From the University of Hawaii ‑ SOEST
“This summer has seen the highest global mean sea surface temperatures ever recorded since their systematic measuring started. Temperatures even exceed those of the record-breaking 1998 El Niño year,” says Axel Timmermann, climate scientist and professor, studying variability of the global climate system at the International Pacific Research Center, University of Hawaii at Manoa.
From 2000-2013 the global ocean surface temperature rise paused, in spite of increasing greenhouse gas concentrations. This period, referred to as the Global Warming Hiatus, raised a lot of public and scientific interest. However, as of April 2014 ocean warming has picked up speed again, according to Timmermann’s analysis of ocean temperature datasets.
“The 2014 global ocean warming is mostly due to the North Pacific, which has warmed far beyond any recorded value (Figure 1a) and has shifted hurricane tracks, weakened trade winds, and produced coral bleaching in the Hawaiian Islands,” explains Timmermann.
He describes the events leading up to this upswing as follows: Sea-surface temperatures started to rise unusually quickly in the extratropical North Pacific already in January 2014. A few months later, in April and May, westerly winds pushed a huge amount of very warm water usually stored in the western Pacific along the equator to the eastern Pacific. This warm water has spread along the North American Pacific coast, releasing into the atmosphere enormous amounts of heat–heat that had been locked up in the Western tropical Pacific for nearly a decade.
“Record-breaking greenhouse gas concentrations and anomalously weak North Pacific summer trade winds, which usually cool the ocean surface, have contributed further to the rise in sea surface temperatures. The warm temperatures now extend in a wide swath from just north of Papua New Guinea to the Gulf of Alaska (Figure 1b),” says Timmermann.
The current record-breaking temperatures indicate that the 14-year-long pause in ocean warming has come to an end.
###
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
listening to: Jonny Lang- “Lie To Me”
As a lay person who has been watching “global warming” data (and adjustments) for years, I don’t think I need to see Bob Tisdale’s comments to understand that Axel Timmermann’s “data” is a bunch of Do-Do (to quote Young Frankenstein).
Since the heat appears not to be coming from the atmosphere, should we be worried about accelerating underwater volcanic activity?
Enigmatic recent increase in Pacific submarine volcanism:
http://www.pmel.noaa.gov/eoi/
We can worry about a decrease of high albedo clouds in the equatorial regions.
can we all just humor the nitwit for a few minutes….
“releasing into the atmosphere enormous amounts of heat”
…times up
“enormous amounts of heat” which seem to be “hiding” somewhere ever since being released. …
http://www.drroyspencer.com/wp-content/uploads/UAH_LT_1979_thru_October_2014_v5.png
http://www.drroyspencer.com/2014/11/uah-global-temperature-update-for-october-2014-0-37-deg-c/
“releasing into the atmosphere enormous amounts of heat”
Isn’t it fantastic that they’ve figured out a way to release the heat in cold water. Climate scientists are obviously much cleverer than all those negative engineers who told me it couldn’t be done. Just think of the applications. I can heat my house with cold water! Better yet we can “releasing the heat” from the cooling water that comes out of a power station and use it to power the turbines. No need to burn fossil fuels or radiate stuff. We can get our power from environmentally clean water which just cycles round and round in the power station generating more energy each time. Marvellous!
It’s the sea surface temperature that’s rising, which means cold water is finding it’s way to the bottom of the container, just like they explained it to us in a Junior High experiment. The difference is that in Junior High we 1) made a prediction, 2) conducted a test, 3) saw that the results of the test were consistent with the prediction, 4) and then we discussed how that process was related to scientific method. Maybe we should send all these bozos back to Junior High
Is Timmermann’s or his Center’s project grant up for renewal?
He’d better get his claim for ending the pause in quick, cos if there’s any sort of La Nina in the next year, it’s going to flatten his temperature increase like a steamroller.
That is what our friend LordM across the pond has been saying. I’m wondering if Bob Tisdale could answer my following SWAG of a question.
Is it possible that the el Nino we were seeing earlier in the year went weak/MIA because the winds moved the warmth north?
If so does that still mean we will get a la Nina next year?
“heat that had been locked up in the Western tropical Pacific for nearly a decade”
So greenhouse gas warming in the atmosphere from the globally well mixed gas, CO2, knows where to hide the heat for a decade………….. in the Western tropical Pacific.
Interesting place for it to go, considering atmospheric global warming is greater at higher latitudes and solar energy, by a wide margin, is greatest in the tropics.
At least that’s what the AMS tells us(an organization that I was a member of for 25 years, including holding the broadcast seal for television)
http://www.ametsoc.org/policy/2012climatechange.html
“The warming trend is greatest in northern high latitudes and over land”
The Pacific ocean may be extra warm this year but the North American continent shows a different picture The pause continues .
CONTIGUOUS US
TREND OF ANNUAL TEMPERATURE ANOMALIES IS DECLINING AT (-0.36 F/DECADE) SINCE 1998
ANNUAL, WINTER, SPRING and FALL have DECLINING TEMPERATURES
SUMMER has RISING TEMPERATURES (mostly due to one month only , namely, June)
10 months of the year show declining temperature trends and only 2 months show rising temperature trends [March, June]
CANADA
Winter trend TEMPERATURE DEPARTURES ARE DECLINING
Spring trend TEMPERATURE DEPARTURES ARE DECLINING
Summer trend SLIGHT RISE IN TEMPERATURE DEPARTURES
Fall trend TEMPERATURE DEPARTURES ARE FLAT
Annual trend TEMPERATURE DEPARTURES ARE FLAT
When you have rogue leadership it does not matter a whit what the science says.
Okay then, the Northwest Passage should be a piece of cake at this point. You go first, we’ll watch via satellite.
We should not get too excited about the current warm oceans without looking at all the seasons and the bigger picture as well.
The observable trend is for the globe is that the winters are getting colder. In the past this has caused colder spring and fall seasons and ultimately colder summers as we have just seen during 2014.in North America.
. According to NOAA, CLIMATE AT A GLANCE data, the trend of GLOBAL LAND and OCEAN WINTER TEMPERATURE ANOAMLIES has been declining since 1998 at 0.6 C /decade. So has the WINTER TEMPERATURE ANOMALIES for the NORTHERN HEMISPHER declined at o.11C /decade since 1998. The trend of WINTER TEMPERATURE ANOAMLIES for CONTIGUOUS US declined at -1.79 F/decade since 1998
Annual Contiguous US temperatures have been declining at (-0.36 F/DECADE) since 1998. This is happening in 7 of the 9 climate regions in United States. Only the Northeast and the West both of which receive the moderating effect of the oceans, had slight warming trend of 0.2 and 0.3 F/decade respectively. Theses 16 year annual temperature declines illustrate that despite any summer warming , the cooling during winter , spring and fall offsets any summer warming resulting in the annual temperature declines .
The WINTER TEMPERATURE ANOMALIES for CANADA declined from an average of + 2.6 C during 1998-2000 to -0.4C by 2014 winter , or a cooling of 3 degrees. A winter cooling trend is also apparent in EUROPE, and NORTHERN ASIA. I see this pattern continuing until 2035/2045
Northern hemisphere winters are indeed getting colder.
http://tinypic.com/view.php?pic=2dj8eau&s=8#.VGZO7PnF_A0
http://data.giss.nasa.gov/gistemp/maps/
If the oceans really are much warmer, it cannot be due to GHG. The ocean heat capacity is so great, that the impact from GHG would be so small to be unmeasurable.
It’s a negative feedback Paul. The ocean will offer its (relative) heat to a cooler atmosphere.
Where the hell do they get the global ocean temperature in 1880?
What is ARGO saying? What’s up with OHC?
Wait a minute…..For years the alarmists have been denying the pause. Now, the pause is suddenly over and they know for certain why? Perhaps they were all waiting for another theory to hitch their wagons to and bilk the tax-payers for several more decades of nonsensicle faux science.
“heat that had been locked up in the Western tropical Pacific for nearly a decade.”
So the relief valve is being blown to cool the planet and he’s making an issue of it.
Oh, so that’s where it was “hiding”. How come someone like Trenberth had not been able to spot it if it was “hiding” in plain sight?
Get a grip, Alex.
Well if I were a younger guy, I would have claimed all that heat was hiding in my girlfriends underwear.
The low math of it:
Jon Gruber =’s Mike Mann
EPA =’s ACA
A lie and a fraud =’s Lies and Fraud
Government grants of our tax money to fund the Jon Gruber’s and Mikie Mann’s.
Obama is going to get to the bottom of it ,,,,
Maybe he finds a mirror.
“The 2014 global ocean warming is mostly due to the North Pacific,'”
One cannot say that the extra warming is due to greenhouse gases alone . North Pacific has had peak warm period well before 2014 ever came along . The North Pacific SST peaked around the1870/1890 era and again 1940/1970. . This latest peak may be just the latest of ocean sst cycle peaks . We should also note that Southern Oceans are at a cold trough and deep oceans are cooling . The ocean conveyor belt brings both warm and cold phases to the ocean SST. Bob Tisdale has presented a number of graphs which show a 60-70 year cycle to the Pacific and Atlantic ocean SST, pole to pole .
I thought I’d take a look at all of the SST datasets that come up to the present … here’s that graph:
Some comments. First, all three of the datasets on the left incorporate the Reynolds data, to a greater or lesser extent for the recent years. So we don’t really have three datasets, we only have one. It appears that the Hawaii folks are using one of the three datasets depending on Reynolds.
Next, what the good folks in Hawaii didn’t say is that since the high reading of the Reynolds dataset in August. temperatures have fallen back to their normal values (HadISST is not as up-to-date as the others).
Next, by how much did the new record beat the old record?
Well … two hundredths of a degree …
Finally, the important question. Does this two-hundredths of a degree portend further warming, the “end of the pause”?
The truth is, there’s nobody on this planet who can honestly answer that question. Given the gradual warming of the last three centuries or so, it seems likely … but there’s a long ways between “likely” and “we’re sure”.
w.
The gradual warming since the depths of the LIA has not been straight up, but in multidecadal cycles, so actually a cooling is more likely as the next move after the warming of the late 1970s to late ’90s. And the long-term trend is down, as it has been for 3000 to 5000 years, so the past three century period of on balance warming is itself just an up cycle in the secular, millennium-scale, post-Holocene Climatic Optimum cooling.
I should add that the Modern Warming Period still hasn’t experienced a single 30 year interval as warm as a number of such intervals during the Medieval WP, let alone the even hotter Roman and Minoan WPs or the HCO, not to mention the last interglacial, the Eemian.
Hi Willis, I thought you may find this one of interest since you seem to have adopted Loess as one of your favoured low-pass filters.
Like any low-pass filter you cannot run it into he buffers but the loess algorithm does not give you the choice.
Look up the code of whatever you use to apply the filter and you will find it changes what it does at either end and is not more valid than a convolution filter to the end or padded by infilling data.
The current SST data has couple of high values which illustrate the problem nicely.
Best, Greg.
Damn, forgot the link:
http://climategrog.files.wordpress.com/2014/11/lowess_crap.png
http://climategrog.wordpress.com/?attachment_id=1023
Dang, 1.2 deg C in two months ! Now that is CAGW on steroids.
The pause is well and truly over, the missing heat has come out if it’s hidey hole and it’s worse than we thought. ™
Greg Goodman November 14, 2014 at 12:46 pm
Thanks, Greg. A few points.
First, if I want the best accuracy at the ends of the data, I use my own algorithm. Why? Because I’ve shown that it beats the other filters at the end points. See here for the details of the analysis.
Second, a 12-month loess filter? Why on earth would anyone use something that short? On your data above, that’s a span of 0.01. I never have used a span much smaller than ten times that, 0.1, and that’s in special situations.
Third, I’m sorry, but the loess filter actually does do better at the end of the data than padding by infilling data. Again, see my analysis linked to above. You can actually test these claims.
w.
Thanks Willis, I remember reading that when you put it up but I’d forgotten some the BS mannian methods. That explains some of the oddities I saw in one of his more recent papers. I could not work out what he thought he was doing. Apparently he’s still using his crap.
Congrats on being better than Mann at home-rolled d.p. but you are not setting the bar very high there. Instead of publishing papers on his latest “discovery” he’d do better to start reading the work thousands of other have done in this field over the last century or more. That way he’d screw up a lot less often.
The trouble with your method ( and you seem to know by constantly adding “for this data” ) is that the selection method is totally subjective to the data being processed and until you do have the rest of the data to fill the buffer you can’t know how it will go.
You could do some kind of monty carlo simulation to estimate the uncertainty, using random data engineered to have similar statistics as the rest of the data but whatever you the frequency and phase response of your filter is changing as you start altering the method at the end.
You suggest adjusting the gaussian coefficients but that is just the same as a reflection about the end of the data. Consider your last point ( window half missing ) : all points get double the weight they would in a full window since total weight is now 0.5 ; in effect, you’ve just padded the window by reflection I think this is what you are calling min slope method.
lowess and lowess suffer from similar but not identical problems. They also change the rules as they get near the end and end up giving undue weight to the final points. Progressively more so as the data runs out. This make what was a symmetrical ( ie non phase distorting ) method introduce a progressibley worse phase distortion as the end is approached.
How this actually works out and which method _appears_ best is largely a function of the data and in particular the bit of the data which is missing.
You used a farily well behaved dataset as your test but some data which more unexpected changes could produce a different result. The test is subjective.
The only solution is not to try to do the impossible. If you don’t have the ata you don’t have the data. You stop processing when you run out. Anything else is in one way or another just injecting spurious data and hoping the real data is not too different when it turns up.
You may as well just fill it in by eye where looks as though it is about to go.
The fact that your error bars are not symmetric should have been a warning too. It’s like thinking that if you have flipped four heads in a row, there’s more chance the next one will be tails.
You would not get away with this in engineering in the same way as you would not be allowed to “homogenise” you data before you started.
Of course in climatology you can do whatever you like because science is irrelevant to this field of study. Here we know what the correct result should be and the challenge is to adjust the data and the method until you can produce it. 😉
Gerg Goodman November 14, 2014 at 5:44 pm Edit
You seem to have misunderstood my method, likely my lack of clarity.
My insight was that you can estimate the end error of any smoothing method for a given dataset. Here’s how. Start by running the method over the full dataset.
Then run it again a number of times starting with the dataset truncated at the shortest possible length (half the filter width plus one) and truncating the dataset each time one timestep further on. This gives you the results you would get at the end of the dataset if it actually ended at each timestep along the line.
Now you have both the estimate if the dataset were truncated at each timestep, and also the actual eventual value of the smooth at that timestep. The standard deviation of the difference between all of these pairs is the standard error of the method at the end of the dataset.
It doesn’t require any monte carlo analysis. You simply test the candidate smoothings on the actual dataset, and pick the best one.
So far, I haven’t found an actual dataset for which my method is not the best … by which I mean it gives the smallest average end error. As a result, that’s what I use.
Regards,
w.
Thanks Willis. I agree that the method is not without merit. It does help show which of these defective methods produces the wildest swings for a particular dataset and quantifies, within certain assumptions, the error margin. That does not change my opinion that one should not be processing data beyond the point were a valid result is possible.
Now consider the two titles:
“On smoothing potentially non-stationary climate time series“
“A closer look at smoothing potentially non-stationary time series”
stationary means having stable statistical properties throughout the record. The definition seems rather woolly since there is not hard and fast rule as to what statistics need to be tested and what can be consider stable, or “stationary”. and it gets rather liberally applied But two of the key features would be mean and variance.
So 20th c. temps are non-stationary since they generally have a rising mean. Also temp anomalies with have notable changes in variance outside the reference period. This is what is happening at the end of most current data where a notable residual annual component remains.
In using the full dataset as you suggest, you are ignoring the effects of non-stationarity. It would seem that you have defined a method estimating the uncertainty of the various “smoothers” on stationary data not non-stationary time series.
That could probably be addressed by a more selective test period but that then assumes that the data is stationary in the reduced test period and that end bit where you don’t have enough data is not part of a new shift, like a change in direction in 2005, for example.
The whole problem is the concept of a “smoother”. This seems to come from econometrics and a more layman approach to messing around with data in spreadsheets.
In an engineering or science context you don’t “smooth” data. You may chose to filter out some of variability that is considered not to be of interest in order to focus on another aspect of the data. This is often a low-pass or high-pass filter. The classic in climatology is the need to remove the strong annual variability that tends to obscure what else is happening.
In this case you would aim to chose a well-behaved filter, with consistent and defined properties that is considered suitable for the job. You would not want to chose a method that does different things at the end of the data. Once you don’t have enough data to apply the filter, it ends.
That is the kind of rigour on which hard science is built and is why engineering usually succeeds and climatology usually fails.
Your method is a way of comparing a plethora of fudged pseudo filter extensions and getting a quantified uncertainty. This is valuable and I think the kind of errors you show illustrates that the none of the methods are much use.
In fact you could use it see what happens if you cheat a little bit and allow up to 25% of the filter window to be empty. For gaussian, I think this would still be pretty good. You could then produce an error estimation for 5,10,15,…..50% and plot error margins showing how the uncertainty mushrooms as you cheat more and more.
That would produce some useful insight for those who like to mess around extending filters beyond the end of valid results.
It’s the consolation prize to not having a super El Nino to crow about. Any cyclical warm pool will do.
No, it’s the consolation prize for not having any scientific integrity, but it’s well rewarded. Please call NSF, your budget for next year has just been approved and expensed for the Paris meeting agreed.
The north Pacific “warm pool” is cyclical because of deliberately disfunctional seasonality correction as pointed out by Bill Illis. Just see how warm it looks next spring.
By January, the north Pacific SSTs will look extremely below “average”.
Something semi-unusual happened in the north Pacific in August/September but the biggest contributor to the big red spots is that the NCDC has screwed up the seasonality of various parts of the northern hemisphere oceans.
The screw-up goes both ways because the January seasonality is way above what the normal SSTs will do so there will be a large cooling showing up in the anomalies by January.
Having said that, there was a rise in the north Pacific SSTs this summer which appears to be gone now.
Gone to space? The OLR maps show increased OLR over some parts of the north Pacific next to land over the August-October period and they usually don’t play around with these measures so far.
The other issue which has not come so far on WUWT, is that the Chinese hacked into several of the NOAA systems (starting about August it seems).
This resulted in the NOAA taking down some of our favorite server systems (like Nomads) in September and completely losing at least 2 days of valuable satellite data etc. etc.
Many stories on the web now. Search it.
Bill,
Not only that but the crooks at NOAA lied to Congress about it. Big surprise.
http://www.dailytech.com/NOAA+Misled+Congress+About+Hack+From+China+Finally+Owns+up+to+Breach/article36884.htm
Wonder how many incriminating emails NOAA has “lost”?
The whole totally corrupt academic-gevernment-climate complex has to be burnt to the ground and genuine climatology rebuilt upon the ruins.
Bill, a layman’s explanation of the “seasonality” issue would be very appreciated.
If air were capable of warming the oceans, in July & August all the swimming pools in Phoenix would be 105 F.
I don’t trust marine temperature data. It all loiks to choreographed. The AGW story is moving to where none but a cabal of activists control the data.
What was the idiot using for a thermometer, his elbow ?
Now if only some of that heat could slide on over to Minnesota where we’re having one of the coldest Novembers ever in the history of earth.