On Fri, Sep 8, 2017 at 4:54 AM, Colin Smale <colin.sm...@xs4all.nl> wrote: > Does anyone have any idea whether the elevations, be they in feet or metres, > are all respecting the wiki definition of being the height above MSL > according to EGM96 (not sure what that would mean in landlocked areas) and > NOT WGS84 or (strictly speaking) relative to local MSL? > > The wiki page for ele says this: > > Elevation (height above sea level) of a point in metres. This is mainly > intended for mountain peaks but could also be used for elevation of airport > runways and many other objects. For OpenStreetMap, this value should be in > meters above above mean sea level as defined by the EGM96 geoid model. This > elevation is usually very close to national "above sea level" systems with > differences < 1m. This is not the height above the WGS84 ellipsoid (see > Geoid) which is shown as raw elevation by some satellite navigation devices > and which can differ from geoid elevation by up to 100m. > > If we can't even rely on the reference point for these elevations, > discussions about feet vs. metres (assuming the unit is indicated properly) > are close to "bikeshedding".
Note that in that definition it refers to the WGS84 ellipsoid, not the WGS84 geoid. If WGS84 is implemented correctly, the ellipsoid is used for horizontal control only and the geoid is used for vertical control. Many widely available libraries, such as https://geographiclib.sourceforge.io/, give reference implementations. I would presume that modern GPS equipment - not bare receivers, but the firmware and software surrounding them - gets it right. I certainly haven't noticed, when using my smartphone, the 30 m elevation difference that I'd see if it were relative to the reference ellipsoid. When I have tagged elevations, it's generally of places I've been. Because I'm in North America, I've generally committed the small sin of using NAVD88 or possibly a published elevation relative to NGVD29. I tend to go with USGS surveyed elevations where they exist and my altimeter doesn't show an egregious error - I presume that the trigonometric elevations are more accurate than an altimeter. I don't bother with orthometric reduction because in my area, the three reference surfaces (NGVD29, NAVD88, EGM96) concur to better than my altimeter's resolution, to say nothing of its repeatability. NAVD88-NGVD29 around here is typically less than a metre. I reset my altimeter at either a benchmark or a tabulated spot elevation several times a day if I'm doing elevations, particularly if the weather is varying. Even my smartphone has a barometer, but ordinarily I carry a separate altimeter. Not perfect, but to do better in the field I'd need better instruments than I carry, or even own. I don't do that sort of work unless you pay me to do it, and if you offer to pay me, I'll happily give you the business card of a licensed surveyor. I tag measurements in SI units because that's the only thing that I know all the tools can handle. (Memo to self - still need to repair those misplaced Catskill peaks. Too much to do, too little time...) I think the only takeaway from all this is "don't use the reference ellipsoid for vertical control." But my understanding is that most of us will get it right without realizing it, because our software does it for us. Even more important is "don't use a GPS-only reference for elevation unless you understand dilution of position." GPS-derived elevations are pretty wonky. (But see above: even my smartphone has a barometer.) _______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging