You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
…TERION_PARTIAL_INTERSECTION if area is specified"
This reverts commit ebe3425.
It was found to break gdalwarp usage in
OSGeo/gdal#3695 when passing a bbox that is
quite large.
Expected behavior and actual behavior.
gdalwarp incorrectly re-projects data in EPSG:27700 to EPSG:3857.
There is 100m offset when re-projecting data from British National Grid to Web Mercator.
I believe the problem was introduced with Proj-8.
Steps to reproduce the problem.
Take the following WMS source xml as an example data source in 27700
To test with
I run the following
Then inside the container /app
reprojects the tile incorrectly
To test with
I ran the following
Then inside the container /app
reprojects the tile correctly
I also tested on my Debian system, with the ubuntugis-stable PPA
Pop!_OS 20.10
Release: 20.10
Codename: groovy
GDAL 3.1.3, released 2020/09/01
Proj Rel. 7.1.0, August 1st, 2020
I've also tested this with a geotiff in epsg:27700 incase there was an issue in the WMS/TMS driver and the error in reprojection remains.
Testing cs2cs
I also tested proj with cs2cs to see if the issue was there. Interestingly
Proj Rel. 7.1.0, August 1st, 2020
GDAL 3.1.3, released 2020/09/01
Result = -286180.06 6703828.59 0.00
Proj Rel. 8.0.0, March 1st, 2021
GDAL 3.2.1, released 2020/12/29
Result = -286180.20 6703828.38 0.00
There is a difference here but not 100m, I do note that the mercator projection was made more accurate in Proj 8 which could account for this.
The text was updated successfully, but these errors were encountered: