Thanks for the link Graeme, I look forward to reading it. Re the comparison between the new modelled data and the NSW road data in the treetracks blog… virtually every road in NSW, except for metro Sydney, has a surface tag in OSM, so the new study should give the same results as the blog since the new study seems to model surfaces based on OSM data (based on my super quick skim of the abstract). It’ll be even more useful in other Aus states where not so many road surfaces are mapped. We can probably expect to get good paved vs unpaved surface data on all Aus roads down the track based on either AI modelling or user interpretations of imagery. Cheers Ian
This is a breakdown of the highways without a surface tag (as at 2025-08-20). I have excluded track and service where there is a service tag. Overall surface tagging is 92% complete.
Highway=
Length (km)
Fraction of highways of this type untagged
Paved
residential
29,943
21%
85%
service
23,519
37%
20%
unclassified
23,019
5%
14%
tertiary
1,935
1%
54%
secondary
458
0%
76%
trunk
424
1%
98%
motorway_link
416
21%
100%
primary
349
1%
90%
motorway
182
2%
100%
living_street
172
28%
95%
trunk_link
130
23%
100%
primary_link
81
20%
100%
pedestrian
59
34%
91%
tertiary_link
57
27%
91%
secondary_link
52
18%
97%
EDIT: I have added a column showing the paved fraction of the highway (filtered as above) that have a surface tag. This is based on tags that cover 99.98% of values. Overall paved to unpaved is about 45:55.
OK maybe my table was a little ambitious. There is 424 km of trunk without a surface tag. This is 1% of the total length of trunk roads (about 42,000 km).
