Over the past few months I have been steadily working on the subdivisions of Scotland but I don’t know what admin_level I should really be using for them.
Currently I have:
admin_level=5: shires (pre-1975), 1975-96 regions, presumably post-1996 council areas
admin_level=7: civil parishes (pre-1930) and burghs (i.e. towns). Note: burghs always overlap with ≥1 civil parish but should be considered separate
admin_level=8: sub-parts of civil parishes where the parish is split between two shires (almost all were eliminated in 1891, except for Croy and Dalcross, which remained split between Inverness-shire and Nairnshire)
There are a few complications though. From 1930 Elginshire/Moray and Nairnshire were administrated sort-of as a single shire (“Moray and Nairn”) but the shires technically continued to exist. (The same was also true of Perthshire and Kinross-shire but this is not yet implemented.) I have mapped this as admin_level=5 for the combined authority, admin_level=6 for the subordinate counties, but this conflicts with the “landward” districts.
Technically, although they lost all administrative functions, the civil parishes continued to exist post-1930 (and continued to appear on OS maps for example); should they still be included and should the admin_level remain constant?
The cities of Aberdeen, Dundee, Glasgow and Edinburgh were made “shires of cities” in 1891 so should be at the same admin_level as the other shires.
There are some parishes already mapped in Forfarshire/Angus by user heatherstacks who has given them an admin_level=8. (These are technically not (exclusively?) the civil parishes created in 1845, but also the ecclesiastical parishes, so I’m not sure if they should really be tagged as administrative divisions before then.) They have also mapped the “landward” districts of Forfar and Monifieth, also at admin_level=8, the burghs of Leith and Portobello at admin_level=7, the city of Edinburgh (1856-95) at admin_level=6 and the shires of the cities of Dundee (1894-1975) and Aberdeen (1891-1936 and 1936-75) at admin_level=5.
The (historic) counties of Wales (equivalent to the shires) have already been mapped but they currently have an admin_level=6.
admin_level=6: Two-tier non-metropolitan counties, Metropolitan counties, Unitary authorities, City of London; Scottish council areas; Welsh principal areas
admin_level=7: Unused
admin_level=8: Metropolitan districts, non-metropolitan districts, London boroughs (all are England-only)
admin_level=9: Unused
admin_level=10: English civil parishes, Scottish community councils, Welsh communities
I can see an argument for admin_level=5 only being used for the 1975-96 regions and maybe “Moray and Nairn” and “Perth and Kinross” and shifting the shires (and council areas) down to admin_level=6; presumably the 1975-96 districts would then also remain admin_level=6, but I’m not sure what to do about the rest and don’t really know how to properly judge this.
Do you mean like Relation: Croy and Dalcross (Inverness-shire) (2845928) | OpenHistoricalMap? Were these “subparishes” considered to be administrative units in their own right? For what it’s worth, there’s no reason a level-7 parish can’t straddle two level-5 shires. Something happens routinely in the United States: over 750 municipalities presently straddle multiple counties, but we generally don’t map distinct subareas for the parts of a given municipality within each county.
If you can eliminate these subparishes without losing valuable information, then it should simplify the structure a bit. You also won’t need to bother with subarea roles. We should avoid subarea roles if we can help it, because they tend to make us duplicate parent relations far too often.
Would it be easier to think of Moray and Nairn as sort of a “supra-shire” containing Elginshire/Moray and Nairnshire? It looks like you’re a bit boxed in with Scotland already at level 4. OSM has a convention of skipping odd-numbered levels by default to leave space for the occasional odd exception. It sounds like this would be a useful strategy here too.
Given the changes in UK administrative structure over the last several decades, I don’t think it would be surprising if some administrative units like parishes need to change levels en masse at the time of an administrative reform such as the Lieutenancies Act 1997. This would have some precedent in other countries too.
Yes, that is what I mean. I don’t think they were formal divisions in their own right but I honestly don’t know. Either way I thought it was probably useful to map them anyway, if only to clarify the situation, especially for exclaves. I suppose it’s a habit I picked up in other projects. They could maybe be treated as informal regions or have no admin_level value or even have boundary=administrative removed but I don’t really know how that would work, and if the general consensus is just to not map them at all so be it.
Is there some other way to represent the hierarchy? It seems like the information is useful even if there are technical reasons to avoid that particular representation of it.
It’s a tricky one. Elginshire/Moray, Nairnshire, Perthshire and Kinross-shire did continue to exist de jure but administratively were effectively replaced with the combined authorities which had the powers of a standard shire. The shires themselves were no longer used as the primary unit of administration, although they could be delegated to; in practice I don’t know how much this actually happened.
Do you mean skip all of them or just leave gaps where appropriate/necessary?
It all depends on whether there’s anything interesting to say about each side. Hypothetically, if each side had its own unique identity (especially its own name), it would be worth mapping as a boundary of some sort, but maybe as a boundary=place instead of boundary=administrative.
There was an informal discussion about it last month in the chat room:
There have been some recent discussions in OSM about getting rid of subarea members of boundary relations, at least in the U.S. I noticed we have over 30,000subarea members in OHM too. Any idea if there’s a strong need for them? I can’t imagine them being very convenient in OHM, since you’d have to create a redundant parent relation any time any one of its children experiences a boundary change, no matter how small.
Kovoschiz
Indeed, it would be extremely more worse in OHM…
Frying_Pan
I agree with stopping using them, maintaining them would be very difficult and the benefit would be marginal.
So far, your use of subarea members has been benign, but if you were to extend this approach up the hierarchy, it would quickly become a scalability issue.
In theory at least, the fact that one administrative boundary is contained inside another and has a numerically higher admin_level=* value should be sufficient for data consumers. I’ve seen some cases where there’s still ambiguity or complex dotted-line relationships, but in general a data consumer shouldn’t be relying on subarea members, and I don’t know of any that do. The OHM tile generator, Nominatim, and QLever all determine containment spatially rather than by subarea.
I just meant leaving the odd levels as gaps until you find a need for them. The gaps are an imperfect solution too: in the U.S., we’ve even encountered one city where the gap has already been filled and we need yet another level. But at least problems won’t come up as easily.
Makes sense, at least for “simple” administrative divisions like these. (It gets a lot more complicated in feudal states, occupied territories etc but that’s another discussion (or several).) I hadn’t intended to go above the shire/region level anyway.
I do wonder how to handle the overlap of burghs and civil parishes. It seems like it should be OK if they both have the same admin_level but if not problems may arise when such automated systems are applied. (I don’t think this is a problem that subarea can solve anyway.)
OK. Here’s what I’m thinking at the moment:
admin_level=5: 1975-96 regions; maybe Moray and Nairn, Perth and Kinross 1930-75
admin_level=6: shires, 1975-96 districts, post-1996 council areas
admin_level=7: maybe Moray, Nairnshire, Perthshire and Kinross-shire 1930-75
