I've never come across this one before now. I can run blunder detect without a problem, but get a "geometric weakness" error when I adjust the network.
I'm adding files one at a time, and the one I've added is really short and simple, but not sure what's causing this. I've always thought that if Blunder Detect flies, so will the adjustment. Ideas?
Hard to say without seeing the network, but if I were to guess, I'd say some point needs another measurement to tie it down from huge variations. Do they all have both distance and angle ties to some point in the "good" network? Could it be an bearing or angle-only intersection that crosses at too shallow an angle?
Bill93, post: 407972, member: 87 wrote: Hard to say without seeing the network, but if I were to guess, I'd say some point needs another measurement to tie it down from huge variations. Do they all have both distance and angle ties to some point in the "good" network? Could it be an bearing or angle-only intersection that crosses at too shallow an angle?
I think you hit the nail on the head. Previously, I had created a Coordinates.dat file with approximate coordinates for all the points I'd observed to date. But then I added a bunch more, but didn't add them to the coordinate file. If I can remember how I got my ascii coordinate file re-formatted into a dat file, I'll update that and see what happens. Thanks.
Sometimes one too many "freed" observations or one too many "fixed" will do that.
Is there something you can change from free ("*") to a more specific standard error (i.e. coordinates to a foot or a tenth rather than free or no weight)? Likewise for anything fixed, can you give it a small standard error rather than leave it fixed? For example ".1 .1 .1" rather than "! ! !" for a 3D coordinate.
What are the largest observation changes in the blunder detect results? That's usually a place to start.
For bootstrapping coordinates into .dat format, there's an option under "other files" that will allow you to output the coordinate list in .dat format, then cut&paste that into your data file.
half bubble, post: 408118, member: 175 wrote: Sometimes one too many "freed" observations or one too many "fixed" will do that.
Is there something you can change from free ("*") to a more specific standard error (i.e. coordinates to a foot or a tenth rather than free or no weight)? Likewise for anything fixed, can you give it a small standard error rather than leave it fixed? For example ".1 .1 .1" rather than "! ! !" for a 3D coordinate.What are the largest observation changes in the blunder detect results? That's usually a place to start.
For bootstrapping coordinates into .dat format, there's an option under "other files" that will allow you to output the coordinate list in .dat format, then cut&paste that into your data file.
Thanks for those additional comments. I never realized that at the bottom of the output of Blunder Detect, there is a bunch of information that it reports, that help with clues to the problem.
In this case, I had traversed from my network several setups around to a distant point that was only tied to the network via that traverse. When I added another traverse via a different route to the same point (tying the entire loop together, everything came up roses. Still have a lot to learn about the program.
rfc, post: 408119, member: 8882 wrote: Still have a lot to learn about the program.
NO! You still have a lot to learn about surveying.
Paul in PA
I've had this issue before, and IIRC it had to do with incorrect angular observation(s) being imported for whatever reason. I had to clean them up manually and it was a bit of a hassle.
Paul in PA, post: 408141, member: 236 wrote: NO! You still have a lot to learn about surveying.
Paul in PA
We all do- surveying's a lifelong learning process.
The only superior evidence is that which you haven't yet found.
Paul in PA, post: 408141, member: 236 wrote: NO! You still have a lot to learn about surveying.
Paul in PA
Completely true indeed, but I'm unsure what the comment has to do with this particular question. If the implication is that I'm placing too much faith in Star*net as the be all and end all of surveying observations, I can emphatically say I don't. It's just one tool. It happens to be a tool that focuses on an area of surveying observations I'm fascinated with (the quantification of errors in a network), but it is just one small piece of what surveying involves. I certainly get that.
I could be erroneously reading between the lines, or overthinking the comment, so if that is so, please forgive me.
That said, I understand what geometric weakness is, how to quantify it's effects, and how to go about correcting in the field with additional observations. As it turns out here, Bill93 came close to hitting on the "problem", and half bubble's proposed solution probably would have worked as well as what did (simply adding additional observations already made to the network).