Trimble GPS Site Calibration

  I have always performed a site calibration by setting my Trimble TSC2 collector to "no projection / no datum",  then observing the points that I want to calibrate to with my Trimble 5700 reciever.  After I have tied my points (at least 4 of them), I then use the calibration routine in the TSC2.  This is the way I know how to do it, and I know it works.  I have a coworker who was taught  a different method of calibrating while using the same Trimble equipment.  He will calibrate to one observed point, then stakeout to the next calc point, once another point is found and observed with GPS, he then "adds" that to his calibration.  This sequence is repeated until enough points are found ( at least 4) to complete a site calibration with good residuals.  So on one site, he may be calibrating two, three, four or more times, whereas I have always performed only one calibration.  My question is this, is his method going to provide a correct calibration?  Will the data collector know how to correctly apply these calibrations?  Also, if his methods are correct, does it need to be processed in TGO any differently than how I would process my single calibration?  Any info will be greatly appreciated.

Calibrations or localizations of any kind are an accident waiting to happen. I can't think of a good excuse to use one.

So what do you do when you have collected "enough" points, do your calibration, and enough isn't "enough"?

I believe that each time you add a new point to the calibration the calculations are remade to include the whole data set. The end results should be equivilant either way. I would tend to prefer your partners way, because he will have a better idea when enough is enough sooner, whereas with your method you may be collecting extra points needlessly. Your mindset also seems to preclude adding extra points and discarding marginal ones. But 4 points would be a minimum, IMO, in either case.

FWIW, I agree with Norm. On the fly site calibrations are a pretty edgy way to go.

I'm with Norm, here.  Accidents waiting to happen. 

Why use an approximation (especially a floating one) when you can simply define your system with no error and go to work?  Use a defined translation instead of a calibration or localization.

JBS

 

At the risk of “ganging up” on this issue, I will throw my 2-bits in with the above responses.

 

I UNDERSTAND that a LOT of modern survey work (GPS and otherwise) is performed and expressed in the ancient “5k/5k on yonder hill (or monument), and aligned with whatever basis of bearing is handy” paradigm.

 

Although this is changing (slowly), I don't really expect everybody (or every project) to move into a georeferenced spatial paradigm (aligned with the NSRS) anytime soon. In many cases, it might be more of a PITA than is warranted by the project.

 

That said however, too many folks, make too many excuses for not taking the next step forward.

 

If you RTK Base Station is setup for longer than 15 minutes, there's really no reason NOT to get on the NSRS (in some form or another).

 

If you are using a VRS/RTN, then you are ALREADY there (or could be if you push the right buttons)!

 

If the project ALREADY has a georeferenced coordinate system, then why not use it?

 

A “calibration/localization” CAN work just fine...it can also blow up in your face, mislead you, and at best is a step backwards.

 

Loyal

Using Trimble tools, my solution is to build required custom projections in a coordinate system definition (CSD) file using Coordinate System Manager (CSM) in the office then upload the file to the data collector for the field crews.  Much easier to build and check prior to the field work being done.  You can define the projections anyway you like.  It also makes them easier to track, distribute and share.  The custom projections are available from the library on the data collector.  With multiple field crews on a project it helps immensely.  Only the specific projections you want the crew to use (custom or published) can be set to be available from the library.

Site calibrations work well with proper training and support of field and office staff, but building them ahead of time with CSM removes most of the headaches.  It also helps to tie them to a known geodetic reference frame as Loyal suggests.

Trust me....it is well worth learning how to do.

Scott

Aaron,
I'm assuming you are doing a "calibration" for your rover or machine control.  If that is correct, either way will work.  It is
"recalibrated" with each additional point.  After collecting all your points (I like at least 5 spread geometrically even around the site), you can "save" the calibration for future work.

I believe your co-workers routine helps if you do not know where the control points are.  starting with one and then 'staking" out the rest helps locate them on the ground.  I have used this method when the surveyor only gives you a coordinate list without a map to where the points are located.

  I really do appreciate all this feedback, and I would like to know what you all think about this hypothetical situation.  Let's start by saying a client has a large tract of land, and contracts me to stake only a small portion of their property line (could be ROW or property line).  The deeds or plats for the subject property, as well as the adjoining properties,  are calculated with assumed coordinates and loaded into the data collector before heading to the field.  Once in the field, I fire up the GPS and log into our VRS network.  I recover and observe all the necessary boundary corners that I can find.  I'll do a site calibration, and can now use the previously calculated points as stakeout points in order to find additional corners to check into, as well as use them for staking the ROW / property line.  For staking a small portion of ROW / property line, I wonder what everybody's thoughts are on this type of procedure.  This forum is full of knowledgeable people, and I look forward to the responses.  Thanks again.

Hi Aaron

That procedure would work, but with a couple of caveats.

1. You are presuming that the calculated coordinates are known accurately enough for the purposes of your survey.  If there are problems with the plan dimensions, you won't know until you are out there and check the residuals.  To paraphrase a quote from Animal Farm by George Orwell, 'all control points are equal, but some are more equal than others'.  Multiple point calibrations work best when the control is well determined. 

2. In my experience, property corners tend to be at places where RTK GPS may be problematic - in rural areas (at least in this part of the world), there are often trees or chain link fences along the property lines. Others might be under forest canopy completely. Some of the property corners you may be relying on for your calibration might be disturbed or obliterated. You won't know until you are at the site.

Just my $0.02 Canadian.

These are a couple of the reasons why - when possible, which is most of the time -  I like to build the custom projection before the field visit. Even for small jobs, I can build one and distribute it to the field crews in just a few minutes. 

A site calibration isreally nothing more than a user defined custom projection with a point of origin, an orientation, and a scale factor.

Scott

Set up your base with a HERE position (if using Trimble, other brands have their own nomenclature for the function). Tie the existing monumentation.

If you find and tie monuments at either end of your line to be staked, and they check, you hold 'em and go. If you don't then do the necessary boundary resolution in the dc and stake.

As a third option, I'd prefer taking precalc'd record coordinates and translate/rotate them onto the near-grid coordinates over a on-the fly site calibration. 

Hello Scott, 
    You are absolutely right that I was presuming that the corners were in areas favorable for GPS observations, but more often than not that is not the case.  Same thing with misclosures on deeds/plats, disturbed corners and the like.  I've often wondered what it would be like to have a project that didnt encounter some sort of challenges......i guess it could possibly get a bit boring if it always went easy out there.  In my opinion thats what makes Land Surveying great, having to find ways to resolve not just the boundaries, but also the different obstacles that lie in wait.  thanks again for everybody's input.

Guys first of all I want to say hi to all the American Surveyors...Good Ole Boys. I need you guys to help me with a question about offset from the bottom of the antenna to the phase center on a R8 type 2. At the bottom of the receivers it  saids the offset is:6.49 centimeters; but I've found a sketch Trimble that called that same offset to be 10.39 centimeters (see attached sketch). Does somebody knows exactly wichone is correct?. I do hydrographic surveying and I don't want to be 4 centimeters off all the times. I've noticed that the Carlson data collectors also called this offset 10.3 cm.

I'll appreciate any help here.

Jimmy Garcia.
Costa Rica/Central America.
(506)8346-8437.
survcollector@hotmail.com

 

Oh boy...this is a tricky question:

 

TRMR8 (Model 1?)

 

http://www.ngs.noaa.gov/ANTCAL/diagrams/TRMR8+NONE.gif

 

Trimble image (above) indicates 0.065 meters, NGS (relative) says 0.093 meters, NGS(absolute) says 0.0742 meters.

 

TRMR8_GNSS (Model 2)

 

http://www.ngs.noaa.gov/ANTCAL/diagrams/TRMR8_GNSS+NONE.gif

 

Trimble image (above) shows no data (but your image indicates 0.1039), NGS (relative) says 0.1039 meters, NGS(absolute) says 0.0851 meters.

 

So:

 

Trimble says 0.065 – 0.1039 = -0.0389 meters

NGS (relative) 0.093 – 0.1039 = -0.0109 meters

NGS (absolute) 0.0742 – 0.0851 = -0.0109 meters

 

NGS Absolute Calibration data can be found here:

 

http://www.ngs.noaa.gov/ANTCAL/images/ant_info.abs

 

Well BEFORE you do ANYTHING, make sure of what antenna that you are using (TRMR8 or TRMR8_GNSS)!

 

Which is right, and which is wrong?

 

I dunno... both Trimble and the NGS appear to agree on the TRMR8_GNSS, BUT NOT on the TRMR8.

 

But here's the kicker...IF you are using TRMR8s at both ends of the vector, it DOESN'T matter. IF you are mixing TRMR8s & TRMR8_GNSSs in the same vector, then it DOES matter. The same can be said for MOST antenna combinations.

 

I have “played” with this a LOT over the last year (Zephyr Base, TRMR8 rover), and the NGS (relative or absolute) values work the BEST! Of course I am comparing stations that have been post processed with either OPUS, or commercial software that also uses the NGS calibration data, with RTK [in-fill] data (using all three versions [combinations/estimates] of L1 phase center offsets).

 

Where this might get particularly TRICKY, is when you are using a Real-Time-Network (RTN). You may KNOW what antenna (and calibration data) that you are using, but what antenna (or antennas) is the RTN using...AND WHAT antenna calibration data is the RTN using. It probably doesn't matter IF the RTN data is based solely on the “actual” L1 phase center coordinate estimates (which I think it is), AND (BIG AND) your data collection software is using the RIGHT L1 phase center offset. I would submit that the NGS data will get you closer to “reality” than the vender data will in most all cases.

 

I just haven't played with RTN data enough to make a definite statement about just what goes on, but Bill Henning, Gavin Schrock, or Cliff Wilke (all of whom post here from time to time), will know the answers to the RTN question.

 

IF (maybe big if) RTN “vectors” work like “conventional” (single base) RTK (which I ASSUME they do), then the rubber meets the road in the Rover data collector, which computes the rover position based on the solved L1 Phase Center Coordinate, AND the Rod Height/L1 Phase Center data supplied by the user (or defaulted by the software in the case of the L1 PC offset).

 

Due to the inherent uncertainty in RTK [derived] heights, most folks (and some vendors) consider this issue to be trivial (and in some, if not many cases, it probably is). On the other hand, I HAVE satisfied myself that it is often NON-trivial, and is by definition, a systematic error (in MOST cases), which should be accounted for.

 

Now I'll put on my flak vest, and hide under my desk...

Loyal

Loyal,

I am under the impression that most RTN's are setup so that the user simply selects a null antenna as the reference station antenna.  I was also under the impression that most of the time the reference station antenna definition is somehow broadcast as part of the RTN correction messages so your receiver will use the right one.

Joe,

That is pretty much the ASSUMPTION that I have as well (see above concerning the L1 Phase Center). That does NOT however remove the ARP-L1_PC offset (at the Rover) from the equation. In the case of a “VRS” solution, there IS NO Antenna per se. BUT you still have to have an “accurate” L1_PC to GRP (monument) value (at the Rover) in the equation.

 

ALL GPS vectors (or positions...chicken or the egg) are INITIALLY derived from L1_PC to L1_PC. Static software (at least every package that I have seen) reduces the L1_PC to L1_PC vectors to GRP-GRP vectors (Mark to Mark), whereas RTK software/firmware (again in every case that I have seen) “returns the derived vector as L1_PC to L1_PC!

 

In a nut shell, here is [pretty much] what the data collector software does (behind the curtain) in the case of a “traditional” Single Base RTK solution.

 

First, the data collector software “solves” for the [geocentric] Cartesian X/Y/Z of the L1PC of the Antenna at the Base Station. This obviously INVOLVES using the Geodetic Coordinate of the Base Station Geodetic Reference Point (GRP) aka...monument, AND the “measured/input” HI + the ARP-L1PC “offset” unique to the Base Station Antenna. Regardless of what YOU entered into the Data Collector to define the Base Station Position (UTM, SPC, LDP, pushed the FUGARWE button, whatever), it all gets transformed to a [geocentric] Cartesian X/Y/Z coordinate in meters (or feet in some cases). Of course your selected coordinate projection and GEOID model factor in here as well.

 

Second, the geocentric Cartesian dX/dY/dZ vector that is solved by the receiver is added to this value. In the case of “conventional” single base RTK, this vector would be expressed in WGS84(G1150) because that's the “datum” of the broadcast orbits from the satellites. I think that most RTNs are using [ultra-rapid] IGS05 orbits, but for RTK purposes, there isn't much difference between WGS84(G1150), IGS05, or NAD83 vectors (due to relatively SHORT vectors). This returns the geocentric Cartesian X/Y/Z of the L1PC at the Rover Station Antenna.

 

Third, (last but not least) the data collector software “solves” for the [geocentric] Cartesian X/Y/Z Coordinate at the GRP of the Rover Station (based of course on the “measured” HI + the ARP-L1PC “offset” unique to the Rover station & Antenna).

 

Fourth, At this point...the Cartesian XYZ Coordinate is transformed to a [Datum specific] Latitude, Longitude, & Ellipsoid Height, the GEOID Model is applied, and whatever Coordinate Projection computations are performed. The result of this LAST computation (or series of computations) is what is burped out on the Data Collector Screen for you to look at.

 

So the “accuracy” of the height returned by this process, is highly dependent on the “accuracy” of the ARP-L1PC offsets used (and of course on whether or not you “measured” the HI [at both ends] accurately).

 

The only [real] difference (as near as I can tell) between the RTN/VRS Solution (in this discussion), and the Single Base Solution, is the omission of the First Step (because the L1_PC is the origin or the Coordinate Estimate to start with (and a geocentric Cartesian Coordinate), instead of the GRP of some point on yonder hill).

 

I may have over-simplified this a little, but [I think] the above is a reasonable accurate description of the goings on behind the curtain.

 

Note...I used “geocentric” above, even though NAD83 is not “geocentric” in the modern [ITRF] sense, it IS geocentric relative to the NAD83 Datum.

 

Loyal

Scott Partridge:
Using Trimble tools, my solution is to build required custom projections in a coordinate system definition (CSD) file using Coordinate System Manager (CSM) in the office then upload the file to the data collector for the field crews.  Much easier to build and check prior to the field work being done.  You can define the projections anyway you like.  It also makes them easier to track, distribute and share.  The custom projections are available from the library on the data collector.  With multiple field crews on a project it helps immensely.  Only the specific projections you want the crew to use (custom or published) can be set to be available from the library.

Site calibrations work well with proper training and support of field and office staff, but building them ahead of time with CSM removes most of the headaches.  It also helps to tie them to a known geodetic reference frame as Loyal suggests.

Trust me....it is well worth learning how to do.

Scott






 

Hi Scott

I have kind of a problem. I can't use that CSD file to calibrate my site. There seems to be no way to add it to my work order. I am using TSC3 with SCS900 Core and Road. The site calibration file needs to be DC file and there are no optios to bring CSD file into the mix. If you could help me with an advice it would be most appreciated.

Heigo

Calibration is a fine tool.  I have used this tool many time, but with reservations.  Ask yourself what does a calibration yield and how it effects your boundary.  Calibration scales and rotates to your found control from call.  After you calibrate, look at the project parameters. 
I realize that you can import your cad calcs into the collector and calibrate to these corrdinates which cuts your office time and a return usually to the field to stake out your points.
In most cases which is 99% of the time, I perfer to set up with an atonomous point, tie what I found during a reconnaissance of the site.  Once I have a few point I can stake-out the appropriate point using the plat or description call to find the next point if it exist.  At the end of the day, I have point tied that are surface and true bearings.  It is a little more work, but nothing is scaled or rotated. 
If I calibrate, I do use the process of collecting new points then re-calibrating.  Once the actual stake-out begins, you cannot do another calibration.  Your calc points will be scaled and rotated to fit the calibration.

my 1 1/2¢ worth....