kmpoppe's Diary

Small Towns in Europe

Posted by kmpoppe on 1 February 2024 in English. Last updated on 2 February 2024.

TL;DR: There are quite a few places on the continent that still need mapping. If you want to know why and how I came to that conclusion, read on ;-)

Preface

Since March 2023 there’s a handy bot called @SmallTownUSA@en.osm.town on the Fediverse, that posts a daily “task” out to its followers, highlighting “small towns” in the United States of America, that, to use the same phrasing, “seem like they could use some mapping”, along with a screenshot of the (Carto) map of the area.

Sparing the full technical details, the program randomly picks an entry from an overpass export of all nodes that have any place tag and a population tag with a value of less than 1000 (hence the name “SmallTown”) in the USA (currently 11389), asks overpass whether there are 10 buildings or less in 800 meters around that node and if so, posts about it.

As the bot has been doing this for about 10 months now, there seem to be at least 300 places that match the above description. Naturally, I wondered, how Europe would fare in comparison. Chatting with the bot’s developer, Matthew Wildon (OSM, Fediverse), they told me, that they checked France for potential candidates and found none and then didn’t look into it any further.

What’s on on the Continent?

My interest was now piqued. Would a similar “SmallTownEurope” bot make any sense, or would it run out of things to post within a week? Or is Pascal Neis’ Unmapped Places of OpenStreetMap Result Map enough to find areas where mapping is needed?

Suffice it to say, I wouldn’t know the answer to that from purely looking at the map, I needed data. So I quickly got myself the above-mentioned Overpass Exports for Areas in Europe - where I was using the English Wikipedia definition of what’s in Europe: “[…] countries falling even partially under any common geographical or political definitions of Europe”. To cover most of the area of those countries, I also checked the “de facto independent countries with limited to no international recognition” (Abkhazia, Kosovo, Northern Cyprus, South Ossetia, Transnistria) as well as “dependencies and similar territories with broad autonomy [that] are also found within or close to Europe” (Aland, Guernsey, Jersey, Faroe Islands, Gibraltar, Greenland, Isle of Man, Svalbard).

Way to go!

263845 nodes fall into the definition of this survey (any place tag and a population tag with a value of less than 1000 and more than 0). I then asked poor Overpass for each one how many buildings there were in an 800-meter radius. That took about 3 days 😅

The beauty of knowing the buildings count for every node now is that we can easily adjust the parameters we’re looking at. When looking through the matches in Germany I found that some very well-mapped places could be considered “false positives”. 5 people living in a dwelling with 7 buildings is a very reasonable idea, but matches the “less or equal than 10 buildings” rule, so to get to my final numbers I added the filter “more people living there than buildings mapped”.

In the end, I arrived at 89338 matching towns (or 33.86%) with the 4 highest countries (Spain, Russia, Ukraine, Romania) making up 82.1845% of those matches (73422).

4 countries do not have a single match: Belgium, Malta, Norway and Switzerland

You want numbers, you say?

Alright then.

Table 1: Ordered by total matches in a country

Rank	Country	Matches	Total	Percent
1	Estonia	1	256	0.3906
1	Iceland	1	80	1.2500
1	Ireland	1	46	2.1739
1	Netherlands	1	1410	0.0709
1	Slovakia	1	1863	0.0537
6	Greenland	2	22	9.0909
7	Latvia	7	97	7.2165
8	France	10	9484	0.1054
9	Northern Cyprus	11	15	73.3333
10	Cyprus	19	55	34.5455
11	Germany	26	9991	0.2602
12	Finland	29	177	16.3842
13	Austria	31	8955	0.3462
14	North Macedonia	32	39	82.0513
15	Denmark	38	3798	1.0005
15	Hungary	38	1915	1.9843
17	Moldova	44	240	18.3333
18	Kosovo	49	95	51.5789
19	Albania	59	102	57.8431
20	Czech Republic	68	12895	0.5273
21	Montenegro	69	197	35.0254
22	United Kingdom	77	710	10.8451
23	Azerbaijan	97	109	88.9908
24	Poland	101	9586	1.0536
25	Slovenia	106	1695	6.2537
26	Bosnia and Herzegovina	115	434	26.4977
27	Armenia	133	527	25.2372
28	Sweden	223	1230	18.1301
29	Bulgaria	304	598	50.8361
30	Kazakhstan	351	479	73.2777
31	Belarus	395	4810	8.2121
32	Turkey	598	972	61.5226
33	Croatia	810	1572	51.5267
34	Georgia	883	1462	60.3967
35	Portugal	952	1562	60.9475
36	Lithuania	2328	16685	13.9527
37	Greece	2365	3156	74.9366
38	Italy	2597	16800	15.4583
39	Serbia	2953	3871	76.2852
40	Romania	7940	9840	80.6911
41	Ukraine	*	*	*
42	Russia	20112	48180	41.7435
43	Spain	28983	52081	55.6499

Table 2: Ordered by percent of towns matched

Rank	Country	Matches	Total	Percent
1	Slovakia	1	1863	0.0537
2	Netherlands	1	1410	0.0709
3	France	10	9484	0.1054
4	Germany	26	9991	0.2602
5	Austria	31	8955	0.3462
6	Estonia	1	256	0.3906
7	Czech Republic	68	12895	0.5273
8	Denmark	38	3798	1.0005
9	Poland	101	9586	1.0536
10	Iceland	1	80	1.2500
11	Hungary	38	1915	1.9843
12	Ireland	1	46	2.1739
13	Slovenia	106	1695	6.2537
14	Latvia	7	97	7.2165
15	Belarus	395	4810	8.2121
16	Greenland	2	22	9.0909
17	United Kingdom	77	710	10.8451
18	Lithuania	2328	16685	13.9527
19	Italy	2597	16800	15.4583
20	Finland	29	177	16.3842
21	Sweden	223	1230	18.1301
22	Moldova	44	240	18.3333
23	Armenia	133	527	25.2372
24	Bosnia and Herzegovina	115	434	26.4977
25	Cyprus	19	55	34.5455
26	Montenegro	69	197	35.0254
27	Russia	20112	48180	41.7435
28	Bulgaria	304	598	50.8361
29	Croatia	810	1572	51.5267
30	Kosovo	49	95	51.5789
31	Spain	28983	52081	55.6499
32	Albania	59	102	57.8431
33	Georgia	883	1462	60.3967
34	Portugal	952	1562	60.9475
35	Turkey	598	972	61.5226
36	Ukraine	*	*	*
37	Kazakhstan	351	479	73.2777
38	Northern Cyprus	11	15	73.3333
39	Greece	2365	3156	74.9366
40	Serbia	2953	3871	76.2852
41	Romania	7940	9840	80.6911
42	North Macedonia	32	39	82.0513
43	Azerbaijan	97	109	88.9908

^* There shall be no mapping in Ukraine for now

Where do we go from here?

I think it’s clear from the numbers, that working on a single match per day isn’t feasible. That would take 244½ years, and we can’t wait that long, can we?

There is now a MapRoulette Project for all matches except Ukraine and Russia.

If you have any other ideas or questions, let me know in the comments.

Have a lovely day.

Discussion

Comment from Friendly_Ghost on 1 February 2024 at 23:39

Awesome data!

If you can update this list on a regular basis, I’m sure some mappers will gladly hunt down cases to “complete” countries. This approach worked very well in the river modernization project.

Comment from chris_debian on 2 February 2024 at 00:18

Hi, K.

This is really good work, and I think the MapRoulette idea is a good one. I’d give the UK a good go, then look at other priorities.

As an aside, the figures for Ukraine may be a false positive, and I believe all map amendments are paused, until the invasion is resolved. This may have changed, but worth checking, before investing any effort.

Well done,

Chris

Comment from watmildon on 2 February 2024 at 05:21

I know it’s very common for new mappers to look at their local area (usually very urban) and wonder “so what’s left”? I originally wanted to help folks find places off the beaten path and I’m so happy to see you’ve found lots of good mapping opportunities!

I am very excited to hear how other ways of presenting this information help drive engagement and help us get to a more complete map.

Comment from kmpoppe on 2 February 2024 at 08:21

@chris_debian, here you go: https://mpr.lt/p/55150 - now also linked in the entry itself. Have fun!

Comment from chris_debian on 2 February 2024 at 11:14

@kmpoppe Hi, just had a go at a settlement in the UK. Could you consider:

Amending the comment that is auto-generated on MapRoulette, so it is more meaningful. At the moment, it looks very automated, and could cause concern from reviewers. I added the additional text “Added buildings to settlement” to my commit.
It’s difficult to know when the task is ‘complete enough’, so when saving the edits, I paused when I was deciding to click on the options. The number of houses/ buildings that can be added, could be huge (many hundreds+ ?).
Related to the last point, and to avoid change sets that are huge, can tasks be split into smaller chunks, that can be completed? I tend to limit my change sets to 50 ‘commits’. Even using AI to assist in the edits (https://rapideditor.org/) the tasks could be potentially endless, and huge. People contributing, will soon lose interest, if each task seems endless.

This is a really good initiative, and my comments are meant to make this work even better, and are definitely not a criticism.

Many thanks,

Chris.

Comment from mmd on 3 February 2024 at 11:56

That took about 3 days 😅

I believe one reason for this long runtime could be your pre-filtering on certain areas (not exactly sure if this how you’ve done it). When using https://overpass-turbo.eu/s/1GQm to analyze all place nodes on a global scale, the runtime should be less than 1.5 hours. It returns about 401k place nodes that would need some additional filtering by location as a post-processing step.

If you like to try this out, I have uploaded the query result here: https://dev.overpass-api.de/misc/test_buildings.xml.gz (file size: 21M)

Comment from osmuser63783 on 14 February 2024 at 18:23

Really interesting analysis, thanks!

That took about 3 days 😅

This sort of thing can be done with Geodesk in a few minutes, though it requires first downloading a planet file or extract and then preprocessing it once (this step takes about half an hour on an SSD for the full planet).

Comment from Skinfaxi on 20 February 2024 at 08:16

Ibwould consider an other aprove: How about a bot that publishes Major approves in small Towns? Together with a “fault-list” it will adress the topic - but in a positiv way of communication.

After 5 years of intensiv mapping i recently lost my Motivation.

I Work in a remote aera and are quiet a bot frustrated about that att discussions tend to adress micromapping Problems instead of focusing on the big picture.

And in the rare cases that i get some Feedback ist cames from “topic-mappers”. They suggest för example that i shall give Details Like the building Material of the benches that i contributed. And they completly ignore the Fakt that the whole area was Not more than “Landuse” totaly without any Infrastruktur.

So i Like to See same “good News” giving me the Feeling that i am Not alone with my idear of making a map instead of a random collection of more or less intressting Details.

In conclution i think ITS worth to try another communication aproach.