Earlier today we had some down-time... I'm not sure why but it seems to have recovered.
Since I had planned on working on OGF during that time, I started messing around with my most recent downloaded backup file (dated 12/31).
There are some interesting utilities that can be used to manage OSM data offline, in order to get manageable data sets in OSM format (or other formats). The two tools I use all the time are osmconvert and osmfilter.
OSM format files can be edited in JOSM and uploaded later, but these utilities have other abilities too.
Out of curiosity, today, I decided to get some statistics on our current OGF map, and make comparisons with my country, Ardisphere (abbreviated FA). I made two OSM files - one was for the entire OGF planet and the other was using my FA polygon file (.poly) which gives me a .osm file limited to my country's borders (roughly).
I found that the OGF planet file (uncompacted .osm) is currently about 4.05 GB, while the Ardisphere, alone, is 618MB.
"Wow," I thought. That looks like the Ardisphere, a relatively tiny country (about the size of Uruguay) on the world map, accounts for 15% of OGF's data volume. I guess I've drawn a lot of nodes! I'm nowhere close to finished, either.
As a point of comparison, OGF's entire Western Hemisphere, where only vandals and admins dare map, is approximately 65MB.
As another point of comparison, the REAL WORLD .osm file is 46 GB, so our imaginary world has less than 10% of the data. Meanwhile, Uruguay's .osm is 342MB, and Switzerland's is 3.6GB. Hmm. Ardisphere has twice the data as its real-world counterpart, while tinier but well-mapped Switzerland is almost the size of the entire OGF planet.
Both of the files I made, OGF.osm and FA.osm, are much too large to edit in JOSM - my desktop computer will crash on .osm files larger than about 100MB. I have other polygon files of smaller areas for if I'm making an extract for editing.
However, seeing those relative sizes, I was interested in compiling some statistics, using a feature of the osmfilter utility. Here is summary of the stats I found (cut off arbitrarily at ogf-counts > 10000) - screenshot from OpenOffice spreadsheet:
Analysis and thoughts:
Most of these stats make sense to me.
I am not surprised to see that Ardisphere accounts for 75% of the world's "natural" keys - I have been working hard to completely cover the country with detailed landuse and land cover polygons, and to create a realistic hydrologic system.
My custom-made tag "ruta:survey" has become a place to record shorthand tags that I use with JOSM filters to control my edit space - when I complete an area I try to remove those. I was somewhat surprised to see that Ardisphere's 13000 "places" account for 25% of the world.
The very low proportion of highways in the country makes perfect sense - many mappers are road-crazy and neglect other features, under the mistaken belief that a road map is a complete map.
I was surprised by the low proportion of buildings, but it's true I haven't gotten down to mapping many individual neighborhoods, so far, while some other users have done quite a bit with this. I will note, however, that we have many cases where users have imported detailed information from OSM (despite policy against it) and much of that detail is dense with buildings. I have no idea what proportion of our map can be attributed to imported data.
Anyway happy new year! / ¡feliz nuevo año! / 새해 복 많이 받으세요!
and as always, happy mapping