Scott mentioned "treating WGS84 as a 'static' or 'plate-fixed' datum in the "+towgs84 approximation error" thread. I'll take that as my segue into the topic of dynamic datums and time dependent datum transformations.
There is a good introduction/overview of the topic in http://ceur-ws.org/Vol-1142/paper6.pdf. This was written from an Australian and New Zealand perspective, but the general issues are global. There are "International Case Studies" of USA and Great Britain.
Mike Craymer has written some nice accessible material on this topic as well. See, for example, http://www.naref.org/transf/nad83_hydroscan2006.pdf.
As a community, we have managed to largely ignore the issue of dynamic datums for a long time. As precise point positioning technology becomes more and more available, and more and more users are able to access cm level positions in the ITRF, we are not going to be able to continue to ignore this issue.
Generally, there are two classes of challenges that we face with.
Firstly, we need to store the epoch of measurement with any position that record - if that is not implicit in the definition of the reference frame being used. That isn't a trivial challenge, if our database and/or our software architecture assumes that a coordinate has only 3 dimensions.
Secondly, we need to use that epoch to perform a time dependent datum transformation.
In general, transforming coordinates from ITRF to a plate-fixed datum will require a 14 parameter datum transformation, and coordinate propagation to the reference epoch for that datum.
The 14 parameter tansformation is straight-forward (as long as you know the measurement epoch).
Coordinate propagation is not so straight-forward.
In GDA94 the coordinate propagation is combined with the 14 parameter datum transformation. But Australia is the "lucky country", geodetically speaking.
An Euler pole approach works well if you are not close to a plate boundary.
For places like California or New Zealand you need a more sophisticated velocity/displacement/distortion model.
So finally, a question for the Proj-4 community, and a plea to the wider geodetic community:
To the Proj-4 community: Are there plans to add time-dependent datum transformations to Proj-4? (Or is that support already there?)
To the wider geodetic community: Are we able to work together on developing/defining standards for representing and sharing velocity/displacement/distortion models, rather than having a "tower of Babel" where every national authority develops and publishes a slightly different model, and every software developer has to try to implement that model.
Proj mailing list
Le 24/03/2017 à 02:09, Nick Mein a écrit :
I'm not sure if this address your question about sharing the
models, but just in case: ISO 19111 is under revision right now,
and one of the main goal for this revision is to support dynamic
datum in their conceptual model. The draft is quite advanced and
the standard may be published in 2018 if everything go well. After
its publication, there is plan to update the ISO 19162 standard
(WKT 2) accordingly.
Proj mailing list
In reply to this post by Nick Mein
Let me answer you last question:
> To the Proj-4 community: Are there plans to add time-dependent datum transformations to Proj-4? (Or is that support already there?)
Yes, and (to some extent) yes. With the introduction of “transformation pipelines” and the new API (previously discussed here on the mailing list and GitHub) in the development version of PROJ.4 it is possible to implement time-dependent datum transformations. The framework that allows you to implement various time-dependent transformations is there, but not much has been implemented yet.
Within the Nordic Commision for Geodesy I am involved in a project about implementing a dynamic reference frame in Iceland, so this is something that is very much in focus on my end. As time allow it, I am working on adding time-dependent transformations to PROJ.4. First of will be extending the Helmert transformation so it also handles 14-parameter shifts. That should take care of most of your needs for ITRF transformations I imagine. Another thing I am working on is time-depending deformation grids. If you are interested in the progress of this pay attention the GitHub page.
> As a community, we have managed to largely ignore the issue of dynamic datums for a long time. As precise point positioning technology becomes more and > more available, and more and more users are able to access cm level positions in the ITRF, we are not going to be able to continue to ignore this issue.
I agree. We are taking the first steps in PROJ.4 now, but there is still a huge implementation effort waiting in downstream projects!
Proj mailing list
|Free forum by Nabble||Edit this page|