I'll grant you that control on centerline is frequently not very convenient, but saying that it can't be used was wording it too strongly.
I've never seen one placed on centerline, is it possible, I suppose anything's possible.
Is "CONTROL"
- a point which the GPS base or total station can be set up over, or
- a well monumented point whose position relative to other monuments and features is well known and documented.
If it's the former, then centerline monuments are not likely to be good control. On the other hand I can show you any number of NGS brass caps that are next to impossible to use as control by this definition. If it is the latter, centerline points can do the job as well as any other.
I'll pass on centerline as a survey control monument. Boundary control for sure.
When I was in charge of creating a survey mark GIS for our state we assigned three classes and we accepted all input. A single observation was equal to unknown quality with no error estimate. Could be balls on. Could be off 100 miles or anywhere in between.
A static survey - network adjusted by a PLS with positional error estimates (high confidence)
B Averaged RTK or conventional with positional error estimates (medium confidence)
C Single observation or unknown with no positional estimates. (low confidence)
In some cases, a centerline Mon in a median is the safest place to leave the instrument.
Check out these YouTube videos regarding GPS accuracy from thr Orange County Chapter of the California Lad Surveyor’s Association. They include qua little nugget on measuring Centerline.
In some cases, a centerline Mon in a median is the safest place to leave the instrument.
Back in my early days, I sat on centerline for everything all the time. We built highways, railroads, dams, powerlines; all of it was centerline focused.
Back in *my* early days we used to set up on the centerline of a railroad track and use the miles-long tangent for meridian. I always hated standing there by the gun, nervously eyeing the distant engine headlight, trying to judge how far away it was -- it could have been a mile, it could have been ten miles, in that flat ground it was impossible to assess reliably. It was always a great relief when the party chief gave me the okay to pick up and move off the tracks.
I never had any close calls doing that, but I heard of another crew that lost a gun to a train that way.
If I were spending three minutes at the point I'd likely extend it and do a static network. If RTK was my only tool I'd split the occupation and spin the rod 180. At least you're getting rod error out of the mix..
I've done a lot of testing and found improvement on RTK positions all but stops at the 20 second mark. The display will tell you it's getting better, but that's due to the abuse of common statistics formulas.
I've done a lot of testing and found improvement on RTK positions all but stops at the 20 second mark. The display will tell you it's getting better, but that's due to the abuse of common statistics formulas.
I'm seeing similar results. Newer GPS units are freakishly accurate.
That describes what I see and my standard observing practice. 20 epochs at 1hz, LF+RF with pole. Pole adjustment at 2m can easily have a few mm in it and even your centring on the mark can vary a few mm, all adds up.
If I were spending three minutes at the point I'd likely extend it and do a static network. If RTK was my only tool I'd split the occupation and spin the rod 180. At least you're getting rod error out of the mix..
I've done a lot of testing and found improvement on RTK positions all but stops at the 20 second mark. The display will tell you it's getting better, but that's due to the abuse of common statistics formulas.
Why not both? Set up a static-only base locally, and collect RTK-logging when using a network rover… post process the rover logging against your local base and shove those vectors alongside your network vectors into the thunderdome for adjustment
You can build some pretty neat networks without going the traditional “static network” route.
Recent example of two 5’ observations under full canopy, visited this past weekend (20 hours apart, rover facing opposite directions for each observation).
They include a little nugget on measuring Centerline.
I believe that he is referring to the referencing of centerline monuments (4:00 minute mark of episode 4) to nearby offsets, not to centerline monuments generally. And while I agree with him generally that such centerline referencing should not be done with RTK alone, I think that he is vastly overestimating the probable errors in so doing. He is showing a 0.09' peak to peak variation in instantaneous RTK positioning and saying that is the potential error in an RTK measurement. Well, that is the reason that one collects RTK for some period of time, to average out those momentary peak to peak errors. True, one should be able to beat RTK for precision on these short measurements by using the TS, but not by as much as Mr. Wooley is claiming. Episode 5 addresses my comments, but I still say I get considerably better than 0.09' with my units in decent conditions. I may have to look in to repeating his experiment.
I would further expand on that and highlight how locked down centerline control is expected to be in Southern California, especially in the Los Angeles area. High expectations for tight distance and angular relationships for monuments and ties- mostly tangency.
I’d argue that you’d need to double determine such work with two separate TS setups- my go to method was to locate ~3 ties or CPs with 2 rounds of GNSS to anchor the intersection to my network and then run two separate TS setups to tighten up relative relationships.
Edit: in my mind: 1 TS setup = relatively confident for relative distances… 2 setups = confident in distances and angles. At the same time I love the concept of overkill and I’ll happily do it on the regular.
Edit2: I would agree that although I wouldn’t necessarily say that Dave is overexaggerating the potential errors regarding RTK, the argument is definitely posed in a purposeful way to lead the audience to his conclusions. Like the canopy example I posted above: you can get obscenely good and repeatable answers with GNSS with proper procedure. Period. For that passion project in particular every monument is getting 4-6 visits… because I’m bored on the weekends.
@meh I have yet to see a usable network solution in Idaho. We routinely see 0.3' of difference between network rtk and validated static positions. When I'm locating monuments that won't cut it...
That’s fair- I remember having similar experiences with CRTN… apparently there’s been improvements as of late but I’m not in state anymore to find out.
SC RTN is the example above, and appears to be giving solid results. You can also download 4 constellation T0 files for their stations, which makes me one very happy camper.
We don't have a network, the only thing available for me is Trimble's ProPoint technology. However, testing by my office got less than ideal results. So I'm not stepping into it. I'll continue to use CORS/OPUS and Base/Rover.
To answer the original question, this is fine to do as an “inventory “ of monuments. A county in Ohio has that mostly for section corners, which position data and what it’s is in its attributes. It’d be great for a surveyor to have a full county inventory database on where centerline monuments are in every road to know how easy or difficult a road might be to survey on, in an ideal world. For “inventory “ purposes, only need maybe 10 seconds because it’s not for use in an actual survey. It’s a GIS data need not survey.