Re: [talk-au] PSMA Administrative Boundaries
I'll look there. Thanks. On Sun, Oct 7, 2018, at 12:31 PM, Warin wrote: > On 07/10/18 11:22, cleary wrote: > > > > In regard to admin boundaries sharing the coastline, I think that would > > also be incorrect but I am less confident of my view on this. > > > > I did update some administrative boundaries in South Australia using the SA > > Government Data and those boundaries did not coincide with the coastline > > (see the coastline around Streaky Bay, Ceduna, Fowlers Bay and near the > > Nullarbor). The coastline in OSM needs a lot of work - I have looked at > > parts of WA and SA but found it very difficult to use the satellite imagery > > correctly to refine the map of the coastline in areas where there are > > extensive mudflats or large tidal flows and even rocky areas just > > under/above the waterline. But I suggest it is safest, and more accurate, > > to map the administrative boundaries and the coastline separately. > > Might be good to look at Broome for hi/low tide .. there is a fair > distance between the two there so it would be easy to pick in that > location. > > > ___ > Talk-au mailing list > Talk-au@openstreetmap.org > https://lists.openstreetmap.org/listinfo/talk-au ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] PSMA Administrative Boundaries
PS While on coast lines ... computer model of the Kimberly coastline over a few thousand years.. looks like it is breathing. http://www.abc.net.au/news/2018-10-07/wa-coastline-transformed-by-sea-levels-over-thousands-of-years/10338500 On 07/10/18 11:22, cleary wrote: In regard to admin boundaries sharing the coastline, I think that would also be incorrect but I am less confident of my view on this. I did update some administrative boundaries in South Australia using the SA Government Data and those boundaries did not coincide with the coastline (see the coastline around Streaky Bay, Ceduna, Fowlers Bay and near the Nullarbor). The coastline in OSM needs a lot of work - I have looked at parts of WA and SA but found it very difficult to use the satellite imagery correctly to refine the map of the coastline in areas where there are extensive mudflats or large tidal flows and even rocky areas just under/above the waterline. But I suggest it is safest, and more accurate, to map the administrative boundaries and the coastline separately. ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] PSMA Administrative Boundaries
On 07/10/18 11:22, cleary wrote: In regard to admin boundaries sharing the coastline, I think that would also be incorrect but I am less confident of my view on this. I did update some administrative boundaries in South Australia using the SA Government Data and those boundaries did not coincide with the coastline (see the coastline around Streaky Bay, Ceduna, Fowlers Bay and near the Nullarbor). The coastline in OSM needs a lot of work - I have looked at parts of WA and SA but found it very difficult to use the satellite imagery correctly to refine the map of the coastline in areas where there are extensive mudflats or large tidal flows and even rocky areas just under/above the waterline. But I suggest it is safest, and more accurate, to map the administrative boundaries and the coastline separately. Might be good to look at Broome for hi/low tide .. there is a fair distance between the two there so it would be easy to pick in that location. ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] PSMA Administrative Boundaries
In regard to admin boundaries sharing the coastline, I think that would also be incorrect but I am less confident of my view on this. I did update some administrative boundaries in South Australia using the SA Government Data and those boundaries did not coincide with the coastline (see the coastline around Streaky Bay, Ceduna, Fowlers Bay and near the Nullarbor). The coastline in OSM needs a lot of work - I have looked at parts of WA and SA but found it very difficult to use the satellite imagery correctly to refine the map of the coastline in areas where there are extensive mudflats or large tidal flows and even rocky areas just under/above the waterline. But I suggest it is safest, and more accurate, to map the administrative boundaries and the coastline separately. ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] PSMA Administrative Boundaries
On 06/10/18 21:34, Andrew Harvey wrote: Thanks for raising that. I'd seen some boundaries in WA defined in legislation as, follow this road, then that road etc. but I think that was for school zones. So the LGA and Suburb/Localities are defined by the cadastral plans then? I hear the points and see there is consensus to not reuse existing roads, rivers in the admin boundaries, so I support that approach. What about admin boundaries that border the coastline? Should they share the existing coastline or not? OSM has defined the 'coast line' as the high tide mark as that is easier to pick than the low or mid tide marks. It is probable that the admin boundaries use the low tide mark? Do a sample comparison? That does simplify the import, as there is much less manual effort needed. I guess what we need now is an OSM XML file with both the Suburb/Localities and LGA boundaries together with shared ways (as many ways are in common). I'll see what I can do to put this together, is anyone else working on this too? On Sat, 6 Oct 2018 at 20:53, cleary wrote: In regard to administrative boundaries being attached to other features such as waterways and roads, I think it is a trade-off between accuracy and convenience. I am most familiar with NSW. Boundaries are not "defined" by words but rather by surveyors' charts. The surveyors may often have been directed to use waterways, roads, mountain ridges and similar features for their surveys. However the waterways and roads have sometimes/often moved but the boundaries have not. Words are sometimes used to describe boundaries such as "it follows the river and then goes south along the main road ... " Such a description is approximate and is near enough for many purposes, especially if one's area of interest is well within the boundaries. However it may not be sufficiently precise if one is concerned with particular locations close to the boundaries. Examples in NSW that might be considered include the boundary on the Murray River west of Tocumwal, the Lachlan River east of Cobb Highway, Willandra Creek south of Roto, Bogan River at Girilambone. If the boundaries were attached to the respective waterways, either the boundaries or the waterways would be incorrect. Where boundaries are mapped on rivers or roads, mappers may re-align the river or road as changes occur and the administraitve boundary becomes distorted, sometimes only slightly but usually increasingly significant over time. Alternatively we could map the waterway or road using the administrative boundary data (as some mappers have done in the past) and ignore the satellite imagery and GPS data but this affects the accuracy of the location of the waterway or road. While I will accept the community's group decision, personally I think accuracy is to be valued over convenience. I strongly advocate for accuracy by mapping administrative boundaries separate from other features on the map, even if they are nearby. The decision in regard to the above issue will affect use of a source tag for the boundary. If the boundary is an approximation and attached to waterways or roads then it would be incorrect to use a boundary source tag, However if boundaries are mapped separately and accurately, then we should record the source of the boundary data. While I would suggest adding the source tag to the relation for the administrative boundary, it might also be added to the way if there is any need to specify the source for the way e.g. if using the administrative boundary for the geography of a river, then also give the source of the boundary data as the source for the waterway. ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] PSMA Administrative Boundaries
Thanks for raising that. I'd seen some boundaries in WA defined in legislation as, follow this road, then that road etc. but I think that was for school zones. So the LGA and Suburb/Localities are defined by the cadastral plans then? I hear the points and see there is consensus to not reuse existing roads, rivers in the admin boundaries, so I support that approach. What about admin boundaries that border the coastline? Should they share the existing coastline or not? That does simplify the import, as there is much less manual effort needed. I guess what we need now is an OSM XML file with both the Suburb/Localities and LGA boundaries together with shared ways (as many ways are in common). I'll see what I can do to put this together, is anyone else working on this too? On Sat, 6 Oct 2018 at 20:53, cleary wrote: > In regard to administrative boundaries being attached to other features such > as waterways and roads, I think it is a trade-off between accuracy and > convenience. > > I am most familiar with NSW. Boundaries are not "defined" by words but rather > by surveyors' charts. The surveyors may often have been directed to use > waterways, roads, mountain ridges and similar features for their surveys. > However the waterways and roads have sometimes/often moved but the boundaries > have not. Words are sometimes used to describe boundaries such as "it > follows the river and then goes south along the main road ... " Such a > description is approximate and is near enough for many purposes, especially > if one's area of interest is well within the boundaries. However it may not > be sufficiently precise if one is concerned with particular locations close > to the boundaries. > > > Examples in NSW that might be considered include the boundary on the Murray > River west of Tocumwal, the Lachlan River east of Cobb Highway, Willandra > Creek south of Roto, Bogan River at Girilambone. If the boundaries were > attached to the respective waterways, either the boundaries or the waterways > would be incorrect. Where boundaries are mapped on rivers or roads, mappers > may re-align the river or road as changes occur and the administraitve > boundary becomes distorted, sometimes only slightly but usually increasingly > significant over time. Alternatively we could map the waterway or road using > the administrative boundary data (as some mappers have done in the past) and > ignore the satellite imagery and GPS data but this affects the accuracy of > the location of the waterway or road. > > While I will accept the community's group decision, personally I think > accuracy is to be valued over convenience. I strongly advocate for accuracy > by mapping administrative boundaries separate from other features on the map, > even if they are nearby. > > The decision in regard to the above issue will affect use of a source tag for > the boundary. If the boundary is an approximation and attached to waterways > or roads then it would be incorrect to use a boundary source tag, However if > boundaries are mapped separately and accurately, then we should record the > source of the boundary data. While I would suggest adding the source tag to > the relation for the administrative boundary, it might also be added to the > way if there is any need to specify the source for the way e.g. if using the > administrative boundary for the geography of a river, then also give the > source of the boundary data as the source for the waterway. ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] PSMA Administrative Boundaries
On 06/10/18 20:52, cleary wrote: In regard to administrative boundaries being attached to other features such as waterways and roads, I think it is a trade-off between accuracy and convenience. +1. If the admin boundaries use other features as there boundaries then it is the other feature that takes priority in accuracy over that of the boundary. The tags on the way will have those of the other feature, possibly including the source of that other feature. If the admin boundary is moved because the other feature is changed then so be it. I have come across a few admin boundaries that are attached to things .. and from now on I'll move them to match the other feature, if that is a problem for you then make the admin boundary separate. I for one am tired of separating them. ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] PSMA Administrative Boundaries
In regard to administrative boundaries being attached to other features such as waterways and roads, I think it is a trade-off between accuracy and convenience. I am most familiar with NSW. Boundaries are not "defined" by words but rather by surveyors' charts. The surveyors may often have been directed to use waterways, roads, mountain ridges and similar features for their surveys. However the waterways and roads have sometimes/often moved but the boundaries have not. Words are sometimes used to describe boundaries such as "it follows the river and then goes south along the main road ... " Such a description is approximate and is near enough for many purposes, especially if one's area of interest is well within the boundaries. However it may not be sufficiently precise if one is concerned with particular locations close to the boundaries. Examples in NSW that might be considered include the boundary on the Murray River west of Tocumwal, the Lachlan River east of Cobb Highway, Willandra Creek south of Roto, Bogan River at Girilambone. If the boundaries were attached to the respective waterways, either the boundaries or the waterways would be incorrect. Where boundaries are mapped on rivers or roads, mappers may re-align the river or road as changes occur and the administraitve boundary becomes distorted, sometimes only slightly but usually increasingly significant over time. Alternatively we could map the waterway or road using the administrative boundary data (as some mappers have done in the past) and ignore the satellite imagery and GPS data but this affects the accuracy of the location of the waterway or road. While I will accept the community's group decision, personally I think accuracy is to be valued over convenience. I strongly advocate for accuracy by mapping administrative boundaries separate from other features on the map, even if they are nearby. The decision in regard to the above issue will affect use of a source tag for the boundary. If the boundary is an approximation and attached to waterways or roads then it would be incorrect to use a boundary source tag, However if boundaries are mapped separately and accurately, then we should record the source of the boundary data. While I would suggest adding the source tag to the relation for the administrative boundary, it might also be added to the way if there is any need to specify the source for the way e.g. if using the administrative boundary for the geography of a river, then also give the source of the boundary data as the source for the waterway. ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] PSMA Administrative Boundaries
>Do you have examples of the overlapping ways? It looks pretty okay around Brisbane to me. Here's an a mesh of the LGA's in GeoJSON which you can import into JOSM with enough memory. https://tianjara.net/data/LGA_mesh.geojson.xz Just import the shapefile into JOSM and use the validation. The GeoJSON file is looking much cleaner. >That's fairly easy to as a preprocessing step, eg with ogr2osm or via other scripts. I've been working on processing the PSMA data to make it easier to import. Since I think we should reuse existing ways where possible, if we did that it's a mostly manual process anyway. Even without reusing existing way, to get relations you need shared ways on the borders. One approach is to use https://github.com/andrewharvey/geojson-mesh to get single ways for the border which we then manually join up into the full relations in JOSM. We might just have to do this. It would be a lot faster then trying to find overlapping ways, at least on the LGA data. >If there's an existing place node, then we should use that as the label member of the relation Yes this is what I was intending. ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] PSMA Administrative Boundaries
On Sat, 6 Oct 2018 at 15:57, Joel H. wrote: > > OK everyone I am currently editing the LGA shapefiles for QLD so no one > should attempt as to not create conflicts (although I'm not currently working > on the suburbs file). Don't worry, no one should actually be doing any importing until we get community consensus and a plan in place. > The PSMA data isn't good IMO. And requires a lot of fixing to be imported, > lots of problems with overlapping ways. Do you have examples of the overlapping ways? It looks pretty okay around Brisbane to me. Here's an a mesh of the LGA's in GeoJSON which you can import into JOSM with enough memory. https://tianjara.net/data/LGA_mesh.geojson.xz > But the biggest problem is that the names are all caps! Does anyone know a > way to automatically convert all of these properly so that "BRISBANE" is > "Brisbane"? That's fairly easy to as a preprocessing step, eg with ogr2osm or via other scripts. I've been working on processing the PSMA data to make it easier to import. Since I think we should reuse existing ways where possible, if we did that it's a mostly manual process anyway. Even without reusing existing way, to get relations you need shared ways on the borders. One approach is to use https://github.com/andrewharvey/geojson-mesh to get single ways for the border which we then manually join up into the full relations in JOSM. > On 6/10/18 12:07 pm, Joel H. wrote: > My first though is to add a fixme notice (in similar fashion to the NSW > import), that tells us to reconsider the place label node already in OSM, and > to integrate it with the boundary data. If there's an existing place node, then we should use that as the label member of the relation. ___ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] PSMA Administrative Boundaries
On Fri, 5 Oct 2018 at 11:43, cleary wrote: > A month ago, we celebrated the news that OSM now has approval to use the PSMA > Administrative Boundaries and there was some discussion, including the need > for a proper import process. I am willing to start adding some boundaries in > areas with which I am familiar/interested but I am waiting for the proper > import process to be determined. I am not aware if anything has been done in > regard to a plan to import this data. If so, please guide me. Glad to see people are interested in bringing this data into OSM. I'm strongly of the view that we should discuss and document the import process before beginning, so naturally that's the first step. >If not, I propose the following: > Individual mappers download most recent data from > https://data.gov.au/dataset/psma-administrative-boundaries and add/check data > in their areas of interest and/or when they have time. (Andrew Harvey has > provided a script to assist working with PSMA Administrative Boundaries Data > and there is a link on > https://wiki.openstreetmap.org/wiki/Australian_Data_Catalogue) > > Administrative boundaries in NSW, SA, VIC and ACT be modified > instance-by-instance if there are discrepancies that require updating. Where > there are discrepancies, the most recent data is usually preferred . If there > are concerns about accuracy of the most recent data, further consultation > with other mappers or relevant government boundary authorities would be > appropriate. > > In QLD, WA, TAS and NT : LGA and Suburb/Locality boundaries be added one at > a time, integrating with existing data where appropriate. Previous data from > unauthorised sources be deleted or the authorised source be added if the data > is accurate. > > LGA Boundaries continue to be tagged as admin_level=6 > Suburb/Locality boundaries be tagged as admin_level=10 I agree. > In both instances source=PSMA_Admin_Boundaries_August_2018 > (or whichever date is applicable to the most recent data being used) > This source information may seem unwieldy but provides accuracy and > completeness of information. I'm on the fence on this, while the source tag on the feature can be very handy for mappers, I've also seen the downside where as other tags are added later on it becomes unclear what actually came from that source. Where do you propose this tag to go, on the relation or on the way members of the relation? Consistent with existing LGA boundaries in OSM I'm proposing we use relations with the tags: type=boundary boundary=administrative admin_level=6 place=municipality with optional tags short_name=Sydney name=City of Sydney wikidata= wikipedia= (optional with wikidata tag) website= phone= email= the admin boundary ways as "outer" members. Consistent with existing Suburb/Locality boundaries in OSM I'm proposing we use relations with the tags: type=boundary boundary=administrative admin_level=10 place=suburb name= with optional tags: postal_code (although there's not 1:1 match between postal areas and suburbs, I'm okay with adding a postal code that generally applies to the suburb so long as it's not coming from copyrighted sources) wikidata wikipedia (optional with wikidata tag) > Administrative boundaries NOT be attached to other ways such as creeks, > rivers, roads, etcetera so that the other features can be modified as needed > without affecting the administrative boundaries which are generally static. I disagree with that. If the admin boundary seems to follow the creek, river, road, coastline then we should use that existing way as part of the relation. I'd like to avoid a "mess" of multiple ways all very close to each other and overlapping. Plus if the boundary is defined as that river, road, coastline then by using the existing way we are able to capture that detail in OSM in a way that almost all other spatial data cannot. > Multiple administrative boundaries (including national_parks and government > approved protected_areas or state forests) be mapped on a single way, where > appropriate, and with multiple relations attached to that single way i.e. a > single way could form the boundaries for localities, LGAs and a national > park, if appropriate in the particular location. I agree. > Electoral boundaries NOT be added at this time. I agree. > Other boundaries which have been discussed, such as regions and indigenous > areas, NOT be added as part of this particular project unless they are LGAs > or Suburbs/Localities (some indigenous areas may be identified as protected > areas or defined localities or LGAs, in which case they are added and > identified on that basis). Mapping of regions could be further discussed > separately if required. They are usually not defined by legislation and > therefore usually not "administrative" in the way that LGAs and > Suburbs/Localities are legislated. I agree, plus these other types of regions aren't covered by the PSMA Admin
