Spurious SST Warming Revisited
Dr. Roy Spencer August 31st, 2009
My previous post described what I called “smoking gun” evidence of a spurious drift in the NOAA sea surface temperature (SST) product when compared to SSTs from the TRMM satellite Microwave Imager (TMI). The drift seemed to be mostly confined to 2001, almost a ’step’ jump. The moored buoy validation statistics of the TMI sea surface temperatures from Frank Wentz’s web site (SSMI.com) suggested that the TMI SSTs had good long-term stability.
But 2001 was also the year that the TRMM satellite was boosted into a higher orbit, which concerned me. I asked Frank about the effect of this event on the TMI SSTs (which also come from his web site). Frank couldn’t remember the details, but said he spent quite a bit of time correcting for the altitude change on the retrieved SSTs since the microwave emission of the sea surface depends upon the TMI instrument’s view angle with respect to the local vertical.
I know from our many years of work together on the AMSR-E Science Team that Frank is indeed a careful researcher, yet it seemed like more than a coincidence that the TMI and NOAA sea surface temperatures diverged during the same year as the orbit boost. So, I went back to see what might have caused the problem. I went back and thought about the different ways in which one can compute area averages from satellite data.
To make a long story short, because the orbit boost caused the TMI to be able to “see” to slightly higher latitudes, the way in which individual latitude bands are handled has a significant impact on the resulting temperature anomalies that are computed over time. The previous results I presented were for the 40N to 40S latitude band, which is nominally what the TMI instrument sees today. But before 2001, the latitudinal extent was slightly smaller than it was after 2001.
As shown in the following figure, if I restrict the latitude range to 38N to 38S, which was always covered during the entire TRMM mission, I find that the divergence between the TMI and NOAA average SST measurements essentially disappears.
Even though I was processing the NOAA and TMI datasets in the same manner, I should NOT have been. This is because there were not as many gridpoints over cooler SST regions going into the ‘global’ averages before the satellite altitude boost as after the boost. So, for example, one must be very careful in computing a latitude band average, say from 39N to 40N, to make sure that there has been no long-term change in the sampling of that band.
Based upon the above comparisons, I would now say there is no statistically significant difference in the SST trends since 1998 between TMI, the NOAA ERSSTv3b product, and the HadSST2 product. And it does look like July 2009 might well have experienced a warmer SST anomaly than July 1998, as was originally claimed by NOAA. (Remember, TMI can not see all of the global oceans, just equatorward of about 40 deg. N and S latitude.)
In the bottom panel of the above figure, I also have a comparison between the TMI and AMSR-E sea surface temperatures, which are available only since June of 2002 from the Aqua satellite. As can be seen, there is no evidence of a calibration (or sampling) drift in that comparison either.
So, what’s the moral of this story? Always question your results…even after finding the obvious errors. And maybe I should eliminate the term ’smoking gun evidence’ from any results I describe in the future.
Oh…and don’t believe everything you read on the internet.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Hey guys thanks for trying and thanks for your commitment to the truth.
It will always come out in the wash anyway, so might as well catch your mistakes.
This is a HARD job [especially dealing with observable data]. To err is human.
Too bad the folks on the other side can never admit their error…rather…they find a way to manipulate, and even flaunt it in some unseemly way.
“To err…is human.” [Martin Luther]
“To extrapolate HUGE “errs”… is climate model.” [Michael Mann’s successor]
Chris
Norfolk, VA, USA
Yours is a hardly imitable example of scientific honesty. More than one person should follow your example. Thank you, Dr. Roy Spencer, for the clarification.
Okay, I’m beginning to feel a little whiplash but as I adjust it seems we are about where we started with differences so small there seems little need to try to explain them.
Graphics: I need to ‘right click’ and open in a new tab to see the right hand edge.
And, Dr. Roy, thanks for the update and for keeping the high ground.
Well done Dr Spencer. You have finally cracked what was going on in the data. I expect the AGW crowd, who have been very quiet on the issue for the last week or so, will be paying us a visit here…
It is reassuring to see such tenacity Dr. Spencer. Thank you for sharing the play-by-play as the story unfolded (and I don’t assume this is the last installment).
–
My main concern is not whether records were set or not; rather, it is distortion that downplays natural factors affecting climate that most concerns me (regardless of whether trends are up, down, or sideways).
Yes we have to thank Dr Spencer for his honesty and not hiding away errors or worse in a file called CENSORED_DATA, as others have been known to have done.
Still this leaves the Hadley data which, as was pointed out I believe by John Finn, shows the SST July 2009 anomaly as less than the 1998 one.
At least the Dr Spencer was quick to point out his mistake and honest enoght to admit to the error… If only Mann, Hanson and others were as scrupulous.
Dr Spencer wrote, “Based upon the above comparisons, I would now say there is no statistically significant difference in the SST trends since 1998 between TMI, the NOAA ERSSTv3b product, and the HadSST2 product. And it does look like July 2009 might well have experienced a warmer SST anomaly than July 1998, as was originally claimed by NOAA.”
Except that for HADSST2…
http://hadobs.metoffice.com/hadsst2/diagnostics/global/nh+sh/monthly
…July 2009 SST anomaly (0.512 degC) was less than July 1998 (0.554 degC). In graph form:
http://i25.tinypic.com/f3xobq.png
It also wasn’t a record in NOAA’s OI.v2 SST anomaly data, far from it:
http://i25.tinypic.com/24g7kwj.png
From my July Update:
http://bobtisdale.blogspot.com/2009/08/july-2009-sst-anomaly-update.html
So, the ERSST.v3b data is the only SST anomaly dataset of those three showing record SST anomalies for July.
And based on the preliminary August SST anomaly data posted by NOAA yesterday, August is also not a record month.
http://bobtisdale.blogspot.com/2009/08/preliminary-august-2009-sst-anomalies.html
Interesting. Thank you Dr Spencer for your integrity and persistence, especially in the face of results that once again seem counter to skeptics’ intuitives, that with a cool Sun the sea must be showing an overall cooling trend.
I don’t think we’ve got to the bottom of it yet. There is plenty of evidence against alarmist AGW, and weather reports definitely seem to indicate the warming of the nineties is gone – so what might be causing the seas to warm – or to show warming? Akasofu might say we’re still in recovery from the Little Ice Age… really? but yes, CO2 rise lags temp rise in the ice records by 800 years or so, which seems to tally with the time taken for the total ocean circulation…
Exacting mistress, Lady Earth. Yet we need to be sure we understand the natural variations before we can be sure what the human ones are.
August here in New Zealand was the warmest on record so Niwa say.
http://www.niwa.co.nz/our-science/climate/publications/all/cs/monthly
we had more northerly winds than normal air temps 2c or more warmer in some places . But the a.v.g temps for these northerly wind were normal for this time of year .we just had more days with this wind pushing up the over all temps.
Since there’s no possibility to post on jour blog, I’ll do it here Dr Spencer (if you ever read the comments). Thanks for the honesty you showed about the data analysis you performed! It’s always great to see that two different means of measurement actually lead to the same conclusions… And I hope it will make people understand that hasty analyses and conclusions should not form the basis of any scientific approach. I wonder where all the rumors of data manipulaiton will go now…
Mark (01:47:56) : August here in New Zealand was the warmest on record so Niwa say.
Yes but this follows 3 months in a row of below average temperatures. The absolute temperatures were not that hot. NIWA: The highest temperature during August 2009 was 25.5°C recorded at Henderson (Auckland) on the 11th. The lowest temperature was recorded at Dashwood (Marlborough) with a minimum temperature of -8.6°C on the 18th.
Its pretty cool in Auckland at the moment. 17 8 and set to fall to 14 5 over the next two days. If it was the hottest there were no complaints. Cant say the same for the fix – cap n trade.
Research long ago showed that really smart people are wrong a little more than 50 percent of the time when deciding on complex issues from incomplete data (no kidding).
So, rule of thumb, one has to check and find the bad 50 percent — and fix them. Of course, half of those fixes will be wrong — now down to 25 percent error. Then, find that 25 percent — and fix them — leaving only 13 percent in error. Now attempt to fix them. Etcetera.
Thus, being wrong is not bad — one is in good company. Not being skeptical of one’s own work is what is bad.
That heuristic has worked for me in actual practice — as I always meet my quota of wrongness when working on hard issues.
Lucy Skywalker:
“I don’t think we’ve got to the bottom of it yet. There is plenty of evidence against alarmist AGW, and weather reports definitely seem to indicate the warming of the nineties is gone – so what might be causing the seas to warm – or to show warming? ”
The answer is extreme Proxigean spring tides – see my May 2009 Melbourne talk at:
http://www.naturalclimatechange.info/?q=node/10
There was a set of Extreme proxigean spring tides in january and June-July of 2009 which usually heralds the onset of an El Nino event.
I have wondered whether there is a time lag between oceanic temperatures and solar activity, since water has a high specific heat capacity and therefore it may warm slowly but surely in response to high solar activity.
Given that we have had about 80 years of very high solar activity, this may have led to warming by mechanisms I don’t understand.
However, it would be surprising to expect rapid cooling of the oceans afterwards, so maintaining higher oceanic temperatures at the start of reduced solar activity would hardly be surprising.
In my amateur opinion, that’s where research must head – understanding the linkages between inputs and modulating features over the 30 – 100 year timescale.
A job for patient researchers?
Great Job.
Rhys Jaggar (04:45:36) : “A job for patient researchers?”
Perhaps it’s not impatience that drives them but rather the need to find the answer before they die. When dealing with relatively long term predictions, most will never know if they were correct.
Thys Jaggar, I think your Question, “A job for patient researchers?” is very correct one to be asked and should have been a statement.
Dr. Spencer it is good to see a good scientist be skeptical of his own findings when they don’t agree with some one Else’s findings and to check and re check himself. It is like a breath of fresh air to see this. I tire of those that make studies and declare that there is no question of the correctness of their findings. It is bad enough when they don’t follow up on their own findings and check for errors but then restrict the data and method so that no one else can replicate the computations.
It would be a good thing if more of our researchers would add a good statician to their “teams” to perhaps increase the validity of their computations.
Very good work Dr. Spencer.
Bill Derryberry
First of all, well done Dr Spencer for following this up and correcting errors in previous analyses – this is research after all.
As a bit of a lurker on this blog, I am interested in the impact that the internet (more specifically – blogs) are having in research. We are – in essence – seeing blog posts replace the discussions co-authors and colleagues would have about results prior to publication. I am sure that if this had been a paper for publication, Dr Spencer would have done this kind of error-checking prior to submission.
This is not a criticism, but a recognition that we can now expand the group of people with which we discuss results prior to reaching some kind of conclusion. The number of people who have read and commented on this series of postings is orders of magnitude greater than would normally have been involved in coffee-room discussions and one can hope that at least some of the comment has been useful. (I am not a climate scientist and cannot follow the maths so I will admit that I am making an assumption here).
This leads me to a major question: Does blogging remove the need for (or value of) working in a collaborative team? For me, research has always been teamwork, both in designing experiments and interpreting results, yet now it would seem we have a credible alternative, at least where the research is mostly analysis of data. What possible impact does this have on our current research (and more importantly, research-funding) models?
I could also see an impact of blog-discussion on scientific publication – at what stage does the on-line discussion constitute prior publication? In the field of interest here (climate science) this may be a moot point, given the political involvement in journal editing, but in other fields, the primacy of published (I won’t say peer-reviewed) work is not seriously challenged.
Maybe this is too philosophical for a Tuesday morning (not enough coffee – or too much?), but are we seeing a fundamental change in the way we do research?
So when does the AP print it’s retraction?
Flanagan (02:44:21) :
And I hope it will make people understand that hasty analyses and conclusions should not form the basis of any scientific approach.
I don’t think find many people here would disagree. Will you applying you new high standards to the AGW crowd?
wonder where all the rumors of data manipulaiton will go now…
There would be no rumors about data manipulation if all researchers were as open and honest as Dr. Spencer.
Interesting how just a couple of degrees of latitude can make such difference in the time series. So the next question is what about the many degrees of latitude north and south of the observed bands (40N-40S and 60N-60S). Do they vary similarly to the equatorial regions or are they decoupled to any significant extent?
Richard (23:46:54) :
Yes we have to thank Dr Spencer for his honesty and not hiding away errors or worse in a file called CENSORED_DATA, as others have been known to have done.
Still this leaves the Hadley data which, as was pointed out I believe by John Finn, shows the SST July 2009 anomaly as less than the 1998 one.
Bob Tisdale (01:03:57) makes the same point about the Hadsst2 data, but the 2 anomalies are not necessarily inconsistent. There isn’t a huge difference between them (see second plot). It’s perfectly possible for one to show a record while the other doesn’t.
Dr. Spencer,
The only problem with using the term “smoking gun” is that sometimes that gun fires blanks…
I like your style. Keep up the hard work, and continue to question your results.
Mike
Dr. Spencer has just demonstrated how a ethical scientist should do business.