Re: [gpsd-users] MSL-WGS84 separation

[Greg Troxel]

address@hidden writes:

> I need an altitude-output in the WGS84 reference system. As it seems I can
> only read the altitude as above mean sea level. So I wanted to read the
> separation from the GPS device (ublox 6) that should be available in the
> struct gps_data_t, but the value returned by the GPSD is always zero for
> the separation. I checked the data-stream coming from the GPS and there
> the separation appeared correctly in the $GPGGA line of the NMEA-output:
> $GPGGA,120521.00,4333.88682,N,00128.90257,E,2,08,1.21,155.3,M,48.5,M,,0000*52

A few semi-independent comments:

0) If you're hoping to get submeter with heights from (other than
double-difference carrier-phase) GPS, you're almost certainly out of
luck.   But 1m+ish is not crazy.

1) It would be nice if gpsd recorded the geoid height value from the GPSr
and returned that, so that one could get back to ellipsoidal height
(above the WGS84 ellipsoid).

2) "MSL" is an obsolete term.  There is no such thing as global mean sea
level in modern height systems.  (It did exist in NGVD29, a ~US vertical
datum from 1929, where many tide stations were held fixed.  But in
NAVD88, only one station was held fixed.)   Instead, modern height
practice measures the difference in sea level among places, called "sea
surface topography".  But I think that's just a phrasing issue and not
about the essence of your question.

3) Besides ellipsoidal height (meters above the purely ellipsoidal
reference surface, as fixed to the earth by the datum), people use
various kinds of heights relative to gravity, in particular orthometric
height, which essentially doesn't change as one moves normal to the
gravity gradient.  In the US the modern one is NAVD88, and I'm sure
there is a similar high-accuracy crust-fixed datum in Germany (probably
Europe-wide), but I'm not familiar with the details.

4) WGS84 includes a geoid model that gives the difference between
ellipsoidal height (measured by GPS) and a form of orthometric height.
GPS receivers have some version of this model and, after measuring in
ellipsoidal height, convert to orthometric height, so that the values
are usable relative to datums like NAVD88.  There have been various
versions of WGS84, and I think various geoid models.   Geoid models are
published every few years and are constantly improved.


So if you want "WGS84 ellipsoidal height", you'll need to spiff up gpsd
to carry the geoid separation from the receiver (because you want what
it subtracted to add back, not some 'right' value).  If you want a WGS84
pseudo-orthometric height, you've got it already.


If you want to know more, go to

  http://www.ngs.noaa.gov/corbin/past_classes.shtml

and listen to all the lectures that have height in the description :-)


If someone from Europe would like to chime in about modern European
vertical datums, I'm very curious.

pgprMOZ0OL5RQ.pgp
Description: PGP signature