NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Almanac Calculation Questions
From: Bill Murdoch
Date: 1997 Mar 24, 12:43 EST
From: Bill Murdoch
Date: 1997 Mar 24, 12:43 EST
From: W. S. Murdoch, Polymers D&C (Kpt. 1027) Subject: Almanac Calculation Questions When I wrote my navigation program, I did not worry so much about the programming language. These were some of the issues I tackled. Why not just buy a $200 GPS and forget the celestial ? I have the GPS, but this is all a hobby and celestial is part of that hobby. Why not just use the Nautical Almanac and Sight Reduction Tables for Marine Navigation or just the Nautical Almanac and a simple trig calculator for the sight reduction ? My experience is that I often make mistakes and that I am lazy. When half sick, tired, and with other things to do, the navigation slips. A few keystrokes is easier for me than a quarter page of figures. Why do you want to write your own navigation programs when commercial versions are available and cheap ? There was no commercial program available for a TI-80 series calculator and to me a calculator seems a more "marine" bit of electronics than a PC. I wanted to learn how almanac calculations were done. I had the time to spare and could not do any real sailing at the time. I was in England and the library system was excellent. What is the proper balance of speed, program length, and accuracy, usability. and bells & whistles ? I wanted a 200 year almanac with no need to update any of the constants. I shot for a one sigma accuracy of 0.1' and achieved about 2 sigma. I kept the program length under 28K because that was the memory constraint. I made the input and output displays as logical as possible using about 10 of the 28K in input and output displays. Bells & whistles needed to include reductions to a lat/long fix both stationary and running. The running needed allowances for changes in both course and speed. Sunrise, sunset, phase of the moon, the almanac data in both N.A. and A.A. formats, unknown body identification, sight planning, Mercator and great circle sailings, Rude starfinder data, time- speed-distance, vector addition, current sailing, ..... This part is fun but uses 10k or more. What should be the source of the lunar and planetary almanac data ? Because of my accuracy needs, I did not use Duffet-Smith or Van Flandern. Because of the need for periodic updating, I did not use Yallop. That left Montenbruck and Meeus. Both are usable. Dr Yallop recommended Montenbruck and I found it somewhat shorter for the equal accuracy. I removed terms and shortened constants not needed for 0.1' accuracy. (I was aware of Emerson and copied his use of a Poisson rather than Fourier series for the perturbation terms to shorten the program length at the expense of calculational speed.) What should be the source of the stellar almanac data ? How many stars should be included ? Dibbons-Smith has J2000 positions in a nice compact format. It has a few mistakes that can be found by comparing his work to Hirchfield. The Astronomical Almanac gives a vector method of converting J2000 positions to epoch of date which is short and quick if some of the smaller corrections are ignored. 92 stars seemed enough. How do you make sure you have no mistakes ? This is a tough one; test and document and hope. Bill Murdoch