You have a nice coordinate that says something is located at somethingsomething northing and somethingsomething easting for such and such state plane and such and such zone. Or whatever the precise terminology happens to be for your area. Then you go set up and pin point where you think that spot is to be found. Over the next nine days, nine other surveyors go out and do the precise same thing. How tight of an ellipse, or whatever, should be expected that would enclose all ten answers?
[sarcasm]something something by something something[/sarcasm]
😀
2 meters or so n-s and 3 to 4 meters e-w...sometimes closer.
Not necessarily a Guru, but I have an answer
Much like the elderly's underwear: Depends. See all the threads on multipath and other factors that affect the magic GPS unit...
6 years ago a guy from another firm and I each took out our GPS units to Eastham, MA. He had a smart stick connected by cell to a base. I had a pair of units with UHF and a 35 watt repeater. I setup on a random point with static for an OPUS solution.
We went miles away and located points all over the National Seashore.
I had more trouble getting a fixed RTK solution than he did as we went out of radio range several times. He lost the cell connection once. I ran short static sessions using a bipod when we were out of range.
After processing mine and importing his, the common results were within 0.05' of each other. I thought this was going to go south and I would see some real differences between our data but did not.
Best part, they still had the beach open and were charging $20 for parking. I drove up to the gate, told the attendant in her green uniform that I was going to set a nail in the sidewalk and locate it. She asked, "Are you allowed to?" and I said "Yup!" and drove in.
Six months later the GIS guy from National Seashore called to see if I had any GPS control in the area. I said, "Yup!" and sent him the coordinate sheet.
> .. How tight of an ellipse, or whatever, should be expected that would enclose all ten answers?
Relative to some nearby known point a few hundreths. Relative to the CORS network probably something under two tenths. Assuming appropriate procedures.
> Over the next nine days, nine other surveyors go out and do the precise same thing. How tight of an ellipse, or whatever, should be expected that would enclose all ten answers?
It pretty much depends upon whether the other surveyors know what they're doing or not. If they don't, you'll see all sorts of idiotic stuff like the coordinate units not all being in US Survey Feet or International Feet or some RTK surveyor "localizing" off of some mysteriously chosen control points and locking in an oddball rotation and a scale factor from Mt. Everest.
You can ask pretty much the same question about any survey measurement technology. Take measurement with a steel tape. If the surveyors don't know what they're doing, you get slope distances reported and distances uncorrected for temperature. You might even get distances measured just by eyeballing the tape graduation over the ground mark, no plumb bob need apply.
> Over the next nine days, nine other surveyors go out and do the precise same thing. How tight of an ellipse, or whatever, should be expected that would enclose all.
Probably a better way of posing the problem would be what the values are by which all of the independently determined positions would not be expected to deviate from the weighted mean at the 95% confidence level.
Assuming competent surveyors truly using similar equipment and methods, the answer lies in what the standard errors of the N, E, and Up components of their separate determinations of the coordinate positions are. All of those standard errors should be quite nearly the same.
If you can determine the NAD83 coordinates of some point with, say, standard errors of 0.015 ft. in N and E components, and there isn't any reason why the other nine surveyors should not be able to do the same, then the answer is a straightforward statistical exercise. In the real world, just because nine other surveyors should be able to do the same doesn't mean that it will necessarily happen.
.04'
with all due respect to other's studies, in my experience 2 tenths on network RTK is about what we achieve around here. Mind you, my study is just myself revisiting multiple points on many different days, sometimes using the same equipment and sometimes using different. It can vary greatly across networks as in Raleigh, I can expect errors less than half that but on the coast we have more error. I can tighten that up to a few hundredths using static and normal procedures.
If proper techniques are used ....
this is probably a very realistic answer.
> It pretty much depends upon whether the other surveyors know what they're doing or not.
Everything relevant was just said in that line; yesterday I found out a former employee I worked with a few times was known to do 2 point calibrations (as I was told there were only 2 control points, apparently too lazy to take the total station out and push some control out so as to have a proper calibration).
Ignorance is the biggest accuracy issue with any piece of equipment or software, and sadly none of us are immune to its effects.
If proper techniques are used ....
I agree with you Dave.
Accounting for the the variable of the changing constellation and types of equipment, that is a justified guesstimation for horizontal..
The reason for my question
The reason for my question goes back to the thread I started a day earlier. The one asking for opinions on setting new survey corners based on controlling corners found or set during different surveys at different times in the past.
If you merely go to a coordinate that is supposed to fall on a straight line between two controlling corners, what is the probability that you are on that straight line and at the distance from them that you think you are?
If the point is being set by non-network RTK, the base could occupy different monuments, in which case the local accuracies of those monuments would have to be taken into consideration as well:
AA5772 FGDC Geospatial Positioning Accuracy Standards (95% confidence, cm)
AA5772 Type Horiz Ellip Dist(km)
AA5772 -------------------------------------------------------------------
AA5772 NETWORK 3.10 4.00
AA5772 -------------------------------------------------------------------
AA5772 MEDIAN LOCAL ACCURACY AND DIST (003 points) 2.92 3.49 3.18
AA5772 -------------------------------------------------------------------
(P.S. I am NOT a guru...)
Are "nice coordinates" the ones that don't bite you in the arse?
B-)
Pablo
:good: :good:
The reason for my question
> The reason for my question goes back to the thread I started a day earlier. The one asking for opinions on setting new survey corners based on controlling corners found or set during different surveys at different times in the past.
>
> If you merely go to a coordinate that is supposed to fall on a straight line between two controlling corners, what is the probability that you are on that straight line and at the distance from them that you think you are?
What are the coordinates of the controlling corners?
What is the classification of the survey as far as relative accuracy and position?
Are we allowed to ask questions too? 😉
> or some RTK surveyor "localizing" off of some mysteriously chosen control points and locking in an oddball rotation and a scale factor from Mt. Everest.
Ding Ding Ding Ding!!!!! We have a winner!
I can count many times where we have provided more than enough metadata for a project (C.F., zone, accurate ortho heights and so on) and we still get calls from other surveyors who can't "get the coordinates to fit" because they went out and calibrated to the 2 easiest points to drive to.
The reason for my question
> If you merely go to a coordinate that is supposed to fall on a straight line between two controlling corners, what is the probability that you are on that straight line and at the distance from them that you think you are?
I'll take surveying koans for $1000, Alex.
The reason for my question
What are your chances of setting it exactly using ANY method? There is error in everything. Pick the method that is good enough for what you are doing, and go with it - just make sure you know what your expected error should be given the method you choose.
