PDA

View Full Version : Best projection for overlapping regional maps?



laevex_esre
09-28-2011, 01:37 PM
I've almost finished creating my world map, here (http://www.cartographersguild.com/showthread.php?15832-World-Creation-WIP-Etarek).

Next I want to get working on mapping each region of the globe in some detail. I've tried this before and found that any overlap between such maps is extremely difficult. Is there a good projection to use for this purpose? Almost every projection type distorts the geology as far as I can tell, making it very difficult to overlap maps. I'd love to know if there's an answer to this.

Or am I doomed to painstakingly redrawing all of the cities and roads of my maps?

feanaaro
09-28-2011, 02:08 PM
isn't the classical mercator, or perhaps equirectangular, good for this purpose?

laevex_esre
09-28-2011, 03:48 PM
Equirectangular distorts distances quite a lot from what I can gather off the internet. Mercator might not distort so much, but atlases usually use Lamber Azimuthal Equal-Area projections these days. The problem with these is they won't overlap very well. Maybe Mercator is my only shot?

moutarde
09-28-2011, 05:22 PM
All map projections distort distance to some degree. Do a wikipedia search for map projection, or ask Hai-Etlik or Cantab on these forums for advice.

cantab
09-28-2011, 06:09 PM
Well two different map projections, in general, aren't going to line up - if they did then they'd be the same projection.

gdalwarp will translate images from one projection to another. However, using it from the command line may be tricky. QGIS lets you call it and then you can select projections from a list of standards. (You may need a plugin, I'm not sure.)

What sort of size regions do you plan on working on?

Hai-Etlik
09-28-2011, 11:33 PM
To get maps to "line up" that way, you need to be using the same projection for all of them, and pretty much the only projection that looks any good "zoomed in" that way is Normal Mercator, which is still far from perfect. That's why real maps are done in projections appropriate to their extent and purpose.

You might want to look into vector graphics, and vector GIS tools if you want to be able to reproject data and then symbolize it. Inkscape is a good vector graphics editor, and for a GIS you might try QuantumGIS, uDig, or OpenJUMP.

cfds
09-30-2011, 05:24 AM
Distortion is always a function of the extent of the map relative to the globe. You can map a whole planet with lots and lots of small equirectangular or Mercator maps, centered on the region in question. They will never overlap perfectly but good enough.

laevex_esre
09-30-2011, 01:52 PM
OK. That's all interesting. How do you guys think HandsomeRob did the overlap in step 4 of this tutorial (http://www.cartographersguild.com/showthread.php?1577-Award-Winner-Atlas-Walkthrough-Fractal-Terrains-amp-Illustrator) then? He was using a Lambert Azimuthal Equal-Area projection, which is quite a nice one I think.

cantab
10-02-2011, 10:03 PM
I think all his maps are just in the exact same projection. That's suitable for mapping regions of a continent, but it may not be ideal for a whole world. (It might be OK if you don't want regional maps of the poles.)

LAEA is good close to the projection centre, distortion being about 2% within 15 degrees, 5% within 30 degrees, and 10% within 45 degrees. The further out one gets, the more squashed the map becomes; by the time you reach 90 degrees (half the planet) it's up at 25%, equivalent to turning a square into the shape of a widescreen TV.

laevex_esre
10-03-2011, 01:19 PM
If HandsomeRob's regional maps are from the same projection, why don't bits of them look horribly distorted? His continents are huge.

cantab
10-03-2011, 06:20 PM
One guess is that he's done all his work in the projection, so of course they won't look distorted on the map, but the shapes would be different on the globe.

Redrobes
10-03-2011, 06:33 PM
He released a KMZ file for Google Earth if I recall.

Handsome Rob knew a lot about the projections and made the map in the projection before starting. He chose the equal area because then he had not savagely distorted polar regions. When viewed as discreet maps these all looked great but they didn't fit together as a 2D sheet - I tried ! However they were accurate to the projection so that when mapped to a sphere they did fit to it perfectly. He mapped the whole globe and had a google earth interface to his world. It was possibly the most impressive atlas style but still unreal map we had here. The attention to detail was superb. He was a day job pro atlas mapper by trade so thats the level of the bar that is set for doing that kind of work.

Edit: Apologies, it was a KML File (http://sorol.wikispaces.com/space/showimage/Sorol.kml) he made from his web site Here (http://sorol.wikispaces.com)

laevex_esre
10-04-2011, 01:21 PM
I saw the KML file. In fact, Sorol was a major inspiration for what I plan to do with Etarek. I understand that he used a lot of projections and joined them together on Google Earth, but how did he manage to import all of the overlapping bits of adjacent maps into projections next to them (my original question, step 4 in his tutorial (http://www.cartographersguild.com/showthread.php?1577-Award-Winner-Atlas-Walkthrough-Fractal-Terrains-amp-Illustrator))

Hai-Etlik
10-04-2011, 04:13 PM
I saw the KML file. In fact, Sorol was a major inspiration for what I plan to do with Etarek. I understand that he used a lot of projections and joined them together on Google Earth, but how did he manage to import all of the overlapping bits of adjacent maps into projections next to them (my original question, step 4 in his tutorial (http://www.cartographersguild.com/showthread.php?1577-Award-Winner-Atlas-Walkthrough-Fractal-Terrains-amp-Illustrator))

Presumably he reprojected so they were in a common projection.

Ideally you would have raw, georeferenced data. DEMs, and raw vector data in formats like GeoTIFF, GML, or Shapefile. This can then be projected as desired using standard GIS tools and then symbolized to get a map. Taking a symbolized graphic and trying to reproject it tends to be messy. It has its uses but won't generally make for an attractive result.

laevex_esre
10-05-2011, 06:09 AM
Presumably he reprojected so they were in a common projection.

Does that mean he opened the previous projection on FTPro and just offset it on the X and Y axis rather than creating a new centre for the projection? Surely that would distort the maps hugely.

Hai-Etlik
10-06-2011, 02:37 AM
Does that mean he opened the previous projection on FTPro and just offset it on the X and Y axis rather than creating a new centre for the projection? Surely that would distort the maps hugely.

Maybe he was using the same projection for the whole set of maps.