-- by Horatio Algeranon
The denialosphere is all abuzz,
Noisier than it ever was.
A paper by Carter and McClean,
(Careening down the passing lane)
Claims that temperature's rise and fall,
Hugs el Nino, as flies hug the wall,
With little room for greenhouse gases,
To explain the melting of glacier masses.
But alas, the overconfident tone,
Is wishful thinking, so hold the phone.
The authors have greatly oversold,
The "correlation", truth be told,
Between ENSO and the global temp,
As pot proponents do with hemp,
And most of the recent temperature rise,
Is not explained as they surmise.
In fact, the results are nothing new
The climate scientists knew this too:
ENSO causes "downs and ups",
In global temp like brief "hiccups",
While over decades, the temperature rises,
From greenhouse gases, with few surprises.
The denialosphere is all abuzz,
Noisier than it ever was.
A paper by Carter and McClean,
(Careening down the passing lane)
Claims that temperature's rise and fall,
Hugs el Nino, as flies hug the wall,
With little room for greenhouse gases,
To explain the melting of glacier masses.
But alas, the overconfident tone,
Is wishful thinking, so hold the phone.
The authors have greatly oversold,
The "correlation", truth be told,
Between ENSO and the global temp,
As pot proponents do with hemp,
And most of the recent temperature rise,
Is not explained as they surmise.
In fact, the results are nothing new
The climate scientists knew this too:
ENSO causes "downs and ups",
In global temp like brief "hiccups",
While over decades, the temperature rises,
From greenhouse gases, with few surprises.
ENSO: "The El Niño/Southern Oscillation (ENSO) phenomenon is the biggest player in the game of year-to-year climate variability." -- "Multivariate ENSO Index (NOAA)
As illustrated above, the year to year "oscillations" or "ups and downs" in global temperature follow the changes in ENSO fairly closely (the two are "correlated"), but the pronounced upward "ramp" (or trend) in temperature (about 0.17 deg C per decade) from around 1975-present is not accounted for by these variations in ENSO.
Scientists who have studied the issue extensively have concluded that most of this "ramp" is attributable to human activities, particularly the emission of greenhouse gases (like CO2 and methane) into the atmosphere. Note that the upward sloping temperature ramp has year-to-year "ups and downs" (noise) "riding" (superimposed) on it.
As explained by Tamino here, the analysis method of Carter et al (referred to in the above ditty and found here) effectively removes this ramp! (ie, removes its impact on the correlation between temperature and ENSO) It is as if the ramp has been laid flat, so all that remains are the "ups and downs" from one year to the next -- fluctuation due to ENSO and volcanic eruptions, primarily.
Though the strength of the correlation between ENSO and global temperature is overinflated by their (effective) " ramp (trend) removal", Carter et al would at least remain on stable scientific ground if they restricted their claims to statements about the correlation between ENSO and the "noise" in global temperature (superimposed on the trend).
If.
But they go on to make some claims that are simply not supported by the evidence and -- because of their "ramp (trend) removal" -- can not even be made in principle.
Composite created from the NOAA ENSO graphic found here and the NASA GISS graphic of global mean temperature anomaly found here.
Scientists who have studied the issue extensively have concluded that most of this "ramp" is attributable to human activities, particularly the emission of greenhouse gases (like CO2 and methane) into the atmosphere. Note that the upward sloping temperature ramp has year-to-year "ups and downs" (noise) "riding" (superimposed) on it.
As explained by Tamino here, the analysis method of Carter et al (referred to in the above ditty and found here) effectively removes this ramp! (ie, removes its impact on the correlation between temperature and ENSO) It is as if the ramp has been laid flat, so all that remains are the "ups and downs" from one year to the next -- fluctuation due to ENSO and volcanic eruptions, primarily.
Technically, the ramp is transformed to a constant value, so the trend no longer has any impact on the correlation. Carter et al actually used the Southern Oscillation Index (SOI, an alternative to the MEI shown above) and UAH tropospheric temperatures for their analysis, but Horatio has used the MEI and GISS land-ocean surface temperatures in the graphic above to illustrate what is going on. Illustration of the basic idea does not depend on the specific index or temperature data set used.Since Carter et al exclude the time periods with large volcanic eruptions from their analysis, it is not surprising that (after removing the impact of the ramp) they find a fairly high "correlation" (synchronous change) between ENSO and global temperature. The effect of ENSO comprises most of the remaining "noise" in the temperature record!
Though the strength of the correlation between ENSO and global temperature is overinflated by their (effective) " ramp (trend) removal", Carter et al would at least remain on stable scientific ground if they restricted their claims to statements about the correlation between ENSO and the "noise" in global temperature (superimposed on the trend).
If.
But they go on to make some claims that are simply not supported by the evidence and -- because of their "ramp (trend) removal" -- can not even be made in principle.
Take the following statement from the conclusion to the paper, for example:
"This study has shown that natural climate forcing associated with ENSO is a major contributor to variability and perhaps recent trends in global temperature, a relationship that is not included in current global climate models."
Worse still is this statement from a press release by Bob Carter, one of the papers' authors:
So, after effectively removing the impact of greenhouse gases (the trend) from his correlation analysis, Carter claims there is "little room for any warming driven by human carbon dioxide emissions"?!“The close relationship between ENSO and global temperature, as described in the paper, leaves little room for any warming driven by human carbon dioxide emissions.”
"Hogwash!" says Horatio.
Listen to "ENSO it goes" by Billy Joel.
The primary motivation for using the “first derivative” (the "slope") of spectra rather than the spectra themselves is removing the effect of the “baseline” on the correlation -- removing the impact of “humps", "ramps", etc, effectively leaving only the “peaks” that were formerly "riding" on that baseline.
While this may be desirable for many spectroscopy cases where the baseline is not important, it is highly undesirable in a case like global temperature in which the "baseline" (eg, the upward sloping temperature "ramp") is important, particularly if one is trying to determine the correlation of global temperature with some factor.
In fact, Tamino has shown here how dramatically the "first derivative method" of Carter et al inflates the correlation, from about 0.036 (for the correlation between SOI [a measure of ENSO] and temperature) to about 0.72 (for the correlation between "first derivative of SOI" and "first derivative of temperature")
But one need not understand the mathematics of "correlation analysis" to see that the "first derivative of SOI" accounts much better for changes in the "first derivative of temperature" than SOI accounts for changes in the temperature itself. The concept is illustrated by the graphs below.
On these, both "SOI" and "temperature" are "simulated" (artificial) -- meant to illustrate the effect of taking the first derivative of temperature and SOI before finding the correlation (effectively, what Carter et al did)
On the first graph, temperature (blue) is comprised of an upward sloping "ramp" (meant to mimic the effect of increasing greenhouse gas concentration) on which a single sinusoidal oscillation with the same "period" as SOI (red) has been superimposed (meant to simulate the effect of an "ENSO-like" oscillation [gauged with SOI]. Note: in reality ENSO is not actually a periodic phenomenon but will be treated as such for the purpose of illustrating the concept involved here). The observed "lag" between temperature and SOI (about 7 months) has been set to zero.
Update (Sept 1)
As Horatio noted in a comment at More Grumbine Science, the method employed by Carter et al in their recent paper essentially amounts to “first derivative correlation”, a standard method used in spectroscopy for "spectral searching/matching” -- ie, for finding the spectrum in a "library of spectra" that comes “closest” to an unknown "test" spectrum, as described here
The primary motivation for using the “first derivative” (the "slope") of spectra rather than the spectra themselves is removing the effect of the “baseline” on the correlation -- removing the impact of “humps", "ramps", etc, effectively leaving only the “peaks” that were formerly "riding" on that baseline.
While this may be desirable for many spectroscopy cases where the baseline is not important, it is highly undesirable in a case like global temperature in which the "baseline" (eg, the upward sloping temperature "ramp") is important, particularly if one is trying to determine the correlation of global temperature with some factor.
In fact, Tamino has shown here how dramatically the "first derivative method" of Carter et al inflates the correlation, from about 0.036 (for the correlation between SOI [a measure of ENSO] and temperature) to about 0.72 (for the correlation between "first derivative of SOI" and "first derivative of temperature")
But one need not understand the mathematics of "correlation analysis" to see that the "first derivative of SOI" accounts much better for changes in the "first derivative of temperature" than SOI accounts for changes in the temperature itself. The concept is illustrated by the graphs below.
On these, both "SOI" and "temperature" are "simulated" (artificial) -- meant to illustrate the effect of taking the first derivative of temperature and SOI before finding the correlation (effectively, what Carter et al did)
On the first graph, temperature (blue) is comprised of an upward sloping "ramp" (meant to mimic the effect of increasing greenhouse gas concentration) on which a single sinusoidal oscillation with the same "period" as SOI (red) has been superimposed (meant to simulate the effect of an "ENSO-like" oscillation [gauged with SOI]. Note: in reality ENSO is not actually a periodic phenomenon but will be treated as such for the purpose of illustrating the concept involved here). The observed "lag" between temperature and SOI (about 7 months) has been set to zero.
As shown above, taking the first derivative effectively removes the impact of the ramp (trend due to greenhouse gas increases) on the correlation. While the correlation between SOI and temperature for this artificial case is only 0.22 (fairly low), taking the derivative of both before calculating the correlation boosts the value to "1" (perfect correlation), again for the artificial case. Of course, in reality, the correlation is less than perfect (even after taking the derivatives) because there is other noise ( in addition to SOI) , closer to the case illustrated below.
For this second case, temperature is comprised of an upward "ramp" on which two sinusoidal oscillations have been superimposed: one with the same "period" as SOI (meant to simulate the effect of ENSO[SOI], again with the caveat that ENSO is not really periodic) and another with a longer period (meant to simulate other temperature "noise"). The observed "lag" between temperature and SOI has again been set to zero.
For this second case, temperature is comprised of an upward "ramp" on which two sinusoidal oscillations have been superimposed: one with the same "period" as SOI (meant to simulate the effect of ENSO[SOI], again with the caveat that ENSO is not really periodic) and another with a longer period (meant to simulate other temperature "noise"). The observed "lag" between temperature and SOI has again been set to zero.
Again, while the correlation between SOI (red) and temperature (blue) for the simulated (artificial) case comes out rather low (only 0.14), the correlation between the first derivative of SOI (green) and the first derivative of temperature (gold) for the same case comes out fairly high (0.76), which again demonstrates how significantly the "derivative method" can inflate the correlation. As Tamino has shown, this was indeed the case -- the correlation was actually boosted from 0.036 to 0.72 by the Carter derivative method.
