Presumably they can only give such different outputs if the raw
meteorlogical data has changed? So what sort of readings fluctuate this
quickly in 12 hour periods because the actual weather type would rarely (and
I know it does sometimes) change in that time scale.
On current form it seems no models are reliable, even for short periods
ahead. What percentage of a M.O. outlook would be based on models compared
to experience, pattern matching, other?

Dave

I would suggest to anyone who is interested in meteorology, that they should
get a few good books on mathematical modelling, fluid dynamics, and chaos
theory.
These subjects are taught well on the intense Met Office forecasting
courses.
It's just not enough that you know how to spot a cloud and name it
correctly, you really have to understand the four dimensional aspects of the
atmosphere, and the physics/chemistry involved.
The computerised Met centres all around the globe all use either gridpoint
or spectral models, but both types are subject to chaos theory, or what some
have named the butterfly effect. This is where small alterations in the
initialisation fields of the model "simulation" can multiply through
iteration, into larger scale and sometimes dramatic changes in the longer
term forecast. Study of a "julia Set" for example will show how this
works.....get a fractal generator off the web and play with small changes in
the initial settings to see how it works.

Hope this helps.

M



Dave. C wrote in message
...
Presumably they can only give such different outputs if the raw
meteorlogical data has changed? So what sort of readings fluctuate this
quickly in 12 hour periods because the actual weather type would rarely
(and
I know it does sometimes) change in that time scale.
On current form it seems no models are reliable, even for short periods
ahead. What percentage of a M.O. outlook would be based on models compared
to experience, pattern matching, other?

Dave

"Dave. C" wrote in message
...
Presumably they can only give such different outputs if the raw
meteorlogical data has changed? So what sort of readings fluctuate this
quickly in 12 hour periods because the actual weather type would rarely
(and
I know it does sometimes) change in that time scale.
On current form it seems no models are reliable, even for short periods
ahead. What percentage of a M.O. outlook would be based on models compared
to experience, pattern matching, other?

All these systems can be represented by models (it might be tricky to
actually code up a met man but you could lock him in a black box and call it
a "black-box" model). Try looking up chaos theory or "chaotic systems" for a
bit of useful background on non-linear systems.

Stephen

I would suggest to anyone who is interested in meteorology, that they should
get a few good books on mathematical modelling, fluid dynamics, and chaos
theory.
.../..
It's just not enough that you know how to spot a cloud and name it
correctly, you really have to understand the four dimensional aspects of the
atmosphere, and the physics/chemistry involved.
...... Study of a "julia Set" for example will show how this
works.....get a fractal generator off the web and play with small changes in
the initial settings to see how it works.

Hope this helps.

M
Hey Michael,fine sentiments,but the British don't think esp about the weather!;-).,.But a julia set
may not be a good example,why not fiddle with a Lorenz attractor?
Google will produce lots of hits,inc
http://archive.ncsa.uiuc.edu/SCMS/Di...or-Hobill.html
http://mathworld.wolfram.com/LorenzAttractor.html
http://archives.math.utk.edu/topics/...rDynamics.html

--
regards,
david
(add 17 to waghorne to reply)

thanks Michael, when I get a spare minute or two ..........

"Michael" wrote in message
...
I would suggest to anyone who is interested in meteorology, that they
should
get a few good books on mathematical modelling, fluid dynamics, and chaos
theory.
These subjects are taught well on the intense Met Office forecasting
courses.
It's just not enough that you know how to spot a cloud and name it
correctly, you really have to understand the four dimensional aspects of
the
atmosphere, and the physics/chemistry involved.
The computerised Met centres all around the globe all use either gridpoint
or spectral models, but both types are subject to chaos theory, or what
some
have named the butterfly effect. This is where small alterations in the
initialisation fields of the model "simulation" can multiply through
iteration, into larger scale and sometimes dramatic changes in the longer
term forecast. Study of a "julia Set" for example will show how this
works.....get a fractal generator off the web and play with small changes
in
the initial settings to see how it works.

Hope this helps.

M



Dave. C wrote in message
...
Presumably they can only give such different outputs if the raw
meteorlogical data has changed? So what sort of readings fluctuate this
quickly in 12 hour periods because the actual weather type would rarely
(and
I know it does sometimes) change in that time scale.
On current form it seems no models are reliable, even for short periods
ahead. What percentage of a M.O. outlook would be based on models
compared
to experience, pattern matching, other?

Dave

Sorry Waghorn, typed "Julia Set" actually mean't "Lorenz"... finger and
brain in non-perfect harmony!

Was just talking Julia Sets to a colleague, and it was fresh in the mind.



"Waghorn" wrote in message
...
I would suggest to anyone who is interested in meteorology, that they
should
get a few good books on mathematical modelling, fluid dynamics, and chaos
theory.
../..
It's just not enough that you know how to spot a cloud and name it
correctly, you really have to understand the four dimensional aspects of
the
atmosphere, and the physics/chemistry involved.
..... Study of a "julia Set" for example will show how this
works.....get a fractal generator off the web and play with small changes
in
the initial settings to see how it works.

Hope this helps.

M
Hey Michael,fine sentiments,but the British don't think esp about the
weather!;-).,.But a julia set
may not be a good example,why not fiddle with a Lorenz attractor?
Google will produce lots of hits,inc

http://archive.ncsa.uiuc.edu/SCMS/Di...tractor-Hobill.
html
http://mathworld.wolfram.com/LorenzAttractor.html
http://archives.math.utk.edu/topics/...rDynamics.html

--
regards,
david
(add 17 to waghorne to reply)