Talk:GNS Import for Yemen and Others
Problems with the GNS dataset
Please look at the GEOnet Names Server page and especially the "Caveats" section and tell us how you intend to mitigate the problems with this dataset, preferably before importing anything. --Lyx (talk) 23:22, 18 April 2021 (UTC)
GEOnet Names Server page was my main guide in importing. I will write about it. --حبيشان (talk) 07:51, 20 April 2021 (UTC)
see "Working with GNS Caveats" section. --حبيشان (talk) 09:54, 20 April 2021 (UTC)
Tagging suggestions
This is an ambitious project and I hope you're able to use the GNS data to improve mapping in this area.
I have some suggestions for you on the mapping from GNS feature designation code to OSM tags:
| Feature Designation Code | Description | Proposed OSM Tags | Suggested OSM Tags | Notes |
|---|---|---|---|---|
| DPR | depression(s) | place=locality | natural=depression? | The natural=depression tag is not formally documented but it does have some organic use. |
| DUNE | dune(s) | natural=dune | landform=dune_system | Mapping individual dunes may be uncommon, since individual dunes are often too small and too mobile to be mapped effectively. Consider landform=dune_system if the name applies to an area of dunes rather than a single individual dune. |
| HDLD | headland | natural=peninsula | natural=cape | The natural=cape definition explicitly includes headlands. |
| HLLS | hills | natural=peak | natural=hills | The natural=hills tag is analogous to natural=mountain_range but for areas of smaller relief. It is partially documented on the natural=hill page and is in use. |
| PLAT | plateau | place=locality | natural=plateau | |
| PLN | plain(s) | place=locality | natural=plain? | The natural=plain tag is not documented but has some organic use. |
| PPLQ | abandoned populated place | place=locality | abandoned:place=hamlet | The Lifecycle prefix makes sense here. |
| PROM | promontory(-ies) | natural=peak | * | This could also be mapped as natural=ridge or natural=cliff depending on the landform. Some manual review is probably required. |
| RF | reef(s) | natural=reef | * | OSM uses natural=reef to refer to a ridge or shoal of rock, coral or similar relatively stable material, lying beneath the surface of a natural body of water. Make sure GNS uses it the same way. |
| RVN | ravine(s) | natural=gorge | natural=valley | A ravine could be mapped as natural=gorge, natural=gully, or natural=valley depending on the feature. Using natural=valley is likely more generic unless each feature is manually reviewed. |
| SALT | salt area | landuse=salt_pond | * | The landuse=salt_pond tag applies only to man-made ponds used for salt extraction. If the feature is natural, consider some of the alternate tagging in Proposal:Dry_lake. |
| SBKH | sabkha(s) | landuse=salt_pond | * | The note above for SALT might apply here too.
As I understand it, the term "sabhka" has been applied to two different types of landforms: a coastal wetland which may dry leaving a surface of salt behind, or an inland dry lake which accumulates a salt surface through evaporation. For the former, consider natural=wetland + wetland=saltmarsh + tidal=yes unless the area is no longer tidal. For the latter or perennially dry coastal areas, consider the alternatives in Proposal:Dry_lake. |
| SHOL | shoal(s) | natural=shoal | * | OSM uses natural=shoal to refer to a natural submerged sandbank that rises from a body of water to near the surface. Make sure GNS uses it in the same way. |
| TRGD | interdune trough(s) | place=locality | natural=blowout?? | The natural=blowout tag is undocumented and experimental, but it does describe this type of feature. |
| UPLD | upland | place=locality | * | This could be mapped as natural=hill, natural=ridge, natural=peak, or natural=plateau and may need manual review. |
| WAD,WADX | wadi,section of a wadi | natural=valley | natural=wadi | You might use the tagging suggested in Proposal:natural=wadi. |
In general, you might consider some of the tags in the Glossary of landforms as ways to make the tagging for imported features richer.
--B1tw153 (talk) 15:25, 30 April 2024 (UTC)
Automated conflation
If the proposal is to search for matching features within +/- 0.0004 degrees of latitude and longitude, that's roughly a search radius of about 45 m. That area may be much too small to find many of the features that should be conflated.
Consider the following nodes which are possibly duplicates of GNS ID -2033134:
These nodes are spread over an area of almost 2.5 km. At a minimum, a much larger search area may be necessary to find potential matches for conflation.
[Edit] There's also the case where a feature has been mapped in OSM as a large area (e.g., a large lake). Attempts to match large areas using individual coordinates may fail because the feature may be too large to intersect the search area. In this case, you really need to search for enclosing features as well. You may be able to do that with PostGIS but you would need to make sure you're including that case in your queries.
If you're interested, I have been working on software that does exactly this type of matching for USGS GNIS records. I have already tried to use it with GNS records and it's clear it could be adapted to handle this data with some minor modifications. The software is described in User:B1tw153/recoGNISer and in the Github repo. If that's something you're interested in, get in touch and we can work together on it.