NavList:

   

Reply

We pretty much thrashed the topic of averaging a run of sights to death, but
we did not reference Dutton.  Just for completeness, Dutton's latest edition
discusses the problem in articles 3013 (Analysis of Uncertainties), 3014
(Use of Multiple LOP's), 3016 (Random Error) and 3018 (LOP Bisectors).

Article 3016, the most relevant, is short but much better than article 1609
in Bowditch.  Dutton's mentions these methods for coping with random error,
two of which I don't think were discussed yet, and a reminder about error in
the DR position used to plot the LOP:
1. Average the times and altitudes (well thrashed here).
2. Make 5 or more quick observations, note the azimuth with gyro, calculate
the rate of change of altitude per second of time = 0.25 arc minute x cos
Latitude x sine azimuth and use one of the sights that has a consistent rate
of change.
3. Make 3 or more quick observations, solve and plot all 3 LOP's, and assume
that the correct LOP lies midway between the 2 lines that lie closest
together.
4. Regardless of the method, draw an error circle around the DR or EP
position used to plot the celestial LOP, the radius being equal to the
navigator's best estimate of the inaccuracy of the DR or EP position.

In article 2244 they mention that STELLA has a drop-down field called
"Quality" that allows the navigator to "weight, statistically, a sight that
will be used, with other sights, in a fix solution.  Specify quality by
clicking the box and selecting 'Good', 'Average', or 'Poor' ... (based on)
... physical factors (such as) atmosphere steadiness, vessel motion, cloud
obscuration, etc.)".

As David Burch pointed out in his Starpath website article on using the
slope to select sights from a run, earlier editions of Bowditch had more
detail on this topic, but the latest edition only mentions it in passing
(article 1609).

Those of us beginners who are looking for clear, authoritative and practical
guidance on this small but very important and complex part of CN need to
separate wheat from chaff very carefully indeed.

Our generation does not have the adavantage of shipboard experience with day
to day CN that previous ones enjoyed.  If I was at sea in a yacht, had no
GPS and no weather window for celestial sights for several days, and then
managed to get short runs on 2 bodies from a bouncing deck on each of three
days apart, then I would certainly want simple reminders for how to deal
with random error.

I thank Alex for starting the thread, and everyone who contributed.  The
"Averaging" thread might be worthy of formal publication, if anyone is so
inclined.  I am still digesting the dense information posted in those 60+
messages, and in other Nav-L threads that discuss random error.  The Nav-L
archives contains a huge wealth of raw detail.

Jim Thompson
jim2@jimthompson.net
www.jimthompson.net
Outgoing mail scanned by Norton Antivirus
-----------------------------------------

   

Reply