NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Re: accuracy of automatic celestial navigation
From: Glenn Geers
Date: 2002 Dec 8, 09:08 +1100
From: Glenn Geers
Date: 2002 Dec 8, 09:08 +1100
On Sat, 7 Dec 2002 05:18 pm, you wrote: The algorithms used in STELLA have been published. The author is George Kaplan former head of the Astronomical Applications Department at USNO. I have the full set available (have to dig them out). The USNO will provide reprints of the papers that aren't available on the web. It's pretty neat stuff. The algorithms take the motion of the vessel into account and can provide very high (30m) accuracy. Also of interest: Polaris subs of the 60's carried an automatic astro-nav unit that they used to reset their inertial systems from time to time. Most auto systems use infrared optics so they can track stars during the day. Regards, Glenn > On Friday, December 6, 2002, at 04:05 PM, George Huxtable wrote: > > For those of us that sail our small craft out at sea, my opinion is > > that if we can achieve a precision of 2 min, we are doing pretty well. > > What do others think? > > I would agree. 2 nmi at sea is very decent; on land we should be able > to get to 1/2 mile or better. > > On land I recently took a string of shots using a Tamaya Jupiter > sextant with a 7X scope. I shot the sun mid afternoon over a period of > 20 minutes. (This was on the coast of Oregon at a beachhouse on the > water.) I used a Garmin GPS in averaging mode to determine my actual > position to 15 feet or so. The resulting intercepts in nmi are > interesting to analyze. Because my assumed position is known to within > 15 feet, these can be taken as errors in my sights. Here they are in > order: > > 0.91 > -0.06 > 0.59 > -0.05 > 0.22 > 0.55 > 0.22 > -0.28 > -0.53 > -0.57 > -0.77 > -0.03 > -0.19 > -0.11 > -0.43 > -0.62 > 0.27 > 0.24 > 0.44 > 0.25 > > The mean of the intercepts is a mere 0.0025 nmi -- very good -- but the > individual readings varied over a 1.68 nmi range -- not so good. > > When we are "warming up" and we haven't used our sextants in a while I > find that the first shots are often in greater error than later sights. > This is in fact the case with this set of shots shown above: the worst > error came from the very first sight. Therefore I have found that it > is wise throw out the the first reading or two, or at least to discount > them. > > Other experiments I have done: > > I took five shots in my backyard using the same sextant and an > artificial horizon and got a 1.27 nmi range of errors and a mean error > of 0.644 nmi. > > I took fourteen shots in my backyard using a Tamaya Venus (not nearly > as powerful of scope) and an artificial horizon and got a 4.3 nmi range > of errors but a mean error of -0.45 nmi because of the greater number > of sights taken. > > And for my finale, one afternoon I used six different sextants in my > backyard and I took a total of 46 sights using an artificial horizon. > I got an 11.2 nmi range of errors with a mean error of > -1.23 nmi. > > From this the statistical nature is shown to be very useful: although > any one shot could be in error up to almost a mile, a series of shots > averaged does indeed improve accuracy to very good levels; STELLA > appears to do this, and thus an automated system is very appealing. > > Dan