[Talk-us] USFS Wilderness data, import with a "lenient transformation?"

stevea steveaOSM at softworkers.com
Thu Feb 21 21:58:17 UTC 2013

I've been working on an amazing national resource, regional to me, 
which I greatly enjoy both in real life and mapping in OSM:  the 
Ventana Wilderness (around Big Sur, California) of the Los Padres 
National Forest (LPNF).  I have discovered that Ventana is only the 
northernmost of ten wilderness areas in this single national forest 
alone, and have found missing from OSM and multiple/conflicting 
sources for these wilderness boundary data, correct and current 
versions of which I very much want to see in OSM.

Recently, I was pointed to the "official" (US Department of 
Agriculture/Forest Service) data source for these wilderness areas, 
in the form of a shapefile containing all wilderness areas in all 
national forests.  This shapefile is about 23 megabytes, which 
becomes about seven times larger in .osm format, due to its xlm-ish 
nature.  However, before initiating the translation from .shp to 
.osm, JOSM informed me:

"JOSM was unable to find a strict mathematical transformation between 
GCS_North_American_1983 and WGS84."

And further warned that the polyconic spherical case for such a 
transformation is not implemented, so the algorithm would fall back 
to the (error-prone) elliptical equations. Such a lenient 
transformation can yield up to 1 kilometer errors. I am explicitly 
warned that choosing this methodology means the results should NOT be 
uploaded to OSM.

Yet, I am not so easily defeated.  I still need to "whittle away" the 
entire national dataset to just the ten LPNF multipolygons relations 
(and their members) representing the LPNF wilderness areas.  Assuming 
I do, I further believe that I can add to my JOSM editing a USGS 
layer which often or always has a wilderness boundary on it, 
potentially allowing me to engage in some smart edit-blending away of 
the (up to 1 km) transformation error.  In other words, I'm using 
multiple data sources to mitigate the potential error.

Question #1:  If the additional USGS layer showing wilderness 
boundaries does indeed allow me to mitigate away the transformation 
error with some careful edit adjustments where necessary (though, it 
is also possible NO or VERY MINIMAL adjustments are necessary), 
should I feel comfortable uploading these boundary data?  This begs 
the additional question of "why not just start with USGS rather than 
USFS data..." but you'll agree it is easier to start with something 
largely done and tweak it, than it is to start from nothing and 
painstakingly take a much longer time and effort to make it exact 
from scratch.

Question #2:  If this "test case" proves successful in uploading 
wilderness boundaries in central and southern California for the 
LPNF, and I document the set of steps I took (largely already just 
done), might the greater OSM community reach consensus that all 
national wilderness areas (and other national treasures it can be 
valuable to have in OSM, such as wild and scenic rivers) are suitable 
candidates for importation and upload?  Even with a potential error 
from a "lenient transformation?"

Question #3:  Is it possible to improve the JOSM's transformation 
algorithm so it uses polyconic spherical instead of the error-prone 
elliptical equations?  Is the best way to do this with a trac bug 
report/feature request?

Thank you for your feedback,


More information about the Talk-us mailing list