Flying Tadpole wrote:
Martin Baxter wrote:
Flying Tadpole wrote:
Gilligan wrote:
? http://msnbc.msn.com/id/15391426/site/newsweek
?
?
. But
of course we have lotsa big computers now, so that's all right, and the
predictions must therefore be much more reliable and accurate. GIGO.

Well things do change, Galileo was eventually proved correct (more or
less):

"The point to remember, says Connolley, is that predictions of global
cooling never approached the kind of widespread scientific consensus
that supports the greenhouse effect today. And for good reason: the
tools scientists have at their disposal now-vastly more data,
incomparably faster computers and infinitely more sophisticated
mathematical models-render any forecasts from 1975 as inoperative as the
predictions being made around the same time about the inevitable triumph
of communism."

Cheers
Marty

Which is a longer-winded way of saying "of course we have lotsa big
computers now, so that's all right, and the
predictions must therefore be much more reliable and accurate." To
which I add "GIGO" because, quite simply, the modelling is a multiple
generation extrapolation (model based on model output based on model
output) using a simply inadequate data base. Too short a time scale
with reliable data.

I was always intrigued when a former client demanded we stop using
simple analytical techniques, with confidence tests based on testable
null hypotheses, on their SO2 problem and start using a surface
modeller, where confidence tests were not applied in the strict sense
and where the model was a generator, not a tester of hypotheses. It was
fun, though, tweaking the various model parameters and treatments to
generate the reverse of what (our) independent observations were indicating.

I discovered, earlier this year, that the absolute best coastal models
applied by supposedly the best Oz modelling scientists could not cope
with longshore drift (==mandatory sailing content) because it was too
close to shore. These are the same people making a "significant global
contribution" to current climate modelling. Yeah. GIGO.

Can you send me the ref on that? Hotmail addy works. I'm interested in
this stuff tho I haven't really been following the field in the last 25
years - more interested in dealing with the 'GI' part of the problem.
With qualified success, I might add (modestly).... pity about the sea
surface temps of 99999 I found yesterday in historic data (ie before my
time).

I agree with the GIGO bit and I supply the modellers with data. In fact
I ran a query the other day for one of the ocean modellers extracting
sea surface temperature blocked by 1 deg lat/lon grid by day for a big
chunk of the Southern Ocean. Interesting in that the timeline is waay
too short to show trends, if indeed there are trends to show. Classic
S/N problem. One of these days I'd like to give up my current role and
go back to playing with data but the money & fringe bennies are too
good ATM.

I'm sort of thinking on building an accurate micro scale data set on
the bay where I live. I'm 90% sure there's a gyre trapping nutrients
from a fish farm just north of us. Hard to explain the algal blooms any
other way - very low runoff, not much agriculture and very few people,
all on reasonable bits of land with septic systems. If I ever finish
building I'm gonna have a good look at what's happening. Meanwhile I'm
building an cough unapproved structure blocking the longshore drift
which has the twin benefits of providing me with my own personal beach
and helping keep the creek mouth open, not burying the oyster covered
rocks.

Thankx, PDW