Time transformation

François Bonnarel francois.bonnarel at astro.unistra.fr
Tue Apr 7 17:51:46 CEST 2020


Hi all,

I agree that changing time representation (rendering) is not a transform 
in the sense of the model.

But changing the frame (time scale + refpos + direction) would be.

The discussion below ios about going from a "World" TimeFrame to another 
one.

But I see at least another use case : "movies" encoded as FITS Cube 
where axis 3 is representing time

These kind of things exist in solar physics as far as I know.

In that case a Transform FROM pixels in axis 3 to a World TimeFrame 
could be needed.

Cheers

François

Le 03/04/2020 à 21:10, Arnold Rots a écrit :
> Yes, the spacecraft-to-earth case is a valid transform.
> I only responded to the line I quoted.
>
> That said, there is more to it (and I realize I will be repeating 
> comments I have made before).
> Defining a time stamp using the (M)JD/UTC combination is a bad 
> (dangerous) idea; UTC measures an angle, it is not a chronometer.
> Earlier Transform designs focussed on spatial coordinates, emulating 
> the FITS WCS development in that area (WCS Paper II).
> Other simple coordinate types were automatically included (because it 
> was easy to do).
>
> Time is a different beast.
> There is the transformation from one time scale to another, with or 
> without a change in reference position.
> But, if done properly, the relative motion of these reference 
> positions should be taken into account.
> And then there is the option of applying a pathlength correction, 
> which is relevant when measuring photon arrival times.
> One may need to take gravitational potentials along the way into account.
> But some of this may not be relevant for transforming spacecraft time 
> stamps, since many spacecraft missions (at least the ones I am aware 
> of) artificially synchronize their spacecraft clocks with TAI on the 
> surface of the earth, thereby zeroing out all relativistic effects 
> (gravitational potential and acceleration) that the spacecraft clock 
> is subject to. In a way, if you want to be pedantic, it means that 
> effectively fundamental physics constants vary on board.
> All this to say that time transformations are a can of worms that is 
> best left to a specialized library. Leave it to David Berry.
>
> As to Mark's comment regarding the timing of FITS WCS Paper IV, if one 
> looks closely, one may realize that that paper is based on HEA time 
> conventions combined with the experience gained in the development of 
> STC1 - which, in turn, had some roots in the ISAIA program. And so, 
> the design of the temporal domain that I introduced in STC2 pretty 
> much continued what was contained in Paper IV. I admit that I have no 
> idea how much of that is left at this point.
>
> Cheers,
>
>   - Arnold
>
> Arnold H Rots
>
> Research Associate
>
> SAO/HEAD
>
> Center for Astrophysics | Harvard & Smithsonian
>
>
> Email: arots at cfa.harvard.edu <mailto:arots at cfa.harvard.edu>
>
> Office: +1 617 496 7701 | Cell: +1 617 721 6756
>
> 60 Garden Street | MS 69 | Cambridge, MA 02138 | USA
>
>
>
> cfa.harvard.edu <http://cfa.harvard.edu/>| Facebook 
> <http://cfa.harvard.edu/facebook>| Twitter 
> <http://cfa.harvard.edu/twitter>| YouTube 
> <http://cfa.harvard.edu/youtube>| Newsletter 
> <http://cfa.harvard.edu/newsletter>
>
>
>
> On Fri, Apr 3, 2020 at 11:51 AM Laurent MICHEL 
> <laurent.michel at astro.unistra.fr 
> <mailto:laurent.michel at astro.unistra.fr>> wrote:
>
>     Arnold ,
>
>     In my example, I was talking (a bit ahead) about transforming
>     spacecraft
>     time to earth time. The might be real a transform (change of
>     scale/ref)
>     isn't it?
>
>     LM
>
>     Le 03/04/2020 à 16:33, Arnold Rots a écrit :
>     >>>> You mean like  DATE => MJD?  GMT - PST?
>     >  > yes I do
>     >
>     > With all due respect, these are not transformations.
>     > They are properties (renderings, if you like) of the pure
>     coordinate
>     > value, without needing any coordinate frame information.
>     > If you want to transform a pixel coordinate to, say, a ICRS
>     equatorial
>     > coordinate, that is a true transformation, requiring information
>     about
>     > the relation between the two coordinate frames involved.
>     > On the other hand, if I have a time stamp value that is kept, for
>     > instance, in JD, I can ask it to be rendered in MJD or ISO-8601. No
>     > information external to that JD value is required, not even its
>     time scale.
>     > True time transformations would be required for changing the
>     time scale
>     > of a particular time instance, or its time frame's reference
>     position.
>     > But that is a much more involved subject.
>     >
>     > Cheers,
>     >
>     >    - Arnold
>     >
>     > Arnold H Rots
>     >
>     > Research Associate
>     >
>     > SAO/HEAD
>     >
>     > Center for Astrophysics | Harvard & Smithsonian
>     >
>     >
>     > Email: arots at cfa.harvard.edu <mailto:arots at cfa.harvard.edu>
>     <mailto:arots at cfa.harvard.edu <mailto:arots at cfa.harvard.edu>>
>     >
>     > Office: +1 617 496 7701 | Cell: +1 617 721 6756
>     >
>     > 60 Garden Street | MS 69 | Cambridge, MA 02138 | USA
>     >
>     >
>     >
>     > cfa.harvard.edu <http://cfa.harvard.edu>
>     <http://cfa.harvard.edu/>| Facebook
>     > <http://cfa.harvard.edu/facebook>| Twitter
>     > <http://cfa.harvard.edu/twitter>| YouTube
>     > <http://cfa.harvard.edu/youtube>| Newsletter
>     > <http://cfa.harvard.edu/newsletter>
>     >
>     >
>     >
>     > On Fri, Apr 3, 2020 at 5:04 AM Laurent MICHEL
>     > <laurent.michel at astro.unistra.fr
>     <mailto:laurent.michel at astro.unistra.fr>
>     > <mailto:laurent.michel at astro.unistra.fr
>     <mailto:laurent.michel at astro.unistra.fr>>> wrote:
>     >
>     >     Hello,
>     >
>     >     FITS paper II and III are listed in the draft (page 6) but
>     not FITS
>     >     paper IV (time representation).
>     >     It's a bit odd.
>     >     To me, time transformations are symmetric with e.g. spectral
>     coordinate
>     >     transformations, even if they are less used in FITS files.
>     >
>     >     Few comment below:
>     >
>     >     Le 02/04/2020 à 17:35, David Berry a écrit :
>     >      > On Thu, 2 Apr 2020 at 15:55, CresitelloDittmar, Mark
>     >      > <mdittmar at cfa.harvard.edu
>     <mailto:mdittmar at cfa.harvard.edu> <mailto:mdittmar at cfa.harvard.edu
>     <mailto:mdittmar at cfa.harvard.edu>>> wrote:
>     >      >>
>     >      >>
>     >      >> You mean like  DATE => MJD?  GMT - PST?
>     >     yes I do
>     >
>     >      >> I think these are in the same level as
>     >     ENERGY-FREQUENCY-WAVELENGTH, which are really standardized
>     >     transforms which are basically considered different forms of the
>     >     same value.  Which is why they are in the Coords model as
>     different
>     >     'flavors' of  Time coordinate rather than having a single time
>     >     coordinate and using transforms to convert.
>     >     You need a transform to go from a flavor to another.
>     >     If I want to use TRANSF to model e.g. a ground segment
>     processing, I
>     >     might have a step converting spacecraft time to earth time.
>     >     This is very similar with converting pixels to RA/DEC by
>     using attitude
>     >     data among other things or converting spectrometer channels
>     to kEv.
>     >
>     >      >
>     >      > I think it depends to what extent we want to limit the
>     use of the
>     >      > Transform model. Being able to describe the mapping
>     between different
>     >      > coordinate systems in a single physical domain seems like a
>     >      > reasonable goal for a Transform model. For instance, if I
>     have an
>     >      > image in which the WCS gives the (ra,dec) as a function
>     of pixel
>     >      > position, I may want to create a copy of that image that
>     gives the
>     >      > (l,b) of each pixel instead of (ra,dec). To do this I
>     would need to
>     >      > modify the WCS by tagging on a Mapping to convert
>     (ra,dec) to (l,b).
>     >      >
>     >      > Original WCS:   (pixel) -- mapping 1 -> (ra,dec)
>     >      > New WCS: (pixel) -- mapping 1 -> -- mapping 2 -> (l,b )
>     >      >
>     >      > i.e. "mapping_1" is the pixel to (ra,dec) mapping from
>     the original
>     >      > image, and "mapping 2" is the (ra,dec) -> (l,b) mapping.
>     >
>     >     Agree, but the time is one of these WCS domains (paper II
>     III and IV)
>     >     isn't it?
>     >
>     >
>     >      > Being able to modify a WCS so that it represents a
>     different physical
>     >      > coordinate system seems like a reasonable use-case. But
>     maybe one
>     >     that
>     >      > can be deferred to a later date.  The beauty of the sort
>     of system we
>     >      > are creating is that it is is easy to extend it. What
>     matters most is
>     >      > that we get the right definition of mappings, transforms,
>     operations,
>     >      > axes and so on.
>     >     Correct me if I'm wrong, there is no need to modify WCS to
>     deal with
>     >     time coordinates.
>     >
>     >     I think that time transformations should be part of the
>     model, also
>     >     because Time Data still are an high VO priority.
>     >     If it is not, e.g. because we urge to go in PR, this must be
>     justified
>     >     in section 2.
>     >
>     >     Cheers
>     >     Laurent
>     >
>     >     --
>     >     ---- Laurent MICHEL              Tel  (33 0) 3 68 85 24 37
>     >            Observatoire de Strasbourg  Fax  (33 0) 3 68 85 24 32
>     >            11 Rue de l'Universite      Mail
>     > laurent.michel at astro.unistra.fr
>     <mailto:laurent.michel at astro.unistra.fr>
>     <mailto:laurent.michel at astro.unistra.fr
>     <mailto:laurent.michel at astro.unistra.fr>>
>     >            67000 Strasbourg (France)   Web
>     http://astro.u-strasbg.fr/~michel
>     >     ---
>     >
>
>     -- 
>     ---- Laurent MICHEL              Tel  (33 0) 3 68 85 24 37
>           Observatoire de Strasbourg  Fax  (33 0) 3 68 85 24 32
>           11 Rue de l'Universite      Mail
>     laurent.michel at astro.unistra.fr
>     <mailto:laurent.michel at astro.unistra.fr>
>           67000 Strasbourg (France)   Web
>     http://astro.u-strasbg.fr/~michel
>     ---
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ivoa.net/pipermail/dm/attachments/20200407/778aed29/attachment-0001.html>


More information about the dm mailing list