Re: my total miserable failed experiment with PPP and rtklib

[Greg Troxel]

You are getting into tricky territory, and will need to dig in to
understand a lot.

First, I think it's totally reasonable to try rtklib.   While there are
PPP services, they aren't, as far as I know, open source, and are thus
not someting you can replicate and understand.

rtlklib itself has a comlicated history, and you should be aware of
rtklibexplorer, and likely use their code.

Missing from your discussion was a mention of datum.   GPS itself will
give positions in WGS84(G2139) at least this week.  But as soon as you
use SBAS you get positions in the SBAS reference frame.  If you are in
.at, vs just using a .at email address, that is likly EGNOS and thus
probably some ITRF2008 or ITRF2014 -- but do look it up and tell us.

Then, for a reference station, questions arise:

  how stable is the antenna mounting?

  is it a geodetic-grade antenna

  is it dual frequency?

  how was the reference position determined? *

  what datum is the reference position in? **


* There is a dreadful culture of using several minutes of "survey in" to
  establish position.   That can be ok for some uses, if understood, but
  it is not ok for a public NTRIP feed.

** While you probably think everything is or should be in WGS84(G2139)
   that is not how professional surveying works.  There are
   national/regional datums that are much more closely tied to tectonic
   plates (often called plate-fixed even though they may not be entirely
   fixed), and these are used instead, because marks have much lower
   velocities in these frames.  In the US, that's NAD83(2011) epoch
   2010.0 -- which is about a meter different horizontally from WGS84.
   In Europe, I expect it's some realization of ETRS89.

   See
   https://en.wikipedia.org/wiki/European_Terrestrial_Reference_System_1989
   and realize that while it says that ETRS89 is a frame, I'd day it is
   a system, and there are frames which are realizations of the system.



If this message doesn't make sense, then I'd suggest reading about
geodesy.

As for distance, generally accuracy degrades with distance, but with
static positioning over 24h of data 100 km baselines should be ok, and
yes, the shorter the better.


Once you have some data, you should certainly process it every way you
can and compare, paying attention to datum.  PPP is one method, and then
there is rtklib with reference data.  In the US there is OPUS, which is
their carrier phase solution with their reference data.  I would not be
surprised to find similar in Europe.