NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Re: Astronomy DLL for .NET: SofaJpl 2.1.2 released
From: Paul Hirose
Date: 2018 Sep 9, 22:23 -0700
From: Paul Hirose
Date: 2018 Sep 9, 22:23 -0700
On 2018-08-29 17:30, I wrote: > Note that JPL's Horizons online calculator won't exactly agree with all > my numbers, mainly due to a different precession / nutation model. Since Horizons is a free and authoritative calculator of solar system body positions, I thought this discrepancy with respect to my SofaJpl demonstration program ought to be remedied. Therefore, the entire precession and nutation computation has been revised, including a Horizons "compatibility switch" in the source code. If turned on, the program is compiled with the IAU 1976 precession model. Then nutation angles are applied to put the pole in the same place as the default 2006/00B precession / nutation model. That yields equatorial and ecliptical coordinates within 1 mas (milli arc second) of Horizons. Azimuth and elevation are within .0001°, which is all the precision available from Horizons. (But for that accuracy you must set the UT1-UTC offset in the place provided in the source code.) The IAU 2006/00B model remains the default, and that's what I recommend unless you need compatibility with Horizons. It is true that you get the same pole in either mode, but the two coordinate systems have a small rotational difference about the pole. The 2006/00B model should better match the true direction of the equinox. For utmost accuracy you can substitute the IAU 2000A nutation model. Combined with IAU 2006 precession, it predicts the pole with an accuracy of about .2 mas in the current era. It's available in SofaJpl. In the demo program, simply substitute 'A' for 'B' in the name of the nutation routine. There's a major increase in computational complexity since the 'A' model evaluates about 1400 trigonometric terms. I can't imagine you'd notice a difference in a "one shot" run like the demo program, but nevertheless I see no need for such elaboration. The default 'B' model is accurate to "only" 1 mas, which is enough for me. And it's even smaller than the old IAU 1980 nutation model. Note that there has been NO change to SofaJpl itself. The current version is still 2.1.2. Only the C#, C++, and Visual Basic source code for the demonstration program have changed. http://home.earthlink.net/~s543t-24dst/SofaJpl_NET/index.html