th river and lake area objects. The most
common case is when borders of neighbouring objects run in and out of each
other. The fragmentation itself is causing lots of problems even in
rendering. Just look at these examples (the well-known light/dark stripes)
here <http://osm.org/go/7WCEND?layers=H> http://osm.org/go/7WCEND?layers=H
or here <http://osm.org/go/7WCzACu--?layers=C>
http://osm.org/go/7WCzACu--?layers=C or here <https://goo.gl/JVI1E7>
https://goo.gl/JVI1E7 or here <https://goo.gl/Xhv1nq> https://goo.gl/Xhv1nq
. Extending the areas within the object classes may help in rendering but
still the fragmentation is there.
Assume, we have managed to remove all redundancy, repair most of the "broken
polygons" and perform full defragmentation within area classes: forests,
lakes, rivers and land masses. Besides, we managed to recognize and replace
missing river sections, missing islands in lakes and rivers. So, within any
of these object classes we have the best data presentation that is
potentially possible from the source data. Yet, we quickly discover that
there are forests overwritten by lakes, rivers running over forests, borders
of lakes running in and out of forests and so on (the inter class
anomalies). While these anomalies are not show stoppers in rendering, they
limit the corresponding GIS's quality, statistics, quantitative analyses and
forecasts (number of trees in forests, CO2 consumption per year, oxygen
production per year and so on). Let us assume, we have managed to repair
all these anomalies by using the topology geometry/calculus as hinted in my
previous mail. Then some of the results are like these:
The country's land area created from the coastline data is here
<http://goo.gl/O1L60r> http://goo.gl/O1L60r , the border polygons are
disjunctive and there are no holes at all. Subtracting all inland water
areas and adding the islands within these, we get the land-masses
illustrated here <http://goo.gl/OM2dqn> http://goo.gl/OM2dqn. The yellow
areas represent a minimal simple/compact land-masses coverage. The inland
waters make only about 0.5% of the land area.
The countries forest coverage is pretty high <https://goo.gl/HU63M7>
https://goo.gl/HU63M7 . The forests cover around 63.6% of the land-masses,
though there are still some forests to be mapped (see the Kyushu island).
The largest compact/simple forest area, here <https://goo.gl/4yzeyC>
https://goo.gl/4yzeyC, by size equals to 24% of all forests. It consists of
one outer/container and 25831 inner/excluding polygons. All polygons are
disjunctive and from any point A to any point B in this area one can go
walking exclusively through the forest (hm, the shortest way?). However, the
holes of this largest simple area contain additional 2892 new (small)
"forests". An extract from this complete, largest reginal forest is
presented here <https://goo.gl/mzgDRg> https://goo.gl/mzgDRg . The light
green is the largest simple forest area while the dark green represents the
smaller forests in holes. One can see that there are even holes in these
small forests and new forests in their holes and so on. Similar inclusions
sometimes go up to 6 levels. The ten largest simple areas make 70.2% of all
forests in the country.
Finally, extending the case to other object types and/or larger areas like
continents or the Planet, one can feel the huge potential of OSM, especially
in the future with growing content. Simply, it is difficult not to be an
enthusiast of it.
Regards, Sandor
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk