TRMM Satellite Suggests July 2009 Not a Record for Sea Surface Temperatures
August 26th, 2009 By Dr. Roy Spencer
UPDATE ADDED: 8/26/09 13:30 CDT see below
NOAA/NCDC recently announced that July 2009 set a new record high global sea surface temperature (SST) for the month of July, just edging out July 1998. This would be quite significant since July 1998 was very warm due to a strong El Nino, whereas last month (July, 2009) is just heading into an El Nino which has hardly gotten rolling yet.
If July was indeed a record, one might wonder if we are about to see a string of record warm months if a moderate or strong El Nino does sustain itself, with that natural warming being piled on top of the manmade global warming that the “scientific consensus” is so fond of.
I started out looking at the satellite microwave SSTs from the AMSR-E instrument on NASA’s Aqua satellite. Even though those data only extend back to 2002, I though it would provide a sanity check. My last post described a significant discrepancy I found between the NOAA/NCDC “ERSST” trend and the satellite microwave SST trend (from the AMSR-E instrument on Aqua) over the last 7 years…but with the AMSR-E giving a much warmer July 2009 anomaly than the NCDC claimed existed! The discrepancy was so large that my sanity-check turned into me going a little insane trying to figure it out.
So, since we have another satellite dataset with a longer record that would allow a direct comparison between 1998 and 2009, I decided to analyze the full record from the TRMM Microwave Imager (TMI). The TRMM satellite covers the latitudes between 40N and 40S, so a small amount of N. Hemisphere ocean is being missed, and a large chunk of the ocean around Antarctica will be missed as well. But since my analysis of the ERSST and AMSR-E SST data suggested the discrepancy between them was actually between these latitudes as well, I decided that the results should give a pretty good independent check on the NOAA numbers. All of the original data that went into the averaging came from the Remote Sensing Systems (RSS) website, SSMI.com. Anomalies were computed about the mean annual cycle from data over the whole period of record.
The results are shown in the following three panels. The first panel shows monthly SST anomalies since January 1998, and as can be seen July 2009 came in about 0.06 deg. C below July 1998. At face value, this suggests that July 2009 might not have been a record. And as you can see from the first 3 weeks of August data, it looks like this month will come in even cooler.
Now, if you are wondering how accurate these monthly anomalies are, the second panel shows the validation statistics that RSS archives in near-real time. Out of the 5 different classes of in situ validation data, I chose just the moored buoys due to their large volume of data (over 200,000 matchups between buoys and satellite observations), and a relatively fixed geographic coverage (unlike drifting buoys). As can be seen, the TMI SST record shows superb long-term stability. The 0.15 deg. C cool bias in the TMI measurements is from the “cool skin” effect, with water temperatures in the upper few millimeters being slightly cooler on average than the SSTs measured by the buoys, typically at a depth around 1 meter.
The third and final panel in the above figure shows that a substantial fraction of the monthly SST variability from year to year is due to the Southern Oscillation (El Nino/La Nina), and the Pacific Decadal Oscillation, PDO. Each of these indices have a correlation of 0.33 with SST for monthly averages over the 40N-40S latitude band, while their sum (taking the negative of the SOI first) is correlated at 0.39. I did not look at lag correlations, which might be higher, and it looks like some additional time averaging would increase the correlation.
I will post again when I have new information on my previously reported discrepancy between NOAA’s results and the AMSR-E results. That is still making me a little crazy.
8/26/09 13:30 CDT UPDATE
I computed the monthly global (60N to 60S latitudes) AMSR-E SST anomalies, adjusted them for the difference in annual cycles with the longer TMI record, and then plotted the AMSR-E and TMI SST anomalies together. Even though the TMI can not measure poleward of 40 deg. latitude (N or S), we see reasonable agreement between the two products.
None of this represents proof that July 2009 was not a record warm month in ocean surface temperatures, but it does cast significant doubt on the claim. But the focus on a single month misses the big picture: recent years have yet to reach the warmth of 1998. Only time will tell whether we get another year that approaches that unusual event.
Re: Bob Tisdale (15:37:31)
Thanks for the notes Bob.
My level of agitation with with the various processing & anomaly conventions is going nuclear.
What do we need to do?
Hire an army of full-time auditors to keep track of all the nonsense?
Note to data agencies:
1.
Post raw data (not anomalies) and log notes. This is ABSOLUTELY ESSENTIAL to responsible data analysts intent on drawing sensible conclusions.
2.
If you want to post anomalies:
(a) BE ABSOLUTELY CLEAR & EXPLICIT about the climatology and any processing right at the top of the plain-text webpages.
(b) ALSO post raw data and log notes. This is ABSOLUTELY ESSENTIAL to responsible data analysts intent on drawing sensible conclusions.
Untangling the mysteries of natural climate is already a consuming task without the needless administrative hurdles impairing our every step like a string of traffic lights. (Administrators’ favorite tactic: Build in delays.)
Claude Harvey wrote:
“…Roy Spencer’s original post got me looking at the ASMU-A plots of global average temperatures at various altitudes. In comparing them I noticed something peculiar. I noted that the annual swings in global average temperature as recorded by satellite follow a curious pattern at differing altitudes…”
…and likewise in the ocean depths, are there any fixed instruments to monitor the temperature trends of vertical currents?
“anomaly
1. The deviation of a property at a particular location from the long-term mean value. Anomalies, such as sea-surface temperature anomalies, are often an extremely significant factor in … ”
http://www.encyclopedia.com/doc/1O16-anomaly.html
It seems to me to be a very weird, if not anomalous, use of the word anomaly. Weird because it assumes that something is out of whack or not normal about the “deviation of the property” being examined. It suggests that “it should not be” as it’s anomalous or broken or out of the normal what ever that is. It assumes that the “long-term mean” value is the normal value that SHOULD BE! This term is clearly a form of BIAS in climate science (and possibly from statistical science as well). This doesn’t bode well for anyone who isn’t aware of this BIAS upon their way of thinking. A more neutral term would be preferable, is there one?
pwl re anomaly
I’m not an expert, but an anomaly is just a deviation from some average (e.g. 1979 – 2009) value. The controvercy lies in what period is chosen.
Thanks Tom, that’s insightful and useful.
I think it’s also clear from the discussions on climate science that what is considered normal is also controversial… which often rests on whether or not the “mean” has any meaning to Natural cyclic data. Sounds like man applying binary thinking (of a sorts) upon a fuzzy Nature. It seems to me that the statistical approaches being taken in climate science at their very core are the wrong ones or should be treated with deep suspicion.
Please correct me if I’m wrong.
Re: pwl
The convention which developed a long time ago in Climatology was to define a “climatology” ….confused yet?
The way this works is that a “base period” is established, commonly a 30 year period that slides by a decade at the end of each decade. The average for each month within the base interval is calculated. These 12 values are then accepted as defining “climate normals” for the period.
Subtracting this “climatology” from the raw series yields the anomalies.
A few important things to realize:
1) The anomalies change when the base period changes. (The changes can be dramatic when the base periods are short.)
2) Most people think of anomalies as being the raw data with the annual cycle removed, but this is a somewhat loose & misleading interpretation (based only on mainstream convention for those who have not thought it through carefully).
Anyone wanting to understand more deeply is advised to role up their sleeves and get their hands into the raw data.
I sincerely hope there is not a single climate-related policy-maker who could not sit down and quickly whip-up climatologies & anomalies with raw series.
A very serious problem for a responsible data analyst who intends to draw sensible conclusions is that one cannot reconstruct the raw data from anomalies without the “climatology” (which is seldom provided on anomaly webpages …AND WHAT’S THE BIG DEAL ADMINISTRATORS??? IT’S ONLY 12 NUMBERS!!! Save us any lame/creative excuses and GET REAL!).
A more fundamental question is: Why not just supply the raw data? It only takes seconds to whip-up a climatology & anomalies if one is operating from the natural starting-point (i.e. raw data).
The problem seems to be that mainstream convention & tradition is overlooking/suppressing/obfuscating something that should be REALLY obvious to any good Stat 101 student who has understood chapters 1 & 2 (…& to be clear I am not suggesting this is being done maliciously).
Everything should start with raw data.
Anomalies, differencing, smoothing, etc. are useful in various data analysis contexts, but it is important to have the raw data on hand so that a responsible analyst KNOWS WITH ABSOLUTE CERTAINTY FIRST-HAND EXACTLY what processing has been done to the data — this is the way to ensure SENSIBLE CONCLUSIONS can be drawn.
Any arguments otherwise are just hurdle-erection based on tradition, mainstream momentum, &/or misunderstanding of Stat 101 chapters 1 & 2. (I’m convinced [in part based on my experience teaching Stat 101] that a lot of otherwise very bright people overlook the latter.)
A more complicated issue, which I will not go into in depth here, is the issue of shared variance. One cannot recover from a bad decomposition and we have to realize that just because diurnal & annual cycles are dominant, that does not mean they do not share variance with other factors. It is not reasonable for agencies to be precluding a wider range of responsible analyses for no reason other than stubborn tradition. It is foolish to try to guarantee that there is no confounding.
To be completely honest: I abhor the tradition used in Climatology because of the flawed assumption that a series can be decomposed into 2 components, one of which is based on such a dominant mode that it does not share any variance with any other factor. Based on analyses I have done, my strong instinct is that this is one of the places where conventional climatology has gone seriously off the rails.
We won’t be able to shake the convention of using anomalies …and we don’t need to do so — anomalies are useful, as I’ve noted – (I use them frequently). HOWEVER: We should be able to succeed in getting the following message across: PROVIDE RAW DATA TOO – WITHOUT EXCEPTION. This message needs to be hammered consistently from now until the day when it happens without exception.
Thanks Dave for the link; I’ll be digesting that. Not too interested in the software, just the physical sensing process, but as a working stiff, I can’t dig much of this up myself.
Thanks again.
George
pwl (16:34:34) :
It seems to me to be a very weird, if not anomalous, use of the word anomaly. Weird because it assumes that something is out of whack or not normal about the “deviation of the property” being examined. It suggests that “it should not be” as it’s anomalous or broken or out of the normal what ever that is. It assumes that the “long-term mean” value is the normal value that SHOULD BE! This term is clearly a form of BIAS in climate science (and possibly from statistical science as well). This doesn’t bode well for anyone who isn’t aware of this BIAS upon their way of thinking. A more neutral term would be preferable, is there one?
I agree wholeheartedly. The most anomalous result possible would be if any component of the climate never deviated from the long term mean of its record. But this is hardly a singular instance of linguistic bamboozling in the climate sciences. Aspects of the climate that have barely demonstrated an ability to influence the climate are routinely described as “forcings” implying not only that their influence is assured, but irresistible. Minor changes in the alkalinity of the oceans are always ominously characterized as “acidification”. Even warmth which has been a positive quality since the time when humanity developed language is a pejorative now. The damage that is being inflicted on all human discourse is profound. In my view the denizens of the climate science swamp could all use some force feeding in Korzybski’s “General Semantics”, although most probably wouldn’t get it. We could at least hope that some number of them could absorb and benefit from his most famous injunction i e “The map is not the territory”.
pwl wrote:
“anomaly
1. The deviation of a property at a particular location from the long-term mean value…It seems to me to be a very weird, if not anomalous, use of the word anomaly. A more neutral term would be preferable, is there one?”
Use the alternate term used in the definition, “deviation.” (Of course “deviants” doesn’t sound that good either!)
Thank you Paul Vaughan (19:09:44) for your excellent post explaining the usage of “anomaly” in the context of statistics and climate science in particular. Obviously I need to learn more statistics. Which book(s) do you recommend or online materials (videos, pdf papers, web sites, audios, etc…)?
Dave Wendt (19:25:12) your post is very interesting indeed. Thanks.
noaaprogrammer (20:36:30), deviation sounds good unfortunately the climate scientists would need to change which they won’t likely do. Do you really work for NOAA programming their systems? Which systems? Model-E or others? On the Parallel Super computer? Systems and Computer Science are my specialties.
Thanks again to those who’ve responded with eloquent and educational posts. I have learned from you. Now where are the statistics book in my collection.
Oh, Dave Wendt (19:25:12), you’ll find the Korzybski quote, “The map is not the territory”, on my web site PathsToKnowledge dot NET (http://www.PathsToKnowledge.NET) where it’s been since I started the site. I have a background in Neuro-Linguistic Programming which has adopted the quote as one of the general principles of NLP. I have had the book “Science and Sanity” for a couple of decades now and recently ordered and received a number of additional books written by Korzybski.
Korzybski, awesome is!
[:|]
Claude Harvey (12:52:29) : .. I noted that the annual swings in global average temperature as recorded by satellite follow a curious pattern at differing altitudes. The annual positive and negative temperature swings measured at near-surface (and at 14,000 feet) are mirror images of the annual temperature swings at 102,000 feet. The lower-altitude global average atmospheric temperature invariably peaks in mid-July, within a month of the northern hemisphere’s summer solstice. At the exact same time, the temperature at 102,000 feet invariably reaches its minimum annual temperature. In both cases, that annual swing is roughly in the range of 3.5 to 4.5 degrees F. At the same time, the temperature half way in-between at 56,000 feet shows little annual variation.
1) Why does global average temperature at 14,000 feet and below track northern hemisphere seasons and, conversely
2) Why does the global average temperature at 102,000 feet track southern hemisphere seasons?..
Well I can answer your first question. As for the 102,000 ft I never looked at it till today, but I’m sure there is an explanation why its the opposite of the lower atmosphere. The temperatures peak and have their minimum very close to aphelion and perihelion day, which are out of sync with the solstices and will get more so as we plunge once more inevitably into another ice-age.
Please look here http://wattsupwiththat.com/2009/08/21/soho-back-up-and-running-didnt-miss-anything-sun-still-blank/#more-10138 and the posts Richard (01:48:07) : Richard (01:52:25) : Richard (10:37:19) : Richard (11:05:32) : Richard (20:54:22) : John Finn (03:32:23) : Richard (11:53:47) : where I asked this very question and then tracked down the answer.
In brief, in Dr Roy Spencers words, “it’s because most of the land is in the Northern Hemisphere, and land reaches a higher temperature than ocean under the same amount of sunlight”.
Paul Vaughan (19:09:44), so the statement “The anomalies change when the base period changes. (The changes can be dramatic when the base periods are short.)” implies that with 30 year base periods that slide every ten years that the anomalies for the last twenty years (decades BC) in the first 30 year sample ABC COULD HAVE, and often WILL HAVE, different “anomalies” (deviations) after the 10 year slide to sample BCD? That’s spooky but understandable with the math involved. If I’m not mistaken it also implies that the anomaly data can’t be understood – it’s utterly useless – unless you have the full context of the original data used to compute the base periods! Is that right?
So just like a high school math question is incomplete if you don’t show your work, climate science is incomplete if the scientists don’t show ALL the STEPS of their work? Didn’t these guys ever go to high school math class? How did climate science get so sloppy?
Without all the steps shown in a science paper, which would include all the data being available along with all the programs, spreadsheets, etc…, used in the paper how could any scientist sign off on a peer review? It baffles the mind. It smacks of bad science and poor ethics.
Re: Richard (22:26:49)
“Well I can answer your first question….In brief, in Dr Roy Spencers words, ‘it’s because most of the land is in the Northern Hemisphere, and land reaches a higher temperature than ocean under the same amount of sunlight’”.
Thanks for that straightforward explanation. As to the second question, I vaguely remember being told the earth is actually closer to the sun during the northern hemisphere’s winter than during its summer. Perhaps the global average temperature at 102,000 feet is driven by “closeness” whereas at 14,000 feet it is, as you indicate, dominated by axial angle and exposed landmass?
CH
noaaprogrammer (20:36:30) :
pwl wrote:
“anomaly
1. The deviation of a property at a particular location from the long-term mean value…It seems to me to be a very weird, if not anomalous, use of the word anomaly. A more neutral term would be preferable, is there one?”
Use the alternate term used in the definition, “deviation.” (Of course “deviants” doesn’t sound that good either!)
“Variation” seems like a reasonably value judgement free word to use.
Re-post (ripost?) for comments,
from an earlier thread, Allan M R MacRae (17:27:29) :
First, it has to be established that this ocean warming measurement is real.
I think it is already safe to conclude that whatever is happening is almost entirely natural and not humanmade.
Why? Because (I’m guessing that) humanity does not have nearly enough horsepower to effect such rapid warming and cooling changes.
The farse continues. I don’t understand how can anyone make science with this kind of data.
While July 1998 may have been a record for July ocean temps it was not particularly warm given the extraordinary El Nino event. A month is only a thirty or so period of days in 365. You can have record 30 day temps spanning two months without any of the months setting a record. You can have a one month period of moderately hot temps that beat both the months that nested the record thirty days.
July 1998 was well into the decline of the ENSO for that period and the tropical Atlantic temperatures had also dropped. A better candidate for a July record would have been 1997, if we only use the Pacific and Atlantic tropics.
http://i599.photobucket.com/albums/tt74/MartinGAtkins/ATL-PAC.jpg
Obviously other latitudes and oceans conspired to rob 1997 of the coveted prize of record July ocean temps. Candidates for the record month for that event would be anywhere between 97-11 and 98-05.
UAH judges the contest.
http://i599.photobucket.com/albums/tt74/MartinGAtkins/UAH-Ocean97.jpg
Re: pwl
People are buying too heavily into assumptions of cyclostationarity.
Sensible interpretation can get tricky with functions of functions of functions of …
The best IntroStats textbook on the market is by DeVeaux, Velleman, & Bock. I’d refer you to my 100s of webpages of online intro-stats notes had my access to them not been mysteriously & suddenly blocked. (Watch your back if you speak truth to power.)
pwl
Looks like your eyes are opening to the true state of official Climate Science. It can take a while to grasp the lot – but re. your interest in statistics and its importance, I’d highly recommend you read McKitrick’s original What is the Hockey Stick debate all about? and that article is a good lead-in to the Climate Audit website which, par excellence, is pushing for proper standards in IPCC science, particularly w.r.t. statistics and transparency – openness to inspection by any intelligent Jo Ordinary.
If you really want to learn the relevant statistics, nose around the CA site and its linked sites, CA101 and the forum and when you’ve done all you can that way, ask for help.
Claude Harvey (23:36:43) : .. As to the second question, I vaguely remember being told the earth is actually closer to the sun during the northern hemisphere’s winter than during its summer. Perhaps the global average temperature at 102,000 feet is driven by “closeness” whereas at 14,000 feet it is, as you indicate, dominated by axial angle and exposed landmass? Possibly but I’m not sure about that. The Earth is closest to the sun at perihelion which is around the 3rd of Jan and furthest at aphelion around the 5th of July. We get 6.9% more radiation from the Sun on the 3rd of Jan than the 5th of july. But in absolute terms the 5th of July is 21.55 % warmer than January.
Ok, I’ll dive in here with a question that has been bugging me, although I suspect I’ll feel like an idiot once its answered…. with regard primarily to surface temperatures from surface met. stations – how in the world do they justify anomalies in the x.y range, when the original data has error ranges in the neighborhood of 1 to 5 degrees?
Whatever happened to keeping the results and conclusions within the same order of magnitude as the data collected? In other words, how can they justify temps to the tenth of a degree or smaller, when that’s at least an order of magnitude more sensitive than the actual data collected??
Then to really show some climatology/technology ignorance and confessing the fact that I haven’t looked at the actual satellite data – what is the sensitivity of satellite measurements of both the surface and higher layers?
Ready with a dunce cap, stool, and handy corner for myself if necessary and looking forward to replies none-the-less.
“None of this represents proof that July 2009 was not a record warm month in ocean surface temperatures, but it does cast significant doubt on the claim…..”
Maybe not as yet but it has the potential of proving exactly such a thing and that, if indeed this is the case, that the warming has been overstated.
As such the significance of such a finding is huge and deserves concerted further investigation.
I think that if Dr Spencer is proved correct this would have far greater impact and consequence than the debunking of the “hockey stick”.
How could the AGW alarmists justify their alarmism if in fact the oceans have heated far less than they have stated?
Do we owe Karl a nod in apology following his dissing of the satellite data?
I recoil at the thought. Thinking of skunky homebrew.