By Steve Goddard
Earlier in the month I wrote an article showing the trend in Arctic ice since 2002.
I took a lot of criticism from people for not measuring “crest to crest or trough to trough.”
Any one schooled in analysis of cyclical data would know that one must go from crest-to-crest or trough-to-trough, to maintain some semblance of symmetry about the x-axis.
It is time now to see how serious people are about their belief systems. We have passed the 2010 El Niño peak, and can see what the “real” trend is since the cyclical El Niño peak of 1998.
http://www.woodfortrees.org/plot/rss/from:1998/last:2010/plot/rss/from:1998/last:2010/trend
“Global warming on decadal timescales is continuing without let-up … we conclude that there has been no reduction in the global warming trend of 0.15-0.2C/decade that began in the late 1970s.”
Talk about cherry-picking! Look at his start point. He chose the worst case trough to crest to measure his trend.
Question for readers. Is Hansen correct, or does he need some serious graphing lessons? Below are the trend graphs from 1998-present for all four sources. GISS is way out of line.
Jean Meeus 7:33 am
Where I live was covered with ice 10,000 years ago. It has been warming since then.
Where I live suffered through drought 75 years ago. It is cooler now than it was then.
To be an AGW believer requires selecting a convenient starting point, setting aside the other drivers of temperature variance, ignoring the creeping biases in the measurement and reporting of temperatures, and blaming CO2 for the resulting trend in the graphs.
Sounds like a pretty solid basis for global policy-making to me…what do you think?
This post seems confused on a number of points.
1. For a cyclical and regular time series, even measuring trends peak-to-peak or trough-to-trough is not meaningful. See here
2. El Niño is not a cyclical phenomenon. It is quasi-periodic. Its amplitude varies enormously. The 1997-8 one was considerably larger than the most recent one.
3. Global temperatures are not cyclical either. The instrumental record has no “trough” to measure any trend from. The late 1970s saw a clear change in climate trends – to measure something from a point of clear change cannot be described as cherry picking. In any case, for every year in the record for which you can derive a statistically significant trend from that year to the present, the trend is positive.
4. 1998 to present is too short to measure statistically significant trends, and too short for variations between datasets to be considered significant. If you are worried about outliers, you should look at a longer trend – like since the change in climate trend in the late 1970s.
” John Finn says:
July 31, 2010 at 7:39 am
Simply lengthening/shortening the period by a few months changes the sign of the UAH trend. We need at least 5 more years before any conclusions can be drawn.”
For the UAH linear trend of monthly means from peak (1998) to peak (2010) I get 0.0003°C per month or 0.036°C per decade. The positive sign might be a result of the fact that temperatures drop so sharply after El Niño. We will see in a few month if the trend from trough to trough is negative.
But waiting a few years will almost certainly result in a negative trend from peak 1998.
A reasonable comparison is to pull-out the variability caused by the ENSO. The 1997-98 El Nino was a little bigger than the last one was. (I also take into account the AMO since it’s impact is similar to the ENSO. Today’s AMO values are very high, about the same as they were at this time in 1998).
When these are taken into account, we find GISTemp is increasing at about the same rate it has always increased at. This rate though is much lower than predicted and will only get us about half-way to the projected warming of 3.25C by 2100.
Here is GISS since Feb 1998 (adjusted for the ENSO and AMO) versus the most current prediction of the IPCC in AR4 (under the A1B CO2 scenario which seems to be very close to the trends to date).
http://a.imageshack.us/img821/8305/gisspeak1998.png
Here is nice big chart of the IPCC AR4 predictions put up by Lucia. (Even in the year 2000, their back-cast models using temperature data up to that time were too high by 0.2C or so and they would now be too high by about 0.32C).
http://rankexploits.com/musings/wp-content/uploads/2008/08/compare-to-copenhagen.jpg
I think a random noise signal should not have any component with a wavelength longer than the resolution. It is very easy to say that short term variations are natural variations which will disappear with a 30 years average. And long term variations are in fact climate feedback with a constant trend.
I thought there was a standard scientific method to analyze a signal you think might be random noise. I thought a settled science built by thousands of scientists should have a clear explanation for all natural variations and the absence of long term natural variations when the record is full of it.
Gavin’s Goalposts
Gavin Schmidt of NASA has a website called RealClimate. A couple of years ago there was a post on signs of climate change.
http://www.realclimate.org/index.php/archives/2007/12/a-barrier-to-understanding/
In the discussion thread Daniel Klein asks at #57:
OK, simply to clarify what I’ve heard from you.
(1) If 1998 is not exceeded in all global temperature indices by 2013, you’ll be worried about state of understanding
(2) In general, any year’s global temperature that is “on trend” should be exceeded within 5 years (when size of trend exceeds “weather noise”)
(3) Any ten-year period or more with no increasing trend in global average temperature is reason for worry about state of understandings
I am curious as to whether there are other simple variables that can be looked at unambiguously in terms of their behaviour over coming years that might allow for such explicit quantitative tests of understanding?
[Response: 1) yes, 2) probably, I’d need to do some checking, 3) No. There is no iron rule of climate that says that any ten year period must have a positive trend. The expectation of any particular time period depends on the forcings that are going on. If there is a big volcanic event, then the expectation is that there will be a cooling, if GHGs are increasing, then we expect a warming etc. The point of any comparison is to compare the modelled expectation with reality – right now, the modelled expectation is for trends in the range of 0.2 to 0.3 deg/decade and so that’s the target. In any other period it depends on what the forcings are. – gavin]
The key words in the criticism you received were “sinusoidal data”. Your graph was sinusoidal, Hansen’s graphs are not!
A little discipline is needed when filtering noisy and/or cyclic data:
1. A way to to deal with end effects is to “window” the data. This also makes point 2 dramatically apparent – current data has little influence.
2. A moving average (or other symetrical filter method) is an estimate for the actual value of the data at the centre of the averaging period, not the end. Filtered data should be time-shifted to display correctly.
Are either of these methods used on climate data?
MG
Jean Meeus says:
July 31, 2010 at 7:33 am
Not the Jean Meeus of ‘Astronomical Algorithms’, I presume? Or, just a pseudonym? On the unlikely chance that it is the same, thanks for a very useful book.
Icarus says:
July 31, 2010 at 8:23 am
“Hansen’s conclusion is based on 30 years of data. Isn’t that enough?”
Based on what? What if there is a strong 60-ish year cycle in the data? (There is, BTW.)
Steve,
You must be enjoying this thread immensely.
You invite readers to observe the GISS data from 1880 to 2010, note Dr. Hansen is quoted drawing significance from the period late 1970s to present, and ask your readers “Is Hansen correct, or does he need some serious graphing lessons?”.
Well, it is as obvious as the proverbial dogs balls that Dr. Hansen references a “trough to crest” period in the data. Which, as per the background context you provided, makes the critique of your analysis on an earlier thread – you used mean to mean – trivial.
Not one of the comments – particularly not one of the comments from the warming-fast side – answers your question. Most of the discussion focuses on the last 10 or 12 years. Some explores the necessary minimum period, concluding correctly that 30 years is not enough (which is true, but you were looking for a simpler observation).
So let me be the first: any information from wandering data (I avoid the term cyclical, since that term implies a regularity not present in climate data) must be crest to crest or trough to trough or mid to mid. Since we are currently at a crest in the GISS data presented above, the only acceptable periods ending in 2010 would be 1940 to 2010 or 1900 to 2010. Anything else is intellectually dishonest.
RW says:
July 31, 2010 at 8:45 am
“2. El Niño is not a cyclical phenomenon. It is quasi-periodic. Its amplitude varies enormously. The 1997-8 one was considerably larger than the most recent one.”
An amplitude modulated signal is not periodic? Allow me to introduce you to a guy named Fourier.
Icarus
You wrote, “You will have to wait about 20 years before you can say anything at all about how accurate that projection was.”
If you compare the global warming rate for the last four decades, you see a definite deceleration for the last period from 2000 to 2010 as shown here:
http://www.woodfortrees.org/plot/hadcrut3vgl/from:1970/to:1980/plot/hadcrut3vgl/from:1970/to:1980/trend/plot/hadcrut3vgl/from:1980/to:1990/plot/hadcrut3vgl/from:1980/to:1990/trend/plot/hadcrut3vgl/from:1990/to:2000/plot/hadcrut3vgl/from:1990/to:2000/trend/plot/hadcrut3vgl/from:2000/to:2010/plot/hadcrut3vgl/from:2000/to:2010/trend
As a resutl, IPCC’s Exageration Factor is 6.7!
GLOBAL MEAN TEMPERATURE CYCLES
http://www.woodfortrees.org/plot/hadcrut3vgl/from:1880/to:2010/compress:60/detrend:0.775/offset:0.518/plot/hadcrut3vgl/from:1880/to:2010/trend/detrend:0.775/offset:0.518
The above graph for the data from CRU shows the following periods for relative global cooling and warming phases:
1) 30-years of global cooling from 1880 to 1910
2) 30-years of global warming from 1910 to 1940
3) 30-years of global cooling from 1940 to 1970
4) 30-years of global warming from 1970 to 2000
If this pattern that was valid for 120 years is assumed to be valid for the next 20 years, it is reasonable to predict:
5) 30-years of global cooling from 2000 to 2030
The above graph also provides the important result that the years 1880, 1910, 1940, 1970, 2000, 2030 etc are (Global Mean Temperature Anomaly) GMTA trend turning points, so meaningful GMTA trends can be calculated only between these successive GMTA turning point years.
Here are then the global warming rates for the last 100 years:
Global warming from 1910 to 1940 => +0.15 deg C per decade
Slight global cooling from 1940 to 1970 = -0.03 deg C per decade
Global warming from 1970 to 2000 => +0.16 deg C per decade
Slight global warming from 2000 to 2010 => +0.03 deg C per decade
http://www.woodfortrees.org/plot/hadcrut3vgl/from:1910/to:1940/compress:12/plot/hadcrut3vgl/from:1910/to:1940/trend/plot/hadcrut3vgl/from:1970/to:2000/compress:12/plot/hadcrut3vgl/from:1970/to:2000/trend/plot/hadcrut3vgl/from:1940/to:1970/compress:12/plot/hadcrut3vgl/from:1940/to:1970/trend
@RW
Showing a 30 year trend as being strongly positive and with agreement between the various sources while the 10 year trend in the main post shows different results simply reinforces the idea that trends should be regarded with considerable suspicion as the data clearly deviates from a linear trend.
As regards using the peak or trough of the trace, surely we are interested in the way that the mean temperature varies, i.e. a zero-phase, low pass filtered version of the signal. While the smoothing using a rectangular average creates something nearer to an integral of trace, it is not ideal for many reasons. There is substantial theory underpinning near-optimal filters. When these are constructed auto-regressively, much more useful responses can be obtained withough having to employ long impulse responses as in the case of a moving average filter so that there isn’t a “dead zone” at the beginning and end of the record.
The reality is, statistically speaking, the temperature record is totally consistent with a warming of 1°C/century. There is a 60-years cycle, see Girma’s comment. And the 90s are about the most warming you can expect in a 10-years period. The latter led to exaggerated claims about climate sensitivity. I challenge anyone to prove that 1°C of warming per century is impossible considering the well known 60-years cycle.
The now common practice of extrapolating “trends” into the distant future is as unscientific as unscientific can become. My college professors in chemistry and physics all discussed this matter with their students, and each and all said, “Never do it”.
Of course, that was back in the 1950s. Wise college professors back then.
If all the natural variation is explained by scientific literature in a quantifiable way, simply remove it all from the trend and show us what is left over as unexplained warming. Just because the trend is X/decade (4X for GISS), doesn’t mean you have successfully explained how much warming is coming from CO2. The natural variability should be removed from the trend. Of course, if anyone does this, they will need to list each component of variability, and then quantify how much was subtracted from the trend. It is the only way to say anything one way or the other about AGW.
Re: Icarus, link to temperature graphs with 3 to 22 year rolling trends.
The 1978 to present 0.18C*/decade trend is certainly clear. What happens when you do that for the 1850 to 1978 period?
And 0.18C*/decade is certainly not catastrophic, and, I understood, not indicative of pCO2 forcing (with H2O enhancement). If water enhancement exists, does this last 30-year consistency of 0.18C*/decade trend not mean that CO2 forcing is really something like 0.06C*/decade? Do any of this last 30-year trend support the IPCC model of CO2/H2O forcing and catastrophic, i.e. 3-5*C/100 year global warming from fossil fuel usage?
Each time I refocus on the times of supposed AGW, i.e. after 1965, I bump into a contradiction in temperature increase and the disaster model used to justify the CO2 tax and punish program. Am I missing something here? Where is the data or observation that supports the model?
Girma: 10 years can only ever show you the natural interannual variability, not the trend, because that variability is ~0.2C from year to year, and the trend is ~0.2C per decade. In other words, interannual variability can completely mask the long term trend in any 10 year period. Make sense?
” Icarus says:
July 31, 2010 at 11:16 am
…variability is ~0.2C from year to year, and the trend is ~0.2C per decade. In other words, interannual variability can completely mask the long term trend in any 10 year period. Make sense?”
What does not make sense is the fact that Hansen observes the trend of 0.2°C per decade (for 1998 to 2009), whereas Jones or Spencer do not.
Steven Goddard wrote: “I took a lot of criticism from people for not measuring ‘crest to crest or trough to trough.'”
And as I noted in that thread:
http://wattsupwiththat.com/2010/07/02/arctic-ice-increasing-by-50000-km2-per-year/#comment-422082
You should have converted to anomalies first:
http://i49.tinypic.com/25jeemw.jpg
Bart: El Niño is not periodic. It is quasi-periodic, occurring at intervals typically of 2-7 years.
RC Saumarez: “trends should be regarded with considerable suspicion as the data clearly deviates from a linear trend” – well, certainly the trend may not be linear, but we could avoid all sorts of misinterpretation problems if no trend was ever plotted without the ±1σ trend lines also marked.
Graphing lessons won’t help Hansen. His choice of 1880 is a cherry-pick, he knows it, and this is all about handing over something to rubber-stamp as “proof of global warming”.
November has a chill in it already.
When I look at the long term HADCRUT and superimpose a linear trend on that, it shows a 0.7 rise over the 160 years. However that is with 3 peaks and 2 troughs of a lumpy sine wave, which correlates very closely with the PDO cycles, and could be said to be entirely attributable to natural variation.
There is every indication that we are past the beginning of a cool cycle, and if the past is any guide to the future, it is likely that we will finish this cool cycle about 0.7C below where we are now. That will drag the linear trend down to about a 0.6C rise over 180 years.
Given we were coming out of a little ice age, I don’t think there’s anything here to be getting alarmed about. In fact, what’s strange about all this, is just how stable the climate has been when you consider all of the factors that contribute to climate change.
Further, it is only this 0.6C rise that we can attribute anything not natural to, such as greenhouse gasses or other forcings, and that is assuming that this 180yr trend isn’t just part of a longer cycle, possibly headed down again given our weak sun.
Dr Lindzen was right. Our descendants are going to have a good old laugh at our version of arguing over angels on pins.
Just for reference, here is a table for an unofficial five-segment line-anomaly plot derived from the monthly NSIDC Arctic ice-extent data from 1978 to 2010. I created the master anomaly curve by subtracting a Fourier series representation of the average annual freeze-melt curve computed over the largest complete integer-year period.
The interior line segment breaks were automatically optimized for best fit to the master curve.