List of Observables for Observation Core components DM

Petr Skoda skoda at sunstel.asu.cas.cz
Fri Sep 24 02:05:42 PDT 2010



Sorry for delay (lot of things after my return from Ukraine)
I will comment both on my previous suggestion and that of Pat:

What I wanted to say is that in many places in optical spectroscopy there 
is a need to express the observable not measured as the instrument gives 
but normalized (divided) by some hypothetical function, that (due good 
reasons;-) we want to have some physical meaning - in our case the 
continuum flux (outside of spectral lines). The continuum is however not
easily seen - so only some kind of approximation is introduced - called 
"pseudocontinuum" - to know the real continuum a theoretical model ,which 
we know only in simple cases :-(  , must be used.

In fact all the Pat's examples seems to express the same problem:
If you start some philosophical speculations, in fact, every measurement - 
that is givin the value of observable is normalized in some sense to 
some reference function (or constant) - e.g. to measure distance you have 
to divide by constant representing the lenght of 1 m (the Pt-Ir rod in 
Sevres ;-) . If you measure SED you have to divide by the SED of standard 
star observed in same condition etc ...

AS I understand the radio examples - they have some function representing 
flux out of the source, some complicated function of averaged intensity 
in given beam as a reference (e.g. antenna temperature).


> You're missing "scaled counts".... You could argue that they are the same
> [as ADU] from a philosophic point of view, but to an astronomer, they mean
> something different.

Scaled count again is something "normalized" by given function.

Example of Francoise - the Halpha filter imaging of solar chromosphere is 
in fact the same like ordinary spectrum - in the data cube we (optical 
spectroscopists) are thinking as having one spatial point studied along 
the spectral axis - this imaging is in fact shifting whole plane along 
spectral axis - but in every point of a prominence you can imagine 
frequency shifting (if you tune the filter by changing its temperature - 
you are scanning the spectral axis).

So the conclusion - I think the crucial point in preparing the model is to 
define exactly the process of normalization - i.e. we have to describe 
formally the nature of the "normalizing function" - then all observables 
may be expressed as a ratio (or in general arithmetic function) of some 
"abstract value of the given variable - which we do not know ;-) " and the 
proper normalizing function . stating this I would be able to ask the 
tables (or registries) for things like

flux normalized by space angle
flux per frequency bin
flux relative to continuum
flux relative to radio sky background "
etc

I think there has to be first prepared such a formalismus
it will solve the purpose of data provenance as well (reduction pipelines 
are doing some normalization as well. - in fact all spectra are first 2D 
influence by instrumental characteristics and then normalized by function 
of flat field and other functions (echelle blaze) to yield the resutl 
after extraction ...


>From the philosophical point there is nothing like the "pure - ideal 
observable" - this is what we want to get using complicated theories .
We have only observable in Bayesian sense - the value of flux in X  is 
such provided we use such a method and detector .....

I understand we need some practical rule to encode in XML -
Currently I am not able to formulate this concept alone - we need to 
involve more general discussion of instrument specialists and 
theoreticians as well.

Lets invite them to collaborate - not only let all on DM group

P. SKODA

*************************************************************************
*  Petr Skoda                         Phone : +420-323-649201, ext. 361 *
*  Stellar Department                         +420-323-620361           *
*  Astronomical Institute AS CR       Fax   : +420-323-620250           *
*  251 65 Ondrejov                    e-mail: skoda at sunstel.asu.cas.cz  *
*  Czech Republic                                                       *
*************************************************************************



More information about the dm mailing list