Re: [Talk-us] Systematic errors in the Maui, Hawaii Data
Since the roads are so concentrated along the coast, it seems fairly simple to load a few hundred ways at a time into an editor to shift them. Not more than a few hours work, surely. Lanai seems to suffer from the same error, so I experimented. Selecting only those with tiger:reviewed=no, there are just 330 ways. Using the street grid in Lanai City as a guide, I dragged them into position. The result is not perfect, but no worse than is normal for Tiger data... /Stellan ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] Systematic errors in the Maui, Hawaii Data
On Wed, Sep 30, 2009 at 8:12 AM, Stellan Lagerstrom lagerst...@blindsight.com wrote: Since the roads are so concentrated along the coast, it seems fairly simple to load a few hundred ways at a time into an editor to shift them. Not more than a few hours work, surely. Lanai seems to suffer from the same error, so I experimented. Selecting only those with tiger:reviewed=no, there are just 330 ways. Using the street grid in Lanai City as a guide, I dragged them into position. The result is not perfect, but no worse than is normal for Tiger data... My strong suspicion is that this widespread displacement has to do with the pre-2008 TIGER data for Hawaii being in a different datum than NAD83, which is what was used for CONUS. I'm not sure there's much that can be done except a reimport (preferably from 2008 TIGER, which is NAD83) to ensure better positional accuracy... The same issue may also affect the Northern Mariana Islands, Guam, and American Samoa which were also non-NAD83 before 2008. Chris ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] Systematic errors in the Maui, Hawaii Data
On Wed, Sep 30, 2009 at 7:17 AM, Chris Lawrence lordsu...@gmail.com wrote: My strong suspicion is that this widespread displacement has to do with the pre-2008 TIGER data for Hawaii being in a different datum than NAD83, which is what was used for CONUS. I'm not sure there's much that can be done except a reimport (preferably from 2008 TIGER, which is NAD83) to ensure better positional accuracy... The same issue may also affect the Northern Mariana Islands, Guam, and American Samoa which were also non-NAD83 before 2008. I spot-checked Oahu, and it looks like it may have already been corrected, manually or otherwise. That isn't too surprising, given the much larger population of this island. I'm still relatively new to the OSM project. What is the best way to proceed to correct the datum errors for these islands? Do we need to achieve some sort of consensus on the best technical solution first? (e.g. manually moving ways in JOSM vs. reimporting) -Scott -- Scott Atwood Cycle tracks will abound in Utopia. ~H.G. Wells ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
[Talk-us] Systematic errors in the Maui, Hawaii Data
I recently returned from a trip to Maui and Big Island in Hawaii in which I collected a large quantity of GPS tracklogs. I was hoping to use this data to tune up the existing maps for Hawaii. However, as nearly as I can tell, very little work seems to have been done on Maui beyond the initial Tiger upload. The problem is, this Tiger data seems to have some large systematic errors, a translation to the north-northwest. I started to manually correct a few ways based on the satellite photos and GPS tracks, but I realized it would probably be far better to either reimport the Maui data with a correction, or apply some automated process to the existing data to translate it to more-or-less the correct location. Does anyone know what needs to be done to accomplish this? -Scott -- Scott Atwood Cycle tracks will abound in Utopia. ~H.G. Wells ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] Systematic errors in the Maui, Hawaii Data
On Mon, 2009-09-28 at 23:23 -0700, Scott Atwood wrote: However, as nearly as I can tell, very little work seems to have been done on Maui beyond the initial Tiger upload. The problem is, this Tiger data seems to have some large systematic errors, a translation to the north-northwest. I started to manually correct a few ways based on the satellite photos and GPS tracks, but I realized it would probably be far better to either reimport the Maui data with a correction, or apply some automated process to the existing data to translate it to more-or-less the correct location. I'd really prefer not to reimport it. Could you point us to a few areas where we can look at your tracks and the existing data together that exemplify this behavior? Can you quantify large systematic errors? I really don't want to blow it away and re-upload. -- Dave ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] Systematic errors in the Maui, Hawaii Data
Here's a good example from the Hana Highway: http://www.openstreetmap.org/?lat=20.86962lon=-156.17746zoom=17layers=B000FTF There is a distinctive set of W-shaped hairpin turns at this location. You can see that the OSM ways are displaced approximately 500 meters to the north-northwest of both the satellite picture and GPS tracklogs. Here is another example near the intersection of Hana Highway and Baldwin Ave. in Paia (My GPS tracklog includes a brief excursion onto Baldwin Ave. at this intersection): http://www.openstreetmap.org/?lat=20.91837lon=-156.38279zoom=16layers=B000FTF The GPS tracklog I uploaded includes data from Wailuku in central Maui, all along the Hana Highway as far as Waianapana State Park near Hana. In all the locations I spot checked (including the ones mentioned above), there appears to be a consistent displacement of approximately 500 meters to the north-northwest of all ways associated with streets. The shoreline ways appear to be more or less accurate. -Scott On Tue, Sep 29, 2009 at 9:41 AM, Dave Hansen d...@sr71.net wrote: On Mon, 2009-09-28 at 23:23 -0700, Scott Atwood wrote: However, as nearly as I can tell, very little work seems to have been done on Maui beyond the initial Tiger upload. The problem is, this Tiger data seems to have some large systematic errors, a translation to the north-northwest. I started to manually correct a few ways based on the satellite photos and GPS tracks, but I realized it would probably be far better to either reimport the Maui data with a correction, or apply some automated process to the existing data to translate it to more-or-less the correct location. I'd really prefer not to reimport it. Could you point us to a few areas where we can look at your tracks and the existing data together that exemplify this behavior? Can you quantify large systematic errors? I really don't want to blow it away and re-upload. -- Dave -- Scott Atwood Cycle tracks will abound in Utopia. ~H.G. Wells ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] Systematic errors in the Maui, Hawaii Data
On Tue, 2009-09-29 at 10:40 -0700, Scott Atwood wrote: The GPS tracklog I uploaded includes data from Wailuku in central Maui, all along the Hana Highway as far as Waianapana State Park near Hana. In all the locations I spot checked (including the ones mentioned above), there appears to be a consistent displacement of approximately 500 meters to the north-northwest of all ways associated with streets. The shoreline ways appear to be more or less accurate. It was pretty easy using JOSM to fix the first example. The second one appears a little less clear cut to me. My best suggestion at this point would be to give JOSM a try. It lets you grab much more interesting sets of data and move them around as a group. I'm a bit hesitant to go moving large swaths of data around in an automated way. First, I don't have any good tools to do it and it'll take some time to code them up. Also, I'm unsure about exactly how widespread and precise the 500m offset is. Does anybody know of any better tools that might help with this? -- Dave ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] Systematic errors in the Maui, Hawaii Data
On Tue, Sep 29, 2009 at 11:04 AM, Dave Hansen d...@sr71.net wrote: On Tue, 2009-09-29 at 10:40 -0700, Scott Atwood wrote: The GPS tracklog I uploaded includes data from Wailuku in central Maui, all along the Hana Highway as far as Waianapana State Park near Hana. In all the locations I spot checked (including the ones mentioned above), there appears to be a consistent displacement of approximately 500 meters to the north-northwest of all ways associated with streets. The shoreline ways appear to be more or less accurate. It was pretty easy using JOSM to fix the first example. The second one appears a little less clear cut to me. I started out trying to fix things in a piecemeal fashion, but translating ways one at a time, but I eventually gave up and reverted my changes, because I noticed that the continuation ways of Hana Highway and all the adjoining ways seem to be affected by the same ~500 meter NNW displacement, and the magnitude of the manual adjustments was overwhelming. The second example may appear less clear cut, because in additional to the systematic displacement to the NNW, the area around Paia seems to also be suffering the typical sorts of errors found in Tiger data: streets that are displaced or misshaped by random amounts. So even if you translated the intersection of Hana Highway and Baldwin Ave, the rest of the streets of Paia won't necessarily line up cleanly without additional manual effort. My best suggestion at this point would be to give JOSM a try. It lets you grab much more interesting sets of data and move them around as a group. I'm a bit hesitant to go moving large swaths of data around in an automated way. First, I don't have any good tools to do it and it'll take some time to code them up. Also, I'm unsure about exactly how widespread and precise the 500m offset is. I spot checked a number of locations all around Maui, and in all the locations I checked, they seem to be affected by the same or highly similar offset to the NNW. Here is another example from the Lahaina area: http://www.openstreetmap.org/?lat=20.88606lon=-156.68522zoom=15layers=B000FTF The prospect of manually shifting every single street on the entire island of Maui (even in batches within JOSM) is extremely daunting, and strongly inhibits me from wanting to do any work on Maui. And the fact that such enormous errors persist suggests to me that few if any people have done much non-automated work on Maui yet. Can we verify how much editing has been done on Maui? If it is as little as I think, there is little to loose by throwing away the existing data and starting over. I still think re-importing (or some kind of automated batch editing) might be a faster and easier way to fix the problems with Maui. -Scott -- Scott Atwood Cycle tracks will abound in Utopia. ~H.G. Wells ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] Systematic errors in the Maui, Hawaii Data
On Tue, 2009-09-29 at 11:20 -0700, Scott Atwood wrote: I spot checked a number of locations all around Maui, and in all the locations I checked, they seem to be affected by the same or highly similar offset to the NNW. Here is another example from the Lahaina area: http://www.openstreetmap.org/?lat=20.88606lon=-156.68522zoom=15layers=B000FTF If you can look into formalizing this, I'd be willing to take a look at it. What I mean is to take some precise measurements from a wide set of samples across the island, probably 20 or 30. Figure out what the displacement vector is at each point, and ensure that it *is* consistent. The worst thing we could do would be to perform such an edit and later realize that we've hurt more than we've helped. I'm just really hesitant to go moving an entire island worth of data based on the problem I've seen so far. The prospect of manually shifting every single street on the entire island of Maui (even in batches within JOSM) is extremely daunting, and strongly inhibits me from wanting to do any work on Maui. And the fact that such enormous errors persist suggests to me that few if any people have done much non-automated work on Maui yet. The question is whether it will take longer for people to do it manually or if doing it in an automated way will be more efficient. Believe me, the prospect of doing a mass edit like that is pretty daunting as well. Can we verify how much editing has been done on Maui? If it is as little as I think, there is little to loose by throwing away the existing data and starting over. I still think re-importing (or some kind of automated batch editing) might be a faster and easier way to fix the problems with Maui. Sure. Just download the entire island and see how many objects have been touched by people other than DaveHansen. JOSM can tell you this pretty quickly. -- Dave ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] Systematic errors in the Maui, Hawaii Data
Hawaii (Big Island) seems to be affected by the same or similar displacement: http://www.openstreetmap.org/?lat=19.09422lon=-155.78022zoom=15layers=B000FTF -Scott On Tue, Sep 29, 2009 at 11:42 AM, Dave Hansen d...@sr71.net wrote: On Tue, 2009-09-29 at 11:20 -0700, Scott Atwood wrote: I spot checked a number of locations all around Maui, and in all the locations I checked, they seem to be affected by the same or highly similar offset to the NNW. Here is another example from the Lahaina area: http://www.openstreetmap.org/?lat=20.88606lon=-156.68522zoom=15layers=B000FTF If you can look into formalizing this, I'd be willing to take a look at it. What I mean is to take some precise measurements from a wide set of samples across the island, probably 20 or 30. Figure out what the displacement vector is at each point, and ensure that it *is* consistent. The worst thing we could do would be to perform such an edit and later realize that we've hurt more than we've helped. I'm just really hesitant to go moving an entire island worth of data based on the problem I've seen so far. The prospect of manually shifting every single street on the entire island of Maui (even in batches within JOSM) is extremely daunting, and strongly inhibits me from wanting to do any work on Maui. And the fact that such enormous errors persist suggests to me that few if any people have done much non-automated work on Maui yet. The question is whether it will take longer for people to do it manually or if doing it in an automated way will be more efficient. Believe me, the prospect of doing a mass edit like that is pretty daunting as well. Can we verify how much editing has been done on Maui? If it is as little as I think, there is little to loose by throwing away the existing data and starting over. I still think re-importing (or some kind of automated batch editing) might be a faster and easier way to fix the problems with Maui. Sure. Just download the entire island and see how many objects have been touched by people other than DaveHansen. JOSM can tell you this pretty quickly. -- Dave -- Scott Atwood Cycle tracks will abound in Utopia. ~H.G. Wells ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us