simon at zymurgy.org
Tue Mar 21 13:02:31 GMT 2006
David Groom wrote:
> Thinking forward to the Isle of Wight workshop, but also relevant
> elsewhere, is the issue of coastlines.
> Two issues spring to mind.
> Firstly is it necessary to have coastline data in OSM?
> Secondly if the answer to the first question is yes, then how should the
> coastline data get into OSM? It will not always be possible to walk /
> drive along a coastline either because the land may be private, or
> because it may be inaccessible due to woodland, cliffs etc. It
> obviously would be possible to use the Landsat imagery and trace the
> coastline for an area that was being mapped, but if the
> orthorectification of the Landsat imagery is not accurate then this is
> not ideal. Also it breaks the idea of having all OSM data derived from
> user entered GPS data.
Think how Ordnance Survey do this sort of thing. They don't go climbing
all the cliffs with theodolites and GPS equipment. The traditional way
of doing this is with trig. points and noting positions of headlands,
inlets and the like.
These days they use aerial photography, and with ortho-rectify them
against known features that they have really accurately surveyed on
land, and then trace the outlines. To get the mean high water and mean
low water marks, you really need to take footage at those times of a
If someone wants to lend me a light aircraft with proper aerial
photography gear, and fuel it, I'd be happy to fly it straight and level
over the Isle of Wight a few times. The sort of aerial photography you
get just by sticking a normal camera against the window is going to be
at the wrong angle and a pain to align properly with other strips.
The professional aerial photographers have one or more downward facing
medium-to-large format cameras with the lens sticking out of the bottom
of a specially modified aircraft, the modification itself needs to be
certified as airworthy. Certainly, if anything protrudes from, or
affects, the body of a certified aircraft that affects the aerodynamics,
it will need to undergo certification.
The US National Geodetic Survey contracts the work out and the details
of how they do it are at
Anyway, we *could* use landsat to do this, but we'll probably want to
make sure that it's properly orthorecified against obvious features that
OSM contributors have surveyed accurately on the ground.
More information about the talk