[Imports] NLS.fi import stage 1 plan

Ilpo Järvinen ilpo.jarvinen at helsinki.fi
Mon Sep 2 21:43:48 UTC 2013

Hi all,

I'm preparing a road import for Finland based on NLS of Finland (nls.fi)
dataset. Since OSM is quite complete in many areas already and even
significantly exceeds the quality of the NLS dataset the import is meant
to be selective. I.e., I've calculated isolated subnets in the road 
dataset that do not overlap significantly with existing OSM data, only 
those parts are to be imported in the phase 1. All subnets are supposed to 
be manually reviewed as there's also some additional road classification 
that cannot be done solely based on NLS classification system. The task 
will be divided by subnet granularity to mappers in Finnish OSM community.
Please check the more detailed plan:


I've done some preliminary discussions/previews with OSM community in
Finland (although I also wanted to be quite selective on where I post
direct links to .osm.gz files as sadly some people tend to be too
trigger-happy). In general, the OSM community in Finland is quite excited
about the dataset and I received some positive and constructive feedback
from there which was useful to improve the toolset to calculate the
subnets in a more useful manner. There are discussions about the NLS
dataset and also about importing it in various threads at forum's Finland

Currently the biggest open question is how to actually distribute the subnets
and do the bookkeeping of the progress?
- The most optimal solution would be to have subnet bboxes to appear on map
and reservation, download, done/problems marking could be done using its
interface but my webside programming skill are not good enough to come up
such system in a reasonable time. This approach is somewhat similar to OSM
tasking manager's but different enough that enabling tasking manager to have
such different task selection interface is beyond my web coding skills (the
griding would be probably easy enough to circumvent by selecting huge cell
- Another potential tool which is somewhat similar is Maproulette but again
it would require some modifications.
- The simplest alternative I've though for the task selection is just have
big tables in wiki and do the bookkeeping but I'm not sure if it would
really work well or not.

Some background about NLS dataset: 

More than a year back NLS opened up quite significant portion of their
datasets under a free license [1] but right from the beginning it was
unclear how strictly the attribution requirements are to be
interpreted (whether all users of OSM data need to repeat it or whether
mentioning it on the /copyright page is enough). There were some
non-written confirmations that /copyright page would be enough but only
very recently there was also email reply stating exactly that [2] which
re-triggered this import planning once again.

I've run earlier some analysis over NLS dataset, done some comparing
between them and based on that there have been improvements to both
dataset which could be performed without deriving the data directly
from the other dataset (NLS has been doing improvements based on our
discoveries too [3]). While playing with both dataset, I've been forced
to look the NLS dataset from many places so that I've become somewhat
familiar with it. From prior experience we do know that the road data
is reasonably correct (many people have used NLS.fi dataset based maps
in Finland) but based on my experience with the actual data, there are
quite many objects that are disconnected although on map they would
appear as joined together (most likely due to inaccurate clicks during
map drawing similar to what all OSM mappers probably have encountered
themselves too). Obviously we want these errors to be fixed rather than
duplicated into OSM.


[1] http://www.maanmittauslaitos.fi/en/NLS_open_data_licence_version1_20120501

Available only in Finnish:

[2] http://forum.openstreetmap.org/viewtopic.php?pid=356296#p356296
[3] http://wiki.openstreetmap.org/wiki/Fi:Maastotietokanta/Virheet

More information about the Imports mailing list