 |
|
| Science Forum Index » Energy Forum » Lindzen and climate feedback... |
|
Page 1 of 5 Goto page 1, 2, 3, 4, 5 Next |
|
| Author |
Message |
| Earl Evleth... |
 Posted: Sun Oct 04, 2009 3:38 am |
|
|
|
Guest
|
On 4/10/09 7:19, in article
bc16ab84-0642-4a91-a355-a317eaca9753 at (no spam) f10g2000vbf.googlegroups.com, "Roger
Coppock" <rcoppock at (no spam) adnc.com> wrote:
[quote:24a0a9f9b4]The deniers
here don't know how to compute correlations
and regressions.
[/quote:24a0a9f9b4]
Curious since they are always regressive |
|
|
| Back to top |
|
|
|
| Rob... |
| Posted: Wed Oct 07, 2009 2:55 pm |
|
|
|
Guest
|
On Oct 3, 2:30 pm, Bill Ward <bw... at (no spam) ix.REMOVETHISnetcom.com> wrote:
[quote:cf694b8caa]On Sat, 03 Oct 2009 09:43:28 -0700,RobDekkerwrote:
This thread discusses a paper published by Lindzen et al. :
http://www.seas.harvard.edu/climate/seminars/pdfs/
lindzen.choi.grl.2009.pdf
It is the continuation of a discussion that was out of place under a
different subject.
"Bill Ward" <bw... at (no spam) ix.REMOVETHISnetcom.com> wrote in message
news:paqdnbz9kLXQXFjXnZ2dnUVZ_vqdnZ2d at (no spam) giganews.com... ........
Have you seen this?:
http://www.seas.harvard.edu/climate/seminars/pdfs/
lindzen.choi.grl.2009.pdf
It makes a pretty good case for negative feedbacks in reality. The
models show positive feedback, but don't match the satellite data.
Thanks for the link. I did not see this study before.
Interesting note:
"Results also show, the feedback in ERBE is mostly from shortwave
radiation while the feedback in the models is mostly from longwave
radiation."
Interesting is that increased water temp causes increased SHORTwave
radiation.
I just assumed it's the satellite seeing reflected SW from the
additional cloud cover when the water is warmer. It must originate
from the Sun - the Earth doesn't emit much SW.
Does your lack of comment indicate agreement?
[/quote:cf694b8caa]
I did not want to make any assumptions, but now that you mention it,
you bring up an issue that I pointed out in some of our previous
discussions :
If you look at the ERBE graphs, you see that SW radiation became lower
in intensity over the decade(s).
So if you assume that SW radiation relates to cloud cover (as you
mention above), then even though the planet's climate and ocean waters
became warmer over that period, the amount of cloud coverage became
smaller.
[quote:cf694b8caa]
I've seen that before. In Wong's paper :
http://ams.allenpress.com/perlserv/?request=get-abstract&doi=10.1175%
2FJCLI3838.1&ct=1
That is Wong et al, 2006, which also does a SW analysis using ERBE
over the same 1985-1999 period !
All I get is the abstract and a paywall. Can you excerpt the relevant
parts under fair use?
Here is the entire Wong et al (2006) paper :
http://asd-www.larc.nasa.gov/~tak/wong/f20.pdf
Thanks. That helps a lot.
Here is the picture presented in this paper :
http://eosweb.larc.nasa.gov/PRODOCS/erbe/quality_summaries/s10n_wfov/
ed3rev1_comparison.jpg
It shows that SW radiation is highly variable, even more so than LW
radiation.
Specifically, the SW radiation spikes around 1991/1992 are quite
difficult to explain if SW radiation is related to ocean temps alone.
Perhaps reflection from high particulates?
[/quote:cf694b8caa]
Probably, from the Pinatubo eruption, most likely...
[quote:cf694b8caa]
Moreover, long term shows actually a DECREASE of SW radiation over the
1985-1999 period (indicating positive feedback). This as opposed to
the paper you present, which shows an INCREASE over the same period,
using the same satellite measurements.
Not having the Wong paper, I'll have to trust the reviewers. Lindzen
was later, so I assume he's aware of Wong.
So these two papers seem to contradict each other using the same data..
Now either the ocean temperatures actually cooled over the 85-99
period, or I misinterpret the data presented here, or there is
something really weird going on between these two papers.
Alternatively, there is so much noise in the data that we cannot draw
any conclusions at this time.
We can put limits on the feedback. It doesn't seem to be positive,
from the sensitivity plots in Lindzen. The models assumed positive
feedback, and they had a negative correlation to the actual data.
OK. I read Lindzen's paper in detail. Here are my findings : Lindzen is
twisting the data from ERBE. In fact, he makes some pretty big mistakes
that falsify his conclusions.
Here we go : Lindzen claims there is negative climate feedback visible
from the correlation between the NET flux of outbound radiation versus
the SST (sea surface temperature) in the tropics.
He presents SST info in figure 1a, top-left corner graph. That is data
from the National Centers for Environmental Prediction (his quote). I
have questions about that graph, but that is not relevant for what
follows.
He presents outbound radiation flux from the ERBE rev.3 data set, just
like Wong et. al.
The LW ERBE data is in figure 1a, second graph from the left top (below
the SST graph).
The SW ERBE data is in figure 1b, first graph in the left top corner.
The NET flux -(LW+SW) is not given in his paper (not sure why not,
because he basis his conclusions on the NET graph). However, Wong et al
(2006) (see link above) gives all three graphs in figure 2 of Wong's
paper.
http://asd-www.larc.nasa.gov/~tak/wong/f20.pdfHere is an amplified
version of that graph :
http://chriscolose.files.wordpress.com/2009/03/
wongetal2006jclimateoceanhtstorage1.jpg
Now pay very close attention to the bottom (NET) graph in this picture.
And compare that to the SST graph (figure 1a, top left corner) in
Lindzen's paper.
http://www.seas.harvard.edu/climate/seminars/pdfs/
lindzen.choi.grl.2009.pdf
At first sight, there does not seem to be ANY correlation between these
two (NET and SST) graphs.
Maybe there is a little bit of long term negative correlation
(indicating positive climate feedback), but the anomalies of the 1988
and 1998 in SST and 1992 in the ERBE graph make any correlation
difficult.
Visual correlations are notoriously unreliable. That's what signal
analysis is for.
[/quote:cf694b8caa]
Agreed. As long as the correct data is used...
[quote:cf694b8caa]
However, Lindzen presents correlation data in Fugure 2 of his paper, top
left corner, to show the 'negative' climate feedback that he bases his
conclusions on.
Note that he uses only 13 data points for that picture.
Also note he explains why:
Begin quote from page 2
The next obvious question is whether fluctuations with
the time scales associated with feedback processes exist in the
observed data and models. Figures 1a and 1b show that such
fluctuations (DFlux) are amply available in OLR and SWR,
although data are not currently available in some periods in
1993 and 1999. However, it is possible that many of the very
small fluctuations are simply noise. Restricting oneself to
fluctuations in SST (DSST) which exceed 0.2 K still leaves
nine cases in the available data (red and blue lines in
Figure 1a). Note that appreciable fluctuations of the anoma-
lies are due to El Nino events (in 1982/83, 1986/87, 1991/92, ̃
and 1997/9 , La Nina events (in 1988/90), and Pinatubo eruption (in
1991) [Wielicki et al., 2002a; Wong et al., 2006].
end quote
Also see the explanation that follows and table 1 on page 4 to see why he
picked those points. None of the statistical analyses resulted in a
positive slope.
[/quote:cf694b8caa]
Sure. But I cannot possibly find the data points that he claims are
responsible for the slope. See here again :
[quote:cf694b8caa]
Also note that
these 13 data points are incorrectly placed !!! For example, the data
point in the top-right corner is a +0.7 C / +5 W/m^2 data point.
In the SST graph, there is only one +0.7 W/m^2 point, which is in 1998.
However, in 1998, the ERBE NET graph shows a close to 0 W/m^2 (and
possibly negative) radiation delta. So that point needs to move way
south in his Figure 2.
Two other points, on the other side, are at -0.5 C. The only points at
-0.5 C in the SST graph are 1989 and 1985. However, the ERBE graoh shows
close to 0 or mildly positive NET radiation flux for these years, while
Lidzen puts them at -1 and -3 W/m^2.
So the points in his graph need to go up north.
The points are selected to have a minimum of .1K difference to eliminate
high frequency noise. Note the axes are dflux and dsst. Look at the red
and blue sections in the fig 1a SST graph. I believe that represents the
0.2 max delta he mentions in the quote above.
Table one shows the thirteen (0.1K) point case he chose had slightly
better statistics.
Lindzen obviously made some major mistakes in putting this Fugure 2,
top-left corner picture in place.
I think you may be misinterpreting it. It seems to me to be exactly what
he says it is.
[/quote:cf694b8caa]
Let's try this again :
Just for the fun of it, focus on the first 'extreme' point in the
right-top corner of his ERBE delta-flux/SST graph : +5W/m^2 delta-
flux / +0.7 C delta-SST.
Try to find that point in the data sets : +0.7 C delta-SST is only
occurring in 1998 (according to his own SST graph).
Then find the delta-flux in the NET graph for 1998 from ERBE edition
3 :
http://chriscolose.files.wordpress.com/2009/03/wongetal2006jclimateoceanhtstorage1.jpg
and you find something close to 0 W/m^2 delta-flux. Certainly not +5W/
m^2 !!
Similar problems with the 2 or 3 other extreme data points that
ultimately determine the slope and 'fitting' numbers from table 1.
So there is something very, very fishy here..
[quote:cf694b8caa]
When all points are adjusted to the ERBE rev 3 data set, then his
picture will look very much like any of the other pictures (from the
models) that he so prominently displays as well.
The bottom line is that the negative feedback that he claims is simply
not there in the ERBE rev 3 data set.
Where do you find a direct correlation between SW and SST in Wong? You
need both to see the negative feedback.
[/quote:cf694b8caa]
Wong did not make any SST analysis, nor any negative feedback
analysis.
He only mentioned that the radiation numbers match accurately with
heat-storage numbers, and that the radiation numbers match with
existing models and current climate predictions from that.
I think I have mentioned once that negative feedback can be deduced
from the the ERBE SW radiation reduction measurements : Climate
warmer : less SW radiation. Thus negative feedback.
[quote:cf694b8caa]
I personally believe that there is too much noise in these signals to
draw any conclusion from. The SST varies over +/- 0.5 C and the ERBE
data set shows variability over 2-3 W/m^2 due to simple weather pattern
changes (ignoring the -6 W/m^2 Pinatubo 1991 eruption anomaly). So to
draw conclusions on climate sensitivity from such noisy signals seems
fairly futile from a scientific point of view.
That's why he reduced the noise by focusing on the ratio of the slopes.
Think of it as a crude but clever bandpass filter.
However, Lindzen could have at least get the data from his own graphs
correct. That blunt inconsistency that invalidates his own conclusions
is a very grave mistake for any scientific publication.
I think you need to read the Lindzen paper again, focusing a little more
on the details. Remember, he cites Wong, so he's aware of the issues
raised.- Hide quoted text -
[/quote:cf694b8caa]
I read it. In great detail.
The slope that he calculates (from ERBE and SST data) depends on a
very small set of measurement points, which do not seem to be
correctly determined (see above).
Consequently, I do not trust this paper one bit.
Rob |
|
|
| Back to top |
|
|
|
| Last Post... |
| Posted: Wed Oct 07, 2009 3:03 pm |
|
|
|
Guest
|
On Oct 3, 12:43pm, "Rob Dekker" <r... at (no spam) verific.com> wrote:
[quote:dee0797b93]This thread discusses a paper published by Lindzen et al. :http://www.seas.harvard.edu/climate/seminars/pdfs/lindzen.choi.grl.20...
It is the continuation of a discussion that was out of place under a
different subject.
"Bill Ward" <bw... at (no spam) ix.REMOVETHISnetcom.com> wrote in message
news:paqdnbz9kLXQXFjXnZ2dnUVZ_vqdnZ2d at (no spam) giganews.com...
........
Have you seen this?:
http://www.seas.harvard.edu/climate/seminars/pdfs/
lindzen.choi.grl.2009.pdf
It makes a pretty good case for negative feedbacks in reality. The
models show positive feedback, but don't match the satellite data.
Thanks for the link. I did not see this study before. Interesting note :
"Results also show, the feedback in ERBE is mostly from shortwave
radiation while the feedback in the models is mostly from longwave
radiation."
Interesting is that increased water temp causes increased SHORTwave
radiation.
I just assumed it's the satellite seeing reflected SW from the additional
cloud cover when the water is warmer. It must originate from the Sun -
the Earth doesn't emit much SW.
I've seen that before. In Wong's paper :
http://ams.allenpress.com/perlserv/?request=get-abstract&doi=10.1175%
2FJCLI3838.1&ct=1
That is Wong et al, 2006, which also does a SW analysis using ERBE over
the same 1985-1999 period !
All I get is the abstract and a paywall. Can you excerpt the relevant
parts under fair use?
Here is the entire Wong et al (2006) paper :http://asd-www.larc.nasa.gov/~tak/wong/f20.pdf
Here is the picture presented in this paper :
http://eosweb.larc.nasa.gov/PRODOCS/erbe/quality_summaries/s10n_wfov/
ed3rev1_comparison.jpg
It shows that SW radiation is highly variable, even more so than LW
radiation.
Specifically, the SW radiation spikes around 1991/1992 are quite
difficult to explain if SW radiation is related to ocean temps alone.
Moreover, long term shows actually a DECREASE of SW radiation over the
1985-1999 period (indicating positive feedback). This as opposed to the
paper you present, which shows an INCREASE over the same period, using
the same satellite measurements.
Not having the Wong paper, I'll have to trust the reviewers. Lindzen was
later, so I assume he's aware of Wong.
So these two papers seem to contradict each other using the same data.
Now either the ocean temperatures actually cooled over the 85-99 period,
or I misinterpret the data presented here, or there is something really
weird going on between these two papers. Alternatively, there is so much
noise in the data that we cannot draw any conclusions at this time.
We can put limits on the feedback. It doesn't seem to be positive, from
the sensitivity plots in Lindzen. The models assumed positive feedback,
and they had a negative correlation to the actual data.
OK. I read Lindzen's paper in detail. Here are my findings :
Lindzen is twisting the data from ERBE. In fact, he makes some pretty big
mistakes that falsify his conclusions.
Here we go : Lindzen claims there is negative climate feedback visible from
the correlation between the NET flux of outbound radiation versus the SST
(sea surface temperature) in the tropics.
He presents SST info in figure 1a, top-left corner graph. That is data from
the National Centers for Environmental Prediction (his quote). I have
questions about that graph, but that is not relevant for what follows.
He presents outbound radiation flux from the ERBE rev.3 data set, just like
Wong et. al.
The LW ERBE data is in figure 1a, second graph from the left top (below the
SST graph).
The SW ERBE data is in figure 1b, first graph in the left top corner.
The NET flux -(LW+SW) is not given in his paper (not sure why not, because
he basis his conclusions on the NET graph).
However, Wong et al (2006) (see link above) gives all three graphs in figure
2 of Wong's paper.http://asd-www.larc.nasa.gov/~tak/wong/f20.pdf
Here is an amplified version of that graph :http://chriscolose.files.wordpress.com/2009/03/wongetal2006jclimateoc...
Now pay very close attention to the bottom (NET) graph in this picture.
And compare that to the SST graph (figure 1a, top left corner) in Lindzen's
paper.http://www.seas.harvard.edu/climate/seminars/pdfs/lindzen.choi.grl.20...
At first sight, there does not seem to be ANY correlation between these two
(NET and SST) graphs.
Maybe there is a little bit of long term negative correlation (indicating
positive climate feedback), but the anomalies of the 1988 and 1998 in SST
and 1992 in the ERBE graph make any correlation difficult.
However, Lindzen presents correlation data in Fugure 2 of his paper, top
left corner, to show the 'negative' climate feedback that he bases his
conclusions on.
Note that he uses only 13 data points for that picture.
Also note that these 13 data points are incorrectly placed !!!
For example, the data point in the top-right corner is a +0.7 C / +5 W/m^2
data point.
In the SST graph, there is only one +0.7 W/m^2 point, which is in 1998.
However, in 1998, the ERBE NET graph shows a close to 0 W/m^2 (and possibly
negative) radiation delta. So that point needs to move way south in his
Figure 2.
Two other points, on the other side, are at -0.5 C. The only points at -0..5
C in the SST graph are 1989 and 1985. However, the ERBE graoh shows close to
0 or mildly positive NET radiation flux for these years, while Lidzen puts
them at -1 and -3 W/m^2.
So the points in his graph need to go up north.
Lindzen obviously made some major mistakes in putting this Fugure 2,
top-left corner picture in place.
When all points are adjusted to the ERBE rev 3 data set, then his picture
will look very much like any of the other pictures (from the models) that he
so prominently displays as well.
The bottom line is that the negative feedback that he claims is simply not
there in the ERBE rev 3 data set.
I personally believe that there is too much noise in these signals to draw
any conclusion from. The SST varies over +/- 0.5 C and the ERBE data set
shows variability over 2-3 W/m^2 due to simple weather pattern changes
(ignoring the -6 W/m^2 Pinatubo 1991 eruption anomaly). So to draw
conclusions on climate sensitivity from such noisy signals seems fairly
futile from a scientific point of view.
However, Lindzen could have at least get the data from his own graphs
correct. That blunt inconsistency that invalidates his own conclusions is a
very grave mistake for any scientific publication.
[/quote:dee0797b93]
Bullshit
Either way short term or long term the data
doesn't support man made global warming?
short term
http://junkscience.com/MSU_Temps/RSSglobe.html
long term
http://junkscience.com/MSU_Temps/Moberg2005.html |
|
|
| Back to top |
|
|
|
| Bill Ward... |
| Posted: Wed Oct 07, 2009 8:35 pm |
|
|
|
Guest
|
On Wed, 07 Oct 2009 17:55:50 -0700, Rob wrote:
[...]
[BW]
[quote:ec777156a8]I think you need to read the Lindzen paper again, focusing a little
more on the details. Remember, he cites Wong, so he's aware of the
issues raised.- Hide quoted text -
[/quote:ec777156a8]
[Rob]
[quote:ec777156a8]I read it. In great detail.
The slope that he calculates (from ERBE and SST data) depends on a very
small set of measurement points, which do not seem to be correctly
determined (see above).
[/quote:ec777156a8]
In table 1, he shows a dozen statistical analyses of scenarios ranging
from 3 to 176 intervals. All show negative feedback. He selected the one
(13 points) with the lowest error as the best estimate, which seems OK to
me.
[quote:ec777156a8]Consequently, I do not trust this paper one bit.
[/quote:ec777156a8]
Do you understand his logic? Roughly, he's using edges in the natural
driving function to see the response of the system in the frequency range
of interest. By focusing on larger events, he can improve the in-band
S/N ratio.
Put on your system analysis hat and read it more carefully. The fact
that all of the actual data analyses show negative feedback, while all
the models show positive feedback seems rather convincing to me.
[quote:ec777156a8]Rob
[/quote:ec777156a8]
Sorry to snip so much, but this post has the Google property that
prevents pan from editing it the normal way. |
|
|
| Back to top |
|
|
|
| Rob... |
| Posted: Wed Oct 07, 2009 9:38 pm |
|
|
|
Guest
|
On Oct 7, 7:35pm, Bill Ward <bw... at (no spam) ix.REMOVETHISnetcom.com> wrote:
[quote:a0399897a1]On Wed, 07 Oct 2009 17:55:50 -0700, Rob wrote:
[...]
[BW]
I think you need to read the Lindzen paper again, focusing a little
more on the details. Remember, he cites Wong, so he's aware of the
issues raised.- Hide quoted text -
[Rob]
I read it. In great detail.
The slope that he calculates (from ERBE and SST data) depends on a very
small set of measurement points, which do not seem to be correctly
determined (see above).
In table 1, he shows a dozen statistical analyses of scenarios ranging
from 3 to 176 intervals. All show negative feedback. He selected the one
(13 points) with the lowest error as the best estimate, which seems OK to
me.
[/quote:a0399897a1]
Once again, that would be fine if he is using the right data sets.
[quote:a0399897a1]Consequently, I do not trust this paper one bit.
Do you understand his logic? Roughly, he's using edges in the natural
driving function to see the response of the system in the frequency range
of interest.
[/quote:a0399897a1]
That's one way of explaining it. In my words :
He wants to show the correlation between radiation balance changes and
SST changes, over a timespan that is long for direct effects (days)
and short for the feedback effects that may 'obscure' cause and
effect (years). He experiments with a month to 7 months as workable
windows.
By itself that is a reasonable goal ; there can be many different
feedback factors of various different timescales, and it is good to
understand and measure the individual effects of each by data
correlation techniques.
[quote:a0399897a1]By focusing on larger events, he can improve the in-band
S/N ratio.
[/quote:a0399897a1]
Makes sense, since the 'smaller' events (within 0.1 K or within 0.4 W/
m^2 deltas) especially on small time scales are very, very noisy, and
virtually uncorrelated (as his table 1, last 3 lines indicates).
But here comes the first problem : The 'correlated' data sets are very
sparse : All well-correlated data sets are less that 10 data points,
most have 5 or less. It's easy to get reasonable correlation with a
straight line with 3 or 4 or 5 random points, so these tests are
pretty useless.
Only one reasonable correlating data set (0.1k unfiltered) is 13 data
points large, and this is the one that be plots in fig 2, top-left
corner.
However, that plot shows the second problem : The 13 dots are fairly
uncorrelated, and the 'fitting' with the slop is mostly determined by
3 or 4 crucial data points in the corners. These are most likely the 3
or 4 data points that he found in the other table 1 lines as well,
because the relate to the 'larger' events.
But that hits the biggest problem : These crucial 3 or 4 data points
seem to be incorrectly placed (as I now explained twice).
Please tell me where he got the first crucial point (+0.6 C / +5 W/
m^2) from ?
[quote:a0399897a1]
Put on your system analysis hat and read it more carefully. The fact
that all of the actual data analyses show negative feedback, while all
the models show positive feedback seems rather convincing to me.
[/quote:a0399897a1]
If we cannot even explain the first point on the first plot of what he
claims is the experiment that best shows his findings, then how are we
supposed to believe any of the drastic conclusions that he draws
(discrediting established models, and adjusting Earth's climate
sensitivity) in the remainder of his paper ?
[quote:a0399897a1]
Rob
Sorry to snip so much, but this post has the Google property that
prevents pan from editing it the normal way.
[/quote:a0399897a1]
Sorry, my regular NNTP server let me down. Forced to work vioa Google
for now.
Rob |
|
|
| Back to top |
|
|
|
| Bill Ward... |
| Posted: Thu Oct 08, 2009 7:53 pm |
|
|
|
Guest
|
On Thu, 08 Oct 2009 00:38:13 -0700, Rob wrote:
[quote:9036f4016b]On Oct 7, 7:35 pm, Bill Ward <bw... at (no spam) ix.REMOVETHISnetcom.com> wrote:
On Wed, 07 Oct 2009 17:55:50 -0700, Rob wrote:
[...]
[BW]
I think you need to read the Lindzen paper again, focusing a little
more on the details. Remember, he cites Wong, so he's aware of the
issues raised.- Hide quoted text -
[Rob]
I read it. In great detail.
The slope that he calculates (from ERBE and SST data) depends on a
very small set of measurement points, which do not seem to be
correctly determined (see above).
In table 1, he shows a dozen statistical analyses of scenarios ranging
from 3 to 176 intervals. All show negative feedback. He selected the
one (13 points) with the lowest error as the best estimate, which seems
OK to me.
Once again, that would be fine if he is using the right data sets.
Consequently, I do not trust this paper one bit.
Do you understand his logic? Roughly, he's using edges in the natural
driving function to see the response of the system in the frequency
range of interest.
That's one way of explaining it. In my words :
He wants to show the correlation between radiation balance changes and
SST changes, over a timespan that is long for direct effects (days) and
short for the feedback effects that may 'obscure' cause and effect
(years). He experiments with a month to 7 months as workable windows.
By itself that is a reasonable goal ; there can be many different
feedback factors of various different timescales, and it is good to
understand and measure the individual effects of each by data
correlation techniques.
By focusing on larger events, he can improve the in-band S/N ratio.
Makes sense, since the 'smaller' events (within 0.1 K or within 0.4 W/
m^2 deltas) especially on small time scales are very, very noisy, and
virtually uncorrelated (as his table 1, last 3 lines indicates).
But here comes the first problem : The 'correlated' data sets are very
sparse : All well-correlated data sets are less that 10 data points,
most have 5 or less. It's easy to get reasonable correlation with a
straight line with 3 or 4 or 5 random points, so these tests are pretty
useless.
Only one reasonable correlating data set (0.1k unfiltered) is 13 data
points large, and this is the one that be plots in fig 2, top-left
corner.
[/quote:9036f4016b]
Note the slope is +4.23 with a SE of 1.54. All the models are negative
slopes.
[quote:9036f4016b]However, that plot shows the second problem : The 13 dots are fairly
uncorrelated, and the 'fitting' with the slop is mostly determined by 3
or 4 crucial data points in the corners. These are most likely the 3 or
4 data points that he found in the other table 1 lines as well, because
the relate to the 'larger' events.
But that hits the biggest problem : These crucial 3 or 4 data points
seem to be incorrectly placed (as I now explained twice).
Please tell me where he got the first crucial point (+0.6 C / +5 W/ m^2)
from ?
[/quote:9036f4016b]
Probably from near the end of the sequence, from around 1998 to 2000.
You could always ask him.
I still don't think you understand the analysis. Each point on the plot
is derived from the ratio between the slope of the ERBE and the slope of
the SST data in specific time intervals where the SST varied by some
threshold (0.1C in this case). The number of those intervals at each
threshold and the resulting statistics are shown in Table 1.
Because of the noise, a visual confirmation is not likely to be
convincing. If you want to replicate the analysis, I think you'll need
to get the cited datasets, perform the specified filtering, and crunch
the numbers to see if you agree with Table 1.
[quote:9036f4016b]Put on your system analysis hat and read it more carefully. The fact
that all of the actual data analyses show negative feedback, while all
the models show positive feedback seems rather convincing to me.
If we cannot even explain the first point on the first plot of what he
claims is the experiment that best shows his findings, then how are we
supposed to believe any of the drastic conclusions that he draws
(discrediting established models, and adjusting Earth's climate
sensitivity) in the remainder of his paper ?
[/quote:9036f4016b]
It's a published paper. If you find an error that changes the
conclusions, you'll be famous, and several reviewers will have egg on
their faces.
[quote:9036f4016b]
Rob
Sorry to snip so much, but this post has the Google property that
prevents pan from editing it the normal way.
Sorry, my regular NNTP server let me down. Forced to work vioa Google
for now.
Rob[/quote:9036f4016b] |
|
|
| Back to top |
|
|
|
| Rob Dekker... |
| Posted: Sat Oct 10, 2009 12:48 am |
|
|
|
Guest
|
"Bill Ward" <bward at (no spam) ix.REMOVETHISnetcom.com> wrote in message
news:Eq-dncuR8b8OClPXnZ2dnUVZ_tSdnZ2d at (no spam) giganews.com...
[quote:696cef5da3]On Thu, 08 Oct 2009 00:38:13 -0700, Rob wrote:
On Oct 7, 7:35 pm, Bill Ward <bw... at (no spam) ix.REMOVETHISnetcom.com> wrote:
On Wed, 07 Oct 2009 17:55:50 -0700, Rob wrote:
[...]
[BW]
I think you need to read the Lindzen paper again, focusing a little
more on the details. Remember, he cites Wong, so he's aware of the
issues raised.- Hide quoted text -
[Rob]
I read it. In great detail.
The slope that he calculates (from ERBE and SST data) depends on a
very small set of measurement points, which do not seem to be
correctly determined (see above).
In table 1, he shows a dozen statistical analyses of scenarios ranging
from 3 to 176 intervals. All show negative feedback. He selected the
one (13 points) with the lowest error as the best estimate, which seems
OK to me.
Once again, that would be fine if he is using the right data sets.
Consequently, I do not trust this paper one bit.
Do you understand his logic? Roughly, he's using edges in the natural
driving function to see the response of the system in the frequency
range of interest.
That's one way of explaining it. In my words :
He wants to show the correlation between radiation balance changes and
SST changes, over a timespan that is long for direct effects (days) and
short for the feedback effects that may 'obscure' cause and effect
(years). He experiments with a month to 7 months as workable windows.
By itself that is a reasonable goal ; there can be many different
feedback factors of various different timescales, and it is good to
understand and measure the individual effects of each by data
correlation techniques.
By focusing on larger events, he can improve the in-band S/N ratio.
Makes sense, since the 'smaller' events (within 0.1 K or within 0.4 W/
m^2 deltas) especially on small time scales are very, very noisy, and
virtually uncorrelated (as his table 1, last 3 lines indicates).
But here comes the first problem : The 'correlated' data sets are very
sparse : All well-correlated data sets are less that 10 data points,
most have 5 or less. It's easy to get reasonable correlation with a
straight line with 3 or 4 or 5 random points, so these tests are pretty
useless.
Only one reasonable correlating data set (0.1k unfiltered) is 13 data
points large, and this is the one that be plots in fig 2, top-left
corner.
Note the slope is +4.23 with a SE of 1.54.
[/quote:696cef5da3]
\I double checked some of the graphs (SW and OLW), and found that at least
the ones that he printed in the paper are from the ERBE edition 2 data set.
Not from the edition 3 data set that he claims they are based on.
The edition 3 data set is the altitude-adjusted set, and should be used for
climate analysis.
Not sure if that makes much of a difference for the short-term analysis that
he is doing, but it is another indication that Lindzen is a bit sloppy with
where he got his data from... And I'm sure that it will at least affect his
analysis a little bit. After all, even Lindzen himself admits that the slope
is very sensitive for small delta adjustments. And the climate sensitivity
factor is very sensitive to the slope.
[quote:696cef5da3]All the models are negative
slopes.
[/quote:696cef5da3]
I'm sure that you have read some of the blog reviews of the Lindzen and Choi
paper. If not, please Google that and read them. The models that he used are
not intended for the type of feedback analysis that he wants to do.
Also, I don't really care much about the models that he brought into this
paper.
The important finding that he reports is about a (read 'one') short-term
feedback mechanism. That's more than enough to deal with in one paper AFAIK.
Climate sensitivity analysis and feedback mechanisms are difficult enough as
is.
[quote:696cef5da3]However, that plot shows the second problem : The 13 dots are fairly
uncorrelated, and the 'fitting' with the slop is mostly determined by 3
or 4 crucial data points in the corners. These are most likely the 3 or
4 data points that he found in the other table 1 lines as well, because
the relate to the 'larger' events.
But that hits the biggest problem : These crucial 3 or 4 data points
seem to be incorrectly placed (as I now explained twice).
Please tell me where he got the first crucial point (+0.6 C / +5 W/ m^2)
from ?
Probably from near the end of the sequence, from around 1998 to 2000.
You could always ask him.
[/quote:696cef5da3]
I do have a day job and a family, so I don't think I can get much done in
short time.
However, I will likely do so if my first analysis does not confirm that data
point.
[quote:696cef5da3]
I still don't think you understand the analysis. Each point on the plot
is derived from the ratio between the slope of the ERBE and the slope of
the SST data in specific time intervals where the SST varied by some
threshold (0.1C in this case). The number of those intervals at each
threshold and the resulting statistics are shown in Table 1.
[/quote:696cef5da3]
I do understand, Bill.
The only thing I am not sure about is how he determined the time interval
lengths.
For example, if the SST is increasing straight-line 0.6 C over one year,
then there are many data points that you can get from that :
One point that increases 0.6 C, or two points that increase 0.3 C or three
points that increase 0.2 or 6 points that increase 0.1 C. All these points
are above the set lower 0.1 C limit. Which ones did he choose ? Or did he
choose overlapping data points ?
Or does he choose the data points (the time segments) by Monte Carlo
approach ?
He is not very clear about that.
I will ask him that once I'm ready to shoot him an email.
[quote:696cef5da3]
Because of the noise, a visual confirmation is not likely to be
convincing. If you want to replicate the analysis, I think you'll need
to get the cited datasets, perform the specified filtering, and crunch
the numbers to see if you agree with Table 1.
[/quote:696cef5da3]
I got the ERBE edition 3 numbers (monthly for the tropics).
I would like to get the SST numbers, but I can't find that. Only individual
ocean section numbers.
I'll look a bit deeper.
Any way, I'd like to do my own data analysis first before I bother him with
questions about his.
[quote:696cef5da3]
Put on your system analysis hat and read it more carefully. The fact
that all of the actual data analyses show negative feedback, while all
the models show positive feedback seems rather convincing to me.
If we cannot even explain the first point on the first plot of what he
claims is the experiment that best shows his findings, then how are we
supposed to believe any of the drastic conclusions that he draws
(discrediting established models, and adjusting Earth's climate
sensitivity) in the remainder of his paper ?
It's a published paper. If you find an error that changes the
conclusions, you'll be famous, and several reviewers will have egg on
their faces.
[/quote:696cef5da3]
I'm not looking for that.
I'm just looking for the truth.
You know that I believe that the case for GW from CO2 has not been made yet.
I believe the theory (GW by CO2) is correct, but we have not seen enough
evidence to support (or discard) that theory yet. Meanwhile, I give the
researchers the benefit of the doubt.
That opinion is based on the lack of evidence that I've seen, and the
discussions we had about this subject.
Same thing with Lindzen. I give him the benefit of the doubt, that there may
be a short-term negative feedback mechanism in place. That does not discard
the vast amount of research that shows equilibrium temps for doubling of CO2
in the range of 4 C, but Lindzen's paper also does not throw that end-result
away at all.
I'll take this one paper at a time. For now, I suspect that Lindzen is
twisting data to serve his incentive of showing negative feedbacks. The
paper has inconsistencies which as suspicious. Will you help me get the data
straightened out ? I'm not very good with statistics. Can you help ?
[quote:696cef5da3]
Rob
Sorry to snip so much, but this post has the Google property that
prevents pan from editing it the normal way.
Sorry, my regular NNTP server let me down. Forced to work vioa Google
for now.
Rob
[/quote:696cef5da3] |
|
|
| Back to top |
|
|
|
| Bill Ward... |
| Posted: Sat Oct 10, 2009 1:01 pm |
|
|
|
Guest
|
On Fri, 09 Oct 2009 23:48:42 -0700, Rob Dekker wrote:
[quote:f9c461171f]"Bill Ward" <bward at (no spam) ix.REMOVETHISnetcom.com> wrote in message
news:Eq-dncuR8b8OClPXnZ2dnUVZ_tSdnZ2d at (no spam) giganews.com...
On Thu, 08 Oct 2009 00:38:13 -0700, Rob wrote:
On Oct 7, 7:35 pm, Bill Ward <bw... at (no spam) ix.REMOVETHISnetcom.com> wrote:
On Wed, 07 Oct 2009 17:55:50 -0700, Rob wrote:
[...]
[BW]
I think you need to read the Lindzen paper again, focusing a
little more on the details. Remember, he cites Wong, so he's aware
of the issues raised.- Hide quoted text -
[Rob]
I read it. In great detail.
The slope that he calculates (from ERBE and SST data) depends on a
very small set of measurement points, which do not seem to be
correctly determined (see above).
In table 1, he shows a dozen statistical analyses of scenarios
ranging from 3 to 176 intervals. All show negative feedback. He
selected the one (13 points) with the lowest error as the best
estimate, which seems OK to me.
Once again, that would be fine if he is using the right data sets.
Consequently, I do not trust this paper one bit.
Do you understand his logic? Roughly, he's using edges in the natural
driving function to see the response of the system in the frequency
range of interest.
That's one way of explaining it. In my words :
He wants to show the correlation between radiation balance changes and
SST changes, over a timespan that is long for direct effects (days)
and short for the feedback effects that may 'obscure' cause and
effect (years). He experiments with a month to 7 months as workable
windows.
By itself that is a reasonable goal ; there can be many different
feedback factors of various different timescales, and it is good to
understand and measure the individual effects of each by data
correlation techniques.
By focusing on larger events, he can improve the in-band S/N ratio.
Makes sense, since the 'smaller' events (within 0.1 K or within 0.4 W/
m^2 deltas) especially on small time scales are very, very noisy, and
virtually uncorrelated (as his table 1, last 3 lines indicates).
But here comes the first problem : The 'correlated' data sets are very
sparse : All well-correlated data sets are less that 10 data points,
most have 5 or less. It's easy to get reasonable correlation with a
straight line with 3 or 4 or 5 random points, so these tests are
pretty useless.
Only one reasonable correlating data set (0.1k unfiltered) is 13 data
points large, and this is the one that be plots in fig 2, top-left
corner.
Note the slope is +4.23 with a SE of 1.54.
\I double checked some of the graphs (SW and OLW), and found that at
least the ones that he printed in the paper are from the ERBE edition 2
data set. Not from the edition 3 data set that he claims they are based
on. The edition 3 data set is the altitude-adjusted set, and should be
used for climate analysis.
[/quote:f9c461171f]
Perhaps he used the wrong graphs (it happens), but I doubt he'd base the
analysis on the uncorrected data. That's kind of the point of the paper,
to answer one of the criticisms of his earlier work, to wit:
"... this result was internally inconsistent since the persistence of the
imbalance over a decade implied a positive feedback. A subsequent
correction to the satellite data eliminated much of the decadal variation
in the radiative balance." (from page 1)
So it seems he was quite aware of the shortfalls in the original,
uncorrected dataset.
[quote:f9c461171f]Not sure if that makes much of a difference for the short-term analysis
that he is doing, but it is another indication that Lindzen is a bit
sloppy with where he got his data from... And I'm sure that it will at
least affect his analysis a little bit. After all, even Lindzen himself
admits that the slope is very sensitive for small delta adjustments. And
the climate sensitivity factor is very sensitive to the slope.
All the models are negative
slopes.
I'm sure that you have read some of the blog reviews of the Lindzen and
Choi paper. If not, please Google that and read them.
[/quote:f9c461171f]
If you'll recommend a specific link, I'll review it. Otherwise, I'd
rather rely on the paper itself rather than a random blogger's comments.
[quote:f9c461171f]The models that he
used are not intended for the type of feedback analysis that he wants to
do. Also, I don't really care much about the models that he brought into
this paper.
[/quote:f9c461171f]
Neither do I. Do you have a favorite you can recommend?
[quote:f9c461171f]The important finding that he reports is about a (read 'one') short-term
feedback mechanism. That's more than enough to deal with in one paper
AFAIK. Climate sensitivity analysis and feedback mechanisms are
difficult enough as is.
[/quote:f9c461171f]
Remember short-term negative feedbacks have lasting effects.
[quote:f9c461171f]However, that plot shows the second problem : The 13 dots are fairly
uncorrelated, and the 'fitting' with the slop is mostly determined by
3 or 4 crucial data points in the corners. These are most likely the 3
or 4 data points that he found in the other table 1 lines as well,
because the relate to the 'larger' events.
But that hits the biggest problem : These crucial 3 or 4 data points
seem to be incorrectly placed (as I now explained twice).
Please tell me where he got the first crucial point (+0.6 C / +5 W/
m^2) from ?
Probably from near the end of the sequence, from around 1998 to 2000.
You could always ask him.
I do have a day job and a family, so I don't think I can get much done
in short time.
However, I will likely do so if my first analysis does not confirm that
data point.
I still don't think you understand the analysis. Each point on the
plot is derived from the ratio between the slope of the ERBE and the
slope of the SST data in specific time intervals where the SST varied
by some threshold (0.1C in this case). The number of those intervals
at each threshold and the resulting statistics are shown in Table 1.
I do understand, Bill.
The only thing I am not sure about is how he determined the time
interval lengths.
For example, if the SST is increasing straight-line 0.6 C over one year,
then there are many data points that you can get from that : One point
that increases 0.6 C, or two points that increase 0.3 C or three points
that increase 0.2 or 6 points that increase 0.1 C. All these points are
above the set lower 0.1 C limit. Which ones did he choose ? Or did he
choose overlapping data points ?
Or does he choose the data points (the time segments) by Monte Carlo
approach ? He is not very clear about that.
[/quote:f9c461171f]
OK, now I think I understand the problem. Look at the SST graph at the
top of page 2. Count the red and blue segments. I get nine, 5 red
ascending, and 5 blue descending. Now turn to Table 1 and note the
second entry, with nine data points. I assume that the colored segments
represent those nine points, selected by the threshold used. I'm not
sure how he took the derivatives, but it seems logical to simply take the
difference between the end points, since the information between is out
of band anyway.
The point is that the sample interval is not constant. Think of each
warming or cooling interval as a separate event yielding a slope, and the
overall result the average of the slopes. He chose the set containing 13
events, probably because it gave the lowest error.
Does that make sense to you?
[quote:f9c461171f]I will ask him that once I'm ready to shoot him an email.
Because of the noise, a visual confirmation is not likely to be
convincing. If you want to replicate the analysis, I think you'll need
to get the cited datasets, perform the specified filtering, and crunch
the numbers to see if you agree with Table 1.
I got the ERBE edition 3 numbers (monthly for the tropics). I would like
to get the SST numbers, but I can't find that. Only individual ocean
section numbers.
I'll look a bit deeper.
Any way, I'd like to do my own data analysis first before I bother him
with questions about his.
Put on your system analysis hat and read it more carefully. The fact
that all of the actual data analyses show negative feedback, while
all the models show positive feedback seems rather convincing to me.
If we cannot even explain the first point on the first plot of what he
claims is the experiment that best shows his findings, then how are we
supposed to believe any of the drastic conclusions that he draws
(discrediting established models, and adjusting Earth's climate
sensitivity) in the remainder of his paper ?
It's a published paper. If you find an error that changes the
conclusions, you'll be famous, and several reviewers will have egg on
their faces.
I'm not looking for that.
I'm just looking for the truth.
You know that I believe that the case for GW from CO2 has not been made
yet. I believe the theory (GW by CO2) is correct, but we have not seen
enough evidence to support (or discard) that theory yet. Meanwhile, I
give the researchers the benefit of the doubt. That opinion is based on
the lack of evidence that I've seen, and the discussions we had about
this subject. Same thing with Lindzen. I give him the benefit of the
doubt, that there may be a short-term negative feedback mechanism in
place. That does not discard the vast amount of research that shows
equilibrium temps for doubling of CO2 in the range of 4 C, but Lindzen's
paper also does not throw that end-result away at all.
I'll take this one paper at a time. For now, I suspect that Lindzen is
twisting data to serve his incentive of showing negative feedbacks. The
paper has inconsistencies which as suspicious. Will you help me get the
data straightened out ? I'm not very good with statistics. Can you help
?
[/quote:f9c461171f]
Thanks for the compliment, but I don't consider myself anywhere near
statistically competent enough to advise anybody. Google would probably
be more help. |
|
|
| Back to top |
|
|
|
| Rob... |
| Posted: Tue Oct 13, 2009 9:46 pm |
|
|
|
Guest
|
On Oct 10, 12:01pm, Bill Ward <bw... at (no spam) ix.REMOVETHISnetcom.com> wrote:
.......
[quote:a6534de74f]\I double checked some of the graphs (SW and OLW), and found that at
least the ones that he printed in the paper are from the ERBE edition 2
data set. Not from the edition 3 data set that he claims they are based
on. The edition 3 data set is the altitude-adjusted set, and should be
used for climate analysis.
Perhaps he used the wrong graphs (it happens), but I doubt he'd base the
analysis on the uncorrected data. That's kind of the point of the paper,
to answer one of the criticisms of his earlier work, to wit:
"... this result was internally inconsistent since the persistence of the
imbalance over a decade implied a positive feedback. A subsequent
correction to the satellite data eliminated much of the decadal variation
in the radiative balance." (from page 1)
So it seems he was quite aware of the shortfalls in the original,
uncorrected dataset.
[/quote:a6534de74f]
Yes, he is aware of the Edition 3 set, but I'm not sure he used it.
Two reasons for being sceptical about this :
(1) He used the Edition 2 set in a paper that he wrote earlier this
year :
http://wattsupwiththat.com/2009/03/30/lindzen-on-negative-climate-feedback
Which was promply followed by a blog, pointing out that he used the
old Edition 2 data :
http://chriscolose.wordpress.com/2009/03/31/lindzen-on-climate-feedback
Now if that mistake was pointed out so quickly, couldn't he have
double-checked that his graphs were correct for the real paper to
Geophysical Reasearch Letters ?
(2) The data points from his figure 2 (especially the +0.6 C/+5 W/m^2
point) may be there in the Edition 2 data set, but I sure can't find
them in the Edition 3 data set. (see my analysis below).
.......
[quote:a6534de74f]The models that he
used are not intended for the type of feedback analysis that he wants to
do. Also, I don't really care much about the models that he brought into
this paper.
Neither do I. Do you have a favorite you can recommend?
[/quote:a6534de74f]
Here are two on his choice of models :
http://julesandjames.blogspot.com/2009/08/quick-comment-on-lindzen-and-choi..html
http://groups.google.com/group/globalchange/browse_thread/thread/796e8951bc6a5b92?pli=1
[quote:a6534de74f]
The important finding that he reports is about a (read 'one') short-term
feedback mechanism. That's more than enough to deal with in one paper
AFAIK. Climate sensitivity analysis and feedback mechanisms are
difficult enough as is.
Remember short-term negative feedbacks have lasting effects.
[/quote:a6534de74f]
That depends on how strong other feedback mechanisms are.
Also, please understand one thing : If there is no feedback at all on
the short (months) term that Lindzen analysed, then we SHOULD see a
slope of 4 (W/m^2 per C), simply because of the Plank function : a 1 C
warmer surface radiates 4 W/m^2 extra.
Incidentally, he found approximately a slope of 4 (W/m^2/C) so that
indicates no short-term feedback.
......
[quote:a6534de74f]I got the ERBE edition 3 numbers (monthly for the tropics). I would like
to get the SST numbers, but I can't find that. Only individual ocean
section numbers.
I'll look a bit deeper.
Any way, I'd like to do my own data analysis first before I bother him
with questions about his.
[/quote:a6534de74f]
OK. I did my analysis. Here we go :
I picked up the ERBE Edition 3 (rev 1) data set for the tropics, for
the monthly, as Lindzen used. Here it is, in understandable (and
parsable) form :
http://earth-www.larc.nasa.gov/erbeweb/Edition3_Rev1/Edition3_Rev1_wfov_sf_monthly_tropics
I can't find the SST data, so as a start, I entered the 9 segments
from Lindzen's graph manually. With delta-SST for each time segment
(red and blue).
Then I wrote a small program that does correlation analysis using the
SW and OLW columns from ERBE, along the 9 time segments indicated.
I calculate the Pearson product moment (R) factor over the 9 data
points, using this algorithm :
http://en.wikipedia.org/wiki/Pearson_product-moment_correlation_coefficient#Computing_correlation_accurately_in_a_single_pass
My first experiments gave correlation factors close to 0. Then I found
that there is an aweful lot of noise in the data. The SW and OLW vary
widely and fast over the months, and the choice of time segments is
crucial to get ANY form of correlation between delta SST and delta
flux. Misplaced time segments and correlation collapses.
After I very carefully placed the 9 time segments according to
Lindzen's SST graph, I got my first 'unfiltered' result (SW+OLW
monthly, no time-averaging) that were showing some correlation :
delta-SST delta-flux
1. 0.300000 6.020000 slope: 20.066667
2. -0.500000 -3.550000 slope: 7.100000
3. 0.300000 -1.760000 slope: -5.866667
4. -0.250000 -2.180000 slope: 8.720000
5. 0.200000 -4.670000 slope: -23.350000
6. 0.250000 -0.880000 slope: -3.520000
7. -0.200000 2.680000 slope: -13.400000
8. 0.600000 3.030000 slope: 5.050000
9. -0.600000 -5.220000 slope: 8.700000
Results :
N:9 average weighted slope : 2.250768 correlation (R): 0.547974
First note that weighted slope is lower (2.25 versus 4.5) than Lindzen
found.
If may be that he calculated the best-fitting slope differently, but
with only 9 data points, and segment slopes going all over the place,
who knows what the 'actual' slope is.
Second, note that the mistery of the +0.6 / +5W/m^2 data point may be
solved !
The only +0.6 C spot is indeed the onramp for the 1998 El Nino,
But the delta-flux over that one-year period is 3 W/m^2.
That's fairly close to Lindzen's 5 W/m^2, so I have to assume we are
talking about the same segment. The difference can easily come from
the Edition 2/3 data set difference, or even choice of revision number
of the data set.
No matter how I positioned that time segment, I could never get more
than 3 W/m^2 increase. Another indication that Lindzen may be using an
older data (Edition 2?)set, where the OLR and SW were larger than in
the correct set.
Third, and most importantly : the correlation is worse than what
Lindzen found.
No matter how hard I tried (move the time segments start and end-dates
back and forth a month or so) I could not get an R larger than 0.6.
Now, Lindzen used 7-month SW averaging to eliminate noise (and
temporal aliasing effects) in the data. I did the same here :
delta-SST delta-flux
1. 0.300000 2.205714 slope: 7.352380
2. -0.500000 -2.531428 slope: 5.062856
3. 0.300000 -1.455715 slope: -4.852383
4. -0.250000 -2.987143 slope: 11.948572
5. 0.200000 -1.714286 slope: -8.571430
6. 0.250000 2.297143 slope: 9.188572
7. -0.200000 2.390000 slope: -11.950000
8. 0.600000 0.405714 slope: 0.676190
9. -0.600000 -2.830000 slope: 4.716667
N:9 average weighted slope : 2.582305 correlation (R): 0.524967
See that indeed neither the slope nor the correlation changes much (as
Lindzen also noticed). However, notice that even with 7-month running
average of SW, the slopes of the data points still go all over the
place.
Overall, correlation is low, and the slope (if you can talk about that
with such low correlation) is lower than Lindzen found. There is way
to much noise for me to draw any conclusions other than that there is
no signal to speak of.
I found something very interesting by doing separate OLW <-> SST
analysis, comparing how long-wave radiation changes with changes in
sea-surface temperature :
0.300000 -0.790000 slope: -2.633333
-0.500000 -2.620000 slope: 5.240000
0.300000 0.080000 slope: 0.266667
-0.250000 -2.740000 slope: 10.960000
0.200000 2.700000 slope: 13.500000
0.250000 0.500000 slope: 2.000000
-0.200000 -0.530000 slope: 2.650000
0.600000 1.800000 slope: 3.000000
-0.600000 -3.250000 slope: 5.416667
N:9 average slope : 4.488889 correlation (R): 0.830673
Correlation of 0.83 is pretty strong, indicating that LW radiations
indeed changes with SST. At a slope of about 4 (W/m^2/C). That's
pretty close to the Planck curve, and thus would be expected in
absense of any short-term feedback.
Separate SW <-> SST shows this fuzzy picture :
0.300000 6.810000 slope: 22.700000
-0.500000 -0.930000 slope: 1.860000
0.300000 -1.840000 slope: -6.133333
-0.250000 0.560000 slope: -2.240000
0.200000 -7.370000 slope: -36.850000
0.250000 -1.380000 slope: -5.520000
-0.200000 3.210000 slope: -16.050000
0.600000 1.230000 slope: 2.050000
-0.600000 -1.970000 slope: 3.283333
N:9 average slope : -4.100000 correlation (R): 0.100434
Correlation of 0.1 indicates that there is no correlation between
changes in SW and changes in SST.
My conclusions from my analysis : OLW increases fairly accurately with
changes in SST, at about 4 W/m^2/C, which would be expected from a
system without any short-term feedback. Other than that, everything is
very noisy, and no other conclusions should be drawn.
If Lindzen did indeed base his data on the Edition 3 (rev 1) ERBE data
set, then he should at least show the correlation numbers of the
separate SW <-> SST and OLW <-> SST numbers. I suspect that we will
then see the same low SW correlation and strong OLW correlation that I
found. Concequently, this means that feedback mechanisms seem not
present for the short terms (months) time segments that were present
in the analysis.
That's my 2 cts.
P.S. Sorry for posting via Google groups. I know this makes replies
difficult, but it's the only way to get my post out right now. |
|
|
| Back to top |
|
|
|
| Bill Ward... |
| Posted: Wed Oct 14, 2009 10:55 am |
|
|
|
Guest
|
On Wed, 14 Oct 2009 00:46:10 -0700, Rob wrote:
[quote:7c2d67d7bf]On Oct 10, 12:01 pm, Bill Ward <bw... at (no spam) ix.REMOVETHISnetcom.com> wrote:
......
[/quote:7c2d67d7bf]
[Rob]
[quote:7c2d67d7bf]\I double checked some of the graphs (SW and OLW), and found that at
least the ones that he printed in the paper are from the ERBE edition
2 data set. Not from the edition 3 data set that he claims they are
based on. The edition 3 data set is the altitude-adjusted set, and
should be used for climate analysis.
[/quote:7c2d67d7bf]
[Bill]
[quote:7c2d67d7bf]Perhaps he used the wrong graphs (it happens), but I doubt he'd base
the analysis on the uncorrected data. That's kind of the point of the
paper, to answer one of the criticisms of his earlier work, to wit:
"... this result was internally inconsistent since the persistence of
the imbalance over a decade implied a positive feedback. A subsequent
correction to the satellite data eliminated much of the decadal
variation in the radiative balance." (from page 1)
So it seems he was quite aware of the shortfalls in the original,
uncorrected dataset.
Yes, he is aware of the Edition 3 set, but I'm not sure he used it. Two
reasons for being sceptical about this :
(1) He used the Edition 2 set in a paper that he wrote earlier this year
:
http://wattsupwiththat.com/2009/03/30/lindzen-on-negative-climate-
feedback
Which was promply followed by a blog, pointing out that he used the old
Edition 2 data :
http://chriscolose.wordpress.com/2009/03/31/lindzen-on-climate-feedback
[/quote:7c2d67d7bf]
Followed by this comment to WUWT at the same link:
<begin excerpt>
UPDATE3: I received this email today (4/10) from Dr. Lindzen. My sincere
thanks for his response.
Dear Anthony,
The paper was sent out for comments, and the comments (even those from
“realclimate”) are appreciated. In fact, the reduction of the difference
in OLR between the 80’s and 90’s due to orbital decay seems to me to be
largely correct. However, the reduction in Wong, Wielicki et al (2006)
of the difference in the spikes of OLR between observations and models
cannot be attributed to orbital decay, and seem to me to be
questionable. Nevertheless, the differences that remain still imply
negative feedbacks. We are proceeding to redo the analysis of satellite
data in order to better understand what went into these analyses. The
matter of net differences between the 80’s and 90’s is an interesting
question. Given enough time, the radiative balance is reestablished and
the anomalies can be wiped out. The time it takes for this to happen
depends on climate sensitivity with adjustments occurring more rapidly
when sensitivity is less. However, for the spikes, the time scales are
short enough to preclude adjustment except for very low sensitivity.
That said, it has become standard in climate science that data in
contradiction to alarmism is inevitably ‘corrected’ to bring it closer to
alarming models. None of us would argue that this data is perfect, and
the corrections are often plausible. What is implausible is that the
‘corrections’ should always bring the data closer to models.
Best wishes,
Dick
<end excerpt>
Note that he explicitly comments that he plans to redo the analysis, and
months later releases the revised article we're talking about. It seems
unlikely he would try to put something so obvious over on anyone,
considering the long standing intense efforts to discredit him.
[quote:7c2d67d7bf]
Now if that mistake was pointed out so quickly, couldn't he have
double-checked that his graphs were correct for the real paper to
Geophysical Reasearch Letters ?
(2) The data points from his figure 2 (especially the +0.6 C/+5 W/m^2
point) may be there in the Edition 2 data set, but I sure can't find
them in the Edition 3 data set. (see my analysis below).
......
The models that he
used are not intended for the type of feedback analysis that he wants
to do. Also, I don't really care much about the models that he
brought into this paper.
Neither do I. Do you have a favorite you can recommend?
Here are two on his choice of models :
http://julesandjames.blogspot.com/2009/08/quick-comment-on-lindzen-and-
choi.html
http://groups.google.com/group/globalchange/browse_thread/
thread/796e8951bc6a5b92?pli=1[/quote:7c2d67d7bf]
So many models, so few Earths. Which model is right?
[quote:7c2d67d7bf]The important finding that he reports is about a (read 'one')
short-term feedback mechanism. That's more than enough to deal with
in one paper AFAIK. Climate sensitivity analysis and feedback
mechanisms are difficult enough as is.
Remember short-term negative feedbacks have lasting effects.
That depends on how strong other feedback mechanisms are. Also, please
understand one thing : If there is no feedback at all on the short
(months) term that Lindzen analysed, then we SHOULD see a slope of 4
(W/m^2 per C), simply because of the Plank function : a 1 C warmer
surface radiates 4 W/m^2 extra.
[/quote:7c2d67d7bf]
I think you meant to say the Stephan-Boltzmann relation:
<http://hyperphysics.phy-astr.gsu.edu/hbase/thermo/stefan.html#c2>
Planck is the spectrum guy.
[quote:7c2d67d7bf]Incidentally, he found approximately a slope of 4 (W/m^2/C) so that
indicates no short-term feedback.
[/quote:7c2d67d7bf]
That is a feedback, due to the nonlinearity.
[quote:7c2d67d7bf].....
I got the ERBE edition 3 numbers (monthly for the tropics). I would
like to get the SST numbers, but I can't find that. Only individual
ocean section numbers.
I'll look a bit deeper.
Any way, I'd like to do my own data analysis first before I bother
him with questions about his.
OK. I did my analysis. Here we go :
I picked up the ERBE Edition 3 (rev 1) data set for the tropics, for the
monthly, as Lindzen used. Here it is, in understandable (and parsable)
form :
http://earth-www.larc.nasa.gov/erbeweb/Edition3_Rev1/
Edition3_Rev1_wfov_sf_monthly_tropics
I can't find the SST data, so as a start, I entered the 9 segments from
Lindzen's graph manually. With delta-SST for each time segment (red and
blue).
[/quote:7c2d67d7bf]
Did you allow for transcription error in digitizing the graph?
[quote:7c2d67d7bf]Then I wrote a small program that does correlation analysis using the SW
and OLW columns from ERBE, along the 9 time segments indicated. I
calculate the Pearson product moment (R) factor over the 9 data points,
using this algorithm :
http://en.wikipedia.org/wiki/Pearson_product-
moment_correlation_coefficient#Computing_correlation_accurately_in_a_single_pass
My first experiments gave correlation factors close to 0. Then I found
that there is an aweful lot of noise in the data. The SW and OLW vary
widely and fast over the months, and the choice of time segments is
crucial to get ANY form of correlation between delta SST and delta flux.
Misplaced time segments and correlation collapses.
After I very carefully placed the 9 time segments according to Lindzen's
SST graph, I got my first 'unfiltered' result (SW+OLW monthly, no
time-averaging) that were showing some correlation :
delta-SST delta-flux
1. 0.300000 6.020000 slope: 20.066667 2. -0.500000 -3.550000 slope:
7.100000 3. 0.300000 -1.760000 slope: -5.866667 4. -0.250000 -2.180000
slope: 8.720000 5. 0.200000 -4.670000 slope: -23.350000 6. 0.250000
-0.880000 slope: -3.520000 7. -0.200000 2.680000 slope: -13.400000 8.
0.600000 3.030000 slope: 5.050000 9. -0.600000 -5.220000 slope:
8.700000 Results :
N:9 average weighted slope : 2.250768 correlation (R): 0.547974
First note that weighted slope is lower (2.25 versus 4.5) than Lindzen
found.
If may be that he calculated the best-fitting slope differently, but
with only 9 data points, and segment slopes going all over the place,
who knows what the 'actual' slope is.
Second, note that the mistery of the +0.6 / +5W/m^2 data point may be
solved !
The only +0.6 C spot is indeed the onramp for the 1998 El Nino, But the
delta-flux over that one-year period is 3 W/m^2. That's fairly close to
Lindzen's 5 W/m^2, so I have to assume we are talking about the same
segment. The difference can easily come from the Edition 2/3 data set
difference, or even choice of revision number of the data set.
No matter how I positioned that time segment, I could never get more
than 3 W/m^2 increase. Another indication that Lindzen may be using an
older data (Edition 2?)set, where the OLR and SW were larger than in the
correct set.
Third, and most importantly : the correlation is worse than what Lindzen
found.
No matter how hard I tried (move the time segments start and end-dates
back and forth a month or so) I could not get an R larger than 0.6.
Now, Lindzen used 7-month SW averaging to eliminate noise (and temporal
aliasing effects) in the data. I did the same here :
delta-SST delta-flux
1. 0.300000 2.205714 slope: 7.352380 2. -0.500000 -2.531428 slope:
5.062856 3. 0.300000 -1.455715 slope: -4.852383 4. -0.250000 -2.987143
slope: 11.948572 5. 0.200000 -1.714286 slope: -8.571430 6. 0.250000
2.297143 slope: 9.188572 7. -0.200000 2.390000 slope: -11.950000 8.
0.600000 0.405714 slope: 0.676190 9. -0.600000 -2.830000 slope:
4.716667 N:9 average weighted slope : 2.582305 correlation (R):
0.524967
See that indeed neither the slope nor the correlation changes much (as
Lindzen also noticed). However, notice that even with 7-month running
average of SW, the slopes of the data points still go all over the
place.
Overall, correlation is low, and the slope (if you can talk about that
with such low correlation) is lower than Lindzen found. There is way to
much noise for me to draw any conclusions other than that there is no
signal to speak of.
I found something very interesting by doing separate OLW <-> SST
analysis, comparing how long-wave radiation changes with changes in
sea-surface temperature :
0.300000 -0.790000 slope: -2.633333
-0.500000 -2.620000 slope: 5.240000
0.300000 0.080000 slope: 0.266667
-0.250000 -2.740000 slope: 10.960000
0.200000 2.700000 slope: 13.500000
0.250000 0.500000 slope: 2.000000
-0.200000 -0.530000 slope: 2.650000
0.600000 1.800000 slope: 3.000000
-0.600000 -3.250000 slope: 5.416667
N:9 average slope : 4.488889 correlation (R): 0.830673
Correlation of 0.83 is pretty strong, indicating that LW radiations
indeed changes with SST. At a slope of about 4 (W/m^2/C). That's pretty
close to the Planck curve, and thus would be expected in absense of any
short-term feedback.
Separate SW <-> SST shows this fuzzy picture :
0.300000 6.810000 slope: 22.700000
-0.500000 -0.930000 slope: 1.860000
0.300000 -1.840000 slope: -6.133333
-0.250000 0.560000 slope: -2.240000
0.200000 -7.370000 slope: -36.850000
0.250000 -1.380000 slope: -5.520000
-0.200000 3.210000 slope: -16.050000
0.600000 1.230000 slope: 2.050000
-0.600000 -1.970000 slope: 3.283333
N:9 average slope : -4.100000 correlation (R): 0.100434
Correlation of 0.1 indicates that there is no correlation between
changes in SW and changes in SST.
My conclusions from my analysis : OLW increases fairly accurately with
changes in SST, at about 4 W/m^2/C, which would be expected from a
system without any short-term feedback. Other than that, everything is
very noisy, and no other conclusions should be drawn.
If Lindzen did indeed base his data on the Edition 3 (rev 1) ERBE data
set, then he should at least show the correlation numbers of the
separate SW <-> SST and OLW <-> SST numbers. I suspect that we will then
see the same low SW correlation and strong OLW correlation that I found.
Concequently, this means that feedback mechanisms seem not present for
the short terms (months) time segments that were present in the
analysis.
That's my 2 cts.
[/quote:7c2d67d7bf]
I'm afraid I don't feel qualified to defend a guy who's spent most of his
life at MIT teaching the subject. That's not an appeal to authority,
just recognition that he knows more about it than I do. If you really
think your analysis shows significant errors in his work, I'm sure he'd
appreciate hearing about them. His work looks OK to me, and fits in with
my personal observations.
But it seems you've confirmed at least the "no positive feedback" portion
of his work, which invalidates all the doom predicting models. That's
fairly significant, I'd say.
[quote:7c2d67d7bf]P.S. Sorry for posting via Google groups. I know this makes replies
difficult, but it's the only way to get my post out right now.
[/quote:7c2d67d7bf]
The bug only affects certain posts, apparently at random. This one
opened OK. |
|
|
| Back to top |
|
|
|
| Rob... |
| Posted: Thu Oct 15, 2009 2:45 pm |
|
|
|
Guest
|
On Oct 14, 9:55am, Bill Ward <bw... at (no spam) ix.REMOVETHISnetcom.com> wrote:
......
[quote:86fc2c27d0]
I found something very interesting by doing separate OLW <-> SST
analysis, comparing how long-wave radiation changes with changes in
sea-surface temperature :
0.300000 -0.790000 slope: -2.633333
-0.500000 -2.620000 slope: 5.240000
0.300000 0.080000 slope: 0.266667
-0.250000 -2.740000 slope: 10.960000
0.200000 2.700000 slope: 13.500000
0.250000 0.500000 slope: 2.000000
-0.200000 -0.530000 slope: 2.650000
0.600000 1.800000 slope: 3.000000
-0.600000 -3.250000 slope: 5.416667
N:9 average slope : 4.488889 correlation (R): 0.830673
Correlation of 0.83 is pretty strong, indicating that LW radiations
indeed changes with SST. At a slope of about 4 (W/m^2/C). That's pretty
close to the Planck curve, and thus would be expected in absense of any
short-term feedback.
Separate SW <-> SST shows this fuzzy picture :
0.300000 6.810000 slope: 22.700000
-0.500000 -0.930000 slope: 1.860000
0.300000 -1.840000 slope: -6.133333
-0.250000 0.560000 slope: -2.240000
0.200000 -7.370000 slope: -36.850000
0.250000 -1.380000 slope: -5.520000
-0.200000 3.210000 slope: -16.050000
0.600000 1.230000 slope: 2.050000
-0.600000 -1.970000 slope: 3.283333
N:9 average slope : -4.100000 correlation (R): 0.100434
Correlation of 0.1 indicates that there is no correlation between
changes in SW and changes in SST.
My conclusions from my analysis : OLW increases fairly accurately with
changes in SST, at about 4 W/m^2/C, which would be expected from a
system without any short-term feedback. Other than that, everything is
very noisy, and no other conclusions should be drawn.
IfLindzendid indeed base his data on the Edition 3 (rev 1) ERBE data
set, then he should at least show the correlation numbers of the
separate SW <-> SST and OLW <-> SST numbers. I suspect that we will then
see the same low SW correlation and strong OLW correlation that I found..
Concequently, this means that feedback mechanisms seem not present for
the short terms (months) time segments that were present in the
analysis.
That's my 2 cts.
I'm afraid I don't feel qualified to defend a guy who's spent most of his
life at MIT teaching the subject. That's not an appeal to authority,
just recognition that he knows more about it than I do. If you really
think your analysis shows significant errors in his work, I'm sure he'd
appreciate hearing about them.
[/quote:86fc2c27d0]
I think there are suspicious inexplicable details in his paper, but
after doing my 'poor-man's' analysis above, I can actually confirm
much of what he reports :
Overall, there is a ~4 W/m^2/K signal in the ERBE data, and it is
almost exclusively from OLW. That means that if SST go up 1 K, that
OLW goes up 4 W/m^2 and visa versa.
There is virtually no correlation between delta-SST and delta-SW.
So far so good.
Now only the conclusions differ.
See below.
[quote:86fc2c27d0]His work looks OK to me, and fits in with my personal observations.
But it seems you've confirmed at least the "no positive feedback" portion
of his work, which invalidates all the doom predicting models. That's
fairly significant, I'd say.
[/quote:86fc2c27d0]
Yes and no.
Indeed we found no significant short-term feedback (neither positive
nor negative) :
ERBE data shows that if SST goes up 1 K, that OLW goes up 4 W/m^2. SW
radiation is not affected.
Radiation theory tells that when there is NO feedback mechanism in
place, that if SST goes up 1 K that OLW goes up 4 W/m^2 and SW is not
affected much.
So, that means that ERBE data is consistent with an Earth that shows
no significant feedback mechanism for the short (months) term. That
means feedback factor 0.
So the paper shows that there is no measurable feedback factor.
Still, Lindzen reports in Figure 3 that the ERBE data implies a
feedback factor of -1, and this paper is now used as proof that there
is strong negative feedback in place, and that thus the
Where did that all come from ? That was not in the ERBE data !!
[quote:86fc2c27d0]P.S. Sorry for posting via Google groups. I know this makes replies
difficult, but it's the only way to get my post out right now.
The bug only affects certain posts, apparently at random. This one
opened OK.[/quote:86fc2c27d0] |
|
|
| Back to top |
|
|
|
| Bill Ward... |
| Posted: Thu Oct 15, 2009 7:48 pm |
|
|
|
Guest
|
On Thu, 15 Oct 2009 17:45:54 -0700, Rob wrote:
[Google Groups apparently corrupted the post, so I'm hand editing it]
[Rob]
I think there are suspicious inexplicable details in his paper, but
after doing my 'poor-man's' analysis above, I can actually confirm
much of what he reports :
Overall, there is a ~4 W/m^2/K signal in the ERBE data, and it is
almost exclusively from OLW. That means that if SST go up 1 K, that
OLW goes up 4 W/m^2 and visa versa.
There is virtually no correlation between delta-SST and delta-SW.
So far so good.
Now only the conclusions differ.
See below.
[Bill]
[quote:a6a0d7a36e]His work looks OK to me, and fits in with my personal observations.
But it seems you've confirmed at least the "no positive feedback"
portion
of his work, which invalidates all the doom predicting models. =A0That's
fairly significant, I'd say.
[/quote:a6a0d7a36e]
Yes and no.
Indeed we found no significant short-term feedback (neither positive
nor negative) :
ERBE data shows that if SST goes up 1 K, that OLW goes up 4 W/m^2. SW
radiation is not affected.
Radiation theory tells that when there is NO feedback mechanism in
place, that if SST goes up 1 K that OLW goes up 4 W/m^2 and SW is not
affected much.
So, that means that ERBE data is consistent with an Earth that shows
no significant feedback mechanism for the short (months) term. That
means feedback factor 0.
So the paper shows that there is no measurable feedback factor.
[Bill]
Actually, you believe your analysis of his paper shows that. Lindzen and
the reviewers apparently believe his analysis of the same data shows
negative feedback. You need to compare notes and see why there's
disagreement. Why don't you write up your analysis and conclusions, then
email Lindzen for his comments? That's the usual way to solve
differences.
********************
Still, Lindzen reports in Figure 3 that the ERBE data implies a
feedback factor of -1, and this paper is now used as proof that there
is strong negative feedback in place, and that thus the
Where did that all come from ? That was not in the ERBE data !!
[Bill]
Your analysis didn't find it. Perhaps Lindzen's analysis was able to
reduce the noise further than yours.
Assuming for the sake of argument that your analysis is correct, and
there is no feedback, does that change your opinion of the models, which
require considerable positive WV feedback to be scary enough to be
effective?
*******************
[quote:a6a0d7a36e]P.S. Sorry for posting via Google groups. I know this makes replies
difficult, but it's the only way to get my post out right now.
The bug only affects certain posts, apparently at random. =A0This one
opened OK.
[/quote:a6a0d7a36e]
[Bill]
This one didn't. Beats me why.
****************** |
|
|
| Back to top |
|
|
|
| Rob... |
| Posted: Fri Oct 16, 2009 8:13 am |
|
|
|
Guest
|
On Oct 15, 6:48pm, Bill Ward <bw... at (no spam) ix.REMOVETHISnetcom.com> wrote:
[quote:0e408cb33a]On Thu, 15 Oct 2009 17:45:54 -0700, Rob wrote:
[Google Groups apparently corrupted the post, so I'm hand editing it]
[Rob]
I think there are suspicious inexplicable details in his paper, but
after doing my 'poor-man's' analysis above, I can actually confirm
much of what he reports :
Overall, there is a ~4 W/m^2/K signal in the ERBE data, and it is
almost exclusively from OLW. That means that if SST go up 1 K, that
OLW goes up 4 W/m^2 and visa versa.
There is virtually no correlation between delta-SST and delta-SW.
So far so good.
Now only the conclusions differ.
See below.
[Bill]
His work looks OK to me, and fits in with my personal observations.
But it seems you've confirmed at least the "no positive feedback"
portion
of his work, which invalidates all the doom predicting models. =A0That's
fairly significant, I'd say.
Yes and no.
Indeed we found no significant short-term feedback (neither positive
nor negative) :
ERBE data shows that if SST goes up 1 K, that OLW goes up 4 W/m^2. SW
radiation is not affected.
Radiation theory tells that when there is NO feedback mechanism in
place, that if SST goes up 1 K that OLW goes up 4 W/m^2 and SW is not
affected much.
So, that means that ERBE data is consistent with an Earth that shows
no significant feedback mechanism for the short (months) term. That
means feedback factor 0.
So the paper shows that there is no measurable feedback factor.
[Bill]
Actually, you believe your analysis of his paper shows that. Lindzenand
the reviewers apparently believe his analysis of the same data shows
negative feedback. You need to compare notes and see why there's
disagreement. Why don't you write up your analysis and conclusions, then
emailLindzenfor his comments? That's the usual way to solve
differences.
[/quote:0e408cb33a]
OK. I wrote a letter to Geophysical Research Letters, which reads
essentially like this :
---
I looked at the Lindzen and Choi paper in detail. I'm not a climate
expert, so I may be wrong here, but I found what seems to be a
fundamental error in reasoning in the paper.
Lindzen did a correlation between changes in outbound radiation (OLW +
SW) from ERBE, against natural changes in sea-surface temperature. He
found a reasonable correlation that shows that total outbound
radiation goes up at about 4 W/m^2 per K increase in sea surface
temperature.
In Figure 3 of the paper, Lindzen shows that the measured 4 W/m^2/K is
almost exclusively caused by an increase in long-wave (OLW) radiation.
The the flux for SW is virtually independent of sea-surface
temperatures (delta-flux/delta-SST is close to 0 W/m^2/K for SW).
Stephan Boltzmann's law says this (increase of OLW radiation at a
slope of 4 W/m^2/K) is exactly what you would expect from a planet
radiating at around 255 K, as long as there is no feedback mechanism
in place.
Still, somehow Lindzen claims that this finding implies a strong
negative feedback, and even claims that the 'models' predict a
negative slope (a decrease in radiation if sea surface temperatures go
up). To obtain a reduction in radiation after an increase in Sea
Surface Temperatures, is essentially physically impossible, with or
without feedback mechanisms.
I think the cause of this error is that he misrepresents the radiative
"forcing" (such as from CO2) with natural changes in surface
temperatures. That confusion leads to an incorrect feedback factor
scale in figure 3 in his paper. In that figure, the SW (short-wave)
graph is off-set by 4 W/m^2. All models, and the right scale (feedback
factor) should move up by 4 W/m^2, so that the 0 W/m^2/K on the left
scale lines up with a feedback factor of 0.
Of course, after correcting this error, the conclusions of his paper
would need to be adjusted as well. Not only is the ERBE data
essentially is in line with the model predictions, but also the ERBE
data shows that there is NO feedback (feedback factor 0) at least for
short-term (months) sea surface temperature changes.
---
[quote:0e408cb33a]
********************
Still,Lindzenreports in Figure 3 that the ERBE data implies a
feedback factor of -1, and this paper is now used as proof that there
is strong negative feedback in place, and that thus the
Where did that all come from ? That was not in the ERBE data !!
[Bill]
Your analysis didn't find it. PerhapsLindzen'sanalysis was able to
reduce the noise further than yours.
[/quote:0e408cb33a]
I don't think the problems are with the data.
It's in the reasoning.
[quote:0e408cb33a]
Assuming for the sake of argument that your analysis is correct, and
there is no feedback, does that change your opinion of the models, which
require considerable positive WV feedback to be scary enough to be
effective?
[/quote:0e408cb33a]
Lindzen only showed that there is no measurable feedback in the short
term, and even that comes with a wide margin of error (due to the
variability of short-term radiation balance).
The feedback in the models is long-term :
Thinks like upper-troposphere water vaper increase and cirrus clouds
and albedo changes due to ice cap melting and methane release from
melting permafrost do not change rapidly.
[quote:0e408cb33a]
*******************
P.S. Sorry for posting via Google groups. I know this makes replies
difficult, but it's the only way to get my post out right now.
The bug only affects certain posts, apparently at random. =A0This one
opened OK.
[Bill]
This one didn't. Beats me why.
[/quote:0e408cb33a]
Sorry about the Google Groups bugs. Which NNTP service are you using ?
[quote:0e408cb33a]
******************[/quote:0e408cb33a] |
|
|
| Back to top |
|
|
|
| Bill Ward... |
| Posted: Fri Oct 16, 2009 1:39 pm |
|
|
|
Guest
|
On Fri, 16 Oct 2009 11:13:45 -0700, Rob wrote:
[This also has the GG bug, and is again hand edited]
[Rob]
[quote:29cbfae42d]ERBE data shows that if SST goes up 1 K, that OLW goes up 4 W/m^2. SW
radiation is not affected.
Radiation theory tells that when there is NO feedback mechanism in
place, that if SST goes up 1 K that OLW goes up 4 W/m^2 and SW is not
affected much.
So, that means that ERBE data is consistent with an Earth that shows no
significant feedback mechanism for the short (months) term. That means
feedback factor 0.
So the paper shows that there is no measurable feedback factor.
[Bill]
Actually, you believe your analysis of his paper shows that. Lindzen
and
the reviewers apparently believe his analysis of the same data shows
negative feedback. You need to compare notes and see why there's
disagreement. Why don't you write up your analysis and conclusions,
then email Lindzen for his comments? That's the usual way to solve
differences.
[/quote:29cbfae42d]
[Rob]
OK. I wrote a letter to Geophysical Research Letters, which reads
essentially like this :
---
I looked at the Lindzen and Choi paper in detail. I'm not a climate
expert, so I may be wrong here, but I found what seems to be a fundamental
error in reasoning in the paper.
Lindzen did a correlation between changes in outbound radiation (OLW + SW)
from ERBE, against natural changes in sea-surface temperature. He found a
reasonable correlation that shows that total outbound radiation goes up at
about 4 W/m^2 per K increase in sea surface temperature.
In Figure 3 of the paper, Lindzen shows that the measured 4 W/m^2/K is
almost exclusively caused by an increase in long-wave (OLW) radiation. The
the flux for SW is virtually independent of sea-surface temperatures
(delta-flux/delta-SST is close to 0 W/m^2/K for SW).
Stephan Boltzmann's law says this (increase of OLW radiation at a slope of
4 W/m^2/K) is exactly what you would expect from a planet radiating at
around 255 K, as long as there is no feedback mechanism in place.
Still, somehow Lindzen claims that this finding implies a strong negative
feedback, and even claims that the 'models' predict a negative slope (a
decrease in radiation if sea surface temperatures go up). To obtain a
reduction in radiation after an increase in Sea Surface Temperatures, is
essentially physically impossible, with or without feedback mechanisms.
I think the cause of this error is that he misrepresents the radiative
"forcing" (such as from CO2) with natural changes in surface temperatures.
That confusion leads to an incorrect feedback factor scale in figure 3 in
his paper. In that figure, the SW (short-wave) graph is off-set by 4
W/m^2. All models, and the right scale (feedback factor) should move up by
4 W/m^2, so that the 0 W/m^2/K on the left scale lines up with a feedback
factor of 0.
Of course, after correcting this error, the conclusions of his paper would
need to be adjusted as well. Not only is the ERBE data essentially is in
line with the model predictions, but also the ERBE data shows that there
is NO feedback (feedback factor 0) at least for short-term (months) sea
surface temperature changes.
---
[Bill]
That should be interesting. Keep us posted on the response.
[quote:29cbfae42d]********************
Still,Lindzen reports in Figure 3 that the ERBE data implies a feedback
factor of -1, and this paper is now used as proof that there is strong
negative feedback in place, and that thus the
Where did that all come from ? That was not in the ERBE data !!
[Bill]
Your analysis didn't find it. Perhaps Lindzen's analysis was able to
reduce the noise further than yours.
[/quote:29cbfae42d]
I don't think the problems are with the data. It's in the reasoning.
[quote:29cbfae42d]Assuming for the sake of argument that your analysis is correct, and
there is no feedback, does that change your opinion of the models, which
require considerable positive WV feedback to be scary enough to be
effective?
[/quote:29cbfae42d]
[Rob]
Lindzen only showed that there is no measurable feedback in the short
term, and even that comes with a wide margin of error (due to the
variability of short-term radiation balance).
The feedback in the models is long-term : Thinks like upper-troposphere
water vaper increase and cirrus clouds and albedo changes due to ice cap
melting and methane release from melting permafrost do not change rapidly.
[Bill]
If you mean by "long-term" feedback, that with an integral term,
remember, it changes sign at the frequency where the delay T is a half
cycle. Short term negative feedback has a long term stabilizing effect.
Delayed feedback can cause oscillation. Non-linear feedback can cause
chaos.
I suspect the deep ocean currents act like a feedback network of delay
lines and are causing a lot of the unpredicted climate swings.
[quote:29cbfae42d]*******************
P.S. Sorry for posting via Google groups. I know this makes replies
difficult, but it's the only way to get my post out right now.
The bug only affects certain posts, apparently at random. =A0This one
opened OK.
[Bill]
This one didn't. Beats me why.
[/quote:29cbfae42d]
Sorry about the Google Groups bugs. Which NNTP service are you using ?
[Bill]
Giganews, with pan as the newsreader.
> ****************** |
|
|
| Back to top |
|
|
|
| Peter Muehlbauer... |
| Posted: Fri Oct 16, 2009 1:49 pm |
|
|
|
Guest
|
Bill Ward <bward at (no spam) ix.REMOVETHISnetcom.com> wrote:
[quote:70f8367e83]Sorry about the Google Groups bugs. Which NNTP service are you using ?
[Bill]
Giganews, with pan as the newsreader.
******************
[/quote:70f8367e83]
If you can find an option for
Content-Transfer-Encoding: quoted-printable
set it to
Content-Transfer-Encoding: text/plain
That's all. |
|
|
| Back to top |
|
|
|
|
|
All times are GMT - 5 Hours
The time now is Mon Nov 30, 2009 10:54 pm
|
|