Hello everybody!

I just recently started to use the Phantom 4 pro v.2 to generate high resolution digital elevation models. I'm just trying different processing workflows so far (I'm using Agisoft Metashape), and even without implementing GCPs, the global location accuracy of the output derived from the camera postions is quite good. In a next step, i wanted to implement GCPs i measured using a differential GNSS (so the accuracy is < 2cm in x/y/z) and discovered, that the altitude of my camera positions is completely wrong - the data generated from the images is ~177m lower then the actual surface. I know that the elevation of the GCPs in right, I compared it with the coarse DEM provided by the state. As I measured the GCPs originally in the local UTM-system (WGS84 UTM 33 N, EPSG 32633) I reprojected it to WGS 84 global (EPSG 4326). I read something about different altitude types (altitude above sea level, altitude above ellipsoid), but even considering that, 177m difference seems too much (when using an online tool to recalculate height data, it shows differences of about 46m between ellipsoidal height and mean sea level).

Did anybody experience similar problems and can anybody tell me something about that issue? The added image illustrates my issue in Agisoft Metashape.

Thanks a lot and cheers from Vienna!


Anna

You P4P2 does not use GPS to measure altitude, as it is far too innacurate. It uses a simple barometric pressure sensor to measure height, and the height is relative to the elevation of the aircraft above the point ar which it took off. The aircraft has no knowledge of altitude above any terrain objects or sea level, and has no sophisticated height measurement methods. It does have an ultrasonic sensor which can measure altitude, called VPS, but thisa is only active in the range from 0 to 3 Metres above ground.

What Geebax said is correct.

However while your raw image altitudes will be subject to the fact that they are barometric pressure-derived, very noisy, and only relative to your home point elevation,  if you properly apply GCP's in any decent photogrammetry software, all your output products should be accurate in not only XY but also Z, to the extent that your GCPs are.

Just using the exif data from the images will only give relative heights which isn't much use.
But as Patiam says, once you bring the GCPs in, that should pull your elevations to match your local datum.
What were your results like using the GCPs?

I have the same problem.. With phantom 4 pro v2..
Have almost always about 45m different between exit elevation data and real elevation of the camera...
What I have to do to solve thus issue...
Thanks

Sorry for the late reply, thanks a lot for the information! Good to know that the elevation data is only from a barometric sensor. I have been flying again on the same field, and the elevation is again different - this explains also that variation. As I was busy with other stuff the last days and did not process again, I haven't tried including the measured GCPs, but I guess it will work then. I'll keep you updated as I try!
Thanks!

anna.i Posted at 4-16 00:49
Sorry for the late reply, thanks a lot for the information! Good to know that the elevation data is only from a barometric sensor. I have been flying again on the same field, and the elevation is again different - this explains also that variation. As I was busy with other stuff the last days and did not process again, I haven't tried including the measured GCPs, but I guess it will work then. I'll keep you updated as I try!
Thanks!

Take off from the top of a hill and fly down it - you'll notice that the altitude reading in the app will be negative - that's why.

I have just had this same issue.

I am processing my 3D Model with Pix4D.  I noticed that my elevations were roughly 200m higher than my actual elevation in my EXIF's.   My thoughts are that on startup the P4P initialized with an erroneous initial elevation and propagated this while flying the site.  Fortunately the elevations are consistent (due to the barometer to manage elevations) so the following is my suggestion.

- it is currently processing so we will see how this helps -

I exported all the EXIF values into a file, took that file and updated the values so they were roughly 50m above my ground height.  I flew this site at 50m, so this is why I used that specific measurement.  NOTE: it appears that the elevations were roughly 185m (That's ±600') above the correct values.

I have now imported these NEW initial positions into my software and will see what the results are.

*** Results are in *** Processed successfully!!!


Take care,

B