Photometry in VO Spectrum Model

[Ed Shaya]

Let's review exactly what a magnitude is.  It is defined to be:

-2.5 log [F(target)/F(zero-point standard)]

where F is the flux in the relevant band and the zero-point standard is 
generally taken to be the flux of the star Vega.  This is the apparent 
magnitude, but the absolute magnitude is the same plus a distance 
modulus correction (5log(d)-5).
If you ban magnitudes then you either quote F(target) in Jy or other 
flux physical unit or you quote it as a ratio to Vega.  You can consider
the latter to be a unitless ratio or you can consider the units to be
"Vega Fluxes" in the given band or wavelength bin.  However, you will 
also have to change the error bars too.  Usually a single magnitude 
error can be associated with a given observation.  This is because the 
errors are fractional.  So, instead of quoting +/- errors, as is 
normally done, one has to transform to percentage errors. Additionally, 
extinction, A_b, is an additive term in magnitudes but a multiplicative 
term in linear form, etc.

Or you can accept magnitudes and simply provide metadata that informs
one that the numbers represent -2.5log of this flux ratio.  And, one 
must provide software to convert back and forth.

Computers can handle it either way, so, besides many historical 
precedences, it really comes down to the fact that as magnitudes, flux 
can be compressed into fixed decimal format, but in the linear form, it 
requires floating point to handle the typical ranges.  The fixed format 
takes up considerably less space in memory and particularly in the human 
readable table.  In the end the convenience of compressing a factor of 
100 into a difference of 5 does carry a noticeable advantage.

Ed

Matteo Guainazzi wrote:
> Dear Doug, Johnathan et al.
> 
> I believe that there are two main reasons not to ban magnitudes from the 
> VO realm:
> 
>   1. historical: magnitudes - like it or not - were, are and will be
>      widely used by observatories and surveys, and will continue to
>      appear in future VO-exposed catalogues
>   2. (more important) physical: the conversion from magnitudes to flux
>      (or from X-ray counts to fluxes) implies some assumptions on the
>      intrinsic source spectrum and foreground absorption corrections.
>      It is potentially misleading to leave the choice of this
>      assumption to the data providers only, a choice which may be
>      hidden in the metadata description. It is definitely better to
>      leave the user a certain control on these assumptions, especially
>      for non-stellar objects
> 
>  Once the filter transmission curve, and the zero point are specified, 
> converting from magnitude to fluxes for a given spectral shape is a safe 
> and deterministic algorithm.
> 
>  Hence, I believe that v1.0 of the Spectrum Data Model document should 
> be submitted as it is, and a specific group with the mandate of writing 
> a Photometry Data Model (intended in the broad sense as above) should be 
> set-up and start working as soon as possible.
> 
>  Regards, Matteo
> 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: eshaya.vcf
Type: text/x-vcard
Size: 257 bytes
Desc: not available
URL: <http://www.ivoa.net/pipermail/dm/attachments/20061024/f0187646/attachment-0001.vcf>