global warming hoax

"OG" wrote in message
...

"Peter Webb" wrote in message
...

"Chris L Peterson" wrote in message
...
On Sat, 28 Nov 2009 16:39:48 +1100, "Peter Webb"
wrote:

You didn't answer my question about whether you had ever seen "rolling
averages" used in a peer reviewed journal outside of climate "science".

A rolling average is almost always used when evaluating data for trends,
because it is the primary statistical method for measuring trends. This
is the case for all scientific data, and non-scientific data as well.
When analyzing a time-series for a trend, you select a boxcar size that
is appropriate for the scale of the noise, or of systematic effects that
you want to tune out. When looking at temperature trends, you ask
yourself two things: what maximum time scale are you interested in, and
what is the time scale of the noise. It is apparent when you look at the
raw, annualized temperature record that noise is approximately decadal.
That's why the most common boxcar applied is a 10-year filter. And since
the maximum time scale is usually a few hundred years, a 10-year filter
is still fine enough to clearly show trends.


Yes, but a ten year rolling average cannot be validly used to show that
temperatures have increased or decreased over the last 10 years, as it
includes data from 10 years ago.

Indeed, a temperature graph showing 1998 temperatures as a 10 rolling
average in fact has equal contributions from 1997 to 2007. If (say) 1999
was a very hot year, it would contribute just as much to the stated
temperature as the temperatures in 2007.

Rolling averages are statistically meaningless. The choice to average
over multiple years is an entirely arbitrary one, and the period is
arbitrary.

Yes, if only I'd put a 5 year rolling average as well so that we could see
how that changed matters - oh Hang on, I did!


The temperature in 2006 was the temperature in 2006. The temperature in
2006 was not the average of the years 1996, 1997, 1998 ..., the
temperature in 2006 was the temperature in 2006.



There's a physical reality to using a decadal filter as well, since many
known short term effects are on this scale: oscillating current patterns
like the El Nino, the solar cycle, and other components of very short
term climate or weather.

The temperature in 2007 was the temperature in 2007. The temperature in
2007 was not the average of the temperatures in 1997, 1998 ...

To attempt to compare 2007 temperature with 1998 temperature using a
measure (rolling averages) which actually constructs a nominal 2007
temperature using 1998 data (equally weighted in the calculation with
2006 data) is obviously and clearly bogus.

That is why it often appears in sales proposals, and never in real
science. Its just playing with statistics. And the fact that they feel
the need to play unjustified and unfounded statistical games in the
presentation of their data makes me wonder why they have to bother ...

So you don't believe in trends ?


I don't believe in massaging figures to show trends which are not evident in
the source data.

The trend could easily be an artefact of the massaging process. That is why
this sort of massaging does not occur in real science.



Nobody is using filters like this to hide things. They just recognize
the reality that what the temperature does over a few years doesn't mean
anything in terms of climate until at least a decade is considered. You
can look at the raw temperature record for the last century, and there
are decadal periods where temperatures increased quickly or slowly,
decreased quickly or slowly, or held steady. Yet the warming trend is
clear and unambiguous, and only by smoothing the data can the actual
structure of that trend be easily seen (in particular the increasing
rate of heating in the last few decades).
_________________________________________________

I see. If they presented the actual data, the trend would not be seen, so
they manipulate the data statistically so it shows what they want it to
show.

The annual data is presented, What makes you incapable of seeing the
increasing temperature trend?


No, the annual data is not presented.

Annual data would show for (say) the temperature in 2007 the temperature in
2007.

As I understand it, the temperature you show for 2007 is not the temperature
in 2007. In fact, it is the result of adding 10 different numbers - being
the temperatures in 1998, 1999, 2000 ... and then dividing by 10. Only 10%
of the value of the temperature in 2007 that you provide derives from
temperature measurements in 2007, 90% of the value of 2007 temperatures
actually comprises different years including some up to 10 years old.

However, as I understand your argument, the warming would not be evident if
you just used the actual temperatures, and that is why you don't.

So if the warming trend is not clear from the raw data, but is clear from
the smoothed data, the trend was caused by the smoothing process and is not
intrinsic to the data.

This is why this sort of "trick" is so popular in sales proposals, and so
rare in real science (as opposed to climate science).



This happen a lot in astrophysics?



Chris L Peterson
Cloudbait Observatory
http://www.cloudbait.com

Incidentally, I am quite sure that the earth warmed from the middle of
the 19th century to about 1998 or 1999. I don't know if it warmed from
1999 to 2009. I asked you before if you thought it had; lots of people
believe that it is basically unchanged over the last 10 years, what is
your opinion?


1998 an exceptionally warm year, but that is on top of an increasing trend
as you can see.

"Peter Webb" wrote in message
...

"OG" wrote in message
...

"Peter Webb" wrote in message
...

"Chris L Peterson" wrote in message
...
On Sat, 28 Nov 2009 16:39:48 +1100, "Peter Webb"
wrote:

You didn't answer my question about whether you had ever seen "rolling
averages" used in a peer reviewed journal outside of climate "science".

A rolling average is almost always used when evaluating data for
trends,
because it is the primary statistical method for measuring trends. This
is the case for all scientific data, and non-scientific data as well.
When analyzing a time-series for a trend, you select a boxcar size that
is appropriate for the scale of the noise, or of systematic effects
that
you want to tune out. When looking at temperature trends, you ask
yourself two things: what maximum time scale are you interested in, and
what is the time scale of the noise. It is apparent when you look at
the
raw, annualized temperature record that noise is approximately decadal.
That's why the most common boxcar applied is a 10-year filter. And
since
the maximum time scale is usually a few hundred years, a 10-year filter
is still fine enough to clearly show trends.


Yes, but a ten year rolling average cannot be validly used to show that
temperatures have increased or decreased over the last 10 years, as it
includes data from 10 years ago.

Indeed, a temperature graph showing 1998 temperatures as a 10 rolling
average in fact has equal contributions from 1997 to 2007. If (say) 1999
was a very hot year, it would contribute just as much to the stated
temperature as the temperatures in 2007.

Rolling averages are statistically meaningless. The choice to average
over multiple years is an entirely arbitrary one, and the period is
arbitrary.

Yes, if only I'd put a 5 year rolling average as well so that we could
see how that changed matters - oh Hang on, I did!


The temperature in 2006 was the temperature in 2006. The temperature in
2006 was not the average of the years 1996, 1997, 1998 ..., the
temperature in 2006 was the temperature in 2006.



There's a physical reality to using a decadal filter as well, since
many
known short term effects are on this scale: oscillating current
patterns
like the El Nino, the solar cycle, and other components of very short
term climate or weather.

The temperature in 2007 was the temperature in 2007. The temperature in
2007 was not the average of the temperatures in 1997, 1998 ...

To attempt to compare 2007 temperature with 1998 temperature using a
measure (rolling averages) which actually constructs a nominal 2007
temperature using 1998 data (equally weighted in the calculation with
2006 data) is obviously and clearly bogus.

That is why it often appears in sales proposals, and never in real
science. Its just playing with statistics. And the fact that they feel
the need to play unjustified and unfounded statistical games in the
presentation of their data makes me wonder why they have to bother ...

So you don't believe in trends ?


I don't believe in massaging figures to show trends which are not evident
in the source data.

The trend could easily be an artefact of the massaging process. That is
why this sort of massaging does not occur in real science.



Nobody is using filters like this to hide things. They just recognize
the reality that what the temperature does over a few years doesn't
mean
anything in terms of climate until at least a decade is considered. You
can look at the raw temperature record for the last century, and there
are decadal periods where temperatures increased quickly or slowly,
decreased quickly or slowly, or held steady. Yet the warming trend is
clear and unambiguous, and only by smoothing the data can the actual
structure of that trend be easily seen (in particular the increasing
rate of heating in the last few decades).
_________________________________________________

I see. If they presented the actual data, the trend would not be seen,
so they manipulate the data statistically so it shows what they want it
to show.

The annual data is presented, What makes you incapable of seeing the
increasing temperature trend?


No, the annual data is not presented.

Yes it is.
Take a look at the second plot on the page.
The one marked Temperature Anomaly 1990-2007, the one with "Annual data"
points.

Annual data would show for (say) the temperature in 2007 the temperature
in 2007.

As I understand it, the temperature you show for 2007 is not the
temperature in 2007. In fact, it is the result of adding 10 different
numbers - being the temperatures in 1998, 1999, 2000 ... and then dividing
by 10. Only 10% of the value of the temperature in 2007 that you provide
derives from temperature measurements in 2007, 90% of the value of 2007
temperatures actually comprises different years including some up to 10
years old.

You understand wrong then. The plot that says "annual data" is just that,
not 5 year averages, not 10 year averages. Single year data, just like you
want.

However, as I understand your argument, the warming would not be evident
if you just used the actual temperatures, and that is why you don't.

What in my words makes you think that?

So if the warming trend is not clear from the raw data, but is clear from
the smoothed data, the trend was caused by the smoothing process and is
not intrinsic to the data.

Just look at the data.

This is why this sort of "trick" is so popular in sales proposals, and so
rare in real science (as opposed to climate science).


Just look at the data

"Martin Brown" wrote in message
...
Peter Webb wrote:

"Chris L Peterson" wrote in message
...
On Sat, 28 Nov 2009 16:39:48 +1100, "Peter Webb"
wrote:

You didn't answer my question about whether you had ever seen "rolling
averages" used in a peer reviewed journal outside of climate "science".

A rolling average is almost always used when evaluating data for trends,
because it is the primary statistical method for measuring trends. This
is the case for all scientific data, and non-scientific data as well.
When analyzing a time-series for a trend, you select a boxcar size that
is appropriate for the scale of the noise, or of systematic effects that
you want to tune out. When looking at temperature trends, you ask
yourself two things: what maximum time scale are you interested in, and
what is the time scale of the noise. It is apparent when you look at the
raw, annualized temperature record that noise is approximately decadal.
That's why the most common boxcar applied is a 10-year filter. And since
the maximum time scale is usually a few hundred years, a 10-year filter
is still fine enough to clearly show trends.


Yes, but a ten year rolling average cannot be validly used to show that
temperatures have increased or decreased over the last 10 years, as it
includes data from 10 years ago.

A rolling average would normally be done centred on the most recent year
with a loss of signal to noise as the end of the dataset is reached since
only the past 5 points can contribute. Whilst one might complain about
using an even number of points in the boxcar there is nothing wrong with
this method for improving signal to noise in raw data.

Indeed, a temperature graph showing 1998 temperatures as a 10 rolling
average in fact has equal contributions from 1997 to 2007. If (say) 1999
was a very hot year, it would contribute just as much to the stated
temperature as the temperatures in 2007.

Rolling averages are statistically meaningless. The choice to average
over multiple years is an entirely arbitrary one, and the period is
arbitrary.

The choice of period isn't arbitrary. There is good reason to average over
something commensurate with the solar sunspot cycle and a few other
roughly decadal natural frequency components.

Why "roughly decadal" ? Why didn't you use 11 years, which would better
match the sunspot cycle, if that was your motivation?

Further, the graph is no longer one of temperatures, but rather temperatures
modified to eliminate solar variability, which is explicit modification of
the data to suppress a cause of climate variability which is not
anthropogenic. If you want to show that the changes are anthropogenic,
massaging the data to eliminate non-anthropogenic causes is going to do
that. Its not science, of course.



The first periodic term not adequately corrected is the 3rd harmonic of
the Saros at roughy 60 years. This is also visible in the HADCRUT time
series from 1850-2009

The temperature in 2006 was the temperature in 2006. The temperature in
2006 was not the average of the years 1996, 1997, 1998 ..., the
temperature in 2006 was the temperature in 2006.

Depends how he did it. But the point remains that the trend so far is ever
upwards with by eyeball from the trendline about 0.03C Gaussian noise
year. You cannot read anything into a single outlier.


Well, you say that after you massage the data its clear, but I gather its
not at all clear if you provide the actual temperatures.

You arbitrarily changed the data (through a smoothing process) to make
evident a trend that is not at all clear from te original data. And you
justify this by saying that the raw data doesn't clearly show a trend, and
that is why you modified it.

You do understand that this is a circular argument, don't you? And that this
sort of thing would never be accepted in real science?



There's a physical reality to using a decadal filter as well, since many
known short term effects are on this scale: oscillating current patterns
like the El Nino, the solar cycle, and other components of very short
term climate or weather.

The temperature in 2007 was the temperature in 2007. The temperature in
2007 was not the average of the temperatures in 1997, 1998 ...

To attempt to compare 2007 temperature with 1998 temperature using a
measure (rolling averages) which actually constructs a nominal 2007
temperature using 1998 data (equally weighted in the calculation with
2006 data) is obviously and clearly bogus.

That is why it often appears in sales proposals, and never in real
science. Its just playing with statistics. And the fact that they feel
the need to play unjustified and unfounded statistical games in the
presentation of their data makes me wonder why they have to bother ...

These filters are used in real science.


Not in this manner.


Nobody is using filters like this to hide things. They just recognize
the reality that what the temperature does over a few years doesn't mean
anything in terms of climate until at least a decade is considered. You
can look at the raw temperature record for the last century, and there
are decadal periods where temperatures increased quickly or slowly,
decreased quickly or slowly, or held steady. Yet the warming trend is
clear and unambiguous, and only by smoothing the data can the actual
structure of that trend be easily seen (in particular the increasing
rate of heating in the last few decades).
_________________________________________________

I see. If they presented the actual data, the trend would not be seen, so
they manipulate the data statistically so it shows what they want it to
show.

This happen a lot in astrophysics?

Boxcar averaging is used very commonly in scientific signal processing. It
is extremely fast and so is used in a lot of realtime control with noisy
raw data to select out the real signal from noise.


But not to compare the last boxcar with previous ones and hence infer a
trend. The fact is, a 10 year boxcar includes 10 year old data, and so is
far too coarse to show changes occuring within a 10 year timeframe. But that
is exactly what you are trying to do here.

Why?

Well, according to you, its because if you provided actual figures there is
no obvious trend, so the AGW solution is just to modify the data (in this
case through 10 year averages) until a way is found to make the data look
like your hypothesis.



And most radio astronomy uses it for certain operations although if more
CPU time is available better low pass filters are used. They are all the
rage for terrestrial planet detection:

http://arxiv.org/abs/0812.1667


Incidentally, I am quite sure that the earth warmed from the middle of
the 19th century to about 1998 or 1999. I don't know if it warmed from
1999 to 2009. I asked you before if you thought it had; lots of people
believe that it is basically unchanged over the last 10 years, what is
your opinion?

"Dittohead science" graphs are designed to mislead the public and are done
with a start date of 1998. This is the highest anomaly ever seen. If you
least squares fit the past decade using that single exceptional point as
your reference datum then you can force a faked cooling trend.



How is the cooling trend derived in this manner any more fake than you
adding 1998 temperatures into the formula for 2007 temperature?

It seems considerably less fake to me.


The reality though is that when you allow for the noise the trend remains
upwards or flat and certainly no significant cooling.


So it may have been flat, huh?

Sounds like your graph is wrong, if in your opinion the raw data could show
temperatures remaining flat over the last 10 years, but your graph with
modified temperatures shows it increasing.



There is one thing you should think about. Australia is the most likely
continent to be very badly affected in the future by AGW with even hotter
and drier conditions in the interior.


So you say.

Again, without evidence.

Regards,
Martin Brown

"Peter Webb" wrote in message
u...

"Martin Brown" wrote in message
...
Peter Webb wrote:

"Chris L Peterson" wrote in message
...
On Sat, 28 Nov 2009 16:39:48 +1100, "Peter Webb"
wrote:

You didn't answer my question about whether you had ever seen "rolling
averages" used in a peer reviewed journal outside of climate
"science".

A rolling average is almost always used when evaluating data for
trends,
because it is the primary statistical method for measuring trends. This
is the case for all scientific data, and non-scientific data as well.
When analyzing a time-series for a trend, you select a boxcar size that
is appropriate for the scale of the noise, or of systematic effects
that
you want to tune out. When looking at temperature trends, you ask
yourself two things: what maximum time scale are you interested in, and
what is the time scale of the noise. It is apparent when you look at
the
raw, annualized temperature record that noise is approximately decadal.
That's why the most common boxcar applied is a 10-year filter. And
since
the maximum time scale is usually a few hundred years, a 10-year filter
is still fine enough to clearly show trends.


Yes, but a ten year rolling average cannot be validly used to show that
temperatures have increased or decreased over the last 10 years, as it
includes data from 10 years ago.

A rolling average would normally be done centred on the most recent year
with a loss of signal to noise as the end of the dataset is reached since
only the past 5 points can contribute. Whilst one might complain about
using an even number of points in the boxcar there is nothing wrong with
this method for improving signal to noise in raw data.

Indeed, a temperature graph showing 1998 temperatures as a 10 rolling
average in fact has equal contributions from 1997 to 2007. If (say) 1999
was a very hot year, it would contribute just as much to the stated
temperature as the temperatures in 2007.

Rolling averages are statistically meaningless. The choice to average
over multiple years is an entirely arbitrary one, and the period is
arbitrary.

The choice of period isn't arbitrary. There is good reason to average
over something commensurate with the solar sunspot cycle and a few other
roughly decadal natural frequency components.

Why "roughly decadal" ? Why didn't you use 11 years, which would better
match the sunspot cycle, if that was your motivation?

Further, the graph is no longer one of temperatures, but rather
temperatures modified to eliminate solar variability, which is explicit
modification of the data to suppress a cause of climate variability which
is not anthropogenic. If you want to show that the changes are
anthropogenic, massaging the data to eliminate non-anthropogenic causes is
going to do that. Its not science, of course.

The graph isn't there to show 'anthopogenic' anything. It is there to show
warming. The warming that you deny exists over the last decade.


The first periodic term not adequately corrected is the 3rd harmonic of
the Saros at roughy 60 years. This is also visible in the HADCRUT time
series from 1850-2009

The temperature in 2006 was the temperature in 2006. The temperature in
2006 was not the average of the years 1996, 1997, 1998 ..., the
temperature in 2006 was the temperature in 2006.

Depends how he did it. But the point remains that the trend so far is
ever upwards with by eyeball from the trendline about 0.03C Gaussian
noise year. You cannot read anything into a single outlier.


Well, you say that after you massage the data its clear, but I gather its
not at all clear if you provide the actual temperatures.

When you say "I gather its not at all clear", do you mean someone else has
told you this? You've not looked at the data yourself?


You arbitrarily changed the data (through a smoothing process) to make
evident a trend that is not at all clear from te original data. And you
justify this by saying that the raw data doesn't clearly show a trend, and
that is why you modified it.

How do you know the trend is 'not at all clear from the original data"?
You've not looked at the original data

You do understand that this is a circular argument, don't you? And that
this sort of thing would never be accepted in real science?




There's a physical reality to using a decadal filter as well, since
many
known short term effects are on this scale: oscillating current
patterns
like the El Nino, the solar cycle, and other components of very short
term climate or weather.

The temperature in 2007 was the temperature in 2007. The temperature in
2007 was not the average of the temperatures in 1997, 1998 ...

To attempt to compare 2007 temperature with 1998 temperature using a
measure (rolling averages) which actually constructs a nominal 2007
temperature using 1998 data (equally weighted in the calculation with
2006 data) is obviously and clearly bogus.

That is why it often appears in sales proposals, and never in real
science. Its just playing with statistics. And the fact that they feel
the need to play unjustified and unfounded statistical games in the
presentation of their data makes me wonder why they have to bother ...

These filters are used in real science.


Not in this manner.


Nobody is using filters like this to hide things. They just recognize
the reality that what the temperature does over a few years doesn't
mean
anything in terms of climate until at least a decade is considered. You
can look at the raw temperature record for the last century, and there
are decadal periods where temperatures increased quickly or slowly,
decreased quickly or slowly, or held steady. Yet the warming trend is
clear and unambiguous, and only by smoothing the data can the actual
structure of that trend be easily seen (in particular the increasing
rate of heating in the last few decades).
_________________________________________________

I see. If they presented the actual data, the trend would not be seen,
so they manipulate the data statistically so it shows what they want it
to show.

This happen a lot in astrophysics?

Boxcar averaging is used very commonly in scientific signal processing.
It is extremely fast and so is used in a lot of realtime control with
noisy raw data to select out the real signal from noise.


But not to compare the last boxcar with previous ones and hence infer a
trend. The fact is, a 10 year boxcar includes 10 year old data, and so is
far too coarse to show changes occuring within a 10 year timeframe. But
that is exactly what you are trying to do here.

Why?

Well, according to you, its because if you provided actual figures there
is no obvious trend, so the AGW solution is just to modify the data (in
this case through 10 year averages) until a way is found to make the data
look like your hypothesis.

Can you please just look at the actual chart that uses the "actual figures"
that were on my original web page (2nd plot) and tell me if you can see a
trend or not.

Can you also describe a 'smoothing' function that can make a declining trend
appear as an increasing trend? You seemed to claim it was possible (indeed
you claimed I have actually done it).

"Sam Wormley" wrote in message
news:wzjQm.150330$5n1.7444@attbi_s21...
Peter, I'm telling you is that greenhouse gases trap heat... the is a
greater concentration of greenhouse gases (observable), surface (land
and sea) temperatures are rising (observable) altering climate
(observable),
altering ice and sea level (observable). It's that simple, Peter.

800,00 year record of CO2
http://edu-observatory.org/olli/800000yrs_CO2.png

Global surface temperature 1845-2007:

http://www.pewclimate.org/docUploads...emp-trends.gif

Land supported ice and sea temperature, both contribute to the rising
sea level. About 20 cm in the last 125 years. More to come.

http://www.wildwildweather.com/forec...level_rise.png
http://earthobservatory.nasa.gov/ima...200606_lrg.pdf


What's worrying is that moderate warming at high latitudes appears to
threaten the initiation of two further sources of greenhouse gases -
methane from warming tundra (when the frozen subsoil thaws, anaerobic
decomposition generates methane) and from the decomposition of methane
clathrates on the Norwegian continental shelf.

On Nov 27, 2:37*am, "Peter Webb"
wrote:
Sam glacier length has shown a shortenong trend since about 1820(before
the
use of coal wa widespread) and the slope of the trend line has not changed
at all with the increases in fossil fuel usage that really took off in the
1920s... if you want to see the 180 year trend graphs of both plotted
together look he

http://www.uwgb.edu/dutchs/PSEUDOSC/GlobWarm0.HTM

if you don't want to see data that conflicts with your belief then by all
means avert you attention and others by calling someone a name

Yes that is precisely what you are doing. There is a preponderance of
evidence that supports the concept of anthropogenic global warming.
Yes please point us to geologist who also teaches history and hasn't
published a research paper in over 10 years and explain to us how he
refutes thousands of scientists in fields as diverse as chemistry,
biology and computer science with his unsupported personal opinion

Here is a homework assignment for you. You wanted to know how we can
determine how we have been responsible for the increase in CO2 in the
air since the industrial revolution. Look up what the ratio of C12 to
C13 isotopes for natural (plant/animal/volcano etc) CO2 generation and
then look up the *ratio of C12 to C13 for fossil fuel combustion.
Describe how those ratios have changed since the industrial
revolution. As a hint to help you along the way look up

Stable isotope ratio mass spectrometry in global climate change
research
Prosenjit Ghosh, Willi A. Brand doi:10.1016/S1387-3806(03)00289-6

After that a google scholar search will provide you with the
appropriate papers to read. Report with the answer. I doubt you will
or better yet you will find yet another site with a unsupported
personal opinion maybe this time from a washed up TV journalist who
takes pictures with a telephoto lens

_________________________________
This long rave of yours has nothing to do with what he said. He made a
statement about glacier retreat. You launched into ad-hominem attack on him,
and then changed the topic, thus fulfilling his prophecy that you would
ignore evidence you didn't like.

By the way, making ad-hominem attacks and changing the subject are two of
the key characteristics of a crank.

DIdn't change the subject, merely pointed out the web site YOU pointed
is full of factual errors because he knows nothing about any of the
topics he talks about. He MAY know something about structural geology,
but he clearly knows ZERO about climatology and again makes factual
errors that a freshman physics major wouldn't make.

I gave you a homework assignment that was related to the errors your
website reference makes AND would provide answers to questions you
asked in an earlier post about how do you we know the CO2 increases
are anthropogenic in nature. You failed to read the content of the
post and wandered off on a weird tangent just like Gerald.

Just like in post #56 you quote several web sites about the little ice
age hoping that will bolster your non-existent arguments. Yet you fail
to read everything and fail to note that the little ice age was a
western and central European phenomenon and not recorded any where
else in the norther hemisphere at the same time. Try reading (yes I
know it's hard)


http://www.windows.ucar.edu/tour/lin....html&edu=high

or

Little Ice Age Volume 1, The Earth system: physical and chemical
dimensions of global environmental change, pp 504–509

"OG" wrote in message
...

"Peter Webb" wrote in message
...


No 'they', it was me that presented the data that way as a means of
smoothing out the data from year to year.

How would you prefer to see the data presented?


I could make many suggestions, quite obviously I would prefer that a
graph which show the temperature in 2005 as being the temperature in
2005, not 10% of the temperature in 2005 + 10% of the temperature in 2004
+ ... + 10% of the temperature in 1996.


Done that, see the graph "temperature anomaly 1990 - 2007"

And how and where do they explain why they did this, and how they
decided which period to average each year's data over? Was it simply
because it made the prettiest graph?

From the annual temperature graph below you can see that 1998 was an
exceptional year, but it lies on top of a generally increasing trend.
You seem to ignore that graph.

How would you rather see the data presented?


As I understand it, your motivation for using rolling averages is that it
makes the graph clearer.

However, if it is not clear from the raw data that a trend is obvious,
then it probably doesn't exist. If you have to massage the data to show
something, then that thing may not (and probably does not) exist.

See the graph "temperature anomaly 1990 - 2007 - do you honestly not see
any trend there?


As I understand it, there is no motivation whatsoever for the decision on
whether to pick 2 year, 5 year, 10 year or any other period for the
rolling average, other than it produces the best graph. Probably
unconsciously, you picked a special way to massage the data - a 5 year
term - because this produced a curve with the result you wanted. This
happens a lot in climate science. You believe, so you don't care or
notice that you deliberately picking a way of massaging the data to
produce the answers (in this case a plausible temperature curve) you
want.

I challenge you to find an averaging that shows a declining trend in
temperatures.

Does the raw data show an increase in temperatures?

If it did, why did you feel the need to massage your data?

Whoops, that's right, you massaged the data because the raw data didn't show
a clear trend.

"Chris L Peterson" wrote in message
...
On Sat, 28 Nov 2009 18:23:04 +1100, "Peter Webb"
wrote:

Yes, but a ten year rolling average cannot be validly used to show that
temperatures have increased or decreased over the last 10 years, as it
includes data from 10 years ago.

And the data aren't used for that purpose.


Incorrect. I specifically asked ... "has the earth warmed or cooled over the
last 10 years", and the graph was provided for exactly the purpose of
proving a warming trend over the last 10 years.



You can always look at the
raw data, but it isn't statistically useful for determining anything at
all about the climate trend.

I see. Raw data is useless for showing that the temperature has increased
over the last 10 years. So you change the data. Got it.



There's no method that can be used to show
if "temperatures have increased or decreased over the last 10 years",
since that's a statistically meaningless period in which to answer that
question. Which is, of course, why it is silly to attempt to use the
last few years of data to decide if global warming is real or not.


You should tell that to the AGW-believer who posted massaged data that
showed the earth had warmed over the last 10 years and used this as evidence
of AGW.

I agree that the last 10 years data to do not allow to conclude the earth is
continuing to warm. I am particularly pleased that you seem willing to (at
least in principle) accept that at least some of the evidence of AGW posted
in this newsgroup is crap.



Rolling averages are statistically meaningless. The choice to average over
multiple years is an entirely arbitrary one, and the period is arbitrary.

These two statements just demonstrate again that you lack the
qualifications to carry on this discussion in a reasonable way.


Ohh, an ad-hominem attack. I was waiting for that. You have continued your
fine tradition of attacking people you disagree with rather than posting
evidence you are correct.


Incidentally, I am quite sure that the earth warmed from the middle of the
19th century to about 1998 or 1999. I don't know if it warmed from 1999 to
2009. I asked you before if you thought it had; lots of people believe
that
it is basically unchanged over the last 10 years, what is your opinion?

I don't believe the question means anything as stated. We understand
very well the mechanism that has primarily contributed to warming since
the middle of the 19th century,

I have asked repeatedly for this, and nobody has told me what it is.

What is it?


and theory does a fair job of predicting
the trend. We can accurately measure the inputs to the system, and we
therefore expect the trend to continue. I have no reason to think
otherwise.


Other than the fact that every warming period in the last 4 billion years
has been followed by a cooling period.


So what I believe is that when we look back on the data in a
couple of decades, we will see the trend continuing, probably rising
faster at the later dates.

I know you believe that. What I don't understand is why you believe that.


For the period from 1999-2009 we'll also see
the trend rising, although it will be a bit flatter during that period.
In other words, it will look like many decadal periods over the last 150
years.

In other words, it will look like many periods in history when there was
very little anthropogenic CO2.

Which really does make you wonder if CO2 has anything to do with it, when
(according to you) we have just finished a decade which looks much like many
other decades when anthropogenic CO2 was definitely not a factor.


_________________________________________________

Chris L Peterson
Cloudbait Observatory
http://www.cloudbait.com

On Sun, 29 Nov 2009 15:18:23 +1100, "Peter Webb"
wrote:

I see. Raw data is useless for showing that the temperature has increased
over the last 10 years. So you change the data. Got it.

No, you don't.

Ohh, an ad-hominem attack. I was waiting for that.

You don't understand the term. If somebody displays ignorance, and then
displays a lack of rationality by failing to pay any attention to the
questions that are answered, it is not an ad hominem attack to point out
that the person is both ignorant and irrational. Both those terms
describe you, based on your many posts here. You are demonstrably not
competent to discuss this topic- you don't understand the science, you
don't understand basic statistical analysis, you don't understand how
climate models work. This is simple fact; there is nothing ad hominem
about it.
_________________________________________________

Chris L Peterson
Cloudbait Observatory
http://www.cloudbait.com

"Sam Wormley" wrote in message
news:wzjQm.150330$5n1.7444@attbi_s21...
Peter, I'm telling you is that greenhouse gases trap heat... the is a
greater concentration of greenhouse gases (observable), surface (land
and sea) temperatures are rising (observable) altering climate
(observable),
altering ice and sea level (observable). It's that simple, Peter.

800,00 year record of CO2
http://edu-observatory.org/olli/800000yrs_CO2.png


Here is a graph of CO2 over the last 600 million years.

http://www.geocraft.com/WVFossils/Ca...s_climate.html

It shows that CO2 levels are very close to record lows. Indeed, the only
time that CO2 levels were anywhere near this low was about 300 million years
ago.

Your graph - with an 800,000 year baseline - attempts to show that we are
entering a period with CO2 far higher than in the past. The reality is that
we are in a period of extremely low CO2 levels (by historical standards).
This doesn't exactly support the premise of AGW, so the answer of the AGW
believers is to pick a cut-off date that suits the theory (800,000 years),
and completely ignore the data that contradicts the theory (the other
599,200,00 years).

You guys do this a lot.

Its not science.


Global surface temperature 1845-2007:

http://www.pewclimate.org/docUploads...emp-trends.gif

Land supported ice and sea temperature, both contribute to the rising
sea level. About 20 cm in the last 125 years. More to come.

http://www.wildwildweather.com/forec...level_rise.png
http://earthobservatory.nasa.gov/ima...200606_lrg.pdf


I'm not arguing that the temperature of the earth increased from 1850 to
2000. I have no idea whether the earth warmed over the last 10 years. If it
has, it is by a very modest amount.

I am also aware that the results of the last decade to not match up with
what was predicted by AGW models. At least no-one has produced a single AGW
model which predicted almost static temperatures for this last decade.

You know what I think of scientific theories where the result of experiments
does not match the results predicted by the theory? I think they are wrong.
Its called the scientific method.