Hi all,
Following Brian's example, and "playing the ball" as Warin pointed out,
I have now taken some action.
To document what I have done:
1) I first selected all features tagged as "natural=heath" using
Overpass Turbo in an area roughly the extent of Wales (but actually
larger as the extent of selection was just the Overpass Turbo default
window zoomed in to some extent).
2) I ran the following Overpass Turbo query:
/*
This has been generated by the overpass-turbo wizard.
The original search was:
“natural=heath”
*/
[out:xml]/*fixed by auto repair*/[timeout:25];
// gather results
(
// query part for: “natural=heath”
node["natural"="heath"]({{bbox}})(user:"USER_NAME");
way["natural"="heath"]({{bbox}})(user:"USER_NAME");
relation["natural"="heath"]({{bbox}})(user:"USER_NAME");
);
// print results
out meta;/*fixed by auto repair*/
>;
out meta qt;/*fixed by auto repair*/
With USER_NAME being one of the three previously named editors. This is
a pretty conservative approach, since the Overpass API only returns
those objects last edited by USER_NAME, so any natural=heath feature
edited or corrected later on by another user is thus excluded. This does
mean the resulting data is a kind of "patch-work".
3) I subsequently exported the data from Overpass Turbo using the
"Export" option with the "raw data data directly from Overpass API"
option selected. This gives a file that is OSM XML that can be read by
JOSM. I only needed to rename the file and add the *.osm extension,
since the download result of the export option did not add an extension,
and the file was thus not recognized as a valid JOSM XML file.
4) I loaded the data in JOSM
5) I selected all natural=heath features again in JOSM
6) The reason I did the latter, was to see if there were any other tags
on the selected features beside natural=heath and maybe source=x.
7) If there was another tag in the combined tagset of all the selected
features (e.g. name=x or wetland=x), I deselected those by using a query
like "name=*" to exclude all heath features having a name. The relevance
of this is that I thought it likely that any feature with further, more
detailed tags, was more likely to be a "true" heath feature, or
originally created by other users. By excluding features with more
detailed tagging, those were left alone during the edit operation
8) Next, for the remaining features, I deleted the "natural=heath" tag
of all the selected features, and I set two similar "note" and "fixme"
tags as Brian did.
9) I did also have a closer look at the history of some of the objects,
and reviewed some stuff in Bing. Nonetheless, it is likely that this
action removed some areas that may be genuine "heaths". I especially
suspect this for some areas in the south of Wales.
Please note that, like Brian, I did NOT delete the features. They are
still there, just with no relevant tag to render them. You can either
re-tag or correct them as you like.
Also note that while having a closer look at the data, I also noticed
other issues I hadn't noted before:
- Areas are inconsistently mapped. Some areas are treated as a whole,
others are cut up in many small pieces often using arbitrary border
lines between the "natural=heath" features, like a straight line between
two mountain tops.
- Areas overlap with each other. Even some of the areas mapped by the
same user, may overlap another area of natural=heath. Even more worrying
is that some of these areas were drawn straight over other natural=heath
features added by other users in the years before this. I actually find
it really hard to comprehend why this has been digitized like this, it
really does still beg the question if much of the data wasn't simply
imported, without much regard to what was already there.
If the British community feels this is all to bold an action, feel free
to revert the relevant changesets, I have no problem with that at all. I
do think it would be worthwhile to continue the discussion here though
before doing that, and possibly rethink how such areas need to be
mapped, if at all. Personally, I think it would be much more useful,
instead of attempting to fill in the map with large swaths of inherently
hugely complex and diverse (semi-)natural landscapes, to start mapping
the surrounding managed countryside and farmland instead, or add more
easily and reliably delineate-able forest cover, or difficult to
navigate swamp/wetland and scree areas, like they are visible in
topographic maps. Those are the things that are really relevant when
navigating the map, much more so then knowing that the entire Snowdonia
National Park may contain "heath"...
The changesets I created are:
- https://www.openstreetmap.org/changeset/46000525
- https://www.openstreetmap.org/changeset/46000280
- https://www.openstreetmap.org/changeset/45999966
Again: since I took a relative conservative approach to selecting the
features, the current rendering result will be patchy. Not all heath
features that may need to be removed, have been removed in this action.
Further clean up and correction is needed if the community feels my
edits were warranted and need to be maintained. As I wrote, I also did
not do this for the entire UK, primarily just Wales for now.
Marco
---
Dit e-mailbericht is gecontroleerd op virussen met Avast antivirussoftware.
https://www.avast.com/antivirus
_______________________________________________
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb
