[GRASSLIST:3346] Re: correcting srtm data with mrsid images (oops)

[tatel at euskalnet.net]

Hi Markus, Bryan, Hamish and all GRASS list subscribers,

Thank you very much for your help.

Now I'm pretty sure there was not shift between srtm and mrsid datasets.
Differences in N and NW shorelines were produced by sun shadows on high
slope areas, which become as black as sea on mrsid image. 

With r.slope.aspect elevation=srtm slope=slope aspect=aspect

and then

r.mapcalc slope=if'(slope<10,null(),1)'

I get areas which allmost coincide with these N and NW "shifts" It would
be better if I could use a lower slope value, but in this case much of
the speckled sea remains. Now I'm missing what I think are little plains
eroded by tides below the cliffs.

By adding this slope layer to the red and green layers mentined on my
previous post, I get a pretty shoreline. Slope layer (allmost) works in
cliff areas, red and green layer are working fine in bare soil coast and
vegetation coast respectively.

The shoreline I get follows coast better than the one I made by hand
with v.digit and also has advantages over the GSHHS dataset, which
missed some beachs, wetlands and docks.

Anyway, I'm quite impressed by the GSHHS quality. It is the best
shoreline I have seen, very much better than any other vector layer
available for free I know of. I'm thinking about combining it with the
red, green and slope layers in order to catch those little plains down
at the cliffs.

Unfortunately there is a true shift with GSHHS and I'd read on the Net
that this dataset's ellipsoid and datum are unknown even by authors.

Is there a way to set correction values somewhere?

Thanks a lot, and best wishes,

Roman