Jump to content

 
Photo

Projection alignment issue between QGIS, NSD, and AI

- - - - - projections

  • Please log in to reply
4 replies to this topic

#1
David Medeiros

David Medeiros

    Hall of Fame

  • Moderator
  • PipPipPipPipPipPipPip
  • 1,311 posts
  • Gender:Male
  • Location:Redwood City CA
  • Interests:Cartography, wood working, wooden boats, fishing, camping, overland travel, exploring.
  • United States

I'm having a projection workflow issue with QGIS, Natural Scene Designer, and AI (w MAPublisher) I need some help with.

 

My workflow is: download GMTED in WGS 84 for Australia > open in GIS and re-project to World Miller Cylindrical (EPSG 54003) > clip out center section > open clipped Miller DEM in NSD > generate shaded relief and save with .tfw > open SR in AI for map compilation.

 

The issue I'm seeing is that when I get the shaded relief into AI with MAPub, its not lining up. If I take the same SR from NSD and open it side by side in AI and QGIS with natural earth land as a reference I can see that in QGIS the SR lines up correctly while in AI it's off. The AI map view environment is set to Miller Projection and all other projected data seems to show up where it should.

 

This is some kind of projection interpretation error, but I can't figure out where the error is occurring (NSD or QGIS). Each program seems to interpret the projection of the same DEM data differently:

 

QGIS: EPSG 54003 World Miller Cylindrical (0 central meridian, 0 Easting, 0 Northing, Sphere radius 0).
ArcGIS: Custom Miller Cylindrical (0 central meridian, 0 Easting, 0 Northing, Spherical radius not shown).
AI MAPub: World Miller Cylindrical, Custom no EPSG. (0 central meridian, 0 Easting, 0 Northing, Sphere radius 0).
NSD: Miller Cylindrical (0 central meridian, 0 Easting, 0 Northing, Sphere radius 6370997 meters).
 

When the same WGS DEM is opened in ArcGIS and projected into Miller I get yet another projection offset, but different from the QGIS one. I can't test the SR placement as this clipped TIF image wont open in my copy of NSD 6 (I think NSD 7 Pro will open it though).

 

Any help is appreciated!

Attached Thumbnails

  • WorldMiller DEM in QGIS.png
  • NSD SR in QGIS.png
  • NSD SR in AI.png

GIS Reference and Instruction Specialist, Stanford Geospatial Center.

www.mapbliss.com
@mapbliss

#2
Melita Kennedy

Melita Kennedy

    Key Contributor

  • Validated Member
  • PipPipPip
  • 99 posts
  • Gender:Not Telling
  • Location:Redlands, CA
  • United States

Unless the PROJ.4 math for Miller has changed recently, the algorithms look the same in the Esri projection engine. 

 

BTW, Esri:54003 uses WGS 1984 for the GCS, so the radius is the semimajor axis: 6378137.0 m. 

 

I expect that where the problem is--that the different definitions are using different radius values. The QGIS definition CANNOT be using radius = 0. You should try to see what the PROJ.4 string is.

 

If the data were directly north-south offset, I would definitely say it's a problem with different radii. It looks like there might be a smaller east-west component.

 

Melita



#3
David Medeiros

David Medeiros

    Hall of Fame

  • Moderator
  • PipPipPipPipPipPipPip
  • 1,311 posts
  • Gender:Male
  • Location:Redwood City CA
  • Interests:Cartography, wood working, wooden boats, fishing, camping, overland travel, exploring.
  • United States

Unless the PROJ.4 math for Miller has changed recently, the algorithms look the same in the Esri projection engine. 

 

BTW, Esri:54003 uses WGS 1984 for the GCS, so the radius is the semimajor axis: 6378137.0 m. 

 

I expect that where the problem is--that the different definitions are using different radius values. The QGIS definition CANNOT be using radius = 0. You should try to see what the PROJ.4 string is.

 

If the data were directly north-south offset, I would definitely say it's a problem with different radii. It looks like there might be a smaller east-west component.

 

Melita

 

I'll have to pull it up on my laptop to check again but looking at the details for the Miller Proj in QGIS I'm pretty sure all I saw were 0s listed for the variables. I didn't show it here, but when running the same WGS DEM through Arc, converting to Miller then rendering in NSD I get a similar offset shift, but in the opposite direction, SE of the correct position. It could be NSD that is not playing well with the various inputs. I'll have to pass this on to Brett.


GIS Reference and Instruction Specialist, Stanford Geospatial Center.

www.mapbliss.com
@mapbliss

#4
David Medeiros

David Medeiros

    Hall of Fame

  • Moderator
  • PipPipPipPipPipPipPip
  • 1,311 posts
  • Gender:Male
  • Location:Redwood City CA
  • Interests:Cartography, wood working, wooden boats, fishing, camping, overland travel, exploring.
  • United States

Here's the parameters I'm getting for Miller World Cylindrical in QGIS. Do these look correct?

Attached Thumbnails

  • Screen Shot 2016-11-24 at 7.56.40 AM.png

GIS Reference and Instruction Specialist, Stanford Geospatial Center.

www.mapbliss.com
@mapbliss

#5
Melita Kennedy

Melita Kennedy

    Key Contributor

  • Validated Member
  • PipPipPip
  • 99 posts
  • Gender:Not Telling
  • Location:Redlands, CA
  • United States

Sorry about the delay in responding. The +R_A option means that the software is calculating the authalic (equal surface area) radius from the WGS84 ellipsoid. The authalic radius is approximately 6371007.2 (EPSG uses 6371007.0). 

 

ArcGIS does have a "Miller Cylindrical (Auxiliary Sphere)" projection algorithm. It has another parameter which lets you set the sphere radius type. 0 is semi-major axis, 1 is semi-minor axis, 2 is authalic radius, 3 is use authalic radius and also convert geodetic latitudes to authalic latitudes).

 

Melita







Also tagged with one or more of these keywords: projections

0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users

-->