Yeah thanks - I suspected the -34 was incorrect but that's 50 meters off the correct mark. Hard to imagine air pressure would do that much. But I will do more tests and report back.
Standard sea level atmospheric pressure is 1013 hPa and generally ranges from around 1000 hPa to 1050 hPa, although the highest and lowest on record are 850 hPa and 1084 hPa. But even the usual range represents over 500 m difference in pressure altitude, which is what barometric sensors measure.
Or ... I notice that metadata from the M3 shows some differences to metadata from earlier drones I used.
For one shot with the M3, it shows:
Absolute Altitude: -51.122
Altitude Type: GpsFusionAlt
They used air pressure in earlier models (discovered by Sar104).
But perhaps they are using GPS now which could give incorrect results due to problems with the geoid not properly representing the actual shape of the earth ?
That's far too much to be a geoid error. If we are talking about southern Australia the errors are only a few meters at most.
DJI switched back in 2019 ago to using true GPS altitude in the EXIF data, but the field name you are seeing perhaps suggests that they have now changed to a GPS/barometric fusion scheme on the M3. Strange choice, if so, since the GPS altitude data are always going to be more accurate once the aircraft has a decent number of satellites locked.