about galaxy "velocity cubes"

[Petr Skoda]

Sorry I do not have large experience in radioastronomy but my subjective 
view is such :

All the aim of a physical observation is to get all the information in the 
ortogonal system of variables ( to simplify the detector and data 
acquisition) - I mean ortogonal in such a sense that you cannot express 
some variable as a function of another (suppose now Euclidian space and 
time not spacetime for the simplicity).  So we have defined the system of 
physical units as well. The question is how to get such a representation 
from measurebles (or Observables in ObsCore) that may be intermingled. 
This is a task of data reduction and interpretation. If I succeed (whole 
the know- how of given field is there even with some subjective 
contribution of method used) I consider the result worth of publication 
(say level 2 data - the instrument dependency is hidden in the 
ortogonalization process as well - say bias subtraction, flat-fielding etc 
).

So I can use the Char to describe the data and all of the level 3 products 
can be derived using only this information (by pure mathematical 
transformation of all data +  all Char (meta)data).

I will use the examle in optical spectroscopy - the primary information I 
have to measure in addition to the location (space) and time information 
is the wavelength (or frequency in radio, energy in X-ray and gamma 
etc...) - I understand the spectral axis exactly like this.

When I publish the spectrum with RV on X-direction ( horizontal) to show 
the movement of mass (e.g. stellar wind) relative to given wavelength - it 
is my arbitrary decision using another information (e.g. laboratory 
wavelenght of given spectral line I may not know it - I have only table of 
ThAr arc lines used in wavelength calibartion - or frequency normals in 
radio case) - but I can express the spectrum in RV values relative to 
another line.  This is just my choice and it does not change anything on 
the nature of the radiation.

Moreover the RV is derived using some principles (in our case Doppler 
law neglected in some terms - often the quadratic) (just a helpful 
estimate of a maximum size of a vector) but does not tell anything about 
the true velocity (the direction of movement is mostly unknown until I 
know the proper motion). Moreover the law itself is just dependent on this 
unknown direction (longitudal or transversal etc ...)

So I do not like the terms like Observable (or independent axis) for the 
variables derived using transformation with arbitrary parameters.

The spectrum with X-axis in RV  is dependent on a line frequency I have 
choosen and it is a derived product. The equivalent width is defined only 
for continuum normalized spectra and depends on a selection of a continuua 
(which may be difficult without  models) but it belongs to analysis part.
If I want to publish it (and maybe important for some research - e.g. 
terminal velocity of winds in hot stars) it is a science level 3 product.

In this respect I support Igor !

The published data stating they are result of observation after 
instrument signature removal (say level 2) should be  Irreducible ;-0

The same holds for Time series vs. Light curves.
The Observable is just intensity in time. But the derived product should 
include the estimated periods and epoc to allow the folding.
But it is Level 3 product.


> The problem with this data type is that the two *observable axes* (RA and 
> Dec) are mixed with the velocity axis (or redshift if you wish) which is a 
> data analysis result. My impression is that conceptually it is horribly 
> wrong.

is is not ortogonal in units

>If one is using radial velocity for the Z-axis, then he/she should be 
> using real spatial coordinates in kpc or other physical units (e.g. meters) 
> for the other two axes.

it is already level 3 implying some other information (e.g. distance) for 
transforming angles

and we still have the unknown direction but it is a JUST constant not 
changing the units (unitless cosines)

  On the other hand, if obervables are used (e.g. 
> RA/Dec, l/b or whatever) for coordinates, than the spectral coordinate 
> (wavelength/frequency/energy) has to be used for the 3rd dimension.

in this case we already have the independent primarily acquired 
information from the measurement -level 2


But sorry as I said I do knot know too much about radio methods and 
jargon

Cheers,

Petr

*************************************************************************
*  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                                                       *
*************************************************************************