From: Josef Reisinger <datafabdriver@ao...>  20120709 14:50:30
Attachments:
Message as HTML

Folks, I assumed somehow this topic is set and understood, but.... I have some skytraqbased GPS data loggers and used gpsbabel with the skytrag module to read the tracks from these devices. There is a Windowsbased software called nTrip, which is delivered with the device to download and visualize the tracks. I was recently asked by a user of gpsbabel and nTrip, why the altitude in nTrip and gpsbabel differs remarkably. I believed this might be a typo in the (pretty complex) formula in the skytraq.c module, which turned out not to be the case. I tried to compute the altitude with the code of skytraqdatalogger, I tried even to code what I fond in various web pages ((1) http://www.joachimbolz.de/projekte/gps/gps.html <http://www.joachimbolz.de/projekte/gps/gps.html>;, (2) http://www.fsd.mw.tum.de/HomepageDaten/Lehre%20WS1112/FSD1/FSD_I_Trafo_WGS.pdf,) describing the transformation ECEF > Lat/lng. I am now at a point where I need real expertise. The big question now is: how could I verify the correctness of a formula given that: * Two independently functions (the one in skytraq.c and (1) above) yield the same result for altitude for a given Lat/Lng pair * The result of these functions deviate from nTrip by >30m * The result of nTrip function is closer to reality (tried a few spots where I can get the altitude information). In particular, nTrip seems to be close to what I can get from a database available through http://www.gpsvisualizer.com/elevation. * (2) gives results close to nTrip for some selected trackpoints. Any expert out there who could have a look at (2) and confirm that this formula should be implemented as the transformation ECEF > Lat/Lng? Kind regards Josef Reisinger 
From: Stefan Bauer <yoltiamaintain@fr...>  20120710 21:46:32

On Monday 09 July 2012 16:50:05 Josef Reisinger <datafabdriver@...> wrote: > Folks, > > I assumed somehow this topic is set and understood, but.... > > I have some skytraqbased GPS data loggers and used gpsbabel with the > skytrag module to read the tracks from these devices. There is a > Windowsbased software called nTrip, which is delivered with the device > to download and visualize the tracks. > > I was recently asked by a user of gpsbabel and nTrip, why the altitude > in nTrip and gpsbabel differs remarkably. I believed this might be a > typo in the (pretty complex) formula in the skytraq.c module, which > turned out not to be the case. > > I tried to compute the altitude with the code of skytraqdatalogger, I > tried even to code what I fond in various web pages ((1) > http://www.joachimbolz.de/projekte/gps/gps.html > <http://www.joachimbolz.de/projekte/gps/gps.html>;, (2) > http://www.fsd.mw.tum.de/HomepageDaten/Lehre%20WS1112/FSD1/FSD_I_Trafo_WGS.pdf,) > describing the transformation ECEF > Lat/lng. > > I am now at a point where I need real expertise. > The big question now is: how could I verify the correctness of a formula > given that: > > * Two independently functions (the one in skytraq.c and (1) above) > yield the same result for altitude for a given Lat/Lng pair > * The result of these functions deviate from nTrip by >30m > * The result of nTrip function is closer to reality (tried a few spots > where I can get the altitude information). In particular, nTrip > seems to be close to what I can get from a database available > through http://www.gpsvisualizer.com/elevation. > * (2) gives results close to nTrip for some selected trackpoints. Just a guess: Did you consider that there's a difference between height above geoid (above sea level) and height above ellipsoid? Ref: http://principles.ou.edu/earth_figure_gravity/geoid/index.html Stefan  PGP public key: https://wwwuser.tuchemnitz.de/~stefb/pubkey.asc Fingerprint: E57E 4776 7AA5 C33A F73C A4E9 D79A 9D84 DAF3 7593 
From: Josef Reisinger <datafabdriver@ao...>  20120711 10:18:35

Hi Stefan, thanks for your response. As said, I am not an expert (others would say I "... have no clue..." ) on this subject area, but my understanding so far was that mentioning "WGS84" (and the corresponding a/b diameters of the model for earth) would lead to the conclusion that an ellipsoid is used in the formulas. Any more hints out of the community? Otherwise I would be in favor of (2) and replace the current formula. /jr On 10.07.2012 23:43, Stefan Bauer wrote: > On Monday 09 July 2012 16:50:05 Josef Reisinger <datafabdriver@...> wrote: >> Folks, >> >> I assumed somehow this topic is set and understood, but.... >> >> I have some skytraqbased GPS data loggers and used gpsbabel with the >> skytrag module to read the tracks from these devices. There is a >> Windowsbased software called nTrip, which is delivered with the device >> to download and visualize the tracks. >> >> I was recently asked by a user of gpsbabel and nTrip, why the altitude >> in nTrip and gpsbabel differs remarkably. I believed this might be a >> typo in the (pretty complex) formula in the skytraq.c module, which >> turned out not to be the case. >> >> I tried to compute the altitude with the code of skytraqdatalogger, I >> tried even to code what I fond in various web pages ((1) >> http://www.joachimbolz.de/projekte/gps/gps.html >> <http://www.joachimbolz.de/projekte/gps/gps.html>;, (2) >> http://www.fsd.mw.tum.de/HomepageDaten/Lehre%20WS1112/FSD1/FSD_I_Trafo_WGS.pdf,) >> describing the transformation ECEF > Lat/lng. >> >> I am now at a point where I need real expertise. >> The big question now is: how could I verify the correctness of a formula >> given that: >> >> * Two independently functions (the one in skytraq.c and (1) above) >> yield the same result for altitude for a given Lat/Lng pair >> * The result of these functions deviate from nTrip by >30m >> * The result of nTrip function is closer to reality (tried a few spots >> where I can get the altitude information). In particular, nTrip >> seems to be close to what I can get from a database available >> through http://www.gpsvisualizer.com/elevation. >> * (2) gives results close to nTrip for some selected trackpoints. > Just a guess: > Did you consider that there's a difference between height above geoid (above sea level) and height above ellipsoid? > Ref: > http://principles.ou.edu/earth_figure_gravity/geoid/index.html > > Stefan > 
From: Josef Reisinger <datafabdriver@ao...>  20120711 13:18:36

I compared the height difference between (1) and (2) and it turns out that the difference in height is pretty constant. I have analyzed tracks with 16554 points in an area of a variation in lat/lng/alt of 0.177277/0.162193/394.858827 and the difference between the two formulas is in the range of [68.010577..68.173726] meters with a clear maximum @68.011222 with 4306 points. I'll be on a business trip next week and try to get some tracks there and redo this investigation. /jr On 10.07.2012 23:43, Stefan Bauer wrote: > On Monday 09 July 2012 16:50:05 Josef Reisinger <datafabdriver@...> wrote: >> Folks, >> >> I assumed somehow this topic is set and understood, but.... >> >> I have some skytraqbased GPS data loggers and used gpsbabel with the >> skytrag module to read the tracks from these devices. There is a >> Windowsbased software called nTrip, which is delivered with the device >> to download and visualize the tracks. >> >> I was recently asked by a user of gpsbabel and nTrip, why the altitude >> in nTrip and gpsbabel differs remarkably. I believed this might be a >> typo in the (pretty complex) formula in the skytraq.c module, which >> turned out not to be the case. >> >> I tried to compute the altitude with the code of skytraqdatalogger, I >> tried even to code what I fond in various web pages ((1) >> http://www.joachimbolz.de/projekte/gps/gps.html >> <http://www.joachimbolz.de/projekte/gps/gps.html>;, (2) >> http://www.fsd.mw.tum.de/HomepageDaten/Lehre%20WS1112/FSD1/FSD_I_Trafo_WGS.pdf,) >> describing the transformation ECEF > Lat/lng. >> >> I am now at a point where I need real expertise. >> The big question now is: how could I verify the correctness of a formula >> given that: >> >> * Two independently functions (the one in skytraq.c and (1) above) >> yield the same result for altitude for a given Lat/Lng pair >> * The result of these functions deviate from nTrip by >30m >> * The result of nTrip function is closer to reality (tried a few spots >> where I can get the altitude information). In particular, nTrip >> seems to be close to what I can get from a database available >> through http://www.gpsvisualizer.com/elevation. >> * (2) gives results close to nTrip for some selected trackpoints. > Just a guess: > Did you consider that there's a difference between height above geoid (above sea level) and height above ellipsoid? > Ref: > http://principles.ou.edu/earth_figure_gravity/geoid/index.html > > Stefan > 
From: tsteven4 <tsteven4@gm...>  20130107 23:42:55
Attachments:
Message as HTML

Joseph, This is most likely an issue with vertical datums, and not ECEF > Lat,Lon,Alt conversion. The ECEF_to_LLA converts from ECEF to the WGS84 ellipsoid. I have found sub 1 meter errors in that routine that are a result of approximations in the method used. I believe the results from your reference (2) should match those of the current routine within this same tolerance. This conclusion disagrees with your report. I am not able to read German, but I don't see anywhere in your reference (2) where there is reference to the EGM96 geoid, which is a very complex surface. I believe your desire is to have an altitude relative to the EGM96 geoid, which approximates mean sea level. The SRTM3 data on gpsvisualizer would have been referenced to this geoid and not the WGS84 ellipsoid. Can you provide the location (lat,lon) of your test data and the expected altitudes? The differences between the WGS84 ellipsoid and the EGM96 geoid can be about 105 to +85m depending on the location. Also see the thread http://sourceforge.net/mailarchive/forum.php?thread_name=50EAD3E5.3090204%40gmail.com&forum_name=gpsbabelcode Steve On 7/9/2012 8:50 AM, Josef Reisinger wrote: > Folks, > > I assumed somehow this topic is set and understood, but.... > > I have some skytraqbased GPS data loggers and used gpsbabel with the > skytrag module to read the tracks from these devices. There is a > Windowsbased software called nTrip, which is delivered with the > device to download and visualize the tracks. > > I was recently asked by a user of gpsbabel and nTrip, why the > altitude in nTrip and gpsbabel differs remarkably. I believed this > might be a typo in the (pretty complex) formula in the skytraq.c > module, which turned out not to be the case. > > I tried to compute the altitude with the code of skytraqdatalogger, I > tried even to code what I fond in various web pages ((1) > http://www.joachimbolz.de/projekte/gps/gps.html > <http://www.joachimbolz.de/projekte/gps/gps.html>;, (2) > http://www.fsd.mw.tum.de/HomepageDaten/Lehre%20WS1112/FSD1/FSD_I_Trafo_WGS.pdf,) > describing the transformation ECEF > Lat/lng. > > I am now at a point where I need real expertise. > The big question now is: how could I verify the correctness of a > formula given that: > > * Two independently functions (the one in skytraq.c and (1) above) > yield the same result for altitude for a given Lat/Lng pair > * The result of these functions deviate from nTrip by >30m > * The result of nTrip function is closer to reality (tried a few > spots where I can get the altitude information). In particular, > nTrip seems to be close to what I can get from a database > available through http://www.gpsvisualizer.com/elevation. > * (2) gives results close to nTrip for some selected trackpoints. > > Any expert out there who could have a look at (2) and confirm that > this formula should be implemented as the transformation ECEF > Lat/Lng? > > Kind regards > Josef Reisinger > > > >  > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats.http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > > > _______________________________________________ > Gpsbabelcode mailing listhttp://www.gpsbabel.org > Gpsbabelcode@... > https://lists.sourceforge.net/lists/listinfo/gpsbabelcode 