I'm a total newbie to least squares adjustments, but just 'cuz I'm old, fat and feeble doesn't mean I can't learn new tricks.
I'm attempting to reproduce another guy's 100+ acre survey. I know from experience that his work is usually top shelf, at least as far as placing monumentation where his map says it's spoda be....
I've computed theoretical, if you will, coordinates for his corners. In the field, I located seven of his apparently undisturbed capped pins from a three-legged open traverse (I'm avoiding traversing across a ginormous gully.
In my .rw5 file, I inserted those theoretical coordinate values for those seven pins as control points numbered with the same point numbers assigned to my field located points.
In Carlson SurvNET, I assigned a standard error to those control points of 0.05' N & 0.05' E, and kept the same values I've always used for the remainder of the standard errors.
When I adjust the network, all of the found points, with one exception, lie within .04' (??!!) of their theoretical positions, and the report file shows:
Statistics
==========
Solution converged in 3 iterations
Degrees of freedom:13
Reference variance:0.97
Standard error unit weight: +/-0.98
Passed the Chi-Square test at the 95.00 significance level
5.009 <= 12.594 <= 24.736
Seems OK to me, but, then again, Colonel Hogan, I know nossing, so I'd like some input from the experts hereon!
Thanks!
SS
Sergeant Schultz, post: 344065, member: 315 wrote: I'm a total newbie to least squares adjustments, but just 'cuz I'm old, fat and feeble doesn't mean I can't learn new tricks.
I'm attempting to reproduce another guy's 100+ acre survey. I know from experience that his work is usually top shelf, at least as far as placing monumentation where his map says it's spoda be....
I've computed theoretical, if you will, coordinates for his corners. In the field, I located seven of his apparently undisturbed capped pins from a three-legged open traverse (I'm avoiding traversing across a ginormous gully.
In my .rw5 file, I inserted those theoretical coordinate values for those seven pins as control points numbered with the same point numbers assigned to my field located points.
In Carlson SurvNET, I assigned a standard error to those control points of 0.05' N & 0.05' E, and kept the same values I've always used for the remainder of the standard errors.
When I adjust the network, all of the found points, with one exception, lie within .04' (??!!) of their theoretical positions, and the report file shows:
Statistics
==========Solution converged in 3 iterations
Degrees of freedom:13
Reference variance:0.97
Standard error unit weight: +/-0.98
Passed the Chi-Square test at the 95.00 significance level
5.009 <= 12.594 <= 24.736Seems OK to me, but, then again, Colonel Hogan, I know nossing, so I'd like some input from the experts hereon!
Thanks!
SS
Why do you think you did anything "wrong"? Looks like it passed. Had the survey been previously adjusted? Do you have the original .rw5(s)? There may be more redundancy to put through Survnet if you do. I'm not sure using just a coordinate file with position uncertainties is as good as using multiple, redundant observations.
Looks good. I might suggest freeing up the one outlying coordinate and tightening down your standard errors on the others a little.
rfc, post: 344104, member: 8882 wrote: Why do you think you did anything "wrong"? Looks like it passed. Had the survey been previously adjusted? Do you have the original .rw5(s)? There may be more redundancy to put through Survnet if you do. I'm not sure using just a coordinate file with position uncertainties is as good as using multiple, redundant observations.
I don't necessarily think I did anything wrong, but when I think that - I'm usually wrong. I was just looking for validation, as my experience with least squares adjustments is limited. The practicalities of the survey (read budgetary constraints) precluded redundant measurements or closing a loop; so, in holding the other surveyors theoretical corners as control points, I was hoping to prove there were no blunders in my measurements, which apparently turned out to be the case.
Thanks!
SS
The only thing proven is that you and the other surveyor agree on the relationship of the points. Your conclusion that all is well comes from the assumption that such agreement would be highly unlikely if either of you had errors in your work. Back in the transit and tape days, on tough, rural surveys we used to "close on record" much more than we need or want to these days.
BajaOR, post: 345106, member: 9139 wrote: The only thing proven is that you and the other surveyor agree on the relationship of the points.
That's all I was trying to prove - am I missing something here?
To continue:
I don't see any diiference between control coordinates derived from another surveyor's published math and control coordinates obtained from any other source, like OPUS. After taking Norman O's advice and modifying the SEs for the control points, I got this:
Statistics
==========
Solution converged in 2 iterations
Degrees of freedom:15
Reference variance:1.00
Standard error unit Weight: +/-1.00
Passed the Chi-Square test at the 95.00 significance level
6.262 <= 15.021 <= 27.488
And......
Relative Error and ALTA Tolerances
==================================
Alta Tolerance Report, Specific Connections, 95% Confidence Interval
Actual Allowable Ratio
Sta. Sta. Dist. Semi Major Semi Major Actual/Allowable Semi Minor Max. Err. Az.
10 42 502.887 0.087 0.095 0.916 0.034 S 10-32'20.3"E
42 48 181.442 0.053 0.079 0.667 0.049 N 15-23'37.9"E
48 51 174.502 0.047 0.079 0.592 0.035 N 19-36'18.2"E
51 52 60.494 0.036 0.073 0.486 0.033 N 47-12'03.2"E
52 61 271.208 0.053 0.084 0.638 0.037 S 01-48'10.6"E
61 10 339.861 0.060 0.087 0.689 0.040 N 89-41'36.7"E
So, what's not to :love: ?
Best regards,
SS
That part of your report looks good. In fact, given your methodology of holding the record dimensions of another surveyor it looks quite good. But it is difficult to fully judge these things on the basis of any single part of the adjustment report. If you would like to post he whole thing, or email it to me ...(start a conversation, and I'll send you my address)
You said "I was hoping to prove there were no blunders in my measurements". I'm not arguing with why, what, or how you did what you did, but the word "prove" caught my eye because it implies certainty. Instead of certainty I think what you have is "highly likely" there are no blunders. Sorry to get overly technical on you. I need to get out of the office more....
By using Norman O's suggestion you assume your measurement to pin #7 is good, and the prior surveyor's reported location for pin #7 is bad, when it really could be the other way around. In adjusting the standard errors of pin #7 you are forcing something nearer a 1:1 measurement to expected relationship, which makes your statistics look good, but they might not be real. If you're confident in your measurements, I'd just remove pin #7 as control.
At the start of this adjustment you put what you thought were logical standard errors on the pins and you haven't done anything to prove that standard still doesn't apply to all of them. Your .05' x .05' actually seems too tight for a 100 acre parcel. By tightening up the standard errors on the 6 pins aren't you doing the same thing you did by loosening them up at pin#7, that is forcing something nearer a 1:1 measurement to expected relationship to make your statistics look better?
I might not have the least squares expert input you were looking for, but I do have questions! 🙂