2011-10-26 10:44:55Eschenbach's BEST?
Rob Honeycutt

robhon@mac...
98.207.62.223

Starting up a new post titled Eschenbach's BEST? in response to his post at WUWT.  Working with Dana and Glenn on this one.

2011-10-26 15:50:22
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Here are some of the Eschenbach errors we need to cover (as discussed here):

1) failing to apply a land mask to the GISS data

2) failing to note that HadCRUT is known to be biased low

3) mistaking the amplication factor between surface and lower troposphere (it's 1.2 globally, but 0.95 over land)

4) assuming that any discrepancy must be a problem with the surface data (and assuming that problem must be UHI) rather than considering other possibilities (i.e. that UAH and RSS are low, as other analyses show, as Glenn will discuss)

5) using a very odd (and short) baseline period, which may mess with the results (we'll see when we do it properly)

2011-10-26 17:56:52
Glenn Tamblyn

glenn@thefoodgallery.com...
124.183.154.151

Guys. I have a very rough draft of some stuff on the sat side of thing. Currently in Word, waiting to upload it to SkS. Unfortunately dometic duties call - if I don't make serious inroads into the forest we call our lawn I may never find our dog again. Will upload rough version within 24 hours. Then we can play cut/paste/edit. And in my usual prolix fashion, precis.

2011-10-26 20:13:18
Kevin C

cowtan@ysbl.york.ac...
144.32.72.165

I want to get the landmask stuff right. Need to check GHCN versions, confirm that my unmasked CCC run gives the same result as GISTEMP, deal with the high latitude mask errors, and write bulletproof explanation of what I've done.

Can we host supplemental data in an SkS article? e.g. provide CCC input and output files so that people can repeat the calculation?

What is the best way to pass files around? I can put them on my webserver, but will need to post a link. Is there any sort of personal message system here so I can notify people who might be interested (although I think that's mainly Dana).

2011-10-27 00:53:39
Kevin C

cowtan@ysbl.york.ac...
144.32.72.165

OK, I've run lots of checks and am happy. Pick up:

http://www.ysbl.york.ac.uk/~cowtan/kevin/ccc-masked-monthly.zip

This contains a csv file (with initial headers to drop, then multi-space delimited items containing:

    Year Month GISTEMP-LAND CCC-LAND CCC-MASK1 CCC-MASK2 CCC-MASK3

The csv header describes the data and gives instructions for regenerating it. There are also three mask files used for 3 different runs.

The GISTEMP-LAND and CCC-LAND columns are there as a check that CCC is giving he same results as gistemp (land index). Most months are identical, a few differ by 0.01C, one month in 1906 differs by 0.02. That's about the accuracy expected.

Then there are the properly masked land temperatures.

MASK1 uses the mask based on the ocean temp data, which marks the arctic ice cap as land: http://clearclimatecode.org/masking-gistemp/
This overestimates the trend because of arctic amplification.

MASK2 uses a proper land mask, but this omits some islands: http://clearclimatecode.org/a-real-land-mask/

MASK3 is my own combined mask using MASK1 for low lattitudes and MASK2 for high lattitudes (|l|>=60)

I would recommend MASK3, as it is more defencible on the basis of the CCC posts, and seems to fall between the other two. But the differences between the different masks are pretty small.

It would be nice to provide the files as supplementary data. If SkS can't host it, an external file drop would do. I'd rather not distribute it publically from my site.

The only other thing to do is to write a little supporting text, probably for a footnote at the bottom. This will depend on which data you use.

2011-10-27 02:35:32
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Can you hook me up with the data in an Excel file, Kevin?  dana1981@yahoo.com

I think SkS can host it if you go to the Author Admin and Upload Graphic.  The files on that page don't necessarily have to be graphics, and I've uploaded PDFs there before.

2011-10-27 02:55:35
Kevin C

cowtan@ysbl.york.ac...
144.32.72.165

Sent. Excellent news on hosting data.

2011-10-27 08:03:37
Andy S

skucea@telus...
66.183.185.188

Out of curiosity, I downloaded Kevin's data, calculated one-year moving averages and plotted out his MASK3 alongside the GISS land data for comparison. See here. I included the annual GISS land and ocean data, just for reference.

I also calculated the difference beween the two land data sets and plotted the "Delta" and fitted a fifth-order polynomial to it, just to illustrate the trend of the difference. See here

2011-10-27 09:19:46
Andy S

skucea@telus...
66.183.185.188

I did a quick check on using different baselines for the GISS-Land data, Kevin's MASK3 data and the RSS LT land-only data. All three are smoothed monthly data (one year running means) but I calculated the baselines on unsmoothed data. I used Willis's short 1979-1984 baseline and a longer baseline, 1980-2009. Graphs here.

Obviously there's a DC shift of about 0.2 degrees between the two plots but there are only minor relative shifts between the three datasets. Relative to the GISS-Land, the MASK3 line shifts down by about 0.05 of a degree, and the RSS slips down by 0.02 of a degree, using the longer baseline. So, based on this limited analysis, I don't think that Willis's choice of short baseline will make a big difference.

2011-10-27 10:25:29
Albatross
Julian Brimelow
stomatalaperture@gmail...
198.53.65.169

Andy,

Thanks good job.  So it looks like the baseline (surprisingly) does not make much difference, but he seems to have screwed up the land mask part, b/c your graphs show excellent agreement between gistemp land and RSS land.  Can you add the BEST data?

2011-10-27 12:28:06
Andy S

skucea@telus...
66.183.185.188

OK I added the BEST data.

Here's a comparison with BEST, Kevin's MASK3 and the GISS-Land only. No baseline corrections at all, annual smoothing. Nothing surprising here.

Here's the baseline comparisons. Now, the baseline change from short to long did make a difference. Relative to the GISS Land, the BEST data shift down by 0.12 degrees, which is appreciable, bringing the curves closer together at the right-hand side of the graph, where the eye is drawn. So maybe the baselining is important, after all.

By the way, I'm getting a little beyond my pay grade here, technically speaking. Please, nobody quote these results. I'm just indulging in a learning exercise.

2011-10-27 13:11:48
Daniel Bailey
Daniel Bailey
yooper49855@hotmail...
97.83.150.37

Dunno if you guys have noticed, but Paul Clark over at Woodfortrees has added BEST to the available options:

http://www.woodfortrees.org/plot/best/mean:60/plot/best/trend

BEST

 

http://www.woodfortrees.org/graph/best/mean:60/plot/best/trend/plot/gistemp/mean:60/plot/gistemp/trend

B+G

2011-10-27 15:20:58graph
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Here's what I get using 1981 to 2010 as the baseline, and Kevin's land mask C for GISS:

land-only

Quite different from Eschenbach's.  Note that I used a 12-month running mean for the satellites vs. annual anomalies for the surface stations.  Didn't have the time to take annual averages for the satellite data.

There is a trend discrepancy between surface and satellites.  The trends in °C per decade are:

BEST: 0.29

NOAA: 0.29

GISS: 0.24

HadCRU: 0.22

UAH: 0.18

RSS: 0.20

The smallest amplification factor is UAH/NOAA = 0.61, which is pretty far off from the model-expected 0.95.  But that certainly doesn't mean that BEST and NOAA fail to remove the UHI effect, which was Eschenbach's absurd conclusion.

2011-10-27 15:29:16Satellite Notes
Glenn Tamblyn

glenn@thefoodgallery.com...
58.168.75.220

All here is rough draft of the satellite iissues with Willis's analysis. Note that my values are global, not land only. Suggest copy @ paste into Rob's post then glue them up. I may have been more detailed than needed, my idea was to write it so that it would be informative for you guys. You can then precis as needed to fit size & tone requirements. Not sure about the quality of the graph of the STAR trands, might need some work.

2011-10-27 15:43:56
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Hmmm you've got enough material there for an entire post Glenn :-)  Maybe we could use part of it or summarize it, then link to that page for further details.  Either that, or turn this into a two-part post.

2011-10-27 17:24:54
Glenn Tamblyn

glenn@thefoodgallery.com...
138.217.126.181

That was my point Dana. If you guys feel it can be edited down into a single post, go for it. It is probably better if an independent eye makes that assessment. Will two posts have the same impact as a single one? And the difference between RSS/UAH and the other 3 assessments has a big bearing on the magnification factor question.

2011-10-27 20:16:20
Kevin C

cowtan@ysbl.york.ac...
144.32.72.165

Good work, Andy. Those graphs look like what we would have expected: Not using a land mask supresses the trend a bit because the coastal stations are being upweighted to cover a 1200km band of ocean around the continents. It also leads to less short term variability, which is also what you would expect due to the greater heat capacity of the oceans. It's good to have another sanity check.

The dip in difference at the beginning is presumably due to the small number of stations.

How do we caption that line? We want to say that it is what GISS would be if landmasked, but not claim that it is a GISS output. How about:

CCC(GISS) land only

or

CCC/GISS land only

2011-10-27 21:57:08
Kevin C

cowtan@ysbl.york.ac...
144.32.72.165
2011-10-27 22:03:50Descriptive text, use/modify as required...
Kevin C

cowtan@ysbl.york.ac...
144.32.72.165

<h4>Long description:</h4>

<p>A problem arises when comparing the BEST temperature results with those from GISTEMP: BEST provides a land-only temperature estimate, however the nearest equivalent dataset from GISS is a <i>global</i> estimate based on land stations alone. This will show reduced tend and variability when compared to the BEST data because the more temperate coastal stations are upweighted to provide estimates for the unsampled ocean regions. The problem is discussed in the <a href=http://berkeleyearth.org/Resources/Berkeley_Earth_Averaging_Process>BEST paper</a> at the top of page 31, and in more detail at the end of <a href=http://rankexploits.com/musings/2010/the-great-gistemp-mystery/>this article</a> by Zeke Hausfather. This issue has been addressed in <a href=http://clearclimatecode.org/>Clear Climate Code</a>'s reimplementation of the GISTEMP algorithm by allowing the application of a land mask. In order to determine the results of a true land-only GISTEMP analysis, we therefore ran ccc-gistemp version 0.6.1 with land stations only and verified that the data matched the GISTEMP land-station estimates. We then repeated the analysis with 3 different land masks, the first based on data from land/sea station locations described <a href=http://clearclimatecode.org/masking-gistemp/>here</a>, the second based on the geographical dataset described <a href=http://clearclimatecode.org/a-real-land-mask/>here</a>, and the third using the station data at low latitudes (-60&lt;&lambda;&lt;60) and geographical data at high latitudes to address the issues noted with the two masks. We present the data from the third test, however the choice of mask makes little difference to the results.</p>

<p>The results of the land-masked calculation, the masks, and documentation on how they were generated are all available for download in <a href=>this file(ADD URL)</a>.</p>


<h4>Short description for captions etc:</h4>

<p>Effective GISTEMP land-only temperatures computed by running Clear Climate Code's implementation of the GISTEMP method with a land mask; see text/footnote for details.</p>

2011-10-28 03:40:52
Rob Honeycutt

robhon@mac...
98.207.62.223

Hi gang.  I'm back on this.  Was lost in bookkeeping duties the past couple of days.

Let me see if I can pull these together into a congruent post.

2011-10-28 03:52:37
Rob Honeycutt

robhon@mac...
98.207.62.223

I just read Glenn's post and I think we should run that as a seperate post.  We can call it "Eschenbach's BEST?  Part 1" and then run another part regarding Eschenbach's Fig 3 graph as Part 2.

How does that sound?

2011-10-28 05:08:39
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

I think I could summarize Glenn's post into a couple of paragraphs if we'd prefer to keep it all as one post.  But I don't have a big preference either way.

2011-10-28 05:15:28
Rob Honeycutt

robhon@mac...
98.207.62.223

Either way is fine with me.  Eschenbach may not deserve two full posts.

2011-10-28 06:39:20
Andy S

skucea@telus...
66.183.185.188

Here's a revised comparison between the various GISS temperature series. I changed the caption following Kevin's recommendation. The Land-Ocean data are annual. The other two are monthly, with a 12-month running mean applied.

I think this is OK for publication (although I can easily modify it more if required). I would recommend it be just linked to, not put as an inline graphic, since it needs to be big for the detail.

2011-10-28 08:08:32
Glenn Tamblyn

glenn@thefoodgallery.com...
58.164.83.15

Perhaps Dana, put in a short version of my stuff for a post on Eschenbach, then we can put my post up shortly afterwards as a follow-on, rather than a part 2,. Getting the graph of Weighting Functions in front of peoples eyeballs so they have a better understanding of what the sat trends really mean is a worthwhile exercise. You have enough in the other data to hit him anyway

2011-10-28 08:30:47
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

That would work, Glenn.

Also Singer has provided a quote that we should use in this post, in the comments to a Nature editorial.

"But unlike the land surface, the atmosphere has shown no warming trend, either over land or over ocean — according to satellites and independent data from weather balloons. This indicates to me that there is something very wrong with the land surface data. And did you know that climate models, run on super-computers, all insist that the atmosphere must warm faster than the surface? And so does theory."

2011-10-30 04:15:09
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Nevermind that last comment, Riccardo is addressing Singer is a separate post.

Rob H, what's the status of this post?

2011-10-31 11:14:37comment
Robert Way

robert_way19@hotmail...
142.162.25.5

I think this should go ahead soon. PS where do I find the data you're using for the comparison with land-only?

2011-10-31 13:11:55
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55
All of them are available from the source groups except GISS, which Kevin C applied a land mask to. If there are any in particular you can't find, I can get the links for you.
2011-10-31 13:25:31comment
Robert Way

robert_way19@hotmail...
142.162.25.5

I'm wondering where is the data we have for Kevin C's land mask?

2011-10-31 14:13:48
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Kevin emailed it to me, and he also pasted it into one of the Gen Chat threads, though it would take me a while to remember which one.  He was planning to upload it here, but I'm not sure if he has.

2011-10-31 14:17:45comment
Robert Way

robert_way19@hotmail...
142.162.25.5

Can you email me it dana?

2011-10-31 14:50:28
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Sent.

Rob H, what's happening with this one?  Don't make me commandeer it :-)

2011-10-31 22:43:08
Kevin C

cowtan@ysbl.york.ac...
144.32.72.165

Draft version here: http://www.ysbl.york.ac.uk/~cowtan/kevin/ccc-masked-monthly.zip

I'm going to tidy up the files a bit and make a README with as much detail as I have time for, slap a dedication to Charlie at the bottom (tee hee) and upload it. Hopefully in a few hours, otherwise tomorrow.

2011-10-31 23:07:23
Kevin C

cowtan@ysbl.york.ac...
144.32.72.165

OK, here you go:

http://www.skepticalscience.com/pics/ccc-masked-monthly.zip

2011-11-01 01:57:45comment
Robert Way

robert_way19@hotmail...
134.153.162.53

is this with GHCN v2 or v3?

2011-11-01 03:04:45
Kevin C

cowtan@ysbl.york.ac...
144.32.72.165

GHCN v2. I first checked the match of the CCC-land result to the current GISTEMP land result - it is exceedingly good - most months give the same number in hundredths of a degC, a few are out by 0.01, only one out of 1400 is out by 0.02. That was the level of agreement I was expecting having read the ccc website, so I infer GISTEMP is still using v2, despite one note on their website expressing an intention to switch a couple of months back.

I tried with v3 as a cross check but ccc crashed.

2011-11-01 07:31:08
Rob Honeycutt

robhon@mac...
98.207.62.223

Sorry Dana.  I've been falling down on this one (along with other work).  I have family in town until tomorrow.  And of course Pielke's been keeping me busy on the side.  I'm about done with him too.

2011-11-01 08:26:52
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

That's ok Rob.  Like I said, I can take over the writing if you don't mind.

Did you email Pielke that WUWT Muller cartoon?

2011-11-01 09:58:51
Rob Honeycutt

robhon@mac...
98.207.62.223

Dana...  I think you probably are going to have to take this one over.  As I'm reading through this stuff I'm realizing it's pretty well above my skill level.  I don't think I'm in a position to be explaining this whole thing to a broader audience.

2011-11-01 09:59:20
Rob Honeycutt

robhon@mac...
98.207.62.223

Oh, and yes I did send that link to Pielke but he didn't comment on it.

2011-11-01 10:19:53
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Ah that's a bummer. It probably didn't get past Pielke's perception filter, to use a geeky Dr. Who reference :-)

No problem, I'll work on this post tonight.  With any luck I'll get a draft up for review tonight, but we'll see.

2011-11-01 10:27:32comment
Robert Way

robert_way19@hotmail...
134.153.162.53

If you need any help Dana let me know

2011-11-01 14:00:56
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

I'd appreciate a review, Rob.

Eschenbach's BEST Shot at the Surface Temperature Record

2011-11-01 14:55:50
Daniel Bailey
Daniel Bailey
yooper49855@hotmail...
97.83.150.37

Eschenbach hits it with his BEST shot


Eschenbach gives it his BEST shot

 


 

Eschenbach channels the Cigarrete-Smoking-Man: BEST, the Surface Temperature Record and Plausible Deniability

(Or how I took up smoking and learned to love burning CO2)

2011-11-01 14:57:32
Daniel Bailey
Daniel Bailey
yooper49855@hotmail...
97.83.150.37

Apologies to Stanley Kubrick...and the X-Files...and to dana...

Dr Strangelove

 

Cancer Man

2011-11-01 16:10:53Comment
Robert Way

robert_way19@hotmail...
142.162.25.5

2011-11-01 16:13:16Comment
Robert Way

robert_way19@hotmail...
142.162.25.5

Maybe the previous image could be of some use somewhere?

Anyways I read the post and it looks great. Only minor thing I might mention is that Table-2 seems a bit screwy and it might be worth making a note like the following in this paragraph:

"Eschenbach then concocts a figure comparing land-only temperature data for satellites (UAH and RSS) and surface stations (BEST, GISS, NOAA, HadCRU) in an attempt to argue that the surface records are all biased by the urban heat island (UHI) effect (Figure 2)."

<To>

"Eschenbach then concocts a figure comparing what he (wrongly) called "land-only" temperature data for satellites (UAH and RSS) and surface stations (BEST, GISS, NOAA, CRUTemp) in an attempt to argue that the surface records are all biased by the urban heat island (UHI) effect (Figure 2)."

2011-11-01 17:50:56Looks good.
Glenn Tamblyn

glenn@thefoodgallery.com...
60.230.147.182

Dana, this looks good. I will go over my post in the next 48 hours to do any tidy-up or strengthening needed to relate it to this post. It may need to be posted at the same time - I'm not sure what the SkS system will do if a published post links to an unpublished post.

2011-11-01 22:34:52
Kevin C

cowtan@ysbl.york.ac...
144.32.72.165

You've lost the links on my footnote.

Otherwise, it's looking pretty good.

2011-11-02 02:50:14
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Rob W - my next post is going to be on short-term "pauses" and how the underlying trend remains.  I was thinking of diong a very similar graph to the one you've got there.  Mind if I use it?

Whoops, had a funky formatting issue in table 2.  Fun with HTML.  Easy to fix, and I'll use your suggested Figure 2 reference text.

Glenn - I changed your post to "Embargoed" which means that anyone who's given the link can read it.  But it would make sense to publish yours around the same time as this one is published.

Kevin - I'll add your links in there.

2011-11-02 06:11:26
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

ooh, John has added the ability for up to 4 authors on a post, so I added Glenn and Kevin as co-authors, along with myself and Rob H.

2011-11-02 09:33:21Comment
Robert Way

robert_way19@hotmail...
142.162.25.5

We should incorporate STEVE MC climate audit rant

2011-11-02 09:39:42link?
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Got a link?

2011-11-02 09:54:39Comment
Robert Way

robert_way19@hotmail...
142.162.25.5

http://climateaudit.org/2011/11/01/closing-thoughts-on-best/

2011-11-02 10:15:45
Rob Honeycutt

robhon@mac...
98.207.62.223

Mmmm....  Very tasty piece.  :-)

2011-11-02 10:16:17
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Oh man, nice Robert.  McIntyre basically echoes everything Willis said, along with most of the same errors (except he doesn't even look at GISS).  McIntyre wrongfully "downscales" the satellite data by a factor of 1.4 to account for the amplification factor, which is actually 0.95.  His conclusion is just absurd:

"If one takes the view that downscaled satellite trends provide our most accurate present knowledge of surface trends, then one has to conclude that the BEST methodological innovations (praised by realclimate) actually provide a worse estimate of surface trends than even CRU.

In my opinion, it is highly legitimate (or as at least a null hypothesis) to place greatest weight on satellite data and presume that the higher trends in CRU and BEST arise from combinations of urbanization, changed land use, station quality, Mennian methodology etc.

It seems to me that there is a high onus on anyone arguing in favor of a temperature reconstruction from surface station data (be it CRU or BEST) to demonstrate why this data with all its known problems should be preferred to the satellite data. This is not done in the BEST articles."

The fake skeptic love affair with UAH and RSS is really bordering on the absurd.  McIntyre provides no justification for favoring their data sets (his "null hypothesis"), nor does Willis.  It couldn't be more obvious that they prefer UAH and RSS because they have low trends.

I'll change the title to "Eschenbach and McIntyre's BEST shot..."

2011-11-02 13:41:03
skywatcher

andycasely@hotmail...
122.107.164.176

Has the title change busted the link to the post?  From the comments, looks like it'll be a good one!

2011-11-02 14:20:51
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Okay, updated the post to include McIntyre alongside Eschenbach.  I also estimated the scaling ratio if Fu, V&G, and Zou are correct about the TMT and (our estimation of their) TLT trends (well, it's V&G's TLT trend, and our estimation of Fu's and Zou's). 

The resulting scaling ratio over land is ~0.94; well within the 0.95 +/- 0.07 range quoted by Gavin.  It's a very very rough estimate of course, but I'm just trying to make the point that there are more plausible explanations than all the surface temp records still being contaminated by UHI and biased high, and it sure looks like UAH and RSS being biased low is a very plausible explanation.

Here's the revised post if anyone wants to have another look.  I think it's pretty damning:

Eschenbach and McIntyre's BEST Shot at the Surface Temperature Record

2011-11-02 14:37:56
Andy S

skucea@telus...
66.183.185.188

Looks good!

Since this is going under the microscope and every molehill will be blown into a McMountain, you should point out that most of the curves in Fig 3 are annual averages, whereas the UAH and RSS curves are monthly data with a 12 month averaging  filter. After the Daily Mail BS people will be looking for that recent bogus low spike on the BEST data.

Also: " UHI and surface station siting do not effect temperature trends". Should be "affect"

You mention the short baseline that WE used but don't point to the consequences. What this does is that WE's 79-84 baseline brings the satellite curves down in these early years (UAH and RSS were running relatively warm to the others at that time), and visually exaggerates the divergence with the other records in more recent years.

2011-11-02 16:58:10
skywatcher

andycasely@hotmail...
122.107.164.176

Looks very good to me.  Andy S has pointed out the only typo I noticed :)  Once again, I have learned a lot from reading a post by you.

2011-11-02 17:35:29Comment
Robert Way

robert_way19@hotmail...
142.162.25.5

Dana thought you might be interested in this document about NOAA. Says their temps work worse in the arctic
ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/blended/ghcnm-v3.pdf

Umm regarding this stuff I think you should see my comment and then the subsequent responses by Gavin and Steve Mc and climate audit (as well as Mosher and JeffID).

Will the UAH and RSS monthly curves impact the trends being analyzed as opposed to the annually averaged series'?

2011-11-02 17:39:21Comment
Robert Way

robert_way19@hotmail...
142.162.25.5

From Gavin *commenting under me at Steves*

Robert
I might need a reference for the model amplification factor of 1.4x. Where can I find that ?

Gavin

Nowhere. The expected land-only amplification of MSU-LT over SAT is close to zero (actually equivalent to a factor of ~0.95 +/- 0.07 according to the GISS model).

The ocean-only tropical amplification is related to the moist adiabat which is not the dominant temperature structure over land since deep convection is mostly an tropical ocean phenomena. Ocean temperatures are rising slower than over land, therefore even if tropical land tropospheric temperatures were being set by a moist adiabat over the ocean, it would still have a smaller ratio with respect to the land temp.

2011-11-02 19:27:40
Glenn Tamblyn

glenn@thefoodgallery.com...
58.167.63.44

This is good to go. Again, it shows up the limits of how many different sources the WUWT 'Team' use. I will 'tone up' my additional post tomorrow and repost it for comment. Suggest they both go live on the same day.

2011-11-02 22:31:52
Riccardo

riccardoreitano@tiscali...
2.33.129.49

Robert

the trend doesn't change when using monthly vs. annual series. The estimated parameter error is much smaller with the monthly series, but it's an artifact of autocorrelation.

2011-11-03 03:25:23
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

As Robert notes, McIntyre corrected his post, although for some reason he still leaves the plot with the amplification of 1.4 alongside the corrected version.  I don't even know where he got 1.4 - aside from being 0.95 over land, it's 1.2 globally, not 1.4.  I guess that's the denialist brain amplification factor.

Have made the other suggested corrections.  Let me know if there are any other comments.

2011-11-03 03:32:32comment
Robert Way

robert_way19@hotmail...
134.153.162.53

Maybe it would be worth showing a bar graph of the land-only temperature trends across all satellite teams and surface temp series...

So a bargraph with the group name on bottom and the trend per decade on the side. I can do so if you like. Plus use the 3 other Satellite series groups and the proper downscaling factor.

2011-11-03 03:39:08
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

More like an upscaling factor in this case!  But yeah, that could be a good graphic.  Go for it if you have time, or I can whip something up tonight pretty quickly.  To get the other 3 groups, I just took the global ratio (0.2 for them vs. 0.14 for UAH/RSS) and applied it to the UAH/RSS land-only ratio (0.19°C/decade on average).  That's where the 0.27°C/decade came from.  Then add the 'upscaling factor' of 0.95, and you get ~0.286°C/decade.  Basically identical to BEST and NOAA.

2011-11-03 03:39:52comment
Robert Way

robert_way19@hotmail...
134.153.162.53

Okay I'll go for it. I'll show the figure soon.

2011-11-03 04:00:36Graph
Robert Way

robert_way19@hotmail...
134.153.162.53

Any thoughts?

http://www.skepticalscience.com/pics/Trend_Graphs.png

Might need to double check the numbers for 100% sure before "climate audit" jumps down my thoat.

2011-11-03 04:01:37comment
Robert Way

robert_way19@hotmail...
134.153.162.53

Should we add error bars with trend uncertainties?

2011-11-03 04:19:07
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

You'll have to apply the 0.95 upscaling to UAH and RSS, and then change the title to "Estimated Land-Only Surface Temperature Trends".  Otherwise looks good.  I don't know what the uncertainties would be for our last 3 estimates there. 

2011-11-03 09:57:06Important point
Glenn Tamblyn

glenn@thefoodgallery.com...
144.131.91.21

Oops, I have been away from the forum a bit too long. I need to stress an important point here, if it wasn't obvious from my post. The figures for Fu, G&V & Zou are Global, not Land Only. I haven't seen any data for land-only for these three. So Roberts bar graph would be misleading.

 

Double oops. Just reread Dana's latest version and that is covered. Note to self, do not write anything before the second coffee has kicked in.

2011-11-03 13:08:41
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55
Hehe Glenn :-) Robert, maybe somehow indicate which of the records are surface and which are satellite, either above or below.
2011-11-03 13:22:32
Rob Painting
Rob
paintingskeri@vodafone.co...
118.92.46.170

Maybe in the bars themselves? Cool graph Rob Way!

2011-11-03 14:34:12comment
Robert Way

robert_way19@hotmail...
134.153.162.53

Hi Dana,
I can't update the graph until tomorrow. I have a lecture to teach in the morning that i'm still not prepared for haha

2011-11-03 14:58:38
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Ok Robert.  We gotta give Glenn a little time to finish his satellite post anyway.  It would be nice to get both up tomorrow, or Friday at the latest though.

2011-11-03 23:08:01
Glenn Tamblyn

glenn@thefoodgallery.com...
58.167.1.112

Latest version of the satellite post here

2011-11-04 05:06:29
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

Looks good to me Glenn.

2011-11-04 08:31:20
Andy S

skucea@telus...
66.183.185.188

Glenn

Seems good to me, although I need to read some background posts to properly grasp what you're saying.

One typo: Oxygen molcules

channels vs sensors Could you explain what the difference is? As I read this quickly I started assuming the tems were interchangeable but they are not.

Willis also seems to think that the expectation from theory is that the air in the Troposphere over the land should show greater warming seems to be a confusion of 2 different ideas. Greater warming is expected at the surface on land compared to the oceans since the oceans can transport heat away from the surface down to the depths. Also greater warming at the surface in higher latitudes than at the equator is expected since major processes within the climate system transport heat from the equator to higher latitudes.

I'm not sure what the underlined word (by me) "surface" refers to, is it the land and ocean surface or just the land surface?

And doesn't evaporative cooling at the surface play a bigger role in oceans than on land? (I'm not sure about that at all, I'm just asking)

Perhaps you and Dana should be consistent about whether you refer to the guy as Willis or Eschenbach. I'd recommend using Eschenbach as Dana has mostly done. There's something about less common first names like Willis and Gavin that encourages everyone to use those names, whereas we don't generally refer to people as Richard (eg, Lindzen), Michael (Mann), David (Archer) or Steve (McI)

2011-11-04 15:05:43
Glenn Tamblyn

glenn@thefoodgallery.com...
138.217.134.13

Thanks Andy.

Typo fixed, reworked use of channels vs sensors, all names are Eschenbach and tidied up the para you referred to.

2011-11-04 16:26:22
Albatross
Julian Brimelow
stomatalaperture@gmail...
198.53.65.169

I hate to be a pain or is that pessimist, but I have an uneasy feeling about this post.  It just seems too good to be true. Yes, very helpful I know ;)  Perhaps me reading this did not help:

http://wattsupwiththat.com/2011/11/03/a-considered-critique-of-berkley-temperature-series/

Mosher:

"Can we conclude ( as Christy, Spencer and Steve do) that the difference

.28 [CRU] – .18C [UAH] or .1C per decade could be UHI?"

I am a little wary of estimating our own TLT data from STAR data.  Glenn any idea how robust is that estimate, or how accurate it is likely to be, could you provide an uncertainty range?  Maybe use the range obtained using UAH and RSS (0.186-0.221). There is a Vinnikov et al. (2006) and a Vinnikov and Grody (2003), be careful to distinguiSh clearly between them. Anyhow, in Vinnikov et al. (2006),  their analysis was for 1979-2004, what about since then though?  Is that rate of 0.20 specified in Table 2 still valid?  If not, maybe the rate for 1979-2004 can be provided in parentheses)

That all said, this is clearly NB, b/c the deniers are digging in their heels and saying that there is a notable UHI impact, BUT they are hanging their hat on that.  So that is their weak point, but at the same time we have to be absolutely sure that we have all out ducks lined up.  

Have people seen/read Thorne et al. (2010)?  They say:

"The controversy over surface and tropospheric temperature trends started in 1990 when the first satellite upper air ‘watch’ was produced and it was na ̈ıvely assumed that it told the correct time. Over the subsequent years, with the advent  of not just two but multiple watches from different ‘manufacturers’ and using two distinct ‘technologies’, a more accurate measure of the structural uncertainty inherent in estimating what the ‘time’ truly is has emerged.

The state of the observational and model sci- ence has progressed considerably since 1990. The uncertainty of both models and observations is cur- rently wide enough, and the agreement in trends close enough, to support a finding of no fundamental discrepancy between the observations and model estimates throughout the tropospheric column."

 

Re "In fact, if we assume a linear scaling of global trend ratios to land-only trend ratios, the Fu, V&G, and Zou analyses would result in a land-only TLT trend of 0.27°C per decade, which gives a scaling ratio of ~0.94 - well within the 0.95 +/- 0.07 range predicted by climate models."

Now if true, that would be a cool finding, a finding worth publishing in fact Glenn!!  A new paper on this is desperately needed. I would suggest stating what the scaling ratio is for scaling between global to land only (is it about 1.4? That is about what I got for RSS), and say "trend of near +0.27 c per decade", you do not know for sure what the value is.

One issue that occured to me is that surely the satellite 'footprint'is not exclusively over land?  Surely that is not a valid assumption and surely that would affect the results and affect the ratio of the satellite versus surface too?  Another is if UHI were an major issue wouldn't the ratio becoming noticeably smaller with time as cities grew?  There have been some major changes/growth the past 33 years. Has anyone looked at that?

Does the figure below help at all?  The mid trop trend is greater than that of the NCDC surface data here, but inlcudes tropical sites, ocean sites etc, so maybe one cannot use it for anything useful?  

Has anyone looked at land radiosonde sites (away from water bodies and ideally away from major tpography) verus models and satellite?

 

Some tips:

i) Remember to say "et al.", not the period after "al".  if one does say Viinigrove et al and decide to abbreviate that, then say something like Vinnikov et al. (2006) (hereafter VIN06)".

ii) Table captions could be a liitle more detailed, specifying time periods involved etc.

 

A note on Mann et al. (2008):

I would also include Ljungvist (2010) too, or at least mention his paper that corroborates the claim that the 90s are the warmest in 2000 years, at least for the N. Hemi.

2011-11-04 19:34:26Back of an envelope
Glenn Tamblyn

glenn@thefoodgallery.com...
58.166.106.25

Albatross.

The Zou TLT estimate I did is just that, a rough estimate. Based solely on adding the same increment to the Zou TMT number as occurs in going from the UAH/RSS TMT values to TLT. I would certainly not claim that it is accurate, let alone assigning uncertainty to it. Equally you could use multiples rather than a fixed offset. Since Zou haven't done their TLT analysis yet (ghrsghrsghrsssshh - that is the sound of teeth gnashing) all I am trying to do is show that the TLT result from Zou would be in this ballpark. The purpose of this post is not to be definitive - not possible. Rather to show that claims by others that the Satellite data is definitive is misplaced. 2 teams using essentially the same methods calling it low, 3 teams using other methods calling it high.  I really hope that in the next 6-12 months the Zou analysis goes head to head with RSS/UAH in the journals.

So if the wording, in either post suggests more certainty than that, we need to rework it. The point is to refute Willis's apparent confidence in the 2 mainstream satellite records. We aren't trying to say that Zou or whoever is right. Rather that Willis's certainty is wrong. And that Satellite issues are far more likely to be an explanation than the ubiquitous influence of UHI.

2011-11-04 23:16:54
Kevin C

cowtan@ysbl.york.ac...
144.32.72.165

A problem I'm afraid.

If you've been following the low discussion over at Moyhu, you'll know some of this.

GISS seem to have supplied a data set to BEST on the basis of which BEST have done an updated comparison:

This shows very good agreement to the BEST data.

I just tried plotting the CCC results against the GISS unmasked and the BEST results. Masking the results in CCC makes a small difference in the right direction, but the result is still much closer to GISS-land than to BEST.

Therefore, I have to admit that I cannot reproduce the above graph using CCC. I've been pretty thorough in checking what Ive been doing with CCC, but there is something else going on. At the moment, we simply don't know what the file GISS provided to BEST was, or why it differs from the CCC landmasked result.

Sorry guys. I know that throws a spanner in the works.

2011-11-05 02:53:52
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

I'll re-word the ending a bit to make clear that these are just rough estimates for Fu and Zou, and V&G only goes to 2004.

Kevin - we can just note that the GISS land-only estimate is somewhat higher than our estimate using CCC.  Frankly since ~1950 it looks almost identical.  I'm guessing we won't be able to be more precise than that, unless we can somehow get our hands on the data GISS supplied to BEST.

2011-11-05 03:09:14
Kevin C

cowtan@ysbl.york.ac...
144.32.72.165

Yes, OK, that's reasonable. You could probably go a bit further and say that the comparison figure given by BEST (give the link but don't embed) show a GISTEMP land estimate which agrees much more closely with BEST, but the data has not been released.

2011-11-05 03:23:39
Albatross
Julian Brimelow
stomatalaperture@gmail...
198.53.65.169

Kevin and Dana,

I agree, just have a clear caveat and be up front (we have nothing to hide and are trying to do this right of course).  I'm on the periphery here, but it seems that even if the descrepancy is resolved it won't make a huge difference and the current plots still make the point right?

2011-11-05 05:18:18
dana1981
Dana Nuccitelli
dana1981@yahoo...
64.129.227.4

It would look better if the discrepancy were resolved.  As it is, our GISS trend (0.24°C/decade) is a fair amount below BEST/NOAA (0.29°C/decade), which had troubled me a little.  I'm actually glad GISS' estimate is on par with the other two, because that's more what I'd expect (i.e. globally GISS and NOAA trends are basically the same). 

I'll be sure to make that point when I update the post tonight or tomorrow, that while our CCC-based estimate is lower, GISS' own is on par with the other two.

2011-11-05 15:20:47
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Post is updated and ready to go tomorrow, so get any final comments in before then.

Eschenbach and McIntyre's BEST Shot at the Surface Temperature Record

Will publish Glenn's at the same time, too.

2011-11-05 18:01:56
Glenn Tamblyn

glenn@thefoodgallery.com...
124.179.27.45

Dana.

I think this strikes the right balance between being cautious about over-confidence while still highlighting the real possibility that the Satellite data is the issue. And paints Willis as being too quick to presume that the explanation must fit the WUWT narrative

2011-11-05 19:30:07viral?
jyyh
Otto Lehikoinen
otanle@hotmail...
85.76.49.217

EDIT:Sorry, didn't check the latest posts before that.

Anyway, on to check if my Anti-virus and anti-intrusion software is constantly updated.

2011-11-06 03:18:53
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55
Thanks Glenn. Agreed, that's the balance I was going for. Will post these a bit later today.