[Imports] Import of coastlines for Greenland

Karl Zinglersen karl.zinglersen at gmail.com
Wed Jul 29 00:23:33 UTC 2015


Hi Chris Hormann,

Thank you for the involvement in this question - and experience with in
coastline mapping.

Concerning the GIMP license, I had an idea, that this could be an issue. I
can ask for direct permission for OSM use from Ian Howat. The vector data
is derived/vectorized from the GIMP raster data by myself as an OSM
contributor , so I suggest the citation to GIMP could be in a source tag
equivalent to e.g. Bing.

On the data quality, I agree, that particularly a portion of North East
Greenland near Daneborg is filled with problems, possibly because of ice
conditions, shadows and more. Unhappily this is more or less the same
regions as the GADM coastlines. Otherwise I feel, that the coastline is
more reasonable in other areas, but I'll happily test it against Landsat
imagery along OSM/PGS - but they would be more or less of the same source,
namely Landsat7 imagery around year 2000, so somewhat outdated. I recall
reading your article a couple of years ago on global shorelines and
vectorization/classification

Positioning: I am afraid that many images, also from DigitalGlobe via Bing
are badly positioned in Greenland. A couple of months ago, I realized a new
hi-res satellite image for Nuuk instead of the older aerial image from
Asiaq Greenland Survey. Furthermore the rest of the Asiaq aerials had
disappeared also. The new imagery was in some places 30 m misplaced. I had
to contact Wolfgang Walcher from Microsoft to bring back the Asiaq aerials
(some 15 years old), I delivered to them some years ago, so Bing imagery
was aligned with OSM vectors and the real world in the towns and
settlements. I have promised Wolfgang to make a list of DigitalGlobe scenes
to be included in Bing - to cover the less covered wilderness areas.

Greenland mapping authorities: The mapping of towns and settlements are
done by Asiaq Greenland Survey, my former employer and a governmental body,
but with commercial activities and map store - still like the old days of
Ordnance Survey in the UK and KMS in Denmark. I and my former colleagues
have for a long time worked and aspired to change this with no luck so far.
The mapping of topography is done by KMS now GST (Geodata Agency) of
Denmark. They haven't done official topographic mapping since the 1930's
more or less, but have now started in small areas which will never cover
the whole country. Topograhic data is now freely available in Denmark, but
the law is not working for Greenland. I have struggled for 2-3 years to
change this with little luck.

Placenames across Greenland are managed by Oqaasileriffik, the Language
Secretary, also a governmental body, whom I know very well. I could
certainly explain and ask for openess and OSM use of their quite
comprehensive data set. I got them started mapping placenames just around 6
years ago, and now they have travelled the country and interviewed local
eldermen on the naming - and the number of recordings has doubled or
tripled. However, North East Greenland is only sparsely covered, as many
are foreign (Danish, English, Norwegian etc.) and of less importance to the
inhabitants of Greenland. The Geological Survey of Denmark and Greenland
(GEUS) has made a comprehensive mapping of placenames the recent years. I
could ask, but I am less confident in getting permission for use for OSM
based on previous experiences on their map of North East Greenland in
1.100.000. We could give it a go in collaboration?
Online map: http://geuskort.geus.dk/stednavnedb/
Online report: http://www.geus.dk/publications/bull/nr21/nr21_p001-368.pdf
Information is copyright to GEUS, but can be cited...

I regularly meet with both the technical and administrative persons of
mapping in Greenland, being a part of this previously myself. I'll move on
with the "free our data" part - but I could use some good advice on how to
approach this issue such as use cases, growth in public use, less
administrative work etc.

Chris, how do you suggest I, and hopefully others, should proceed for
Greenland? Using new Landsat8 imagery for classification? Follow up on
getting Microsoft to add more WorldView scenes to the map service, and map
manually? Digitizing coastlines on the baselayer of Bing with only Landsat7
mosaic is probably not recommended.

Karl

2015-07-28 17:10 GMT-02:00 Christoph Hormann <chris_hormann at gmx.de>:

>
> First of all the license situation of GIMP is somewhat unclear, OSM does
> not qualify as academic purpose and it is not clear if the citation OSM
> can provide on the contributors page on the wiki is sufficient - so you
> would need explicit permission.
>
> The current situation of the Greenland coastline in OSM en large (with
> the exception of some local mapping in smaller areas) is the following:
>
> - west coast is mostly PGS based
> - east coast is mostly GADM based
> - north coast is based on MODIS imagery, partly improved by Landsat in
> the far north.
>
> The GIMP land ocean mask is higher resolution than most of the current
> data in OSM but of varying quality.  You'd have to deal with a number
> of problems:
>
> - wrong classification due to shadows and wet snow.
> - positional accuracy problems due to bad alignment of Landsat images
> used.
> - outdated ice coastlines due to 15 year old data basis
>
> Examples showing this (from East Greenland as magenta mask in comparison
> to up-to-date Landsat imagery and the OSM coastline (blue line)):
>
> http://www.imagico.de/files/coastline_greenland1.png
> http://www.imagico.de/files/coastline_greenland2.png
>
> So to get actual improvements out of this compared to current data you'd
> need to manually fix a lot of problems and would still end up with data
> significantly worse than what you could produce from available imagery
> by hand (or where water is ice free in summer with semi-automated
> mapping).
>
> By the way if you have contacts to Greenland mapping authorities getting
> data and permission to use for names of islands, mountains, fjords,
> bays etc. would be most valuable.  For a mapper working in essentially
> unpopulated areas like northern and eastern Greenland such are
> difficult to determine reliably.
>
> --
> Christoph Hormann
> http://www.imagico.de/
>
> _______________________________________________
> Imports mailing list
> Imports at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/imports
>



-- 
Karl Brix Zinglersen
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/imports/attachments/20150728/89db5689/attachment-0001.html>


More information about the Imports mailing list