[Data Characterisation]: Spatial domain models comparison

[Alberto Micol]

>
>
>Dear all,
>
>I have a couple of immediate comments:
>
>0.) The suggested column does not always give enough detail on how it would be 
>done, e.g. Pat's way, Doug's way or yet another one?
>
Generally speaking, I reviewed the various suggestions trying to see 
which ones
are the quantities that should be  part of the Data Characterisation 
aspects of the DM.
I haven't gone much furhter than listing such quantities, a part from 
indicating a
first order approximation, what Doug calls the "summary dm", that is the 
layer
that will be immediately useful for eg SIAP and the like.
Further thoughts, ie real modelling, are absolutely needed.

>
>1.) How will the astrometric accuracy of the spatial coverage be represented 
>in a single floating point value? 
>
It's the first order approximation only. A query might ask data where the
astrometric accuracy is better than 0.5 arcsec; we need to summarise such
quantity with a single number for practicity.

>A real WCS contains many different 
>parameters which all have their own error associated and they can be quite 
>different. Moreover the parameters for the spatial coverage have different 
>units (fill_factor and shape are dimensionless, fov is probably in degrees.
>
I absolutely agree, we need to define much more precisely the errors 
associated
with all the other WCS related quantities.
 

>2.) What is a spatial sensitivity?? In my view the only way to represent 
>something like this is a sensitivity map as Alberto mentioned. The parameters 
>given in the table are certainly not representing anything like spatial 
>sensitivity, but global flux sensitivities deducted (how?) from the overall 
>FOV. 
>  The *how* is actually important here because if this is done differently 
>the values are not comparable. Note that I'm not against these parameters in 
>general I'm just saying that the name 'spatial sensitivity' should be 
>changed.
>  
>
Correct. A distinction between "absolute" and "relative" sensitivity is 
needed.
Absolute is to do with overall photometric accucary, while relative has 
to do with
pixel to pixel variations.

>3.) The concept name 'Photometric Accuracy' (first column) should also be 
>changed to flux uncertainity to be more consistent with the other concept 
>names in the same table.
>  
>
The concept was just meant to describe what the quantity should 
describe, hence
I considered it just a free text description.

>Cheers,
>Andreas
>
>  
>
Alberto