[Spice_discussion] Delta_t from Earth PCK?

Nat Bachman nathaniel.bachman at jpl.nasa.gov
Thu Jun 18 19:30:49 PDT 2015

Hi Ernie,

It's good we're having this discussion. :)

In a word, in the long term, LOW ACCURACY predict earth PCK file


the value of




after July 2007.

It's pointless to try to recover UT1-TAI from the predict region of
the file.

We (NAIF) have no ability to predict UT1-TAI. I've asked around the
JPL Navigation section; nobody is willing to offer a means of
predicting this quantity, even at a low level of accuracy.

The best NAIF can do is update the long term predict file to use
available data up to the current date.

The fact that you're getting good agreement with others' calculations
is notable, but it's not because the long term PCK is accurate.
I'd be cautious about getting a false sense of security from
the agreement of the predictions.

As you say, the time conversion routines you're using depend
only on the leapseconds kernel.

Best regards,


On 06/18/15 19:06, Ernie Wright wrote:
> Hi Nat.
> Thanks for replying!  I should've explained what I'm trying to do.  I've
> calculated the circumstances of the 21 Aug 2017 solar eclipse using
> earth_070425_370426_predict.bpc and I'd like to get some idea of the
> delta_t (ET - UT1) that's implicit in that kernel for that date, for a
> couple of reasons.
> Your comments in the kernel say that beyond the end of the EOP's
> coverage (July 2007), some things including UT1-TAI are "held constant."
>    Based on the really good agreement I'm getting with eclipse
> calculations others have done, I'm guessing that doesn't literally mean
> that the implicit delta_t is frozen at the 2007 value.  Or I got lucky
> and made two mistakes that cancel each other out.  Anyway, I wanted to
> check.
> The other reason is that I'd like to be able to tell people what delta_t
> I used.
> I'm willing to do the work necessary to back out the polar motion if
> that's necessary, but I don't need high accuracy.
> For ET <-> UTC, I've been making good use of ET2UTC and TIMOUT.  My
> understanding is that those are independent of the PCK, depending only
> on the leap seconds kernel.
> - Ernie
> On 06/18/2015 07:43 PM, Nat Bachman wrote:
>> Hi Ernie,
>> Presuming by delta_t you mean
>>      TT - UTC
>> you can get this quantity by calling the SPICE routines
>>      DELTET
>>      UNITIM
>> DELTET will give you
>>      TDB - UTC
>> and UNITIM can convert TDB to TT (called TDT in
>> SPICE documentation) and vice versa.
>> A restriction on the above approach is that DELTET will
>> not give you correct results for times during leapseconds.
>> The conversion performed by DELTET is accurate to about
>> 40 microseconds.
>> SPICE doesn't provide a way to obtain UT1 from
>> a time specified in another time system. Trying to recover it
>> from the high-precision earth PCK would require you to model
>> polar motion and back that out from the orientation of the
>> ITRF93 frame relative to the ICRF.
>> Best regards,
>>     -Nat Bachman (JPL/NAIF)
>> Nathaniel.Bachman at jpl.nasa.gov
>> On 06/18/15 14:22, Ernie Wright wrote:
>>> It seems like I should be able to infer a delta_t from any of the binary
>>> Earth PCKs, but I can't seem to wrap my head around how to do it.
>>> If I PXFORM from Earth-fixed (supplied by the PCK) to true equinox and
>>> epoch of date equatorial coordinates, I should be getting something like
>>> UT1, I think.  (The TOD frame should cancel out the precession and
>>> nutation built into the Earth orientation, leaving the diurnal rotation.)
>>> I have the vague notion that the next step would be to subtract the "TAI
>>> rotation", but I really don't know what that means.
>>> - Ernie
>> _______________________________________________
>> Spice_discussion mailing list
>> Spice_discussion at naif.jpl.nasa.gov
>> https://naif.jpl.nasa.gov/mailman/listinfo/spice_discussion

More information about the Spice_discussion mailing list