5

I am attempting to re-project a vector shapefile from Natural Earth, following a qgis tutorial. Another user reported this error, but a solution has not yet been reported.

Given this post, and the last line of the error, I suspect that there is something wrong with the CRS setting, etc. Though, I have not been able to identify it.

   Export to vector file failed.
Error: Failed to transform a point while drawing a feature of type 'ne_10m_admin_0_map_units'. Writing stopped. (Exception: forward transform of
(3.141593, -1.570796)
PROJ.4: +proj=longlat +ellps=WGS84 +datum=WGS84 +no_defs +to +proj=merc +lon_0=0 +k=1 +x_0=0 +y_0=0 +ellps=WGS84 +datum=WGS84 +units=m +no_defs
Error: tolerance condition error)

I am using qgis 1.8.0 on Ubuntu 12.05. Thanks for the help!

Community
  • 1
ryanjdillon
  • 433
  • 1
  • 7
  • 19

2 Answers2

4

If you look at the extent of the layer, you get:

xMin,yMin -180;-90 : xMax,yMax 180;83.6341

Google Mercator projection is only valid between +/- 85.0511 degrees North/South, but the dataset contains also the south pole.

EDIT: The value of yMin=-90 forces the error in projecting to Google Mercator. The tutorial has a dataset that ends at -89.9998, which is nearly the same, but does not compute to inifity.

So you have to delete the objects around the south pole, to save the layer in Google Mercator projection.

AndreJ
  • 76,698
  • 5
  • 86
  • 162
  • Great to know. So 'Google Mercator' is the same as WGS 84 Mercator then? Thanks! – ryanjdillon Dec 15 '12 at 21:32
  • In double checking my method here, it looks like the projection I attempted to save to was in fact WGS 84 (EPSG:3395) and Google's Mercator is EPSG:900913. Is this still the issue, and where might I check the extent of each projection? – ryanjdillon Dec 15 '12 at 22:11
  • 2
    EPSG:3395, 900913 and 3857 are not exact the same, but use the same projecting method, where 85.0511 or arctan(sinh(π)) is the maximum bound to the north and south. For use with Google or OpenStreetMap, you should use EPSG:3857. You find the extent in the Layers properties in the Metadada tab. – AndreJ Dec 16 '12 at 06:40
1

The tutorial does not use ne_10m_land but 10m_admin_0_map_units.

The dataset used in the tutorial can be reprojected just fine. I'm not yet sure why this is the case because the working layer's extent is 

xMin,yMin -180,-89.9998 : xMax,yMax 180,83.6338

which would also be outside the bounds Andre posted.

underdark
  • 84,148
  • 21
  • 231
  • 413
  • I just gave it another shot with the ne_10m_admin_0_map_units dataset, and it is giving me the same error. I have updated the error in my post to reflect the correct dataset. – ryanjdillon Dec 15 '12 at 21:59
  • Hmmm... I have found the above mentioned dataset (also with an xMax,yMax 180,83.6341), but not the 10m_admin_0_map_units data that is mentioned in the tutorial. I thought that the ne must just stand for Natural Earth and was left off of the name of the file in the tutorial. – ryanjdillon Dec 15 '12 at 22:06
  • The file might be from a previous release. But you could have a look in Natural Earth's quickstart kit. – underdark Dec 15 '12 at 22:20
  • -89.9998 is not the same as -90, if you do certain mathematical functions... – AndreJ Dec 16 '12 at 07:21
  • 1
    Thanks underdark, it appears the Natural Earth quickstart kit contains the same files, so I suppose this error has developed with the new release of Natural Earth's data. – ryanjdillon Dec 16 '12 at 08:48