Dear Doug, Silvia, Elizabeth, Phil,
Doug Tody recently presented a very interesting draft document describing
access to 3-D/N-D Image Data in VOs
http://www.aoc.nrao.edu/~dtody/sia/3D-use-cases.pdf
Here are some suggestions for further consideration:
TIME SERIES The draft appeals for more exampls of data, especially multi-dimensional time-series data, which is why this is copied to people who are experts in this. I understand that one problem so far is the lack, in practical use, of standardised time formats or tools to convert between formats. I can't believ that it is any more difficult than providing tools for converting between the various bizarre celestial coordinates, just that no-one has done it yet.
Phil can correct/add more details but as I understand it, this is a time series of all-channel squashes from monitoring of SiO masers (i.e. axes RA, Dec, Time). The original data are actually a large number of observations at intervals of a few weeks, each of which have axes not only for RA, Dec, multi-channel Frequency, but also for polarization. The image datacubes can have axes (RA, Dec, Freq) in which the 'observable' is flux density in total intensity, linear or circular polarization, Stokes parameters etc. or polarization angle. They can also have axes (RA, Dec, Velocity). In this example there is a unique mapping from frequency to velocity but in similar cubes there can be two or more molecular transitions within one band, i.e. more than one possible velocity axis. One can also present cubes with axes of RA, Dec, Stokes parameters, which could be used either combined to give polarized intensity and angle, or to view I, Q, U or especially V separately.
3D VISUALISATION TopCat (www.starlink.ac.uk/topcat) now offers a number of 3D visualisation options for tabular data which recognises celestial coordinates etc. My feeling is that as we go to more or less than 2D data the distinction between tabular data and 'image' data is more and more blurred. On the one hand, many 1D spectra and time series are simply tables of freq (or time etc.) v. flux or other observable, possibly plus uncertainties. On the other hand, 3+D maser data or other observations of collections of compact sources are often presented as tables with headings something like TIME RA DEC Bmaj BMin BPA VEL Iflux Qflux UFlux Vflux (errors.. ...)
CHARACTERISATION
These data have at least 5 types of axes, some of which are themselves
multiple (Space, Time, Frequency, Velocity, Polarization). The problem is
that there is not yet any easy way to visualise or analyse more than a
selection at any one time. Thus where the 3D Acess Draft says 'in
practice data with more than 3 dimensions is rarely seen...' that is true
literally, because tools don;t exist to visualise it, but not true in
spirit, in that we do need to describe multi-dimensional data in order to
extract any desired combination of axes. The Characterisation model
(draft at
http://www.ivoa.net/internal/IVOA/IvoaDataModel/Characterisation_Note_20051129.pdf)
should allow this, so we need to make sure that it meets the needs of the
3D Access model - comments welcome to dm-at-ivoa.net
OTHER COMMENTS Other comments on 3D draft:
Para 2
I am not sure why we need _uniformly_ sampled data, since in some cases
(e.g. frequency/wavelength) this is only the case if you pick the right
units. In other cases - such as some of the time series examples given
above - the original data may be taken irregularly in time and the
extraction and visualisation tools may or may not interpolate onto a
regular grid. There is also the case of adaptively smoothed data.
As such tools do exist (or it may not always be necessary) I
suggest that we leave out 'uniformly'. The Characterisation model
provides (possibly too much!) detailed methods for coping with various
irregularities in sampling.
Access modes
Spectrum extraction - add - or an analogous extraction parallel to any other axis, e.g. to produce a light curve or variability with time.
We also need to consider extracting 3 dimensions from >3D data sets, including collapsing higher number axes - e.g. the collapse of each epoch's frequency axis in the maser movie example above.
The other common requirement (especially for ALMA!) is to convert velocity conventions and to generate a velocity access in response to a rest frequency and velocity convention.
Generic data set discovery
I will try and compare the proposals against some specific use cases and come up with words for this section and elswhere in the light of comments on what of the above is relevant/suitable at this stage.
best wishes
Anita