The great thing about complex data is that one can basically come up with any number of headlines describing it – all of which can be literally true – but that give very different impressions. Thus we are sure that you will soon read that 2008 was warmer than any year in the 20th Century (with the exception of 1998), that is was the coolest year this century (starting from 2001), and that 7 or 8 of the 9 warmest years have occurred since 2000. There will undoubtedly also be a number of claims made that aren’t true; 2008 is not the coolest year this decade (that was 2000), global warming hasn’t ‘stopped’, CO2 continues to be a greenhouse gas, and such variability is indeed predicted by climate models. Today’s post is therefore dedicated to cutting through the hype and looking at the bigger picture.
As is usual, today marks the release of the ‘meteorological year’ averages for the surface temperature records (GISTEMP, HadCRU, NCDC). This time period runs from December last year through to the end of November this year and is so-called because of the fact that it is easier to dice into seasons than the calendar year. That is, the met year consists of the average of the DJF (winter), MAM (spring), JJA (summer) and SON (autumn) periods (using the standard shorthand for the month names). This makes a little more sense than including the JF from one winter and the D from another as you do in the calendar year calculation. But since the correlation between the D-N and J-D averages is very high (r=0.997), it makes little practical difference. Annual numbers are a little more useful than monthly anomalies for determining long term trends, but are still quite noisy.
The bottom line: In the GISTEMP, HadCRU and NCDC analyses D-N 2008 were at 0.43, 0.42 and 0.47ºC above the 1951-1980 baseline (respectively). In GISTEMP both October and November came in quite warm (0.58ºC), the former edging up slightly on last month’s estimate as more data came in. This puts 2008 at #9 (or #8) in the yearly rankings, but given the uncertainty in the estimates, the real ranking could be anywhere between #6 or #15. More robustly, the most recent 5-year averages are all significantly higher than any in the last century. The last decade is by far the warmest decade globally in the record. These big picture conclusions are the same if you look at any of the data sets, though the actual numbers are slightly different (relating principally to the data extrapolation – particularly in the Arctic).
So what to make of the latest year’s data? First off, we expect that there will be oscillations in the global mean temperature. No climate model has ever shown a year-on-year increase in temperatures because of the currently expected amount of global warming. A big factor in those oscillations is ENSO – whether there is a a warm El Niño event, or a cool La Niña event makes an appreciable difference in the global mean anomalies – about 0.1 to 0.2ºC for significant events. There was a significant La Niña at the beginning of this year (and that is fully included in the D-N annual mean), and that undoubtedly played a role in this year’s relative coolness. It’s worth pointing out that 2000 also had a similarly sized La Niña but was notably cooler than this last year.
While ENSO is one factor in the annual variability, it is not the only one. There are both other sources of internal variability and external forcings. The other internal variations can be a little difficult to characterise (it isn’t as simple as just a super-position of all the climate acronyms you ever heard of NAO+SAM+PDO+AMO+MJO etc.), but the external (natural) forcings are a little easier. The two main ones are volcanic variability and solar forcing. There have been no climatically significant volcanoes since 1991, and so that is not a factor. However, we are at a solar minimum. The impacts of the solar cycle on the surface temperature record are somewhat disputed, but it might be as large as 0.1ºC from solar min to solar max, with a lag of a year or two. Thus for 2008, one might expect a deviation below trend (the difference between mean solar and solar min, and expecting the impact to not yet be fully felt) of up to 0.05ºC. Not a very big signal, and not one that would shift the rankings significantly.
There were a number of rather overheated claims earlier this year that ‘all the global warming had been erased’ by the La Niña-related anomaly. This was always ridiculous, and now that most of that anomaly has passed, we aren’t holding our breath waiting for the ‘global warming is now back’ headlines from the same sources.
Taking a longer perspective, the 30 year mean trends aren’t greatly affected by a single year (GISTEMP: 1978-2007 0.17+/-0.04ºC/dec; 1979-2008 0.16+/-0.04 – OLS trends, annual data, 95% CI, no correction for auto-correlation; identical for HadCRU); they are still solidly upwards. The match of the Hansen et al 1988 scenario B projections are similarly little affected (GISTEMP 1984-2008 0.19+/-0.05 (LO-index) 0.22+/-0.07 (Met-station index); HansenB 1984-2008 0.25+/-0.05 ºC/dec) – the projections run slightly warmer as one would expect given the slightly greater (~10%) forcing in the projection then occurred in reality. This year’s data then don’t really change our expectations much.
Finally, as we’ve discussed before, what climate models did or did not predict is available for all to see. Despite many cautions about using short-term changes to imply something about the long-term trend, these comparisons will still be made. So just for fun, here is a comparison of the observations with the model projections from 1999 to 2008 using 1999 as a baseline. The answer might be surprising for some:
You can get slightly different pictures if you pick the start year differently, and so this isn’t something profound. Picking any single year as a starting point is somewhat subjective and causes the visual aspect to vary – looking at the trends is more robust. However, this figure does show that in models, as in data, some years will be above trend, and some will be below trend. Anyone who expresses shock at this is either naive or … well, you know.
As for the next few years, our expectations are not much changed. This coming winter is predicted to be ENSO neutral, so on that basis one would expect a warmer year next year than this year (though probably not quite record breaking). Barring any large volcanic eruption, I don’t see any reason for the decadal trends to depart much from the anticipated ~0.2ºC/decade.
Update: Just FYI, the same figure as above baselined to 1990, and 1979.
Re: #47
Wayne
The GISS trend at the Arctic was about the same for the 1910-35 period as it was for the 1978-2007 period. Admittedly the starting point in 1978 was higher than it was in 1910 so the final temperatures (as you pointed out) ends up higher in 2007 than in 1935. But the Arctic warmed at pretty much the same rate. Unfortunately, the long-term surface records are the only comparisons we can make because there were no satellites in 1910, so we have no way of knowing what the Arctic troposphere temperatures were back then.
I’m not sure if this addresses any of your points, since you seem to be attributing comments to me which I never made.
Between 1935 and 2007, the Arctic appears to have warmed about half a degree which, in the absence of any other proven cause, I’ve assumed is due to increased GHGs. Though I wouldn’t be surprised if there’s a solar signal in there as well. After all, the 3 strongest solar cycles ever recorded all occurred in the second half of the 20th century. Whether, its GHGs or solar, I’m not convinced a ~0.5 deg increase in 70 years is particularly alarming.
@33…
Julius it could be the ‘swithin’. Weather lore says that the weather for St Swithin’s day sets the pattern for the next 40 days. Weather, a week. Climate, and whole bunch of trends over years. Anything in between, the swithin.
In the graphs, it appears that the GCMs have higher variability than reality. Is this supported by statistics, or is the eye just picking out the outlying GCMs?
[Response: Some do, some don’t – however ten years is a little too short to define what the base standard deviation is, so the relative smoothness of the obs over this period is not necessarily representative of the true variance. – gavin]
Gavin (or someone): could you say a few words about the effects of an El Nino or La Nina episode on the heat balance of the Earth.
Suppose, for example, an El Nino raised the average global temperature by 0.2C above what it would otherwise be for a number of months.
A number of positive (and possibly negative) feedbacks seem likely:
Increase water vapour in the atmosphere => increased GH effect.
Change in cloud cover => changed albedo/night heat loss.
Decreased CO2 uptake to oceans => increased GH effect.
Changes in vegetation => changes in albedo/CO2 levels.
Melting ice/snow => decrease in albedo.
Has anyone tried to quantify these, or are the effects just too small/short-lived or unknown?
I assume the overall effect would be a positive feedback. That is, although an El Nino episode is primarily concerned with re-distributing (but not changing the total amount of) heat in the ocean/atmosphere system, the combined feedbacks would tend to increase the heat in that system. (And for a La Nina the feedbacks would tend to decrease the total heat in the system.)
Re: #51 (John Finn)
Looks to me like it’s closer to a full degree. The 5-year average centered on 1935 is 0.518, the 5-year average centered on 2005 (the most recent available) is 1.636 — a difference of 1.118 deg.C.
Even if you take the highest 5-yr average from the first half of the 20th century (centered on 1945), it’s still only 0.808, fully 0.828 deg.C less than the most recent 5-yr average.
The net warming of the arctic over the entire time span of GISSTEMP data is OVER 2.5 deg.C.
You’ve lived up to your usual habit by fulfilling the prophecy of this post’s title: spin.
I’m wondering about el nino & la nina. Is there some absolute temps that determine whether we are in one of these, or are they relative temps. (Or are these just particular configurations of temperatures, with the average temp of the entire oceans remaining the same?)
It seems to me with the ocean warming, there would be more el nino years as GW progresses, and less la nina….unless it’s based on relatively greater or lesser temps year to year.
Is it just me, or does attempting to “explain” warming trends in terms of cycles really just tacitly abdicate explanation?
More specifically, how does a cycle such as PDO affect the planet’s energy budget? If it can’t, then it isn’t really a “cause” of warming or cooling, though it may be part of a larger mechanism, I suppose.
Aren’t such cycles just emergent distributional patterns within the various energy fluxes, all of which are driven by solar input and modified by the effective rate of radiational output at TOA?
RE #29 & “won’t taking latent heat of melting out of the equation — if the ice disappears early enough to make a difference — mean that temperatures go up faster for the same net energy change..”
A physics teacher I know wondered the same thing, and spoke of how a glass of water with ice cubes stays fairly cool until the last cube completely melts, then the water warms fairly quickly to room temp. I think he said that the reason the water didn’t warm linearly as the ice melted, is because energy was going into the melting process, rather than the warming process; but once there was no more ice the energy went directly into the warming.
Seems this might hold for larger scale events, such as the arctic ice melting (i.e., there would be more warming in the arctic ocean in our current times, except some of the “warming” energy is going into the melting process rather than warming).
Back online after five days without power thanks to bizarre ice storm in northeast. Every other ice storm I have experienced is light freezing rain that goes on and on. This was a major downpour that still managed to freeze. Good post, I wonder what the next substantial el nino will bring given that la nina could not remove 2008 from the top ten?
#51: 0.5 degrees in 70 years might not be (to you – it will be to my grandchildren!), but 0.5 degrees in 20 years (see figure 1: http://earthobservatory.nasa.gov/Features/ArcticIce/arctic_ice3.php) is considerably more alarming. The problem is the ice, or rather the lack thereof. As this quote (from here: http://earthobservatory.nasa.gov/Features/ArcticIce/arctic_ice4.php) explains rather succinctly:
“…as sea ice melts, Arctic waters warm, greatly altering ocean processes, which in turn have an effect on Arctic and global climate, says Michael Steele, senior oceanographer at the University of Washington, Seattle. As the oceans warm and ice thins, more solar energy gets absorbed by the water, creating a positive feedback that leads to further melting and warming.”
Re: #55
You’ve lived up to your usual habit by fulfilling the prophecy of this post’s title: spin.
I don’t do spin,Tamino . I did a quick and hurried least squares fit on 1935-2007 (GISS 64N-90N) then multiplied the annual rise by 72 (number of years). I might have made an error. I did it in my head and the GISS values are in 0.01 deg, so it’s more than possible. I haven’t got access now but I’ll check it again later and if I’ve made a mistake I’m quite preapred to admit it.
More specifically, how does a cycle such as PDO affect the planet’s energy budget? If it can’t, then it isn’t really a “cause” of warming or cooling, though it may be part of a larger mechanism, I suppose.
It doesn’t but it can allow more heat from the oceans into the atmosphere. This is where our measurements have been taken over the past 100 years. As I said to Wayne, above, these are the only comparisons we have.
#49 Richard
“The arctic sea ice will ~finish melting and then everyone will agree. Two to four years should be long enough.”
If you can guarantee an ice-free Arctic, you might want to advise those who are currently building Arctic rigs and ice-breakers. I’m sure they’ll pay you a bunch. Then come and see me, I’ve got a bridge for sale.
Looks at the facts of sea ice cover, not the waving and frothing of alarmists:
http://arctic.atmos.uiuc.edu/cryosphere/IMAGES/current.area.jpg
Re 3, 24,
ref: Abrupt Climate Change: Will It Happen this Century?
http://www.climatescience.gov/Library/sap/sap3-4/final-report/sap3-4-press-release.pdf
Re: #61 (John Finn)
The arctic trend from 1935 to 2007 is demonstrably nonlinear — a linear trend can easily be shown to underestimate the warming. Severely. In fact, the arctic trend from 1975 to the present is also demonstrably nonlinear, and a linear trend again underestimates the warming. This is in sharp contrast to the global trend from 1975 to the present, which is statistically indistinguishable from a linear trend plus stochastic (random) noise.
An ice-free arctic in the summer won’t get rid of the need for ice-breakers in fall, winter, and spring. Nice little strawman you’ve got there.
I have seen a couple of papers, admittedly only in the blogosphere, which try and model the bit between climate change and weather based on AMO ENSO and CO2. These attempts give much better fits than ENSO plus C02 alone and suggest the CO2 component is nearer 0.7c for a doubling which is below the low end of IPCC by quite a lot.
Do you have a view on this Gavin. The obvious criticism is that given enough variables you can get anything to fit anything but its clear there are “natural variations at work” which are more than noise and CO2 signal
[Response: The main issue is that in the real world many things trended up in the 20th Century – sulphate aerosols, CO2, black carbon, maybe even solar for the earlier part at least. Therefore any correlation analysis will conflate these things and so you can get any one of them you like to explain all the trend. None of these correlation methods of climate attribution are robust especially if they use just the global mean surface temperature – which is why physics-based methods are generally preferred. There are some analyses that are worse than others though – for instance using a trend+AMO to match the N. Hemisphere temperatures is simply correlating the temperature with itself. Not much predictability there. Similarly, correlating T against CO2 and expecting the coefficient to give the equilibrium response is just foolish. – gavin]
57 Kevin, you’re right that it’s primarily just a masking of the underlying trend. Technically, I think the common knowledge is actually backwards! An El Nino warms the surface of the planet, so more heat will escape to space. La Nina cools the surface, so less heat escapes. On a first-order level, El Nino cools the total ocean/atmosphere system while La Nina warms it. As you said, the oceans count and atmospheric heat content is negligible compared to OHC.
62 Red, note the ~ in front of “finish”. It means approximately. Also note that the ice is thinning more rapidly than the extent is declining. The extent figure is extremely misleading. The only reason it is used is that it used to be impossible to measure volume. When I asked them a few months ago, the NSIDC said that within a year or two volume measurements will be up and running.
Dr Maslowski made an estimate in 2007 which used data through 2004 and came up with 2013 as the first year with a total meltdown of arctic sea ice. Add in 2007 and 2008 data, and that prediction seems pretty solid. http://news.bbc.co.uk/2/hi/science/nature/7139797.stm
As for icebreakers, that most of the ice will clear in September doesn’t negate the need for November-July. With more utilization of the Arctic Ocean, I’d guess icebreakers will be as busy as ever.
Why do deniers ask for a guarantee when parsing statements which are caveated with “~” and “should”? If I was gonna give a guarantee, I’d a done it. The caveats were carefully inserted and the phrase was in the context of convincing people. I stand by my statement that essentially everyone will be probably be convinced within 4 years.
Re: #61 (John Finn)
The arctic trend from 1935 to 2007 is demonstrably nonlinear — a linear trend can easily be shown to underestimate the warming. Severely. In fact, the arctic trend from 1975 to the present is also demonstrably nonlinear, and a linear trend again underestimates the warming. This is in sharp contrast to the global trend from 1975 to the present, which is statistically indistinguishable from a linear trend plus stochastic (random) noise.
Ok – I’ll give you that. Though, the non-linearity essentially results from the ocean cycles.
The fact remains that the rate of warming in the early 20th century is comparable to that in the late 20th century whether you look at the Arctic in isolation or the globe as a whole and since CO2 levels were markedly different in the 2 periods there must be another significant factor. Incidentally you appear to have written something which contradicts itself but I think I understand your meaning.
Re: #67
57 Kevin, you’re right that it’s primarily just a masking of the underlying trend.
Surely a strong El Nino phase amplifies the trend while a La Nina phase masks the trend. The underlyimng trend, therefore, probably lies somewhere between the 1945-75 trend and 1976-2008 trend.
John Finn (68) — A good approximation formula for the CO2 is logarithmic in the concentration, as found here:
http://forecast.uchicago.edu/samples.html
but the emissions have increased approximately exponentially:
http://cdiac.ornl.gov/trends/emis/tre_glob.html
and the combination, leaving all other factors aside, gives (nearly) a linear trend line. Which, given I’ve left everything else out is about what you’ll eyeball using the decadal averages from the HadCRUTv3 global surface temperature product:
http://tamino.files.wordpress.com/2008/04/10yave.jpg
Re: #68 (John Finn)
All you have to do is look at the linear fit compared to the data, to see how obviously bad that model is. So either you looked and went with that model anyway, or you didn’t even look.
And now I’m supposed to believe you when you say “the non-linearity essentially results from the ocean cycles,” and “there must be another significant factor”?
Deviations from trend also provide useful information. Define weather as annual temperature observations and climate as n-year average weather. The maximum deviation from (linear) trend in the Hadcrut2 (winter) data 1850-2007 occurred in 2006 for n=25 and in 1884 for n=35. The recent warming, despite the accompanying unambiguously higher emissions, is not unambiguously exceptional.
[Response: .. in terms of variability. However, you have to factor in a larger uncertainty in the earlier data… – gavin]
#68 John Finn.
“The fact remains that the rate of warming in the early 20th century is comparable to that in the late 20th century whether you look at the Arctic in isolation or the globe as a whole and since CO2 levels were markedly different”
warming not comparable to recent days, especially with Arctic ice, I am not aware of any historical descriptions comparable to ice conditions now a days, from 1900 to 1950, or from 1985 backwards…
#62 Red Etin, “Looks at the facts of sea ice cover, not the waving and frothing of alarmists:”
Good idea, wrong adjective:
http://www.natice.noaa.gov/pub/West_Arctic/Canadian_Arctic_West/2008/canwcurrentcolor.pdf
http://www.natice.noaa.gov/pub/West_Arctic/Canadian_Arctic_West/1996/canw961203.gif
Even with Polar ice completely melted during the summer in the future, Arctic ice extent will
be quite high during the long night till mid March. Its volume which matters. If you understand the links above, volume is wholesale a lot less compared to 1996, along with open water, with very little multi year ice in the archipelago. We are mostly talking about Climate Change, which is happening, not” Climate Armageddon”… Never read once, not once, Gavin and peers trying to frighten the willies out of anybody.
Some have asked about the El Nino and La Nina trends over time (and the AMO index as well).
The Raw data shows there is slight warming trend in both of these indices going back to 1871 and 1856 respectively. There is an increase of 0.035C per decade in the Raw Nino 3.4 region data and a 0.023C per decade trend in the Raw AMO data. Yes, the oceans are warming but these trends are 10% to 20% of the predicted surface warming trend of about 0.2C per decade.
If you pull the trend out of the ocean indices, as you should do, you can get something like a natural climate variable. The ENSO can affect temperatures by as much as +/-0.2C. The AMO has a bigger impact at as much as +/-0.3C impact.
Here is the detrended Nino 3.4 anomaly back to 1871 – lots of rapid ups and downs. 80% of the peaks (up and down) occur around December.
http://img218.imageshack.us/img218/6444/nino34anomalyft6.png
Here is the detrended AMO index which has much longer cycles of 25 years or so but shows much less overall variability than the ENSO has (+/-0.6C versus the ENSO at +/- 3.0C). Note that it does seem to match some of the bigger climate cycle swings such as the downswing from 1900 to 1919, the upswing from 1920 to 1945, the downswing from 1946 to 1975 and the upswing from 1975 to 2006.
http://img234.imageshack.us/img234/2853/amoanomalyrc3.png
For comparison purposes, here the untrended Raw AMO index which shows an definitive increase over time but the scale and the lower magnitude of the AMO variability makes it look bigger than it really is.
http://img357.imageshack.us/img357/900/trendedamoindexkq9.png
John Lang, pointer please to the source of those images and the data used; if the images are your own work, where did you get the data?
Always enjoy the posts on this site.
On #72: Gavin, isn’t the real point that the magnitude of individual year deviations from the (rising) trend says nothing about the “exceptionality of recent warming”?
Hank, 75, you may want to hold back on the “links links LINKS!!!” mantra.
Try saying what you think is going on and then asking what shows your version to be wrong and theirs right. If you think they are right then ask and say it is so you can see whether it’s a “gut feeling” or whether there’s real information backing it up.
I mean, if you don’t know what else it should be, why do you think it could be wrong? That’s a denialism.
NCAR publishes NAO data going back to 1860. You can find it here. http://www.cgd.ucar.edu/cas/jhurrell/indices.html
This does show an upward trend but it is pretty small as John Lang suggests
An analysis of the NAO by Hurrell can be found on the same website. His paper, some years ago, was agnostic on whether AGW drives/forces/influences the NAO.
Re: #70 David B. Benson
If the increase in CO2 over the last century is exponential, the resulting forcing wouldn’t be linear with time, due to the pre-existing CO2 concentration.
log(A + B*exp(k*t)) is not close to linear in t unless B*exp(k*t) >> A, which is not currently (and hopefully ever) the case.
[Response: You haven’t got it quite right. Exponential growth in this context is related to a fixed time for doubling or a constant percentage growth. This can be written as C=C0 * (1+a)^t or C=C0 * exp(kt). Forcing is ~log(C/C0) = t*log(1+a) or k*t and thus is linear in time. – gavin]
Philip and Lynne, On the question of whether all the melting ice could be slowing warming, the headline today saying we’ve lost 2 trillion tonnes of ice since 2003. A very rough back of the envelope calculation suggest that energy required to melt that ice is equivalent to 1-2% of the additional energy due to greenhouse warming over the same period. Not insubstantial, but not all that significant.
74,
Linear detrending doesn’t make a lot of sense, particularly when you are trying to control for anthropogenic forcings, which has certainly not been linear over the 1856-2008 period. It’s just a local temperature proxy to estimate global temperatures. Just about any local temperature proxy (covering the same amount of area) will fit as well as AMO. ENSO variability helps explain global temperatures, as it combines statistical correlation with physical science. the fact that AMO has low variability suggests you have to be especially careful in attributing cause to a correlation. It probably trends the global temperature anoamlies well, becuase, well, global temperature changes are causing unadjusted AMO variability.
For kicks, I tried incorporating AMO in my 30 year model, and it improved the fit somewhat. The problem is that the shape of AMO doesn’t differ significantly from anthropogenic forcings, and so one can be substituted for the other. One way to interpret this is that anthropogenic forcings are a significant cause of AMO variability (in addition to solar, ENSO, volcanic and anything else that affects climate).
Todd, you’re claiming the AMO — the “O” stands for “oscillation” — matches the result from increasing CO2.
Wha-a-a-at?
More back-of-the-napkin trivia – two trillion tons of ice over 5 years is roughly equivalent to the total discharge of the Mackenzie River, which flows directly into the Arctic basin, over a 6 year period (annual mean of ~10,000 cms).
Re: #80, Gavin’s response
C = C0 * exp(k*t) is of course pure exponential growth, but an equation of that form doesn’t well represent the historical atmospheric CO2 concentration during industrialization, which was the original context (See #70). An added constant, A ~= 280, corresponding to pre-existing CO2 improves it greatly, but removes the linearity of the final expression. Gavin, I assume it’s uncontroversial that the CO2 forcing has increased at an increasing rate over that period.
[Response: No-one is claiming that the entire history of CO2 change has been exponential – that wouldn’t be a good fit at all. The issue is how it is increasing now. (By the way, your expression has the concentration at t=0 (pre-industrial) C0=A+B (not A)). But in any case, do an expansion about today (t1) and you will see that the forcing increase is linear in time i.e. F ~ F(t1) + k*(t-t1)*B*exp(k*t1)/C(t1) (minor correction made) – gavin]
Aren’t there 4 different global temperture sets? Why don’t you include the satellite data from UAH MSU?
[Response: The satellite data measure something different. They are not equivalent to the surface temperatures and comparisons with the models would need to be with those diagnostics specifically. I would also point out that there are multiple versions of those data products as well, and so any comparison should be with all of them (RSS, UAH, UMD (not up to date unfortunately), not just one that you might prefer for some reason. I’ll put up a post on the differences at some point, but the description here is a useful start. – gavin]
Gavin, instead of selecting 1999 or 1998 as the reference year could you show us what the graph would look like if you pick the baseline used by the IPCC in the AR4?
Thanks
[Response: There are lots of different baselines used in IPCC for different reasons and there is no objective way to prefer one over another. But the figures using 1979 or 1990 are linked above already. The basic picture is the same – 2008 is a cool anomaly on the back of a warming trend and is very analogous to similar cool anomalies that occur in the models at random intervals. If you want something more specific let me know. – gavin]
83,
I don’t think there’s as much oscillation going on there as indicated by the linear detrended data. The linear detrending was intended to control for GHGs, but since net anthropogenic forcings are not linear, the untrended data is suspect.
There’s some discussion on this issue here: http://en.wikipedia.org/wiki/Atlantic_Multidecadal_Oscillation.
The question I have is how much of the current AMO warming period is due to a faster local effect of Anthropogenic Global Warming? How would the shape differ if the detrending was done with Net Climate Forcing data, rather than just linear detrending over a 152 year period of time?
Since I’m not an expert, I will defer to those who are. I’m just giving reasons why I think the theory is suspect.
Perhaps Gavin can comment on this?
[Response: I commented on this very issue in an earlier comment thread a few months back: “In control simulations with GCMs such as used in Delworth and Mann (2000) and Knight et al (2005), there is no need to separate the internal multidecadal variability from the forced long-term trend, because there is no change in radiative forcing, and thus no forced (including anthropogenic) large-scale trend to contaminate the estimate of internal multidecadal variability. No such luck in the real world, where both are present. In that case, one needs to use some technique for separating the multidecadal variability from the long-term trend. In many papers, this is simply done by subtracting off a linear trend and defining the residual as e.g. the “AMO”. I don’t particularly like that approach, because the radiatively forced temperature trend is extremely unlikely to be linear in time [this is an issue we discussed in Mann and Emanuel (2006)]. I prefer frequency-domain signal detection techniques such as the “MTM-SVD” technique (for obvious reasons) which was employed by both Delworth and Mann (2000) and Knight et al (2005).” -mike]
As scientific men we should set up a common set of parameters for verification. What kind of temperatures in the next ten years would match the data and what would not?
Re: #85, Gavin’s response
… No-one is claiming that the entire history of CO2 change has been exponential …
Sorry to go on about this minor point. Post #70 was suggesting that CO2 emissions followed an exponential course over a long period, and that therefore the corresponding forcing was linear over the same period (at least since the early 20th century, as this was a response to #68). I was just pointing out that this does not follow, as the forcing goes with the log of total concentration, not the increase (leaving aside how the increase in concentration relates to emissions). Of course the CO2 forcing can be represented as locally linear, but that wasn’t the issue.
[Response: Well, it’s worth pointing out then that the history of CO2 forcing has not been perfectly exponential in the past. However, it is close to exponential now, and the forcing is increasing roughly linearly in time. – gavin]
“There will undoubtedly also be a number of claims made that aren’t true; 2008 is not the coolest year this decade (that was 2000), global warming hasn’t ‘stopped’…”
However, 2008 was the second coolest year of this decade (using your graph), so that statement would be correct.
[Response: A decade is 10 years. The statement is wrong. – gavin]
If global warming hasn’t “stopped”, then why isn’t it getting warmer? Your 1999 figure shows 2008 almost returning to 1999 levels. Do we see any reason to expect an upturn in 2009?
[Response: Yes. Based on our understanding of climate physics, I think the likelihood of a warmer 2009 is very high. – gavin]
Anon, Ya just don’t get it, do you? You can’t tell much of anything from looking at a decade–any decade. Climate is noisy. Period. It is only on multi-decade timescales that long-term trends emerge from the noise.
X, what matters is the change in concentration and the change in forcing–and that is pretty close to linear.
Gavin,
When I look at the comparisons of temperature change vs. model prediction that you showed us, I see something different from what I think that you see.
The first thing that I see is a fallacy of the faggot. You have plotted the output of what looks like a couple of dozen different model simulations. None of them does a particularly good job of tracing the actual temperature changes, but in aggregate, the eyeball average of all of them appears to fall at approximately the ending temperature for the 10-year graph. Even if the aggregate of the models falls close to the 20- and 30-year graphs, the aggregate of a couple of dozen weak arguments is not a strong argument for model accuracy, just as a faggot, a bundle of sticks, is not as strong as a log of equal diameter.
Secondly, the eyeball average of the model outputs does not fall particularly close to the endpoint of the temperature record. It falls somewhat above the ending temperature of the 20-year graph and it falls far above the ending temperature of the 30-year graph.
Finally, I see that the eyeball line of best fit through the model outputs on the 30-year graph implies a trend of approximately 1.5 C degrees per century. If this trend continues, I cannot see the need for panic. That would mean that the temperature change during the next century would be just a bit larger than we have seen in the past century; a change that has caused no obvious harm. Or, perhaps you consider a 30-year trend to be a short-term change that probably does not predict the long-term accurately. Which brings me to ask, what length of change can we assume will predict long-term changes? Is the 130 or so year time that humans have been measuring temperature adequate?
One more thing. All of these comparisons use temperatures between 1980 and the present. Why did you not include the satellite data in your comparisons?
[Response: see above. Satellite data are a different diagnostic of the climate – related, but not commensurate. – gavin]
“There will undoubtedly also be a number of claims made that aren’t true; 2008 is not the coolest year this decade (that was 2000), global warming hasn’t ’stopped’…”
However, 2008 was the second coolest year of this decade (using your graph), so that statement would be correct.
[Response: A decade is 10 years. The statement is wrong. – gavin]
(JW response – OK, if you are including 1999, your “base” year, than 2008 is the third coolest this decade. My point is that calling 2008 the third coolest (rather than the second coolest) this decade is just as accurate as calling it the “…the coolest year this century (starting from 2001)..” (a correct statement) )
[Response: That wasn’t the statement in question. Your point? – gavin]
If global warming hasn’t “stopped”, then why isn’t it getting warmer? Your 1999 figure shows 2008 almost returning to 1999 levels. Do we see any reason to expect an upturn in 2009?
[Response: Yes. Based on our understanding of climate physics, I think the likelihood of a warmer 2009 is very high. – gavin]
(JW response – In that same manner, were you at this time last year predicting that 2008 would be cooler or warmer than the 6 or 7 previous years?)
[Response: There was a substantial La Nina brewing this time last year and 2007 was (or very close to) the warmest year on record. Thus I would have predicted 2008 to be a cooler year than 2007 – Now, I don’t recall actually making such a prediction (though possibly I did somewhere in the comments?) – but the GISTEMP authors did. Their statements seem to have been vindicated. – gavin]
“Even if the aggregate of the models falls close to the 20- and 30-year graphs, the aggregate of a couple of dozen weak arguments is not a strong argument for model accuracy, just as a faggot, a bundle of sticks, is not as strong as a log of equal diameter.”
Incorrect analogy. Model realizations will (and should) differ; their clustering *does* tell us something.
“Secondly, the eyeball average of the model outputs does not fall particularly close to the endpoint of the temperature record.”
Who cares? The endpoint is not a privileged datum.
“That would mean that the temperature change during the next century would be just a bit larger than we have seen in the past century; a change that has caused no obvious harm.”
Perhaps you should go back and have another look at AR4.
“a change that has caused no obvious harm”
Tell that to the polar bears.
snorbert zangox (93) — I’ll not attempt here to list all the harms which global warming has already caused to those living on some edge of the cryosphere, i.e., below glaciers or near the sea of the wwest coast of Alaska. Nor will I attempt to relate droughts and floods to global warming, others have already done that.
Instead I just encourage you to read Mark Lynas’s “Six Degrees”.
JW, Gavin,
The year 2000 was the last year of the 20th century. Think of it as 199ten. The first year of the 21st century was 2001.
JW,
The problem is this whole ranking of individual years, which is not a good way to think about climate. What’s more, the uncertanties over an individual year may be greater than the differences between the nth greatest, nth-1, nth-2, etc years on record. Gavin’s statement about next year being warmer than this probably reflects the La Nina at the beginning of the year rather than the extra years worth of CO2, which is very small. IF you’re addicted to which year comes in 1st and second and third, etc like the media seems to be, then ENSO variations play a large part in that, but it is very small compared to the longer term (~0.7-0.8C) trend.
Gavin,
What is not commensurate about the satellite temperature data?
[Response: They measure something different (MSU-TLT is a weighted average of temperatures reaching from the surface to 10km, peaking at around ~4km and with significant influence from surface type depending on elevation and polar latitude). Therefore comparing them with surface temperature anomalies from the models is not comparing like with like i.e. they are incommensurate. There are ways to create synthetic MSU data from the models, and I’ll discuss this in another post. – gavin]