Mark & Juanita wrote:
On 3 Jul 2005 20:51:22 -0700, wrote:



Mark & Juanita wrote:
On 3 Jul 2005 00:41:19 -0700, wrote:



Mark & Juanita wrote:
On 2 Jul 2005 21:57:11 -0700, wrote:


... snip


... alright fred, present a credible source for how global temperature
change in tenths of a degree (which is the amount and rate being cited for
global warming evidence) can be identified for periods before accurate
weather records were kept.

Why should I? You haven't presented any credible source indicating
that anyone claims to be able to do so.



As you are so fond of saying, Google is your friend.

My first response was snide and I've deleted it.

What cuaght my eye was your statement "The idea that by measuring
tree ring size, one can determine the average temperature of an
area to within tenths of a degree is ludicrous."

A quick google search using the search terms "tree ring" and "average
temperature" does not yield anyoone making such a claim.

So, I remain skeptical that such a claim has been made.


Try to hearken
back to various debates in which the infamous "hockey stick" chart is shown
that attempts to show departures from average temperatures in 1961 to 1990
for the years 1000 AD to current time, showing this sudden jump of +.5C
when the other tempertures were below. The difference is less than 0.5 C.
The "measurements" from tree rings, corrals, ice cores and "historical
records" (remember that no calibrated met stations existed in 1000 AD) are
all being pegged at less than 0.5 C increments.

A Google search for "hockey stick chart" yields a few pages that come
up 404, perhaps due ot the NHL strike, and a few that criticize
"the hocky stick chart" but I haven't found any explanation of the
chart itself.

One example is found he

http://www.greeningearthsociety.org/...4/wca_24c.html

The chart appears to be a graph of temperature as a function
of time. Note the caption on the left side which indicates
the temperature origin is a "1961 to 1990 average." What is
meant by "1961 to 1990 average" is a mystery to me but inasmuch
as choice of origin is arbitrary let's not worry about
that.


You have the correct chart. This is the chart that various ("Earth in
the Balance") former presidential candidates have used to highlight the
future devastation to be caused by the alarming increase in temperature in
only the past several years.

Have you found anything written about the chart by the persons who
(allegedly) created it?

The 1961 to 1990 average temperature was
taken as a baseline and is the zero bar of said chart.

The choice of origin is still as arbitrary
as it was when DesCartes introduced (or popularized) x-yplots.
E.g. they could have used 0 degrees C as their origin, the inter-
pretation would be the same, though John McCain would need stilts
or a very long pointer when using the chart.

The numbers below
zero indicate average temperatures below the reference bar and those above
indicate average temperatures greater than the reference. The large spike
at the end of the chart is intended to cause alarm due to a) it's large
slope and b) the fact that it is fully 0.5 C above the average from the
previous 30 years and well above the average for the past millenia.

"Intended to cause alarm?" That implies MOTIVE. Can you show that
the shart was drawn that way "to cause alarm" rather than to
conform to the data?



It looks to me like the error bars (in grey--if those are
not error bars I don't know what they are) are about +/-
.5 degree for observations prior to about 1600, perhaps +/-
.3 degrees from 1600 - 1900 and I won't hazard a guess as
to what they are in the more recent data.


Given that the grey bars are error bars, then the overall exercise and
alarmism raised by the presentation of said chart are beyond simple
hysterics and border on fraud.

How so?

Planning for the future should be based on predictions for the future
from Climate models that are validated by close fits to historical
data. You can't extrapolate by 'looking at' a plot, for any but the
simplest of linear models. I don't think climate models fall into
that category so I don't see how the chart in question fits into
the scientific debate.

The blue and red lines are those focused
upon the by the Chicken Little crowd. The error bars indicate that this
entire exercise is attempting to extrapolate future climate from noise.
Having spent the last 15 years of my career in various development projects
that rely heavily upon integration and test and data collection, I can
categorically state that attempting to extrapolate performance from noise
measurements is a fool's errand.

That does't make any sense. How well the data fit the model over
the period of observation how one tests validity of a model. Not
how noisy it 'looks'. Very seldom can one look at a plot of real
world data and see somethign meaningful. The question one
needs to ask as a first step to deterimining the predictive value
of the model in question is how well it fits the data.

As you know, mathematically valid results may be extracted from data
that to the human eye, appear to be randomly distributed, even as
the human eye may 'see' trends in data where mathematics tells us
there are none.



So, what again is your objection? Do you feel the the variance
in the data prior to 1600 was underestimated? If so, what
do you allege has been mishandled in the error estimation?


That the error bars are only 0.5C is the first part that anyone with some
degree of skepticism should focus upon.

As you know, whether or not the estimated uncertainties in the data
are correct can be objectively tested. So, have they been?

The second is the deltas that are
being extrapolated for periods before the advent of the thermometer are
being assessed at less than 0.5C, when the exact causes for tree ring size,
ice core sample depth, and other "indicators" are hardly precise enough to
estimate global average temperature to such a degree of precision.

I do not see that the chart extrapolates any deltas anywhere.
The charts shows mean temperatures (still undefined) vs time.
As you know the standard deviation of a mean is inversely
proportionate to the square root of the number of observations
The size of the statistically correct error bars on any 'average'
can be made arbitrarily small simply by gathering enough
data.



What do you mean by "pegged at less than 0.5 C increments"?

"Pegged" is usually used to mean a hard limit, for example met
sensor data showing relative humidity inexcess of 100% may be
arbitrarily adjusted to ("pegged" at) 100% at ingest, though
the term more often refers to a hard limit on the measurement
device itself (e.g. "pegging the meter").


Oh please, let's not play games with semantics, you know darned well what
I meant,

Please do not lie about me. I am not playing games with semantics.
In Mathematics and Science words are carefully defined so as to
facilitate communication. When someone starts thowing them around
without regard to those defintions communincation is obstructed.

i.e. that the error bars shown are at best 0.5C, the attempt to
show increments of less than 0.1C are simply ludicrous.

If you claim the errors are underestimated, what is your basis?
Are his chi squares too small? Also consider that we are only
GUESSING that those are error bars and even if we are, we do not
know for what confidence interval. Still if you show your
arithmetic, I'll check it out.

Are you SERIOUSLY objecting to the _tic spacing_ on the vertical
axis? If so, you'd better be tough that is just as arbitrary as the
choice of origin.

BTW, by 'arbitrary' I mean 'has no effect on interpretation',
hence my comment regarding toughness.

Substitute
"represented", or "reported" for "pegged" if that makes you feel any
better.

None of them do. Please explain what you mean, rather than play
games with semantics. Do you accuse the author of understimating
the uncertainty in his data? If so, show your evidence.


The only 0.5 degree C increments I see are the major tick spacing
on the vertical (temperature) axis. Again, like the choice of
origin, that is arbitrary.

IOW, I don't see anything here to the effect of "that by measuring
tree ring size, one can determine the average temperature of an
area to within tenths of a degree is ludicrous."


Take a closer look a the graph, the numbers for the era before the
thermometer was invented are being estimated based upon tree ring
measurements, ice core samples and historical records (i.e, some current
era literati writing, "dang, it's cold this winter!" or "We had to order 5
more pairs of longjohns this winter"). Now, look at the blue lines, look
at the zero reference line, this graph is trying to tell you that global
average temperature was moving around 0.2 to 0.5C below the global average
reference line.

The chart is not a claim by anyone "that by measuring tree ring size,
one can determine the average temperature of an area to within tenths
of a degree." THAT is obvious, just by looking at it. We don't
even know what each 'point' being plotted actually represents. It
is possible that each point on the chart is actually extracted from
its own database each with a large number of observations. As you
know the variance decreases in inverse proportion to the number of
data points. E.g. the chart may represent a so-called 'meta'
study, a examination of an ensemble of other persons' resuults,
treating their conclusions as data.

Why don't we know these things? Well for starters, we haven't
yet found anything written about this chart by the author, have
we?

I also don't believe any scientist basing predictions on future
climate on THAT chart. That's not the way scientist make pre-
dictions, especially about the future. That the chart gets
presented a lot, does not mean that anyone who knows a burro
from a burrow actually uses it for anything other than illustrative
purposes.

Do you claim that the chart is a fake, not supported by data?

If so, what is your evidence?

--

FF