Leading Paleoclimatologist: Prepare for the Next Ice Age

[skierinvermont]

Well, let's assume about a .4C total expected rise from 2000 to 2020 (completely in line with IPCC). We are on pace to see maybe .2C (ENSO-adjusted). So if that happens, seeing just 50% of the expected warming over 20 years is a significant issue, I think.

Well .1C/decade 2000-2020 would probably still be more than 25% of the ensemble members. Over 25 or 30 years it probably becomes invalidation. At some point the warming's got to pick up though, and I think it will.

There are other ways of calculating confidence intervals and I'm not entirely sure which is best.

[skierinvermont]

IPCC graphs don't exlude ENSO, because it cannot be done. AGW is expected to create a +ENSO feedback anyway, which is why it is pointless to attempt and incorrectly remove ENSO.

We have been rapidly warming at .2C/decade over the last 12 years

[nzucker]

No not at all.. I think negative-neutral next year is most likely which would probably put us around +.1+.2 on UAH for 2012.

Most of the IRI models seem to be showing something in the ballpark of negative-neutral, and this La Niña seems to be in no hurry to die with Region 3.4 remaining at -0.8C. I wonder if we'll see a slightly colder version of 07-08 where the Niña faded quite a bit, we came up to positive-neutral over the summer, and then the La Niña made a comeback for 08-09 with a weak cold ENSO event. Since we're starting a bit lower this year and with a more entrenched -PDO, in addition to lower solar activity, I wouldn't be surprised if we saw the same pattern as 07-08/08-09 but with everything being shifted colder.

IPCC graphs don't exlude ENSO, because it cannot be done. AGW is expected to create a +ENSO feedback anyway, which is why it is pointless to attempt and incorrectly remove ENSO.

I think the exclusion of ENSO is reasonably accurate statistically for figuring out short-term global temperature trends. GISS shows .12C/decade, UAH shows .08C/decade, and RSS shows .06C/decade for the last ten years or so. Obviously you can never remove any component of the climate system completely, since as you say everything is interwoven, so these are just estimates. ENSO oscillates in a fairly regular pattern so it's a good candidate for correction to eliminate problems with starting in a warm/cold year...obviously you're not going to get an exact temperature trend in this manner, but close enough to see how the IPCC models are doing. In this case, we're on the very low end of the confidence interval with 2011 likely to be a cold year.

[skierinvermont]

I think the exclusion of ENSO is reasonably accurate statistically for figuring out short-term global temperature trends. GISS shows .12C/decade, UAH shows .08C/decade, and RSS shows .06C/decade for the last ten years or so. Obviously you can never remove any component of the climate system completely, since as you say everything is interwoven, so these are just estimates. ENSO oscillates in a fairly regular pattern so it's a good candidate for correction to eliminate problems with starting in a warm/cold year...obviously you're not going to get an exact temperature trend in this manner, but close enough to see how the IPCC models are doing. In this case, we're on the very low end of the confidence interval with 2011 likely to be a cold year.

Explained perfectly. It's not perfect, but it's a good estimate.

[tacoman25]

Well .1C/decade 2000-2020 would probably still be more than 25% of the ensemble members. Over 25 or 30 years it probably becomes invalidation. At some point the warming's got to pick up though, and I think it will.

There are other ways of calculating confidence intervals and I'm not entirely sure which is best.

Maybe, but 20 years is a long enough time period that I think the ensemble mean becomes more meaningful. I mean, here we are at 10-13 years and close to falling outside of the confidence intervals.

[nzucker]

Explained perfectly. It's not perfect, but it's a good estimate.

Did I just hear you agree with me?

In any case, if we see a long multi-year Niña event that stretches into 2013, the IPCC models are going to start looking silly. I wouldn't be surprised if this happens given that we had a similar PDO switch in the late 1940s and a string of La Niñas including 48-49, 49-50, 54-55, 55-56, 56-57. It wasn't until 10 years after that switch that we finally got a strong El Niño for Winter 57-58, and even that was a low-end strong peak like 09-10, not in the style of mega Niños like 82-83 and 97-98. It's going to be hard to get that type of strong event in such a strong -PDO pattern, and that would be the only thing that gets us near IPCC estimates of warming. Just as Hansen's 1988 predictions were revised downward, I think the 2007 predictions will be revised downward to account for less climate sensitivty (a subject of many new papers) as well as the long solar minimum, which Landschedit seemed to think would last at least till the 2050s.

[skierinvermont]

Yeah, but 20 years is a long enough time period that I think the ensemble mean becomes more meaningful. I mean, here we are at just over 10 years and close to falling outside of the confidence intervals.

Well we're not close to falling outside the confidence interval if we use the ensemble spread to calculate the confidence intervals, which seems like a reasonable method to me.

Like I said, there are other methods though and I have a terrible memory so I'd have to go back and read them.

[BethesdaWX]

I think the exclusion of ENSO is reasonably accurate statistically for figuring out short-term global temperature trends. GISS shows .12C/decade, UAH shows .08C/decade, and RSS shows .06C/decade for the last ten years or so. Obviously you can never remove any component of the climate system completely, since as you say everything is interwoven, so these are just estimates. ENSO oscillates in a fairly regular pattern so it's a good candidate for correction to eliminate problems with starting in a warm/cold year...obviously you're not going to get an exact temperature trend in this manner, but close enough to see how the IPCC models are doing. In this case, we're on the very low end of the confidence interval with 2011 likely to be a cold year.

I think the point was AGW is supposed to create a Positive ENSO feedback, so really removing it makes no sense...........which is why IPCC includes it.

In Major ENSO events, it can probably be done. In weak events, there are just so many factors invloved its quite hard.

The issue is, GCC, HLB, IOD/AMO/PDO....will all alter trends in their own ways. GCC is definitely the largest, IOD/AMO combo and PDO cancel eachother out.

[tacoman25]

Well we're not close to falling outside the confidence interval if we use the ensemble spread to calculate the confidence intervals, which seems like a reasonable method to me.

Like I said, there are other methods though and I have a terrible memory so I'd have to go back and read them.

Um, didn't you just quote some reasonable numbers for UAH this year and next, and say those would put us outside 95%? I'd say we are dangerously close.

[skierinvermont]

I think the point was AGW is supposed to create a Positive ENSO feedback, so really removing it makes no sense...........which is why IPCC includes it.

In Major ENSO events, it can probably be done. In weak events, there are just so many factors invloved its quite hard.

The issue is, GCC, HLB, IOD/AMO/PDO....will all alter trends in their own ways. GCC is definitely the largest, IOD/AMO combo and PDO cancel eachother out.

The hypothesized ENSO positive trend due to AGW is like .01C/decade (if it even exists).. whereas by selecting 2002 as a starting point you are using a period which has an ENSO trend of -.2C/decade. Different orders of magnitude. You are selecting a period with a strongly negative ENSO tendency the effect of which can be reasonably removed statistically.

[WeatherRusty]

WTH is a "carbon feedback"???? If you are talking about "global temp. feedback" we have no history of such "feedbacks" getting out of control during "natural" temperature fluctuations over the last few millenia.....niether in a negative or positive role...

Please refer to Post #132 of this thread and look at the links for explanation and examples of carbon cycle feedbackl

[VAwxman]

Before removing ENSO we find large variation in the temperature trends to present. After removing ENSO we find that this variation is washed out and the trends to present are much smoother. This is an indication that we are adjusting for a real phenomenon. When we perform a correlation analysis we find that ENSO explains X% of the variation in temperature trends. We can then remove that X% and see what remains. What remains is a much smoother consistent temperature trend.. indicating that we have adjusted for a real phenomenon. It is not perfect because there are other factors at play, and because each ENSO event might act slightly differently.

Also you raise the concern that some ENSO events are stronger, but in my formula stronger ENSO events are adjusted more. The formula used is created based off the correlation analysis and adjusts GISS temperature by .105C for every 1C of ONI. This is also a method used in peer-reviewed studies. We see that trends to present from 1998-2001 are adjusted downwards because the ENSO trend to present is positive, whereas 2002-2005 is adjusted upwards because the ENSO trend to present is negative. The result is a much more consistent trend that gradually falls from .18C/decade to .1C/decade before being taken over by even shorter term variation. It's not perfect, but it removes much of the variation based on ENSO.

Basically, Bethesda doesn't like this method because he can no longer claim that temperatures have been dropping since 2002, because the ENSO trend from 2002-present is strongly negative. It also prevents AGWers, for example, selecting trends to present starting in 1999 and claiming that we have been warming .2C/decade.

remember, this is a graph of trends to present

Thanks.

I gotta admit I am not comfortable with this, since ENSO events even with the same ONI will lead to different global temperature responses due to variations in PDO state, AO / NAO, etc. I fear those difference could skew the ENSO approximations enough in either direction to make attempts to remove it prove quite difficult and not really tell us what we need to know. And it would not take much. If the estimate for ENSO correction is +0.12 C per decade since 2000, and that is off by even as little as 0.05 in one direction or the other, that's notable, since your actual range then would hypothetically be +0.07 (falling off IPCC estimates) to +0.17C (well within their range).

Not arguing for or against anything to do with AGW here, just making a personal assessment of such methodology, which still doesn't yet touch on the impacts to the ENSO circulations that AGW in theory should have as well, making it even more difficult to remove.

[skierinvermont]

Um, didn't you just quote some reasonable numbers for UAH this year and next, and say those would put us outside 95%? I'd say we are dangerously close.

That was for the narrower version of confidence intervals which I was just remembering off the top of my head.

The wider version of the confidence intervals we're not close to falling out.

Like I said, I'm confused and we should read more to decide what the best method is. It's also possible to separate out the confidence intervals for the low sensitivity and high sensitivity ensembles, which could be a useful exercise.

[BethesdaWX]

The hypothesized ENSO positive trend due to AGW is like .01C/decade (if it even exists).. whereas by selecting 2002 as a starting point you are using a period which has an ENSO trend of -.2C/decade. Different orders of magnitude. You are selecting a period with a strongly negative ENSO tendency the effect of which can be reasonably removed statistically.

The IPCC's trendline includes ENSO with the +.2C/decade because El Nino is not only supposed to dominate, but the feedback will not support La Nina Cooling at significant levels as El Nino.

Thus, ENSO included trends in the IPCC are about +.2C/decade at this point....NO Need to remove it. You cannot do it regardless.

[nzucker]

The hypothesized ENSO positive trend due to AGW is like .01C/decade (if it even exists).. whereas by selecting 2002 as a starting point you are using a period which has an ENSO trend of -.2C/decade. Different orders of magnitude. You are selecting a period with a strongly negative ENSO tendency the effect of which can be reasonably removed statistically.

Different IPCC climate models have come out with different predictions for how ENSO will change with global warming; there isn't one definitive answer because eventually the deep ocean will warm to match the surface more closely, and this will reduce any potential gradient caused by climate change. Some theorize that El Niño will become more common at the onset of rapid temperature increases, but then level off as the deep ocean absorbs the extra heat. So I don't think we can say AGW and ENSO are really connected at this point. Certainly not to the extent that starting your trend line with a strong Niño or Niña influences the resulting global temperature.

I think it's reasonable to expect a bit more La Niña in the 2010s and 2020s, which probably argues for a slightly cooler scenario than what the IPCC depicts in terms of surface warming. Also, some of the GISS models expected a continuation of high solar activity, which hasn't been the case.

[BethesdaWX]

Thanks.

I gotta admit I am not comfortable with this, since ENSO events even with the same ONI will lead to different global temperature responses due to variations in PDO state, AO / NAO, etc. I fear those difference could skew the ENSO approximations enough in either direction to make attempts to remove it prove quite difficult and not really tell us what we need to know. And it would not take much. If the estimate for ENSO correction is +0.12 C per decade since 2000, and that is off by even as little as 0.05 in one direction or the other, that's notable, since your actual range then would hypothetically be +0.07 (falling off IPCC estimates) to +0.17C (well within their range).

Not arguing for or against anything to do with AGW here, just making a personal assessment of such methodology, which still doesn't yet touch on the impacts to the ENSO circulations that AGW in theory should have as well, making it even more difficult to remove.

Yes, Skier's GISS Graph starts showing wild spikes when the -NAO/-AO phase began once "ENSO corrected" with its Arctic Warm Bias, at least I think thats the culprit.

Is ENSO really at +0.12C since 2000?

[nzucker]

Here is what Anthony Watts shows for the temperature trend since 2002:

[skierinvermont]

Thanks.

I gotta admit I am not comfortable with this, since ENSO events even with the same ONI will lead to different global temperature responses due to variations in PDO state, AO / NAO, etc. I fear those difference could skew the ENSO approximations enough in either direction to make attempts to remove it prove quite difficult and not really tell us what we need to know. And it would not take much. If the estimate for ENSO correction is +0.12 C per decade since 2000, and that is off by even as little as 0.05 in one direction or the other, that's notable, since your actual range then would hypothetically be +0.07 (falling off IPCC estimates) to +0.17C (well within their range).

Not arguing for or against anything to do with AGW here, just making a personal assessment of such methodology, which still doesn't yet touch on the impacts to the ENSO circulations that AGW in theory should have as well, making it even more difficult to remove.

It's true that some ENSO events of the same strength would have larger effects than others. But since I am using the average effect derived from a 60 year training period, it is unlikely to be biased in a particular direction over an extended period.

Moreover, what we find is that ENSO-correcting doesn't change the average trend to present for raw data, it simply smooths out the variation in trends to present. So for example, instead of having +.2C/decade since 1999, and +0.0C/decade since 2002, we have a consistent trend of .1C/decade. It's not changing the average trend to present in the data, so it cannot be causing a persistent bias. It is simply smoothing out the data.

We don't even need to use all the complex statistics, we can just see from the raw uncorrected data that trends to present have oscillated from 0 to +.2C/decade over the last 15 years, and that the average trend to present is .1C/decade.

[BethesdaWX]

Here is what Anthony Watts shows for the temperature trend since 2002:

I wish those were updated.

[BethesdaWX]

The IPCC's trendline includes ENSO with the +.2C/decade because El Nino is not only supposed to dominate, but the feedback will not support La Nina Cooling at significant levels as El Nino.

Thus, ENSO included trends in the IPCC are about +.2C/decade at this point....NO Need to remove it. You cannot do it regardless.

Skier, you forget IPCC includes ENSO in their predictions...

[nzucker]

Here is what Watts shows for temperature trends if we simulate a multi-year Niña using 1999-2001 as the example:

We'd clearly have a downward trend since 2002 if we go into a multi-year La Niña....that would be an obstacle for the IPCC and its climate models.

[skierinvermont]

Different IPCC climate models have come out with different predictions for how ENSO will change with global warming; there isn't one definitive answer because eventually the deep ocean will warm to match the surface more closely, and this will reduce any potential gradient caused by climate change. Some theorize that El Niño will become more common at the onset of rapid temperature increases, but then level off as the deep ocean absorbs the extra heat. So I don't think we can say AGW and ENSO are really connected at this point. Certainly not to the extent that starting your trend line with a strong Niño or Niña influences the resulting global temperature.

I think it's reasonable to expect a bit more La Niña in the 2010s and 2020s, which probably argues for a slightly cooler scenario than what the IPCC depicts in terms of surface warming. Also, some of the GISS models expected a continuation of high solar activity, which hasn't been the case.

Yeah which is why I said "if it even exists" .. the AGW-ENSO connection is just speculation at this point.

[VAwxman]

It's true that some ENSO events of the same strength would have larger effects than others. But since I am using the average effect derived from a 60 year training period, it is unlikely to be biased in a particular direction over an extended period.

Moreover, what we find is that ENSO-correcting doesn't change the average trend to present for raw data, it simply smooths out the variation in trends to present. So for example, instead of having +.2C/decade since 1999, and +0.0C/decade since 2002, we have a consistent trend of .1C/decade. It's not changing the average trend to present in the data, so it cannot be causing a persistent bias. It is simply smoothing out the data.

We don't even need to use all the complex statistics, we can just see from the raw uncorrected data that trends to present have oscillated from 0 to +.2C/decade over the last 15 years, and that the average trend to present is .1C/decade.

Yeah if you have a long enough dataset, that would help, no doubt. I just wonder if what we have is sufficient in that regard, or enough to capture ENSO events not only of different intensities, but in different PDO base states, etc too. Interesting stuff to ponder.

[skierinvermont]

Skier, you forget IPCC includes ENSO in their predictions...

They include ENSO variation.. they don't include decadal forecasts of what the ONI will be in 2020. When the ensemble members are averaged, the ENSO variation is washed out.

You are correct that we can't compare ENSO-corrected data to the ensemble members, because the ensemble members include simulated ENSO variation.

But if we are comparing to the ensemble member mean, then we need to correct for ENSO, because in the ensemble member mean the ENSO-variation is averaged out to zero.

So for example, when I compare the 1998-2011 temperature trends to the ensemble spread, I did NOT correct for ENSO.

When I compare the 1998-2011 temperature trends to the ensemble mean, then I correct for ENSO.

[BethesdaWX]

They include ENSO variation.. they don't include decadal forecasts of what the ONI will be in 2020. When the ensemble members are averaged, the ENSO variation is washed out.

You are correct that we can't compare ENSO-corrected data to the ensemble members, because the ensemble members include simulated ENSO variation.

But if we are comparing to the ensemble member mean, then we need to correct for ENSO, because in the ensemble member mean the ENSO-variation is averaged out to zero.

Yes but you'd need to take out much more than just ENSO is the problem. And when you do that, you start tampering with climactic feedbacks, and everything is thrown off.

Also, the ENSO trend is positive this past decade now that I look. So you'd need to correct the trend more negative.

[LakeEffectKing]

Please refer to Post #132 of this thread and look at the links for explanation and examples of carbon cycle feedbackl

All are "thermally" driven feedbacks......using the term "carbon feedbacks" implies a direct feedback specifically related to carbon!

Most of the links you posted are hypothesized "thermal feedbacks"....and in pre-industrial times, there is no such evidence of such feedbacks governing the climate in some out of control fashion with natural (sometimes more pronounced) warmings or coolings, nor even a measurable contribution for that matter. You think wildfires were held more in check some 1000 years ago, somehow migating any potential "wildfire" feedbacks that man can supposedly induce today by his CO2 machines???? And what about the MWP and the "bog permafrost" feedback? Any evidence of such?..... Refer to my post yesterday here: Skeptical of Feedbacks....Negative and Positive

[skierinvermont]

Yes but you'd need to take out much more than just ENSO is the problem. And when you do that, you start tampering with climactic feedbacks, and everything is thrown off.

Its a mess. Also, the ENSO trend is positive this past decade now that I look.

As zucker explained, the ENSO effect is fairly consistent and predictable and extremely large in magnitude. Removing it represents a reasonable approximation.

The ENSO trend is positive if we start in late 1998-2000. It is negative if we start in 2001-2005. It's fairly neutral if we start in early 1998 or earlier.

[BethesdaWX]

As zucker explained, the ENSO effect is fairly consistent and predictable and extremely large in magnitude. Removing it represents a reasonable approximation.

This response has nothing to do with my post

you'd need to take out More than just ENSO is the problem. And when you do that, you start tampering with climactic feedbacks, and everything is thrown off.

Also, the ENSO trend is positive this past decade now that I look. So you'd need to correct the trend more negative.

was my quote. I dont see how your response is at all related.

[skierinvermont]

This response has nothing to do with my post

you'd need to take out More than just ENSO is the problem. And when you do that, you start tampering with climactic feedbacks, and everything is thrown off.

Also, the ENSO trend is positive this past decade now that I look. So you'd need to correct the trend more negative.

The ENSO trend is positive if we start in late 1998-2000. It is negative if we start in 2001-2005. It's fairly neutral if we start in early 1998 or earlier.

The ENSO trend IS corrected more negative for start dates between late 1998 and 2000.

Look at the Graph. I've posted it a half dozen times. Slow down and think.

[nzucker]

This response has nothing to do with my post

you'd need to take out More than just ENSO is the problem. And when you do that, you start tampering with climactic feedbacks, and everything is thrown off.

Also, the ENSO trend is positive this past decade now that I look. So you'd need to correct the trend more negative.

was my quote. I dont see how your response is at all related.

He did correct the temperatures to account for the slightly positive ENSO trend; this is mostly because the multi-year El Niño between 2002-05 would have pushed temperatures higher than in a neutral ENSO environment. All of the graphs skier has posted are ENSO-corrected, and so are his comparisons to the IPCC mean. ENSO corrected shows about .1C/decade on the surface measurements and .07C/decade on the satellites.

I think it's fair to take out large events like ENSO and volcanic activity to try to approximate the warming trend/climate sensitivity. It's never going to be perfect because ENSO is linked to GCC, PDO, AO etc...but the fact that ENSO averages out to zero in the corrected data means these changes in indices should also balance out.