7 replies to this topic

[David Medeiros]

Attempting to transform WGS 84 to Winkel II. When I do this in MAPub 8.4 I get the issue shown below, where a portion of Eastern Russian is moved in the wrong direction relative to central meridian. Tried to remedy by setting meridian at -169 (between Russia and Alaska) but get even more of these cross overs. I seem to recall a discussion of this problem with MAPub but can't find the posts. Any ideas on how to fix?

Attached Files

•  Screen_shot_2011_03_14_at_12.08.12_PM.png   148.18K   22 downloads

[Gord D]

Hi David,

You indicated you were going to supply a screenshot. I'd be interested to see the extent of the crossover line work you are seeing and the movement of central Russia you are describing. If you would like to have Avenza Support have a look at this issue please send your dataset along to us and we can come up with a workaround.

[David Medeiros]

Hi David,

You indicated you were going to supply a screenshot. I'd be interested to see the extent of the crossover line work you are seeing and the movement of central Russia you are describing. If you would like to have Avenza Support have a look at this issue please send your dataset along to us and we can come up with a workaround.

Good grief, this is what happens when you try to do work with a cold! Sorry about that, here's the screen shot of the cross over line work. On re projection a portion of eastern Siberia crosses over to the western edge of the map. The data set is the 110m Land from Natural Earth.

I know I've read about why this happens in the past but couldn't find the post. I thought I remembered reading that Avenza had fixed this issue with the transformation tool at some point?

Attached Files

•  Screen_shot_2011_03_14_at_12.08.12_PM.png   148.18K   19 downloads

[Andrew Patterson]

I know I've read about why this happens in the past but couldn't find the post. I thought I remembered reading that Avenza had fixed this issue with the transformation tool at some point?

I'm the one who 'fixed' it I actually rewrote our transformation engine so it supported splitting polygons when appropriate.

When I finished, the most common problem I got was exactly the one you're seeing now. We used to get much worse results of course, but there are still some examples of the one you've managed to catch. I'm trying to remember what the underlying cause was -- projection envelope boundaries I think -- and more importantly, whether there was a work around. As Gord said though, if there is, Support will help you find it

Either way, if you can help Support replicate it I'll make sure it gets filed to the appropriate person (I handled the Illustrator end; I have no idea how to do the actual projection math). Given the amount of time I spent on the new projection engine, I'd like it to work as well as possible!

[David Medeiros]

Andrew,

I just tried to re create the error using a clean data set and a new AI doc, but this time it worked just fine. I took the exact same steps: add data layer (110m_land shape-file from Natural Earth); double click MapView; select "perform coordinate transformation; select Winkel II from Projected > World; hit ok.

The only difference I can see here is that in my first attempt I added the 110m_admin_0_boundaries and the 110m_lakes shape-files along with the land dataset and performed the transformation on the Map View with multiple data sets (something I do all the time). Could be user error, but I can't see what I would have done apart form those steps outlined above.

Attached Files

•  Screen_shot_2011_03_15_at_11.55.11_AM.png   64.49K   11 downloads

[cartdeco]

David,

This issue happens when your data extends outside of the 180W - 180E envelope, even if it is just by a fraction of a degree. If you follow the thread on the Natural Earth Data forums - http://www.naturalea...topic.php?id=81

The issue appears to have been fixed in the 1.3 release. I occasionally get this happening when you change the central meridian to something other than 0. Crazy things happen when you choose a global projection that is not cylindrical, and it's difficult to correct. One method I use is to adjust the features that cross the 180 meridian (eastern Russia and Fiji) first, then reproject.

Hope this helps.

Craig

[kay]

I have seen this issue before. Like cartodeco says, the data extent may be exceeding +/- 180 degrees. You can check this by using the MAP Location tool or Advanced Import dialog window (you will have to select the option "" in the MAP Preference to see the extent of the dataset: MAP Preference > Import Map Data > select the option "show extents in Advanced Import table). If it is exceeding 180 degrees, you can crop it with MAP Vector crop tool, then transform. The result will be nice and good!

I hope this helps,
kay

[Andrew Patterson]

This issue happens when your data extends outside of the 180W - 180E envelope, even if it is just by a fraction of a degree.

That's my memory of the problem too. I think I remember that if I scaled the map to 99% (using Illustrator) and tried it, no problems. Obviously that's not a solution, but it did demonstrate that it's a boundary problem. A proper solution is, of course, that it should be more tolerant along the boundaries, particularly when its that close. That's something we've been trying to work into the transformation engine for a while, but its an ongoing project.

It probably worked the second time because the boundaries came in just a smidgen somehow and solved the boundary case. I wouldn't classify any of this as user error though