Single BM projects tend to be more prone to issues than those where 2 or more are tied in. The RTN people rely on the accuracy of the data provided, and quite frankly are kind of hanging in the wind IMVHO. The level loop people using only one published BM are also a bit hanging, but likely would have a better leg to stand on.
I don't even own a level. I use GPS or my robot, but always check into 2 or more to make sure that things are at least "good enough" so I am comfortable affixing my state issued RLS seal. At the end of they day, it all becomes about liability and repeatability. Isn't that why we think we get paid the "big bucks"?
Fifty years ago next week Col. Epps said to the class, "How accurate should your measurement be?" My reply was, "It depends on what you are trying to accomplish."
A lot of things have changed since then, but the answer is still appropriate.
If they want to play fast and loose, they will eventually be bitten.
There's a fast and loose firm in the next Town over. We've had disagreements on more than one occasion; fortunately they have been wrong in each instance. Let's hope the trend continues in my favor.
Is the orthometric correction significant for your area and the length of level run you made? Did you apply it? I recall Loyal giving an example from his area where there was a serious error in just a few miles if you ignored it.
Does their software do a proper calculation from measured ellipsoid height to orthometric height taking into account the proper geoid?
Hang On.
First, did the job specifically require NAVD88 (or some other flavor) elevations. If not, then he (or you) could have assumed 100 and made it work, right? Whether or not the NAVD88 is EXTREMELY accurate or not is kind of irrelevant as long as the elevations are accurate within themselves.
Also, you live in a charmed world where there are benchmarks handy. I recently did a topo in an adjoining town and I used OPUS to bring elevations to the site (then I topo'd it with a 1" robot) but my elevation basis is NAVD88 per OPUS. It's VERY repeatable as this isn't the first time I've done it and I have HUGE networks in the oil patch done this way and the elevations work.
Then, even though I could see the site, I used RTK to transfer my coordinate base and elevations about 1500' down a busy road so that I knew, again, that my elevations were real world elevations. I shot multiple points, multiple times, under multiple constellation configurations, and verified that none were different more than 0.05'. With three shots, I meaned them all out and then used the 1" robot to run between them all and found everything within 0.02' elevation wise.
Don't knock it till ya try it. Am I gonna topo with RTK? I have, but I don't like to for a myriad of reasons, but it's still a damn good tool to have and use when done properly. Hell even the NGS produced a paper on it.
I want to know how different your level loop from BM is from there RTK Network as fixed by the CORS stations and latest and greatest GEOID model is though. 🙂
Here is a tidied up version of the post that Bill mentioned above.
Real world example:
Invar Rods, Balanced Turns, Micrometer Level, read to .001 feet.
Four Miles containing 78 turns between two Stability 'A' NGS Bench Marks (U-173 & P-173, both are disks in bedrock) situate alongside a Railroad (circuit run along the access road parallel to the tracks).
Vertical Difference = +53.727 Meters or +176.269 feet (NGS record NAVD88, First Order Class I 1970)
Raw Closure w/o Helmert Corrections:
0.0470 Meters or 0.1540 feet
Raw Closure WITH Helmert Corrections:
0.0006 Meters or 0.0020 feet
Now “maybe” this is compensating error, but I don't think so. A number of other NGS Bench Marks were observed (see below) although their stability were “C' and 'D' (except for T-173). Note the data below is in METERS.
Station S T Leveled Record Rec-Lev Gravity Helmert Rec-Hel
U-173 (TC-16) A 44 1356.0650 1356.065 0.0000 979.7970 1356.0650 0.0000
Turns 16 10.5149
T-173 (TC-18) A 60 1366.5799 1366.589 0.0091 979.7915 1366.5863 0.0027
Turns 16 4.8292
BM_1 (TC-01) ? 76 1371.4091 none n/a 979.8255 1371.3674 n/a
Turns 16 14.0465
R-173 (TC-19) C 92 1385.4556 1385.479 0.0234 979.7788 1385.4779 0.0011
Turns 12 10.6477
Q-173 (TC-20) D 104 1396.1033 1396.137 0.0337 979.7737 1396.1316 0.0054
Turns 18 13.6417
P-173 (TC-21) A 122 1409.7450 1409.792 0.0470 979.7599 1409.7914 0.0006
S = Stability
T = Turns
Leveled = RAW Field heights (no Corrections, in Meters)
Record = NAVD88 Height from NGS Data Sheets.
Rec-Lev = NGS Record minus RAW Leveled Height (in Meters)
Gravity = NAVD88 Gravity estimate in gals (NGS NAVD88 Gravity Model)
Helmert = NAVD88 Height in Meters, (Helmert Corrections applied)
Rec-Hel = NGS Record NAVD88 minus corrected Helmert Height (in Meters)
The above data is BEFORE final adjustment of the circuit (which didn't actually change these numbers much at all).
Now it should be noted here that the gravitational gradient that heavily influenced these results, is not generally typical for a ~180 foot elevation change over 4 miles, and reflects the situs of this circuit relative to a significant Mountain Range (which the circuit is more or less paralleling along the Western range front).
Now I would submit that this type of gravitational situation is NOT all THAT uncommon in the Western HALF of CONUS, and in some areas it is even MORE pronounced. I mean shucks man, we are only talking about a “grade” of about 1% along this route.
Now maybe .15 feet in four miles is within the warm and fuzzy of some folks, but as EASY as the Helmert Corrections are to do, I just don't think that it's something that should be neglected as a matter of convenience or [canned] software limitations.
SHORT loops with minimal height/gravity variations are NOT going to benefit much (if any) from the Helmert corrections (EXCEPT when they DO), and you don't really know for sure until you run them.
Large (long) "out and back" loops where you don't have NGS Control at each end, can REALLY bite you on the ass when the Helmert corrections are NON-trivial (such as this case). Just because you "go out and back" and "close back in" at your start point, does NOT mean that you have a valid NAVD88 height at the other (far) end! Had we "looped back" in the 4 mile example above, we would probably "close" within a few millimeters, but without the Helmert corrections, our numbers would all be wrong.
The above data represents the first four miles of a 40 mile LOOP that ranged in height from about 1483 meters (4865 feet) to about 1289 meters (4230 feet).
Loyal
Large (long) "out and back" loops where you don't have NGS Control at each end, can REALLY bite you on the ass when the Helmert corrections are NON-trivial (such as this case). Just because you "go out and back" and "close back in" at your start point, does NOT mean that you have a valid NAVD88 height at the other (far) end! Had we "looped back" in the 4 mile example above, we would probably "close" within a few millimeters, but without the Helmert corrections, our numbers would all be wrong.
:good: :good:
And if you start on a Bench Mark that's been disturbed you can really be wrong. I've seen two benches used on control networks recently-one that was .4' off and another that was .8' off. I couldn't check them with GPS and ran through them with levels to "prove" their "error". Both of them looked fine on the ground, but they had probably both been removed and reinserted by railroad construction without any record, so I sent the data into the NGS to update the marks, but I always suspect any standalone bench mark.
Another example
The one I was thinking of was [msg=42933]here[/msg]
wherein Loyal demonstrated about 2 tenths of a foot in 2.8 miles.
It would depend on the requirements and specifications of the individual projects.
I always use a published A.H.D. mark for datum where one is available though.
Mighty and Gavin are spot on!
Aligning your survey (horizontal OR vertical) REQUIRES at least two Monuments (datum points) that are expressed in the SAME datum (and hopefully the same adjustment), that ARE verified in the field as being spatially consistent.
Even stability 'A' Bench Marks (disks in BEDROCK) within the same USC&GS/NGS Circuit/Adjustment, can MOVE relative to each other (think faults and [even minor] earthquakes over the years). High-end (stainless steel rods driven 30-40 meters to refusal) Bench Marks in the valleys can also MOVE quite a bit over the years. In the Tooele Valley (Utah), “depth to bedrock” can exceed 12,000 feet, so a hundred feet or so is pretty 'superficial,' and anything set on the surface (even “massive structures”) are a crap shoot when your are talking decades since the last level run.
Checking between TWO (or more) Bench Marks does NOT insure “absolute accuracy,” but it will at least confirm “local accuracy” and identify most stinkers.
Loyal
In the Tooele Valley (Utah), “depth to bedrock” can exceed 12,000 feet, so a hundred feet or so is pretty 'superficial,'
I've driven quite a number of control points to "refusal". I finally figured out that "refusal" was when my hand and arm refused to swing the sledge anymore;-)
Mighty
Roger that!
B-)
Loyal
