2011-07-31 10:36:32Loehle and Scafetta Play Spencer's Curve Fitting Game
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.106.190

Post about the L&S paper is ready for review.  In particular, make sure my second paragraph (about how real climate modeling is done) is accurate.

Loehle and Scafetta Play Spencer's Curve Fitting Game

2011-07-31 11:43:52
grypo

gryposaurus@gmail...
108.34.130.66

Great work!

 

"The only way we will prevent man-made warming from accelerating is if we take significant action to reduce our greenhouse gas emissions, which "skeptics" like L&S oppose. "

 

You should either source this or change it to "which most skeptics oppose" or the entire thread will turn into accusations of false accuasations.  I know Loehle opposes it, but i have no source for it.  Scafetta, I have no idea about.

2011-07-31 11:47:10
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.106.190

Thanks grypo, I thought about that (I did say "skeptics like L&S").  I'll see if I can find any specific statements from the two on the subject though.

2011-07-31 19:23:54
Rob Painting
Rob
paintingskeri@vodafone.co...
118.92.78.28

What year does your Moberg reconstruction finish? My brain was melting trying to read that paper, but from what I recall they suggest volcano aerosols affect the climate, but man-made ones don't.  

Other than that looks good.

2011-08-01 01:43:12
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.106.190
Moberg ends in 1979 (and starts in 1). I'll have a look at their paper again - in the quote I provided, they didn't specify the source of aerosols. I don't see why it should really make a difference. Sulfates are sulfates'
2011-08-01 01:55:42
Rob Painting
Rob
paintingskeri@vodafone.co...
118.93.252.69

I don't know that you should cover it, just pointing out the inconsistency in their so-called reasoning about aerosols. It's schizophrenic.

As for the Moberg graph is just looks like th MWP was warmer in your version.         

2011-08-01 04:30:00
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.106.190

Moberg should be the same.  I just took the data from NOAA's page.  The peak MWP anomaly is right around zero in mine and here, for example.

2011-08-01 06:21:49
Alex C

coultera@umich...
67.149.101.148

>>>two "skeptics", Loehle and Scafetta (L&S), perform the same sort of curve fitting using a simple model. [comma after "(L&S)"]

>>> In this case, L&S use a model with a very simple formula:

Uh, the formula is not showing up for me (Safari glitch?). - EDIT: nevermind, came up after I refreshed... weird.

 

>>>Because of the linear term, you might expect the model to match the observations back to the Little Ice Age (LIA) and then rapidly diverge from observations.  If so, you would be right (Figure 1).

Actually I see good agreement during the instrumental period and then terrible agreement during the rest before, even before it diverges.  The cycle is very often anti-phase with the Moberg reconstruction, the peaks don't match up at all.  There's a bit of (lucky) match-up at ~1750-1780, but before and after it's pretty bad.  Can you graph the differnece between the two (the model and the reconstruction) over the 1500-1850 time period, or blow up that portion?  I'm interested to see what the pattern would be like.

2011-08-01 07:40:17
Rob Painting
Rob
paintingskeri@vodafone.co...
118.92.93.45

Dana, re - Moberg, right you are. It was difficult to discern in the graph linked to on the NOAA page.

2011-08-01 08:35:57
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.106.190

Alex - a very good point.  I meant that the trend was pretty close back to the LIA, but you're right that the 60 year cycle in the model doesn't match the Moberg reconstruction worth a damn.  Considering that the 60 year cycle is the basis of the paper, that's an important point which I'll add.

Here's the data zoomed in at 1500-1900 (which I'll add to the post):

1500-1900

And here's the residual over that period:

residual

2011-08-01 08:52:35
Rob Painting
Rob
paintingskeri@vodafone.co...
118.92.93.45

More ouch! Great work guys.

2011-08-01 09:36:29
Andy S

skucea@telus...
66.183.179.249

I haven't had a chance to read the L&S paper in detail. After my quick read of Dana's rebuttal, I think it's really good. It must have been tough knowing what criticisms to cut out.

It's interesting that they essentially don't model the first half of the LIA or MWP at all, although they probably could if they added a 1000 year wavelength cycle term (not that that would make any more physical sense than what they have already done and it would just be wiggling the elephant's trunk). They also don't think the PDO or ENSO have any influence. Imagine the fuss in the denialosphere if an "alarmist" did that.

Why do they assume round-number periods for their cycles (20, 60 Earth years)? Is it because they need them to be harmonics and in a constant phase shift? Yet they calculate several significant figures for their coefficients? Since they speculate that these planetary cycles affect climate by acting on the sun, why don't they show up in observed sunspot cycles, which apparently have 11 and 22 year periods? I have to stop now, I'm getting a headache.

 


Just a small niggle about the tone of one passage, which might be a little alienating for some...

 

In the rebuttal, you say:

In this case, L&S use a model with a very simple formula:

You can probably guess how their model will perform just by examining this formula.  

That's probably appropriate if you are addressing physics freshmen in a seminar but it might be assuming  too much expertise from most SkS readers. I know that you go on to explain things very clearly but it does imply a little bit that if the reader doesn't grasp this right away, then they shouldn't be reading this. I'm sure you didn't intend that at all.

2011-08-01 09:40:42Correlation versus causation
Andy S

skucea@telus...
66.183.179.249

One of my favourite xkcd cartoons

 

2011-08-01 12:05:06
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.106.190

Fair point Andy, I'll change the phrasing from "probably" to "may be able to".

2011-08-01 12:22:39
Alex C

coultera@umich...
67.149.101.148

Yeah, those residuals should be near-zero if the cycles they conjured up have such an effect on temperature as they propose.  Even if the residual showed a strong sinusoidal pattern, which could theoretically also convey a large lag-to-period ratio, that would still not suffice as they purposefully built the model to be coincident with the instrumental record (i.e. assumed no lag).

Actually, that being said, the model and the reconstruction seem very close to one another in behavior within the instrumental time frame shown at the end, just with the reconstruction following the model by what appears to be a consistent interval (off by, mm, 10 years?), are you sure you set up the time periods correctly for both?  I would think they'd be much closer over 1850-1900.

Thanks Dana, post looks good to me.

2011-08-01 12:31:54
Alex C

coultera@umich...
67.149.101.148

One more thing though, you have this section "Why Does the L&S Model Fail the Hindcasting Test?" which starts to delve into the issue of whether or not the cycles even have an effect, but leave a bit of a hole between that and the later discussion of astronomical cycles where you reference Riccardo's post.  It doesn't flow well, perhaps an additive at the end of the above section like "(more on this below)."

2011-08-01 13:42:10
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.106.190

I set them up correctly.  The reconstruction actually doesn't match the instrumental temperatures all that closely in terms of year-to-year or decade-to-decade variations, which isn't that surprising.  But still, if there were a strong 60 year cycle in there, you should be able to see some evidence of it.

I added the "more on this below" suggestion.

2011-08-01 15:34:49
Alex C

coultera@umich...
67.149.101.148

OK, gotcha.

2011-08-01 15:43:57
Albatross
Julian Brimelow
stomatalaperture@gmail...
199.126.232.206

What would be really amusing is if L&S's "model" hindcast didn't match Loehle's own paleo reconstruction?  Is there time to look into that Dana?

2011-08-02 01:36:59
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Heh that's a good point Albie.  I don't have Loehle's data though.  Anybody got it or know where to find it?  There's time if we can get the data - this one's not scheduled to publish until Wednesday US time.

2011-08-02 01:50:19
Alex C

coultera@umich...
67.149.101.148

Here's the data for his 2007 version's Figure 1.  I don't know if it's close enough to the most updated version to do a comparison.

Personally I question whether adding in the comparison to Loehle's data to the post would be appropriate or not.  To me it seems too... snarky?  It doesn't stick to the science, it's a poke in the ribs.  Adding it in along with comparisons to still yet other reconstructions could be an option (Ljungqvist, Mann 2008), as a "he can't even agree with his own data, let alone others'."

2011-08-02 02:04:51
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Thanks Alex, I think I found the data for his updated version (supplemental data zip file here).  I might just add it to the Moberg comparison figure.  Will play with it tonight.

You have to consider that Loehle's response to the '60 year cycle doesn't show up in Moberg' may very well be 'Moberg is wrong, my reconstruction is better'.  I think it would make sense to show both.  I don't want to show too many though, because I don't want it to get too cluttered.

2011-08-02 02:22:46
nealjking

nealjking@gmail...
91.33.112.100

Alex,

I don't think it's snarky at all to expect a scientist's views of the same phenomenon to be scientifically consistent: They're supposed to represent reality, and the last I checked there was still only one reality. Either that, or admit that one's views are still somewhat speculative.

2011-08-02 03:14:10
Alex C

coultera@umich...
67.149.101.148

All right, that's fine; and yeah, I can also understand the clutter issue.

2011-08-02 04:13:01
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Yeah when you think about it, the only two climate-related bits of research Loehle has done (as far as I'm aware) are his millennial temp reconstruction, and this modeling paper (coincidentally, I noticed Loehle claimed it was observationally-based and not modeling, but I don't know what else you call representing climate changes numerically if not modeling). 

You would (and should) expect his two forays into climate science to be consistent with eachother.  If they're not, then as neal says, he's not presenting a consistent view of reality.

2011-08-02 09:19:09
Albatross
Julian Brimelow
stomatalaperture@gmail...
199.126.232.206

Yes, the point of me suggesting that is to demonstrate that Loehle is contradicting himself.  The incoherence and inconsistency of the " skeptic"/denialists continues.

Dana thanks for accommodating my request!

2011-08-02 12:39:53
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.106.190

Well thanks for the good idea Alby.  So now the question is, does the zoomed in version look to cluttered with both Moberg and Loehle?

comparison

Alternatively, I could just show the comparison to Loehle, and then either link to or show the comparison to just Moberg.

LS vs Loehle

That way it's clearer that there's no 60 year cycle in Loehle's data.  Thoughts?

2011-08-02 13:03:47
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.106.190

I tried sticking them both in there seperately.  Have a look and let me know what you think.

2011-08-02 13:28:11
Albatross
Julian Brimelow
stomatalaperture@gmail...
199.126.232.206

Yes, that first one does look cluttered...sigh.  Better to show L&S versus Loehle and calc. the residuals for those data, and also just sow Moberg for comparison separately.

 

2011-08-02 14:17:29
Andy S

skucea@telus...
66.183.179.249
That second chart is very clear. This is turning out to be quite a devastating rebuttal. You are providing a masterclass in auditing for McIntyre. This should make quite a splash!
2011-08-02 14:38:56
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.106.190

Thanks Andy.  L&S made it easy by giving me so much material to work with.  Really it's an embarassment that this paper was even published.  I think we can add Bentham Open Atmospheric Science to the growing list of joke journals that will publish bad "skeptic" papers.  For a nominal fee, of course.

2011-08-02 14:49:55
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.106.190

I think next I'm going to do a more mainstream article about the curve fitting exercises "skeptics" like Spencer and L&S keep playing, and how they (and certain media outlets like Forbes and Fox News) draw totally unjustified conclusions from them.  Maybe we can get it into the Guardian, since it's relevant to current events.

2011-08-02 15:15:46
Albatross
Julian Brimelow
stomatalaperture@gmail...
199.126.232.206

Dana, fantastic job. Kudos!

At one point you note:

"And as we saw in Figures 2 and 3, their assumed 60 year cycle does not show up in millennial temperature reconstructions - not even in Loehle's own previous work!  This is a rather glaring contradiction. "

I would consider reinforcing/reiterating that important point in the concluding section-- it is a fairly long post and people may not read everything and miss it.

Re the Guardian--yes, the more exposure this can get the better.

2011-08-02 16:12:48
Rob Painting
Rob
paintingskeri@vodafone.co...
118.92.85.150

Yeah, Dana just gave Loehle and Scarfetta an atomic wedgie!

2011-08-03 03:52:25
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Hah Rob :-)

Alby - yeah I thought about putting that in the conclusion, but it didn't really fit.  I'll think about it some more though and try to squeeze it in there.  It's definitely worth mentioning in the conclusion, I just need to figure out a way to put it in there without messing up the flow and focus on the main issues.

2011-08-03 05:49:27
Albatross
Julian Brimelow
stomatalaperture@gmail...
199.126.232.206

Dana,

The revised concluson looks good-- I think that works.  When does this go live?

2011-08-03 06:12:57
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Thanks Alby, I think I found a good spot for it.  I'll leave Trenberth/Fasullo in the spotlight today, and post this one tomorrow morning.  Nicely related series of posts - Bickmore on Spencer's curve fitting Monday, then Trenberth/Fasullo on Spencer's curve fitting Tuesday, then mine on L&S's curve fitting Wednesday.

2011-08-03 07:02:42
Albatross
Julian Brimelow
stomatalaperture@gmail...
199.126.232.206

Thanks Dana. In cricket terms the good guys will have scored a hat-trick.