Hello, I am working on a project where I want to use overpass to export all hamlet nodes in Ireland (the island), as I have already done in Great Britain and elsewhere. Before I do this, I wanted to query the use of place=hamlet in County Longford, which I believe is incorrect compared to everywhere else. As per overpass, there are 1808 place=hamlet nodes in Ireland (country), but 735 of these are concentrated only in County Longford[1]. You can see from this screenshot how County Longford stands out with its amount of place=hamlet nodes.
From what I can see, a place=hamlet node has been placed in the middle of nearly every locality=townland relation, effectively duplicating them. This is not the practice in the rest of Ireland, and presumably is incorrect tagging as a hamlet is a small named settlement. Most of these place=hamlet nodes do not seem to correspond to any settlements. For example, Node: Stonepark (1883246128) | OpenStreetMap is certainly wrong as it duplicates a nearby village node of the same name. However, I assume some of these place=hamlet nodes may be correct (a hamlet could have the same name as the surrounding townland) so I’m not sure what solution to propose aside from a manual review of each. As I am of course not particularly knowledgeable about the whole townland concept, I wanted to raise this query amongst the Irish community, rather than start deleting ones I think are wrong.
Thank you.
It looks like the fake hamlet nodes were added as labels so each townland could render on the map, they’re definitely wrong but I can’t really think of a good way to sort them out other than manually.
You could maybe check for hamlet nodes that are within a certain distance from highways of any kind and keep them as “real”, as nucleated settlement will obviously be along roads.
As I understand things, place=hamlet was used in the early days of OSM as a proxy for townlands. A lot of these seem to have been mapped in 2012. @mackerski@amapanda_ᚐᚋᚐᚅᚇᚐ@boggedy might be able to give you better information.
That said, there are places that have been genuinely mapped as place=hamlet, as they lack the infrastructure for them to be considered villages. We should put together an agreed points scheme to help decide which tag to apply.
Interestingly (or oddly) I found a hamlet called Kiltyreher which is beside another hamlet called Kiltyreher, which are in two different but neighbouring Administrative Boundaries, both called Kiltyreher.
So which came first… the hamlets or the admin boundary… because both the hamlets and the admin boundaries appear to have been scripted to exist together.
Adding some detail to my finding above - the two Kiltyreher hamlets were created by the same user within 20 mins of each other (29 Jul 2012, 23:46 to 00:09), and the first version of the admin boundary was also created by that same user around the same date/time.
My interpretation therefore is that the user who created these hamlets did so so that the townland name would clearly display a name on the rendered map, rather than the name be hidden within the boundary line as per the image below:
So is it ok to get rid of the Kiltyreher nodes and anything similar? It doesn’t look like there’s enough here to be called a hamlet and as the townlands are mapped it wouldn’t remove Kiltyreher from the map completely. Would it be ok to suggest that any hamlet node that has the same name as the enclosing townland (and has no significant settlement visible on aerials) can be deleted?
Another thing to try is to look for documents by Longford County Council that list settlements (eg planning and development strategy documents - things like this Longford County Development Plan 2021 -2027 - LongfordCoCo ) and see if that gives you a more reliable list of what small settlements the powers that be think actually exist in the county.
If someone originally experimented with adding the hamlet nodes to get townland names to render on Carto, you could always consider changing them to ‘label’ nodes (i.e. tag them as part of the townland relation, with the role of ‘label’). Or you could just delete them
Based on the discussion in this thread, I have gone ahead and done a manually-confirmed deletion of all hamlet nodes which duplicated the name of the enclosing townland. I used bing to confirm that there were no obvious real hamlets that the place node was referring to. A total of 9 place=hamlet nodes have been retained. 8 of these have a different name to the townland, and the remaining one predates the townland imports and is located on a concentration of buildings (Node: Carrickboy (52220443) | OpenStreetMap). I also moved a couple of wikidata and name tags from hamlet nodes onto village nodes where there was already a village with the same name (Node: Ballinalee (52244765) | OpenStreetMap & Node: Abbeyshrule (52260995) | OpenStreetMap), and added a missing island (Way: Saints Island (1557210710) | OpenStreetMap). Essentially, anywhere that the node was more than just place=hamlet & name=*, I made sure the extra tags were copied to the corresponding real feature. I hope my work is satisfactory. Link to changeset: Changeset: 188689738 | OpenStreetMap.