I’m glad that a forum for Turkey has been established now, because while travelling and mapping in Turkey, I found some issues.
The first one I’d like to mention is the mapping of some junctions in Turkish trunk roads that have circular islands in them but are otherwise normal junctions of trunk roads with more minor roads. However they are often mapped as circular junctions. An example is here. Because there is a way tagged with junction = circular as part of the main course of the trunk road, a navigation app will tell a driver approaching this junction to “enter the roundabout and take the second exit” when he wants to go straight on the trunk road. However in reality the trunk road just goes straight on without interruption, and without needing to do anything or even slow down. That the navigation system gives instructions where they are not necessary gets annoying after some time. I think such junctions should be mapped like this one. It used to be mapped as described above, but I edited it in this changeset. I repaired quite few of them in western Turkey, but I’m sure there are plenty left. Maybe mappers in Turkey should put it on their to do list to check such junctions in their local area and edit them when necessary? They should not be difficult to find using an overpass query for junction = circular in major roads. Or maybe a Maproulette quest could be created for them? Maybe @sbaido could help?
Please reply in Turkish. I’ll hit the translate button to read it.
Hi @rhhs, thanks for dropping by to Türkiye subcategory!
I am also part of TomTom’s Community & Partnerships team and working closely with @sbaido, also active in the Turkish community and moderation as @unen.
You have brough up an excellent point which was also previously discussed in the Telegram channel as well. I also think the circular islands are very misleading and a lot of editors are interpreting the junction as circular.
I do my best to update this misinterpretations if I encounter them, and have in my mind to take up editing country-wide.
Sadece etiket değil, geometriyi de düzenlemek gerekli.
Dairesel geometriyi iptal edip iki parçayı da ayrı ayrı düzenlemek, junction=* etiketini kaldırmak, diğer etiketleri korumak, ve gerekiyorsa oneway=yes etiketi eklemek.
As I have noted earlier on the wiki, the Turkish highway department (KGM/T.C. K.) has a concise explanatory PDF and some videos on the topic available on the official website. (You’d likely not be able to reach the document as Turkish government sites are usually geoblocked—DM and I’d provide an archive.org link. It’s public domain)
TL;DR: Such junctions are usually considered as Main Road-Side Road Junction (Ana Yol-Tali Yol Kavşağı), where the state road (devlet yolu) chausee has the right-of-way and not the turning lanes. Such junctions are not marked with the roundabout sign, but sometimes are marked with U-turn signs if it only is a detour; and its segments shall be marked on OSM with junction=circular. Whereas a Modern Roundabout (Modern Dönel Kavşak) is marked with a roundabout sign and its exits/ramps are pretty visible from satellite&must be marked with junction=roundabout.
Google translates the text as At such intersections, unlike modern roundabouts, vehicles coming from the secondary road must stop and give way to vehicles coming from the main road, i.e it just describes the right of ways and that they are different from those on a roundabout.
I think it is wrong to map such junctions with junction=circular because navigation software will interpret them as circular roads with an entrance and exit. However this is not the case with these Turkish junctions: a driver on the main road will not experience them as junctions with an entrance and exit. Maybe the wiki text about Ana Yol - Tali Yol Kavşağı should be revised? Or are you saying that navigation software is interpreting junction=circular wrong?
A driver in Turkey would barely sense a difference between a circular/roundabout and turning left/right. “Exit from nth exit” always seemed like a bad translation to me and most people are just confused, scrambling to count the exits. Some of the existing examples also fit circular—side road just directly connects into the main road without a physical barrier for oneway separation, thus not the same nodes for entrance-exit (unless one does an arbitrary separation on OSM for convenience.) There is very rarely a signed priority in Turkey, it’s usually based on which road is wider. Only on the newer roundabouts drivers sense a difference—and feel that it’s dangerous and useless
Saw this today, which is something not uncommon. The mapper thought the circular island makes it a roundabout, then couldn’t figure out how vehicles exiting through the junction would connect to the secondary road, and improvised Quite old too, dating back to 2019.
The level crossing is not correctly mapped either in my opinion. I would map this as 2 one-way ways. This is a common configuration in Turkey to prevent road users zigzagging across the (usually) half barriers.
I updated there as this already. I think too, we are adding circular geometries unnecessarily a lot. Also the two connections were not tagged as trunk links.
The roundabout you gave as an example was edited 7 years ago. The intersection appears in the 2014 satellite images, and the data is outdated. However, in some areas, there may be highway arrangements that are not reflected in the satellite image. It is necessary to check the arrangement date.
For example, this roundabout drawing is inconsistent with the satellite image, but the satellite images are old.
Hi there, this has been a very interesting discussion that I’ve been following closely, thank you @rhhs for bringing up the issue and everyone for participating.
I encountered very similar intersections in Casablanca, Morocco. I know this is the Türkiye forum, but the situation is essentially the same, so I thought it might be helpful to share.
This is a satellite image of one of the intersections. It has a round small island, but in reality people on the main road just go straight and don’t give way to anyone:
You are quite correct. Bing satellite picture showing in OSM relative to this roundabout is dated 2021 and this area changed a lot since that time. Hopefully, we have an Esri picture dated 2024.
I was myself updating OSM to insert a new bridge replacing a railway level crossing at that location. This bridge was completed in 2019.
Maybe not, because the way you suggest, the routing tools would say:
”Make a right turn to the main road, then make a left turn, then make another left turn, then make a right turn to the residential road” instead of continue straight.
If there was no physical separation with the small island, life would have been much simpler as a mapper, but probably the one like below would be optimal:
I still haven’t found out what the rules are…
