Checking UK Towns

classic Classic list List threaded Threaded
7 messages Options
Reply | Threaded
Open this post in threaded view
|

Checking UK Towns

Gregory Marler
You might spot my recent map edits with a changeset such as...
"Checking data and relations for towns in Lincolnshire. This changeset forms part of paid work to improve OpenStreetMap data. #UKTownCheck #OrganisedEditing"

A small amount of my time has been funded by Open Cage Data to check towns in the UK. Ideally it should be possible to get a town as both a node and a relation.
I've been going through a list of towns where this isn't the case. I've been doing a single county per changeset to avoid it being a mess to follow.


In a lot of cases the towns nicely relate to parish wards (admin_level=10). Sometimes I just need to add the town node as an admin_centre member of the relation. Other times the "outer" parts of the relation are not in order.
In some cases, the town has a distinct area but not a simple parish ward (the previous parish might have ceased to be). I have created some relations with boundary=place, place=town, admin_level=11.
I've even found some town nodes that are complete tagging for the renderer. They should have been something like place=suburb.


So far it's been very insightful to do this for several areas around the UK. I don't think I'll manage the whole country within the funded time. I intended to properly report on what i've done and what I've found.

Gregory.


--
Gregory Marler

_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
Reply | Threaded
Open this post in threaded view
|

Re: Checking UK Towns

sk53.osm
A few points:

* I believe you should complete the relevant organised edit page on the wiki.
* I think any use of admin_level=11 or indeed any use of admin_level at all for un-parished areas needed to be discussed up front. Places like Mansfield, Maidenhead etc  simply do not exist as administrative entities. To represent them as such is erroneous. 
* In the case of the two Ashfields and Mablethorpe/Sutton-on-the-Sea there is a single unparished area so how can someone verify the boundary? 
* Saxilby ought to be a village
* Sutton-on-the-Sea probably ought to be a village; not sure about Mablethorpe.
* Banstead as a town is rather marginal. Unlike Epsom and Ewell which were towns before London grew, Banstead is more a London exurb. I'd guess village or suburb may be just as appropriate.
* I'd have anticipated Dorking's W boundary to be pretty much follow the line of the former Urban District which wiggles about more than the one you mapped.
* I don't think place=town should be duplicated on the node and the relation, one or the other (it makes for awkward processing for what should be very simple queries: how many towns in East Anglia?). I do realise that associating a town with a boundary is the objective, but I believe there should be a way which doesn't break "one feature one element".

Regards,

Jerry

On Wed, 30 Jan 2019 at 15:09, Gregory Marler <[hidden email]> wrote:
You might spot my recent map edits with a changeset such as...
"Checking data and relations for towns in Lincolnshire. This changeset forms part of paid work to improve OpenStreetMap data. #UKTownCheck #OrganisedEditing"

A small amount of my time has been funded by Open Cage Data to check towns in the UK. Ideally it should be possible to get a town as both a node and a relation.
I've been going through a list of towns where this isn't the case. I've been doing a single county per changeset to avoid it being a mess to follow.


In a lot of cases the towns nicely relate to parish wards (admin_level=10). Sometimes I just need to add the town node as an admin_centre member of the relation. Other times the "outer" parts of the relation are not in order.
In some cases, the town has a distinct area but not a simple parish ward (the previous parish might have ceased to be). I have created some relations with boundary=place, place=town, admin_level=11.
I've even found some town nodes that are complete tagging for the renderer. They should have been something like place=suburb.


So far it's been very insightful to do this for several areas around the UK. I don't think I'll manage the whole country within the funded time. I intended to properly report on what i've done and what I've found.

Gregory.


--
Gregory Marler
_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb

_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
Reply | Threaded
Open this post in threaded view
|

Re: Checking UK Towns

Tom Hughes-3
In reply to this post by Gregory Marler
On 30/01/2019 15:08, Gregory Marler wrote:

> A small amount of my time has been funded by Open Cage Data to check
> towns in the UK. Ideally it should be possible to get a town as both a
> node and a relation.

The fundamental problem with this, as Jerry has just said, is that
many towns in the UK have no defined boundary.

Even where there is an administrative entity there is no guarantee
that it's boundary equates to what most people would view as the
boundary of the town - it may under or overstate things.

Equally there is no clear way of even determining what is, or is
not, a town. Just a variety of rules-of-thumb...

This has all been discussed a number of times before ;-)

If you want a challenge look at my local area - it's unparished
so the smallest administrative unit is the district council:

   https://www.openstreetmap.org/relation/2677978

Looking at the builtup area on the right hand side along the Lea
Valley how many places are there, what are their boundaries, and
what type is each of them ;-)

Hilariously in doing that I've just noticed a town (well that's
what wikipedia says it is anyway) that is completely missing
from OpenStreetMap... Waltham Cross should be somewhere below
Cheshunt and west of Waltham Abbey.

Tom

--
Tom Hughes ([hidden email])
http://compton.nu/

_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
Reply | Threaded
Open this post in threaded view
|

Re: Checking UK Towns

Will Phillips
In reply to this post by Gregory Marler
I've already raised concerns I have in a changeset comment about these
edits adding admin_level=10 administrative boundary relations for voids
between civil parishes. They are tagged with
designation=non-civil_parish. This has been discussed on this list
previously. My main objection is that these areas aren't really
administrative entities at all. Gregory correctly points out they are
sometimes used for statistical purposes, but I don't think that
justifies tagging them as administrative.

Another concern is that these admin_level=10 voids often include several
former civil parishes, so they cover a wider area than the name given to
them suggests. An example is Beeston:
https://www.openstreetmap.org/relation/9246079
This new relation covers five former civil parishes (Attenborough,
Beeston, Bramcote, Chilwell and Toton) and so includes a wider area than
what is usually considered to be Beeston. If Beeston is mapped as an
area I think it would be better to use something closer to the area of
the former civil parish.

Gregory has already agreed to think about alternative tagging for this,
but I thought it was worth raising here, in case other people have any
thoughts.

Cheers,
Will

_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
Reply | Threaded
Open this post in threaded view
|

Re: Checking UK Towns

Colin Smale

I also have a couple of observations about these changes.

1) Sometimes an admin_centre is being added to a boundary=political (e.g. parliamentary constituencies, electoral wards). I am not sure this is appropriate.

2) There are multiple definitions of "town", and I don't know which definition Gregory is using here for place=town. A parish council can simply call itself a town council if it so resolves. This is independent of other definitions of "town" based on population, market charter, letters patent etc.

3) Unparished areas are recorded and numbered in the governments GSS coding system (codes start with E43). Maybe we should follow their level of (dis)aggregation w.r.t. multiple contiguous parishes? Do they possibly qualify as statistical areas?

 

Colin

On 2019-01-30 18:01, Will Phillips wrote:

I've already raised concerns I have in a changeset comment about these edits adding admin_level=10 administrative boundary relations for voids between civil parishes. They are tagged with designation=non-civil_parish. This has been discussed on this list previously. My main objection is that these areas aren't really administrative entities at all. Gregory correctly points out they are sometimes used for statistical purposes, but I don't think that justifies tagging them as administrative.

Another concern is that these admin_level=10 voids often include several former civil parishes, so they cover a wider area than the name given to them suggests. An example is Beeston: https://www.openstreetmap.org/relation/9246079
This new relation covers five former civil parishes (Attenborough, Beeston, Bramcote, Chilwell and Toton) and so includes a wider area than what is usually considered to be Beeston. If Beeston is mapped as an area I think it would be better to use something closer to the area of the former civil parish.

Gregory has already agreed to think about alternative tagging for this, but I thought it was worth raising here, in case other people have any thoughts.

Cheers,
Will

_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb

_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
Reply | Threaded
Open this post in threaded view
|

Re: Checking UK Towns

Mark Goodge
In reply to this post by Tom Hughes-3


On 30/01/2019 16:00, Tom Hughes wrote:

> The fundamental problem with this, as Jerry has just said, is that
> many towns in the UK have no defined boundary.
>
> Even where there is an administrative entity there is no guarantee
> that it's boundary equates to what most people would view as the
> boundary of the town - it may under or overstate things.

One useful thing here is the ONS concept of a "Built Up Area". That aims
to give the normal human name of a defined place - for example,
Mansfield and Maidenhead are both in the list, and mean what someone
living there would expect them to mean. And it has subdivisions for
major settlements - for example, the BUA of Greater London includes
BUASDs of Bromley and Camden.

Shapefiles are available from the ONS Geography website, and are OGL, so
they're compatible with OSM.

The downside of ONS BUA data is that it's only updated once per census,
so the current dataset is now a bit out of date - it doesn't take
account of new developments on the edge of existing settlements, for
example. But, as a simple source of names, it's very valuable. And
alterations to boundaries can be mapped on the ground, provided you've
got an existing boundary to work with.

> Equally there is no clear way of even determining what is, or is
> not, a town. Just a variety of rules-of-thumb...

As far as local government is concerned, there is a defined meaning of a
town. That is, any settlement which has a town council (eg, Evesham,
Newmarket). And, for larger settlements, the word "town" is what most
people would call what is, officially, a non-metropolitan borough (eg,
Bromsgrove, Ipswich).

More generally, while there's no single definition of a town, it can be
reasonably assumed to be the default terminology for any built-up area
unless you know for certain that it isn't a town (eg, because it's a
city, or it has a parish council, or is too small to have a council at
all).

Mark

_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
Reply | Threaded
Open this post in threaded view
|

Re: Checking UK Towns

jc129
In reply to this post by Gregory Marler
On 30/01/2019 15:08, Gregory Marler wrote:
> In a lot of cases the towns nicely relate to parish wards (admin_level=10).

I understood admin_level=10 was for whole parishes, not parish wards. Surely parish wards should be boundary=political just like district and county electoral wards/divisions?

I also agree with Colin, admin_centre should only be associated with administrative boundaries not political ones.

Jez C

_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb