CDED (usgs) wrong extents north of 68°

[matt.wilkie@…]

The cded export screws up north of 68° and exports the west half twice.

gdal_translate --optfile yt_cded.opt -co NTS=117d12e \
-co InternalName=117d12DEMe dem 117d12_e.dem

gdal_translate --optfile yt_cded.opt -co NTS=117d12w \
-co InternalName=117d12DEMw dem 117d12_w.dem

East:
Upper Left  (-140.0002083,  69.7501042) (140d 0'0.75"W, 69d45'0.38"N)
Lower Left  (-140.0002083,  69.4998958) (140d 0'0.75"W, 69d29'59.63"N)
Upper Right (-139.4997917,  69.7501042) (139d29'59.25"W, 69d45'0.38"N)
Lower Right (-139.4997917,  69.4998958) (139d29'59.25"W, 69d29'59.63"N)

West:
Upper Left  (-140.0002083,  69.7501042) (140d 0'0.75"W, 69d45'0.38"N)
Lower Left  (-140.0002083,  69.4998958) (140d 0'0.75"W, 69d29'59.63"N)
Upper Right (-139.4997917,  69.7501042) (139d29'59.25"W, 69d45'0.38"N)
Lower Right (-139.4997917,  69.4998958) (139d29'59.25"W, 69d29'59.63"N)

In this example east top left should be 139.5 (plus/minus the cell offset)

[matt.wilkie@…]

>bump<

[maphew]

I've forgotten the rationale for why this is "wontfix". Can you refresh my memory? thanks.

[warmerdam]

Matt,

I don't think this is wontfix - it is just not followed up on. If you can provide a dataset, and control file that goes with the above commands, I'll try and look into this soon.

[maphew]

option file used for the posted commandline

[maphew]

test case showing bug: source data & script

[rouault]

Although I'm totally unfamiliar with CDED, I've attached a patch that probably fixes the issue.

With it, gdalinfo reports :

gdalinfo 117d12_e.dem :
Upper Left  (-139.5002083,  69.7501042) (139d30'0.75"W, 69d45'0.38"N)
Lower Left  (-139.5002083,  69.4998958) (139d30'0.75"W, 69d29'59.63"N)
Upper Right (-138.9997917,  69.7501042) (138d59'59.25"W, 69d45'0.38"N)
Lower Right (-138.9997917,  69.4998958) (138d59'59.25"W, 69d29'59.63"N)

gdalinfo 117d12_w.dem :
Upper Left  (-140.0002083,  69.7501042) (140d 0'0.75"W, 69d45'0.38"N)
Lower Left  (-140.0002083,  69.4998958) (140d 0'0.75"W, 69d29'59.63"N)
Upper Right (-139.4997917,  69.7501042) (139d29'59.25"W, 69d45'0.38"N)
Lower Right (-139.4997917,  69.4998958) (139d29'59.25"W, 69d29'59.63"N)

[warmerdam]

I have reviewed and applied Even's patch which *seems* reasonable. Matt, would you be able to test this out (from trunk)? If so, I would be willing to back patch it to 1.4 branch after testing.

Patch was r12320.

[maphew]

I can't test from code right now. My linux machine at work is offline and cygwin does not compile out of the box. I'll try from home this weekend, kids and life permitting. :)

[maphew]

The patch seems to have fixed the problem, thank you!

There is a warning message I don't recall seeing before, "Warning 6: 'CDED50K' is an unexpected value for Product creation option of type string-select.". I don't have an old version to test if my memory is correct or not. It appears harmless as diff'ing the output of today's gdal_translate and the hand crafted 117d12_e.dem from a couple of years ago shows no difference.

[warmerdam]

Matt,

Even recently introduced validation of creation parameters based on the driver metadata describing the available creation parameters. But it seems there was a small typo in this metadata for the USGS DEM driver. I have corrected this (r12593). In the meantime there is no problem, you can just ignore the warning.

Thanks