Hello, during a recent flight for photogrammetric data, I encountered a strange error in the EXIF GPS altitude recorded by the Mavic-2 Pro. As you can see it registered as 74 m below sea level in the example given. The entire flight (hundreds of photographs) suffered the same issue. What causes this please and how to rectify it? I realise that the GPS is less accurate in Z than in X, Y but I did not expect to find heights below 0 m.a.s.l.

I am not so concerned with correcting these data but I want to avoid similar issues in future flights. Any advice or insights appreciated.

As you can see it registered as 74 m below sea level in the example given. The entire flight (hundreds of photographs) suffered the same issue. What causes this please and how to rectify it? I realise that the GPS is less accurate in Z than in X, Y but I did not expect to find heights below 0 m.a.s.l.

If you looked deeper into the image metadata, you'd find two different altutude numbers.
The one labelled GPS altitude doesn't actually come from GPS at all and isn't used by most photogrammetry software.
The "GPS" altitude is useless and actually comes from a rough approximation of altitude above sea level calculated from the air pressure, but using some incorrect assumptions.
Like I said, it's useless.

The altitude figure you see on the screen is also in the image metadata (but you might need a different data viewer to find it).
That's the altitude figure that photogrammetry software would use.

Thank you for the reply Labroides - but as you can see from the above screenshot in metashape, actually, Metashape is using this -70m ish height from the exif data.

KA77 Posted at 5-14 04:47
Thank you for the reply Labroides - but as you can see from the above screenshot in metashape, actually, Metashape is using this -70m ish height from the exif data.

Does this make any difference to the finished product?

How do you get that display in Metashape?
I'm running a test and can't find it.

Thanks for the reply
Since I have used ground control, the output DSM appears in almost the right place in 3D, but without ground control, this would deliver an upside down surface model. I am just more interested in why the GPS might be doing this and giving such an improbable value for Z

KA77 Posted at 5-14 05:17
Thanks for the reply
Since I have used ground control, the output DSM appears in almost the right place in 3D, but without ground control, this would deliver an upside down surface model. I am just more interested in why the GPS might be doing this and giving such an improbable value for Z

It's not the GPS.
AS I pointed out it's DJI's misguided attempt to derive an above sea level height from barometric data.
And just to confuse everyone, they mis-label it as GPS data.

Can you tell me how to get that display of the photos and heights so I can compare?
Also it would be of interest for comparison, if you could post one of the original images  so I can check the full metadata.

Picture size too large to upload... here is a link to one example https://drive.google.com/file/d/1jJ78YC1TdErmIWT99gLLmcTerWveRFB5/view?usp=sharing

That screenshot from metashape comes by using 'reference' when importing images and aligning them in the workflow

This shows that the reference appears to be 'below sea level' but I don't understand how or indeed if it is possible to change this

KA77 Posted at 5-14 05:33
[view_image]

This shows that the reference appears to be 'below sea level' but I don't understand how or indeed if it is possible to change this

OK ... I assumed that Metashape used the altitude that you see on your tablet while you fly.
But checking here, it does use the "GPS altitude" as you've found.
I usually don't use GCPs, but haven't noticed any upside down issues.
The program just plots the model at the minus altitude ... as if it was in Death Valley.
And if you use GCPs, that ties it to the local dataum.


Here's part of the XMP section of your image metadata where it shows both altitude figures.

Labroides Posted at 5-14 05:48
OK ... I assumed that Metashape used the altitude that you see on your tablet while you fly.
But checking here, it does use the "GPS altitude" as you've found.
I usually don't use GCPs, but haven't noticed any upside down issues.

Ah, OK thank you this is quite helpful