The whole use of state plane scaled to ground is what makes my blood boil. ?ÿAs others have stated it is not and just because the numbers look similar doesn??t make them state plane .
Again, that??s why I??d rather have ??project/site coordinates? that look nothing like SPCS with no projection. Very little confusion IMO. Also, Topcon recommends at least 4 points for a localization even though you start to get residuals after 3 points.?ÿ
I??m curious though if I set 4 points with RTS, then run levels between them and adjust the elevations, then what would the vertical residuals look like when I localize the gps? ?????ÿ
Also, from now on I will not call them ??ground coordinates? but rather ??site coordinates? so thank you for the clarification.?ÿ
"5000,5000,100 coordinates are on a flat plane too. Plane geometry = rectangular grid."
Are you saying that (5000, 5000, 100). (5050, 5100, 110), (5200, 5250, 130) and (5150, 5150, 125) are all on the same plane?
?ÿ
@on_point ok 4 points now that good. Now no matter which brand of equipment or software. When you do perform a localization or site calibration and a minimum of 4 points please understand that those points need to be around your site. Once you go outside of the calibration your error increases. Now that??s horizontal and vertical. But lets get you to thinking a little more. You say you get residuals with 3 points. ?ÿNow lets picture a 3 legged stool vs a 4 legged chair. You know those chairs that made your teachers throw the chalk board eraser at you for wobbling in it and making noise on that hard flat floor. Now lets say on the 4 legs i have a slight error in one and or the floor might have a little un flatness error. We now have residuals with redundancy we can see. ?ÿNow set that 3 legged stool no matter what it doesn??t wobble in theory now all three feet will hit the floor. No redundancy no way to truly check. If we look at the seat and its a flat seat on that stool it can tilt in any direction of the error in the leg or legs. ?ÿThat seat is still flat but tilts. ?ÿSo look at the ppm as well not just the deltas of the points. ?ÿThis is a elementary example not implied to be perfect but a small visual. ?ÿI don??t do well writing it but better at showing it. ?ÿHopefully this gives a little insight. Maybe some of the better writers can help explain better. On e had a professor tell me anyone can memorize or regurgitate information but the true person with knowledge can explain the toughest problems to a child to get the concept. My problem is typing on this Not So Smart Phone and poor grammar here lol.
@on_point if you ran levels you would eliminate error on the control side vertically. Because levels if done correctly are more accurate and precise than trig levels. ?ÿNow you still have the error in your gps measurements. ?ÿErrors are like freckles they can appear anywhere anytime. As a surveyor we try to eliminate as many errors as possible. ?ÿBy what we can control and what we know. So why do we turn multiple angles with a total station. That same total station why do we turn direct and indirect (face 1 and face 2 measurements historically) just to name a few errors in a set up. ?ÿSet up error sighting error error in the plate error in the build of horizontal and vertical axis. Parallax refraction etc etc etc. ?ÿthis we try and eliminate by procedures and routines in field. Not as much talked about these days but a good i man would keep a eye on facing in a direction and sun was mostly on one side heating instrument more on one side than the other. When crew chief said smokem if ya gottam. You checked level bs zeroes or azimuth and moved gun to ge sun on the other side while he was doing whatever. Always checking. ?ÿI bet if we sat on a point for two hours straight or longer we checked our back-sight often. ?ÿ Now the equipment today has become so good it might be a non issue but when i am in field i ck at those times i am pausing between task or that truck or dozer is online and just tell gun to look back and ck zero . It is a non smoker ????.?ÿ
one other thing as you develop your skill set with localization etc. study strength of figure. The earlier you acquire that knowledge the less your traverse or anything you do will bite you. Strength of figure and balancing out control will solve more problems than most will ever truly understand. ?ÿIt??s a study well worth the time. And learning to apply it on your job sites.?ÿ
When you do perform a localization or site calibration and a minimum of 4 points please understand that those points need to be around your site. Once you go outside of the calibration your error increases.
Yes, I like to keep the control I localize to surrounding the site as far as I can. I believe it reduces the magnitude of error since you will always have some error.
I??m curious though if I set 4 points with RTS, then run levels between them and adjust the elevations, then what would the vertical residuals look like when I localize the gps?
?ÿ
That would depend on both the relative accuracy of the RTK vectors, and the method of vertical adjustment chosen.
For average-size sites, though, doing a vertical shift plus a geoid model will get you there 99% of the time. I would expect to see cm-level vertical residuals for tight levelled control and solid GNSS procedures.
Are you saying that (5000, 5000, 100). (5050, 5100, 110), (5200, 5250, 130) and (5150, 5150, 125) are all on the same plane?
Ahhh... I see what you did there. (D'oh?!)
Although (by luck) I was technically accurate with my post, it would appear that I was dangerously close to "lazy, sloppy, and imprecise", opening the door to interpretation.
Guess I need to lock my doors...
Touch??
Since linear distances in three dimensions are calculated in a Pythagorean way, it's easy to carry the plane concept over.
But when we introduce the third coordinate axis, a plotted point is no longer required to rest in any one single coordinate plane. Any three points will still define their own unique plane but beyond that, all points can't be assumed to lie in the same plane.
Hey?!
I (subtley) kinda, sorta conceded your point (interpretation).
In fact, take a look at my last post on the "ground distance" thread.
?ÿ
Come at me, bro!
LOL! No, just reinforcing the point. My heated arguing days ended years ago when i took note of how often I was wrong.
Great discussion, though, providing much to ponder.
I??m curious though if I set 4 points with RTS, then run levels between them and adjust the elevations, then what would the vertical residuals look like when I localize the gps? ?????ÿ
Are you localizing your own observations and control to better get GPS on vertical leveled data?
?ÿ
?ÿ
I??m curious though if I set 4 points with RTS, then run levels between them and adjust the elevations, then what would the vertical residuals look like when I localize the gps?
Don't do this unless you absolutely have to. When you include those 4 points vertically in your calibration, you will create an inclined plane. If there is a bust in your elevations or your RTK when collecting those points, the inclined plane you created will cause a problem for that site. If you use a geoid and 1 point vertically in your calibration, you will end up with a constant separation from the ellipsoid. This is much safer and has 99.9% of the utility of an inclined plane. If you are in that 1 chance out of a thousand that an inclined plane needs to be used so be it. Otherwise there is no need to take the additional risk.
Don't do this unless you absolutely have to. When you include those 4 points vertically in your calibration, you will create an inclined plane. If there is a bust in your elevations or your RTK when collecting those points, the inclined plane you created will cause a problem for that site. If you use a geoid and 1 point vertically in your calibration, you will end up with a constant separation from the ellipsoid. This is much safer and has 99.9% of the utility of an inclined plane. If you are in that 1 chance out of a thousand that an inclined plane needs to be used so be it. Otherwise there is no need to take the additional risk.
Unfortunately this is exactly the way the Survey suppliers train the contractors to setup machine grading gps setups.?ÿ I've been round and round (in a good way) with the 6 or so contractors i have worked with that utilize this technology.?ÿ I repeatedly hear that's the only way to get their calibration setup.?ÿ
Are you localizing your own observations and control to better get GPS on vertical leveled data?
I??m localizing the rtn gps to the site coordinates/no projection because occasionally I use the gps get some quick topo, locations, or layout where precision isn??t as critical and I like to keep everything on one coordinate system and job in the DC. If I can.?ÿ
If you use a geoid and 1 point vertically in your calibration
That was another question I wanted to ask. Also, How much of a vertical plane would you get if did a really good level checks and adjustments and was careful about what control you used in your localization?
I??m curious though if I set 4 points with RTS, then run levels between them and adjust the elevations, then what would the vertical residuals look like when I localize the gps?
Here's what the gnss calibration people aren't telling you:
So, as long as you are "within the box", there is not a better way to tie gnss orthometric heights to a local site. "Exaggerating the tilt" would still happen if you used total station, etc.
If you level (and hold those heights) the 4 corners of your project site (the box), your "tilt" from gnss would (should) be almost exactly the same as the "tilt" from total station, which is exactly the same (related) to the leveled heights.
Leveled heights are the best 1d accurate.
Total station positions are probably the best 2d relative positions, and pretty darn good 1d precise/accurate.
GNSS is really good x,y,z, and, if site calibrated, with repect to your 1d levels and 2d traverse.
(It is difficult to get much better than that.)
?ÿ
1 point vertically fixed plus geoid model still depends on things outside your control.
You can site calibrate without a geoid model, and that is okay to do so, as long as you have good (local) vertical control to calibrate to.
Control/calibrate to the (county) extents. Work inside of those constraints, and you're fine.
And "technically" you're on the best low distortion projection for the site you can get!
I??m curious though if I set 4 points with RTS, then run levels between them and adjust the elevations, then what would the vertical residuals look like when I localize the gps?
Control/calibrate to the (county) extents. Work inside of those constraints, and you're fine.
MI-Other-Left, are you calibrating to control county wide?
?ÿ
Before this thread departs into cyberspace let me add my last take on field calibrations. I think we know RTK is used the great majority of the time to calibrate to existing control. Some will average RTK observations and some won't if they're in a hurry. Even if I did believe in calibrating, I wouldn't do it using RTK observations. I wouldn't do original project control with RTK observations either. Control -existing or new- requires a solid static survey with the proper number of repeated sessions. This is particularly true in the vertical component.?ÿ
A number of years ago I was approached by a technician I did not supervise wondering what was going on with his project control checks. One time his check would be good and the next time it would be off several tenths of a foot. After asking about his field procedures, I discovered that as he moved about his project if he missed a check too much he would add that point to the calibration and eureka! It would check great. But now the point that checked before was way off.?ÿ
You can say this is just inadequate training which is true, but I saw this nonsense repeated too often even among people who you think would know better than to allow field calibrating under my watch. Maybe I've been out of the game too long and this doesn't happen anymore but I kind of doubt it. Friends don't let friends calibrate.?ÿ
Even if I did believe in calibrating, I wouldn't do it using RTK observations. I wouldn't do original project control with RTK observations either. Control -existing or new- requires a solid static survey with the proper number of repeated sessions. This is particularly true in the vertical component.
?ÿ
I'd say that all depends on the project goals and vertical accuracy requirements, as well as how terrestrial work is handled. With proper procedures, RTK/NRTK observations can absolutely be used in lieu of static work.
The joint ODOT/OSU study that came out last year is pretty compelling, and based upon my experience processing control networks with NRTK and terrestrial observations over the past ~5 years, I'd say that the report findings are conservative, and that using the methods outlined brings better results than the report itself claims.
Full-constellation support along with improved RTK engines has really changed things in the past 10 years.
?ÿ
I'm still a static proponent, especially for any medium to large-scale job. But RTK ain't the dog it used to be, and when properly employed with terrestrial work it is a viable tool in the control toolbox.
