Thanks for the response, Peter.
Sorry if I was a bit cryptic. The notion is to take various measures
that have been acquired at a series of points on the ground, using
traditional methods of land surveying, such as angles between and
distances to other points in the survey network or to points of known
location, and to use these data to compute the location of each
point. Typically, the computation looks like an overconstrained
system of linear equations, and the output of the computation is the
set of coordinates for each point that minimizes the sum of the
squared distances between the location for each point and the
locations as determined by the various combinations of equations that
give a single, deterministic result for each point. (Was that
obscure enough? The idea is that there are typically several sets of
equations that will produce a location for each point, but the
results are not consistent: depending upon which equations you use,
you usually come up with a different location for each point. This
is because the equations come from measurements that contain a
certain amount of error.)
I believe that there are many libraries that include one or more
functions for least-squares fits to overconstrained systems of
equations. What I have not found anywhere (yet) is open-source code
that does the necessary data handling, i.e., that accepts the
measurements taken at each point, converts them into a form suitable
for doing the least-squares fit, and then provides output showing the
coordinates for each point in the network, and the associated error
(based on the difference between overall computed location and the
various inconsistent, "partial" results for that point). This can be
quite cumbersome if you do it by hand, and in general, the problem
grows faster than linearly as the number of points increases.
FYI, my application involves old data (definitely pre-GPS) from
biological field surveys that relied on bearings taken with a
compass, and often, from a combination of pacing and tape
measurements between points. I anticipate a need for further
hacking, which is one reason I'm looking for an open-source solution.
There definitely are some similarities between my requirements and
that of various visualization technologies, and in fact, I have found
some open-source solutions for the latter. They don't quite fit my
needs, but perhaps at some point, one will. Did not know/recall that
the X3D Earth folks had an implementation yet, but I will inquire.
Were you among the people at the X3D face-to-face at Ames that I
attended? Apologies - I just don't recall.
Thanks again for the message and the request for clarification. In
the mean time, I'll be happy to hear from anyone who has a good lead.
Flip
On Apr 21, 2007, at 1:52 AM, Peter Strømberg wrote:
Flip, could you describe what solving a survey network entails and
how it would appear visually?
I've just returned from the second meeting og X3D Earth working
group who have big amibitions on
the 3D visualisation side and an open specification and an
excellent Java open source implementation.
Please tell us more...
Cheers
Peter Strømberg
www.hypergis.com
From: [EMAIL PROTECTED] [mailto:geowanking-
[EMAIL PROTECTED] On Behalf Of Phillip C. Dibner
Sent: 21 April 2007 00:45
To: [email protected]
Subject: [Geowanking] Open-source least squares network solution?
Hello all. I have been looking for some open source code that
solves an arbitrary survey network in 2D. Least squares solution,
with the ability to weight points, would be ideal. 3D wouldn't hurt
either, but isn't necessary.
I've been looking for a while, and found nothing that suits, or
even comes close. Any pointers? This isn't readily available
through any of the "well-known" packages like GeoTools, GRASS,
PostGIS etc., is it?
Thanks,
Flip
Phillip C. Dibner
Ecosystem Associates
+1 650 948 3537
+1 650 948 7895 Fax
_______________________________________________
Geowanking mailing list
[email protected]
http://lists.burri.to/mailman/listinfo/geowanking
Phillip C. Dibner
Ecosystem Associates
+1 650 948 3537
+1 650 948 7895 Fax
_______________________________________________
Geowanking mailing list
[email protected]
http://lists.burri.to/mailman/listinfo/geowanking
