What's The Current Status Of Altitude Data?

[Consultant]

I've read multiple threads over the past couple of years with changing information regarding how the P4P records altitude. There was talk DJI went from GPS based to Barometric Altimeter Based. And there was additional talk that it was now storing the same altimeter altitude reading in two different fields in the EXIF data (GPS and non-GPS). I admit I never really dug around in the EXIF data until now but I'm using windows and I just brought up a photo in Windows Explorer and looked at the EXIF properties and I only saw one GPS altitude and it was obviously true GPS because the photo was (purposely) taken while the drone was on the ground and the elevation was like 270 feet.

I'm also testing Agisoft Photoscan and when it brings the photos in the altitude it shows for the camera positions is the GPS altitude. There is a known issue with the barometric altimeter having a "drift" while the drone is heating up. I see the drift examing the change in the GPS altitude. So I would assume that even though the GPS altitude isn't very accurate, at least the changes in altitude due to the drift appear to mitigated since the relative changes in altitude as measures by the GPS are fairly accurate even if the absolute altitude above sea level can be quite off. (The relative changes are what is important in the photogrammetry processing)

Does anyone have a solid current understanding of what's going on with the current firmware regarding recording of altitude? Am I just not seeing the barometric AGL altitude because I need a better EXIF viewer? Are some of the other processors (such as DroneDeploy or Pix4D) using a different field with the AGL altitude whereas Photoscan uses GPS altitude?

 

[Fly Dawg]

It is my understanding there is actually no such animal as GPS Altitude, per say. The Phantoms have always used Barometric Altimeters for the Altitude, completely un-related to GPS. The only way to have what you might consider a GPS altitude would be with a topography interface of some sort. True GPS knows nothing about altitude. Others can chime in more accurately, but this is my basic understanding. On an added note, if there were to be a Radar Altimeter as some future date, then you would have your topography interface.

 

[DanMan32]

GPS does have Z axis but not very accurate. My Garmin Nuvi can provide altitude.
But I too understood Phantoms used barometer exclusively for altitude. Internally it is recorded in meters, and shows that way in video caption.

 

[bsartist]

GPS does have Z axis but not very accurate. My Garmin Nuvi can provide altitude.
But I too understood Phantoms used barometer exclusively for altitude. Internally it is recorded in meters, and shows that way in video caption.

Agreed. Enough visible sats at a low azimuth can give altitude data. But, yes, often inaccurate.

 

[Meta4]

I've read multiple threads over the past couple of years with changing information regarding how the P4P records altitude. There was talk DJI went from GPS based to Barometric Altimeter Based. And there was additional talk that it was now storing the same altimeter altitude reading in two different fields in the EXIF data (GPS and non-GPS). I admit I never really dug around in the EXIF data until now but I'm using windows and I just brought up a photo in Windows Explorer and looked at the EXIF properties and I only saw one GPS altitude and it was obviously true GPS because the photo was (purposely) taken while the drone was on the ground and the elevation was like 270 feet.

I'm also testing Agisoft Photoscan and when it brings the photos in the altitude it shows for the camera positions is the GPS altitude. There is a known issue with the barometric altimeter having a "drift" while the drone is heating up. I see the drift examing the change in the GPS altitude. So I would assume that even though the GPS altitude isn't very accurate, at least the changes in altitude due to the drift appear to mitigated since the relative changes in altitude as measures by the GPS are fairly accurate even if the absolute altitude above sea level can be quite off. (The relative changes are what is important in the photogrammetry processing)

Does anyone have a solid current understanding of what's going on with the current firmware regarding recording of altitude? Am I just not seeing the barometric AGL altitude because I need a better EXIF viewer? Are some of the other processors (such as DroneDeploy or Pix4D) using a different field with the AGL altitude whereas Photoscan uses GPS altitude?

What you've said is how DJI exif data displays altitude. It's from GPS with baro altitude hidden in there as well.

For the others commenting, the Phantom uses baro altitude data for flight but the OP was asking about data recorded with each photo.

 

[African wildlife]

I'm ignorant about the inner workings of the P4P, with the exception of camera/video settings.

That said, we frequently use an old Garmin GPS bit of equipment for recording exact locations on the nature reserve. This includes a precise altitude reading from GPS, not least because we're in a mountainous/plateau terrain. If this technology was around 10 years or so ago, then.....

 

[Fly Dawg]

Not used to seeing that information in an EXIF. Other information yes, but never altitude data until I just checked an image to verify.

 

[sar104]

It is my understanding there is actually no such animal as GPS Altitude, per say. The Phantoms have always used Barometric Altimeters for the Altitude, completely un-related to GPS. The only way to have what you might consider a GPS altitude would be with a topography interface of some sort. True GPS knows nothing about altitude. Others can chime in more accurately, but this is my basic understanding. On an added note, if there were to be a Radar Altimeter as some future date, then you would have your topography interface.

To clarify, GPS knows nothing about altitude AGL unless combined with a DEM. However, the GPS data used by the FC is 3-dimensional in that it describes a location in space relative to a reference ellipsoid that approximates the shape of the earth. That's typically regarded as representing mean sea level but, in fact, it is not quite the same. Mean sea level is more accurately approximated by an empirical geoid that does not assume a gravitational equipotential ellipse. That discrepancy is a significant source of error in GPS MSL data.

For modern GPS receivers, the mean vertical error (relative to the ellipsoid) is approximately 1.5 times the mean horizontal error. The DJI FCs record both the 3-D GPS location, which includes an altitude above MSL, and a barometric altitude relative to the takeoff point. Both those are added to the EXIF dataset, but you need to view the complete EXIF set, including the XML fields, to see the relative barometric altitude.

 

[Fly Dawg]

@sar104 Well, that clarifies things a bit. You have to remember I come from old school telemetry when we used manual chart recorders and 1 inch tape reels for storage of not more than 256 channels of data and could only view them 16-64 channels at a time. And that's after waiting for the chart printouts. These days its a lot more data than I can handle sometimes........But still giving it a shot!

 

[sar104]

@sar104 Well, that clarifies things a bit. You have to remember I come from old school telemetry when we used manual chart recorders and 1 inch tape reels for storage of not more than 256 channels of data and could only view them 16-64 channels at a time. And that's after waiting for the chart printouts. These days its a lot more data than I can handle sometimes........But still giving it a shot!

These are the GPS and barometric data fields extracted from an example DJI JPEG EXIF. Some repetition. Relative altitude is barometric height above takeoff, not AGL.

GPS Version ID : 2.3.0.0
GPS Latitude Ref : North
GPS Longitude Ref : West
GPS Altitude Ref : Above Sea Level
Absolute Altitude : +3233.03
Relative Altitude : +248.10
GPS Altitude : 3233 m Above Sea Level
GPS Latitude : 35 deg 54' 22.70" N
GPS Longitude : 106 deg 24' 15.49" W
GPS Position : 35 deg 54' 22.70" N, 106 deg 24' 15.49" W

 

[Fly Dawg]

These are the GPS and barometric data fields extracted from an example DJI JPEG EXIF

Yes, I saw that in the EXIF of one of my images that I looked at before I posted. Just never have had the need for use of that data nor did I even think of looking, although others might. Data overload my friend.....

 

[David_Cambridge]

If you have at least 3 satellites, GPS Can calculate altitude as well as position. It suffers a wider accuracy error than barometric altitude, which is probably why barometric is used. Note as well that barometric altitude is relative. It is determined at the moment of power up relative to the launch site. It has no connection to height above sea level.

 

[N017RW]

But isn't the the major reason why you need a fourth satellite is for timing corrections?
If you know the exact position and speed of the satellites, tri-lateration will give you 2 points, but won't one be impossible or with an impossible speed?
Since a receiver uses the time (to receive a satellite signal) to determine the distance to that satellite, minor time errors will cause huge positioning errors and thus a large [position] uncertainty when you have only three satellites.
So four or more are needed and DJI likes 6+.

 

[sar104]

But isn't the the major reason why you need a fourth satellite is for timing corrections?
If you know the exact position and speed of the satellites, tri-lateration will give you 2 points, but won't one be impossible or with an impossible speed?
Since a receiver uses the time (to receive a satellite signal) to determine the distance to that satellite, minor time errors will cause huge positioning errors and thus a large [position] uncertainty when you have only three satellites.
So four or more are needed and DJI likes 6+.

The answer is more basic. A GPS receiver trying to determine position in 3D space actually needs to solve for four unknowns, x, y, z and t (where technically t is Δt, the receiver clock deviation from GPS time), not for just the obvious three unknowns (x, y, z). In a set of linked simultaneous equations, that requires knowledge of four independent variables. Each of four satellites provides one of those, in the form of signal TOA relative to the receiver clock. Three satellites do not provide enough information.

 

[N017RW]

Yea, that's my understanding too... just said it less succinctly.

 

[David_Cambridge]

The absolute time can be calculated from the time data from 3. 2 points are needed for a 2d fix, and 3 for a 3D fix. More satellites increase the accuracy by allowing averaging, but 3 are all that is required for co-ordinates on a surface, and altitude comes from a 3rd.

 

[N017RW]

‘Absolute’???
That does not coincide with easily searched trilateration info.
3 SVs ID two possible locations with 1 unlikely. This is low confidence.

Please post or cite your reference for such 3D certainty with 3 SVs.

 

[sar104]

The absolute time can be calculated from the time data from 3. 2 points are needed for a 2d fix, and 3 for a 3D fix. More satellites increase the accuracy by allowing averaging, but 3 are all that is required for co-ordinates on a surface, and altitude comes from a 3rd.

I'm not sure how that's possible. If you have four unknowns then you need more than three equations to solve for them.

 

[sar104]

‘Absolute’???
That does not coincide with easily searched trilateration info.
3 SVs ID two possible locations with 1 unlikely. This is low confidence.

Please post or cite your reference for such 3D certainty with 3 SVs.

It's worse than that, I think. With an unknown offset from GPS time, each TOA sphere is actually a shell of finite thickness and uncertain center, so three satellites can only provide two line segment solutions, not two points.

 

[N017RW]

It's worse than that, I think. With an unknown offset from GPS time, each TOA sphere is actually a shell of finite thickness and uncertain center, so three satellites can only provide two line segment solutions, not two points.

Ok. The diagrams I recall highlited the overlapping points but there is at least 1 line between them. A bit less descriptive than your explanation.