[Therion] Problem with new default settings and recent proj changes

[Martin Budaj]

On Tue, Apr 26, 2022 at 11:45 AM Bruce Mutton <bruce at tomo.co.nz> wrote:

> I am a little confused, as Therion automatically downloaded the grid and
> yet my therion.ini does not have any of the proj-auto or proj-missing-grid
> settings mentioned in the 14fac78 Therion Book pg 86-87, which raises for
> me many questions.  I assume the functionality of those ini statements is
> now superseded by the proj.ini that is now present in the install folders
> (as you described as "use the best transformation and download the grids if
> needed") and that the Therion Book is yet to be updated?
>

Indeed, thbook needs an update. Now, proj-auto is 'on' by default and
proj-missing-grid is 'download'.


> Or maybe I am not looking I the correct place for the ini files that
> Therion is actually using?
>

If the setting is missing in therion.ini, therion just uses the defaults.
You can always override them in the therion.ini file.


> Can Therion be run without having internet access (or without first having
> had internet access for a particular survey dataset at some previous time)?
>

Definitely. With the new defaults, it would attempt to download the grid
(only if proj can't find it locally) and end with an error message if there
is no internet connection.

You can do either of these:

* change the proj-missing-grid setting to warn, e.g. (this prints a warning
stating which grid you need to download manually, then download it and put
somewhere where proj finds it)
* set proj-auto off (this is the old behaviour where proj doesn't look for
the optimal transformation)
* get online, run therion once, and the grids would be reused in the
subsequent runs without internet connection


>  In the attached file, do the transformations marked [no] and [yes] relate
> to whether they are used or not?
>

All listed transformations are used. AoU: [no] or [yes] indicates, whether
proj optimised a transformation for your particular area (defined by the
fixed stations in your data set) [yes] or not [no].


> And presumably the accuracies listed are the estimated accuracies?
>

Yes, estimated by proj.


> Can I use proj-missing-grid to improve the accuracy of my example further
> or is 4m accuracy the best I am likely to get for this particular dataset?
>

Sure, the accuracy when using grids should be in centimeters. It's
interesting that in your log file the grid is downloaded (twice!) but not
actually used by proj, it would be worth investigating why.

Anyway, for cases when you think you can do better than proj (including
using a grid file), you can use 'cs-trans' to define a custom
transformation between two coordinate systems; see an example in thbook.
Probably your local GIS community has already defined such transformations;
in the case of Slovakia, our cartographic office published such a list for
QGIS users, which can be easily adopted for therion.

Martin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.speleo.sk/pipermail/therion/attachments/20220426/545865e4/attachment.htm>