Re: [gpsd-dev] [gpsd] Altitude in TPV

[Peter Liu]

I cannot find any definitive answer to whether the Altitude is HAV or

MSL. Also, which datum is assumed here?

No real definitive answer is possible. Different GPS report altitude
differently. gpsd has no way of knowing how your GPS is programmed
and/or configured. So it just passes on what your GPS told it.

Most GPS will provide MSL calculated internally from ECEF and WGS-84.
But many can, and do, use other datums.

Check your GS doc, and then compare to a known benchmark.

I often put two GPS side by side. Their lat/lon will agree closely, but
their altitude often differ by 20 feet. I had one $8k GPS that was off
by 64 feet!

gpsd tries to program u-blox to output ECEF. Then gpsd uses the ECEF
and its own WGS-84 conversion to get to MSL. When compared to USGS
benchmarks the agreement is very good.

Great. Thank you for the quick response. I can pick what I wanted to use.

ECEF is nice. I wish I can get ECEF from the INS.

Regarding MSL vs. Height Above Ellipsoid (HAE). I always like to go back to the point MITRE's Mike Butler makes in the "Developer's Guide to Cursor on Target CoT)" http://www.dtic.mil/dtic/tr/fulltext/u2/a637348.pdf

Below is the excerpt on why HAE is preferred over MSL for data exchanged between systems. I can see why MSL is good for human consumption.

"MSL is empirically derived (as is AGL from DTED). MSL is based on an estimate of “sea level”
which is affected by the density of the Earth’s crust. “Sea level” is actually an irregular,
undulating surface that’s higher where the crust is densest. The shape of this surface is modeled
by a big 2-D table of measurement, an Earth Gravity Model (EGM). The model is refined
periodically (EGM84 and EGM96 are both in current use). Models are available in multiple
resolutions (10x10 degree grids, 15x15 min grids, etc.), and there are multiple interpolation
algorithms in use.
In short, MSL is an interoperability nightmare. You have to have the same model, gridding, and
interpolation to exchange information, or errors will result. All this is avoided if we use HAE.
HAE is a mathematical model with just a few parameters for ellipse axes. It’s simple, accurate,
and stable."