Anthony Watts makes a huge fuss about miniscule changes in atmospheric temperature data

A few days ago an astute reader wrote to me, asking if I knew why the RSS historical temperatures for the lower atmosphere varied month to month. I in turn asked Dr Carl Mears, who is the scientist behind the RSS temperature analysis (among other things).

To see what changes month to month, compare the charts for RSS TLT v3.3, which is just of the lower troposphere. The data is from that provided in November 2013, March 2016, July 2016 and October 2016:

Figure 1 | Lower troposphere temperature RSS v3.3. The chart shows data as reported in November 2013, March 2016, July 2016 and October 2016. Source: RSS

Do you see the big changes? No? So what is all the fuss about at WUWT (archived here) you might wonder.

This is what it's all about. Anthony's reader compared the difference between the data provided in September (to end of August) and that provided in October (to September). I don't have September data (to August) but I do have earlier months, so I can show the difference between the latest data and earlier data. The chart below shows differences over time: March 2016 minus November 2013; July 2016 minus March 2016 etc.

Figure 2 | Difference in lower troposphere temperature RSS v3.3 between selected months. The chart shows differences over time: March 2016 minus November 2013; July 2016 minus March 2016; October 2016 minus July 2016. Source: RSS

Before you gasp, look at the left axis. It goes from minus 0.02 to plus 0.04. To put that into perspective, the month of June this year was changed from 0.467 C to 0.485 C - hardly earth shattering and likely well within uncertainty.

Look up close. This next chart is the same as Figure 1 above, except it homes in on the period from 2009 onwards, so you can see the differences more clearly.

Figure 3 | Lower troposphere temperature since January 2009 - RSS v3.3. The chart shows data as reported in November 2013, March 2016, July 2016 and October 2016. Source: RSS

If you look very closely you'll see some very slight differences between the data sets provided. So what is the reason?

The reason for month to month changes

Anthony Watts didn't bother to enquire of RSS why the difference. As I said I did, when I was asked a few days ago. I asked Dr Mears why the difference and if the numbers were recalculated each month. Carl Mears kindly replied within a couple of minutes and said, yes they are. He explained:

You are right.  Each month the target factors and inter satellite offsets are re-calculated.  Because the some of the satellites I use for V3.3 are undergoing calibration drifts of some sort (see the V4.0 paper for my arguments about excluding NOAA-15 late in its life for TMT, for example) and because the diurnal adjustment is not perfect, each additional month of data tends to change the merging parameters a little bit as the regression "tries” to  remove these drifts.

The changes should be visible all the way back to mid 1998, when the NOAA-15 data started.  The MSU data is fixed and no longer changes since AMSU data is not used to determine the MSU merging parameters.

If you don't quite see what happens, perhaps a chart from an earlier article will make this more clear. The chart below shows the satellites that have been used over the years to estimate atmospheric temperature changes:

Figure 4 | The time scale at the bottom shows when each of the satellites were providing data. The rest of this caption is as in the paper: Ascending local equator crossing time (LECT) for each of the satellites used. The LECT drifts over time for all satellites except AQUA, METOP-A, and METOP-B, which are maintaining at constant local time by orbit keeping maneuvers. For the drifting AMSU satellites, the thinner lines denote the portion of the missions excluded in the MIN_DRIFT analysis. (That's different to the time excluded from v4.) Source: Meares and Wentz 2016

NOAA 15 was excluded from the latest version 4, but it's still included in RSS v3.3. TLT hasn't been migrated to v4 (yet) and is still on v3.3. In version 4 it was excluded from 2011 onwards.

Each month the whole data set is re-calculated. Each month the new data allows a re-calibration for satellite drift. That means that each month there will be changes that can affect every month before it for which a particular satellite is in operation.

This last month's data has bigger changes than previous months. Still very small changes and barely noticeable even when homing in on recent years (as in Figure 3 above). Earlier years included satellites that have since been replaced, so any drift corrections would probably have a bigger impact on the data from the most recent years. In other words, the more recent years will have greater changes because that's when the current (drifting) satellites are still in operation.

WUWT outrage

Anthony Watts is quite outraged. His conspiratorial headline was:

Remote Sensing Systems apparently slips in a ‘stealth’ adjustment to warm global temperature data

Instead of asking RSS about it, he asked Roy Spencer, who can't get his own data right let alone be expected to know what's happening at RSS. Anthony wrote:

I asked UAH scientist Dr. Roy Spencer about it today, showing him the data and he replied:
“We suspected they have a revised LT in the works, after they came up with a new MT.

 Which doesn't answer the question at all. Whether RSS is going to move TLT data to version 4 or not doesn't explain the month to month differences in version 3.3.

Following on from his conspiracy-ridden headline, Anthony went further and basically accused RSS scientists of fakery and fudgery, in true denier fashion. He wrote:

Of course, the unannounced LT adjustment discovered by Ablitt also makes the trend warmer, some thing that isn’t entirely unexpected given the remarks last year by RSS chief scientist Carl Mears:

What Anthony objected to was Dr Mears pointing out that one shouldn't rely just on satellite data to see how temperatures are changing. One should also look at surface temperatures. Not something that Anthony Watts has had much success at.

Anthony added, peevishly:

Mears uses the term “denialist” so there goes his objectivity when he feels the need to label people like that.

Yep, anyone who calls a science denier a denialist can't be trusted to call a trough a trough. Only fake sceptics who wrongly refer to themselves as "skeptics" can be trusted in the upside down world of climate conspiracy land.

PS Anthony Watts has learnt a new word: "execrable". He wrote: Readers may recall a video produced by the execrable “Climate Crock of the Week” activist Peter Sinclair that we covered here where the basic premise was that the “satellites are lying“. Peter Sinclair isn't execrable, except to science disinformers because he shows how bad their disinformation is. And of course the video's basic premise wasn't that the satellites are lying. Satellites are incapable of lying. The data produced is quite difficult to analyse for various reasons.

Added by Sou: 7:54 pm AEDT, 11 October 2016

UAH changes over time

Now it's not fair to compare the UAH changes between versions with RSS changes for the same version, however UAH doesn't seem to do a whole recalculation each month to adjust for drift in the same manner as RSS. The charts below do put the changes in RSS into some perspective.

The first chart compares three different versions of UAH. You can easily see the difference. Compare the chart below with Figures 1 and 3 above.

Figure 5 | Lower troposphere temperature UAH versions. The chart shows data from version 5.2, 5.6 and beta 6.05. Source: UAH

This next chart is of the actual differences. Version 5.6 minus v5.2, and v6 beta 5 minus v5.6. Note that v 5.2 probably has a different baseline.

Figure 6 | Difference between lower troposphere temperature UAH versions. Source: UAH

Now see what happens when RSS differences are plotted on the same scale as the UAH differences above.

Figure 7 | Difference in lower troposphere temperature RSS v3.3 between selected months, scaled to match UAH version differences. The chart shows differences over time: March 2016 minus November 2013; July 2016 minus March 2016; October 2016 minus July 2016. Source: RSS

 I think the WUWT outrage is called a storm in a teacup.

From the WUWT comments

At WUWT, the use of terms like "watermelon", "alarmist", "warmista", "commie", "fascist" etc don't reduce the credibility of conspiracy theorists (in the eyes of conspiracy nutters). But in tomwys1 eyes, calling a denier a denialist does. It makes him sad:

October 10, 2016 at 3:15 pm
Unfortunately the pejorative label use by Dr. Mears undermines his credibility. Sad, very sad.
In the meantime we’ll wait for Dr. Spencer’s more detailed explanation for the change; hopefully with his usual high level of erudition!

Tom Halla may have meant that Dr Mears has nothing to be ashamed about and everything to be proud of. Or maybe not. Tom's posting on a shameless conspiracy nuttery blog:

October 10, 2016 at 3:21 pm
Mears is shameless. 

John MacDonald doesn't read the literature or NASA's website, or he'd know that RSS and NASA do provide explanations. Update: The RSS info is all here at NCEI (h/t Carl Mears.)

October 10, 2016 at 3:26 pm
Is not it the responsibility of outfits like RSS and GISS, who provide the base data for many, to concurrently provide explanations of the their work product when changes are made? I see this as analogous to the imperative to provide data, equations, work methods, etc for any published scientific paper in a journal.

There were the usual false accusations. Javert Chip insinuated fakery:

October 10, 2016 at 6:49 pm
Only if they are ethical… 

And the expected conspiracy theories. Bloke down the pub insinuated forgery for money:

October 10, 2016 at 3:28 pm
Someone worried about their funding? 

chilemike likens the straightforward reporting of monthly data to Soviet propaganda. Has he been Trumped?

October 10, 2016 at 3:44 pm
Hooray for the alarmists! Finally they are getting on par with the Soviets in terms of propaganda. Why don’t they just get it over with and just make up whatever numbers they want. They clearly aren’t real scientists as they do not seek truth so at this point it doesn’t really matter. 

Anthony Watts, responding to a comment about UAH changes, wrote:

October 10, 2016 at 4:29 pm
You must not be reading carefully. Actually Dr. Spencer announces each change, unlike what Mears appears to not have done..
“NOTE: This is the eighteenth monthly update with our new Version 6.0 dataset. Differences versus the old Version 5.6 dataset are discussed here. Note we are now at “beta5” for Version 6, and the paper describing the methodology has been conditionally accepted for publication.
Source: http://www.drroyspencer.com/2016/10/uah-global-temperature-update-for-september-2016-0-44-deg-c/

Anthony Watts hasn't been paying attention (not even to his own blog).  RSS publishes a paper to coincide with each new version. UAH hasn't yet got their paper published, even after 17 months!

Uber conspiracy theorist Rudd Istvan incomprehensibly wrote something about publishing. What? Seriously? ristvan wrote:

October 10, 2016 at 4:16 pm
It has been apparent for years that Mears doesn’t like his RSS result, once it started showing no warming. He first argued that surface measurements were better, even though the original documented premise of RSS was the opposite (remove UHI, remove microsite bias, all of that). So he has now Karlized RSS. But learning from the Karlization brouhaha, has decided not to publish where skeptics could point out the flaws?

TonyL can look above to see the comparison and put the magnitude of the change into perspective:

October 10, 2016 at 4:18 pm
It would be interesting to compare this change to the change in UAH going from v5 to v6. At least it would put the magnitude of the change into context.
Also, on this animated plot, the slopes are listed as 1E-5 and 2E-5. Listing a slope to 1 significant figure is not really helpful. If you let a computer spit out too many digits, people will pick on you for false precision. But at least people can round the numbers off themselves.

rishrac decides on the strength of nothing but his or her "nefarious intent" conspiracy theorising, that the numbers were adjusted without any reason. rishrac isn't the only person at WUWT who lacks the capacity to reason. He's off with the wizards.

October 10, 2016 at 4:50 pm
Say it ain’t so ! Adjusting numbers without any reason. How can that possibly be a problem ? Thank you climate change people !! Yea, I’ll never be wrong again, I can change the numbers! One small change for climate science, one huge step back to the dark ages.
Why bother to do anything ? We can all fit in some dark deep room with smoke and mirrors and conger up any kind of world we want. Hocus pocus Abra ka dabra.. poof. They learned their science at Hogswart school of Wizarding.. 

Pat Michaels, resident disinformer from the Cato Institute, promises "more on this tomorrow". I wonder will he mention satellite drift?

October 10, 2016 at 5:46 pm
It appears to be a step-change around the beginning of 2009, with all subsequent readings adjusted upwards the same amount. More on this tomorrow. 

References and further reading

Carl A. Mears and Frank J. Wentz, 2016: "Sensitivity of satellite-derived tropospheric temperature trends to the diurnal cycle adjustment." J. Climate doi:http://dx.doi.org/10.1175/JCLI-D-15-0744.1 (subs req'd)

• The RSS Middle Tropospheric Temperature Now V4.0 - blog article by Carl Mears about the paper.

Mears, Carl A., and Frank J. Wentz. "Construction of the RSS V3. 2 lower-tropospheric temperature dataset from the MSU and AMSU microwave sounders." Journal of Atmospheric and Oceanic Technology 26, no. 8 (2009): 1493-1509. DOI: 10.1175/2009JTECHA1237.1 (open access)

From the HotWhopper archives

• 101 conspiracy theories about troposphere temperature: the RSS love affair is over - March 2016 - with references
• The mid-troposphere has been warming faster than you thought - March 2016