For the last couple of weeks I have been replacing the portions of the BC
coastline which were PGS or CT-dirty imports with the latest CanVec/GeoBase
data. I elected to use GeoBase since it's coastlines require less manual
processing to get into workable form. This process is now finished, with the
exception of the basins that are in the US and have data in GeoBase and the
08OA000 and 08OB000 basins (Haida Gwaii) which are in progress.

Before replacing any coast, I compared it with the GeoBase data for
accuracy. For most of BC's 25000 km of coastline the data was PGS data and
easily replaced. The only places where there was substantial coastline
retained were in Greater Vancouver, Victoria, Nanaimo and Sidney. The
quality of the OSM coastline data exceeds that of GeoBase/CanVec in these
areas.

All the data was run through JOSM's simplify function with a max-error of
1.5m. On later runs, short ways were combined into larger ones after
simplification. The search expression "(parent child selected | parent
selected) nodes:-1000" was used in doing this.

In the process I identified many unmapped rivers where I needed to join two
sections of coastline together. Where bing imagery showed a river, I added a
brief tracing. Most of the bing imagery was low-resolution Landsat or
higher-resolution imagery from the BC government, the same as is found on
their openmaps WMS server.

In the process, I ran JOSM's validator tool over most of the coastline and
fixed other unrelated errors.

Statistics:
The .osm files were 232 MB on disk before simplification
The total number of objects uploaded is approximately 1.5 million
Surprisingly, no errors were found by coastcheck in the coastline,
indicating my procedures worked
There are 47 NHN basins with coastline, varying in complexity and size from
5k nodes before simplification to 150k nodes before simplification


_______________________________________________
Talk-ca mailing list
Talk-ca@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-ca

Reply via email to