At 04:37 PM 21/08/2014, you wrote:
"Quality control." From talking to them in the past, the actual
observations can be pretty noisy and lossy.
Yeah nature is a bitch ain't she?
If the model is accurate, the actual numerical values ought to be the same.
So let me get this straight - the model which is initialised and
updated by the earlier noisy observations should be relied upon to
produce "better" observations than the actual MEASURED physical
variables at a present time and place, right? uh-huh.
What they really mean is that the models smooth the data I think. Did
they happen to tell you the grid size horizontally, vertically and in
time that the models use? The models also ignore physical variables
that are difficult to measure (do they really measure albedo over say
10km x 10km grids in real time? Or trace gases with interesting
properties, emitted by biology)
If the models smooth the data and then compute the state at future
time from this it is my understanding that this can get you into
trouble. Along with measurement imprecision it seems to be one of the
reasons that models drift from reality over a few days at most and
cannot predict the weather for a given week at a given location even
3 months into the future. I'd still like to see a mathematical proof
that the average weather for a year for a complete hemisphere in say
2050 is predictable by the same models and isn't just integrated
random noise. When you ask this question the answer involves arm
waving and unbacked assertions about "averaging" fixing the problem.
The bad news is that averaging and then integrating random noise
doesn't work (bitter experience). It is why inertial navigation
systems drift with time and must be reset to known physical states
either continuously (artificial horizon) or by some other navigation method.
Virtually everything the BoM produces comes out of the simulation
model, up to and including the rain radar views.
If the rain radar views are model outputs why do we see artifacts on
the view from time to time at various locations which appear to be
returns from hills etc? A model would surely remove these if properly
programmed. There may be some confusion here. The model being
referred to may be the model relating strength of radar return to
rainfall rate which is a completely different animal from the global
circulation models.
> And why then do you sometimes see that at a certain location
there is no latest F160 because the balloon didn't fly?
I've never been able to get a good answer for that.
The page link I gave clearly says aviation OBSERVATIONS. The BoM
seems careful to say where things are forecasts or model outputs in
other parts of its website. The bloke in Brisbane I talked to this
morning seemed to think that the things on that page were in fact the
measurements from the balloon flight. I hope so.
The latest radiosondes don't rely on radar returns for winds BTW.
They have a GPS chip in the sonde so get wind speed and direction off
that. I'm waiting for the penny to drop and somebody puts the
electronics package in a foam glider which flys to a collection point.
BTW I love how some people are free with your time, in spite of your
considerable contribution so far. Maybe somebody could volunteer to
do these negotiations with BoM for you ? :-)
Mike
Borgelt Instruments - design & manufacture of quality soaring
instrumentation since 1978
www.borgeltinstruments.com
tel: 07 4635 5784 overseas: int+61-7-4635 5784
mob: 042835 5784 : int+61-42835 5784
P O Box 4607, Toowoomba East, QLD 4350, Australia
_______________________________________________
Aus-soaring mailing list
[email protected]
To check or change subscription details, visit:
http://lists.internode.on.net/mailman/listinfo/aus-soaring
