[gdal-dev] Discrepancy in proj values between ubuntu-small and ubuntu-full images

Even Rouault even.rouault at spatialys.com
Mon Jan 10 06:35:44 PST 2022


Rajsekar,

One of the reason that the -full image is larger than the -small one is 
that the former includes > 500 MB of geodetic grids, whereas the later 
only includes very basic ones. (I've just amended 
https://github.com/OSGeo/gdal/blob/master/docker/README.md to make it 
clearer)

As you do a WGS 84 to NAD83 transformation, the ubuntu-full image can 
use the HPGN grids, and in particular the Pennsylviana grid

$ echo -77.75 40.16666666666666  | PROJ_DEBUG=2 bin/cs2cs 
'+init=epsg:4326' '+to' '+init=epsg:2271'
[ ... snip ... ]
Using coordinate operation axis order change (2D) + Inverse of NAD83 to 
WGS 84 (57) + SPCS83 Pennsylvania North zone (US Survey feet)
pj_open_lib(us_noaa_pahpgn.tif): call 
fopen(/home/even/proj/PROJ-data/us_noaa/us_noaa_pahpgn.tif) - succeeded
1968499.39    -0.15 0.00

Whereas the -small image has no such grid and will assume that NAD83 == 
WGS84.

In theory the result with the -full image should be more accurate, but 
it depends if somewhere in your workflow someone has not assumed that 
NAD83 == WGS84 ...

Anyway WGS 84 is literaly a moving and somewhat fuzzy target, so you 
shouldn't expect in general results to have an accuracy better than ~ 2m 
in any transformation involving it.

Even

Le 10/01/2022 à 15:21, Rajsekar Manokaran a écrit :
> Hi gdal devs and users,
>
> I came across a discrepancy in proj. values between ubuntu-small and 
> ubuntu-full docker images even for the same gdal version tag.  I tried 
> to convert a few coords from epsg:4326 to epsg:2271.  The coords are:
>
> Input:
> -77.75 40.16666666666666  # this is the center meridian, lat of 2271
> -75.18094433657971 40.25904959199275  # another sample point
>
> The outputs of all ubuntu-small-* images seem to be correct, but the 
> ubuntu-full-* images have about 1 feet error (2271 uses US_SURVEY_FEET 
> units).
>
> Here's how I run it:
>
> $ docker run --rm -i osgeo/gdal:"$tag" cs2cs '+init=epsg:4326' '+to' 
> '+init=epsg:2271' < input
> where tag is the docker tag: "ubuntu-full-3.2.2", or 
> "ubuntu-small-3.4.1", etc.
>
> Here are the outputs for the ubuntu-full images:
>
> output-ubuntu-full-3.2.2
> 1968499.39      -0.15 0.00
> 2685535.97      44295.90 0.00
> output-ubuntu-full-3.3.3
> 1968499.39      -0.15 0.00
> 2685535.97      44295.90 0.00
> output-ubuntu-full-3.4.1
> 1968499.39      -0.15 0.00
> 2685535.97      44295.90 0.00
>
> Here are the outputs for the small images:
>
> output-ubuntu-small-3.2.2
> 1968500.00      -0.00 0.00
> 2685536.18      44296.27 0.00
> output-ubuntu-small-3.3.3
> 1968500.00      -0.00 0.00
> 2685536.18      44296.27 0.00
> output-ubuntu-small-3.4.1
> 1968500.00      -0.00 0.00
> 2685536.18      44296.27 0.00
>
> The first input is the center and so should translate to "(false east, 
> false north)".  Going by that, the small versions seem to be more 
> correct.  Any idea why this difference in the values?
>
> -
> Rajsekar
>
>
>
> _______________________________________________
> gdal-dev mailing list
> gdal-dev at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/gdal-dev

-- 
http://www.spatialys.com
My software is free, but my time generally not.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20220110/9976a770/attachment-0001.html>


More information about the gdal-dev mailing list