The Blackboard

Its interesting that most of 2010 to date is lower than the max in the latest version. If I recall correctly, it was at or above the max for most of the year in the proir version.

Zeke– It was above the max for individual days, but did not reach the all time max. I’ve inserted the graph above.

I’m a bit puzzled about your anomaly plot since AQUA was launched in 2002 so when did the record occur? That plot should have all the data available, so I guessing that the max and min are from the earlier satellite not AQUA.

Phil–
I don’t know when the data set measurements were taken. If there are no data outside of 2002-now, then the non-anomaly plot is equally puzzling. After all, the “record high” and “record low” should overlap one of the 2002-now traces.

I guess this is a question for Roy.

RSS does not exceed 1998 either. Oops.

One significance of the temperature this time of year is that the ENSO influence seems to be weakest between about July and September. However the temperature this time of year can still be influenced with my interpretation being that 1998 was still warming temps well into this time period, and cooling for the 07/08 event was well underway during this time period as well.

So it might be a bit hard to say whether temperatures now reflect warm, neutral or cool ENSO, but my guess is continuing warm, with a fairly fast change to cool around November/December.

Lucia,
I do not quite understand what you did, in the needed detail. But it appears to only be me.

You took some data; its original absolute values are around -20 C. You then subtract something to get an anomaly. Fine, but I can’t figure out from your post what that something is.

Is the it mean temperature over some years for that particular calendar day? For that calendar week? Is it the former, calculated indirectly after smoothing and interpolating from the latter (I’ve seen things like that for daily anomalies)?

The data are daily, so presumably the anomalies are daily. But something is left unsaid here; some description of how the daily anomalies were found.

Carrot–
When you visit the AMSU site, you can click to read the daily data underlying their graph. One of the columns is “average”. I subtracted the “average” reading for day “n” from the reading for day “n” of each year to create anomalies for 2002, 2003, 2004 and so on. Those are plotted.

I then computed the average for day “n” based on the data given for years 2002-2009 (or 2003-2009 if there is no data for 2002 for that day). Note: This average does not come out to equal ‘o’, which means the “average” in the file at the site is not the average based on years 2002-2009.

Actually though, you aren’t the only one puzzled. I’ve also emailed Roy Spencer to explain precisely what the columns mean. We aren’t quite sure what the average, min or max in the files are because Phil says the AQUA was only put in place in 2002.

When you visit the AMSU site

I don’t do that, actually. I don’t find it very user friendly. Nor do I fully appreciate what it is that I’m looking at, in the first place. Someday I’ll immerse myself in the details of those calcs.

OK, so there is an average given for each day, but if you compute the day’s average yourself, you don’t get the same number – is that pretty much what you’re saying? (albeit indirectly)

The thing with anomalies is, if you have too few years in the baseline, things will get noisy when you plot all the days together, as one would do. So there could be some smoothing of the baselines in there, like what I suggested. Edit: on further reflection, this is only useful if not every day is recorded every year.

I wonder what happens to Leap Day. That’s where you’d need it, I’d think.

Carrot

OK, so there is an average given for each day, but if you compute the day’s average yourself, you don’t get the same number – is that pretty much what you’re saying? (albeit indirectly)

Yes. That’s what I’m saying. But this didn’t used to be a mystery because the “average” was for the baseline from 1988-1998 (or so. The break was in july.) Then, data were given for years 1998, 1999, 2000 etc. So, the average for years 1998-now was warmer than the baseline– a different period.

I’d assumed that was the still the reason for the difference– but then Phil pointed out AQUA wasn’t launched until 2002. That makes the difference a bit mysterious. I’ve emailed Roy to find out what’s up.

I wonder what happens to Leap Day. That’s where you’d need it, I’d think.

Well… for this graph which is in now way close to a “major blog result”, I just delete it. But I don’t think the baseline on any of these was ever smoothed. It was left noisy– which was fine given it’s purpose which is just an illutration.

There is not much detail to the calcs of what I do with those figures. I just use it to eyeball what seems to be happening. These don’t have a lot of meaning– the are noisy. I mostly use these as a heads up for the betting and to start seeing what the month might have been like before Roy gets his numbers out.

But this didn’t used to be a mystery because the “average” was for the baseline from 1988-1998 (or so. The break was in july.) Then, data were given for years 1998, 1999, 2000 etc.

…they didn’t used to provide the data from their then baseline period – they only gave data from 1998 on? that’s weird, if so.

you delete Leap Day? You treacherous monster.

they didn’t used to provide the data from their then baseline period

On that web page, the daily data were provided from 1998-on. The min, max and average were provided for the baseline.

I suspect if you wanted the daily data for other periods, you could get it from Roy. I never asked. I just used the AMSU page, which I take to be a “information only” type page.

I got an answer from Roy–

The record highs and lows are based upon the entire MSU MT record back to 1979. To do this, we intercompared our official UAH TMT product, including the average annual cycle added to the anomalies, with just the Aqua averages over the Aqua period of record since mid-2002. This provided an intercalibration between Aqua-only and UAH official. This then allowed us to go back through and use all of the UAH TMT data to find record highs and lows back through 1979.

So, the average and the records are based on the entire MSU record.

Whatever. I don’t trust leap year deniers. You just want to raise our taxes like the rest of them, so you can give our hard earned money to the spherists.

Geez… Can this be made any more confusing? The UAH data seem to be changing faster than anyone can keep up with the changes. We have different satellite data being replaced for previous satellite data, and at the same time there are adjustments being made for a spurious seasonal effect. The result is a spectacular downward revision in the UAH monthly anomalies with version 5.3 reading so very different from the previous version (much lower monthly anomalies). Spencer and Christy are already having enough trouble getting accurate monthly anomalies out, but now this post says we can place our trust in daily anomaly data. Yikes!

I am still very confused by the adjustments made in February, that were stated not to effect the annual anomaly average, just lower some months, and increase others. But it appears that what happened, is that some of the temperature rise for early 2010 was shifted back to 2009, and where we would have expected the downward adjustments to the January and February anomalies (the weird seasonal high in the “old” UAH dataset) to be counteracted by upward adjustments in May and June (the weird seasonal high in the “old” data set), this didn’t happen. At this point in time, I don’t think anyone (other than Spencer and Christy) have any idea what adjustments were made and why they were made. We are being asked to trust their professional competence and not question the reported data.

OK, I get this: my first instinct is to trust the professional experts in this field of study, and to believe Spencer and Christy, even though we have no way of knowing how they are adjusting the raw data and why. Lets hope they have the monthly anomalies correctly reported; but why jump off the cliff here and start talking about daily anomalies? I am not sure that we can trust any calculated daily anomalies, especially since we have seen recent large divergences between the daily raw data, and the reported UAH monthly anomaly over the last couple of months.

OTOH, the daily satellite data show measured temperatures on Channel 5 that are warmer than any prior temperatures measured on that Channel 5… I.e. the high temperatures during the real seasonal upswing in the troposphere temperature in July have produced readings for a number of days, that have equalled or exceeded the previous high temperature for Channel 5. In short, this month we are getting temperature readings for the troposphere that are higher that any temperature ever previously measured… and the highest temperature on Channel 5 was measured just last July (2009).

Statistically, any one measurement isn’t high enough to state unequivocally that the troposphere has higher temperatures than before, but as the number of record days this month rack up, there may be a statistically meaningful result if 10+ measurements this month exceed the highest measurement ever taken in the previous 30 years.

BTW, the July 15 date (latest data) show the highest temperature ever measured for a single day on Channel 5. It seems to me, that there is quite a bit of hand waving trying to explain away the high readings this month.

It’s a good thing Spencer en Christy aren’t AGW scientists. They would have been skinned alive.

PaulK2

but now this post says we can place our trust in daily anomaly data. Yikes!

Trust it? I don’t say trust it. I just subtract to try to anticipate what the monthly value will be before it’s published.

Can we deconvolute this AMSU-A average data at the top of the article to tell us something general… about the seasons or the hemispheres?

Hmm.. I’m not sure you’re using it correctly, Hank.

Re: carrot eater (Jul 16 18:22),
CE, I’ve been running a daily update of the AMSU data here. I use the same kind of anomaly that Lucia does. When I started it, Channel 5 didn’t have the mean values quoted, so I averaged the AQUA values only, and calculated the anomaly from that.

Then I had the same worries about what the quoted mean for Channel 4 represented (I still do), so I used the average of the stated figures for that (1999-2009). I probably should have started from 2002 for AQUA-consistency.

I don’t think it makes much difference. For me, the main benefit of using anomalies here is the better scale resolution. I’m not devoting y-axis space to tracking the seasonal variation. It’s true that the average itself is not very smooth, and injects some small kinks into the anomaly. I’ve considered smoothing it.

Nick,
thanks for the thoughts.

I’m not devoting y-axis space to tracking the seasonal variation.

That’s the only pressing reason to use anomalies here, that I can think of.

carrot, I’m asking what’s the meaning of this curve. why is it warmer in the summer months? Is the northern hemisphere warmer or does it warm more readily?

It appears the 16th may have broken the record again but the red line that gives the reading has dissapeared for me.

Vexing.

carrot, I’m asking what’s the meaning of this curve. why is it warmer in the summer months? Is the northern hemisphere warmer or does it warm more readily?
.
more land. which warms much faster, than water does. (just my educated guess…)

sod (Comment#49262) July 19th, 2010 at 4:40 am
Apparently you all don’t understand solar radiation after all!! LOL

Lucia wrote:
In comments BenjaminG noted:

I’m not reading precisely those numbers for AQUA-Channel 5 for July 5th and 6th, I read 254.317K and 254.301K, but I am reading 254.333K for July 10, which is a record all time high.

The readings for July 5th and 6th are 254.315 and 254.317. I rounded to 254.32.

We are at new absolute high again with the latest reading for July 17th, at 254.443K.

Still not a record anomaly, but I’ve been a little surprised by this latest uptick, which has given us the highest anomaly of the year so far, IINM.

Regarding Michael Hauber’s musing about the impact of El Nino at this stage, it’s my understanding that the maximum effect of El Nino on global tropospheric temperatures is delayed by 3-7 months. The recent McLean et al. paper, for instance, identified a 5-7 month delay in response of the UAH estimate to changes in the SOI.

Which means that we are still definitely seeing the effects of El Nino in the satellite temperatures, and it could be in the ballpark of the peak effect we will see. Or, that peak effect could have already past, but we are almost certainly still seeing some effect.

Holy cow, look at that 2010 trend towering over all the others. Why aren’t Christy and Spencer doing anything? Where is version 5.8beta to take care of this?

Neven,
Be sceptical, but don’t act like the ‘sceptics’..

liza (Comment#49267) July 19th, 2010 at 7:34 am

sod (Comment#49262) July 19th, 2010 at 4:40 am
Apparently you all don’t understand solar radiation after all!! LOL

.
liza, i am looking forward to your explanation of the annual cycle in the graph.

let me see, whether i get this one right:
.
the graph includes maximums for all years on record, but only shows the years since 2002.
.
this is the reason, why the “max” is often much higher than all other graphs that are displayed.
.
2010 has been tracking the max graph pretty much exactly during (parts of) several months. (march, may, june)
during those months, the max number would be 2010. (or it is extreme chance of a very close track)
.
but how can 2010 move above the max number? leaves me puzzled!
.

Be sceptical, but don’t act like the ‘sceptics’..
.
Yes, I’m sorry. It is frustrating. Soon La Niña will do her thing, and the whole charade starts all over again. In the meantime we don’t talk about global temperatures.

sod wrote:

2010 has been tracking the max graph pretty much exactly during (parts of) several months. (march, may, june)
during those months, the max number would be 2010. (or it is extreme chance of a very close track)
.
but how can 2010 move above the max number? leaves me puzzled!
.

As they were fiddling with the graph software over the last several weeks I observed that before the latest changes, in early July the 2010 temps were over the max temps. Then a few days later, along with some other minor changes, those early July max temps were updated to be the new max set in 2010.

My interpretation: the ‘max temp’ data is created by a somewhat manual process. I.e. they have to run a script to update it. In between the updates, the max temps are static and new data can come in that is higher than ‘max’.

I’d prefer if they did not include any of the current year’s temps in the ‘max’ calculation, so that we could more clearly see what the previous max was and when and how much the 2010 trace surpassed it.

As an aside, one thing I’ve noticed that is surprising: look at how far below the max trace January 2010 runs. Yet, in the final dataset, January 2010 is the highest, by a sizable margin. Goes to show just how much the adjustments can affect the outcome.

I’m saying, don’t insinuate that Christy and Spencer are fiddling things with ill intent.

The good thing about the satellite data is that anybody can do their own analysis; the raw data are there. So if the UAH guys are doing something wrong, it’ll eventually catch up to them. There is a history of this, after all.

A somewhat dramatic spike in channel 5 the last few days considering that the spike started from record temperatures.

At the same time the channel 4 and sst channels have dropped a little….

Neven

In the meantime we don’t talk about global temperatures.

Who doesn’t talk about them?

Ok, here is my take on the number of new record daily high temperatures on Channel 5… (Hopefully there is a statistician here who understands the statistics of treating outlying data points better than I, who can do a more accurate analysis.)
.
Since the temperature data is being reported in the reports on the AMSU site to only two decimal places, this implies the accuracy of the measurement is only to one hundredth of degree C. So the discussion about the third decimal place (one thousandth of a degree) is misplaced. Based on the reported data, the highest temperature prior to 2010 was measured on July 23, 2009 at minus 18.84 deg C or 253.31 K. This year the readings on July 5th, 6th, and 11th were 253.32 K. The readings on July 10th and July 12th were even higher at 253.33. This week the reading on July 15 hit a substantial new high of 253.36 which was wiped out the next day July 16 at 253.40. And July 17, the previous record high was demolished with a 253.44 measurement.
.
This makes eight (8) record high measurements this month.
.
OK, let me try to roughly estimate the probability of this happening due to random variation. Record high temperatures in the satellite data are almost certainly likely to be set during July. Since the satellite data began in 1979, that means that including this month, we have had 31 months of July with 31 days in each month, so there could be 961 days that possibly could have been the record setting day. So the chance of a single record high being set this July of 2010 is 1/961 for each day, but there are 31 days in the month so the probability of ONE record high this month (in a totally random system with no underlying warming trend) would be 31/961 or about 3.2%. So ONE record high wouldn’t be too surprising.

But the chance of TWO record highs is (31/961) x (30/960), which is only one chance in a thousand.

We have had EIGHT record highs, so the probability of that happening in a random system is the product from n=1 to n=8 of each probability which can be expressed generally as:
[(31-n/2)/(961-n/2)]^n
.
Plugging in n=8 gives us a probability of only four chances in ten trillion of getting eight record highs this month in a totally random system. No doubt about it; the data clearly show the troposphere is at a higher temperature than ever before in the satellite data, and that this must be due to a strong warming trend, or to a particularly unusual warm event this July.

But contrary to the hypothesis of some unusual warming event this year, data show this year’s El Nino was fairly ordinary, not anything special like the 1998. And we know the El Nino has already been replaced by La Nina conditions in the Pacific. We also know that solar insolation is still in the lowest portion of the eleven year cycle. So clearly the hypothesis of some unusual warming event this year can be rejected, leaving us with the likely cause of these record highs being a strong underlying warming trend.
.
If someone has a better understanding of the statistics, it might be possible to estimate how strong the warming trend would have to be to get eight records this month (and the month is not over yet).
.
Any comments?

Neven (Comment#49232) July 18th, 2010 at 9:49 am
It’s a good thing Spencer en Christy aren’t AGW scientists. They would have been skinned alive.

They’d deserve it too. As soon as their seasonal anomaly problem started to show high values in Jan/Feb(which they’d ignored when it showed very low values in May, June), they introduced an arbitrary fix. This year their on-line data was inconveniently running above their 20year record (79-98) so a couple of months ago they switch to Aqua channel 5, (no records, means etc.). Now they’ve pulled in the previous records (from a different satellite) which remarkably are above the Aqua data! If McIntyre wasn’t so tied up with his politicking it’s something he’d be all over (if it were from ‘the other side’ of course).

An interesting exercise is plotting channel 5 values for 2009 vs average. Looks like late in the year was quite close to average despite official surface monthly values well above average.

Yeah, Phil, as CE says the raw data is there. Get busy, or do you prefer to insinuate?
===============

Steve got busy without the raw data. What’s stopping you?
=================

Is there a claim that we can replicate the UAH reports? How? We don’t know what the adjustment algorithm that UAH is using to remove the spurious seasonal trend. I would have expected that after the substantial downward revisions in the monthly anomaly that were applied to January and February (and likely March), that we would have seen upward adjustments in May and June, but that doesn’t appear to have happened.

And we don’t know how the switch to the new satellite dataset was handled by UAH. Plus, all the historical data back to 1998 has been substantially revised (such that online data sets like woodfortrees don’t have the correct historical data, the last time I checked). The data keeps changing; the reports keep changing; the metrics keep changing…
.
Until Christy and Spencer “fix” their broken UAH system, there isn’t much we can do except take their results as the work of competent professionals who know what they are doing. That I am willing to do.
.
I am not willing to take their results yet, and extrapolate too far with the results. If anything, we seem to have found that the satellite measurements show a great deal of seasonal and yearly variation in anomalies not present in the other temperature records, and it isn’t clear at all, if this variation actually exists to the extent the UAH reports seem to show.

Nope, not to replicate the series, but to create a new one from the raw data. I realize this is difficult. Much more difficult than insinuating.
========================

Phil. reflects, suspects,
And states of what he knows not.
That’s climate science.
=================

kim (Comment#49391) July 19th, 2010 at 10:19 pm
Yeah, Phil, as CE says the raw data is there. Get busy, or do you prefer to insinuate?
===============

Who’s insinuating, that source is starting to become a joke, how many arbitrary changes are we going to get? We don’t know what’s going on with that daily site, after Spencer told us that the way to monitor mid-tropospheric temperatures was to follow AMSU Ch05, on his website for a while we had the last 13 years data accompanied by the 20 year stats (Max, Min, Mean for 79-98).
Of course as the betting competitions on here revealed that channel 05 data didn’t bear much similarity to the data that S&C used to create their monthly stats (Aqua Ch 05). Then suddenly the AMSU data disappeared and were replaced with the last 9 years of AQUA 05 data ( no stats)! Then suddenly some stats appear but it’s obvious that they are not the stats of the data shown (as I told Lucia). Apparently Spencer has told Lucia that it’s the older data, it certainly isn’t Aqua data! The whole thing’s a dog’s breakfast, you have no idea what you’re looking at, the documentation is useless.

Heh, keep digging. Your words are there at 9:14 PM.
==========================

What happens at the end of january and first half of february? Why is the average temperature increase? It doesn’t look natural.

July 17,2010: -18.71, average: 19.38
Just sayin’

kim (Comment#49402) July 20th, 2010 at 12:49 am
Heh, keep digging. Your words are there at 9:14 PM.
==========================

Yes they’re fairly clear, Spencer’s undocumented chopping and changing is making a mess of that site, that’s hardly insinuation.

Yep, everyone read them. What you’ve written since is cowardly.
=============

Doesn’t RSS put out anything useful, for comparison?

Unlike the land near-surface records, this is clearly a realm where methodology differences are still pretty important.

Kim,
The UAH guys do seem to be making fairly important changes on the fly, without much clear documentation. It’s hard to know what to make of it, until they write it up someplace. For anybody using their records in the meantime, it can surely be frustrating.

Kim,
Doing a parallel calculation to double-check the UAH guys is possible, but much harder than what a half-dozen bloggers have just done to double-check GISS, CRU and NCDC. In addition to programming skill, it requires some understanding of radiation physics. There are also issues with the data (orbital drift, how to merge data from different satellites) which are non-trivial and the choices made affect the end results substantially.

That said, UAH are not the only ones doing it. There is also RSS. On at least a one-time basis, there are also publications from some others (Fu, Prabhakara, I forget who else).

It’s taken a lot of corrections to UAH just to reach where they are now, and many of them were found/suggested by these other workers.

So it would be useful if some bloggers moved onto looking at the MSUs, but it would be much more difficult terrain for them.

CE, yes, I understood what you have to say. I’m pointing out that Phil. insinuates something which could be checked. And now we see Phil. backtracking as if butter doesn’t melt in his mouth.
============

And one more question. Why some parts of different years are so higly correlated? See UAH CH05 for 5 JAN – 31 MAR in 2006 and 2009.

Kim,
The UAH guys are making various changes to their methods, and the documentation for this seems inadequate, as of yet. If in the meanwhile somebody is trying to use their records, this will be quite frustrating.

Phil is correct in all that. I just wouldn’t go so far as to hint at ill intent. But given the history of the UAH record, scepticism is required.

Look, I know what Phil. is correct at and I know when he is hinting at ill intent. He knows it, too, the coward.
================

kim,
I don’t read your comments very carefully, but I have gotten the idea that you hint at ill intent constantly.

Maybe that isn’t accurate.

Damned with your own words, fella.
====================

I’ll be watching for you then, kim. You hint at ill intent, and you’ll get called out.

Oh, I’m so threatened. So why do you object when I call out Phil. for hinting at ill intent?
=================

Psst, CE, that’s a hint at ill intent on your part.
==================

I’ll take that back, CE. I know you saw what he did, and I have found you capable of going against your tribal instincts and calling it like it is.
==============

It’s also not as if allegations of ill intent are out of place in this whole climate mess. What I really objected to with Phil. was his insinuation of ill intent when the work to prove ill intent can be done. This is not science; this is tribal drumming.
===================

You’re tripping over yourself.

Nope, I still think it is ill intent on your part, but I’ll withdraw the insinuation. Drum on.
=============

Ill intent on my part? My recollection of the nature of your comments is that they drip with insinuations of ill intent. That is all. I’ll watch more carefully in the future. But you seem to be justifying such behavior, so maybe I’m not so far off base. We’ll see.

Nope, you miss even that. The ill intent I’m talking about is your stated intention of reflecting on my comments with a jaundiced eye. Keep drumming, it’s getting colder and we all need to dance.
=========

I know the impression I have of the nature of your comments. We’ll see.

I have worn my fingers to the bone trying to explain that this false paradigm of strong climate sensitivity to CO2 is a result of an ‘Extraordinary Popular Delusion and Madness of the Crowd’ and not a great international conspiracy. Nonetheless, there are clearly some who’ve ‘breathed together’ with ill intent. What we will see in the future is who so breathed.
====================

To the Moderator: How many infractions of the code of conduct does it take, before you start snipping out comments? I am seeing comments using slurs and personal attacks; other comments seem to have deviated strongly from any discussion of the UAH temperature record. Why are are these comments being allowed?

Paul K2,

This place isn’t really moderated per se, though this is getting a tad rowdy.

Folks: play nice. Or Lucia might have to scold everyone again

PaulK2

How many infractions of the code of conduct does it take, before you start snipping out comments?

Depends when I read them. Usually, I post an other article, and the rowdiness dies a natural death. . .

People– try to be nice or I’ll close the thread.

Having had the dubious distinction of attending what was/is effectively a private school modelled on the English ‘public schools (long long ago in a galaxy far far away), I can attest to the fact that the word ‘defalcation’ was used, often, as a synonym for ‘deficiency’. That usage can be viewed as a noun construct from the verb form, where the verb form refers to the action of reducing (embezzlement) and the noun construct form to the result. (Only available in English… to paraphrase the advert.)

And this usage of defalcation was learned some years before the classics master enlightened his class as to the Latin source of the word…..as our defalcations in learning Latin were many, varied and repeated.

And this usage was learned easily a dozen years before I started learning about ‘breach of fiduciary duties’ in law school.

Summary: actually a common word in a civil society where the concepts of duty and responsibility have not vanished. (And so NOT a cromulent word!).

On topic again:
.
kim wrote:
.

Steve got busy without the raw data. What’s stopping you?
=================

.
Did I get this right? Is Steve McI auditing Christy and Spencer? That’s fantastic news!

guys and girls, you are reading the data wrong.

Steven Goddard has done a careful analysis, and came to this conclusion:

atellite temperatures in 2010 are showing no signs of setting a record.

http://wattsupwiththat.com/201.....e-missing/

that AQUA channel 5 july stuff is pure imagination…

Neven (Comment#49534)-I kinda thought that Mears and Wentz already did that. But surely you tell me, they are in on Christy and Spencer’s conspiracy.

What? GISSTEMP, HadCRUT, and NCDC were all said to be “correct” because they agreed with one-another. Surely the less extreme claim that there is nothing seriously wrong going on with how the satellite data sets are put together is just as supported with the same argument. Unless, you wish to be inconsistent.

Not only that, but the Team has been trying to show UAH is wrong for years. Forcing several revisions in response to criticisms (you know, how science is done?) but never succeeding in causing the complete radical change they wanted, to say it had been completely “wrong”.

Andrew_FL wrote:
.
Forcing several revisions in response to criticisms (you know, how science is done?) but never succeeding in causing the complete radical change they wanted, to say it had been completely “wrong”.
.
For years, before the revisions, the UAH dataset showed a cooling trend over the length of the record. The forced revisions, which corrected items that were most definitely ‘wrong’, flipped the trend to positive.
.
Don’t know how much more radical a change you think people hoped for. I guess some would hope for a few more tenths of a degree so that the dataset agrees better with everything else.
.
I kinda thought that Mears and Wentz already did that. But surely you tell me, they are in on Christy and Spencer’s conspiracy.
.
What? GISSTEMP, HadCRUT, and NCDC were all said to be “correct” because they agreed with one-another. Surely the less extreme claim that there is nothing seriously wrong going on with how the satellite data sets are put together is just as supported with the same argument.
.
The trend found in the RSS dataset by Mears and Wentz agrees more closely with the surface datasets you mention than with UAH. I don’t know whether the .16°C/decade reported by RSS is more or less accurate than the .14°C/decade reported by UAH.

Andrew_FL (Comment#49569) July 21st, 2010 at 4:10 pm
Neven (Comment#49534)-I kinda thought that Mears and Wentz already did that. But surely you tell me, they are in on Christy and Spencer’s conspiracy.

Mears and Wentz have published in detail how there calculations etc. are done, when a change is made it’s documented.

What? GISSTEMP, HadCRUT, and NCDC were all said to be “correct” because they agreed with one-another. Surely the less extreme claim that there is nothing seriously wrong going on with how the satellite data sets are put together is just as supported with the same argument. Unless, you wish to be inconsistent.
Not only that, but the Team has been trying to show UAH is wrong for years. Forcing several revisions in response to criticisms (you know, how science is done?) but never succeeding in causing the complete radical change they wanted, to say it had been completely “wrong”.

There have been 3 substantial changes in the UAH products this year, none of them documented apparently, consequently I have much more confidence in the RSS product.

i just dug up the Spencer comments on the adjustment, to see what to expect from July.

http://www.drroyspencer.com/20.....-unveiled/

Spencer has posted this table for 2009:

YR MON v5.2 v5.3
2009 1 0.304 0.213
2009 2 0.347 0.220
2009 3 0.206 0.174
2009 4 0.090 0.135
2009 5 0.045 0.102
2009 6 0.003 0.022
2009 7 0.411 0.414
2009 8 0.229 0.245
2009 9 0.422 0.502
2009 10 0.286 0.353
2009 11 0.497 0.504
2009 12 0.288 0.262
2010 1 0.721 0.630
2010 2 0.740 0.613

so July should be pretty close to the actual readings.

Phil-
When Spencer announced the change, he wrote:

For a more detailed discussion of this issue written last July, email John Christy at christy@nsstc.uah.edu for the document.

I haven’t emailed for the document. I’m content to wait for a fully released document. I don’t mind UAH implementing the changes on the web and documenting in parallel, but it can be a bit confusing. (To explain what I mean by confusing: Hansen’s method of computing anomalies is confusing when observed– what with all the changes in anomalies propagating back. Of course, we understand why it happens. All that means is somethings that are fine can be confusing.)

Those aren’t comparable, Lucia. One is a change in algorithm, which needs to be documented somewhere. The other isn’t.

This is the problem with live tinkering with the algorithm in a record that’s used by many other people. There’s a tension between putting out what you think is the best version of the record, and waiting to release a new version until you’ve written it up an adequate description. On the whole I think it’s a somewhat unusual position for a scientist to be in, so it’s not surprising this isn’t always handled as well as the users might want.

A better comparison is GISS’s latest algorithm change. They switched from using the GHCN urban/rural information to using satellite night brightness for non-US stations. I think they noted this on the website as soon as they went live with it, and if you’re already very familiar with GISS, a short note is all that was needed to understand what they did. Within pretty short order (I forget how long), they put up a draft of the paper they’re working on to describe and analyse the surrounding issues. Personally, I’d have waited to implement the change until they put up the draft. The world didn’t end because you hadn’t made that change up to that point; the world can handle you waiting another month before you can write a decent description, if not publish a formal paper. I’d apply that standard to UAH as well; maybe the thing Christy would email you would be adequate, I don’t know.

Sod, to be fair to Steven Goddard, isn’t it possible that his was a clumsy way of saying that the satellite temperatures for the whole year 2010 are unlikely to set a record?
With the current onset of a la Nina that’s probably a safe bet.

Niels, I agree with your assessment. I think it’s pretty clear that’s what Goddard was saying, and he’s probably right.

BenjaminG: That was a very good reply, and I agree all around. I wonder if Andrew realises the scale and scope of all the corrections that have been made to UAH over the years. They’re pretty massive. The remaining discrepancy with RSS is minor in comparison, but I think everybody would like to see that discrepancy sorted out, one way or the other.

Carrot

This is the problem with live tinkering with the algorithm in a record that’s used by many other people.

Sure. I agree. But I don’t think the channel 5 daily display is used by many people. The UAH record is, so that does represent a problem for anyone who wants to write papers using monthly data after the change. It needs to be documented pretty soon.

I’m not sure there is a huge problem with the tinkering going live.

Whether or not the tinkering happened live, the UAH record being in transition would represent a problem for end users whether or not we saw the tinkering live. If they tinkering was done but out of public view, then those using the UAH record would be in a position of ignorance making conclusions based on data Spencer and Christy posted but consider superceded. With tinkering live, one at least knows there is an issue. These are the new data, but the method isn’t documented. Maybe you can’t use it– but on the other hand, you shouldn’t have been using the superceded data either. Them’s the breaks.

I actually prefer the latter situation (tinkering live) even if it causes some people who dislike Spencer and Christy to make snide remarks. Presumably, Spencer and Christy prefer to take the better course and endure snark than to take the poorer course to avoid having their detractors say snippy things.

BenjaminG (Comment#49573)-”For years, before the revisions, the UAH dataset showed a cooling trend over the length of the record. The forced revisions, which corrected items that were most definitely ‘wrong’, flipped the trend to positive.”

This is very misleading, as the record has also gotten longer over time, so the trend will tend to change just from that. However, to my knowledge, none of the revisions constituted “significant” changes with respect to the published uncertainty.

“Don’t know how much more radical a change you think people hoped for. I guess some would hope for a few more tenths of a degree so that the dataset agrees better with everything else.”

UAH agrees very closely with all datasets which measure the same thing, including RSS. Neither of these datasets is appropriately compared directly to the surface, so the “agreement” is hardly impressive. The changes people want/need for UAH to be in agreement with expectations are rather large.

“The trend found in the RSS dataset by Mears and Wentz agrees more closely with the surface datasets you mention than with UAH. I don’t know whether the .16°C/decade reported by RSS is more or less accurate than the .14°C/decade reported by UAH.”

RSS and UAH again are very close to one another and comparing them directly to the surface record is not appropriate. But if you want to know which data set is better supported by evidence, there is extensive literature on this topic, and in general UAH is better supported.

Phil. (Comment#49574)-”Mears and Wentz have published in detail how there calculations etc. are done, when a change is made it’s documented.”

This is also true of Spencer and Christy.

“There have been 3 substantial changes in the UAH products this year”

Not sure where you get 3 from, but “substantial” is a funny word to use for adjustments which have had ZERO impact on the trend.

“none of them documented apparently”

They are called “readme files” ever heard of them?

“consequently I have much more confidence in the RSS product.”

This is unfounded, but I’m glad to see that you are applying the same standard as you would to the surface datasets…oh, wait, you aren’t doing that at all! Believing a dataset is more accurate on the basis of better documentation is quite frankly a really stupid reason, when one can instead apply independent evidence.

And don’t say “well the skeptics having been saying to me for years that they don’t believe NCDC GISS and HADCRUT because they aren’t documented well enough!” Because I don’t care what someone else has used to justify questioning those data sets, because I have never personally done any such thing.

This “Audit Spencer and Christy!” BS is really idiotic.

Carrot-”I wonder if Andrew realises the scale and scope of all the corrections that have been made to UAH over the years. They’re pretty massive.”

Before assuming my ignorance, you might consider that I was saying that the changes where never large compared to the published uncertainties. I would be interested if you could show how this is incorrect.

Lucia,
If it’s still in the tinkering stage, I think it absolutely should not be live. Don’t go live until you’re happy with the change, and are reasonably sure you’ll keep it for at least a while.

For the problem of people using something that might have inadequacies, and will about to be superseded: simple, simple, simple. Just put up a note saying you’re working on some major/minor revisions, related to x. One-two sentences would suffice.

Take the GHCN. Everybody using it knows it’s going to get a facelift soon. It’s noted on the webpage somewhere. So particularly if you’re going to use v2.mean.adj, you know you’re using something that’s about to become outdated. Yet they aren’t stage-wise implementing it live, before it’s ready.

Well, I don’t know if quite “everybody” is aware; you get sceptics who still seem unaware of the nature of changes GISS made a decade ago, and were published in formal papers.

Andrew

Before assuming my ignorance, you might consider that I was saying that the changes where never large compared to the published uncertainties.

If that’s the case you’re going to make, then the published uncertainties would have had to have been mammoth. I’ll have to go back and dig those up from the late 90s, when UAH was reporting something like +0.03 C/decade (this from memory, I’ll have to go back for the actual numbers), when after revisions the trend over that time span ended up a full order of magnitude higher.

If I was a betting man, I’d bet on your ignorance here, but OK, we can go back and see what uncertainty bounds were at the time. But even if the bounds were that huge, I think an order of magnitude change can still be called to be much bigger than the remaining discrepancy with RSS.

Carrot–

Don’t go live until you’re happy with the change, and are reasonably sure you’ll keep it for at least a while.

Well.. define “tinkering stage”? I assume Spencer and Christy think they are happy with the change and are reasonably sure they are keeping it for at least a while. But they haven’t formally documented. Is that “tinkering stage” or “not tinkering stage” in your book?

I don’t have any big problem with them posting the monthly data based on the new method. I have no problem at all with them posting daily data that way– that was never anything other than For Information Only. But, if you have a problem with either, I can understand that.

I don’t know if they’re still ‘tinkering’ in my book, or not. But I would define ‘tinkering’ as I put it above – you yourself don’t think you’re done yet.

Again, I don’t think a formal paper needs to be published before you implement. Minor changes don’t deserve a formal paper, anyway. But some level of documentation is required. I think GISS’s note and draft paper (and revised publicly available code, for that matter) are sufficient, to that end. I think the note by itself was insufficient.

carrot eater (Comment#49603)-That’s not quite what you need to do, IMAO. You need to show that the changes in the trend from revision to revision were large compared to the uncertainties, and you need to look at just the periods that the data covered at the times the adjustments were made, directly before and after. You can compare the 1979-1992 trend before the revisions to the 1979-present trend after if you want, and get an “order of magnitude difference” but that is extraordinarily disingenuous.

Niels A Nielsen (Comment#49598) July 22nd, 2010 at 7:36 am
Sod, to be fair to Steven Goddard, isn’t it possible that his was a clumsy way of saying that the satellite temperatures for the whole year 2010 are unlikely to set a record?
With the current onset of a la Nina that’s probably a safe bet.

The problem is that Goddard says something other than he meant and instead of saying mea culpa he lashes out at anyone who picks him up on it.
In this case he says:
“Phil, as usual misquoting

When your belief system depends on BS, why bother?”

In fact I quoted him accurately, his post says:
“satellite temperatures in 2010 are showing no signs of setting a record.”

To which my response was:
“Steve claims that “satellite temperatures in 2010 are showing no sign of setting a record”….

This is a misquote? Note that in his response he inserts “(for the whole year)” inside quotation marks implying that it was in the original post, which is of course a bald-faced lie.
That’s the problem with ‘Steve Goddard’ he’s a serial liar.
Of course the mods at WUWT have covered up for him and appear to have blocked all my posts since then.

Andrew

You can compare the 1979-1992 trend before the revisions to the 1979-present trend after if you want, and get an “order of magnitude difference” but that is extraordinarily disingenuous.

Go back and read what I wrote.

after revisions the trend over that time span ended up a full order of magnitude higher.

Notice the “over that time span” part? Apples to apples.

You might not remember the days when the sceptics were using Spencer and Christy’s results to say there was essentially no warming at all.

When I get a chance, I’ll have to go see if they give numbers in their papers. I doubt ancient results of their analysis are archived online anywhere.

Satellites originally showed cooling from 1979-2000 or so due to the failure to correct for orbital decay, among other factors. One of the reasons RSS emerged was to provide an alternate analysis to UAH that corrected this error. UAH gave in and changed their record as well circa 2005 or so.

See http://www.nature.com/nature/j.....164a0.html for example.

Thanks Zeke – I thought I remembered they actually had a negative trend before they got to essentially zero but barely positive. There you have it.

Looks like there’s an apples to apples comparison right there in the abstract for you, Andrew.

I think in the original 1990 paper, they (Spencer/Christy) didn’t have long enough data yet to calculate long time trends. But we can go through their ensuing papers to see how it developed.

Sod, to be fair to Steven Goddard, isn’t it possible that his was a clumsy way of saying that the satellite temperatures for the whole year 2010 are unlikely to set a record?
.
no. Goddard has made multiple errors of this sort over the last couple of weeks. he made the claim that the sea ice at barrow was
unchanged, when it was already gone. and he did this TWICE.
.
so the claim that there is no record, when temperatures are showing a record, fits a pattern.
.

With the current onset of a la Nina that’s probably a safe bet.
.
there was a la nina after the high temperature in 1998 as well.
.
———————–
.
the real reason, why we can t see a record this year, is the adjustment to v5.3 by Spencer.
.
Steven of course prefers the new version. but as people have shown, the main problem with UAH was June being too LOW.
.
http://deepclimate.org/2010/03.....t-uah-5-3/
.
but instead of correcting June up, Spencer corrected other months down.
.
so the graph shown by Steven is a little bit misleading:
.
http://climateinsiders.files.w.....alies1.png
.
as i showed above, v5.2 for february was on par with 1998. (and January was nearly 0.1° higher than shown in that graph..)

anyone got a link to v5.2 data for this year after February?

SOD, thanks for the links. DC does not show or say in the linked post that the main problem with UAH was that June was too low, though.

I note that on the 20th of April DC writes in an inline comment:

“I’m pretty sure that 2010 will end up #1 or #2 in most data sets, unless ENSO flips into La Nina really fast.

The adjustments in UAH 5.3 should not affect the annual average at all. The Jan-Feb peak is now lower, but May-June will be higher than it would have been (you can see this in 2009″

ENSO has flipped quite fast so I guess DC would agree that a record is now unlikely in most data sets.

Zeke (Comment#49610)
carrot eater (Comment#49613)

Sorry but your facts don’t seem to check out. Christy et al 2000 found a positive trend, as did Christy et al 2003. Both of these papers came out BEFORE Mears and Wentz 2005 and their big correction factor. And “order of magnitude” change in the trend is a pretty ridiculous claim to make and really I think you know it. But I don’t think your analysis of the historical record is quite accurate.

BTW, I’m starting to sense some hostility from you CE. Let’s not let this escalate. This is a perfectly honest disagreement.

Phil: “The problem is that Goddard says something other than he meant and instead of saying mea culpa he lashes out at anyone who picks him up on it.”

I agree.

SOD, thanks for the links. DC does not show or say in the linked post that the main problem with UAH was that June was too low, though.
.
well, actually i think that this graph illustrates the problem quite well:
.
http://deepclimate.files.wordp.....rends1.gif
.
the global monthly trend for may and june is completely different, than in the other FOUR (4!!!) datasets.
.
so most people would have expected an UPWARD correction for those months for recent years.
.
I’m pretty sure that 2010 will end up #1 or #2 in most data sets, unless ENSO flips into La Nina really fast.
.
DC already had the number #2 in that sentence. it looks like a pretty save bet now, but anyone reading the “cooling” misinformation spread by “sceptic” blogs would not have guessed for a “semi” record year in 2010.
it is a strange effect, that a year being the second hottest on record is claimed to be supporting the “sceptic” side in this debate. very strange.
.
The adjustments in UAH 5.3 should not affect the annual average at all. The Jan-Feb peak is now lower, but May-June will be higher than it would have been (you can see this in 2009″
.
well, the problem is, that we have not seen an upward correction at work, so far. UAH was lower than RSS in every month, starting with april.
.
http://www.woodfortrees.org/pl...../from:2007
.
ENSO has flipped quite fast so I guess DC would agree that a record is now unlikely in most data sets.
.
i also agree with that statement.

There were a series of corrections and revisions Christy and Spencer made. There are a bunch of papers finding fault with Christy/Spencer, beyond just Mears/Wentz. C+S fought it for quite a while. I don’t know if all the criticisms ended up being adopted or not.

Here, from 1995, from C+S in Journal of Climate 8:888-896,

“After adjustments, the decadal trend of the lower tropospheric global temperature from January 1979 through February 1994 becomes -0.058 C, or about 0.03 C per decade cooler than previously calculated.”

So let’s see, apples to apples, what does UAH say for that period now? Looks like -0.01 C/decade. Still negative, but rather less so. But not a full order of magnitude. Looking for the uncertainty bounds.. and let’s check other mileposts.

Sod: “well, actually i think that this graph illustrates the problem quite well:
.
http://deepclimate.files.wordp…..rends1.gif
.
the global monthly trend for may and june is completely different, than in the other FOUR (4!!!) datasets.”

So? In theory Both UAH and RSS should trend completely different than the other three surface datasets. About 1.2 times higher. RSS has a problem too then. Or is it the surface datasets? Or theory? Or is it a combination of problems? Who knows.

Andrew FL:

au contraire
http://journals.ametsoc.org/do.....3E2.0.CO;2
http://www.springerlink.com/co.....851756253/
http://www.ncdc.noaa.gov/oa/cl.....ss-msu.pdf
(the last is free and the most important)
.
Jan 1979 – Feb 1995
Prior version trend: −0.06 C per decade
Current version trend: -0.01 C per decade
.
Jan 1979 – Jan 2001
Prior version trend: 0.009 C per decade
Current version trend: 0.094 C per decade
.
Those are some pretty big (one might even say order-of-magnitude) adjustments.

The last one is certainly 1 OOM difference. I can’t find uncertainty bounds given by S+C, but it’s hard to imagine that such a difference was within them, whatever they were.

Zeke

Those are some pretty big (one might even say order-of-magnitude) adjustments.

The usage problem is “order of magnitude” is an odd term when applied to something that was negative and is now positive. Had the trend changed from -0.5 c/dec to +0.5 c/dec that would be a huge change, but I wouldn’t call it “order of magnitude”, because the absolute value of both is 0.5 — we just didn’t know the sign.

In a related but not quit identical issue of usage, if we had a Taylor series expansion like:

y= a + bx + cx^2 + O(x^3),
I’d might say “a” is leading order, “b” is the coefficient for the 1st order “correction”, “c” is the coefficient for the 2nd order correction and so on. But “c” will retain it’s order whether it has a positive or negative sign.

Lucia, it went from 0.01 to 0.1. That’s an OOM, by any meaningful usage.

Lucia,

Fair enough; “large” is certainly a sufficient modifier. I think the tendency to use OOM in conversations to emphasize a point results in a bit of overuse.

Carrot– Yeah…but it’s still a number that spans zero with some frequency. So, the usage is odd. I’m not saying your usage is wrong, but I wouldn’t use it this way because I find it odd under the circumstances.

The change is enough to make a difference to whether or not we would consider the trend meaningful. I think you and I can probably agree on that– and that’s more important than the issue of using the term “order of magnitude” in this circumstance.

I’m just saying – I chose that wording deliberately because I knew it was in that ballpark, if I could just track down the references. So I’ll not have being called ridiculous or disingenuous or whatever because of it.

I think the spanning zero part mainly came into the picture when Pinatubo was weighing heavily on the record; you can see that even in today’s UAH you get a negative trend for one of the time spans in question. With long enough trends, that shouldn’t be such an issue, but yes, it does make the usage a little difficult.

In any event, it’s a bigger shift than the remaining difference between UAH and RSS.

What exactly is the nature of the difference between UAH and RSS today? Is it still a question of how some of the separate satellites are stitched together? I’m hoping we’re past the point of simple arithmetic errors.

bleh. It used to be, if you copied your comment before posting, you could save it if the server messed up. Now that isn’t even true anymore.

You are still comparing them to the present instead of step by step.

Incredibly old apple to fresh apples, will look like a big deal. Comparing the new apples to the least old apples, and those apples to the old apples, that makes more sense. I never said that the corrections considered all together didn’t matter. But you need to consider the progression.

“fought all the way” what a load of BS. All the criticisms which had any merit were indeed accounted for. That’s how science is done. Shows what your attitude is: Christy and Spencer aren’t in your tribe, so demonize them and make them out to be fools who hadn’t a clue what they were doing. Imply that they still must be wrong (which of course you need to be the case).

Again, the evidence supporting the current iteration of the UAH dataset is strong and extensive. But even though it is easily better supported by independent evidence than the surface datasets, they back each other up but the evidence that the satellites are right does not back up anything. Right. What a load.

Oh, I suspect the satellite and surface records are both mostly correct. After all, they are pretty closely in line with each other. However, the difference between RSS and UAH is considerably larger than the difference between any of the surface reconstructions, and that alone suggests that the case is far from closed that UAH should be the record of choice.

Measuring TLT from satellites is a much less direct method of estimating the surface temperature than simply using weather stations, issues of UHI, sensor changes, station moves, etc. notwithstanding.

Also, Mears et al came out in 2003. It was 2005 or so when UAH finally made dramatic revisions to the record, if I recall correctly.

Andrew,
That’s an utterly bizarre requirement you are making. The fair comparison is: trend computed for a certain period by UAH in the past, vs the trend for the same period now. That is a measure of all the adjustments and corrections that have been made, and they are substantial. They are more substantial than the remaining difference between UAH and RSS. So the things that you’ve been objecting to here (and calling me ridiculous and disingenuous, to boot – what was that about hostility?) have been perfectly true.

But now you’re apparently saying it doesn’t count, because there were a series of corrections, instead of a single one that spanned the entire gulf. That’s moving the goalposts in a quite weird way.

Nowhere in this thread have I demonised C+S. All I’ve done is noted that their analysis has required many corrections in the past, and these corrections are quite substantial in effect. I also noted that they long defended against the need for suggested corrections; whether that was justified or not is in the eyes of the beholder. At least in the case of the arithmetic error, I think they admitted it pretty quickly; I don’t quite remember.

I take no position on the remaining difference between UAH and RSS because I don’t even know what the source of that difference is.

I’m not directly comparing UAH to the surface records because… I directly compare it to RSS.

What a load, indeed.

Zeke

I think you’re mixing up your Mears papers.

Also, Mears et al came out in 2003. It was 2005 or so when UAH finally made dramatic revisions to the record, if I recall correctly.

Not quite. The big one (basically a sign error) was pointed out in Mears et al 2005, and was accepted by C+S in 2005. I think that took UAH from having a ~0.09 to ~0.12 C/decade trend, from 1979 to that time.

Andrew_FL (Comment#49636) July 22nd, 2010 at 3:34 pm

You are still comparing them to the present instead of step by step.

Incredibly old apple to fresh apples, will look like a big deal. Comparing the new apples to the least old apples, and those apples to the old apples, that makes more sense. I never said that the corrections considered all together didn’t matter. But you need to consider the progression.

“fought all the way” what a load of BS. All the criticisms which had any merit were indeed accounted for. That’s how science is done. Shows what your attitude is: Christy and Spencer aren’t in your tribe, so demonize them and make them out to be fools who hadn’t a clue what they were doing. Imply that they still must be wrong (which of course you need to be the case).

Christy and Spencer get the benefit of respect for their work, but not other climate scientists such as Jones, Briffa and Mann. Let me know when you accord them the same respect.

Phil.

Hmm. what comments do you think have been blocked

Zeke (Comment#49637)-”Measuring TLT from satellites is a much less direct method of estimating the surface temperature than simply using weather stations”

This, and quite a lot else in your same comment, suggest to me that you STILL don’t know that satellites do not estimate the surface temperature!

carrot eater (Comment#49639)-my requirement is “bizarre”. You should have said something when I first said that was my requirement:

Andrew_FL (Comment#49606)-”That’s not quite what you need to do, IMAO. You need to show that the changes in the trend from revision to revision were large compared to the uncertainties”

I was pretty clear. It’s bizarre that you decide to argue against a different thing, and when I raise that the requirements that I want were in fact different from those you met…THEN you question my requirements? I’m not moving the goalposts, I told you were I put them, and then you complained when I reminded you what they were.

bugs (Comment#49641)-”Christy and Spencer get the benefit of respect for their work, but not other climate scientists such as Jones, Briffa and Mann. Let me know when you accord them the same respect.”

Um, first how about you let me know when I didn’t accord them such respect?

The people with the double standard think that I have a double standard? Now that’s rich!

Andrew

Measuring the lower troposphere temps from the satellites is a bit indirect as it is. There isn’t a “I’m the entirely unadulterated LT signal, pick me!” signal to work with.

You need to show that the changes in the trend from revision to revision were large compared to the uncertainties

OK, I’m sorry that I didn’t notice at that point that you were using different goalposts from those I had set earlier in the discussion:

wonder if Andrew realises the scale and scope of all the corrections that have been made to UAH over the years. They’re pretty massive. The remaining discrepancy with RSS is minor in comparison, but I think everybody would like to see that discrepancy sorted out, one way or the other.

Notice the “all the corrections”.

Even using your goalposts, the 2005 correction by itself was big. Was it bigger than the published uncertainty bounds? I have no idea, because I’m having trouble finding where C+S had published them. Can you find them?

But in any case, looking at the flow of conversation, I don’t see what meaning your goalposts have, anyway. We simply said there’s been a bunch of corrections to UAH over time, and they’ve added up to being a big deal. You object to this, for whatever reason, but it’s true. If you think it’s somehow worth pointing out that a few of the individual corrections were minor along the way (and I wouldn’t say they’re all minor, just adding up in aggregate as I do think the 2005 correction was major in any reasonable sense, if it was ~0.09 to ~0.12), then knock yourself out.

I really don’t think there is much point in continuing this any longer, as I no longer see any conflict which needs to be resolved.

As I seem to do quite often, I have read tone into words which did not necessarily carry them. I still think that you are being unfair in and vaguely suggesting that Spencer and Christy did not engage in the best of scientific practices-I believe this distorts the record. But since those who wish to argue against Spencer and Christy, whether you are one of them or not, will never cease to do that…who cares.

I do however remain irritated that the standard used to defend the surface datasets “they all agree with each other” (Duh, they are the same thing)-does not apply to the satellite data, even when truly independent evidence supports them. Would you agree that such a double standard is ridiculous?

Andrew_FL,

Satellites don’t measure surface temperature. TLT tends to be -close- to surface temperature, and it is often compared to it. There is supposed to be some amplification globally between the surface temps and TLT.

I still think that you are being unfair in and vaguely suggesting that Spencer and Christy did not engage in the best of scientific practices-I believe this distorts the record.

I’m just saying many corrections have been needed, and the effect has been substantial. Does that inherently mean S+C are not good scientists? Absolutely not. It’s a difficult field. Their 1990 Science paper was a landmark, even with all the ensuing problems they had. Trying to figure out how to merge different satellite records, how to deal with drift, how to sniff out the LT signal – none of this is easy. Rome wasn’t built in a day; neither is any scientific discipline.

Now, were C+S too slow to respond to other people’s criticisms, and/or should they have found some of the issues themselves? (the ones that were found by others, that is; they diagnosed some on their own) I am not qualified to answer that question, but that’s the one that others might insinuate. It seems to me that C+S quickly corrected the sign error once it was pointed out to them, so they do have that going for them. A cynic would say that they should have found it themselves, out of due diligence, given how much debate there was over their record. I don’t know.

However, I do find it ironic that many sceptics say on one hand that the satellites are most reliable, and on the other hand that the surface measurements are bedeviled by horrible methodology. As it happens, any reasonable methodology gets you the same results on the surface, but it’s taken a couple decades of substantial changes to the satellite methodology to get to the current position.

I do however remain irritated that the standard used to defend the surface datasets “they all agree with each other” (Duh, they are the same thing)-

You miss the point. So many sceptics suspect the methodology – that somehow global warming is largely an artifact of the meatgrinder or the adjustments or … something. So it is important to show that you can take any reasonable methodology and get to the same place, more or less.

does not apply to the satellite data, even when truly independent evidence supports them. Would you agree that such a double standard is ridiculous?

As has already been pointed out to you, the agreement between GISS/CRU/NCDC and now also a bunch of bloggers is better than that between UAH/RSS.

So no, there is no double standard, and there is nothing ridiculous. The gulf between UAH and RSS may be relatively small in some sense, but it’s still the biggest disagreement in the picture. So people will want that disagreement resolved. How that will go, I don’t know.

Well Bugs,

Christy and Spencer shared their ‘code’ with RSS, worked closely with RSS to get the issue ironed out.

Meanwhile, Jones and company whispered about the correction behind the scenes and tried to take the greatest advantage of it.

Dont you read the mails.

In contrast, Jones wrote to Mann and explained that he knew WHY Mcintyre could not replicate the results. It came down to a step,apparently, made in the code, but, apparently, undocumented in the paper.

So. The difference is NOT in making mistakes or not. All science is a “mistake” The difference comes down to transparency and openness.

carrot eater (Comment#49648)-All these things about what “many skeptics” have said…why is this used to bludgeon me, and anyone else who never made any such arguments? What is the dang point?

“As has already been pointed out to you, the agreement between GISS/CRU/NCDC and now also a bunch of bloggers is better than that between UAH/RSS.”

So because there is a tiny difference, but it’s bigger, it is not a good argument that the data isn’t complete bunk, right. That’s ridiculous whether you think so or not. But I wasn’t even referring to RSS. All the tropospheric data is in very close agreement:

http://i23.photobucket.com/alb.....podata.png

(This graph was compiled by John Christy, BTW)

But that’s not good enough, by some elusive, vague protection for data that is in good agreement with similar datasets, which is quantitative in some way, which excludes the satellite data, but includes the surface data. But all the weather balloon records are completely independent from the satellite data, and the surface temperature products are not independent at all (they come from the same underlying data). But apparently the degree of agreement is still insufficient.I say again RIDICULOUS.

Andrew

All these things about what “many skeptics” have said…why is this used to bludgeon me, and anyone else who never made any such arguments? What is the dang point?

Nobody would bludgeon you, if you didn’t voice the idea that you thought it wholly unremarkable that GISS/CRU/NCDC agree with each other. “Duh”, I think it was.

Even if the sceptics weren’t saying what they say, it’s still a significant thing to note. After all, UAH and RSS have at various times disagreed pretty badly. So it’s never a given that you’ll get the same results with different choices of methodology. If you do, it’s worth noting.

So because there is a tiny difference, but it’s bigger, it is not a good argument that the data isn’t complete bunk, right. That’s ridiculous whether you think so or not.

I have no idea what you’re trying to say. There’s in a broad sense a small difference, but it’s big enough that we should try to reconcile it. That one is a lot closer to expectations than the other is definitely part of the picture, for all concerned.

All the tropospheric data is in very close agreement:

That’s a curious plot, hadn’t seen it before. Will have to look into it.

You realise he was claiming that his satellite record was in strong agreement with balloons and whatever else, even before the revisions that substantially changed his results? That to me says that the uncertainty bars on the radiosondes and whatever else are really wide.

But apparently the degree of agreement is still insufficient.I say again RIDICULOUS.

What are you, talking to yourself today? I don’t understand how you got onto this, and wound yourself up. But in any event, I’m not so sure how much agreement there actually is, even before you get to documented biases in the radiosondes, etc. I don’t want to get into a whole discussion about Douglass/Santer, I didn’t really get into that but I know others here did, but if you get to give a graph, then I get to give a graph, and this is not so neat:

http://www.realclimate.org/ima.....8_fig6.jpg

carrot eater (Comment#49652)-”You realise he was claiming that his satellite record was in strong agreement with balloons and whatever else, even before the revisions that substantially changed his results? That to me says that the uncertainty bars on the radiosondes and whatever else are really wide.”

Problems were indeed found with the radiosonde records. These have now been dealt with. Many believe that there are more issues, which still need to be dealt with, I am not convinced that they matter that much.

“but if you get to give a graph, then I get to give a graph, and this is not so neat:”

Haha, the Santer Graph. Couple things: firstly, restricting oneself to the tropics certainly adds noise, to begin with. So, too, does ending your trend lines in 1999. And so, too, does looking at each pressure level seperately. Considering that radiosonde data sets come cobbled together from various places around the world, one certainly needs to iron out the noise carefully for global comparisons. But the data sets which diverge markedly in the tropics from what most other analyses are saying are ROABOCORE and RICH. John Christy had this to say about those:

“RICH is influenced by the ERA-40 model forecast scheme which has a clearly demonstrated spurious warming due to improper assimilation of HIRS channel 11 (which renders RAOBCORE v1.2-1.4 obsolete, see below.)” “the RAOBCORE datasets, which rely on the ERA-40 forecast cycle, have been shown to be spuriously warm in the upper air due to an error in the assimilation of HIRS channel 11 in 1991-2 (noted in earlier papers, but specifically identified in Sakamoto and Christy, 2009).”

Note that the thing the datasets with big upper air trends all have in common is (wait for it)…They are reanalyses! With specific documented biases. And yet, in the global, bulk atmospheric averages, these biases don’t result in much divergence from the satellites (see the graph I showed above).

Sakamoto, Masami, John R. Christy, 2009: The Influences of TOVS Radiance Assimilation on Temperature and Moisture Tendencies in JRA-25 and ERA-40. Journal of Atmospheric and Oceanic Technology, 26, 1435-1455

Re: carrot eater (Jul 22 17:43),

As it happens, any reasonable methodology gets you the same results on the surface, but it’s taken a couple decades of substantial changes to the satellite methodology to get to the current position.

To be fair to the point some critics of the surface sites make, any reasonable computation based on measurements available gets more or less the same result. But at some critics worry about distortions based on stations moves, periods with slow drift due to creeping UHI between moves, slower changes in micro-climate between moves etc. The fact that different methods of treating the same data get the same result doesn’t full resolve this and partly because of the possibility of the creeping changes rather than the sudden moves.

There are other pesky things that happen in the field. My husband used to monitor radiometer for ARM located in various places. His automatic error detection algorithms are designed to detected something “weird” about one. The normal weird things are instrument failure– due to factors that include spiders building nests. But one month, the “weird” thing was that– unbeknowst to the ARM project, someone had erected a tall structure of some blocked the radiometers view. (I don’t remember if it was a corn crib or what. It was tall, it happened to be in the line of sight for the radiometer and screwed something up.)

So, something about the radiometer had to be shifted slightly. (It’s not my project, so I don’t remember precisely. It might be they just had to rotate the instrument so it could sample over a full 180 arc.)

With a recent project that includes automatic error detection, these sorts of things can be detected. But you can’t go back to the 40s and discover that some unacceptable thing no one noticed was in place, and then vanished. You can only hope that these sorts of events average out. These sorts of things can give people vague anxieties over the accuracy of station measurements.

I think Zeke is doing a lot to help explain how one might sniff out problems due to these sorts of factor if they existed. But the issue of station moves, creeping UHI, local human induced changes in micro-climate is conceptually a pesky problem and does need to be explained in a way that the public can understand. Like it or not, the brevity of peer reviewed papers unsuits them to this task.

For my part– I’ve never preferred satellites to surface stations. I don’t necessarily prefer the surface stations over satellites. I’m glad we have both.

Carrot and Andrew_FL– If you find yourselves escalating, remember it’s better to argue with a person who is actually present in comments than to argue against what “many skeptics” or “many AGW advocates” say. (Mind you, I’m guilty of arguing against what “many” say too. It’s hard not to, ain’t it? But it’s best to try to avoid it. )

steven mosher (Comment#49649) July 22nd, 2010 at 5:46 pm

Well Bugs,

Christy and Spencer shared their ‘code’ with RSS, worked closely with RSS to get the issue ironed out.

Meanwhile, Jones and company whispered about the correction behind the scenes and tried to take the greatest advantage of it.

Dont you read the mails.

In contrast, Jones wrote to Mann and explained that he knew WHY Mcintyre could not replicate the results. It came down to a step,apparently, made in the code, but, apparently, undocumented in the paper.

So. The difference is NOT in making mistakes or not. All science is a “mistake” The difference comes down to transparency and openness.

Openess and transparency is a two way street. Scientists generally have an open policy, and there is plenty of evidence in the emails of co-operation. For some reason, they didn’t feel like co-operating with someone who repeatedly abused and pilloried them publicly. I think there is a lesson in there.

bugs,

“Openess and transparency is a two way street.”

No it’s not. It the case of Science, it’s a one way street… from The Scientist out in all directions.

Unless of course, you’re a “Scientist” who doesn’t want people to know what he’s done.

Andrew

Lucia,

To be fair to the point some critics of the surface sites make, any reasonable computation based on measurements available gets more or less the same result. But at some critics worry about distortions

Yes, some critics worry about the distortions that may exist in the underlying data. But many also have expressed concerns that the signal in the underlying data is distorted due to the mathematical processing. Perhaps with malicious intent, no less. I am simply making clear that the latter criticisms definitely exist, and therefore it is worth noting that everybody gets the same results. I will make this clear anytime somebody says it’s completely irrelevant or unsurprising that everybody gets the same results. No, that does not in itself speak to worries about flaws in the underlying data, before anybody starts doing any math. But you can’t be expected to answer all questions with one sentence.

Andrew_KY (Comment#49661) July 23rd, 2010 at 8:02 am

bugs,

“Openess and transparency is a two way street.”

No it’s not. It the case of Science, it’s a one way street… from The Scientist out in all directions.

Unless of course, you’re a “Scientist” who doesn’t want people to know what he’s done.

Andrew

There are more rules in science than just being open, not abusing people and personalising issues is one of them. You can’t say the scientists have to be open, then break other rules. Try to get a paper published that involves juvenile taunts and puns on people’s names, it will get rejected every time. Just as a matter of common courtesy and maturity, people like Mcintyre should just grow up any way, he’s more than old enough now.

bugs,

“There are more rules in science than just being open, not abusing people and personalising issues is one of them.”

“maturity”

I agree any Scientist who withholds his Science because of personal issues or politics, needs to grow up/ Put His Big Boy Pants On, and help increase our knowledge by sharing all the information about the good work that he has done.

Andrew

Andrew_KY (Comment#49665) July 23rd, 2010 at 8:17 am

bugs,

“There are more rules in science than just being open, not abusing people and personalising issues is one of them.”

“maturity”

I agree any Scientist who withholds his Science because of personal issues or politics, needs to grow up/ Put His Big Boy Pants On, and help increase our knowledge by sharing all the information about the good work that he has done.

Andrew

Yeah, that’s about what I’d expect. They made rules about not abusing scientists and harassing them years ago for a good reason. You can see the effects of ignoring that rule at work right now.

Andrew_KY (Comment#49665) July 23rd, 2010 at 8:17 am

bugs,

“There are more rules in science than just being open, not abusing people and personalising issues is one of them.”

“maturity”

I agree any Scientist who withholds his Science because of personal issues or politics, needs to grow up/ Put His Big Boy Pants On, and help increase our knowledge by sharing all the information about the good work that he has done.

Andrew

McInytre doesn’t get to pick and choose the rules he wants to follow. If you want to play science, you don’t victimise people and pillory them.

“McInytre doesn’t get to pick and choose the rules he wants to follow. If you want to play science, you don’t victimise people and pillory them.”

bugs,

I think the Rule of Maturity should apply to everyone in Science. If the rule gets broken, it’s up to the rest of Science to maintain that maturity, and make sure the Science is done properly and with openness and transparency. Otherwise, the system of openness and transparency might break down. And we wouldn’t want that.

Andrew

Getting lots of server errors when trying to converse with bugs, ahem – FYI.

Carrot

I will make this clear anytime somebody says it’s completely irrelevant or unsurprising that everybody gets the same results. No, that does not in itself speak to worries about flaws in the underlying data, before anybody starts doing any math. But you can’t be expected to answer all questions with one sentence.

Fair enough. But obviously, the relevance of getting the same result addresses accusations of people trying to distort the results of computations based on the existing data and not some issues related to the quality of the underlying data themselves.

And… some people are going to point out that some skeptics focus on the latter appropriate concern.

With respect to engaging Andrew_FL specifically– and that’s the person you seem to be discussing things now– you might want to at least address the issue he is talking about rather than the issue some people discussed somewhere else. That will tend to keep things on a bit more even keel. (Mind you, I go off into having arguments with “people in comments at other blogs” too… but then I have to remind myself.)

The surface analyses use largely the same data, and they more or less agree with each other. The satellite analyses also use the same data, yet their differences are much larger. Not only the difference between RSS and UAH, but also between UMD and UW (although they don’t have a TLT product).

The surface data is made up of thousands of measurements from thousands of instruments. Temperature measured in Russia is independent of temperature measured in Kansas. By contrast, there have been only a few satellites that have measured tropospheric temperatures, and the simple act of stitching them together requires very large offsets and corrections. The lower troposphere data that everyone focuses are not even actual measurements. They are derived from other channels. The trends extracted from satellite data depend very heavily on the assumptions of the various scientists. The trends contained within the surface data are there no matter how the data is processed.

For some perspective, the 2005 UAH diurnal correction was about 50 times larger than the GISTEMP Y2K error. This is based on the area (“degree months”) between the original and corrected curves.

Lucia,
I’m going to make the significance of the agreement between GISS/CRU/NCDC/ccc/Ron/Mosh/Zeke/Chad/Tamino/whoever else clear, when somebody says it has no importance at all. Or when somebody expresses irritation that people emphasize this point. Because there is a context where it is important to emphasize this point.

Now, if somebody makes the more specific point that the agreement has no importance towards the specific question they are discussing, then that is totally and perfectly fine. I would not object. But that context has to be clear.

That is all.

Carrot–
Fair enough.
Andrew_FL–
Now for you. What context did you intend?

All I meant was that I found it to be a “duh” personally and was hardly surprised or particularly interested in the results. I would have predicted ahead of time that the results would turn out similarly. I never meant “Duh, nobody anywhere ever in any way, no matter how ill informed, has ever made that argument”. How could I? Someone is bound to have done so. And indeed I know some have. But it’s like, some have also argued that they can create perpetual motion devices. Why do I care when I know that it can’t be correct? If someone said, “I proved this to be incorrect” and I said “Duh” I simply mean that I never took the supposed device seriously in the first place. Not that nobody had.

It’s like bugs “taking me to task” for “not according Mann, Jones, etc” with “respect”…Except I’ve never disrespected them in any way. In fact, he should produce an example or apologize. He simply assumed I had disrespected them because, well, some of his other “enemies” have done so. I don’t particularly like to be pigeonholed. And I don’t like to be lectured about things I already know well enough (like, say, the definition of the Holocene…).

Now, I can understand that you don’t mean to try and pigeonhole me, but from my perspective that is how it looked.

Andrew,

It goes beyond what some sceptic somewhere has said. Anytime you have some complicated analysis, it’s worth noting that other people can independently reproduce it, even if they’re working from mostly the same data. You might note that this has not been the case with UAH/RSS, which is what we’re talking about here anyway. Similar issues have arisen when putting together satellite TSI measurements.

And then you bring in the sceptic brigade, saying there’s fraud in adjustments and station selection and whatever else.

It appeared to me that you were complaining that people make a big deal about this agreement. Well, it doesn’t surprise me either, but there are very good reasons why people make a deal of it. So long some/many sceptics continue insinuating processing fraud, you will continue to hear people making a big deal of it, in response. Maybe it bores you to hear it, but that’s why you’re hearing it.

Anyway, going back to the point: as of July 2010, UAH and RSS agree with each other pretty well, much better than they did years ago. But there remains a difference between the two. An honest person would chalk this up to unresolved methodological questions, and cite both sources.

Somebody else might just pick one and ignore the other, if that selection helped them make whatever point they were trying to make.

Re: carrot eater (Jul 23 10:17),

Anytime you have some complicated analysis, it’s worth noting that other people can independently reproduce it, even if they’re working from mostly the same data.

I agree with CE. I would not have predicted the level of concordance among the analyses of GISS/CRU/NCDC/ccc/Ron/Mosh/Zeke/Chad/Tamino/whoever else. As Zeke has pointed out numerous times, this shows that the anomaly methods really are robust. Now the focus can shift to the data issues themselves–without backtracks of, “Aha, but what about the methods?”

This is genuine progress.

carrot eater (Comment#49675)-To the extent that there are differences between RSS and UAH, I don’t really see them as rising to the level of saying that some serious gulf exists. Are the differences statistically significantly different from white noise, for instance?

If we are going to have a standard of what constitutes sufficient agreement, that seems to me to be a good one, and I don’t think that the differences between RSS and UAH rise to that level. But I would be interested if someone could show me how otherwise.

Now, as to why slight differences exist between the two, John Christy has certainly taken interest in this, as we might expect for someone who examines the data he uses-close scrutiny and attention even to slight, statistically unimportant detail. The difference appears to mainly be associated with the transition from NOAA-11 to NOAA-12, where RSS appears to “jump” relative to UAH. Now, one might say, this could be a UAH problem. Several things, however, suggest that it is not. (is is worth noting that a downward drift has crept into RSS relative to UAH during the recent period where AQUA is in use (along with NOAA-18 just recently) by UAH, and this drift mitigates the difference between the two. In this case, RSS is almost certainly doing something wrong, since AQUA does not require the adjustments they have been using for their different satellite. My understanding of the orbits and the corrections lead me to believe that this would probably meant that, during the earlier transition, RSS warmed too much, so that supports the idea that the problem is with RSS) For one thing, RSS increases not just relative to UAH but all the balloon data sets . Another indication is that this same discontinuity is found between RSS and surface datasets. I could discuss this further, but I’d like to get to something else.

Another point about the differences between the two reminds me of the much discussed differences between GISS and HADCRUT that arise from GISS’s interpolation over the Arctic, I think (if Zeke is reading, he can correct me on this ). Well, the RSS and UAH domains are not identical, either-RSS snips out data near the Antarctic, for fear that sea ice will contaminate the signal (don’t ask me how, I don’t remember). According to John, this makes the difference between the two datasets smaller when it is accounted for, because the small area that UAH includes is marked by cooling. So the differences aren’t even as large as they appear.

Also, a while back, John said this:

“The recent BAMS Climate Summary of the Globe 2008 indicated global LT trends of +0.14 C/decade +/- 0.02 C/decade with the range representing the full range of all of the versions of tropospheric data sets available to us.”

While the +-.02 C per decade range is probably larger by some amount than the 95% confidence intervals would be, so is the difference between RSS and UAH when comparing them on exact apples to apples basis likely to be smaller than .02 degrees C per decade. Which is smaller? Hard to say, but I suspect that the differences between RSS and UAH are at best marginally significant.

Andrew,
UAH and RSS are further apart than GISS and CRU.
Looks like 0.137 C/dec, 0.161; 0.158 and 0.166.

Before accounting for autocorrelation, the 95% confidence intervals do however overlap each other, if that’s going to be your bar. So they are in a broad sense consistent, but they could definitely be tighter in.

As you’ve mentioned, what discrepancy there is between GISS and CRU has been definitively identified and quantified; it becomes easier to deal with. We don’t know what is “truth”, but we know the issue comes down to lack of observations in the high arctic. Knowing that adds a comfort level.

It’d be nice if that were that were true of UAH/RSS; instead of being a black box difference. I never really knew, but I always had the impression that it came to how to merge satellites, and you appear to support that. I’ve not looked at this at all before, but looking at the residuals, and the slopes on either side, it certainly looks like something happened around 1992, and that does look like the time frame of transition from NOAA11 to 12. So that checks out, from what you said.

There may also be a similar issue of spatial coverage, you say.

OK, I would love to see it if somebody’s done a definitive calculation here – meaning, clearly show that if you take UAH’s approach but use RSS’s area and their way of merging those two satellites, or vice versa, you get the other result. With the different choice of satellite of late, that’d be another thing to swing back and forth, to see how it mattered.

UAH and RSS correlate with each other very well. It’s just the difference in the long term trend between the two that comes up. That’s not surprising, really, given the possible reasons for the difference.

It’s definitely worth looking at different ways to combine the satellites together, different ways of doing various adjustments. For instance, RSS uses one particular climate model to estimate the diurnal cycle for correction purposes. Not only could one question whether simulating the diurnal cycle instead of measuring it is a good idea (UAH certainly does) but if we are going to estimate a diurnal cycle that way, there are tons of different models that could potentially estimate it.

I know, incidentally, that RSS has validation comparisons with radiosonde records which are subsampled. The problem is they are sub-sampled to the radiosonde’s domains, not the satellites, since the radiosonde domains are more limited.

With HadAT sampling, RSS has .216 K per decade, UAH has .203 K per decade, (HadAT incidentally, .202 K per decade), ROABCORE sampling RSS .210 K per decade, UAH .209 K per decade, (ROABCORE has .198 K per decade), RICH sampling RSS has .215 K per decade, UAH has .205 K per decade (RICH .219 K per decade), and IUK sampling RSS has .209 K per decade, UAH .185 K per decade (IUK .175 K per decade). So coverage does matter, and in many cases this may lead to a large amount of the difference.

Andrew,
Are those sonde sample domains integrated over some range of altitudes that you’d expect the LT channels to be sampling?

How good do you think they are now at isolating the altitudes they’re targeting? I remember one of the early issues was whether they were picking up part of the stratosphere. And to this day I think, some groups that do other satellite work don’t attempt a lower trop record, I think because they are wary of such problems.

carrot eater (Comment#49756)-I’m not clear on exactly what it is that RSS does to get the radiosonde records they compare with, although my point was mainly about sampling and the satellite records.This is RSS’s page on the subject:

http://ssmi.com/msu/msu_data_validation.html

But I think they know better exactly what they did than I do.

I will be stunned if July 2010 does not come in with the highest July anomaly on the UAH record now.

It seems to me that March 2010 with 0.66 already HAS set a record; March 1998 was only 0.53. So if somebody (S.Goddard) says that 2010 will set NO record, it seems obvious to me he cannot be talking about monthly or daily data, but the average for the whole year.
July 2010 is currently at about 0.66, so the record for one month will probably stay with February/April 1998 (0.76). But it will be the hottest July.

Lucia

I was updating a plot of the Channel 5 data up until June, when UAH changed their data file layout.

I’ve restarted my daily updates with the new Channel 5 data file, link.

I’m open to suggestions on how to make the daily UAH Channel 5 chart more informative.

Opps

Here’s the link

http://processtrends.com/image.....latest.png

Hi Kelly-
It looks informative. I’m assuming you want to highlight how todays value compare to temperatures over time and aren’t particularly trying to highlight how this month (i.e. july) compares to other july’s. The july/july comparison shouldn’t matter, but the discussion of an annual cycle have circulated on blogs. So, I like to highlight that. I don’t think that makes my layout any more informative– it’s just a choice.

It’s just a bit difficult for people with old eyes (like me) to read think lines and small type on a monitor.

Even if I click to enlarge, I can’t read the little inserted daily graph at all. But people with better eyes often think smaller types and thinner lines are more elegant, and that’s what you’ll see in journals. So, once again, that’s a choice. Over time, I’ve had lots of visitors request thicker lines and bigger symbols, so I use thicker traces for my lines.

Over time, I’ve also discovered that a fair number of guys visiting the blog are red/green color blind. So, you might want to avoid having people compare red to green.

Overall your graphs look nice though. So, just organize them to show whatever it is you want to discuss in your narrative.

Lucia

Thanks for the feedback. I’ve incorporated several of your suggestions.

I find the daily satellite data for July very interesting. The satellite data measured on channel 5 is considered the most reliable of the lower troposphere measurements, and sometime in July each year, the global average temperature from channel 5 hits its annual high for the year. But this year the channel has hit an all-time high: and not just once, but as of July 28, nineteen days have exceeded the highest temperature ever recorded on channel 5. The last fourteen days average almost a tenth of degree hotter than the highest temperature previously recorded.

The chances of getting 19 record highs in a single July in a random system are extraordinarily low; less than one chance in a thousand trillion, times one chance in a thousand trillion (less than 1 in 10^31 attempts would result in 19 record highs in 31 attempts, when the 930 days over 30 previous Julys failed to produce an equally high result.)

Clearly the lower troposphere is measurably hotter than at any point in the last 30 years. So much for ‘global cooling’, we can dismiss that as an imaginative fabrication and misuse of data. This evidence also casts a great deal of doubt on the ‘luke warming’ theory. We have obliterated the record high in the troposphere without a strong El Nino and near the low end of the solar cycle. What happens when we get another strong El Nino that coincides with the stronger portion of the solar cycle?

July has been a scortcher to be sure, dont think it will take the all time anomoly record on this dataset though. Could be wrong mind.

Id be real interested if anyone can point me to a bit of an ‘after action’ report on where and what caused the July UAH anomolies. Events like this need a proper debrief.

Re: dorlomin (Aug 2 03:36),
There will be a NOAA Monthly report for July.

Re: Paul K2 (Jul 31 15:21),

We have obliterated the record high in the troposphere without a strong El Nino

I don’t know what planet you’re living on, but the current El Nino on the one where I live looks pretty strong to me.

Nick– I also doubt this will be an all time anomaly record. It will probably be an all time july record. But we’ll see.

PaulK2– This was a pretty strong El Nino looking at things like MEI and ONI. It’s not 1998, but still, pretty strong. Of course, it all depends on how you use the adjective “strong”. Do you mean: Record breaking MEI’s? Top 5%? 10% Top third? Half?

For some reason, you keep focusing on absolute temperatures and then drawing conclusions about warming. In absolute temperatures, July is always warmer than January. If we hit a record low or very low temperature in January and someone decreed this evidence of cooling, you or others squawk about this and they’d be justified. It’s equally silly to notice July is hotter than January and make claims about records that tell use something about a warming climate. The anomalies are still high, and they are high even for July. But they have been higher in the past.

Re: Paul K2 (Jul 31 15:21),

The chances of getting 19 record highs in a single July in a random system are extraordinarily low; less than one chance in a thousand trillion, times one chance in a thousand trillion (less than 1 in 10^31 attempts would result in 19 record highs in 31 attempts, when the 930 days over 30 previous Julys failed to produce an equally high result.)

Is that calculated using independent, identically distributed statistics? If so, then you’re way, way off. Daily temperatures are highly autocorrelated. If the statistics are Hurst-Kolmogorov, it’s even worse. Off the top of my head, if you set one record high, the probability of another in the same month is actually quite high.

More adjustments to hide the incline! The AQUA-15 satellite clearly showed this July to be the warmest on record, running around .6C above the 2002-2009 mean. And yet the reported anomaly is only .49C above the 1979-1998 mean! And Dr. Spencer is reporting that they integrated yet another satellite (NOAA-18) into the dataset to get this value.

Also, if I’m reading his comment from this morning correctly, these satellites only make a single pass during the early afternoon. Considering the surface data shows overnight minimums warmer more quickly than daytime highs, as predicted by AGW theory, this omission of morning temps could explain why the satellites show a little less warming than the surface data.

The anomoly came in at 0.489

http://www.drroyspencer.com/20.....-49-deg-c/

Interesting UAH result of 0.489. I guess we have a long way to understand how the UAH model works, given the very high raw data points.

Meanwhile, the other satellite measurement system came in with an anomaly of 0.608, only slightly higher than the previous highest July reading (July 98) of 0.606, a statistically meaningless difference.

Can somebody explain why we should have a lot of confidence in these two readings? How is that the two models get such different results from the raw data? I am really interested to hear the explanation.

PaulK2–

I guess we have a long way to understand how the UAH model works, given the very high raw data points.

You keep looking at the raw temperatures and then wondering why anomalies are not high. . The july anomaly is obtained by finding computing the average for july and subtracting the july average from the baseline years. This is designed to take out the annual cycle. July is near the top of the annual average cycle, so that means average july during the baseline is also high.

If you look at my anomalie graph posted in the update and missing only the final days of July, you’ll see that the reported UAH anomalie is pretty close to the value you get by just averaging over the daily values.

The general answer to your final questions are complicated. But it seems that some of your concerns stem from your not understanding the difference between anomalies and raw temperatures. To extent that your reservations might be based on this misunderstanding, we first need to get you to understand why anomalies are used.

Then after that we can try to discuss whether we think there are difficulties with the satellite systems more generally and what we think about the meaning of these difficulties.

Maybe I don’t understand anomalies, particularly if differences exist between the the way anomalies are calculated in the satellite records versus the HadCRUT and GISS temperature records. I would like to step through a general understanding of how the satellite global temperature anomalies are measured and calculated, in a series of comments.

Do you want me to do that on this thread, or would you prefer a different post? I am just giving you the option, before I dive in.

the the way anomalies are calculated in the satellite records versus the HadCRUT and GISS temperature records.

The basic method is to pick a baseline, and the average temperature for each month in that baseline and subtract. For reasons having to do with deployment of instruments, the details of how they do this varies. For the surface measurements, it’s a big complicated because they never actually determine the raw temperatures, they only get the anomalies. (This is mathematically possible.)

I have discussed some of this in other posts. But before I send you to one, do you at least understand this general idea:

1) Meausure temperatures for 20 years you chose as the baseline.
2) Compute the average july temperature over days July1-31 for each of the 20 years.
3) Compute the baseline for july by averaging over those 20 years.

That general process gets you the baseline for july. You can do the same for Jan, Feb, March etc. (You will discover that July is hotter than January in raw temperatures.)

Then, to get the July anomaly for any individual year, N: Measure the temperatures on july 1-31 during year N. Compute the average over july 1-31 measurements. Subtract the July baseline. This is the july anomaly for year “N”.

Does that more or less answer your question?

This gives you the anomaly.

Lucia wrote: “If you look at my anomalie graph posted in the update and missing only the final days of July, you’ll see that the reported UAH anomalie is pretty close to the value you get by just averaging over the daily values. ”
—————————————

Did I miss something? The only one I can find is at the top of this post, which only runs through July 13th, which was before the spike in mid July that provided real separation above previous record highs. There is no doubt that the raw data ran at or above previous highs for the entire month of July.

Er … I just opened http://discover.itsc.uah.edu/amsutemps/ and find that it has changed. Now it shows the previous record highs higher than this July for the first half of the month, and July 31st is back down in a tie with record high as opposed to .15°C above as it was earlier today. Well, I guess we just have to accept that everything coming out of those offices is a work in progress and subject to change at any time.

As the graph stands now, I’m not surprised that the final number comes in a bit below the previous record. As it stood just this morning, I was very surprised, since it appeared we were on course to beat the old record by .05-.1°C or so.

BenjaminG

Er … I just opened http://discover.itsc.uah.edu/amsutemps/ and find that it has changed. Now it shows the previous record highs higher than this July for the first half of the month, and July 31st is back down in a tie with record high as opposed to .15°C above as it was earlier today. Well, I guess we just have to accept that everything coming out of those offices is a work in progress and subject to change at any time.
= = = = = = = = = = = = = = = = = = = = = = = = = ==
I had the exact same thing happen and thought I was losing it.

BenjaminG
Well here is a screenshot taken earlier in the month
http://rankexploits.com/musing.....annel5.jpg

Yes, Dolomin, that’s just before the temps spiked in the middle of the month.

If Lucia’s graph above is from the graph on the amsutemps site, then the final anomaly should have been significantly above the 0.533 shown. Obviously Dr. Spencer is doing lots of adjustment, we just don’t know what.

Benjamin– No, I missed something. I scrolled down, saw the lower graphs, and experienced a ‘lack of coffee’ moment.

I’ll grab the final data and compare Roy’s number to the average over all values in Aqua. Mind you, I think the algorithm never was based on a single channel. Aqua is one component and previously the “channel 5″ was one component. But that’s the data we have, so we watch that to get a best guess.

Lucia, OK. lets look at the use of anomalies in the different sets of temperature data.

The sets that collect surface temperatures using thermometers, actually don’t collect temperature data and calculate an average temperature, as you pointed out in your comment. They actually compute an anomaly for each measurement point, then use a grid system to collect the anomaly data into a global anomaly calculation, as the very detailed posts on this site discuss. The key advantage of this system, is the large variability of temperatures from different stations based on local terrain, environmental factors, elevation etc., that it make it very difficult to calculate an average temperature for a given area (say the State of Colorado, or the 48 contiguous US states, or the Mediterranean Sea, or the African continent). The use of anomalies allow us to examine temperature changes, without having to calculate a mean temperature.

But the satellite data is measured and collected differently. I am not well versed in this, but will try to summarize correctly. The satellites measure infrared from oxygen molecules in the atmosphere using a scanning detector on the satellite as it passes over the Earth. Since the infrared entering the detector may have originated from oxygen at different altitudes within the atmosphere, the measured signal must be processed to calculate the signal from the lower troposphere using a model. Once the relevant signal has been determined, the infrared signal is used to calculate a temperature, and these temperature readings are averaged over the area scanned to determine an average temperature for that portion of the troposphere. (Alternatively, the signals may be averaged for that area of the earth’s lower troposphere, then converted to an average temperature.)

As a result of this process, the satellite data include absolute temperature data that can be averaged for different areas of the troposphere, prior to calculating anomalies. Temperatures are typically determined for the Tropics (20S to 20N), the portion of the Northern Hemisphere that can covered (the satellites have difficulty measuring IR if the surface is covered by ice, so generally the polar regions are excluded), and the Southern Hemisphere, as a well as a global average troposphere temperature.

Once these temperatures have been calculated, then the anomalies are calculated. As you described, the baseline period average for the relevant region or global average is subtracted from the temperature readings to get the anomaly (either positive or negative) that compares the current period (month) with prior periods (month).

However because the satellite collect different data, there is a wealth of information in the data that could be used to answer questions the other temperature records can’t address. For example, we could ask the question if the troposphere rose to record temperatures sometime in July, that have not been recorded before. If the satellite data cover enough of the Earth’s troposphere every day, then calculating daily mean temperatures for the global troposphere, or areas of the troposphere may be possible. Since the number of individual data points would be much higher than the anomaly data, these questions could be answered with a much stronger statistical certainty, or to a higher accuracy, then is possible with the anomaly data.

For example, the Aqua satellite clearly measured troposphere temperatures in July, that had never been measured before. This information was lost in the averaging process. It is possible that the daily data posted on the internet site hadn’t been adjusted for measurement issues, or globally adjusted, or was offset by data from other satellites when the anomaly was calculated. To illustrate, both the UAH and RSS data show the NH anomaly statistically tied with the July 98 reports (UAH has Jul 2010 at 0.65 with Jul 1998 at 0.67 and RSS has Jul 2010 at 0.919 and Jul 1998 at 0.910). If the Aqua satellite spent more of the month over the NH, then the daily numbers may be reflecting the higher readings from the NH, which ended up getting adjusted out in the final anomaly.

Ok… I will continue in the next comment.

Again please correct me, if I get this wrong, but statistics involve types of statistical tests that can define the probability of a given event. One statistical test using time averaged means may show a 95% probability, but other statistical tests of individual data points, may answer a similar question covering a different (often even shorter) time period with a probability exceeding 99% or even higher.
.
Let me use the Dow Jones Industrial Average (DJIA) as an example. If the question was, “what was the highest close for the DJIA in 2009?”,
and if the only data available was the monthly average DJIA, then the ability to predict the highest close is rather poor, with an uncertainty of over 1000 points. If the weekly average was available, then there still is an uncertainty of at least several hundred points, but if the daily close information is available, then there is no uncertainty at all in the answer.
.
So looking at the July 2010 and asking the obvious question: Did the global troposphere in July 2010 reach temperatures hotter than at any previous time in the satellite record?
.
Looking first at the monthly global anomaly data, the UAH data seem to say NO or MAYBE, with an anomaly of 0.489 versus 0.52 in July 1998, while the RSS data seem to say likely YES (with a chance the answer is MAYBE), with a monthly anomaly of 0.653 versus 0.606 in July 1998.
.
But if daily mean absolute temperatures as measured by the satellites were used, the question could be answered with greater certainty. In the same way, the question: Did the NH troposphere hit record high temperatures in July 2010? could be answered more definitively. It seems to me, that if the satellite troposphere temperature data were processed more effectively, there is much more information that could be gleaned from the data.

PaulK2

It seems to me, that if the satellite troposphere temperature data were processed more effectively, there is much more information that could be gleaned from the data.

Maybe. What question to do want to learn the answer too? If you can answer it using that data, go ahead and process it to get that answer.

Other than that, I read a lot of details, but I don’t know whether you were trying to make some sort of actual point. Are you disputing something someone said? Trying to clarify for someone? I’m not seeing any particular connection to the previous conversation.

the UAH data is a complete mess.
.
where are the auditors? ah, forgot. lower tempertures get a “pass” from them everytime…
.
i was shocked by this statement by Spencer:
.
ALSO…we have added the NOAA-18 AMSU to the data processing in v5.3, which provides data since June of 2005.
.
http://www.drroyspencer.com/20.....-49-deg-c/
.
so the hottest months on records were those in 1998 WITHOUT the NOAA data added in. and they are comparing it to 2010, WITH that data added in?
.
this does not make any sense at all!

a main difference betwen the old AMSU graph (lucia’s picture above) and the current one (could someone please post a screenshot for comparison?!? lucia?) is massive.
.
the picture obviously factors in 2010 for the maximum. this is indicated by max and 2010 following exactly the same line for some periods.
.
the current AMSU graph does NOT factor in 2010 for maximum. this is demonstrated by the 2010 line being ABOVE the max for significant periods.
.
i do not know, why they change between those two ways of displaying it. (both make sense, but jumping betwen them does not!) it just does not increase my convidence in the UAH numbers.

ah, Roy has given some answers n the comen section to the July data.

Looks like Dr Spencer just shifted the whole of 2010 down by about 0.2C… I wonder why?

Measuring temperature by satellite is a recent thing and a work in progress. There will be more and better satellites, and improved algorithms. NOT using these would be really shocking, as would be not using a new, different GPS- or WX-satellite.

And ch5 is not the same thing as “troposphere”.

I was wrong about July 2010: July 1998 remains the hottest July on UAH-record.

Measuring temperature by satellite is a recent thing and a work in progress. There will be more and better satellites, and improved algorithms. NOT using these would be really shocking, as would be not using a new, different GPS- or WX-satellite.
.
well, adjustments and new methods are needed. but “sceptics” would find such adjustments in one direction and with obvious flaws dubious.
.
they don t attack Spencer, because they like the outcome (lower temperatures).

Looks like Dr Spencer just shifted the whole of 2010 down by about 0.2C
.
well, i would say the shift is bigger than 0.1 for july. (actually looking at the v5.2 v5.3 comparison i posted far above, we would not expect any change to July AT ALL!!!!) for the whole year, i am not so sure.
.
but it is obvious, that the changes were a game changer. in january and february, UAH was very hot and sort of following the AMSU data as we saw it. 2010 was on serious track to break records in the UAH set.
.
after the changes a record became unlikely.
.
but second hottest year on record would silence some of the “cooling” fanatics, at least if they were open to reason….

“sod (Comment#50396) August 4th, 2010 at 3:55 am
“sceptics” don t attack Spencer, because they like the outcome (lower temperatures).”

But you could do it? Of course you would have to first inform yourself about what he is actually doing, and why, and not just on the result.

sod–
You don’t know that this adjustment will lower temperature for the year. Before the adjustment, Deep Climate was complaining that the method resulted in a large annual cycle which had to be wrong. Evidently, Spencer and Christy agree. Since the issue is annual cycle, we obviously cannot even begin to guess whether the change results in a lower values over all until we see at least a year worth of data.

It looks as if March will be the hottest month of 2010, and we can now apply Goddard’s crest-to-crest-principle with more confidence:
The linear trend for UAH-data from February 1998 to March 2010 is 0.0003°C/month. One cannot infer a significant rise in temperature from that.
In a few month we will see if the trough-to-trough trend is slightly positive or negative.

Lucia,

“Deep Climate was complaining that the method resulted in a large annual cycle which had to be wrong. ”

Well, it was at the very least ‘odd’. Especially as the same cycle wasn’t in RSS or the surface records. Didn’t RSS fix that up for UAH though?

Alexej

“Goddard’s crest-to-crest-principle with more confidence:
The linear trend for UAH-data from February 1998 to March 2010 is 0.0003°C/month. ”

This is pure genius…

Nathan

Didn’t RSS fix that up for UAH though?

I don’t think so. Why do you think this?

Alxej–
What’s the advantage of the crest-to-crest principle over least squares? If it is better, wouldn’t it be good to find a series of crests– say 5?

Lucia

I had the impression a couple of years ago RSS had helped UAH identify a yearly signal in their data… But looking at the new Ch5 ‘average’ trace the big bulge at the end of Jan – early Feb is gone… And the bulge that was around Oct.

So I think I must be thinking of some other correction…

Nathan–
We are talking about change in the past few months, not years ago. Whatever change RSS may have helped UAH out with, that resulted in the product with an annual signal because it’s the one that UAH changed recently.

Crest-to-crest principle? I really hope that’s a joke.

http://deepclimate.org/2010/03.....t-uah-5-3/

This seems to indicate they told John Christy of the problem a couple of years ago, and what we see now is the correction.

Chris–
I suspect the “crest-to-crest” notion is designed to overcome objections that someone started an analysis in a trough and ended in a peak. So, “crest-to-crest” or “trough-to-trough” might be useful as a reality check when arguing about least squares.

But still, I don’t think this is any recognized principle as a superior way to do anything. Also, there are good reasons to expect 1998 was a particularly strong peak in temperature based on things like MEI and ONI. So, obviously a “really big outlier crest” to “fairly strong crest” trend estimate is not “better” , than least squares. Certainly, it’s not “better” than least squares corrected using MEI or ONI.

Nathan-
DeepClimate’s use of the passive voice sure makes that update uninformative, doesn’t it? Do you have any clue who might have told Spencer or Christy about some problem in 2008? I’m not motivated to hunt through comments for that tid-bit.

But you could do it? Of course you would have to first inform yourself about what he is actually doing, and why, and not just on the result.
.
well, i did point out an obvious problem. (change of data in 2005, comparison with 1998)
.
i asked Spencer for a reply. i asked people here, about the v5.2 data. (still can not find it)
.
let us see.
.
sod–
You don’t know that this adjustment will lower temperature for the year. Before the adjustment, Deep Climate was complaining that the method resulted in a large annual cycle which had to be wrong. Evidently, Spencer and Christy agree. Since the issue is annual cycle, we obviously cannot even begin to guess whether the change results in a lower values over all until we see at least a year worth of data.
.
i disagree. the deep climate graph shows a long term roblem in temperture betwen april and august. it suggetst serious pwards corrections ae necessary for those months. what will the rest of the year do to to spring/summer data?
.

sod:

they don t attack Spencer, because they like the outcome (lower temperatures).

Or, because the result represents a higher altitude, you’d expect the trend to be lower.

I actually have an objection with the TLL RSS trend showing too close a value to the surface measurement. The physical effects are different within the atmospheric boundary layer than above it…

Lucia

Not sure if either of these people have anything to do with RSS but…

from DeepClimate
“Tamino’s first post came on October 21, 2008 on the matter was entitled simply UAH and RSS. In that post, Tamino plotted the difference between the UAH and RSS monthly record and noted two anomalies, namely a “step” around 1992, and a strong annual cycle in the UAH-RSS difference over the last decade.”

then

“In comments, Richard Steckis said he had run the post [Tamino's blog post] past John Christy and that Christy had responded that the “step” difference was a problem in RSS, not UAH, citing a number of references.”

Later

“In July of last year, Anthony Watts discussed a high value in the NASA GISTemp June 2009 surface temperature anomaly, which struck Watts as unrealistic, especially compared to the UAH value for that same month.”

then

“Watts soon got in touch with John Christy, who acknowledged in a July 2009 “readme” file that the annual cycle problem should be addressed, as well as elucidating its source.”

So he was told of it a while back, and this is the result.

sod:

i disagree. the deep climate graph shows a long term roblem in temperture betwen april and august. it suggetst serious pwards corrections ae necessary for those months. what will the rest of the year do to to spring/summer data?

I happen to think that the satellite data are pretty much useless for anything shorter than annual, especially UAH. Annual trends should be reliable though, IMO.

On the other hand, a lack of one-year peak with RSS in this figure (nice hot linking, DC) doesn’t necessarily indicate that the problem is with UAH. How does DC “know” that it is not RSS? Annual peaks are pretty pervasive (I agree UAH is suspiciously large).

Even GISS has issues with coherency at shorter time periods:

See e.g.. this.

Carrick

“Or, because the result represents a higher altitude, you’d expect the trend to be lower. ”

Hmm, yet Lucia tests it against 0.2C/decade like she does the surface temps.

Carrick

The answer to your question is in the same post – UAH needed to make a correction. They moved from 5.2 to 5.3

An annual 0.1°C is seen in other reconstructions too:

see this.

The lack of an annual peak could be the result of a “smearing” effective from the area averaging in some of the algorithms.

Carrick

This is what John Christy wrote about the source of the annual signal:

” It was brought to my attention by Anthony Watts that there has been some discussion about the noticeable annual cycle in the LT and MT trends when done by months. In other words, the trend for Februaries is on the order of 0.12 C/decade warmer than the trend for Mays.

The feature arises when the AMSU data are adjusted and merged into the MSU data stream beginning with NOAA-15 in Aug 1998, then carries forward with NOAA-16 and AQUA (both of which are AMSUs too). The process involves at one point the removal of a mean annual cycle in the anomaly differences from one satellite to another. It turns out that all satellites have a residual annual cycle due to each instrument’s peculiarities. In the end, all annual cycles are matched to NOAA-6 and NOAA-7 …

I’ve tested a number of alternate processing methods (basically versions of not removing the annual cycle in the difference time series from the first AMSU onward) and the range from the highest to lowest is reduced to just under 0.09 C/decade. This in effect establishes a new annual cycle for the AMSUs based on the first AMSU.

I think the magnitude of the annual cycle in the monthly trends is alegitimate problem to address. The range in the current v5.2 LT looks too large (about 0.12 C/decade) …”

That’s from Deep Climate’s post on it.

Nathan (Comment#50408) August 4th, 2010 at 7:06 am
http://deepclimate.org/2010/03…..t-uah-5-3/
This seems to indicate they told John Christy of the problem a couple of years ago, and what we see now is the correction.

The correction to remove the seasonal behavior was applied in March this year, version 5.3, now we apparently have a correction to the correction.

Nathan:

Hmm, yet Lucia tests it against 0.2C/decade like she does the surface temps.

Probably she should use a higher value, because the models predict a larger trend with increasing altitude.

The answer to your question is in the same post – UAH needed to make a correction. They moved from 5.2 to 5.3

I am aware of that of course. My point is a bit different. We know there is a systematic error in UAH. The question is “what is the right value” for the annual amplitude? It appears to me that GISS “gets it wrong too”.

Waiting for the apologists to jump in and say less than annual trends “don’t matter” unless we are describing the enemy’s work.

(Caveat emptor: I’m using a very different period for my spectral periodogram than Tamino used in his analysis.)

Phil

Perhaps. What I think is that the correction was only just applied to the graphic showing the AMSU Ch5 temp. Just yesterday (or the day before) I had looked at the graph and the ‘bumps’ around Jan and Feb and in Oct were still in the ‘average’ reading. They’re not there now.

Carrick

“It appears to me that GISS “gets it wrong too”.”

Really? The amplitude of the annual cycle is much lower..
I don’t what the two lines are in this:
http://farm5.static.flickr.com.....82897d.jpg

But it would seem to suggest it’s about the quarter of the size of the UAH one.

Sorry Carrick

I should have ended with:

How big do you think the annual cycle should be?

Nathan–
Deep Climate is anon, but rumor has it he’s a programmer (or similar ) from Canada. That means no affiliation with RSS. Tamino is not affiliated with RSS. I don’t think Steckis or Watts are affiliated with RSS.

So, your speculation about RSS “helping” on this appears misplaced.

Nathan–
No, I don’t “test” lower troposphere against surface trends. I just show a nominal value on a graph so people can get a rough gauge. The IPCC didn’t provide projections for TLT, otherwise, I’d post those.

We do have a paper in the endless review cycle that discusses satellite trends and compare those to measurements, but I’m not posting those yet.

“Nathan (Comment#50402) August 4th, 2010 at 6:41 am
This is pure genius…”

Anybody can copy the monthly data from the UAH website to Excel. The program draws the curve, calculates the trend, plots the trend and prints its formula. But yes, Excel is very, very good.

“lucia (Comment#50404) August 4th, 2010 at 6:49 am
What’s the advantage of the crest-to-crest principle over least squares? If it is better, wouldn’t it be good to find a series of crests– say 5?”

Excel does use “least squares”. “Crest-to-crest” stops the user from cherry-picking the start and the end the interval. If you use a series of crests, you get the trend for the series, e.g. for the last 60 years instead of the last 12 years. What we want to know here is whether you can use UAH-data to show a rise in temperature since 1998.

(Actually, it is not Goddard’s principle; it is what his critics told him to do when he calculated the trend of something like a sine from mean to mean for 8 periods. 8.5 periods would have been OK, peak to peak and trough to trough would have been OK)

Nathan, I apologize… I posted a figure without a legend… The cyan is global GISTEMP the magenta is NCDC.

The other figure is land-only, so the values should be higher (land surface amplification effects like winter snow cover, winter-time deciduous die back etc)

If I had to pick the “most likely to be right” for annual global and certainly the most systematic of any of the attempts, it would be NCDC. <a href="http://lwf.ncdc.noaa.gov/oa/climate/research/Smith-Reynolds-dataset-2005.pdf"Good reference here.

Whether it is correct in longer trend is a separate issue (mostly related to care in data handling and how well they address issues like station moves and UHI), but I trust their approach for the “high frequency” more than I do the purely ad-hoc cobbled together algorithms used by GISTEMP.

Lucia:

No, I don’t “test” lower troposphere against surface trends. I just show a nominal value on a graph so people can get a rough gauge. The IPCC didn’t provide projections for TLT, otherwise, I’d post those.

I sometimes wonder if the IPCC would have made projections, had the comparison of data to model been a bit more in agreement?

They cheated a bit on this part, if you ask me.

Alexej Buergin

(Actually, it is not Goddard’s principle; it is what his critics told him to do when he calculated the trend of something like a sine from mean to mean for 8 periods. 8.5 periods would have been OK, peak to peak and trough to trough would have been OK)

I thought that might be the case.

Lucia

“So, your speculation about RSS “helping” on this appears misplaced.”

Yes, I implied as much above, when I said that I didn’t know if any were part of RSS. However, it is clear that UAH was alerted to the problem in 2008.

lucia (Comment#50379)
August 3rd, 2010 at 11:45 am

Lucia

I read your comments re: calculating anomalies and agree with all you wrote. However, there seems to be an inconsistency between the reported anomalies and the raw temepratures. For example, the anomaly for July 2009 was +0.44 while the anomaly for July 2010 was +0.49 – a difference of +0.05. But the raw temperature average for July 2010 was more than 0.2 deg higher than July 2009 (unless I’ve done something horribly wrong).

I think I’m using up-to-date data. Also, I can’t recall a day when
when the daily AMSU reported that the current day was much less than 0.3 deg F (0.18 deg C) warmer than the same day in the previous year.