[OSM-talk] poor quality of PGS coastline data

[David Ebling]

Hi all,

This is more a series of observations than a question,
but I'd like to hear other people's views on the
subject.

I've come across areas where the PGS coast data is
clearly of very poor quality. eg:
http://www.openstreetmap.org/?lat=38.2806&lon=-76.0748&zoom=13&layers=0BFT
-roads are running through sea. If you look at landsat
you can see that the algorithm has followed completely
the wrong lines.

I've done some work to improve things in nearby areas
such as:
http://www.openstreetmap.org/?lat=38.1655&lon=-75.9238&zoom=12&layers=0BFT
but it's very time consuming to do manually though,
and i never managed to get the lakewalker plugin to
work.

I've also noticed that most of the PGS data that has
been uploaded to OSM contains a massive number of
redundant nodes (eg 10 evenly spaced nodes in a
straight line is not unusual), and is also margingally
less accurate than can be achieved by manual tracing
of Landsat images (though this depends on the level of
zoom at which tracing is carried out, obviously). Data
traced from high-res images are of course massively
superior to PGS.

Of course the PGS data is hugely better than no
coastline data, and I've uploaded quite a bit of it
myself. What concerns me is that I've seen comments in
diary entries etc where people have been replacing
manually traced coastlines with PGS data. I hope that
this is only taking place where very rough tracing has
been done, and where the PGS data is of acceptible
quality!

Regards,

Dave


      __________________________________________________________
Sent from Yahoo! Mail.
The World's Favourite Email http://uk.docs.yahoo.com/nowyoucan.html