[Qgis-user] Trimble GeoXT 2005 Accuracy

[Springfield Harrison]

Hi Greg,

Comments inserted below -

-----
Cheers, Spring


On 08/Mar/2021 08:36, Greg Troxel wrote:
> [I'm not sure how on-topic this is for qgis-user, but I'm guessing it's
> relevant enough, at least until a Moderator comment otherwise.]
You're right but I took my que from other GPS related threads and lack 
of response from Trimble.
> Springfield Harrison <stellargps at gmail.com> writes:
>
> To figure out what's going on this description needs to be tightened up,
> which is probably going to require trimble documentation or support to
> clearer (maybe the docs are fine; I haven't looked).
I'm familiar with the docs and no help from Trimble.
>> For many years my work flow has been:
>> Trimble Receiver + RTCM/SBAS ->
> This is unusual phrasing.  SBAS is a class of things, not a particular
> corrections source.  I'm guessing you are in North America and this
> really means WAAS.
>
> RTCM is a family of protocols for transporting corrections.  I am
> unclear on whether the WAAS format uses RTCM but I don't remember seeing
> it.  Within RTCM there is RTCM2, usually used for pseuorange
> corrections, and RTCM3, usually used for carrier phase reference data.
>
> So I wonder if the ProXR is doing RTK, what the correction source is,
> and what the distance to the base station is (or if it's VRS).
>
> To me the most important thing to be precise about is pseudorange
> solutions (often called navigation solutions) vs carrier phase solutions
> (post processed or RTK).  Then, there's PPP which is harder to
> understand.
>
> So I think you mean that while you were collecting points the receiver
> was calculating pseudorange solutions and using the pseudorange
> corrections delivered by WAAS.

Yes!  I use SBAS because Trimble does in TerrasSync but it does mean 
WAAS in southern BC, I think.  They don't use WAAS because the SBAS may 
use a different correction source elsewhere in the world.  I enable 
carrier measurements in TerraSync most of the time.

Again, following Trimble terminology, RTCM in this case is a marine 
beacon.  I have no source for RTK.

>> Pathfinder Office [+ RINEX Post Processing] -> SHP files -> GIS (QGIS
>> or Manifold GIS).
> You say "RINEX Post Processing" but that leaves out a ton of important
> details.   I am guessing you mean that the receiver records carrier
> phase observables, and then there was a double-differenced carrier phase
> solution for each ooccupation.   The obvious question is where the
> reference station data is coming from, how far away it is (or is it some
> VRS), and what datum the reference station coordiantes are in.
>
> Or maybe this is doing PPP.  There, similar questions apply about where
> the orbit data is coming from and what frame that is in.
I use the Pathfinder Office Post Processing engine which looks after the 
technical details.  I select a base station from their provided list, 
sorted by distance and it fetches the (RINEX?) files and applies the 
corrections.  I think the distance was about 30km.  PFO makes no 
provision for specifying or modifying the base station CRS.  It is 
encoded in the RINEX files.  It will utilize carrier phase data if it 
available in the rover file.
>>   The CRS is NAD83 UTM 10N throughout, for my home area at least.
> There are many realizations of NAD83.  Anything you did last year is
> likely in "NAD83(2011) epoch 2010.0" if you are in the US and firmly on
> the North American plate, and in Canada likely in NAD83(CSRS) apparently
> in a province-dependent epoch.  The recent realizations are close enough
> that neglecting this won't hurt too much, but if you are going to call
> 1m an error rather than a match, you should be careful about this.
My concern is not the absolute error but the systematic shift of all the 
points to the NW; the precision seems quite good, it is the accuracy 
that I question.  I think the different flavours of NAD83 UTM zone 10N 
differ by centimeters at most (?)
> If you are in an area with ground motion (e.g. Pacific Plate) then you
> have a lot more to do.
I am but how is this relevant, this is mm per year?
> The other big thing that's missing is what the trimble software is doing
> about datum transformations (or not, if the reference station
> coordinates are in NAD83(2011) as I'd expect).
Datum transformations where?  Their post processing process is "under 
the hood", QGIS datum transformations are much the same.  I have some 
faith (but not total) that they are being handled correctly.  This is 
supposed to be mapping, not geodesy.
>>      None of these steps offer any option to choose or modify the Base
>>        Station CRS so I don't think that would be the culprit in my NW
>>        data offset, although maybe I've missed something.
> I think it's necessary to understand what's going on inside the software.
Maybe, but that would be Trimble proprietary stuff, presumably. Their 
intent is to provide a streamlined workflow that incorporates all the 
myriad technical gymnastics but only exposes me to the NECESSARY 
details.  I'm sure that research grade differential software would 
reveal most of the operating parameters.
>>      Last fall I collected quite a few points in an attempt to
>>        quantify the problem, if that's what it is.  Here are some
>>        summaries:
> As Nicolas said, you are talking about "known" but have not explained
> where those coordinates came from and their expected errors.   I also
> can't figure out your color scheme, and whether you are measuring 2
> marks or one, etc.
I'm sorry but the presentation is complex and not overly clear. This was 
not intended for presentation so lacks certain clarity.  I can't give it 
much more time but do appreciate the time that you and others have 
spent.  Both MAP B and Map A are referencing one point (each).  
Positioning of the known source for Map B is from the orthophoto which 
is not always reliable, although I think it is in this case.  However, 
Map A uses the property cadastre for reference which does fit the 
orthophoto very well (actually the other way around).  Both situations 
are very comparable. *The thing to note is that the GeoXT points are all 
clumped closely together implying precision but are removed from the 
actual point implying lack of accuracy. * This is the puzzle.  Truly 
imprecise results would be more scattered.
> Data from the receiver in no-differential pseudorange solutio mode (not
> sure what "uncorrected" means) is going to be in WGS84(G1762), probably
> labeled as WGS84, which is different than NAD83(2011), and that needs a
> datum transform.  In QGIS, that often ends up with a null transform,
> which is wrong.  So there's a good question about what is going on
> within Pathfinder Office with datum transforms, which you need to find
> out from Trimble docs or support.

Uncorrected means just that, no real time corrections or post 
processing.  Wherever I have control over the CRS, (receiver, PFO export 
to SHP and QGIS) I specify NAD83 UTM zone 10N.  Beyond that, the 
translation is not in my hands.  I don't know exactly what PFO or QGIS 
are doing under the hood but make the best choices wherever they give me 
an option.

Thanks very much for these questions, a good learning experience for 
sure . . . .


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-user/attachments/20210308/d6712105/attachment.html>