I should point out that the Microsoft Open Maps team did a systematic comparison between OSM and government open data specifically to ensure OSM isn't missing roads or road names. Which did help fill in any gaps that organic mappers may have missed. However that systematic comparison was still constrained by which states make their data open data, so didn't cover all states equally.
On Tue, 7 Jul 2020 at 23:37, Andrew Hughes <[email protected]> wrote: > Hi All, > > Thanks for your responses. I'd love to speak more about our scenario but > I'd end up writing a 10 page email. I'm not joking. Maybe theres a virtual > meetup or something like that? > > Continuing with the thread of conversation... > > When I say "correlation", we really like to know that a satisfactory > amount of (Govt) gazetted roads exist is OSM. If we could identify missing > roads that would be ideal. > > Everyones responses have been very insightful. It would seem possible to > geographically align (albeit temporarily) OSM and gazetted roads, but only > practiable with "simple" roads. The granularity I encountered comparing > Govt and OSM data were worlds apart in populated areas where roads are more > complex (which is exactly what Andrew said). Expecting OSM to align with > Gazetted roads would be an exercise in dramatically downgrading OSM data. > > A "map match algorithm" might be helpful but only to a point. > > I don't believe this is a blocker for OSM within the NHVR. It simply means > we have to approach this from another angle. > > Thanks for everyones responses while I get up to speed. > > Andrew Hughes > > > > On Tue, 7 Jul. 2020, 7:31 pm Andrew Harvey, <[email protected]> > wrote: > >> Hi Andrew, >> >> Just to add what others have said already, >> >> > OSM adoption is largely dependent on a minimum correlation between the >> OSM ways and the streets found in Government centerline/road datasets >> (States and/or LGAs) >> >> Could you elaborate on what exactly you mean by that or what specifically >> you're aiming to achieve? >> >> For example could you use a map matching algorithm to match OSM data to >> other road network data but since there will always be differences between >> multiple datasets that's never going to be perfect. >> >> Graeme mentioned on/off ramps, but more generally there could be >> differences in how the centerline is represented. For example, OSM splits >> ways where there is a physical separation, so you might find other datasets >> just have one centerline, but OSM would have two parallel ways for each >> direction. The transition between these two modes is never ideal either. >> This isn't saying one is necessarily wrong but rather have different >> guidelines on where the line should be drawn. >> >> The other consideration is that most states we only have access to GPS or >> satellite imagery to derive road geometries from so the geometry accuracy >> in OSM isn't necessarily as great as survey grade data. >> >> On Mon, 6 Jul 2020 at 12:00, Andrew Hughes <[email protected]> wrote: >> >>> Hi All, >>> >>> First time poster and very new to OSM so please feel free to throw >>> anything at me you think I should educate myself on. >>> >>> I'm currently the GIS Lead at the National Heavy Vehicle Regulator >>> (HNVR). We're very serious about adopting OSM for some of our needs. >>> However, our OSM adoption is largely dependent on a minimum correlation >>> between the OSM ways and the streets found in Government centerline/road >>> datasets (States and/or LGAs). >>> >>> Q: Would anyone be able to provide me with some insight as to what we >>> might expect when looking to achieve the correlation we need? Please be >>> aware, our intent is to contribute and "close the gap" but we need to know >>> if/how this can best be done in a cohesive way within the OSM community. >>> *I'm >>> also aware there may be licensing issues, please overlook these for now.* >>> >>> The NHVR are quite serious about what it hopes to achieve in the next 12 >>> -24 months through GIS and we are very enthusiastic to learn and contribute >>> to OSM. I hope to be speaking with you a lot more in the near future. >>> >>> Thanks in advance, >>> Andrew Hughes >>> >>> _______________________________________________ >>> Talk-au mailing list >>> [email protected] >>> https://lists.openstreetmap.org/listinfo/talk-au >>> >>
_______________________________________________ Talk-au mailing list [email protected] https://lists.openstreetmap.org/listinfo/talk-au
