2011-12-07 16:48:37Huber and Knutti Quantify Man-Made Global Warming
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.137.110.252

Post on the Huber and Knutti paper.  I still need to use this to update a few rebuttals.

Huber and Knutti Quantify Man-Made Global Warming

2011-12-08 08:54:02
Jsquared

jfjanak@myfairpoint...
71.173.93.170

Odd how the discussion about this paper stopped over on the chat forum when you made up a blog post.  About the only thing I could suggest for inclusion in the post are some words to the effect that they are applying energy conservation locally - grid point by grid point - something that the modeling people haven't consistently done in the past, apparently, and it has to be true from the basic physics. 

It's impossible to apply such a local constraint without a computer, so it had to wait both until models were good enough and computers fast enough.  If you trust the model (and there's the rub, as usual), it seems definitive.

         

2011-12-08 10:30:32
Tom Curtis

t.r.curtis@gmail...
112.213.212.189

In the section on natural variability, you say that:

"What the authors have concluded is that natural variability can account for no more than 26% of the warming since 1950, and no more than 18% since 1850, and that it cannot account for the observed global warming"

But what the author's concluded was:

"For global surface temperature it is extremely unlikely (<5% probability) that internal variability contributed more than 26 +/- 12% and 18 +/- 9% to the observed trends over the last 50 and 100 years, respectively.  Even if models were found to underestimate internal variability by a factor of three, it is extremely unlikely that internal variability could produce a trend as large as observed."

As I read it, the <5% refers to the probability of error in attribution, while the error range (+/- sigma) represents measurement error of the size of the effect.  That still  means there is an approx 17% probability that natural internal variability accounted for 38% (27%) of the trend based on this paper.  In other words, you have failed to allow for uncertainties, a point deniers will jump on to create a large distraction.

 

 It also appears to mean that there is only a 2.5% probability that natural internal variability accounts for less than 2% (0%) of the trend.  That makes for an apparent discreprancy between the stated values and the grey shaded area on the graph.  I would interpret that as indicating that the grey shaded area shows the range of natural variability, but that in the 2000's natural variability tended to the upper end of the range (frequent El Nino's) thus contributing to the trend.  If this is correct, it is certainly not clear in the text of our article.  If not, I do not know how that discreprancy is properly interpreted.

I am very uncertain of my interpretations on these points.  Never-the-less, I think we need to put a flag on this section and make sure we have not made an error before publication.  Possibly an email to Huber is in order. 

 

2011-12-08 12:39:14
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.137.110.252

Agreed on the first point Tom - I had a similar thought.  Why 17% probability and not 5% though?

I'm not following this bit:

"It also appears to mean that there is only a 2.5% probability that natural internal variability accounts for less than 2% (0%) of the trend."

Where does that come from?

2011-12-08 12:44:57Contributed how much?
muoncounter
Dan Friedman
dfriedman3@comcast...
76.30.158.238

This line is sure to raise some eyebrows:

"the authors find that greenhouse gases contributed 166% (120-215%) of the observed surface warming"

I see that's more or less a quote from the article and you go on to mention the negatives, but surely a word of explanation about that percentage is in order.

 

Also, in 'The Future' section, approximately is misspelled.

2011-12-08 12:52:38
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.137.110.252

Thanks muon - simple enough.

2011-12-08 13:43:51
Tom Curtis

t.r.curtis@gmail...
112.213.212.189

Dana, the 17% (which should be 16%) comes from the fact that only 68% of values fall within one sigma of the mean in a normally distributed curve.  That means 32% fall outside the mean, with half (16%) being more than 1 sigma above the mean, and half being more than 1 sigma below the mean.  The 17% comes from faulty memory.

 

The 2.5% is because (on a normally distributed curve) 95% of values fall within 2 sigma of the mean, leaving 5% outside the 2 sigma bound.  Of course, half (2.5%) lies above the bound, and half below the bound.

I have of course assumed a normally distributed curve, and if the curve is not normally distrubuted different values will apply.  I have also used approximations, with more accurate figures being given here.

2011-12-08 16:23:52
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.137.110.252

Well, my weak statistics background is showing here, but the way I read it, the 26 +/- 12% internal variability is the maximum value, so I don't think the 2.5% probability of less than 2% is right.  But I could certainly be wrong.

2011-12-08 20:15:35
Tom Curtis

t.r.curtis@gmail...
112.213.212.189

The text states:

 

"Our results show that it isextremely likely that at least 74% ([+/-]12%, 1[sigma] ) of the observed warming since 1950 was caused by radiative forcings, and less than 26% ([+/-]12%) by unforced internal variability."

I assume the 26% ([+/-]12%) represents the converse of the 74%, and hence that the error given is also 1 sigma.  I am unsure of the convention here, but the simplest convention would also be that errors are stated to the standard of the last stated error, on which basis it would also be 1 sigma.  

2011-12-09 03:17:53
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

If you look at Figure 2c in my post (their Figure 3c), the anthropogenic component is almost identical to the net warming.  So 74% is the minimum, ~100% is the most likely value (not sure where the +/- 12% fits in there).  Thus natural variaiblity is most likely 0%, with a maximum of 26%.

Maybe they're saying 74% is the mean (~98%) minus 2 sigma?

2011-12-09 09:06:16
Tom Curtis

t.r.curtis@gmail...
112.213.212.189

dana, at this stage I reiterate my first post, "I am very uncertain of my interpretations on these points".  However, it is certainly not straight forward that the text of your article is correct.   I strongly recommend a direct query to Huber to clarrify the issue. 

2011-12-09 09:58:24
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Okay.  For the record (so I can boast a bit if I get it right ;-), my interpretation is that the statement means that radiative forcings are extremely likely (95% probability) to be responsible for between 74% and 122% of the observed warming.  But I've just emailed Huber and Knutti with a link to the post and this specific question, so hopefully one of them will confirm.

2011-12-09 10:11:11
Rob Painting
Rob
paintingskeri@vodafone.co...
118.93.19.81

What if you get it wrong? Grovelling apology?

2011-12-09 10:27:31
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

No groveling.  Maybe a bit of grumbling ;-)

2011-12-10 03:12:28time to gloat
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.137.110.252

Got a response from Knutti:

Dear Dana,

Thanks for your message and for writing about our paper. I like the way you present it.

You interpretation of the “at least 74%” is correct. Our best estimate is that close to 100% is forced, but because the whole method is probabilistic, the forced component could be a bit smaller, with some variability contributing to the warming, but the forced part could also be larger than observed, with variability opposing it.  More technically it’s saying that 95% of the PDF is above 0.74 times the observed warming. A few news articles didn’t get this unfortunately. But it’s not simple to explain these things: every number has an uncertainty, and then additionally we need to specify the confidence level

And now for the gloating - who da man?  :-)

So I'm going to update several rebuttals that this paper applies to (natural variaiblity, co2 effect is weak, and it's the sun come to mind).  Then maybe post later today.

2011-12-10 04:50:48
Sphaerica

Bob@Lacatena...
76.28.5.93

Did they explain what the +/-12% was?  That was the only thing I'm really unsure of.  The number of layers of uncertainty are unsettling.  it's almost like saying "34% +/2% +/- 8˚C +/-3% +/-12 hamsters... give or take").

Good post, by the way, although it did take me a couple of gear-turns to realize that numbers over %100 meant that natural forcings wre negative and anthro forcings were compensating for that.

2011-12-10 05:19:22Where's the Introduction???
John Hartz
John Hartz
john.hartz@hotmail...
98.122.98.161

Your draft lacks a power-point introduction that tells readers what the article is all about and entices them to read it. In addition, people who either do not have the time, or do not wnat to take the time, to read the entire article, benefit from having the "take-ways" embedded in the opening paragpraph.

2011-12-10 05:30:45
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.137.110.252

Sphaerica - Knutti confirmed that my interpretation of the 12% is accurate.  Namely it's one sigma, so two sigma is 24%, meaning the forcings can be more simply described as 98% +/- 24% at the 95% confidence level (two sigma):

the statement means that radiative forcings are extremely likely (95% probability) to be responsible for between 74% and 122% of the observed warming.

Their 74% (+/- 12%) minimum meant that 74% was two sigma (24%) below the most likely value.

I'll try to clarify the over 100% bit.  It's actually that aerosols are negative - natural forcings are very slightly positive.

John, I'll try to add a bit of an intro too.

2011-12-10 05:46:23
Albatross
Julian Brimelow
stomatalaperture@gmail...
23.17.186.57

Hi Dana,

Thanks for contacting Knutti.  This is good.  Glad that is all sorted out.  I would offer to do my usual review, but it has been a bad week at work and otherwise. Maybe I'll have a quick read this afternoon.

2011-12-10 06:10:05
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.137.110.252

No worries Alby.  I just published the post, but if you review it and see any changes that should be made, I can still make them.

2011-12-10 06:20:06
Rob Painting
Rob
paintingskeri@vodafone.co...
118.92.80.66

Nothing wrong with gloating. Now you have to rub it in Tom's face! (just kidding). 

2011-12-10 11:21:11
Tom Curtis

t.r.curtis@gmail...
112.213.212.189

Dana, you da man!  I grovel before your awesomeness.

 

Or more seriously, good job on the post, and glad we got the correct answer from the source.

2011-12-10 12:31:06
dana1981
Dana Nuccitelli
dana1981@yahoo...
71.137.110.252

Heh thanks Tom.  It was a confusing point, and a good idea to contact the authors about it.  Plus we got the thumbs-up from Knutti, which was cool.  And he actually suggested a few little changes (he read it carefully enough to notice three typos).  And we got an exclusive quote from Knutti explaining the 74% figure.  So it all worked out well.

2011-12-10 14:54:54
thingsbreak

things.break@gmail...
98.204.66.145

Contacting one or more of a paper's authors seems like good practice. Good work!