I do a lot of mapping in Alaska and the quality of the Tiger roads in rural Alaska is simply horrendous. For some of the small communities I've worked on the Tiger ways are so far out of position I simply delete them and start fresh. There is no way to know where they should actually be. The main reason to keep them and why I sometimes make a guess about their positions is to preserve the street names. I'm never going to be able to visit these tiny communities because they are remote and exceedingly difficult to reach. Nobody can ever drive to them, there will probably never be any OSMers living in a town like Alakanuk, for example, nor is any resident likely to use the OSM data but it looks exceedingly bad to leave those ways as they are. So I often try to fix them. It's a tedious job at best. Along with that, there are the ugly coastlines to fix everywhere as well. Take a look at Alakanuk here (62.684314, -164.652965) to get an idea of what Alaska OSM mappers are up against.
Anyway, my point is that most Tiger data I've encountered is so bad I wish it wasn't there at all. I would rather sketch in the roads (roads that I and 99.9999% of data consumers will never see or use), and toss the Tiger data completely. In my example town, I can make a reasonable guess about the position of the roads but it's a ton of work, especially when you realize that _all_ of Alaska's rural towns have data like this or, more often, worse. On Tue, Feb 13, 2018 at 8:23 AM, Brian May <b...@mapwise.com> wrote: > I have spent a very large amount of time cleaning up TIGER in rural areas > of Florida. I agree with others that the vast majority of untouched TIGER > ways in un-populated rural areas classified as residential are forest roads > for logging trucks at best and pure fantasy at worst, with tons of barely > grass paths all over the place. Many of these roads are on private lands > that you can't (or shouldn't) access anyway. Spatially accuracy is often > horrific. I've reviewed a decent amount of 2017 TIGER and many areas have > not been fixed. The best secondary source of data I have found are county > streets from the county GIS departments, but those vary widely in quality. > At the very least the county data shows you where all the forest tracks, > farms tracks and imaginary TIGER streets are, because they are not there in > the county data sets. Many times, a residential street having no name is a > strong tip as well that it is not a residential street. > > Kevin, I hear where you are coming from, but I think your case is somewhat > unique. Most people aren't going to look at a GPS with OSM data in it, see > a bunch of residential roads in a rural un-populated area and think, OK, > that must be unedited TIGER, but I know there's a few navigable roads in > there somewhere, I just need to find them, record what I found and make > some OSM edits. If they know the area, they are going to think this data is > junk. If they don't know the area and they head into it they will then > figure out pretty quickly the data is junk. I agree with others that these > roads should probably not be in OSM at all - let the locals add the real > roads and tracks. But we are living with the old TIGER, and there is some > potential usefulness that can come from it. So as others have said, we are > willing to leave them there, downgrade them to track without a grade > assigned for now, maybe make some spatial corrections, delete roads that > are obviously pure fantasy, etc. > > I don't think there should be any requirement to cover a certain size area > when reviewing these areas. We need to be thankful that someone has taken > the time to look at even a small area of rural areas that don't get much > attention normally at all, especially private lands. > > Brian > > On 2/12/2018 6:02 PM, Kevin Broderick wrote: > > If you can cover an entire area (which I'd define as a swath between the > nearest state highways), I agree that downgrading to track absent other > clues is one reasonable solution. One of my key points is that anyone who's > spent a fair bit of time trying to use GPS maps (of any origin) in > poorly-mapped areas will quickly recognize an area that is clearly an > unverified TIGER import, which signals both (a) that the data is clearly > questionable and (b) that it might be an interesting place to explore to > find out if the roads do go through or not. The questionable map data can > be very useful, especially in conjunction with other data sources, in > attempting to piece together a route through an area that lacks fully > maintained roadways. If a track doesn't actually exist, yes, then it should > certainly be deleted, but I've ridden right-of-ways that were damn near > impossible to see with leaf-on imagery and also found other routes that > looked more road-like via the same imagery impassable, so I definitely > wouldn't delete anything unless you can get there in person and look for > evidence of a roadway, perhaps one that hasn't been maintained in decades > (e.g. Class IV roads in Vermont and Class VI roads in New Hampshire). > > Downgrading some ways to tracks without doing so to a whole localized > network creates the appearance of a higher level of data accuracy than > actually exists, which IMO is more likely to bite someone in the ass than > having a localized network of roads that are mislabeled. I know it would > make some of the exploring I've done via on/off-road motorcycle more > difficult. > > I'd also suggest that leaving tiger:reviewed at no is appropriate if you > haven't been able to travel the road/track in question and determine > whether it is really an unclassified road or a track, so it remains flagged > for further review if someone has the time and proximity to do so. > > On Mon, Feb 12, 2018 at 3:39 PM, Martijn van Exel <m...@rtijn.org> wrote: > >> I am very happy to see this rekindled interest in TIGER cleanup! >> >> Having done a fair amount of backcountry exploring, I know that there is >> a wide range of road grades and aerial imagery alone is not enough to >> decide how navigable a roads is for a particular type of vehicle. Or, for >> that matter, what its access limitations are. I do agree with Clifford that >> leaving them as poorly aligned 'residential' roads is the worst possible >> situation. Yes, worse than deleting the road altogether. What I usually do >> is mark the road as track without a track grade tag. This seems to me to be >> the most acceptable generic solution for a remote mapper: acknowledging >> that something that could potentially be navigated by a 4 wheeled vehicle >> exists, without being more specific. Local knowledge can then come to the >> rescue to upgrade to unclassified if appropriate. >> >> Another note on the MapRoulette side of things: I would very much >> appreciate your feedback on the new MapRoulette version Clifford linked to. >> Just email me, join #maproulette on slack, or file an issue at >> https://github.com/maproulette/maproulette3/issues. >> >> Martijn >> >> On Mon, Feb 12, 2018 at 3:08 PM, Kevin Kenny <kevin.b.kenny+...@gmail.com >> > wrote: >> >>> On Mon, Feb 12, 2018 at 12:55 PM, Kevin Broderick < >>> k...@kevinbroderick.com> wrote: >>> >>>> Please, please, please don't convert rural roads to tracks based on >>>> imagery alone unless it's incredibly clear (and that would exclude anything >>>> with forest cover). >>>> >>>> While many of them should definitely be unclassified, not residential, >>>> downgrading the main rural routes to tracks doesn't match local usage nor >>>> the functional topology of the road network in such places. There are a lot >>>> of USFS and BLM roads around here that are the only way to access >>>> significant areas, that commonly see normal passenger-car traffic and that >>>> can be traveled at reasonable speed in a sedan (or at 30+ MPH with a little >>>> ground clearance and driving skill),. Having these differentiated from true >>>> tracks (where even a stock 4x4 is likely going to be operating at 15 MPH or >>>> less) is incredibly helpful for routing and visual use of the map, and it's >>>> a lot easier to recognize what I'd call "areas of questionable data" when >>>> they haven't been aggressively armchair-mapped. Also, the smoothness key is >>>> really helpful for tracks and impossible to map from orthoimagery. >>>> >>>> >>> Yes, yes, yes. >>> >>> In the rural areas that I can travel to readily, TIGER is downright >>> hallucinatory (and there are few enough mappers that cleanup has been >>> agonizingly slow). TIGER has roads in places where no road is, ever was, or >>> even ever could be. (I've seen one going up a series of cliffs totalling >>> about 2000 feet of ascent!) But even in 'leaves down' images, it's nearly >>> impossible to see the forest roads, much less trace them, and there is >>> definitely a wide variation in quality. Some of them are well-compacted >>> sand and shale, that once they've been rolled in the spring, support >>> driving at 30+ MPH. Others, I wouldn't bring my Subaru on. (Although I've >>> been on a few of those in the ancient Ford Explorrer that the Subaru >>> replaced.) Some are gated, some, you simply have to decide for yourself >>> that they're not drivable. >>> >>> The 'dirt roads' range from 'highway=path abandoned:highway=track >>> smoothness=impassable' to 'highway=tertiary surface=compacted >>> smoothness=intermediate', with no way for an armchair mapper to tell among >>> them. >>> >>> The old road maps that they used to give out at gas stations had, on >>> many of these roads, "inquire locally for conditions," which is still good >>> advice. The signage may say, "LIMITED PURPOSE SEASONAL-USE ROAD: No >>> maintenance November 1-April 15" - but in practice, they'll keep it open >>> later in the Autumn unless the snow comes early, and when they open it in >>> the spring depends on when the crews can get it clear - it could be weeks >>> late if there's been a bad washout or rock slide. There's absolutely no way >>> to tag and encode that sort of thing. Inquire locally for conditions. >>> >>> >>> >>> _______________________________________________ >>> Talk-us mailing list >>> Talkfirstname.lastname@example.org >>> https://lists.openstreetmap.org/listinfo/talk-us >>> >>> >> > > > -- > Kevin Broderick > k...@kevinbroderick.com > > > _______________________________________________ > Talk-us mailing > listTalkemail@example.com://lists.openstreetmap.org/listinfo/talk-us > > > > _______________________________________________ > Talk-us mailing list > Talkfirstname.lastname@example.org > https://lists.openstreetmap.org/listinfo/talk-us > > -- Dave Swarthout Homer, Alaska Chiang Mai, Thailand Travel Blog at http://dswarthout.blogspot.com
_______________________________________________ Talk-us mailing list Talkemail@example.com https://lists.openstreetmap.org/listinfo/talk-us