Correcting for more than just el Niño

I recently showed a version of “el Niño-Corrected” temperature from Gavin Schmidt at NASA. My own calculation suggested that el Niño caused about the same contribution to 2015’s heat as Gavin’s estimate. However, there are some pronounced differences between our calculations.


Here’s Gavin’s el Niño-corrected result compared to the raw data:

Here’s mine:

nasa1

My corrected data show a lot less wiggling around. For example, my correction accounts for the 1998 heat nicely, showing that its extremity was entirely due to el Niño, while Gavin’s still has 1998 well above the trend line even after correction. Also, mine doesn’t show some of the dips in Gavin’s curve, such as the extreme cooling in 1992.

There are two reasons for the differences. One is that I’ve used a more complex model of the el Niño effect, one which matches the 1998 outburst (and others too) much better. The other is that I’ve corrected for more than just el Niño; I’ve adjusted for solar variations and volcanic aerosols too, so my correction removes the 1992 dip which was because of the eruption of the Mt. Pinatubo volcano. The result of correcting for three factors instead of just one, and of a more sophisticated el Niño correction, is a much steadier warming for about the last forty years. We can see this more clearly by plotting just the corrected data (black lines are a piecewise-linear fit by change-point analysis):

nasa

I’ve performed the same calculation for the other main surface temperature data sets. Here are the results for data from NOAA:

noaa1

noaa

Here they are for data from the Hadley Centre/Climate Research Unit in the U.K.:

cru1

cru

Finally, for the estimate from Cowtan & Way:

cw1

cw

The corrected data sets have some things in common. They all still have 2015 as the hottest year on record, despite having removed the el Niño impact effectively. After correction, only the NOAA data still show 2014 as being 2nd-hottest. Most important, they all make claims of a “pause” or “hiatus” in global warming look even more silly than we already knew they were.

When it comes to preparing for the climate change to come, it’s the 2015 heat that tells the most important tale. But when it comes to denier propaganda, the big story is that the 1998 mega-heat really was due to el Niño, and with that factor accounted for 1998 is shown to be just another year on the trend line, part of the continuing march of temperatures upward.

If 2016 shows a similar outburst (as it well may), expect it to awaken many people to the reality and risk of man-made climate change. But you should also expect it to revive the deniers when, a few years later, they begin to crow about how its mega-heat was the end of global warming. It’ll be the 1998 story all over again — not the end of global warming, just a lot of extra heat from el Niño and a lot of hot air from the usual suspects.


If you like what you see, feel free to donate at Peaseblossom’s Closet.

18 responses to “Correcting for more than just el Niño

  1. “But you should also expect it to revive the deniers when, a few years later, they begin to crow about how its mega-heat was the end of global warming.”

    True. The deniers, however, are already in the position of the boy who cried “No wolf!” To relink:

    http://hubpages.com/politics/When-Did-Global-Warming-Stop

    Another go-round won’t help what credibility they have left with the general public. Thankfully.

    • That would be a perfectly logical conclusion if the general populace were perfectly logical. Sadly though, there is still a large contingent of people who get there ‘news’ from FOX News , blowhard AM radio hosts and rightwing bloggers. These people are virtually impervious to facts, and indisposed to admitting any errors in their world view.
      They assuredly will repeat the same blunder/lie about the AGW trend once the spike has reduced. Count on it. The only saving grace, is that those people are dying off little by little, but do not expect many minds to change.

      • There’s certainly truth in what you say; I’ve encountered quite a few specimens of the breed.

        However, I think that their numbers are smaller than one might think. Hope I’m right…

  2. Is the difference between the GISS 2014/2015 ENSO corrected data outside the .05C margin of error?

    • The error in the corrected dataset will be larger than the error in the original data due to uncertainty in the ENSO (etc.) coefficients. It’s probably not likely that the corrected 2015 is “significantly” larger than the corrected 2014, but it’s important to note that there is a difference between the model values and the “corrected” values. The corrected values are the “raw” minus the sum of the products of the exogenous variable coefficients with the corresponding variable values (properly lagged, so on). The purpose of doing this is to illustrate what the model says are the underlying trend and error components. However, one of the effects is to reduce variance between consecutive points. If 2015 in the corrected series is not warmer than 2014 in the corrected series, we should not be particularly surprised. Especially since the error inherently increases, the rate at which we would rate consecutive years in the corrected series as “not significantly different” is much higher than in the “raw” data. For these reasons I don’t think it’s even appropriate to do such a comparison.

  3. I’m quite impressed by the denoising you’ve managed here, as well as the fairly consistent trend change point in the mid-1970s it reveals. Does the correction account for anthropogenic aerosols as well?

    I suspect Gavin Schmidt’s ENSO correction was intended as a very quick and dirty adjustment to answer the predictable “but what about el Nino?” questions about the 2015 record.

    People have asked about the impact of ENSO. In the following figure I took the ONI index and found the lead with the greatest correlation to the monthly GISTEMP values (2 months), regressed that out, and recalculated the annual means.
    http://www.realclimate.org/index.php/archives/2016/01/2015-temperatures/

  4. The big question is whether carbon feedfack will take over, resulting in runaway global warming.

    Melting permafrost and methanogenic critters do not care whether the warming comes from El Nino.

    The heat in El Nino is heat in our climate system, and plus or minus 20 years, all the El Nino heat is in our weather system. El Nino attribution of heating is only of interest if your planning horizon is less then 18 years. If you own property, and/or have kids, or an interest in the world, your planning horizon should be long enough that El Nino attribution does not matter, because that heat is in the system and will affect weather within your planning horizon.

    Even in La Nina years, the heat is still accumulating in the system, and needs to be accounted.

    • Well, technically, El Nino actually *reduces* heating in the system, while La Nina *increases* it. That may seem counterintuitive, since Global Mean Surface Temp does rise during the former, and fall during the latter.

      But it actually makes sense: El Ninos raise atmospheric temps because a deep pool of warm water in the western Pacific gets spread out over a larger area, raising sea surface temperatures over a big chunk of the Pacific. The warm surface waters radiate more heat to the atmosphere and even directly to space, so there’s more net cooling going on than in ENSO-neutral or La Nina conditions. Conversely, of course, for La Nina. So, for long-term heating of the system, it’s the Nina that you need to be worrying about.

      At least, that’s the way I’d been thinking of it. There’s another explanation on the relevant NOAA page, which says that the reduced OLR during Ninos is mostly due to increased cloud coverage. (Oh, well.) That page is here:

      http://www.ncdc.noaa.gov/teleconnections/enso/indicators/olr/

      More detail on Science of Doom:

      http://scienceofdoom.com/2013/02/07/ceres-airs-outgoing-longwave-radiation-el-nino/

      • Trying to figure the basics of this out…

        SW in… how is it changed by the relaxation of surface winds? Does not seem like it should be other variation in TSI.

        In an El Nino, Eastern Pacific warms; Western Pacific cools eventually. LW out, should be slightly more, which means there would also be more LW back in… therefore, more evaporation.

        How does heat leave the oceans? It leaves from the thermal skin layer: sensible heat; latent heat; net LWR. In an El Nino the thermal skin layer is sitting on top of a column of water that is warmer because warm water has sloshed back from the Western Pacific to the Eastern Pacific… less wind.

        The skin layer is being warmed by SW from the sun, and it is not being aggressively blown to the Western Pacific. So during an El Nino this layer should get very warm just below the evaporation activity, where it would get cold. This sets up best circumstance for Minnett’s notion for how GHGs warm the oceans. Heat goes to cold. In the skin layer, heat from below runs into a very hot layer immediately below the evaporation layer. It moves through that very slowly. More SW held below; upward influence on OHC.

        What is the major way El Nino warms the atmosphere? The major one looks to be additional latent heat provided by evaporation which is being caused additional back radiation and the warmer water column.

        Seems like there would also be more sensible heat. Net radiation is higher.

        What does all of that do to OHC during an EL Nino? In 1998 OHC went down as the result of the El Nino (Trenberth). In many other cases it appears to me that OHC, on net, goes up during an EL Nino. Look at the period from 2000 to 2006: lots of El Nino activity, and OHC shot up. My conclusion seems to be wrong versus what is commonly written on the internet.

      • Ocean Heat Content estimates produced by the ex-NODC at NCEI show that OHC anomalies in each quarter of 2015 were the highest on record for each quarter. … – John Kennedy

  5. What’s the slope of the (last section of the) linear fit in each of those four fits? By eyeball it looks just a hair under 0.2 degC/decade, except the NCEI one looks closer to 0.17. Thanks!

  6. Tamino, where do you get your aerosol time series?

    [Response: I used the AOD data which is input to the NASA GISS climate model.]

  7. Very similar to the CSALT model at Context/Earth

    Here is a machine learning version
    http://contextearth.com/2015/01/30/csalt-re-analysis/

    Also have a multiple regression version which is the workhorse

    Agree that Gavin is being too conservative about how much ENSO and other features can compensate the variations in temperature.

  8. I don’t understand the cause of the apparent change in the warming trend at the end of the 1970s. Since you have removed most of the noise from the record, presumably leaving only CO2 concentration in the atmosphere, there seems to be no reason for such a sigificant change at that time. Surely if aerosols (often used to account for the upkick) were so significant, it would be sensible to have removed their impact as well?

    Is there any other comparable shift in the rate of warming in the record? Perhaps 1910 might bear some comparison?

  9. http://www.woodfortrees.org/plot/hadcrut4gl/mean:240/plot/esrl-co2/normalise/offset:0.5/plot/esrl-amo/mean:240

    Maybe answering my own question? Is there another non-greenhouse influence on atmospheric temperatures that hasn’t been fully accounted for in your analysis?

    [Response: Yes there is, but it’ not AMO. It’s primarily *anthropogenic* (rather than volcanic) aerosols.]

  10. johnsaintsmith,
    Elaborating on Tamino’s answer, the influence of anthropogenic aerosols on global temperatures was appreciated at the time. It is the origin of the stories the denialists love to trot out about scientists predicting an ice age. Yes, there were scientists (including Stephen Schneider) who thought we could be bringing on a mini ice age. This was because they underestimated the forcing due to CO2. So in actuality, the whole “scientists predicted an ice age back in the ’70s” supports a higher CO2 sensitivity–and greater alarm over our current situation. I love it when the denialati bring this up, since it is a very nice own goal.

    So what happened in the late ’70s? Clean air legislation and the introduction of the catalytic converter among other things. Catalytic converters are wrecked by sulfur compounds, so gasoline had to be reformulated to remove the sulfur. Acid rain became a concern–again sulfates were an issue.

    Also, aerosols probably still suppress warming to some extent, particularly due aerosols emitted by Asian economic tigers like China and India.

    • Would you care to offer some evidence to support the assertion that aerosols were reduced dramatically in the mid-seventies, in order to explain the rapid and sustained return to warming?
      Why hasn’t the massive increase in aerosols and dust in India and China had any similar effect on global warming – given that China is now burning more fossil fuel that the US?

      [Response: First things first: go read this.

      Based on the comments you submitted (which I deleted) pushing your attachment to AMO and touting your expertise by claiming you “have as clear an understanding of the impact of Greenhouse gas driven global warming as you have – probably much better,” you’ll then be back to argue with us about it. Perhaps you should consider that your understanding isn’t as clear or complete as you flatter yourself to think it is.]

  11. This is all excellent, and like the best scientific analysis it provokes many further questions in my mind. Some of which are:

    1. What percentage of the interannual variability is explained by the three factors: ENSO, volcanic and solar?

    2. What other factors might explain the remaining interannual variability? Could it be variations in NH spring snow cover? Cloud cover? Stratospheric water vapour?

    3. What does your analysis tell us about the contribution of these natural factors to recent global warming, and how that compare with the results from climate models when they are run for natural forcings only to the present day.

    4. If you were to apply this methodology to the climate model projections for the 21st century, would it say anything interesting about the influence of ENSO on future global surface temperatures?

    5. Could you apply this methodology to the Met Office decadal forecasts?