At 07:32 AM 10/11/2009, you wrote:
The NS article pointed at ways, through the use of cascading
fractals, that we could overcome the limitations of even the very
short term weather predictions available at present.
MAYBE. The idea seems to have been around for a quite a while and
hasn't really gone anywhere. The thrust of the article was that
someone noticed that the GCMs exhibit fractal behaviour. Stunning, as
they exhibit what they are designed to exhibit. Oh and you have to
change exponents in the fractals as you go.
Be wary of science by press release. I think successful scientists
could make very good livings as salesmen. It is a vital skill for them.
RASP is currently running on a 12km grid - and soaking up all the
computational resources that our sponsor can provide. One of the
issues that this produces is that the geography in he model is also
"smeared". For the Darling Downs, this means that Toowoomba range is
not a steep escarpment but a relatively gentle slope. The effect of
this in the model is to significantly over-emphasise the maritime
incursion should there be any on the coast. It also makes it very
hard (effectively impossible) to model the Bunya wave system. These
situations arise not as some longterm future date of a weather
prediction, but within 24 hours of the forecast date.
What is the vertical grid size? If it can't model the Bunyas it
probably doesn't handle the scenic rim near Boonah either. Or the
surface roughness and UHI of Brisbane. How about aerosols? Are they
included? Note also that computer produced weather charts don't have
fronts. They have trouble with discontinuities. Still! Don't forget
also that the models for a limited area like south east Queensland
need boundary initialisation from the hemispherical GCM. You may not
be doing what you think you are doing. All I've seen out of these
models so far is "the thermals will be stronger and higher inland
and over higher ground".
There is nothing surprising in there being multiple, different
models of the atmosphere. First of all, as soon as you build any
mathematical simulation of a physical world system, even very simple
ones, it is an abstraction - a model. Science allows us to build
very many very successful models - including really complex ones.
With a system as complex as a planetary atmosphere, it is also
unsurprising that there are multiple models as many different teams
will build models for a whole variety of reasons (a major one being
the "Not Invented Here" syndrome). The point in the article was that
all the major models appeared to demonstrate the fractal cascade effect.
For chrissakes this modeling has been going on since the 1960s. I can
believe the number of models can multiply initially as people try out
different things but if they get tested against observation they
should get winnowed to one by now which incorporates all the
successful features. That's if they are based on physics which they
aren't beyond a shallow level. We have figured out how aeroplanes fly
after all and even the airfoil design programs have converged as
better ones came along. Yes even they have fudge factors for laminar
to turbulent transition but even that is no longer controversial. (
Xfoil works very well for very low to very high Reynolds numbers. You
can even download it for free and run it on your PC. Somebody even
did a nice user interface that isn't reminiscent of early 60s geek.)
Oh - and all the various major models provide significantly better
short term results than reading tea leaves or chicken entrails!
I never said they did. The real question is are they better than
person with local experience and recent data on actual winds and
temperatures vertically and some sat photos can do.
Beware also of a very successful forecasting technique known as
"persistence" the situation is about the same as yesterday so we'll
forecast the same. This is usually modified by "perturbation". aka
"it will be a bit hotter/cooler than yesterday. It really is
*reasonably* successful. What doing this with computer models really
does is give people the perfect excuse for why the forecast was wrong
- the models failed! The BoM uses 3 different models. You'll notice
the forecasters get all vague when the 3 models say different things
and confident when they say the same thing. Still sometimes wrong.
Mike
Borgelt Instruments - manufacturers of quality soaring instruments since 1978
phone Int'l + 61 746 355784
fax Int'l + 61 746 358796
cellphone Int'l + 61 428 355784
email: [email protected]
website: www.borgeltinstruments.com
_______________________________________________
Aus-soaring mailing list
[email protected]
To check or change subscription details, visit:
http://lists.internode.on.net/mailman/listinfo/aus-soaring
