My ASTER GDEM2 has a average elevation error of -10m for my area. (I testet it using 90 defferential GPS reference points and compared it with old Topo25, SRTM, and GoogleMaps).

I would like to correct it by +10m for every pixel. Any idea? Whats the reason for that "error"? It is measured at a different height reference?

  • Are the differential GPS points, SRTM and other data sources in perfect agreement? Is the ASTER data -10 for all the points? – Simbamangu May 21 '12 at 12:29
  • not in perfect agreement, but the amount of reference points from different sources are trustable enough, I guess :) -10m is the average for every pixel. Some of them are also some meters above the reference points. – MAP May 21 '12 at 12:51
  • 1
    Its due the technical constraints. It was made with stereo images taken from space at ~15m, if I am not mistaken. GDEM covers whole Earth! Have a look at different techniques of DEM creation. Each one is time and labor consuming, and offers some level of accuracy (usually for smaller area than whole Earth). Even from most accurate LiDAR (ALS) may be produced DEM that is not accurate (vegetation etc.). – Tomek May 22 '12 at 5:55
  • 1
    You compared GDEM to SRTM, GoogleMaps(how?)- SRTM is less accurate and if remember well 90m a pixel. TOPO maps are usually saddled with some kind of error as well (especially 25K). Is your GPS RTK? Where the 90 points were observed? All of those data will have errors due technical constraints :) How did you calculated 10m of average error? – Tomek May 22 '12 at 6:08
  • @Tomek I took the elevation manualy from GoogleEarth (not maps, sorry) at the coordinate-location. I devided the reverence points into different types of relief such as "steep mountain relief", "flat relief" "continous relief" and "valleys". The topos btw. have the most errors :) – MAP May 22 '12 at 6:41

This is a job for the Raster Calculator, it's under the Raster menu (don't confuse this with RasterCalc!). The expression:

snowdon_dem.tif@1 + 10

will do this, where 'snowdon_dem.tif@1' is the name of the original DEM. You can also find additional informations here.


i think this is a normal elevation error. if you check any validation report of GDEM2, you always see that it is reported Standart Deviation (SD) is to 12.7 m. whereas version 1 is 15.4 meter.

The ASTER GDEM version 1 was released in July 2009 and the version 2, now under processing, will be released in the October 2011. The GDEM version 2 is reproduced using the updated algorithm. Validation study of the GDEM version 2 confirmed that elevation offset and horizontal resolution will be greatly improved in GDEM version 2 and the enhanced horizontal resolution will serve to reduce the standard deviation of elevation and horizontal error.


  • 1
    Yes, thats the point! The results of the validation report are compareable with mine. But why they are in average 10 to 13m below the real alititude? Why they are not corrected? Or is it because a other definition of the sealevel? – MAP May 21 '12 at 12:49

Likely you will have set your GPS device to WGS84 ellipsoid, while the ASTER GDEM data are referenced to the EGM96 geoid:

See: "The data are posted on a 1 arc-second (approximately 30–m at the equator) grid and referenced to the 1984 World Geodetic System (WGS84)/ 1996 Earth Gravitational Model (EGM96) geoid." https://lpdaac.usgs.gov/content/view/full/11033

  • I must be missing something - isn't the WGS84 datum for the GPS the same as that used by the ASTER GDEM? The WGS geoid is EGM96 ... – Simbamangu May 22 '12 at 5:22
  • But you usually use an ellipdoid in GIS environments... See also here: en.wikipedia.org/wiki/WGS84#Main_parameters Citation: "The deviations of the EGM96 geoid from the WGS 84 reference ellipsoid range from about -105 m to about +85 m.[9] EGM96 differs from the original WGS 84 geoid, referred to as EGM84." – markusN May 22 '12 at 9:15
  • For differences, see this map: grass.osgeo.org/wiki/… – markusN May 22 '12 at 9:17

There is a program f477 which is a Fortran program that generates the correction that you need in fact you have to create a map like a frid that in turn the fortran program uses to create a grid with the correct values. Then you import that grid to GIS and make an algebraic sum (-) of both values, then you have your map.

I'm a Linux user and found it hard to find a solution to format the "Gridd" extracted from grass 15" (apperture) so the Fortran program can read it. Finally I made it with the "cut" command.

  • This answer won't help anybody unless you can supply us a better reference to the program you're talking about. – whuber Sep 22 '12 at 15:45

You could use the libLAS library to do the vertical transformation. You have here a reference to this possibility.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.