Activity Feed › Discussion Forums › GNSS & Geodesy › Vertical Control
Jim Frame, post: 419459, member: 10 wrote: Last week I did a 4-hour session on a First Order Class I mark for the GPSonBM program. I got a note back from OPUS saying that the OPUS solution missed the published ortho height by 24.4 cm. That’s about par for the course around here.
Wow – That’s horrible.
Sent from my iPhone using Tapatalk
NGS leveling order accuracy is determined more by procedure than closure. I have seen level loops run with poor procedure that close great but when you run to the far end with proper procedure you get a bust.
FrancisH, post: 419620, member: 10211 wrote: really? the error brought about by just levelling your base/rover setups and getting your HI using a tape would introduce an error or what? 5mm??
how many times have you measured an HI and repeated it and then made your own reading at where the tape curls up when you placed it over the rod?the curled up portion already has several mm of inherent error.I see your problem. I use a fixed height rod and every setup has an HA of 2.083m. Some days it measures 2.084m but I am inclined to let that slide.
Paul in PA
FrancisH, post: 419606, member: 10211 wrote: Paul,
I have checked static/RTK sessions with level loops (<10km) and difference NEVER EVER came to <1 cm.
Most are in ~5-10cm mark.
This is reason why GPS height transfers are ok for general topographic work but never for flood/mean sea level requirements or
for projects that are stretched out over several phases with different surveyors.Not to mention using RTK for verticall values for road engineering surveys that run for +100km.
Maybe you need to check your loops. We do not see what you’re talking about at all with static vectors.
FrancisH, post: 419620, member: 10211 wrote: really? the error brought about by just levelling your base/rover setups and getting your HI using a tape would introduce an error or what? 5mm??
how many times have you measured an HI and repeated it and then made your own reading at where the tape curls up when you placed it over the rod?the curled up portion already has several mm of inherent error.It is undeniable that a correctly executed level loop is a more precise way to transfer elevations over small to moderate distances than the best GPS vector. If I had a large supply of cheap labor at hand I might tend to do them more often. But doing it with RTK is often much faster and frequently more than adequate to meet the needs of a project. Measure ups are the weak point. You might try improving yours by using folding rulers, which flex less than tapes – especially on breezy days. Or fixed height rods.
FrancisH, post: 419620, member: 10211 wrote: the error brought about by just levelling your base/rover setups and getting your HI using a tape would introduce an error or what? 5mm??
The tape is the problem. I have a Trimble 4-section aluminum measuring rod that came with the kit, and I can easily hold 2 mm repeatability, maybe 1 mm with care. Mark’s suggestion of a folding scale would be nearly as good.
.FrancisH, post: 419618, member: 10211 wrote: how can you have 1st order closures and not have 1st order accuracies? if your forward and backward runs differ by 1st order limits then you are accurate within 1st order requirements.
Bad closure is proof of a problem. Good closure is proof of nothing – it is the absence of evidence of a problem, not evidence of absence. Errors may randomly add OR cancel along the way.
If there are no blunders or systematic errors, then the closure is well-modeled as a gaussian normal random variable. If the true standard deviation of a two-way run for the quality of work you do is 1 unit (millimeters, tenths, feet, or smoots), then 10% of your closures will be within 0.125 units, 5% of the closures will be within 0.062 units, and 1% of the closures will be within 0.012 units.
Regardless of the closures, if that’s your quality of work, the true elevation difference between marks still has a standard deviation of 1 unit.
And regardless of the apparent accuracy of your work, you should not claim first or second order unless you followed all the guidelines for that order.
.Bill93, post: 419692, member: 87 wrote: the true elevation difference between marks still has a standard deviation of 1 unit.
Oops – I didn’t say that right. Your round-trip still has sd=1. Each run has sd=0.7 and the mean of two runs has sd 0.5 unit.
.Mark Mayer, post: 419641, member: 424 wrote: It is undeniable that a correctly executed level loop is a more precise way to transfer elevations over small to moderate distances than the best GPS vector. If I had a large supply of cheap labor at hand I might tend to do them more often. But doing it with RTK is often much faster and frequently more than adequate to meet the needs of a project. Measure ups are the weak point. You might try improving yours by using folding rulers, which flex less than tapes – especially on breezy days. Or fixed height rods.
Here’s something: In Oklahoma, Published NGS Benchmarks are notoriously inconsistent with the results we get from the CORS. I’m talking 0.3 foot or so. Not all of them, but you never can tell where the bad ones are. If your project MUST include a long level run from one of these passive marks, it pays to get a few OPUS results on surrounding marks, so you can weed out the bad one(s) and run your level loop only to the ones that fit best. They tell me the new 2022 Datum will make things much better.
Jim Frame, post: 419459, member: 10 wrote: Last week I did a 4-hour session on a First Order Class I mark for the GPSonBM program. I got a note back from OPUS saying that the OPUS solution missed the published ortho height by 24.4 cm. That’s about par for the course around here.
Jim, I would be more suspect of the bench over the CORS. A couple of years we did a 400 mile long, from Richmond to Barstow, LiDAR verification survey for BNSF. The reason we did it was because a local former BNSF surveyor did an initial check somewhere near Fresno and told my client the LiDAR was off by over 37 feet vertically. It turns out it was the benchmark that was off. The amount of water and oil being pulled out of the ground has cause some serious subsidence in areas. After that we did some work for the high-speed rail segment around Fresno (I’m spending way to much time in that town) only to find significant subsidence in the year or so since the original control was established. The beauty utilizing active control is that they are continuously monitored to see how they fit the adjacent CORS. In theory they their coordinates should be adjusted when a threshold of change is observed. You would be running way more that a couple of miles of level to find the 37 feet of subsidence.
John Putnam, post: 419760, member: 1188 wrote: Jim, I would be more suspect of the bench over the CORS.
There’s no question that the BM has subsided, though there are other problems in realizing NAVD88 from CORS around here. The last level run on that line, as I recall, was in 1968 or thereabouts. (Now that NGS has disabled LINETOOL it’s kind of hard to check.) My purpose in submitting the GPSonBM session was to ensure that NGS isn’t tempted to use that line in its geoid modeling.
jakethebuilder, post: 419712, member: 9380 wrote: Here’s something: In Oklahoma, Published NGS Benchmarks are notoriously inconsistent with the results we get from the CORS.
That’s true everywhere. You have to first check the datasheet for confirmation that the elevation was determined by differential levelling and not by some lesser means (up to and including scaling from a USGS quad map). Then you have to accept that the levelling was maybe last done in the 1930’s by the WPA. Stuff settles, even in OK. It’s never perfect and some are less perfect than others.
Bad closure is proof of a problem. Good closure is proof of nothing – it is the absence of evidence of a problem, not evidence of absence. Errors may randomly add OR cancel along the way.
for 1st order leveling, we need to run forward/backward runs and if the PC is skeptic of the results either due to terrain route or temperature or he just feels like making you do it again, we do a 3rd run.
Believe me, in all my 30 years of doing field work, if a run is wrong it’s wrong. I have not yet encountered a forward run/backward run that met order limits that was erroneous. A blunder will be evident in the 1st/2nd run and caught by the 3rd run.
If you have encountered a 1st/2nd run that were within order limts and is erroneous, it is more in my opinion the data was doctored.
Since using digital levels, doctoring of field data has been eliminated or is easily caught by the person doing the QC.The only way to call something truly first or second order leveling is to use invar rods and follow specs.
I have heard a lot of surveyors say that leveling is easy. Once you get into higher order leveling, you will find there a lot of nuances.
There are all kinds of small systematic errors that good procedures and proper equipment will minimize or eliminate. We run a lot of second order levels to spec, using invar.
My first digital level was a leica, with fiberglass/wood rods (early 90’s). The joints would wear, and we would start getting poor results on one way leveling, but good closures on loops. This would happen if there were a lot of high rod/low rod shots (i.e. going up hill). When you come back down, it cancels out. First time I saw it was on a run up a 500 foot hill. GPS at top and bottom disagreed by 0.25 feet, loop closed much better. New rod proved that the joints loosening had caused the problem. The solution was to buy a new rod every year. Not sure if they have improved the rod design.
Since I had my own company (1998), I have been using Zeiss/Trimble. Much better rod design, still have my original wooden/fiberglass rod from 18 years ago, still closes good, compares well to invar rods (I have 3 invar rods).
Have not used invar rods, we got into 1st order work directly using Sokkia digital levels coupled with their fiber glass rods that Sokkia said would not shrink/expand under normal noon heat. PC made us run the forward run at 6am and the backward run either the next morning or from 3pm onwards as sun was on way down.
One requirement that rodmen needed to follow was NOT to use turning points of 1st run for the 2nd/3rd runs. This rule was to make sure that the 2 runs are made on different paths even if they traverse the same route. Usually loops are made along roads as much as possible.It’s not possible to do a 1st order loop without invar
It’s not possible to do a 1st order loop without invar
I think you are reading into the old standards. Digital barcode rods are capable of 1st order standards.
If you read on the various international standards on 1st order leveling procedures there are now clauses
for digital levels & barcode rods.
Technology changes so standards have to adopt to these changes.FGCS specs for both First and Second Order leveling require invar rods. They can – and generally are, these days – barcode rods, but they have to be invar.
Again never used invar rods in my life. Weight alone based on specs is 2x or 3x heavier than fiberglass staves.
For anyone that has done actual fieldwork in their professional life as against sitting in the office doing the
computations of field crew would know, weight of equipment gets heavier for the men as the day grows on.
And anyone with field work experience would also realize is that accuracy goes down as fatigue grows up.
Our office has done 1st order level results using digital levels & fiberglass staves. Please don’t tell me
it’s not 1st order because it’s not invar. There is also a regulation that says foresight/backsight distances should
also be equal. Sure, try to have equal BS/FS distances going up/down a hill. You follow that requirement and you
won’t see your crew back in the office for several weeks.I have a feeling that those regulations were written by office sitting pencil pusher surveyors.
FrancisH, post: 419799, member: 10211 wrote: Again never used invar rods in my life. Weight alone based on specs is 2x or 3x heavier than fiberglass staves.
I have 2 barcode rods: a GKNL4F 3-section fiberglass rod for general work, and a GPCL2 invar rod for geodetic work. The invar rod weighs almost half a pound less than the fiberglass rod.
Most geodetic leveling is done with 3-meter rods, which weigh about a pound and a half more than my 2-meter GPCL2, but being in the flatlands I can get away with the short one.
FrancisH, post: 419799, member: 10211 wrote: I have a feeling that those regulations were written by office sitting pencil pusher surveyors.
On the contrary, they were written by men who through experience and determined — one might say obsessive — experimentation were able to codify the requirements of achieving reliable accuracies.
If you ever get a couple of spare days, I highly recommend the NGS geodetic leveling class.
Log in to reply.