Hi VLD103, thank you for improving the map here in Papua. We appreciate having more mappers at work adding the many missing roads, streets and rivers.
I noticed that you added the tag "import=yes" to many roads. This tag should only be used as a changeset tag when you import data from an outside source, like another database. If you are tracing roads from aerial imagery, don't use this tag, just set the source.

Also, I'd recommend considering using other types of highway, not usually highway=service or highway=track, for any roads that have houses or other buildings along them in rural areas. If there is a road that connects several villages, it's probably a highway=tertiary. If it's just a road through to one tiny village or a few houses, use highway=unclassified. And use highway=residential for roads with houses directly on them, like village streets.

Highway=track should be used for roads in plantations, farmland and forests that are just used for agriculture or forestry. On aerial imagery, these roads usually won't have any buildings along them and will not be the main way to access a town. They may be rather overgrown.

The tag highway=service is used for driveway, alleys, aisle in parking lots, and similar service roads within one property; usually these are private, not public roads. In remote areas there can be service roads at a mine or plantation, but they won't connect a village to another.

Sorry for the criticism. It took me a long time to figure these things out myself, and highway classification can be a little debatable sometimes. If you have no idea what type of road it is (or if its really a road, not a path for just walking and bikes), then you can always use highway=road, which lets other mappers know to add the right class later.

Have fun mapping!

> " Deleting the Pacific Ocean may make things easier for someone editing a beach in the California... but that hardly justifies it"

Fortunately oceans are not yet mapped as multipolygon relations. Can you imagine how difficult it would be to download and edit such a monster? But that will be next if the multipolygon supporters for bays and seas get done mapping those features.

I don't believe way 678904551 provides any real data: it's just an imaginary line used to create a sea polygon, and couldn't be verified on the ground. Can you imagine going to on of the end points and asking locals: "so the Timor sea is on my right hand when I stand right here, yes?". In contrast, you can do that with features like national borders.

Mapping seas as nodes is better and provides the same information (the coastline is the verifiable border) except for these imaginary lines that are sourced from some authority.

Sorry, I shouldn't have deleted that way without also deleting the relation and checking that there was a place=sea node in the right place. This was the case for the other seas that I fixed, but I wasn't careful with this one.

Furaha,

Terima kasih atas bantuan Anda dalam proyek OSM di tanah Papua.

Saya ada satu usulan. Kalau ada jalan ke satu dusun atau daerah dengan rumah-rumah, lebih baik pakai highway=residential atau highway=service daripada highway=track. Tag highway=track artinya satu jalan lewat kebun, sawah, lapangan tanian, atau lewat hutan, bukan jalan ke rumah-rumah ata dusun.

Terima kasih banyak, - Joseph Eisenberg

Furaha, thank you for all your work mapping roads in Papua for OSM. It's very helpful.

I have one suggestion for you about the tags to use for roads. If you are mapping a road that goes to a small village or hamlet or residential area with some houses, such as a dusun, please use highway=residential or highway=service, but not highway=track. The tag highway=track should be used for roads that lead to areas used for agriculture and forestry only, such as small tracks that lead through rice fields, gardens, orchards or areas of forest/woodland. If there is an inhabited village or even one house along the road, it's probably not a track.

Thanks, Joseph Eisenberg

Terima kasih, saya senang sekali bahwa Anda sedang menambah banyak gedung-gendung ke database Openstreetmap.

Saya ada satu usulan. Bisanya "name=" dipakai untuk nama yang dipakai pada umumnya. Kelihatanya bahwa Anda sedang menambah banyak gedung dengan nama "name=A", "name=B", "name=C" dan lain-lain. Mungkin ini bukan nama.

English:
Thank you for all your contributions to Openstreetmap. I really appreciate how man buildings you have been adding to the database.

I do have a suggestion. The field "name=*" should be used for the common name of a feature, only. Most buildings do not have names. I see you've added many buildings with "name=A", "name=B" etc - these aren't the locally used names. Perhaps you could use a different tag for these?

Sincerely,
Joseph Eisenberg,
in Wamena, Papua Province

Sorry about that. I was a new mapper at the time and had not tried to improve this sort of thing. I now know that this is a big pain when you want to go back and map the landuse precisely to the fence line or road edge.

Rather than ungluing and adjusting the nodes, I find it can be easier to cut the closed way and draw a new one, then combine again - at least if the landuse is a complex shape with many nodes, for example woodland and natural areas along curving roads or streams.

(Ungluing and using the "Improve way accuracy" mode in JOSM works fine for rectangular areas.)

My goal was to check all of the towns/cities in the area to make sure that they were tagged properly. I removed the import_uuid and gnis:* tags because I don't think they have any use, other than showing that the node was imported? Since I reviewed the nodes, I think the import tags can be removed.

I checked the census figures on Wikipedia. It looks like there are 17902 inhabitants in the urban area, while the whole population of Municipality of Igoumenitsa is 25814, including some villages. This still is still quite small for a city by global standards in openstreetmap, and even for Greece: only 6 places in Greece are tagged place=city with less than 20,000 inhabitants, and most Greek place=city have >50000. And in the whole Mediterranean region there are only 27 out over almost 800 place=city with less than 20,000 people - and all these are in Libya, Egypt, Greece and Albania. The western countries (Italy, Spain, Morocco, France etc) have no cities <20,000 people, nor do any countries in northern Europe. Worldwide, only 2.5% of place=city have less than 20,000 people and most are in really remote areas like Greenland, Canada, Australia, Siberia, etc, where the map is pretty empty. Most of place=city with a population listed have at least 100,000 people (3820 out of 6119)
But having said all that, please feel free to change the town back to a place=city if you think this is more correct. It's rather subjective. And I've updated the population figure to include the whole urbanized area, which will help in the future.

Hi Warin, I noticed that you used natural=mountain_range and landform=mountain_range for a bunch of features in this area.
Do you think they meet the definition for natural=ridge instead? I would think of a mountain range as containing several end-to-end ridges, eg: ways 535024744, 535024743, 535024741... etc. ... 535024732 ... to 535024728 - if these are considered one mountain range locally.

Thank you for your response, Olyon. I appreciate your willingness to communicate in English.

> "Many things in osm do not have clearly defined borders"

These should not be mapped as areas. The coastline is well-defined: it's the waterline at high tide. While much of the coastline in OSM is still traced pretty roughly, the coast of much of Europe has been mapped very precisely. This is possible because there is a clear definition. If someone maps the coastline at the low tide line, then another mapper will come along and fix it later.

Landuses and natural landcover like forests, woods, farmland and meadows can be precisely mapped from good aerial imagery or on-the-ground survey, with accuracy within a few meters.

In contrast, the borders of the Bay of Bengal that you've mapped are debatable within 100's of kilometers. Even the source that you linked, "International Hydrographic Organization S-23 draft 2002", only provides a rough outline of the area, not even precise to the nearest kilometer.

Re: "I think there is no doubt, along the east coast of sri lanka it i the bay of bengal"

Is this based on local knowledge? I doubt that local people on the east coast of Sri Lanka say "that's the Bay of Bengal" if you ask them the name of the sea off of the coast of their village. I'd expect "that's the Indian Ocean", or just "the sea" or "the ocean".

And I'd be very surprised if anyone in Aceh, Indonesia says "The Bay of Bengal starts a few kilometers northwest of here".

> "2) As the name suggests it is a gulf and a gulf is a big bay."

Merriam-Webster Dictionary defines bay as "an inlet of the sea or other body of water usually smaller than a gulf"

If the tag natural=bay is going to have a useful meaning in openstreetmap, it needs to be defined in a clear way. Bays should be a part of the sea (or a lake) that goes into the land, and so it should be bordered by continuous coastline on most of the sides. An archipelago or chains of islands shouldn't be the border of a bay.

In contrast, place=sea can be bordered by islands; eg the Caribbean Sea and Java Sea - but this means that it isn't possible to map a sea as a precise area.

> "It is delimited on 3 sides by land even if on one side it is partly islands."

Most of the east border of the Bay of Bengal as you have mapped it is made up of open sea with a few islands.

The concept of verifiability means that 2 different mappers who observer the same feature should agree on how it is drawn and tagged (osm.wiki/Verifiability). Clearly this is not the case for this feature, either for it's borders or for the tag natural=bay.

But I think most mappers would agree that a node tagged place=sea and placed south of West Bengal and Bangladesh would be reasonable. This is my recommendation for how to map this sea, and other place=sea

Olyon, I see that you changed the Bay of Bengal place=sea from a node to a multipolygon relation a few days. ago.

I disagree with two things about this:
1) Seas should not be mapped as multipolygon relations.
a) Their borders are not clearly defined; while you've linked to a IHO document, we map "on-the-ground", de facto information, not what's officially prescribed. The IHO's definition leads to a very strange outline of this area: look at the southeastern coast of Sri Lanka. How can that thin area be part of the Bay of Bengal, which is associated with West Bengal in India and Bangladesh?
b) Huge multipolgyons like this are very hard to maintain and download. This one crashed my laptop
c) Mapping seas as nodes is the established practice. Until a few months ago, almost 90% of seas were mapped as nodes
2) This object should not be tagged natural=bay. It's about the same size as the Mediterranean sea, which has a dozen smaller seas within it. While there isn't a 100% certain division between bays and seas, this is certainly so large that it should be a sea. Also consider that half of the boundary that you've mapped is along islands or open ocean water; seas like the Caribbean and the South China sea are often delineated by island arches, but bays are not.

Oh, I should mention that natural=atoll is a great tag for the closed ways that you have drawn! I'm going to make a wiki page to describe this tag, it's a great idea.
I would just recommend removing the place=archipelago tag, and making new relations for those features, just including the islands.

Hi Olyon, I noticed that you have edited or created a number of place=archipelago areas in the Maldives. Thanks!
I would recommended to map an archipelago with a multipolygon relation, by selecting all of the coastlines of each of the islands that make up the archipelago.
This will make it easier for database users, including Openstreetmap-Carto (the standard map style on the openstreetmap.org website) to render archipelago names.
See: osm.wiki/Tag:place%3Darchipelago

I don't see these two rivers in the Bing aerial imagery that you listed as a source, and I don't think I can see them from the nearby hill in Sentani. Did you base this off of some other aerial imagery?

Thanks for responding, fatur. What project are you doing? Is it with HOT / Humanitarian?

About the roads, if you don't know the area well, you can just use highway=road with surface=unpaved or whatever is appropriate.

But the Indonesian guidelines suggest using highway=primary for the main provincial roads, eg Jalan TransPapua (or TransSulawesi, etc) which link the ibukota kabupaten to each other. These towns are all tagged place=town, I think. Myself, I also use highway=primary for other major roads that link place=town to another place=town, even if they are not paved.

But if you find roads that go out to a village (desa), those can be highway=secondary. If they go to very small villages (dusun) or hamlets, they can be highway=tertiary. And if they go to a just a few houses or farms, that's highway=unclassified.

highway=track is for roads in perkebuan or hutan, plantations, farms and forests, which don't connect to inhabited places and are used just by farmers and woodcutters.

(Do you speak Bahasa Indonesia? I can write in Indonesian if that's helpful)

Indonesian tagging guidelines:
osm.wiki/Id:Indonesian_Tagging_Guidelines

English version:
osm.wiki/Indonesian_Tagging_Guidelines

(Bahasa Indonesia di bawa)

Hello fatur, thank you for helping improve the map in Papua!
I see you've been upgrading some roads to highway=primary in Papua. Have you visited this area recently? Normally a highway=primary would connect a town to a village, and a place=village should have more than 100 people. See osm.wiki/Tag:highway%3Dprimary

I think most of the roads in this changeset are probably highway=secondary which connect villages to other village, or a village to a hamlet (with less than 200 people), or highway=tertiary, which connect hamlets and other very small places.

Bahasa:
Terima kasha atas kerja keras ini untuk memperbaiki peta dan data Openstreetmap di tanah Papua.

Saya sudah lihat bahwa Anda sedang menganti jalan-jalan ke highway=primary. Tetapi menerut wiki.openstreetmap.org, tag "highway=primary" artinya "Jalan raya yang menghubungkan kota-kota atau "town" besar"

Saya belum kenal kota kasonaweja dengan baik, tapi menerut foto satellite dari Bing, saya berpikir bahwa jalan-jalan ini adalah highway=secondary atau highway=tertiary. Karena highway=secondary ada di antara desa-desa, dan highway-tertiary jalan dari dusun kepada desa atau dusun lain.

Hello fatur, thank you for helping improve the map in Papua!

(Bahasa Indonesia di bawa)

English:
I see you've been upgrading some roads to highway=primary in Papua. Have you visited this area recently? Normally a highway=primary would connect a town to a village, and a place=village should have more than 100 people. See osm.wiki/Tag:highway%3Dprimary

I think most of the roads in this changeset are probably highway=secondary which connect villages to other village, or a village to a hamlet (with less than 200 people), or highway=tertiary, which connect hamlets and other very small places.

Bahasa:
Terima kasha atas kerja keras ini untuk memperbaiki peta dan data Openstreetmap di tanah Papua.

Saya sudah lihat bahwa Anda sedang menganti jalan-jalan ke highway=primary. Tetapi menerut wiki.openstreetmap.org, tag "highway=primary" artinya "Jalan raya yang menghubungkan kota-kota atau "town" besar"

Saya belum kenal kota kasonaweja dengan baik, tapi menerut foto satellite dari Bing, saya berpikir bahwa jalan-jalan ini adalah highway=secondary atau highway=tertiary. Karena highway=secondary ada di antara desa-desa, dan highway-tertiary jalan dari dusun kepada desa atau dusun lain.

Thanks, I didn't know about that. That sounds much easier than trying to move a selection to a different layer, which is what I attempted.

Hello Berobispo. Thanks again for all the great mapping you are doing in Papua!
We just discussed coastline vs river areas in mangroves on the Tagging mailing list, and the consensus is that these small channels in the mangroves are actually tidal channels, not true rivers. They are salt water, and flow both ways depending on the tide. Christoph Hormann recommended mapping these as natural=coastline. Of course they can also be tagged with natural=water salt=yes (as a multipolygon) in addition, but since they are salt water channels they are part of the marine environment.
I made this simple mistake in many places, and I'm going to work on changing them.
(See https://lists.openstreetmap.org/pipermail/tagging/2018-September/038912.html)

I believe you should revert the change.

We discussed this on the Tagging mailing list: https://lists.openstreetmap.org/pipermail/tagging/2018-September/038912.html

One comment from Christoph Hormann: "The coastline closure there is both below the lower limit of the proposal and below the the range i
can imagine a meaningful coastline closure rule to allow."
...
"That is largely not really an estuary but more of a ria. I have no data for this at hand but you can likely see an abrupt change in the elevation profile near Totnes where the submerged section of the former river valley starts. So in this case it would make a lot of sense to place the coastline closure near the upper end of the tidal section because this is much better defined in terms of physical geography."

Also, on the GP tagging list this issue was discussed recently: https://lists.openstreetmap.org/pipermail/talk-gb/2018-August/021885.html (and following messages):

"The wiki suggests the coastline should be
the high water line going up to the tidal limit (often a lock or a wier) but this can be a substantial distance inland. This is AIUI the general scientific approach." - Colin Smale

"The River Dart [Way: 194211894 waterway= riverbank] is a really arbitrary line across the river from Dartmouth Castle, this offends my view of what a coastline is." -TonyS999

Please fix this

Oops, you are right, it should be old_name, thanks for catching that.
I'm using "old_name" for older spellings. Indonesian used the letter "j" more often in the past, due to Dutch influence I think. But in this case "alt_name" would also be fine, since it is still used on some government maps