[PROJ] NTRIP-catalog

[Greg Troxel]

Mircea Neacsu via PROJ <proj at lists.osgeo.org> writes:

> I find your project a valuable initiative. I see it mostly as a
> verification tool, not as an automatic CRS selection tool. In most
> cases the CRS selection is imposed by other considerations and the
> NTRIP corrections have to match that selection.

I don't understand your comments at all.

Certainly, a program that uses coordinates is going to have a working
CRS, what we'd call "project CRS" in qgis.   That's a choice by the
project/program.

Then, using a GNSS receiver, it's going to get coordinates in some CRS.
Even without using NTRIP and injecting corrections/reference-data,
exactly what CRS is a difficult question.  Today, it could be
WGS84(G2296), or it could be some ITRF if using WAAS or EGNOS.  Once you
inject corrections, it could be quite a variety of CRSes.

I don't see how "the NTRIP corrections have to match".  What's available
is going to be constrained.  Around me, the only real option without
paying is EPSG:6319.  So then a program receiving coordinates in the
frame of the correction source, can transform into the project CRS.
That's pretty straightforward, and one just has to know the correction
source CRS.

Regardless of how you deal with the above, the key point is that one
wants to know, for a given source, what is the frame of the corrections.
That can be verification if you configure what you expect, and it can be
automation if you don't.

> In regard to other comments on the list: the way I see it, information
> about the CRS associated with a corrections stream should have been
> included in NTRIP protocol. The fact that now it exists in RTCM 3.4,
> is good but it's not the proper place. Once you have selected a
> corrections stream you can only check that corrections match your
> intended CRS. Besides, what if you need corrections in a different
> format (CMR or something else)? My view is that this information
> should be provided by the NTRIP casters.

NTRIP is simply a way to ask for a corrections stream.  The frame of a
stream is just as much a part of it as is the base station coordinates.
If you are getting a corrections stream not over NTRIP, because it's raw
TCP, or serial data radio, you still want to know the frame.  So I think
it absolutely belongs in RTCM.  People using CMR should ask their
proprietary equipment vendors to create a CMR+crs definition, but
probably the plan would be to figure out how to stop using CMR :)