The site above by John Daly of Australia is a good resource for skepticism
about global warming. As with any advocacy site (such as the EPA)
you have to take what it says with a grain of salt. I've gone to the
original article and responses in Nature to get more information.
The specific point made on the web page above is somewhat misleading,
although the larger point about time spans does seem valid. Daly shows
a chart representing temperature data from the period studied in the
original paper, followed by another chart showing temperature data for
a longer period. An upward trend is clearly visible in the first chart
but vanishes in the second.
The misleading aspect is that it might appear that the first chart was
used in the original paper. In fact this is not the case. Daly's two
charts are based on a set of measurements by Angell which cover the time
span from 1958-1995. The original paper used different measurements,
by Oort and Liu, which cover only a narrower time span. The original
authors, in their response to the critique, claim that the Angell
data "has instrumental biases and known deficiencies in its spatial
representativeness." So they did not use that data set supposedly
because it was not accurate enough for them. In particular, they
did not select a subset of the Angell data and display it, which is
the implication you might take from Daly's web page.
Actually, the original paper has no charts to show temperature increase,
and surprisingly enough, the original paper is not about warming at
all! Rather, the paper is looking for changes in the *distribution*
of temperatures by altitude and lattitude. Global temperature changes
are completely averaged out and discarded in its analysis.
What the original paper looked at was changes in the distribution of
temperature, based on the data by Oort and Liu. It does not show any
simple one dimensional graph such as Daly has on his web page. Instead
there is a two dimensional chart showing temperature change over the
period as a function of lattitude and altitude. What the chart shows
is that the upper atmosphere has gotten cooler compared to the lower
atmosphere, and that the northern hemisphere has gotten cooler compared
to the southern hemisphere.
The purpose of the paper was to look at climate models and see if any
of them predicted the same distribution of temperatures. It appears
that models which include carbon dioxide along with sulfate aerosols do
reproduce the data relatively well. Adding an ozone contribution makes
the fit even better. CO2 cools the upper atmosphere and warms the lower,
while the sulfate aerosols are concentrated in the northern hemisphere
near industrial sources, and provide cooling.
So simple analyses in terms of global temperature changes over the years
in question don't go to the heart of the original paper. What does matter
is that the differences in temperature between northern and southern
hemispheres also appear to go away when you extend the time frame. That's
the actual point of the chart shown by Daly, although he doesn't say so.
It represents southern hemisphere temperatures, and their decrease in the
past decade effectivelly eliminates the previous temperature differential
between north and south which the models explained so well.
In short, extending the time frame causes the measurements to *depart*
from the model predictions, while within the time frame of the original
paper the measurements were becoming *closer* to the model predictions.
In their response, the original authors admit this. They don't like the
Angell data, but they use a different set of measurements from Parker
and they confirm that the temperature distributions depart from the
predictions after 1988. However they claim that this is OK, that some
departures can be expected.
I didn't find this part convincing. Yes, broadly speaking there can be
variations, for example if the sulfates get out ahead of the CO2 you
could have some cooling. But the whole point of their original paper
was the agreement with the models, and even more it was the increase
in correlation between models and data. Showing that the correlation
immediately turned around and began to decrease after the time span
studied by their paper has to make you question their results. At a
minimum they have to explain why the models don't predict the temperature
distribution after 1988. But they don't go into that at all.
Hal