On Sat, Jan 27, 2024 at 7:06 PM Greg Troxel <g...@lexort.com> wrote:
> As someone not happy about the deprecation of mailinglists, a few brief > comments here: > > First, I think this proposal is fine, as documenting widespread > practice. Regardless of my further comments, I think it's solidly > progress to adopt it. > > While yonur comments about survey feet are valid, modern elevations > (NAVD88) are as far as I can tell actually in meters, and when > expressed in feet, in international feet. Elevations are small enough > that 2 ppm is less than the errors in the values. > > I would expect the proposal to give an example. It seems that one > would have a tag > ele=6288 ft > for Mount Washington (showing my East Coast bias). > > It would be good to explicitly state that in keeping with convention, > ft means international feet, perhaps with a parenthetical comment that > if someone meant US Survey Feet they would have written ftUS. Maybe > this is already documented. > As far as I know, Survey Feet are used only for horizontal measurements, mostly in state plane coordinates. And I don't think that there are yet any elevations, anywhere, for which precision to 2 parts per million matters. > Further, WGS84's first height definition is ellipsoidal height, and > that simply is not elevation. Obviously elevation should be in "WGS84 > Orthometric Height", which is what GPS receivers provide as elevation. > But elevations are not published in WGS Orthometric Height; they are > published in a national or regional datum which is pretty close, as > all datums at least roughly target a similar origin. > In newer datums than WGS84, even horizontal position is reported relative to an epoch, with corrections for phenomena like continental drift. I've hiked to one of the stations used for such geodesy - a copper bolt placed in 1870 in truly ancient and stable rock (Hawkeye Gneiss, ~1.15 Ga) with accuracy that would have been considered first-order even a century after its placement. The difference between survey feet and international feet for horizontal coordinates is significant, because in UTM coordinates, the 2 part-per-million error adds up to 20 metres in the polar regions (or 10 or so in the mid-latitudes). It's somewhat less significant over the smaller range of state plane coordinates, which is why most states don't plan to do anything about it until 2025 at the earliest. In a few 'bad' planes like Nevada North and Michigan East, the origins are far enough outside the state that the errors can be about 15 m. Many "authoritative, official" data sets that I've worked with appear to suffer from mixed horizontal datums, with NAD27 coordinates mistakenly transferred over into WGS84 uncorrected. When working with county tax maps, for instance, I always try to spot-check things like street intersections or monumented corners, to make sure what datum I'm working with, because it's often not what the map claims to be. Sometimes the error is backward - the county has never made the conversion of its systems off of NAD27, but WGS84 data have crept in! But for vertical coordinates, I'm willing to wager that no mapper has the technology to measure absolute elevation to less than a mm. In fact, I doubt that it's even meaningful to discuss that sort of precision in elevation. The geoid isn't defined to that accuracy. So I can easily live with the ambiguity in which 'ft' are used. (Moreover, I can't think of any OSM tag at the moment in which a measure in feet would need a few parts in 10**-6 precision.) I'd certainly approve of a proposal optionally to tag elevations with the vertical datum used - but since the differences are typically on the order of 1-2 metres, I surely don't insist on such a tag! I understand that most GPS units use GRS80+EGM1984, but I'm sure that NAVD88 has crept in, even among objects that I've mapped. And I betcha there's a ton of uncorrected NGVD1929 in TIGER. For a metre or two, who cares (yet?). But I'll roundly condemn anyone who confuses height-above-ellipsoid with elevation! (I'm looking at you, Android Location Services!) Changes in vertical datum together with LIDAR data have wrought confusion among some of the hiking clubs around here, who maintain lists like the Adirondack 46'ers (the 46 summits in the Adirondacks thought at the time of the list's compilation to exceed 4000 feet elevation) or the Catskill 3500's (the 34 summits thought at the time of compilation to exceed 3500 feet, plus Leavitt Peak - which had escaped the notice of the people who compiled the list, and minus Doiubletop and Mt Graham - which are now off-limits to hikers). The uniform decision of the clubs was to ignore the problem and simply say that the lists are by now traditional - it's _these_ summits, and no others. Most Adirondack 46'rs also climb Mt MacNaughton, which was not on the list, but was revealed in a 1953 survey to exceed 4000 feet. But then the change of vertical datums to NAVD88 put it below the 4000 foot threshold again, and it's listed as 3,983 feet. Four of the original list also fall short of 4000 feet but are still required for the award. -- 73 de ke9tv/2, Kevin
_______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
