Precession, Nutation, Epoch, Time Scale

[Sergey K]

Hello,

I am still new to this field, sorry for any mistakes in advance....

My interest is in accurately finding pointing of a image pixel in absolute terms relative to local horizontal, where i know the lat,lon,alt,utc for my image.

My understanding is that all RA,Dec coordinates are moving with time (regardless of proper motion and parallax) because the RA,Dec coordinate system itself is moving (because it is a non-inertial frame, defined relative to The Earth). Most of the movement is due to precession (gravitational interaction of The Earth and The Sun "Precession" > The Moon "Nutation" >> planets "Planetary Nutation", etc).

Precession rate is changing with time but highly predictable and on average is somewhere in the order of 0.1 deg per 18 years.. ie significant for many applications.

So a RA,Dec coordinate is not complete/meaningful/accurate without also specifying the time/epoch, eg J2000.0.

(Note: refraction near horizontal at sea level is about 0.5 deg and can be accounted for to some degree with some basic assumptions, see wikipedia.)

1) The star catalogue here http://broiler.astrometry.net/~dstn/4200 and here http://broiler.astrometry.net/~dstn/4100 does not explicitly state the epoch, but I presume its J2000.0 and will be stated in the literature, somewhere. When I upload images to Astrometry.net it does not ask for image date and returned html does not state the epoch. I assume that Astrometry.net does not account for precession and returns RA,Dec in same epoch as that of the catalogue. It is then up to the user to apply precession to the solution (or the input catalogue) to get an accurate absolute RA,Dec for his particular date?

Should the epoch be stated explicitly in the star catalogue ReadMe, to remove ambiguity?

Should the epoch be explicitly stated with all RA,Dec results,... ie "RA,Dec,Epoch"?

Should there be disclaimers warning users about the need to correct for precession if they are using RA,Dec in an absolute sense, not just measuring relative angles on the image?

Does Astrometry.net have any code/tools to apply precession to a RA,Dec coordinate to propagate from one date to another?

2) I am still grappling with the different time scales (UT1, TAI, UTC, TT). I have seen some astronomers using TT (Terrestrial Time = TAI+32.184s) but I suspect that Astrometry.net like most others assumes UT1 (Universal Time corrected for polar wandering, which is not uniform and defined in terms of Earth's orientation and not the SI atomic second), is that right?

Should the star catalogue ReadMe and any results explicitly state the time scale used?

Should there be disclaimers warning users about the difference in time scales and need for conversions, if RA,Dec are to be used in absolute terms?

Does Astrometry.net provide any code/tools to convert between different time scales?

Note: difference between TT and UTC is 68.184s (2017-01-01) which is 0.28488 deg in RA (360/86164.1*68.184).

Note: I am a MatLab user, confined to Windows OS and I have only used the online version of astrometry.net, so far.

Best regards,
Serge

[Dustin Lang]

Hi,

You are correct, Astrometry.net returns results in the astrometric frame of the reference catalog, and the 4100- (Tycho-2) and 4200- (2MASS) series index files are both in  J2000.0.  As such, we have no need to deal with time systems.

For precessing coordinate systems you could check out astropy.  They also have a bunch of code to convert between time systems, eg
http://docs.astropy.org/en/stable/coordinates/transforming.html
https://github.com/astropy/astropy/blob/master/astropy/utils/iers/iers.py

I just added a paragraph to the README files in the index directories mentioning that they are both J2000.0

cheers,
--dustin

[Bryan]

Serge

An aside from your original question.

You have a couple options for running a local version of astrometry under Windows.  I highly recommend the local install.  It is faster, allows more control over index files and output, and doesn't depend upon having internet access.

1.  AstroTortilla (https://sourceforge.net/p/astrotortilla/home/Home/) - AstroTortilla (AT) is a Windows GUI wrapper on astrometry.  This installs cygwin, a Linux "box" in Windows and astrometry.net locally.  You can run astrometry through the AT GUI or in command line mode under cygwin.  There is a tutorial in the sourceforge Discussion sub-section.

2. If you have Win 10, then you can install a Linux Ubuntu distro (http://www.zdnet.com/article/how-to-run-run-the-native-ubuntu-desktop-on-windows-10/), and then install astrometry there.  

If you are new to Linux, then I recommend option 1.  

Bryan

[Eric SIBERT]

> You have a couple options for running a local version of astrometry
> under Windows.  I highly recommend the local install.  It is faster,
> allows more control over index files and output, and doesn't depend upon
> having internet access.
>
> 1.  AstroTortilla (https://sourceforge.net/p/astrotortilla/home/Home/) -

AstroTortilla is outdated concerning the astrometry.net install.

You would better install ansvr :
https://adgsoftware.com/ansvr/
It's part of Sequence Generator Pro but can be installed as a
standalone. It is also recommended by AstroTortilla's author.

Eric

[Bryan]

There is a thread on the AT discussion about how to upgrade to 0.80, using ansvr

Bryan

[walde...@gmail.com]

You could also check the NOVAS routines from the USNO- United States Naval Observatory, and the SOFA- Standards of Fundamental Astronomy routines from the IAU.  Their manual, samples, and user guides are great background reads.  A quick google will find them.