Doug,
I'm in complete agreement with all the excellent points you make! A
GPS-derived bearing can't be accurate to 0.1 arcsecond -- that's absurd. I
just meant that the latitudes and longitudes of an accurate GPS fix agree with
Google Earth's coordinates (on the WGS84 datum) to roughly that level, in my
experience.
For several years I've been trying to determine azimuths of distant landmarks
as seen from choice (unobstructed) observing locations. So if a neat planetary
conjunction, comet, eclipse, thin crescent Moon, etc., is going to be visible
low in the sky on a certain date, I can use my "azimuth notebook" to pick a
spot, in advance of traveling there, where a certain landmark (lighthouse,
distant cliff, or tall building) will be correctly located to be included in a
dramatic photograph of the event. Using spherical trigonometry and the GPS
coordinates of the end points, an azimuth calculated using spherical trig
should be accurate to better than 0.2 degree or so, since the landmarks are
usually at least a half mile away.
If you are trying to get the azimuth of the wall of a building, then an
astronomical method (such as measuring the Sun's shadow on the wall) should be
much better than looking at a fuzzy image on Google Earth -- or the GPS
coordinates of the building's corners!
You're right about the danger of relying on a printed map, where grid north may
differ from true north. This seems to be especially true of "plot plans"
prepared by land surveyors of house lots, at least in the USA.
-- Roger
At 09:49 AM 1/27/2007 +0000, Douglas Bateman wrote:
>Dear Roger,
>
>I have been following this thread for some time when questions were
>raised last year about finding north-south.
>
>Like John C, with my own house as the test object, I have used plumb
>lines, slot-in-a-card methods, and large scale plans. The latter uses
>our acclaimed Ordnance Survey, but even at the 1:1250 scale the house
>on the plan is only 12mm long. Against the grid system and correcting
>for geographic north ("convergence") I obtained 9.5 deg west of north
>(bearing 350.5 deg).
>
>Google map gave me 8.5 deg with a printed image length of 25mm, and
>aligning against the eaves. I did another print out and used the ridge
>line, and got 9.0 deg. To me, this stresses the point that the way the
>light and shadows fall can change the appearance and choice of best fit
>for the protractor. Further enlargement only brings more blurring. As
>an aside, I am lucky with my Google map - a mere 10km away the area has
>yet to be photographed at the customary high resolution and one can
>barely make out the streets, let alone buildings.
>
>Pursuing optical methods with an old theodolite on the sun and Polaris
>(at its transit) I obtained 9.28 and 9.20 deg. I am still working on
>this method, which requires more practice to eliminate 'operator
>error'.
>
>I am therefore interested in the GPS method and how to obtain the
>extraordinary precision of 0.1 arc sec. For example, I cannot believe
>that a single hand held device can be pointed to this accuracy, and
>what I have seen of the screens of such receivers, the compass effect
>is crude. Similarly, what base line is used to get the high precision?
> If we take a building 100 feet long and the corners can only be found
>to within 10 feet, this is not much better than finding the moss on the
>north side of a tree. I jest, because if the readings at each end are
>taken close together in time, then the same cluster of satellites will
>be in view, then the RELATIVE positions should be found to, say 10,
>times the accuracy. Even so, this gives a bearing from one end to the
>other to an order a degree. Incidentally I did some experiments with a
>very good magnetic compass that could be read to 0.2 deg; with care and
>the current (website derived) deviation you can achieve better that 0.5
>deg.
>
>What then, is the secret of such alignment precision, and neglecting
>survey GPS equipment or differential GPS against precise pre-surveyed
>locations?
>
>Or, am I misunderstanding the point being made, for I do assume that,
>at worst, the global Google 'grid' converges on the sub-north and south
>polar points to within a few metres.
>
>Doug
>
---------------------------------------------------
https://lists.uni-koeln.de/mailman/listinfo/sundial
