Hi !

I'm student in social modelling and I would like to learn more about
the existing modelling solutions. I suppose that meteorology is a
typical domain were one has to deal with huge parameters space, high
complexity and model strength calculation.

Could you recommand me some lecture for reaching a better understanding
of these problematics ?
- how to determine how essential a parameter is for the whole
simulation run
- determine the quality of prediction
- about kinds of models: iterative (need calulations of all steps from
T(0) to T(x-1) for predicting T(x)) or other

Thanks by advance,

Samuel Thiriot

Samuel Thiriot wrote:
Hi !

I'm student in social modelling and I would like to learn more about
the existing modelling solutions. I suppose that meteorology is a
typical domain were one has to deal with huge parameters space, high
complexity and model strength calculation.

Meteorological modeling is arguably a paradigm for some
kinds of modeling, but models are used in a large variety
of studies and I don't know anything about social modeling
and how it is similar or different. IMO the best models are
those that reproduce the phenomenon of interest in the
simplest way, because it is easiest to understand the
workings of those models. Of course, a complicated
system may need a complicated model and there may
be no way around it (at least at our present level of
knowledge).


Could you recommand me some lecture for reaching a better understanding
of these problematics ?
- how to determine how essential a parameter is for the whole
simulation run

I'm not sure exactly what you're asking. In my expereince
usually parameters are fixed for the entire model run. If the
model performs poorly without whatever process to which
the parameter is related, then that process and appropriate
values of the parameter needs to be added. If the process
is in the model and it doesn't perform well, parameters may
need to be tweaked. If the model performs well, the values
of the parameter may be good or the results may be
fortuitous.

- determine the quality of prediction

In meteorology at least, there are a variety of ways to
verify a forecast. Chapter 7 of _Statistical Methods in
the Atmospheric Sciences_ (2nd Ed.) by Danial Wilks
covers the subject, and other references are in the
literature as well. The best method in a given case
depends somewhat on the form of the forecast and
verification data.

- about kinds of models: iterative (need calulations of all steps from
T(0) to T(x-1) for predicting T(x)) or other

Thanks by advance,

Samuel Thiriot

Again the form of the model depends on what one is
trying to model and the resources one has. A book
that might give you a bit of a sense of some atmospheric
science models, including a CD with a few models you
can play with on a PC, is _A Climate Modeling Primer_
(3rd Ed.) by McGuffie and Henderson-Sellers.

Cheers,
Russell

I had a look at the book you referenced -

http://cires.colorado.edu/steffen/cl...ng%20Primer%22

Global climate modelling based on axial tilt is actually modelling of
hemispherical weather patterns (seasons) and the original description
is found in De Revolutionibus Chapter 11 -

http://webexhibits.org/calendars/yea...opernicus.html

Global climate modelling using the motions of the Earth is entirely
different and requires a radical modification to the original
Copernican descriptions.Temperature signatures reflecting the constant
radiation received by the Earth's surface is bound up in the change in
orbital orientation against fixed axial orientation -

http://www.climateprediction.net/ima...ges/annual.gif

The specific modification to global climate modelling centers on the
transition from explaining hemispherical weather patterns (seasons) by
way of axial tilt to a more satisfactory situation where hemispherical
weather patterns are a subset of global climate ,and global climate is
due to the changing relationship between axial and orbital motion with
constant solar radiation rather than the Sun's position taking a more
prominent role.

If climatologists have difficulties discerning the changing orbital
orientation of the Earth against fixed axial orientation (and it is
notoriously difficult) they can use the pronounced axial orientation of
Uranus as a guide to what ours with our planet in terms of the change
in orbital orientation against fixed axial orientation -

http://www.nordita.dk/~steen/fysik51...s/AACHCVO0.JPG


It is an entirely different and more accurate way to astronomically
approach global climate modelling,it keeps all things local such as
constant radiation,constant axial rotation and variable orbital motion
which current models do not take into account.












wrote:
Samuel Thiriot wrote:
Hi !

I'm student in social modelling and I would like to learn more about
the existing modelling solutions. I suppose that meteorology is a
typical domain were one has to deal with huge parameters space, high
complexity and model strength calculation.

Meteorological modeling is arguably a paradigm for some
kinds of modeling, but models are used in a large variety
of studies and I don't know anything about social modeling
and how it is similar or different. IMO the best models are
those that reproduce the phenomenon of interest in the
simplest way, because it is easiest to understand the
workings of those models. Of course, a complicated
system may need a complicated model and there may
be no way around it (at least at our present level of
knowledge).


Could you recommand me some lecture for reaching a better understanding
of these problematics ?
- how to determine how essential a parameter is for the whole
simulation run

I'm not sure exactly what you're asking. In my expereince
usually parameters are fixed for the entire model run. If the
model performs poorly without whatever process to which
the parameter is related, then that process and appropriate
values of the parameter needs to be added. If the process
is in the model and it doesn't perform well, parameters may
need to be tweaked. If the model performs well, the values
of the parameter may be good or the results may be
fortuitous.

- determine the quality of prediction

In meteorology at least, there are a variety of ways to
verify a forecast. Chapter 7 of _Statistical Methods in
the Atmospheric Sciences_ (2nd Ed.) by Danial Wilks
covers the subject, and other references are in the
literature as well. The best method in a given case
depends somewhat on the form of the forecast and
verification data.

- about kinds of models: iterative (need calulations of all steps from
T(0) to T(x-1) for predicting T(x)) or other

Thanks by advance,

Samuel Thiriot

Again the form of the model depends on what one is
trying to model and the resources one has. A book
that might give you a bit of a sense of some atmospheric
science models, including a CD with a few models you
can play with on a PC, is _A Climate Modeling Primer_
(3rd Ed.) by McGuffie and Henderson-Sellers.

Cheers,
Russell

The OP seemed to have a pretty vague knowledge of modeling, so
the reference I gave was just to provide an example of *some* kind
of modeling. Your comments reveal another good point for the OP
to learn about modeling, which Cox put as "All models are wrong,
some are useful."

Cheers,
Russell

oriel36 wrote:
I had a look at the book you referenced -

http://cires.colorado.edu/steffen/cl...ng%20Primer%22

Global climate modelling based on axial tilt is actually modelling of
hemispherical weather patterns (seasons) and the original description
is found in De Revolutionibus Chapter 11 -

http://webexhibits.org/calendars/yea...opernicus.html

Global climate modelling using the motions of the Earth is entirely
different and requires a radical modification to the original
Copernican descriptions.Temperature signatures reflecting the constant
radiation received by the Earth's surface is bound up in the change in
orbital orientation against fixed axial orientation -

http://www.climateprediction.net/ima...ges/annual.gif

The specific modification to global climate modelling centers on the
transition from explaining hemispherical weather patterns (seasons) by
way of axial tilt to a more satisfactory situation where hemispherical
weather patterns are a subset of global climate ,and global climate is
due to the changing relationship between axial and orbital motion with
constant solar radiation rather than the Sun's position taking a more
prominent role.

If climatologists have difficulties discerning the changing orbital
orientation of the Earth against fixed axial orientation (and it is
notoriously difficult) they can use the pronounced axial orientation of
Uranus as a guide to what ours with our planet in terms of the change
in orbital orientation against fixed axial orientation -

http://www.nordita.dk/~steen/fysik51...s/AACHCVO0.JPG


It is an entirely different and more accurate way to astronomically
approach global climate modelling,it keeps all things local such as
constant radiation,constant axial rotation and variable orbital motion
which current models do not take into account.












wrote:
Samuel Thiriot wrote:
Hi !

I'm student in social modelling and I would like to learn more about
the existing modelling solutions. I suppose that meteorology is a
typical domain were one has to deal with huge parameters space, high
complexity and model strength calculation.

Meteorological modeling is arguably a paradigm for some
kinds of modeling, but models are used in a large variety
of studies and I don't know anything about social modeling
and how it is similar or different. IMO the best models are
those that reproduce the phenomenon of interest in the
simplest way, because it is easiest to understand the
workings of those models. Of course, a complicated
system may need a complicated model and there may
be no way around it (at least at our present level of
knowledge).


Could you recommand me some lecture for reaching a better understanding
of these problematics ?
- how to determine how essential a parameter is for the whole
simulation run

I'm not sure exactly what you're asking. In my expereince
usually parameters are fixed for the entire model run. If the
model performs poorly without whatever process to which
the parameter is related, then that process and appropriate
values of the parameter needs to be added. If the process
is in the model and it doesn't perform well, parameters may
need to be tweaked. If the model performs well, the values
of the parameter may be good or the results may be
fortuitous.

- determine the quality of prediction

In meteorology at least, there are a variety of ways to
verify a forecast. Chapter 7 of _Statistical Methods in
the Atmospheric Sciences_ (2nd Ed.) by Danial Wilks
covers the subject, and other references are in the
literature as well. The best method in a given case
depends somewhat on the form of the forecast and
verification data.

- about kinds of models: iterative (need calulations of all steps from
T(0) to T(x-1) for predicting T(x)) or other

Thanks by advance,

Samuel Thiriot

Again the form of the model depends on what one is
trying to model and the resources one has. A book
that might give you a bit of a sense of some atmospheric
science models, including a CD with a few models you
can play with on a PC, is _A Climate Modeling Primer_
(3rd Ed.) by McGuffie and Henderson-Sellers.

Cheers,
Russell

wrote:
The OP seemed to have a pretty vague knowledge of modeling, so
the reference I gave was just to provide an example of *some* kind
of modeling. Your comments reveal another good point for the OP
to learn about modeling, which Cox put as "All models are wrong,
some are useful."

Cheers,
Russell


The transition from variable axial tilt as a working principle to a
more accurate viewpoint based on changing orbital orientation is a
massive task to undertake.Copernicus was describing hemispherical
weather pattern ( seasons) and using variable tilt against the Sun to
achieve this objective whereas contemporary concerns would require a
complete modification of this explanation in order to distinguish the
difference betwen oscillation of temperature signatures arising from
the motions of the Earth and variations due to human activity or other
events not included in the astronomical cycle,the distance from the Sun
and other astronomical data.

The recognition of the daily change in orbital orientation is probably
initially difficult to discern and I thought the pronounced axial
orientation of Uranus would help to highlight that the Earth
orientation changes also.An additional helpful image showing change in
orbital orientation is this -

http://upload.wikimedia.org/wikipedi...easonearth.png

Climatologists may be so accustomed to regarding variable axial tilt
to the Sun as a viable working principle that the image above showing a
change in orbital orientation against fixed axial orientation will make
them feel uncomfortable and I concede that it is difficult to consider
things from an orbital point of view.Those images, although accurate,
have a less severe look when the upcoming Equinox is presented in terms
of orbital and axial orientations.On Sept 21st,the orbital orientation
of the Earth will align with axial longitudes sweeping through it at
dawn and dusk.

The single greatest difficulty in making the transition to a better
astronomical working principles for global climate modelling is not
with Copernicus but with later men in the 17th century who assigned a
variable tilt of the Earth to explain the Equation of Time
correction.The main component of the Equation of the Time is the rate
of change of orbital orientation of the Earth in accordance with
Keplerian orbital geometry.We are back at trying to get climatologists
and astronomers to recognise the changing orbital orientation of the
Earth.

Thank you for your response.











oriel36 wrote:
I had a look at the book you referenced -

http://cires.colorado.edu/steffen/cl...ng%20Primer%22

Global climate modelling based on axial tilt is actually modelling of
hemispherical weather patterns (seasons) and the original description
is found in De Revolutionibus Chapter 11 -

http://webexhibits.org/calendars/yea...opernicus.html

Global climate modelling using the motions of the Earth is entirely
different and requires a radical modification to the original
Copernican descriptions.Temperature signatures reflecting the constant
radiation received by the Earth's surface is bound up in the change in
orbital orientation against fixed axial orientation -

http://www.climateprediction.net/ima...ges/annual.gif

The specific modification to global climate modelling centers on the
transition from explaining hemispherical weather patterns (seasons) by
way of axial tilt to a more satisfactory situation where hemispherical
weather patterns are a subset of global climate ,and global climate is
due to the changing relationship between axial and orbital motion with
constant solar radiation rather than the Sun's position taking a more
prominent role.

If climatologists have difficulties discerning the changing orbital
orientation of the Earth against fixed axial orientation (and it is
notoriously difficult) they can use the pronounced axial orientation of
Uranus as a guide to what ours with our planet in terms of the change
in orbital orientation against fixed axial orientation -

http://www.nordita.dk/~steen/fysik51...s/AACHCVO0.JPG


It is an entirely different and more accurate way to astronomically
approach global climate modelling,it keeps all things local such as
constant radiation,constant axial rotation and variable orbital motion
which current models do not take into account.












wrote:
Samuel Thiriot wrote:
Hi !

I'm student in social modelling and I would like to learn more about
the existing modelling solutions. I suppose that meteorology is a
typical domain were one has to deal with huge parameters space, high
complexity and model strength calculation.

Meteorological modeling is arguably a paradigm for some
kinds of modeling, but models are used in a large variety
of studies and I don't know anything about social modeling
and how it is similar or different. IMO the best models are
those that reproduce the phenomenon of interest in the
simplest way, because it is easiest to understand the
workings of those models. Of course, a complicated
system may need a complicated model and there may
be no way around it (at least at our present level of
knowledge).


Could you recommand me some lecture for reaching a better understanding
of these problematics ?
- how to determine how essential a parameter is for the whole
simulation run

I'm not sure exactly what you're asking. In my expereince
usually parameters are fixed for the entire model run. If the
model performs poorly without whatever process to which
the parameter is related, then that process and appropriate
values of the parameter needs to be added. If the process
is in the model and it doesn't perform well, parameters may
need to be tweaked. If the model performs well, the values
of the parameter may be good or the results may be
fortuitous.

- determine the quality of prediction

In meteorology at least, there are a variety of ways to
verify a forecast. Chapter 7 of _Statistical Methods in
the Atmospheric Sciences_ (2nd Ed.) by Danial Wilks
covers the subject, and other references are in the
literature as well. The best method in a given case
depends somewhat on the form of the forecast and
verification data.

- about kinds of models: iterative (need calulations of all steps from
T(0) to T(x-1) for predicting T(x)) or other

Thanks by advance,

Samuel Thiriot

Again the form of the model depends on what one is
trying to model and the resources one has. A book
that might give you a bit of a sense of some atmospheric
science models, including a CD with a few models you
can play with on a PC, is _A Climate Modeling Primer_
(3rd Ed.) by McGuffie and Henderson-Sellers.

Cheers,
Russell