Here's a tale of two surveys, both made recently and both reporting bearings and distances between a number of monumented points that remain in place on the ground.
I computed coordinates of the monuments in an arbitrary system using the bearings and distances reported by the prior surveys and then separately transformed each set of coordinates (translation, scale, and rotation of arbitrary system) using accurate coordinates in the Texas Coordinate System of 1983 as determined by me.
The standard errors in N and E components of my set of coordinates were all on the order of +/-0.015 ft. with respect to the local segment of the CORS network.
I used those control values to compute the parameters of a Helmert tranformation that translates, rotates, and scales the record coordinates in an arbitrary system to best fit the actual coordinates of the markers.
Survey No. 1 (Made in 2003)
Five markers found distributed along approximately half a mile of boundary
[pre]
Marker Residual (ft.)
1217 0.04 ft
1218 0.15 ft.
1219 0.14 ft.
1220 0.04 ft.
1123 0.01 ft.
[/pre]
The residual errors are quasi-random. The best fit transformation required a rotation by an angle that was within about a minute of the convergence (the record bearings were on average within about a minute of true) and scaling the record distances by only -15ppm to agree on average with actual surface distances
Survey No. 2 (Made in 2013)
Eight markers found distributed along approximately half a mile of boundary
[pre]
Marker Residual
1262 0.28 ft.
1248 0.21
1249 0.37
1250 0.16
1251 0.47
1253 0.21
1254 0.10
1255 0.05
[/pre]
The residual errors appear to be random. The best fit transformation required a rotation of only 0°00'13" (although not stated in the record description, the bearing basis was very close to Grid North of the Texas Coordinate System of 1983) and scaling the distances by +55ppm to best agree with actual surface distances.
So, which of these surveys is more likely conventional and which is more likely RTK?
Both are probably the results of RTK.
In East Texas, I'd remove 1251 from the transformation as an outlier (I'd also consider 1249 as a possible outlier). Likely the residuals will drop substantially for the remaining points. I'd also use a rigid body transformation (scale factor of 1) as opposed to the overdetermined (scaled) transformation you are using because it is unlikely (or should be unlikely) that these surveyors are using a different definition of the foot than you are.
Without more information we can't compute the CSF for either project. At the same time both scale factors are possible relationships between grid and ground. He is also following apparently published surveys, which would undoubtedly be at ground. As Kent is at SPC Grid with his ties they essentially are using a 'different foot'...
I wouldn't trust the transformation to ferret that out. I'd try either 1 or the SPC CSF for the project area.
If the conventional survey was ran around the perimeter, adjusted without a Networked Least Squares including cross ties, I would vote for Survey 2. This would be based on good procedures being used for both methods.
I see folks do it both ways. My preference is either LDP or pre-computed CSF. Too many run-ins with botched maps on a bogus scale or left on SPC Grid by accident...
> Without more information we can't compute the CSF for either project. At the same time both scale factors are possible relationships between grid and ground. He is also following apparently published surveys, which would undoubtedly be at ground. As Kent is at SPC Grid with his ties they essentially are using a 'different foot'...
Actually, what I did was to transform a set of coordinates to best fit the control values and then compared the surface distances between the transformed coordinates to those reported by the prior surveys.
The point of mentioning the scale factor is that one of the surveys has an apparent scale error that is within a range where a conventional survey would be expected to fall and the other is wacky.
> Both are probably the results of RTK.
I don't think that Survey No. 1 was RTK. That line has too much tree cover to be suitable for GPS occupations. I think it was done entirely conventionally. The approximation of true bearings is from having used a survey of an adjacent property made about twenty years before for the bearing basis. That earlier work was probably oriented from solar observations, I'd think.
> So, which of these surveys is more likely conventional and which is more likely RTK?
I'm sure you point is that the sloppier survey was done with RTK, and you are very probably right. But I assure you that much better results than that can be had with RTK if it is used properly.
> I see folks do it both ways. My preference is either LDP or pre-computed CSF. Too many run-ins with botched maps on a bogus scale or left on SPC Grid by accident...
That's not what I meant. Regarding the SPCS CSF and the 1 as the scale factor, I meant specifically regarding the transformation. Either the distances were SPCS grid or they were ground. The first survey likely was ground, and the second could be either. I would guess maybe Kent knows from the surveys he is working from. I wouldn't just let the Helmert transformation determine the scale factor - that was my point. I'd use a rigidly defined factor of 1 or the CSF and I'd look to see which resulted in lower residuals. I might also look to see if the reciprocal of the CSF worked better (perhaps the CSF was misapplied).
> In East Texas, I'd remove 1251 from the transformation as an outlier (I'd also consider 1249 as a possible outlier). Likely the residuals will drop substantially for the remaining points. I'd also use a rigid body transformation (scale factor of 1) as opposed to the overdetermined (scaled) transformation you are using because it is unlikely (or should be unlikely) that these surveyors are using a different definition of the foot than you are.
Actually, removing 1251 from the transformation does virtually nothing to the other residuals. They are just a hot noisy mess.
As for doing a transformation with SF=1.000000, that will give even larger residuals and since it is RTK it isn't a bright idea anyway. The folks who use RTK tend to use odd scale factors when turning out a work product and the best way to figure out what scale factor relates their work to reality is to solve it as a transformation parameter.
Your blanket statements about RTK and people who use it are terrible. Put your money where your mouth is and activate the RTK on your 5700 and see what it can and can't do.
> > So, which of these surveys is more likely conventional and which is more likely RTK?
> I'm sure you point is that the sloppier survey was done with RTK, and you are very probably right. But I assure you that much better results than that can be had with RTK if it is used properly.
The question that interests me is why all the RTK work I follow is crap. Right now, I think the answer is that it gets both pushed into areas where a person wouldn't want to employ GPS at all and is used in the fastest manner possible, without any of that time-consuming checking.
> Your blanket statements about RTK and people who use it are terrible. Put your money where your mouth is and activate the RTK on your 5700 and see what it can and can't do.
Actually, what I posted was:
"The folks who use RTK tend to use odd scale factors when turning out a work product and the best way to figure out what scale factor relates their work to reality is to solve it as a transformation parameter."
That is a correct description of what I see. The scale factors are seldom mentioned and the only way to determine them is to actually compute the scale parameter by transformation as I have done here.
> > >
>
> The question that interests me is why all the RTK work I follow is crap. Right now, I think the answer is that it gets both pushed into areas where a person wouldn't want to employ GPS at all and is used in the fastest manner possible, without any of that time-consuming checking.
So all of the work you follow is crap? Or is it just the crap work that you "suspect" is RTK? Are you saying that you don't have at least one peer in the State of Texas, who uses RTK, that you don't mind following their work? Are you sure that all of the work that you follow that is good, isn't RTK? Tsk Tsk.
> The question that interests me is why all the RTK work I follow is crap.
I detailed in another thread this morning what had been done on a job which I jumped into in Oklahoma. Many, many surveyors have swallowed the salesman's schtick hook line and sinker. It's a magic coordinate generator. "Oh, they're real all right. And they're spectacular".
> > Your blanket statements about RTK and people who use it are terrible. Put your money where your mouth is and activate the RTK on your 5700 and see what it can and can't do.
>
> Actually, what I posted was:
>
> "The folks who use RTK tend to use odd scale factors when turning out a work product and the best way to figure out what scale factor relates their work to reality is to solve it as a transformation parameter."
>
> That is a correct description of what I see. The scale factors are seldom mentioned and the only way to determine them is to actually compute the scale parameter by transformation as I have done here.
Actually, you're wrong again. A good many people that I follow around here leave it on the grid and use RTK well. Maybe you just follow crap because that's what the meat grinders in the metro-mess' tend to turn our, regardless of the type of equipment used.
So, what about it? Are you gonna activate the RTK or just continue to bash people who use it? Put your money where your mouth is big boy and test drive it. Hell, you at least did that with OPUS-RS and determined that you found it lacking in certain areas. Folks listened when you said that, yet you seem to be deaf in a sea of people who say contrary to you.
That would be the thread below?
Man, I must be blessed to live in the country. When we follow those who use GPS, we routinely hit within 0.02'. A company in Tyler set control for a highway project and we were doing a Brownsfield project and checked their control against ours. It was flat!
Shawn and J.D. Billings set some corners we tied to in an adjoining county. Holy crap it was close. I'm not talking the magical 0.04, no. I mean I was looking in the third decimal place to find the small change.
FWIW, they controlled that project with static (if I remember correctly) and we tied in their point with RTK using proven procedures. We were literally within the same dimple.
