Because this is the area of my local knowledge and I prefer talking about things that I can relate to. And why not? When a key applies globally, should regions matter? Unless it applies not the same everywhere, of course.
We do see that in the data. Below again reverse view, the top three each. This view is immune against skewing the result through size of sample set.
| tracktype | surface | ratio |
|---|---|---|
| grade1 | asphalt | 82.90% |
| paved | 4.10% | |
| concrete | 2.50% | |
| grade2 | gravel | 55.00% |
| compacted | 15.10% | |
| fine_gravel | 7.50% | |
| grade3 | ground | 44.00% |
| gravel | 20.80% | |
| grass | 7.00% | |
| grade4 | ground | 44.90% |
| grass | 23.80% | |
| gravel | 9.50% | |
| grade5 | ground | 41.60% |
| grass | 40.00% | |
| dirt | 6.40% |
I am a bit puzzled by the large numbers of “mostly firm grass” here - Or does it come from the centre strip of tractor tracks? The table shows some other properties of the data that implies that documentation progresses more rapidly than mappings. BTW: @ftrebien Which Wiki are you looking at?
PS: From an even smaller area of my local knowledge, by the end of 2014 most tracks were already mapped and most of them already had a tracktype assigned. The plotted values read from top to bottom: grade 5, 1, 4, 2, 3, remainder. Source ohsome dashboard.
