[talk-au] Why not to change coastlines automatically to ABS data.

Ross Scanlon info at 4x4falcon.com
Wed Oct 7 00:55:27 BST 2009


> I would be cautious about preferencing "survey" and satellite/aerial
photography data over ABS.
>
> I have found errors in both of these. Survey data from GPS seems at
times
> to
> have been either traced pooly from gpx tracks or based on innacurate
position data, especially where there are tall objects like buildings and
> hills nearby. Similarly, imagery can be misleading when there is
vegetation,
> like mangroves on the shore, not to mention to low resolution of the
yahoo
> imagery itself.

You've missed the point here.

What I'm saying is don't just go and change it from "PGS coastline" to the
ABS boundary data without looking what's there.

In the example given (Hamilton Island), at the points given, the ABS
boundary data was grossly in error, more than 200m near the restaurants
and approximately 500m near the airport.

The ABS data more than likely came from aerial photos anyway as there's
never been anyone actually survey (professional surveyor style) the
coastlines in this area.

I think everyone should have a read of this:

http://74.125.155.132/u/AustralianBureauOfStatistics?q=cache:ijmG6hPI8egJ:www.abs.gov.au/Websitedbs/D3110122.NSF/4a255eef008309e44a255eef00061e57/8e860540d4a7505cca256bf300055f0d/%24FILE/technical%2520paper.pdf+%22digital+boundary%22+accuracy+2006&cd=1&hl=en&ct=clnk&ie=UTF-8

It is the html version of a pdf file from the abs website, as the pdf is
corrupt and won't load (at least on windows).  It's from 2001 but I could
not find an equivalent document for 2006.

The main area of interest is Appendix B and the section on topographic
features, as below in part:

"A typical use of digital basemap in GIS is to select features which lie
within, intersect, or are adjacent to other features. In most GIS these
spatial relationships are determined by the latitude and longitude of the
objects being analysed. If an object is close to a boundary then the
absolute accuracy of the latitude and longitude becomes important. The
PSMA dataset is digitised from maps at scales of from 1:4,000 to 1:250,000
and the accuracy of a latitude or longitude can therefore vary from 4
metres to 250 metres. Cartographic licence and data integration issues can
all further erode the positional accuracy of basemap features."

So there can be very significant discrepancies in the ABS data in regards
to topographic features.

Given that the only topo maps for this area are 1:250000 then the errors
can be in excess of 200m in the ABS data.


Cheers
Ross












More information about the Talk-au mailing list