[Taxacom] Inappropriate accuracy of locality data
dlmcdonald at alaska.edu
Sat Dec 4 00:09:30 CST 2010
Since this is such an important topic and because there is a lot of
misinformation in this thread, I though I'd gather and report some actual
data. Please, don't take my word for any of this - grab your GPS and your
mapping program of choice and try it for yourself.
I found my mailbox on Google Earth, made sure my GPS (Garmin GPSmap96) was
set to WGS84, plunked it onto my mailbox, and saved three waypoints, labeled
One, Two, and Three, a minute or so apart. The first waypoint was created
within a minute or so of turning the GPS on, and the unit was last powered
up a couple states over. There was no attempt to let the GPS stabilize, and
I did not use the "create waypoint as average" function. Then I rounded the
coordinates to whole seconds, creating an additional waypoint named "Bob."
Mailbox Actual is the red cross, cleverly labeled "mailbox."
All three test points are within a couple meters of where Google thinks my
mailbox lives as the data came from the GPS. My GPS was reporting between 17
and 24 feet of error during this process. Had I been collecting mailboxes,
I'd have probably called it 10m error, just because that's a nice round
number that's larger than 24 feet. Had I rounded first, that error would not
include my mailbox - my data would be erroneous simply because I had some
strange idea about discarding perfectly usable data. There are four other
mailboxes closer than mine to Waypoint Bob, which clearly limits the
usability of the data for answering some questions (like "exactly which
mailbox are we talking about here?"). The average of my actual waypoints
would indeed be a new point, and there does seem to be a systematic bias, to
the west in this case. The waypoints are all still much closer to Mailbox
Actual than to the rounded point or any other mailbox. Your method of "For
GPS readings where the unit's declared 'accuracy' is less than 20 m, I
personally use 25 m" would, from the rounded point, entirely miss the actual
mailbox. Surely it's not hard to imagine more biological questions that
require the same sort of precision, nor to see the implications of
misrepresented data in attempting to answer those types of questions.
Now please tell me how this:
> 41 52 38.84 N 87 39 08.48 W +/- 25 m
> is different from this:
> 41 52 39 N 87 39 08 W +/- 25 m
> as an *estimate* of the true position.
They're different because they describe different points. A point 25m from
one could be more than 25m from the other. The red line in the above image
is 25m long, the yellow is 37. The following is a likely scenario.
* The GPS reports 41 52 38.84 N 87 39 08.48 W
* Actual error is ~25m
* The actual position of interest is ~25m SE of 41 52 38.84 N 87 39 08.48 W
* You round to 41 52 39 N 87 39 08 W, and use 25m as error
* You have just created an error shape that does not include the actual
point of interest.
At the very least, your report of error is inaccurate.
More information about the Taxacom