4

I'm an archaeologist and I'm struggling with the generation of a reference coordinate system for georeferencing the Spezialkarte der Osterreichisch-Ungarischen Monarchie (XIX century). I've found the parameters I attach to this message but I'm not able to create the correct proj4 definition.

This is a link to the map I'm trying to georeference.

Can anyone please help me?

Parameters

Improve this question
3
  • It doesn't have any ProjCRS tics that I could see so you'll have to georeference directly to Bohnenberger ellipsoid/whatever datum was being used. However, I've never seen any transformations published so you're better off trying to georeference directly to a modern datum/geographic coordinate reference system using known points.
    – mkennedy
    Commented Jul 14, 2016 at 0:07
  • 1
    I'm also not sure I agree with the references. Cliff Mugnier lists other information, and the map is dated 1882. This could be a difference between "Austria" and the empire though.
    – mkennedy
    Commented Jul 14, 2016 at 0:08
  • @mkennedy I found very useful your link to Cliff Mugnier document... would you be able to create a proj4 definition of the Austrian Militärgeographisches Institut Hermannskogel following the parameters reported in the article? I really have some problems understanding which parameters I need and how to write down the proj4 definition.
    – SaraZanni
    Commented Jul 15, 2016 at 9:47

4 Answers 4

Reset to default
2

Same as mkennedy, I suggest to georeference against a map of today, like OpenStreetMap or Google satellite. Road junctions, railway crossings and river bridges have not moved over the time. On the other hand, surveying has improved a lot. Keep in mind that they had no satellite imagery, and many things were just collected by sight.

The reference meridian of ferro never had a fixed point like Greenwich, and the location was adjusted over the centuries while geodesy improved.

I have georeferenced your map once against the coordinates in EPSG:4805 MGI(Ferro), and once against recognizable points from OpenStreetMap (EPSG:3857). They offset near the town of Sisek about 400m:

enter image description here

You can play with the +towgs84 values to get them together:

+proj=longlat +ellps=bessel +towgs84=0,0,-100,0,0,0,0 +pm=ferro +no_defs

but this does not fit across the map. The prime meridian of ferro is hardcoded 17°40'W. You could use the value given above, but need a different +towgs84 parameter set. Alignment does not improve though.

Reading through http://rcin.org.pl/Content/40957/WA51_58447_r2013-t86-no4_G-Polonica-Affek.pdf I think that your map is part of the third austrian military survey, which was not based on the Bohnenberger ellipsoid (as the second survey), but already on Bessel 1841. This leads more towards EPSG:4805 as GCS. The map projection is described as polyhedric.

This source could provide some details of the same map type: http://mapire.eu/static/pub/cadastral_timar_biszak.pdf

This page also gives some details converting towards WGS84: http://spatial-analyst.net/wiki/index.php?title=MGI_/_Balkans_coordinate_systems

Improve this answer
3
  • Thanks, @AndreJ indeed I opted out for using a modern Croatian WMS topographic map as basemap for georeferencing. The maps I'm using are not part of the 3rd Austrian Military Survey, it's another one, made with a different scale, but maybe the projection would be the same... I'll try and let you know. By the moment, georeferencing against a modern map is working quite well and I'm obtaining an errore lesser than 100 m.
    – SaraZanni
    Commented Jul 19, 2016 at 7:05
  • BTW you can view georeferenced version of the Austrian maps at mapire.eu/en; together with some useful publications on digitizing them.
    – AndreJ
    Commented Jul 20, 2016 at 18:31
  • Yes, but it's not the same map ;)
    – SaraZanni
    Commented Jul 22, 2016 at 6:15
2

As #mkennedy and @AndreJ suggested, I tried simply georeferencing each map of the Spezialkarte der Osterreichisch-Ungarischen Monarchie on a modern 1:25000 map of Croatia: the results are precise enough for my goals. I attach a picture so that I can give you an idea.

Georeferencing on modern map

I also tried using @Wernfried Domscheit projection, but the map is georeferenced in a very different place. I'd like to thank you all for your precious help!

Improve this answer
2

After some more work on the topic, since I found that georeferencing each map in an enough precise way through the identification of many GCPs was much time consuming, I came up with this last solution that is also applicable to other military survey maps of the Habsburg Empire. Using as SR EPSG:4883, I used the same North coordinated indicated on the sides of the map, while I calculated the East subtracting 17.4 degrees from the Ferro Meridian Coordinate. The result was a geo-referenced map with an error below 10 meters and it took me only 4 points (the corners) to finish the job.

Improve this answer
1

Have a look at this page: PROJ.4 - General Parameters

I think at least partly the projection string is this one:

+proj=latlong +pm=17d39'37.5"W +a=6376033 +b=6356354 +rf=324

I am not a geodesist, so I don't know if this definition is complete. But it could be a starting point.

btw, you can list available datums, projections, etc by

cs2cs -lP
cs2cs -le
cs2cs -lu
cs2cs -ld
cs2cs -lm

Also have a look at Datum Austria to get some EPSG Codes to start with.

Improve this answer
3
  • The projection definition is not complete, it misses the +towgs84 parameter. Unfortunately, noone has calculated it when WGS84 was established. The Datum Austria of 1901 is based on the bessel ellipsoid, and they put ferro at 17°40'W. Both not the same as this map.
    – AndreJ
    Commented Jul 17, 2016 at 6:16
  • 1
    I was expecting such comment. Just a note, the scale of your map let to a resolution of app. 10m per pixel, so an accuracy of 10m would be sufficient.
    – Wernfried Domscheit
    Commented Jul 17, 2016 at 13:40
  • Without +towgs84, it might be more than 100m. Apart from that, surveying in the mid 19th century was not as accurate as today.
    – AndreJ
    Commented Jul 17, 2016 at 13:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.