> Seriously though (& not arguing :-)), "a semi-trailer truck poking out of a
> roof. Pavement staining and tire marks from fork lifts" wouldn't really be
> enough to say definitely whether you're looking at a roof over an open area
> or an enclosed building, would it? That truck could be poked out of the
> doors of an enclosed loading dock, & the forklift could be doing the same?
Usually, no single thing will lead to a conclusion. The identification 'keys'
basically indicate the other things depending on the context one should
look for. Hey, it could be a short truck, too :-) Preponderance of evidence.
Which is why it's useful to use multiple imagery sources.
Bing and Google Maps are aggregators of imagery - they license it from
other companies, at some price, at various resolutions, and what shows
can change over time - urban areas get updated very frequently, not so
much when you get out into the sticks. Even this isn't an absolute, by some
freak of availability we observed a bear on the road of my brother's
property on the Front Range in Montana. Also, the orthorectification
can sometimes be crap ( adjusting for terrain, etc. )
> Yes, if you've got slanting or night time imagery that may help, but I've
> never seen it in the areas I map in :-(>
You can add your own sources to JOSM, the presets are the more or
less globally useful services. Sometimes, you don't need to actually
add it to JOSM to assist, like the building overhang issue. Our
county has phenomenal lidar point data available ( 6"), I can make out
the location of the picket fence on my front lawn. Our state flies
regular oblique photography to monitor coastal conditions, also.
There is an upfront effort of finding what's available, but once you
have the endpoints and sources for your area of interest, it can
> On the subject of clarity of images. Was mapping the other day (using iD),
> marking buildings in an industrial area. As I said, the photo's weren't the
> clearest, but I was also peering through the purple haze of the mapped
> area=industrial, which certainly doesn't help matters either :-(
Yeah, we have
308 cloudy days a year. Which underscores the value of
having sources that have multiple times available. Leaf on and leaf off
is also useful. If Bing's blender picked a day when an inversion layer
was occurring, it's nice to pick another day. Night imagery is rare.
The gold standard for this is SAR ( synthetic aperture radar )