On Sat, Jun 1, 2013 at 6:14 PM, Claudius <gpsbabel.list@wolki.de> wrote:
Hi everyone,

i did not find a bugtracker or any mention of one. So i'm trying this
one here. Maybe i'm just doing something wrong.

Doubtful this is pilot error.


Today i created a bunch of tracklogs with a Wintec 202 (tes-files). For
the first time, i actually cared about the altitude information. Usually
i'm just clamping the tracks to the ground.

I did two conversions (one to GPX, one to KML using GPSbabel 1.4.4), and
in both, the altitudes were far off. Not only were they off, they were
actually the wrong way round?

Actual height profile from Google Earth:

Recorded height from Wintec, converted to KML:

The track starts in the upper left corner. I was riding a bus, wich went
downhill. So i'm pretty certain that the Google Earth profile from the
first image is more or less correct.

If anyone cares to take a look, here's my original recording as well:
And the conversion being displayed on Google Earth in the screenshot:

The Wintec TES file format was reverse engineered about 3 years ago.  The code treats the two bytes after longitude as altitude in meters and I don't think anyone before you has challenged that, though there are conflicting comments in the code on how well altitude was actually understood.

I don't have one of those units so I don't know if the alt is barometeric (notoriously inaccurate in some cases) or GPS derived or if it's using internal tables to correct to the geoid or what.  You can see the code that reads these as https://code.google.com/p/gpsbabel/source/browse/trunk/gpsbabel/wintec_tes.cc#68  (You can see the comments that say we don't know what the units are, but then say that it's meters...and that we ignore negative values, but we're putting it in an unsigned value.)

If you have a better understanding as to what those two bytes actually represent - if they're altitude at all - whether you find it empirically, via the source of other free software, or via specs obtained from Wintec and can help us improve that, that'd be great.