Basic HTML Version
the highly rated scientific journal Physical
Review Letters, the Rypdals show that the
observations that Scafetta and West used
as the basis for their hypothesis can be ex-
plained in another and far more plausible
way.
"In short, Scafetta and West argue that
although the correlation between solar
activity and the Earth's global temperature
is very weak, there is a common statisti-
cal signature in the two time series that
cannot be arbitrary. They believe that this
means that there is a complex relationship
between solar insolation and the Earth's
climate. What we have discovered is that
these common traits are in fact random
and that the relationship disappears when
you correct for the trends in the time
series."
Conceptual Confusion
"The fundamental weakness of Scafetta
and West's method is that they do not con-
duct a critical test of their own hypothesis.
It is not sufficient to prove that a hypothe-
sis provides predictions that match obser-
vations. A good hypothesis must provide
many accurate predictions," says Rypdal.
Rypdal believes some of the problem
results from the difficulty in defining the
term statistical trend. He says it is impor-
tant to stick to the precise definition of
what a trend is.
"There is a need to establish new mathe-
matical concepts in the climate debate.
Ambiguity in the terminology results in
confusion in the discussion amongst rese-
archers. If we are going to overcome this,
we will have to have many interdiscipli-
nary meetings," says the physics professor.
Conceptual confusion in the scientific
community also leads to greater confusion
in the public debate, and helps to blur the
distinction between science and politics.
"We must be able to discuss each others'
scientific arguments, because as soon as
political motives come to the fore, conspi-
racy theories prevail. And then we confuse
the public even more," he says.
FACTS
• A TIME SERIES is a series of observa-
tions or measurements, often performed
at regular intervals. Example: tempera-
ture measurements over many years.
• A STATISTICAL TREND is a long-term
trend in a time series. Example: increase
in global temperature over the past 40
years.
• STATISTICAL NOISE is random va-
riation in a time series. Noise can have
different types of "memory" and if the
degree of memory is strong, it can be
misinterpreted as a trend.
• STATISTICAL SIGNATURES are
different properties found in random
variations. An example of a signature is
"memory."
• A HYPOTHESIS is a statement about
reality that can be tested. A correctly for-
mulated hypothesis can never be verified
(confirmed) completely because alterna-
tive explanations for an observation can
always be found. But it must in princi-
ple be falsified (disproved) by the new
observations, otherwise the hypothesis is
self-fulfilling.
University of Tromsø –
Labyrint E/11
•••
15
Text: Maja Sojtarić
Photo: NASA