How do we measure OpenStreetMap’s data coverage? One approach is to compare it against established references like the CIA World Factbook. As a US Government work, the Factbook is publicly available and freely licensed, and it contains measurements of paved & unpaved road lengths in every country in the world.
We downloaded the Factbook’s road length numbers and compared them against OpenStreetMap’s road data using Tile Reduce and OSM QA Tiles. We found for the United States about 6.2 million miles of motor roads and 7.1 million miles of any type of road or path. OpenStreetMap performed well across all the continents, but still needs work in places like India and China.
One caveat is that the Factbook doesn’t clearly define what constitutes a road, so we’ve provided numbers for both roads in general and only those fit for motor vehicle traffic.
Country metrics page showing OpenStreetMap coverage.
Check out the metrics below. Tim and I would love to talk more about measuring OpenStreetMap coverage on twitter, or in person today at Geography 2050.
Country | OpenStreetMap all roads and paths | OpenStreetMap motor roads only | CIA factbook |
---|---|---|---|
Afghanistan | 67,845 | 25,400 | 26,191 |
Argentina | 389,569 | 370,007 | 143,769 |
Australia | 631,146 | 487,207 | 511,523 |
China | 977,075 | 935,100 | 2,551,590 |
Egypt | 61,943 | 55,397 | 85,395 |
Germany | 1,341,791 | 554,803 | 400,784 |
Indonesia | 181,731 | 169,230 | 308,577 |
Japan | 976,655 | 828,283 | 756,288 |
Kenya | 58,628 | 47,813 | 99,965 |
Mexico | 335,719 | 308,266 | 234,667 |
South Africa | 233,758 | 207,005 | 464,173 |
Thailand | 193,083 | 164,660 | 111,880 |
United Kingdom | 489,640 | 320,376 | 245,086 |
United States | 7,147,209 | 6,220,189 | 4,092,728 |
All comparisons on a zoom level 12 tile basis where one tile is either counted fully for a country or not. This leads to a margin of error that is negligible for large countries but can be significant for smaller countries.
View our searchable metrics page, or browse the full list.