My 2012 temperature prediction: +0.65 C (for GISS)

It's mid-February 2012 and the various groups reporting global surface temperature data have all posted numbers that are on the low side compared to thus far in the 21st century, though still several tenths of a degree above the 20th century average. NASA's Goddard Institute for Space Studies (GISS) in particular, which I've been following and posting guesses/predictions on for several years now, has just posted a January 2012 number of 0.36 as the global land + sea temperature anomaly, 0.15 degrees C below the 2011 annual average of 0.51 and 0.27 degrees below the 2010 (and 2005) record of 0.63 degrees.

So naturally, those who deny the scientific evidence for human-caused warming have been going on about cooling, how warming has stopped for many years, and so forth. Though not as adamantly as they were back in 2008 - that cooling proved very brief; perhaps they have learned to be more cautious. If I was in mind to go with simple linear trends and didn't believe the science on warming myself, the January 2012 number of 0.36 sounds like a pretty good bet for the whole of 2012 at this point, maybe it should be even lower.

But I understand the science and know quite clearly warming will continue (with some minor ups and downs) until we stop with the fossil carbon burning - and it may take a while after that to actually stop. I've plugged some numbers in and come up with a science-based prediction for 2012: 0.65 C (plus or minus about 0.07). That middle value would break all previous records. Even on the low end it would put 2012 at the 5th or 6th warmest year on record (just slightly cooler than the remarkable early warm year of 1998 in the GISS index). Why am I so sure 2012 will be so warm?

First, let's review my previous record of "predictions" for GISS. From this post last January and then my April 2011 prediction for last year we have:

Year Date of prediction prediction GISS - February 2012 Difference
2008 02-15-08 0.41 0.44 -0.03
2009 02-15-08 0.55 0.57 -0.02
2010 02-15-08 0.65 0.63 +0.02
2011 04-04-11 0.58 0.51 +0.07

Now, that is pretty remarkable accuracy, well under 0.1 degree difference every year, though I'm a little sad that my 2011 number was more than twice as far off as the others, as I thought it was the most soundly assessed of all the "predictions" up to that point. My April 2011 post used the fitting technique published last year by Foster and Rahmstorf, which found surface temperatures pretty reliably follow a steady warming trend with variation that can be accounted for by slightly time-shifted multiples of the El Nino-Southern Oscillation (ENSO) index, volcanic aerosol estimates, and the solar cycle (as measured by sunspot counts). The solar cycle is reasonably predictable though the current one has ramped up a couple of years later than originally expected. Volcanoes aren't, except that for the most part the influence is essentially zero except when there is a major eruption. Barring a significant eruption putting sulfates in the stratosphere, we can predict the future influence of those two components reasonably well for a year or so in advance.

The tricky one is ENSO - past behavior is known from the MEI index available for download here. There is a typical two to 5 year cycle of ups and downs, with positive phase representing El Nino (higher surface temperatures) and negative phase La Nina (lower temperature). The fit that I did following Foster and Rahmstorf found an MEI coefficient of 0.073, so an MEI value of +2 (intense El Nino) would mean higher surface temperatures (4 months later) by 0.146 degrees C, and an MEI value of -2 (intense La Nina) would mean lower temperatures by -0.146 C. So if you can't predict the future values of the ENSO index very well, you have a range of about -0.15 to +0.15 of uncertainty in any estimate of future temperatures.

Fortunately, as Foster and Rahmstorf found, the value of temperature next month is most closely related to the ENSO value several months previously - so we can have a few months worth of prediction just based on already measured ENSO values. Furthermore, there are a large number of different models that attempt to predict future ENSO values. Based on the average prediction as of mid-March 2011, it looked like the La Nina that had begun in late 2010 would peter out and conditions would return to ENSO index values of close to neutral or perhaps even slightly positive. So that's what I based my 2011 prediction on.

Unfortunately, most of those models predicting ENSO in late 2011 were wrong. ENSO turned to La Nina conditions late in the year, and has continued down to this past month. As you can see from the image below from the IRI ENSO site, the black line (what really happened) is near the low end of all the model predictions (other colors) from about June to December 2011:


The following graph compares what I used for the value of the ENSO index through the end of 2011 (based on what the models were saying back then) with what actually happened:


The effect of the more-negative-than-expected ENSO conditions was to add about -0.5 to the average MEI term for the year, thus contributing about -0.04 to global average temperature for the year. So if I'd known what ENSO was going to do for the rest of the year that would have made my final prediction close to 0.54, within 0.03 again of what actually happened. Still the ENSO drop doesn't explain the full discrepancy - obviously there are terms affecting global average temperature anomaly that aren't accounted for by the Foster-Rahmstorf approach (extra anthropogenic aerosol contributions, or just some simple variation in weather patterns and cloudiness, might be the explanation).

Forging ahead to 2012, using pretty much the same numbers as for the 2011 prediction but with the latest ENSO model average numbers from the IRI site (and 0 for ENSO values beyond 2012) I get the following graph of GISS temperatures from the fit:


So the model predicts 2012 annual global average temperature anomaly in the GISS index to be 0.65 - despite recent cooling, warmer than any preceding year. That should be good to within about 0.07 degrees (given this past years difference, compounding possible errors in the ENSO estimate with whatever the other source of variance was). For 2013 with no ENSO variation included the model gives 0.72, and 0.75 degrees C for 2014 - both of those numbers have to be given higher error bars of plus or minus about 0.15 degrees thanks to ENSO. Including that uncertainty means there's no guarantee of a new global high temperature record in the next 2 years - but there's certainly a very high likelihood of it, and it is almost certain all three years will be hotter with the GISS index than every year of the 20th century except 1998; 2014 should be hotter than 1998 even with a large La Nina.

Of course there's a caveat here - a large volcanic eruption, or a significant increase in human-caused aerosols or some other effect along those lines could cool things off. But all else equal, it looks like we're in for some more warmer years ahead, despite the slightly cooler start we've had so far this year.


Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Rahmstorf predicts Hansen

Rahmstorf predicts Hansen will come up with record temperature. There's a surprise.

Sounds like you believe in a

Sounds like you believe in a conspiracy? And yet the GISS temperature code is the one most easily reproduced by outsiders - at least until the new BEST temperature set came out (and that is land-only, not global, so far). The code and data are all publicly available, and the code itself was rewritten by the Clear Climate Code project, and Nick Stokes has his own "TempLS" that reproduces the same technique in detail - see his blog for examples. Hadley's source data is still not all completely available as far as I'm aware. UAH keeps updating their approach (latest version is "5.3" but supposedly a new "6.0" version will come out with who knows what new adjustments).

Aside from which, this prediction is mine, not Rahmstorf's. I don't know if he would even agree with the estimate I've made.

But if you believe one of the other temperature series is better please go ahead and make your own guess for what it will show by the end of this year. All the global temperature measures have been tracking one another pretty closely for the past 30+ years.

February GISS temperatures

February GISS temperatures are out - anomaly is 0.40. So the first two months of 2012 have been cooler, globally, than every single month of 2011. And yet I stand by my expectation that 2012 will end up hotter than 2011 by the end. Anyone willing to put their name in asserting the cooler side? I notice very few "skeptics" seem to be willing to put numbers to predictions of future temperatures. After the hottest decade in our records, it's perhaps not surprising...

And March GISS anomaly is

And March GISS anomaly is 0.46 - heading up. Also the MEI ENSO values for February/March are up - -0.41 relative to -0.702 for January/February. The latest predictions for the rest of the year aren't up yet but it looks like the La Nina is definitely close to over. Assuming the lingering La Nina suppression has been about the 0.07 degree level we should see temperatures up about that much (i.e. in the mid 0.5's) in coming months. Of course to hit 0.65 for the year the average will have to be 0.73 for the rest of the year. I guess we'll see how things play out...

And now the February MEI

And now the February MEI number is out and it improved to -0.7 for February 2012, so this winter's mild La Nina is vanishing quickly. Most of the ENSO predictions going forward look to see positive territory by June or July, through the end of the year. Nobody's predicting a La Nina return this time, so I continue to believe 2012 will be quite warm.

March is certainly living up to those expectations in the northern hemisphere at least!

I'm glad to see that you've

I'm glad to see that you've backed WAY off your original guess of 0.80 from back in 2008. Based on the increasing difference between your guess and actual temperature I'll guess that 2012 will come in at 0.51 or .14 lower than your guess.

It looks like we're well on our way to ENSO neutral conditions as forecast by several of the models. The big question is whether that will continue or whether we will move into a weak El Nino.

Given that the April GISS

Given that the April GISS anomaly is already 0.56, and the readings the rest of the year should trend higher due to the influence of ENSO conditions, your guess of 0.51 seems certain to be quite low.

I'm not very skeptical of

I'm not very skeptical of another degree C of warming in the 21st century, so maybe I'm not a proper skeptic; however, I am very skeptical of four degrees, and I'm also skeptical of two degrees. After a decade, I see no evidence of an accelerating warming trend. Outside of talk radio, the controversy involves theoretical feedbacks driving this acceleration, not warming vs. cooling. If temperatures don't rise more rapidly soon, I predict that skeptics of AGW skepticism will lower the bar even further.

As I posted in the other

As I posted in the other thread, hae you accounted for the change in Gistemp? If you rerun your Feb 2011 run with the new numbers, do you still get .58?

Also, do you think Romm is

Also, do you think Romm is likely to win his bet with Fuller that 2010-2019 will be at least .15 warmer than 2000-2009? What would you say is the over-under number?

Yes, I'd say Romm is likely

Yes, I'd say Romm is likely to win. I wasn't even aware of this bet though, do you have a link? There's a chance from pure random variation that the average will be low though - but I'd put the odds around 2:1 in favor of Romm on that. Around 10:1 it will be more than 0.10 degrees C warmer.

Examining the current status

Examining the current status of your prediction, we find the GISS anomaly for the first four months (Jan-Apr) has averaged 0.44, so in order to hit 0.65 for the full year, the GISS anomaly has to average 0.75 for the rest of the year, quite a bit higher than last month's 0.56. To get eight months with an average that high means we would need to likely see most of the months over 0.70 and historically there has only been seven months in the GISS data so far that high. Therefore your forecast looks a bit high at this point.

However, ENSO has gone neutral, and in fact the last weekly reading of the Niño 3.4 anomaly is 0.0 ℃, and the last two month reading of the MEI for Mar/Apr is already at +0.06, so we are rapidly transitioning to warmer SSTs in the Pacific. Furthermore, one of the presentation slides in NOAA weekly report shows subsurface heat spreading out across the Pacific, and the CFS.v2 ensemble mean in the same weekly report predicts El Niño conditions to develop by JJA 2012 (3rd slide from the end of the weekly report ...but note the link changes to a new page every week).

So your assessment of ENSO neutral for the year may be a bit low, and for a neutral 2013 considerably too low. It seems you may miss to the downside, perhaps to the edge of your confidence interval for 2012 (say down at 0.58 to 0.60) because of the cool four months we just went through. But the prediction for 2013 of 0.72 is looking pretty good at this time. We should be able to tell by the end of July, if an El Niño is developing, so the next couple of months should be interesting. The fact that we can predict the global temperature anomaly fairly accurately over the next several years (excepting major tropical volcanic eruptions), seems to increase the credibility of the model forecasts for longer terms, even though I realize that may not be the case.

So as of September... * GISS

So as of September...

* GISS has been pretty low so far this year. August number is not out yet, but through July the average is 0.49. Nick Stokes has posted TEMPLS for August and it is up a little, the equivalent GISS number I believe would be about 0.53 (GISS for July was 0.47). That puts the average for the year through August likely right around 0.50. To hit 0.68 by the end of the year would require an avarege of 0.95 the next 4 months! Just to bring the average up to 0.58 (lower bound of my guess for 2012) by the end of December would require GISS to average 0.74 for the next 4 months - that seems unlikely so I have to agree with PaulK2's guess that it's going to be lower than my predicted range this year. Some other factor besides ENSO is keeping temperatures lower than expected? I have to say it's kind of interesting this method is not working out so well this year, I'd love to understand better what's going on...

The interesting question I think is whether the projected January guess of 0.36 will end up closer to the final number than my prediction of 0.68. So far 0.36 is winning by a hair, but if the average bumps over 0.52 by the end of the year then at least it's closer to my number in the end. Small satisfaction there if it happens...

* The latest ENSO MEI number has headed down - July/August is shown as 0.576 down from a real El Nino level of 1.139 in June/July. Maybe the El Nino everybody's been expecting won't show up after all, and we'll be back in La Nina territory and 2013 won't be so warm after all? Interestingly this is something Michael Mann has been predicting for a while - increased predominance of La Nina's as things warm up. Well, prediction is always hard... especially about the future, as Yogi Berra is supposed to have said.

August GISS number is out at

August GISS number is out at 0.56, higher than I just guessed, so good. Except the early numbers for the year have dropped - January 2012 is now 0.32 and February 0.37 !! So the average for the year so far is still only 0.49.

and the final number for 2012

and the final number for 2012 was 0.56 - a tad below the predicted range, but I'm actually rather pleased - much closer to the range than I'd expected given how things were progressing early in the year. My guess for 2012 was off by 0.09 C - but the "projected January value" of 0.36 was off by 0.19 degrees, more than double that difference. So really, once again, reality much more closely hews to science than to the "cooling" proclamations of the deniers.

I do want to do a bit more thorough analysis on what happened in 2012 - something seems to be slightly off with the Foster-Rahmstorff fits, but ENSO has also been acting unusually - once again being in negative territory for much of the year and ending in neutral with no significant period spent positive. ENSO's unpredictability is a big factor in why I've been somewhat off the last couple of years, but maybe there's more to it than that.

Anyway, shortly it will be time to see the January 2013 numbers and try my luck at a prediction for the coming year including some ENSO and solar projections...