1-point vertical calibration accuracy

The purpose of multiple points for a vertical calibration is to help mitigate the unavoidable error that will occur. What you're doing seems more like a simple adjustment than anything, being that you're not spreading the vertical out across the project area but simply tying it down to one point. Seems to me that the most that would or could do is apply a constant adjustment to your verticals, and not much more. Sure, that'll be fine in small areas like you said, but start expanding the project area and you'll get some undesired results.

Also good to remember, checking a calibration against itself with the same equipment that was used to establish it will almost always yield phenomenal results. If you really want to prove it up, run a level loop or conventional traverse across the calibration control afterwards & check your differences.

Arthur,

I am not too familiar with Trimble gear. I briefly have used TBC and know it does show similar information on the vertical plans as my Topcon gear shows. If you have TBC, review some of the old calibration files with the four or more point vertical calibrations. Look at the "Maximum Slope of Inclined Plane". With a multiple point vertical calibration this plane will be a computed value. It can get grossly tilted, and have an adverse effect on large projects spanning a mile or more. Adjust one elevation by 0.05, then recompute and review the new values for the Inclined plane. This will give you a feel for how much a error you are comfortable with. On a single point vertical calibration the plane will be 0.0000%.

Good luck.

Here is an example of a 0.05' vertical error from Topcon.

algallop, post: 368370, member: 8394 wrote: My boss, a licensed surveyor for many years, taught me that in order to get accurate vertical readings while performing an RTK survey, you have to calibrate on four or more points with good vertical solutions. However in practice, I have done RTK surveys calibrating on only one point vertically and checked it with a level and/or total station, and found that the WGS '84 projection (both in SurveyPro software and Trimble Access) works pretty well establishing an accurate vertical plane.

I would like to perform topographic surveys with our R10 equipment using a 1-point vertical calibration, but first I have to convince my boss that it is accurate... Can you guys give me some convincing feedback that it's safe to do this?

I'm sure it's not as simple as I am making it sound, and I'm not an expert in GPS technology or anything. I just know that in my experience, a RTK 1-point vertical calibration is just as accurate as a level or a total station (at least when you are only working within a relatively small area, maybe not over miles and miles).

Thanks in advance,

Arthur

Using only one point is not the best practice. You are totally dependent on whatever geiod model you are using plus how do you know that the point you chose has a good elevation? Also you are restricted to a very small project area.

4 points was the norm before we had as accurate geoid models. Now I calibrate to one and check the other 3 if possible. It all depends on the accuracy or work that is required. We also run digital levels when required. It is just a tool in the box, do your own tests and build your own confidence. I still remember when my boss did not trust the edm over the steel tape :). My 2 cents, Jp

Offer to do a test localization (on your own time) with a one vertical point on a site for which your boss has pre-existing data on a multi-point vertical localization. That should give you a valid comparison.

I have seen multi-point vertical localizations (in the field) get severely inclined with bad observations.

For small-ish sites I like a single point for vertical; with 4 horizontal points. Then I check a 5th point for confidence.

Maybe bet him a lunch that it will work well; if you are confident, that is!

FWIW... Your suggestion that GPS is as accurate vertically on a small site as a total station or level is false. It probably could be, on occasion. If you are looking for hundredths GPS is not the correct measurement technique.

The conditions affecting vertical accuracy for GPS are too complicated for generalizations. If it has to be tight, levels should be run. I will always lean towards optical instruments when vertical is critical and grades are minimal.

algallop, post: 368370, member: 8394 wrote:
I would like to perform topographic surveys with our R10 equipment using a 1-point vertical calibration, but first I have to convince my boss that it is accurate... Can you guys give me some convincing feedback that it's safe to do this?

What makes a level loop the gold standard for vertical? When you run the loop, you get a check. It CLOSES.

What makes the total station elevation only a silver standard? It isn't precision, I have tons of data that would indicate it is as precise as reading a rod on a level loop. It is because you have not turned through the stations. That hub set for that minimum slope sewer, it isn't CLOSED.

"Safety" is found in the checks, the closure, etc.

So, that being said, you can actually do this in the office. Your office or data collector should be able to apply localizations, and then update the coordinates by applying the localization after the fact.

You should be able to turn on and off points, generating coordinates at each saved change in the localization.

Do the localization for the 4 points. Then look at a point in the middle.

Now do a single point, using WGS 84, and compare the point in the middle, and the as shot values on the 4 points (if you shot them with RTK).

Now do a single point using the appropriate GEOID model.

Now you can see the difference, for the same shots, the same observations, etc. Now run a level loop. I will tell you what you will find. Your VRMS, double it, and that will be about the size of your random placement from the gold standard. Likely it will be about .2' (.1' up and .1' down).

A few questions

1. Are you using a geoid.
2. Is this with base and rover or VRS
3. How big is the site.
4. Is the point being calibrated/localized to in an established vertical datum.

I think the big thing to take away and what a few other posters have mentioned is this is almost never a 1 solution fits all circumstances scenario. There are times to localize and times not to in general I try to avoid it unless tying into an established project datum even then it's good to check how said datum was established. Even so I can't imagine shooting 3 more points could do any harm even if they aren't used in the localization.

While I'm [reasonably] sure that there ARE places where "inclined plane" [vertical] calibrations work okay (at least on SMALL projects), I'm not sure that I have ever seen one. But then, I don't generally work on "small" projects, or in areas where such a critter might be well suited in the first place.

I HAVE however, seen [far] too many abominations directly caused by inclined plane vertical calibrations. Most of these were caused by doing something stupid, but some were caused by folks who TRIED to do it right, BUT simply didn't understand some of the basic check, check, check, and oh by the way, you (they) used el crappo (or mismatched) data to constrain (generate) the incline plane in the first place.

Bottom line here (as alluded to by several above), is that IF you need GOOD NAVD88 Heights, you need to get the level out (or even the total station with proper procedures). Now just how "good" GOOD is, depends on many variables too numerous to address. Kinda like how "small" SMALL is, and what the definition of "is" IS!

Loyal

I'm with your boss. While others have good points as far as very small sites, checkins, good geoids etc....why do something half ass. Calibrate the right way. Why put your reputation on the line? I've followed way too many of these on construction projects, one arrogant idiot made me drag him in front of half dozen city engineers, contractors VP, the engineer he worked for, my 2 bosses....all to show I can't follow his 1 point survey. He had done an autonomous job, even worse. Funny thing was I think my whole fee was 4k. Small job. Big problems.

You start this with "my boss", he's the one with the stamp who certifies the work you perform. If my employees did a half ass job of verifying their work they wouldn't have to worry about work for long.

Boy o boy. This is a great topic.
Lots of good advice.

What I want to know is:

Does your localization software fit the geoid to the control?
or
Does the software fit a plane to the control?

If it fits an inclined plane Then:
Does the geoid in the project area have much undulation?
If so, do the errors generated by the difference in the plane and the geoid exceed your error budget?

I believe dmyhill's aproach will tell you this. I also agree with Kevin and Loyal.

GPS does not do well in low slope/drainage critical areas.
I believe that vertical localizations are only valid when either:
vertical is not critcal
or
the control is vertically sound (levels or other gravity based method) and there isn't much geoid undulation across the project area.

I suspect that localizations and GPS are responsible for many bird baths and other puddles.

Does ANY controller software fit the geoid to the control when localizing/calibrating?

Here's an image of what Geoid12a looks like in Utah. Basically a contour map (relative to the NAD83 Ellipsoid) expressed in 0.1 meter contours. Not too many "planes" in the larger sense, but in the postage stamp sense, one could probably get away with an incline plane locally.

If you're using a Base/Rover setup then you don't do anything, set up on your elevation point and go. The file should be correctly projected, the geoid model applied and you hold the elevation of the point you are on.
Never, never, never calibrate, it's a bad idea, especially vertically. Now if you are in a network??
Don't use them so I don't know, but I would advise to use Base/Rover for tight site work anyway.

Loyal, post: 368455, member: 228 wrote: Here's an image of what Geoid12a looks like in Utah. Basically a contour map (relative to the NAD83 Ellipsoid) expressed in 0.1 meter contours. Not too many "planes" in the larger sense, but in the postage stamp sense, one could probably get away with an incline plane locally.

Loyal, we tested this pretty extensively, the first was the local (small) airport. That area is clearly too large for calibration, we had a network of levels so it was simple to check. It was very frustrating trying to get calibrations to work vertically, basically impossible. Next were two campus sites and a hospital. 60 acres to 200 acre sites.

Didn't work very well with them either. It's clear to see why looking at your map. But modern geoid models are making it better. However, they aren't perfect. Probably will never be.

Hang a plumb bob, in a plain. Then hang it near a big mountain. The mountain will pull sideways, toward the mountain.
Take a gravometer. Gravity is not the same everywhere. This piles up the water, where gravity is stronger. GPS works off a model, where it uses a General idea of where it thinks water would run... But it is not a perfect model. This, you have discrepancy between actual water flow, and what the GPS thinks.
Remember this: Sewer runs downhill. Pick up your check on Friday.
It's NOT just a smart alec comment.if the sewer does not run downhill, no paycheck.

I, personally, love calibration/localization. For the horizontal, it works great, and is fairly painless. For the vertical, it requires planning, judgement, and willingness to adapt.

Arthur,
If you intend to show your boss a new technique, you really need to work on communicating it. I'm not trying to pick on or one up you. Just pointing you to some important tidbits.
WGS84 is not a projection. It is an ellipsoid and a datum. While you can generate a plane tangent to it, it will not represent any portion of the earth (vertical or horizontal).
Holding one point vertical and running a geoid is a valid technique for some applications. Never a good idea if absolute vertical is important, but relative will be superior to nearly every inclined plane. You still need to check in to several other points.
Years back I was discussing a geodesic problem with a tech support guy at Leica. I explained my solution and asked if he had any suggestions how to express it. He responded 'read a book'. Ouch, but he was right. I took it to heart and read lots of books. With GPS and geodesy it's the best advise anyone can give.
Good luck, Tom

I don't want to get off topic here but with all the talk of sewer slopes. Has anyone seen a modern sewer crew without a pipe laser.

MightyMoe, post: 368456, member: 700 wrote: If you're using a Base/Rover setup then you don't do anything, set up on your elevation point and go. The file should be correctly projected, the geoid model applied and you hold the elevation of the point you are on.
Never, never, never calibrate, it's a bad idea, especially vertically. Now if you are in a network??
Don't use them so I don't know, but I would advise to use Base/Rover for tight site work anyway.

I agree with Moe. There is no reason to calibrate or localize today with the onboard software. Usles you are trying to cheat project, and do a quick and dirty job. When I get to a construction site where control is already established by others, be it in NAD83 or local coordinates. I always setup using the current SPC projection and latest GEOID12B. My base is always storing static data. I measure the control to the best of my ability using a combination of static, rtk, total station and digital level. Then I take the data back to the office (trailer or truck) and compare my measurements with the control data supplied to me. After analysis of Pdop, location, stability, and visual inspection, I can determine which points fit best. These are something your controller can NOT do. Even if the original control is said to be on SPC, I find very often it is not. Vertical control does need to be verified with on site benchmarks and seldom agrees with NAVD88 from an OPUS. Horizontal is usual closer to NAD83. Last week I was in southern Jersey where a project control was å±2ft north of NAD83 as per my check with a 4 hour OPUS and verified with post processing with CORS data. Had another project in central Georgia, again å±2ft from NAD83. In Georgia, the original surveyor explained to me they were using control from 30 years ago, and never checked it with OPUS or CORS data, until I brought it to their attention. In these cases I do have to stay on the local datum. Therefore I manually create the shift from local to WGS84 or grid. I can manually create a calibration or localization file using WGS84 lat/lon/ell and the residuals will be 0.00 as a result.

This method also allows me to remotely verify control on my clients base station, by collecting static data to be checked with OPUS and CORS on a a daily or weekly basis.