Going on with observation model
bonnarel at alinda.u-strasbg.fr
Wed Sep 26 08:07:51 PDT 2007
While characterization model is in the last steps of the recommendation process
why don't we take a break?
While building characterization version 1, we faced a lot of times use cases
including descriptions of observations which were falling outside the scope of
our original limited goals in this model.
If IVOA standards are to be usefull we have now to rapidly take into account
these use cases, prioritarize them and start to adress them. Some still belong
to the characterization effort at detailed levels . Some others will address
only the Provenance description of observations.
1 Spectroscopic and Photometric data for SED building and cross calibration
Apertures, Beam (or PSF) are needed for comparison.
as well as Photometric calibration in the form of transmission curve of the
filters and zero points
2 Transmission curves in general will be also very usefull for characterizing
spectral and flux axes of 2D images, or catalogs seen as "datasets" either to get
good photometry of extracted sources or surface photometry.
3 Many datasets can be seen as compositions of simple observations ( multi CCD cameras,
IFU) But both the characterization of the whole and of the component may be
necessary. Resolution may change from one block to the other one.
An IFU may have varying spectral range and resolution on the field, we may need
to characterize that, as well as keeping the spatial and temporal characterization of
the whole IFU in common
4 the Dataset we consider is generally a moment in a process of transformation from
very raw data to high level extracted information or visualisation.
For quality assesment it could be usefull to get raw data and calibartion data
linked to the reduced dataset in order to make some comparisons or do some kind
of reprocessing. Description of the programs is also part of this
4 bis A similar case is the stacking of reduced data in one "association" to improve
the SNR or remove artifacts. links between the stack, the members and a
description of the stacking algorithm is usefull.
4 ter Dataset transformation may also be needed to provide convenient visualizations
usefull for data discovery or analysis
Again these visualizations have to be linked to the transformation description as well
as to the progenitor dataset.
WCS information is generally conveyed with the data using Pre-VO technology such
as FITS keywords. It is probably usefull to be able to describe this kind of
mapping observation within the frame of a general Observation datamodel.
6 A full understanding of the quality of the data and of the problem occuring
during the reduction phase probably requires a knowledge of observing atmospheric
conditions and instrumental configuration (light pass). This is also a usefull
part of provenance.
Use case 1,2 3 belong to what would be called in IVOA vocabulary enhacement of
charcaterisation model by better specifiaction of level 4 or characterisation of
complex data. Use case 4 and 6 and maybe 5 belongs more to what is called by many
people "Provenance". of course this is not and cannot be exhaustive.
We have to start reviewing exsiting modelling of such aspects: (original IVOA observation data
models, the small Acces and Views datamodel, Pat Dowler and G Lemson domain model,
the new CADC Observation model, plus more specific models such as Radio data models, etc.)
and decide a strategy of reuse and development to perform this next step of our modelling effort
Many of these use cases will also have consequeneces ifor the definition of new
generation DAL protocols such as SIA2, TAP and Complex data Access.
Francois Bonnarel Observatoire Astronomique de Strasbourg
CDS (Centre de donnees 11, rue de l'Universite
astronomiques de Strasbourg) F--67000 Strasbourg (France)
Tel: +33-(0)3 90 24 24 11 WWW: http://cdsweb.u-strasbg.fr/people/fb.html
Fax: +33-(0)3 90 24 24 25 E-mail: bonnarel at astro.u-strasbg.fr
More information about the dm