This week I flew through Istanbul for the third and fourth times in a month, this time en route to Doha from Ashgabat. Since Fly Dubai no longer serves Doha, it takes 24 hours for me to get to Doha via Istanbul on Turkish Air, versus four to five hours via Dubai in the good old days, due to the lengthy layover in Istanbul. The good news is I love Istanbul, having lived there 30 years ago, and use the layovers to visit old haunts and old friends..

I was surprised to find somebody had mapped the American Consulate General in Istanbul as a office=commercial, and set about correcting that. I then discovered that every other consulate in Istanbul was mistagged amenity=embassy. Embassies are in capitals, are defined by the Vienna Convention on Diplomatic Relations, and are headed by an ambassador or chargé d’affaires. Consulates are a different animal, as defined by the Vienna Convention on Consular Relations, and are headed by a consul (if a consulate) or a consul general (if a consulate general, which is just a bigger consulate). .I have gone through Istanbul with my electronic red pen, correcting this mistagging plus adding multilngual names (for some reason the Chinese consulate general was identified solely in Ukrainian).

I turns out the wiki entry for the diplomatic=consulate tag contains erroneous information and people are systematically mistagging consulates. I need to take this up with the Data Working Group. Consulates are not embassies, and we need to strive for accuracy in our map.

Ann and I took a break from work to fly via Istanbul to Dushanbe, Tajikistan, where after 40 years of variously studying, living, working, and traveling in the former USSR I finally made it to the 15th former Soviet republic. Somehow I had never managed until last week to visit Dushanbe. Of course I could not roam the city without collecting Mapillary imagery (mostly uploaded by now) and adding to OSM interesting sites we visited, or updating and augmenting the tags on existing entries.

One of the more interesting and engaging sites we updated was the three-room Museum of Musical Instruments. Watch this video with the sound turned up!

We also stopped in Istanbul en route and I visited the old neighborhood where I lived 1988-1990. I’ll be adding names of some streets in that section of town based on Mapillary imagery.

Today I finished adding the M37 route relation to the entire M37 highway inside Turkmenistan, having started a few days ago at the Turkmenbashy International Seaport ferry terminal and ending today at the Farap border crossing with Uzbekistan. This is the definitive route designation according to the UN Asian Highway Network database. While at it I added some U-turns to the newer parts of the dual carriageway and tweaked the track of the highway based on updated sat imagery as well as my own Mapillary ground imagery.

I have now done this for most of the national highways in Turkmenistan. The P-8 is outstanding. I need to find someone here who can tell me which ways are part of the P-8 highway. Local knowledge is key in this country. If I can knock out the P-8 highway, I’ll have a pretty good fix on the national highway network of Turkmenistan and can move on to another systemic gap.

For the first time I used the smoothness=* tag. A day trip to Sarahs took me via Tejen and the P-7 national highway–pretty good road–then I returned via the P-9 national “highway” via Hauz Han. Holy smoke! It took us 2-1/2 hours to cover about 90 kilometers, for an average ground speed of 36 kph. There were potholes the size of a truck with potholes inside them! We drove on the sand in a few places because it was faster and smoother than the P-9.

I think this is the first time I have felt compelled to use the smoothness=* tag, and after reading the wiki article, I assigned it “very bad”. It is driveable in a light-duty SUV and perhaps even a passenger car, if one drives gingerly, but going is slow. It is, I think, the worst road I have seen in this country.

Sarahs itself has some interesting things to see. Polish and Turkmen archeologists are excavating the site of Ancient Sarahs, a city with layers going as far back as 5,000 years ago. On a more mundane OSM-related note, we discovered two new gas stations and collected the station number of a third one on the P-7 highway.

Monday my schedule was inexplicably scrapped. An all-day conference failed to materialize. We went for a drive, a day trip first to Nohur, where Alexander the Great left behind a garrison some thousands of years ago, and where some inhabitants to this day have green eyes bequeathed by their Greek ancestors, then onward, to the Sumbar River valley. I identified two mysterious villages that appear on imagery but till then in OSM had no names: Kürüždeý and Duzlydepe. We drove over dirt and gravel mountain roads, down steep valleys, and confirmed the tracks of some roads, including one that follows a riverbed for a little way (the water was not deep this time of year).

I also found that a road someone drew “speculatively” (his word) on the map about four years ago doesn’t really exist, at least not where it was drawn. Upon return I adjusted it to the GPS trace as far as we went, but deleted what I could not confirm. Locals said the road goes no farther any more due to mud- and rock slides, so that part of the OSM map is more accurate now.

Mapillary imagery is being uploaded as I type this. The scenery was nice, but in springtime it must be spectacular. I will have to go back when the grass is green and the flowers are blooming.

Based on my meetings with the minister of motor roads and the head of the Turkmen Motor Roads State Concern “Türkmenawtoýollary” plus interviews of local drivers, hunters, fishermen, and others here, I have identified many of the segments of the national highway system in Turkmenistan and as of this evening have mapped all but one of them. The P-8 “highway”, really a dirt road, needs GPS traces to be collected before it can be properly identified, and that is problematic since it is in a border zone more or less off-limits to foreigners. I know people who have driven that route but they are unable to verify its route from satellite imagery.

We are still missing the route number for the highway from Kerki to Ymamnazar. It overlaps with the P-36 part of the way, but not all of the way.

There may be other “national highway” routes out there and I’ll continue to keep my eyes peeled for them.

Some local mappers have taken it upon themselves to add street names in some towns in Turkmenistan following my master classes and presentations on mapping. That is gratifying. New POIs are popping up, as well. If we could only find all the gas stations!

Today I arose at 0400 hours to catch an early flight to Turkmenbashy, then upon arrival there rode a bus for three hours to Garabogaz over the bumpy P-18 “highway” (actually a 1-1/2 lane country road most of the way) to witness the opening of the Garabogaz Carbamide (Urea) Plant. This will be an important augmentation of the export potential of Turkmenistan so is something that needs to be mapped, including contact information, which I added this evening upon return to Ashgabat. A small number of Mapillary images are being uploaded as I type this.

A friend with connections in Riga obtained for me reprints of uncopyrighted Soviet military maps from the 1970s of all of Turkmenistan. I’m not sure how useful they will be as a lot has changed here since the 1970s (particularly the names of municipalities) but at least they will provide some historical names that may be useful and of interest. If anybody has experience working with Soviet-era maps and has tips or suggestions, please let me know!

Proper mapping requires ‘‘en situ’’ observation of the POIs to be mapped. Ann and I accordingly took a break from our other duties to investigate the Argentina Pub, conveniently located about 10 minutes’ driving time from our house. Good pizza and salad (I was driving so alas could not partake of any nectar). It also offers steaks, which we will also investigate in due course. osm.org/way/621972265

That said we have broken 300,000 Mapillary images and 11,000 kilometers of imagery collection.

Last week I met the head of the Turkmen Motor Roads State Concern, and he confirmed my list of national highways was nearly complete (one highway was missing, which he provided but which still lacks a route number; I have added it to the list in the wiki). My next “armchair” project when duties preclude getting out on the road will be to identify the national highways with appropriate tags. If Turkmenistan is to become a transportation hub for Central Asia, it will need a decent online road atlas. OSM is currently the best map of the country available to the general public.

Last week I took two days to tour parts of southern Balkan Province in Turkmenistan, hitting the towns of Etrek, Esenguly, and Ekerem, as well as the villages of Oboy and Daneata (the villages are now on the OSM map based on this ground truth). En route for the first time I filled a 32 Gb memory card with Mapillary imagery and had to install a backup SD card in the Samsung Galaxy smartphone I use for collecting ground-level imagery. Those images are being uploaded now. It may take a few days, since the internet here is so slow.

It will take a while to enter all the edits from data collected on this trip. Much of the M37 is now four-lane dual carriageway (divided highway) so I need to add all the U-turns I marked on the Garmin GPS as well as lots of POIs.

The route took me within a stone’s throw of the border with Iran, but we only saw a few Iranian tractor-semitrailer rigs on the roads. Road conditions between Etrek and Ekerem were rough to very rough, with speed averaging 25 to 30 mph (40 to 50 kph). GPS traces have been uploaded from the Garmin nuvi, Pocket Earth, and Mapillary gpx file.

We discovered two new gas stations and confirmed a suspected third one. In Ekerem (formerly called Okarem, but the signs we saw are clearly labeled Ekerem) we lunched on Turkmen meat-filled somsas that cost 2 manat (about 57 cents) apiece from the only cafe in town while tailgating on the beach.

Hmm, OSM Belgium published an interview with me about my mapping in Turkmenistan. You can read it here: https://openstreetmap.be/en/motm/2018/08/10/allan-mustard.html

Today’s newspaper carried news and a map of an internal boundary shift of two of Ashgabat’s boroughs (“etraplar”). About 1,376 hectares of land has been shifted from Buzmeyin etrap to Bagtyyarlyk etrap. More map editing ahead.

The newspaper also said the mayor proposed annexing more territory into the city, which means it is highly likely the outer boundary of Ashgabat will change again in the foreseeable future. Even more map editing ahead!

Saturday we went for a drive, to scope out the border crossing to Iran at Artyk, about an hour and a quarter’s drive east of Ashgabat, mostly out of curiosity. I’m glad we did–there is a new bypass road from the M37 straight to the border crossing that had not been mapped and does not appear on any satellite imagery in the OSM editors. We collected GPS traces and Mapillary imagery (now being uploaded) and will work on updating this approach. While there, we mapped most of the village of Artyk, collecting street names (lots of signage!)

Then on the return we stopped in Balykçylyk to find out if we can buy fresh fish from the fish farm there (yes, we can!) and mapped most of that village as well as the fish farm itself. Our final major mapping foray was Ýaşlyk, site of the Berk brewery as well as, it turns out, a rather large jail.

I had never driven the old highway that used to be the only road connecting Ashgabat and Mary, so we decided to take it back to Ashgabat instead of the M37. We discovered that the road stops abruptly at what used to be a railroad crossing, but now is a dead end at the railroad tracks. We used OSM to navigate on back roads till we came out on the M37 again, but had we not taken the alternate route, we would not have noticed in satellite imagery that the railroad crossing had been decommissioned.

MAPS.ME is a great app, and Ashgabat users suffering from high internet costs (the highest in the world, we are told) have found it a cost-effective way of uploading POIs. In certain important respects, this is a boon to mapping Ashgabat and other cities in Turkmenistan.

There are, however, two issues. First, the mappers often mis-tag POIs, either because they don’t know English too well (some mappers tag all medical or health facilities “amenity=hospital”, for example, including pharmacies, clinics, and ambulance stations), or because MAPS.ME’s limited universe of point tags doesn’t lend itself to proper tagging. Second, because MAPS.ME’s offline maps are updated relatively infrequently, multiple mappers add a POI multiple times, so you end up with two or three new entries for the same POI. This has created a fair amount of work for me, cleaning up after entry of POIs that either duplicate an existing node, usually consolidating all data from multiple entries, or simply correcting tags. I’m not complaining–there are worse problems (like vandalism), but it does take time.

The good news is that this weekend temperatures in Ashgabat will be well over 100 Fahrenheit, mid-40s Celsius, so I have a great excuse to stay close to the air conditioner and edit the map.

Yesterday I had too many other obligations to travel far, so I took about three hours of Saturday afternoon to cruise parts of Ashgabat I’d not visited for a while and to collect Mapillary imagery where there were gaps (uploads currently underway). The dacha community north of the Karakum Canal on the far east end of the city is gone. It was apparently demolished in the past months and has been bulldozed. I had no idea, but will have to revise the map. The road into the community has been blocked with a dirt pile. This is why ground truth is so important.

I have completed corrections to the A-381, P-1, P-15, P-16, and P-18 national highway routes in Turkmenistan. I have consulted with local sources to nail down exactly which roads are part of each route and think these five are now pretty accurate.

I have found an authoritative but slightly out-of-date list of provinces, districts, and municipalities in Turkmenistan, have corrected and updated it based on recent press releases, and posted it to the OSM wiki:

osm.wiki/Districts_in_Turkmenistan

I am also working with my embassy’s motor pool to identify the national highways, which are either not tagged or are in some cases mistagged. This will be a bit of a project over the summer as we research the Turkmen national highway system and bring the OSM map up to date. My ambition is to have a complete map of Turkmenistan’s national highway network, all properly tagged.

I have been unable to find an authoritative official list of the districts (‘‘etraplar’’) in Turkmenistan’s five provinces, so have cobbled together a list for each province based on whatever official press releases, news items, and other sources (including personal travel to some of them) I could find. The results have now been typed into the wiki pages for each of the provinces:

• Ahal Province
• Balkan Province
• Dashoguz Province
• Lebap Province
• Mary Province

If anybody has corrections to any errors of omission or commission, please make them and let me know! These lists are accurate and up to date so far as I can tell, but nobody’s perfect.

As of today I have broken a quarter of a million images uploaded to Mapillary that any OSM user can use for helping map Turkmenistan. I still have over 20K more images to upload from the Serhetabad trip and last Monday’s day trip to Magtymguly (Garry Gala).

Sometime this year I will depart Turkmenistan, so am feeling some pressure to collect as much data as possible before moving back to the United States.

Monday we drove to Magtymguly, a village in Balkan Province, formerly known as Garry Gala, to check out the area. You can see my route in the Mapillary GPS trace here. Mapillary imagery will be uploaded as Internet connectivity permits, so check back in a couple of weeks (it will take a while).

We added some villages, a gas station, some other POIs, and I mapped a river, the Aydere (which in Turkmen means “Moon River”). The Sumbar River valley is beautiful, very scenic, and there are some gorgeous mountain scenes that will eventually show up in Mapillary for those interested.

One section of the route took us through a landscape known as the “Mountains of the Moon”. I shot a Mapillary sequence with my iPhone, which you can see here: https://www.mapillary.com/app/?lat=38.45498688522582&lng=56.20510593993236&z=17&pKey=F1UZu4iMblKil4hVAcwPeA&focus=photo

Just came across this in my news feed. Don’t know if he uses OSM data, but in any event the video is compelling.