Bug report #9070

+towgs84 wrongly ignored when OTF reprojection is on and project CRS is WGS84 Pseudo-Mercator (EPGS: 3857)

Added by Mathieu Pellerin - nIRV about 6 years ago. Updated about 6 years ago.

Status:Closed
Priority:Severe/Regression
Assignee:Marco Hugentobler
Category:Projection Support
Affected QGIS version:master Regression?:No
Operating System:all Easy fix?:No
Pull Request or Patch supplied:No Resolution:
Crashes QGIS or corrupts data:No Copied to github as #:17716

Description

Issue seems to have appeared following the (cool) datum transformation commits. Steps to reproduce:
1. Create a new project
2. Load the roads_partial-indian1960_48n shapefile
3. Open the project property, set OTF reprojection to on and the project CRS to EPSG:3857
4. Load the roads_partial-wgs84 shapefile
5. You'll notice a 500m shift between the two layers, when those two layers should be identically placed on top of each other

The 500m shift is what happens when the +towgs84 parameter is not applied (I learned that from an much older bug :) ). The reprojection issue here won't happen if you repeat the above steps but set the project projection to WGS84.

This is a pretty impactful issue here as it affects @the openlayers plugin (which sets the projection to EPGS:3857). Confirmed over here on two linux machines and windows.

roads_partial-indian1960_48n.zip (68 KB) Mathieu Pellerin - nIRV, 2013-11-15 11:14 PM

roads_partial-wgs84.zip.zip (65.2 KB) Mathieu Pellerin - nIRV, 2013-11-15 11:14 PM

Associated revisions

History

#1 Updated by Marco Hugentobler about 6 years ago

  • Assignee set to Marco Hugentobler

#2 Updated by Marco Hugentobler about 6 years ago

Maybe a problem with the null grid parameter that is removed now. Looking into that problem on Monday.
Do know which of the two layers is misplaced?

#3 Updated by Mathieu Pellerin - nIRV about 6 years ago

  • Target version changed from Version 2.0.0 to Future Release - High Priority

Marco,

I think the indian 1960 48n layer is the one being misplaced/misprojected, based on fact that if you load that shapefile, then add a openlayers' google satellite imagery layer, you'll notice the vector layer is clearly 500m to the southeast of where it should be. Steps to reproduce that:
0. Download and activate the openlayers plugin
1. Create a new project
2. Load the roads_partial-indian1960_48n shapefile
3. Add a google satellite imagery layer via openlayers (it'll activate OTF reprojection and change CRS to EPGS:3857
(If the google layer shows the whole world, just zoom in and out once)
4. You'll notice a 500m shift between the two layers, when the visible roads on the google layer should match the vector layer

#4 Updated by Marco Hugentobler about 6 years ago

  • Status changed from Open to Closed

#5 Updated by Mathieu Pellerin - nIRV about 6 years ago

Marco, thanks, glad this was uncovered and fixed early on :)

Also available in: Atom PDF