen #2: Check out Figure 1. Decadal averages are moving up pretty consistently with prediction. Do you have a source for your RSS “analysis”? Not anti-science blog I hope …
[Response: Sure: what I called additive (at least to good approximation) are the radiative forcings. Not additive are ENSO and ocean heat uptake, since they are two ways at looking at the same thing (again to reasonable approximation). -Stefan]
Re Paul S at #1 and Fred Moolten at #6:
Prolonged suppression of Planck feedback; I tried to explain this here:
Thought experiment
It looks as if there would be a greater sea level rise for a given amount of added greenhouse gas.
RB (#8) – ORAS-4 may have overestimated OHC uptake, but perhaps not as much as implied in the link you cited – see, e.g.,OHC data. Regarding ECS (“equilibrium climate sensitivity”), I think there are difficulties estimating anything truly resembling a Charney-type ECS from data involving OHC uptake and forcing estimates, because these estimates are fraught with so many uncertainties, and because the values that are calculated, even if accurate, bear an uncertain relationship to how the climate would behave at equilibrium. My preference would be to refer to these as estimates of “effective climate sensitivity” rather than ECS. Even the conventional notion of ECS involving the short-term (Charney) feedbacks doesn’t represent an equilibrium result, which is better represented by “Earth System Sensitivity” estimates. Maybe the word “equilibrium” should be omitted from all climate sensitivity estimates, from the shortest term values (TCR) to the longest and most comprehensive (Earth System), since all the different forms of sensitivity estimation seem, in my view, to be looking at somewhat different phenomena and should not necessarily yield the same values.
Mal Adapted:
Tamino’s reference to “statistically significant” refers to the warm period that ran from 1992 to 2006, not the current “hiatus”.
Matti Virtanen,
Ever hear of physics? Try it sometime. It’s great!
As always, a fine, clear, well written piece, Stefan. But, err, apropos of nothing much … would “hiatus” be one of those big words for ordinary things?
Fred at #6:
That is what Kosaka and Xie found in their elegant little model experiment — see their supplementary Figure 3. I hope all you modellers out there are busy replicating / extending…
I’m wondering if anyone has collected the model runs with long hiatus periods in them and looked for commonalities…for example extended periods of La Ninas or anything else. It would be fascinating if there was only a small group of associations- could lead to targets for further research. Apologies if this has already been done. If not, anyone looking for a slightly pre-retirement Chemist to take on as a visiting scientist on a project like this?
I think Matti Vertanen (#2) probably got his claim about the RSS data from Monckton on WUWT, and while Monckton is not a particularly credible source in my opinion, I think the RSS data are probably as claimed. I also agree that model predictions of 0.2 C surface warming per decade were clearly inaccurate, but on the larger question of climate trends, they were probably not very far off. What has happened over recent decades is that planetary warming has continued unabated, as evidenced by ocean heat content (OHC) increases, while surface warming has slowed, and tropospheric warming as measured by RSS has halted. The reasons clearly lie in the shift in the distribution of the accumulating heat in the ocean to greater depths with less remaining on the surface. Models are still not skillful enough to anticipate the timing of these shifts, but they are not too bad at getting the planetary trends right, or at least keeping their estimates within reasonable proximity to observed trends, even if on the high side of observations.
The OHC data are critical to any analysis of surface temperature change. In particular, as discussed above, internal climate oscillations warm the surface by losing OHC, while external forcing by CO2 or other modalities warms the surface by increasing OHC. This permits us to apportion surface temperature change over long intervals on the basis of OHC change. An important quantitative consideration of this principle has been discussed by Isaac Held at Heat Uptake and Internal Variability. It indicates that the current combination of post-1950 warming, OHC rise, and the geographical distribution of temperature change renders a substantial contribution to the warming from internal variability highly unlikely, and the role of greenhouse gas forcing accordingly the likely contributor to a very large fraction of the observed warming.
Stefan:
//”They argue that an explanation for the “pause” in global warming would have to explain this particular pattern. But this is not compelling: there could be two independent mechanisms superimposed.”//
I don’t understand how your objection to the Cohen recommendation follows from the fact that multiple mechanisms may be important. Isn’t this whole problem one of trying to tease out the small contributions from external forcing, internal variability, instrumentation uncertainty, or mis-specified forcing (in the case of model world)? The relevance of the underlying spatial pattern may be tied very much to the contributing terms. I’d agree if the hiatus is simply a product of external forcing and we’re comparing the internal variability of observations vs. a model (since the variability is very high in Eurasian winter), but if we’re interested in things like the ocean state, prevailing modes of variability, etc then this may project onto the Eurasian DJF signal seen in the obs.
[Response: Just saying that “the explanation for the pause” need not have anything to do with the pattern of winter cooling over Eurasia. It might have an ENSO/PDO style spatial pattern as shown by Trenberth. And the Eurasian winter cooling could be a separate, superimposed effect, with no impact on global mean temperature and hence nothing to do with the “pause”. I think it is important to be clear about this. -Stefan]
Who is Monckton? Let the numbers do the talking – the RSS data is here: http://data.remss.com/msu/monthly_time_series/RSS_Monthly_MSU_AMSU_Channel_TLT_Anomalies_Land_and_Ocean_v03_3.txt
Mal, Hank & Michael, many thanks for the response, Taminos article is on a slightly different topic (although the point he makes is spot-on as usual).
Tamino also had an excellent set of three articles on step changes, which is the sort of thing I am interested in, but he didin’t look at the hiatus.
http://tamino.wordpress.com/2012/01/09/steps/
Here’s an interesting item from the CryoSat people. Oct 2013 ice volume is about 9000 km3, up from 6000, in 2012.
http://www.esa.int/Our_Activities/Observing_the_Earth/CryoSat/Arctic_sea_ice_up_from_record_low
Watching the 2010-2013 Oct. data, it look like it’s the highest in the 4 years of taking data.
Also definitely higher then the PIOMAS model predicts.
http://www.esa.int/Our_Activities/Observing_the_Earth/CryoSat/CryoSat_reveals_major_loss_of_Arctic_sea_ice
To bad we don’t have a longer record.
Matti Virtanen #24: thanks for pointing to the numbers. If you try to find a trend over that period, it is close to zero, but so is the correlation. In other words, the period is too short for the trend to break out of the noise. If you are not familiar with the concept of statistical significance, look it up.
Matti Virtanen, Now why would you pick 1996 and why RSS? I mean you have a plethora of choices of start date and date series. Why that combination? Could it be because RSS seems to be much more strongly affected by ENSO than the terrestrial data sets or even UAH?
Tamino has also looked at this, and found that RSS was odd man out. Cherrypickers cherrypick. Film at 11:00.
At least
he’s consistent, always turning in the same direction.
‘oogle: “Matti Virtanen” climate
Fred Moolton #6 – The Planck response (or lack of) is the counter-balance which would make any effect small. My reasoning for expecting a reduction in energy accumulation is based on the observation that models with greater sensitivity tend to transiently develop greater imbalances, given equivalent forcing time series. This means the rate of energy accumulation is dependent on feedbacks as well as forcing, even though higher sensitivity models trigger larger Planck responses.
During a period in which surface warming is stifled by internal variability the rate of energy accumulation would be influenced only by the forcing – there would be no difference between a high-sensitivity model and a zero-feedback model (assuming zero-dimensional models; the reality, with regionally varying temperatures and feedbacks, would be more complex). Given that greater sensitivity matters for the magnitude of imbalance a period during which this higher sensitivity is not activated should mean a slower rate of energy accumulation than would be the case during a period of “normal” variability.
Short-term temperature changes can be interesting as to their specific causes. However, we shuold be looking at longer-term trends. Over the past century, both a linear fit and polymeric fit culminate at a similar value for 2013 Even though they diverge by over a tenth of a degree at times. Indeed, the short-term temperature may even dip below the long-term trend (as indicated by the polymeric fit). However, this does not indicate that the long-term trend will deviate significantly from the established trend.
http://www.climate4you.com/
Paul S (#31 – Paul, I may be misinterpreting your explanation, but the comparison I was making (and has been made previously by others, including Stefan, Trenberth and more, apparently supported by observations) is that a failure of the surface to warm due to a La Nina-like process will increase energy accumulation by reducing OLR relative to the OLR that would have been produced by a warmer surface. Positive feedbacks attenuate Planck Responses but don’t overcome them. In the case of a failure of the surface to warm due to a La Nina-like process, the OLR reduction (and hence the energy gain) will be lessened by the reduction in water vapor and other feedback moieties, but it will still be greater than occurs with a warmed surface. As an example of the level of Planck Response attenuation from positive feedbacks, see Table 1 in Soden and Held 2006.
On a more frivolous note, Matti Virtanen (#24) asked me, “Who is Monckton?”. Matti – he’s the person referred to near the bottom of this page – Matti Knows Monckton. Aren’t you glad to have your question answered?
Dan H. has achieved the impossible–unifying statistics with chemistry: Behold the polymeric fit!
Thanks for a good clarifying blog post.
I think it is a bit strange how twisted analysis perspectives sometimes have become, even IPCC seems rather fond of the ‘hiatus’ concept (e.g, Chapter 9).
But picking shorter parts of noisy global temperature curves, giving them suggestive names and then try to “explain” them is not what you’d expect from year 2000+ science methodology (neither would sacrificing a goat or two to the ‘hiatus’ solve any real problems…).
In chapter 11.3.6.3 they conclude:
“…it is concluded that the hiatus is attributable, in roughly equal measure, to a decline in the rate of increase in effective radiative forcing (ERF
) and a cooling contribution from internal variability
(expert judgment,medium confidence)”.
So they mean that the ‘hiatus’ stands out because it follows a forced trend with internal variability around?
Better put the primitive intuition in the back seat and use more advanced formal statistical methods. There is just a number of data series with fuzzy observations which you can test your preferably physics-based models against. And in statistics all intervals are created equal, you don’t pick only one for analysis, your model has to explain all intervals.
Ray Ladbury suggests that we try physics.
But our problem is planetary physics: a whole bunch of sometimes coupled not exactly oscillators along with a handful of forcing functions and debates (see above for instance) over which is which. All on a rotating sort of sphere where every thing pushes on every thing else. It gets hard.
I think many commenters would be helped by an RC article on La Niña as a physical phenomenon. Given a purely statistical discussion, many miss that during La Niña the trade winds blow strongly away from the coast of South America, leading to an upwelling of deep ocean water off the coast. This water happens to be cold. The deep ocean does not develop a hole. Water that comes up is replace by water going down. The water going down is warmer than the water that came up. The strong winds blow water across the Pacific. This water becomes warm and produces heavy rain back onto itself. But across the Pacific water piles up a bit and the rain clouds are blown beyond. The remaining water becomes salty (evaporation caries water away but not salt). This warmed salty dense water is some of the water that sinks to replace the cold water that came up near South America. More heat is stored in the ocean. Due to conservation of energy there is less energy to warm the land.
But why did the wind blow so hard in the first place? Oh well, I said it gets hard.
The war against the Greenhouse Effect started in December 2004 – Richard L. Mit’s interview in the NY Times – the warming stopped after 1998. It has been revived and refueled by just about every pro-pollutionist individual and group since then. No explanation will end their claims – they win just by getting the science bent out of shape when it curries (no pun intended) to their treatment of the atmosphere as a conventional oven, and the oceans as a bathtub.
ENSO neutral conditions are in effect, and NOAA reported November as the hottest November in the record. The next El Nino is going to heat the globe into new records territory. It won’t cool back down. There hasn’t been a statistically significant cooling period since the 1890s – and that’s not natural variability or a co-incidence.
If you think the pinheads of pollution won’t spin more blarney looking for sticky-stuff, save your energy. They’re already crowing up the record cold reported in East Antarctica (-135.8dF in August 2010 and -135.3dF in July 2013).
The real number that matters is the heating driver: CO2e. Despite the worst economic shrivel in a century, the CO2 pollution has accelerated – averaging over 2 ppm annually for the last decade (CO2Now.org). It sets records just about every month of every year … and has for half a century. The 395.3ppm for November is just another record-breaking November. Get the pro-pollutionists to pause that.
RC & stefan: Thanks. This reference should prove useful.
“First an important point: the global temperature trend over only 15 years is neither robust nor predictive of longer-term climate trends.”
As an older slower physics guy trying get a handle on all this, can someone tell me how many years equals a robust and predictive interval?
Pete Dunkelberg,
Focus first on the spherical cow–the basic fact behind climate change is energy conservation. Unless Matti is willing to deny that basic fact, he doesn’t have a leg to stand on. The rest is just how the energy gets distributed, and it merely determines precisely how screwed we are.
C. Town Springer, see http://moregrumbinescience.blogspot.ca/2009/01/results-on-deciding-trends.html
And http://tamino.wordpress.com/2012/07/06/how-long/
Thanks, Stefan.
I kind of hate the whole idea, though–the ‘hiatus’ meme is being driven by sheer repetition. Rebunking, squared.
I usually point to the inconsistency between the denialist claim that ‘we’ ignore everything but CO2 as forcing, and the denialist claim that the ‘hiatus’ disproves CO2 as forcing–that last only makes any sense if CO2 is ‘the only forcing.’ But I have no idea how many times I’ve had to post that line so far, so I’m grateful for something I can link to by way of variety.
Of course, I could always update this:
http://hubpages.com/hub/When-Did-Global-Warming-Stop
In line with my way of thinking, Judith Lean is putting the jigsaw puzzle together with her NRL statistical climate model. Inspired by that model as well as by Foster & Rahmstorf, Kosaka & Xie, and Cowtan & Way, I have been putting together what I refer to as the CSALT model. This pieces together all the natural variability terms in a way that we are better able to extract the CO2 control knob signature. I will keep leaning on it until it breaks, but so far it is robust as it uses all the factors that skeptics seem to think are important. The latest check is to see how the solar cycle fits in to the multivariate analysis:
http://contextearth.com/2013/12/18/csalt-model-and-the-hale-cycle/
So Stefan, keep pursuing this approach.
Old physics guy @ 39:
BPL explains the 30 year approximation here
http://bartonpaullevenson.com/30Years.html
Spice it with
https://www.realclimate.org/index.php/archives/2008/01/uncertainty-noise-and-the-art-of-model-data-comparison/
and
http://tamino.wordpress.com/2013/09/21/double-standard/
With a constantly changing climate I don’t know how well the 30 year rule holds up.
As a physics guy you expect causative physical factors behind the observed effects. You can learn about the main forcings (factors that increase or decrease outgoing longwave radiation (OLR)) at Makiko’s page
http://www.columbia.edu/~mhs119/
But of course, physical and human factors that change the forcings must be considered. The physical factors rapidly stray from physics.
The chemical equilibrium of CO2 between air and ocean dictates that the seas take in nearly half of the CO2 we create by burning reduced carbon. Ocean CO2 in turn strongly involves sea creatures and their shells.
The current climate change, unlike the iced age cycles of the last million years or so, started with CO2 forcing instead of orbital forcing. This big CO2 increase running ahead of the likely consequent ocean warming is a strong stimulus to Arctic plant growth. This is one reason why the tundra CO2 bomb scare is premature.
Sulfate aerosols from coal plants = a major cooling forcing. This forcing may decrease rapidly one of these days. This may leave the 30 year approximation of climate behind. But the climate models, which take forcings as input and calculate from there, are expected to keep up.
What gives climate scientists confidence that climate science is on the right track? The Big Three are paleoclimatology, modern data and models. Paleoclimatology is #1 according to Gavin Schmidt, one of the scientists in the RC group and someone who would know. Paleoclimatology itself has lots of unclear questions, but whether all details are understood or not earth responds to changes in atmospheric CO2. Bye the way physics guy, increased CO2 warms earth some, leading to more water vapor which has a greater greenhouse effect than the CO2 as such.
> Old physics guy
More physicists, some names you’ll likely recognize, at:
http://www.azimuthproject.org/azimuth/show/HomePage
I think it’s a mistake to refer to changes in global average surface air temperatures as if they were definitive measures of the change to the climate system. Averaged over long enough periods, sure, but year to year, decade to decade they retain the capacity to deceive. Personally I think global heat content better reveals actual, accumulated change.
Stefan I think you and Grant Foster already answered the “hiatus” question quite well; surely just one or two more la Nina years than el Nino’s over a period as short as 15 years would be sufficient to create the illusion of a “hiatus”. Just as a couple more el Nino’s than la Nina’s would create the illusion of accelerated warming. And surely the fact that having just that circumstance – with strong el Nino at the start of the cherry picked period and la Nina’s at the tail – should see a clear temperature ‘drop’ instead of merely levelling off confirms the existence of an underlying warming trend.
Tying the short term variability to physical climate phenomena and processes is going to be valuable, and when it comes to fighting to have the climate problem taken seriously we need explanations that are easy to understand. ENSO stands out as perhaps the single most potent phenomena affecting year to year, decade to decade variability in surface air temperatures, with TSI and aerosols adding or subtracting their shares.
Are there any other physical phenomena that are or can be known well enough to reduce the unknowns and reduce the ‘noise’ in similar fashion, that if adjusted for can allow the underlying trend to be revealed more clearly?
At WUWT where Monckton’s article appears, I tried to post the following:
When I tried to post, it said the comment could not be posted. WTF’s up with That?
Due to natural variability, trends based on short records are very sensitive to the beginning and end dates and do not in general reflect long-term climate trends.
Correct, but the short term natural variability may provide an insight into origins of natural variability, then extended to many decades even centuries.
In recent years, number of articles published suggest that solar wind which exerts well known, defined and measured magnetic pressure on the Earth’s magnetosphere, also has some less defined effect on the upper layers of the atmosphere (known as the Mansurov effect).
It may be obvious that such an effect would vary in step with the sunspot cycle, but that is not the case for a simple but a lesser known reason.
The Earth’s field is not constant either, it shows similar decadal variability, as shown in the data from and used by number of distinguished geo-magnetic scientists and researchers (Jault Gire, LeMouel, J. Bloxham, D. Gubbins, A.Jackson, R. Hide, D. Boggs, J. Dickey etc,)
Since changes in either of two fields affect strength of the magnetosphere, it would be expected that the ‘magnetospheric variability’ time function could be produced by combining two sets of available data.
That is exactly what I did some months ago introducing terms ‘Geo-Solar Oscillation’ and ‘Geo-Solar Cycle’.
Comparing the GSC to two well known climatic sets of data opens a way into an unexpected and fascinating direction for climatologists’ research
http://www.vukcevic.talktalk.net/GSC1.htm
The above graph when back extrapolated to 1700, gives a favourable comparison to two other well known AMO reconstructions.
Not that I expect, but if the author of the above article would be tempted to add these data to the ‘hypothesised’ CO2 effect, the GSC data would be available.
vukcevic, The solar cycle contribution may be there but it is definitely small if one teases the Hale cycle factors out of the temperature time series.
http://contextearth.com/2013/12/18/csalt-model-and-the-hale-cycle/
Letting a multiple linear regression do the work is certainly more robust than your eyeballing estimates. The eyeballing always exaggerates the parts that fit and tries to obscure the parts that don’t. So if you have something, specify the formulation clearly and those of us that can do the numerical fitting can try it out.
WebHubTelescope: “The eyeballing always exaggerates the parts that fit and tries to obscure the parts that don’t.”
QFFT! Thank you. This is why economists have predicted 12 of the last 5 recessions! Cyclicity is particularly problematic because any bounded series will appear to cycle when it nears the bounds. Quasi-periodic behavior is a whole lot more common than actual cyclic behavior. Even the Solar Cycle is not a true periodic phenomenon.