[Proj] *SOLVED*: Creating a nadgrid shift file with GDAL

[Jan Hartmann]

On 01/21/2013 08:21 PM, Hermann Peifer wrote:
>
> I just successfully tested your procedure using the 31 test points 
> given in BETA2007testdaten.csv [1]. And indeed: it is somewhat 
> puzzling that one has to do step 8. I "solved" the issue by defining 
> the grid's Y extents as: gdal_grid (...) -tye ymax ymin # rather than 
> providing the values in the documented order: -tye ymin ymax. Wild 
> guess: This looks to me like a bug in gdal_grid.cpp, perhaps around 
> line 1000.
>
> Hermann
>
> [1] 
> http://crs.bkg.bund.de/crseu/crs/descrtrans/BeTA/BETA2007testdaten.csv
>
Good to hear!

It would be interesting to compare the grid you produced with the 
official BETA2007 grid. From what I know, those grids are made from a 
few real measurements, that are interpolated to a grid. Of course there 
are many interpolation methods. The default one for gdal_grid is inverse 
distance squared, but more advanced geostatistical methods (kriging?)  
could be more appropriate for these data. As we are talking about 
accuracies in the centimeter domain, the choice of method could be 
important. With these gdal procedures, testing them has become really 
simple.

Jan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/proj/attachments/20130122/056587c1/attachment.html>