[Talk-GB] GB Coastline - PGS vs OS

Jason Woollacott woollac at hotmail.com
Mon Dec 12 07:40:33 UTC 2016


I've been doing some coastline updates around the South West, and have been basing my coastline on the OS_StreetView map layer which also shows MHW.  however the GPX file does not seem to match this.   In Cornwall, which has a very jagged coast, you can never rely on Bing as the angles are quite often wrong,  but on flatter ground, say around, Looe, https://www.openstreetmap.org/#map=17/50.35216/-4.44840

The GPX around the pier goes well into the sea (about 6m), compared to the OS SV layer which links fairly close to the bing image at that point.

Agreed that the coastline extract is much better than the PGS, which is a still straight lines in some places.  however, one other area that the GPX extract doesn't seem to cover is the islet and rocks which are above sea level even at high water level.  (See the Looe link again)

Jason (Unieagle)

From: David Groom <reviews at pacific-rim.net>
Sent: 12 December 2016 00:51
To: talk-gb at openstreetmap.org
Subject: Re: [Talk-GB] GB Coastline - PGS vs OS


I was more talking about the actual shape of the MHW rather than its position; if that makes sense.

some examples of problems in the Isle of Wight

1)  There's a section here  http://www.openstreetmap.org/#map=17/50.66636/-1.48566, where the Bing imagery seems reasonably aligned to the gps tracks of the main road, but the gpx file for MHW seems to be too far to the north on the cliff area, and too far to the south on the area to the east.  this beach shelves relatively steeply so there is unlikely to be much difference between MHWS & MHWN

2) Even clearer is an area http://www.openstreetmap.org/#map=17/50.69439/-1.09414, OSM is much more accurate here than the OS Boundary Line

3)  The car park and ice rink here http://www.openstreetmap.org/#map=18/50.73237/-1.15736  were built sometime around 1990, but Boundary line  MHW would show these as flooded

4)  More inaccuracies here   http://www.openstreetmap.org/#map=19/50.76650/-1.30029


------ Original Message ------
From: "Colin Smale" <colin.smale at xs4all.nl<mailto:colin.smale at xs4all.nl>>
To: talk-gb at openstreetmap.org<mailto:talk-gb at openstreetmap.org>
Sent: 11/12/2016 22:17:44
Subject: Re: [Talk-GB] GB Coastline - PGS vs OS

Hi David,

Looking at the spot you indicate on Bing imagery does indeed look like MHW should be above the salt-marsh areas. Looking at Google[1] it is however possible that the grass doesn't quite get submerged, even at the highest tides, so it might also be possible that it is strictly correct.

The Bing imagery is of course just a snapshot, and we don't know the state of the tide at the moment the photo was taken, so it can also be misleading. Even a personal visit is not really enough as MHW is apparently calculated over a 19-year cycle (not sure if OS use this though) and things could change a lot in that time. As MHW is an average, many tides will of course be higher.

The OS data looks a definite improvement for steeper coastlines, where combining OS admin boundaries with PGS coastlines produces many anomalies (admin boundary=MLW inland of coastline=MHW). I would definitely suggest applying the OS MHW data to address this kind of issue. But I agree, use of the OS data would need case-by-case judgements. However I still think the OS data is probably a better base to work from than (unimproved) PGS for reasons I mentioned earlier.

Could you give a couple of examples of problems you saw in the IoW?


[1] https://www.google.com/maps/@51.5386032,0.6292606,3a,24.7y,277.12h,84.08t/data=!3m6!1e1!3m4!1sMl8cwBlLLuOVtPES_DfkOQ!2e0!7i13312!8i6656

On 2016-12-11 22:30, David Groom wrote:

I suspect that even though much of the coastline is tagged "source=PGS" is has been amended by reference to Yahoo and after that Bing imagery, but the subsequent editors did not remove the "source=PGS" tag.

Certainly comparing your gpx file for the Isle of Wight with the coastline currently in OSM there appear a number of places where the gpx file does not accurately represent MHW.

I certainly would not want to see a wholesale replacement of what is in currently in OSM with OD Boundary Line data.

Looking here http://www.openstreetmap.org/#map=19/51.53546/0.60580 an area near Southend, unless the Bing imagery is outdated, the Boundary Line data seems to be an odd representation of the coastline.

On 11/12/2016 10:43, Colin Smale wrote:


Most of the coastline is currently tagged as "source=PGS". As part of the Boundary-Line open data set OS provide MHW lines which look to be significantly better than the PGS data:

  * Much newer - updated twice a year, although I am not sure how old
    the actual underlying survey data is (PGS coastlines seem to be
    from 2006)
  * Better resolution - more nodes, smoother curves
  * Consistent with admin boundary data, so MLW never appears above
    MHW (often a problem on rocky coastlines like Wales and Cornwall)

There are a couple of caveats when working with the OS data:

  * Where MHW=MLW, i.e. the MHW is colinear with the admin boundary at
    MLW, there is a gap in the MHW data
  * The MHW data goes miles inland in tidal estuaries, which is
    correct from the MHW standpoint, but for coastlines I think we
    need to cut across the estuaries at the right point to form the
    correct baseline
  * The MHW data is organised by area - down to constituency level.
    Every time the line crosses the area boundary, it simply stops and
    you need to load the adjacent area to continue the line

I have uploaded GPX versions of the October 2016 OS MHW data to http://csmale.dev.openstreetmap.org/os_boundaryline/mhw/ with a file per county / unitary area (I have not produced the files for the higher-level regions or the lower-level constituency areas).

In the Thames estuary around Southend and on the north Kent coast I have replaced the PGS data with the new OS data and to me it looks much better (in Potlatch) although the changes are not yet showing through on "the map". I think coastline changes are processed less frequently.

Any comments?


Talk-GB mailing list
Talk-GB at openstreetmap.org<mailto:Talk-GB at openstreetmap.org>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-gb/attachments/20161212/5ac4fb82/attachment-0001.html>

More information about the Talk-GB mailing list