Did my first true 3d survey with a p4rtk recently. I ran the images through drone deploy with good results, but most of the resulting map is recorded as 120ft BELOW sea level. Checked the Exif data on the images, and although it's wrong they seem to support that it wasn't a mistake with the stitching software. I flew the flight at 55m and the elevation in the exif data is around 20m. I guess my first question is why am I getting numbers a couple hundred feet off from the images, and if it is off that much can I still expect the elevation data to be consistent on dronedeploy? I.e. if a point on the resulting map is -120.8, and another -110, am I safe in assuming there is 10.8 ft of elevation difference?

I've poked around here on the forum and elsewhere without finding a real answer to the problem, but I'm also new here and very likely could have missed what I'm looking for. Sorry in advance if that's the case

Hi bac123-

Glad to help figure out what's going on if we can, but much more info is needed:

Did you use RTK and have it fixed for the whole flight?

If so, what correction method did you use (D-RTK 2 or other base on a known location? NTRIP?)

If you used an on-site base station, how did you enter the base coordinates? Decimal Degrees and height HAE in m? Something else?

If you used NTRIP, what are the specifics of the provider you used?

What does Drone Deploy expect in terms of input datums/coordinate systems? What are options for output and what did you you choose? It sounds like you're assuming the vertical of the output is relative to MSL, but do you know that?

You also may have some incorrect assumptions about the elevation values in the exif, but the answers to some of the above questions will help figure that out.

patiam Posted at 1-26 14:59
Hi bac123-

Glad to help figure out what's going on if we can, but much more info is needed:

sorry for not including earlier. I did use RTK, it was operating the full flight, and I was using the drtk2 mobile station.

I've read on here somewhere that manually entering base coordinates can get a touch finicky, and I have no way to get an exact long/lat/elev that would be better than the mobile station so I chose to leave it as it was.

My thought was that even if there was some difference in the drtk2's assumed location and it's actual location the map would come back ok because it's basing the other elev numbers off of the drtk2. Does that make any sense?

truthfully the rest is a little over my head. What I can definitively say is that I've got what looks like a pretty solid DSM, DTM, 3d model, and pointcloud. I don't have any previous map of the area as this was my first so I have nothing to go back to, but the little checklist that they send post processing said everything went smoothly.

RTK GNSS-based elevations in the exif are in Height Above Ellipsoid (HAE).

While you had RTK "fixed" during your flight, your accuracy is compromised due to the fact that the base didn't know it's position with high confidence. If you did not enter precise coordinates in the D-RTK 2, it was in autonomous mode so your uncertainty is likely in decimeters (at least) or possibly as bad as 1-2 m horizontal, and worse vertical.

There's lots more to say, but really you need to understand how RTK GNSS works, and a bit more about geodesy (datums, coordinate systems, etc.) if you want to use this equipment to do survey-grade work. You can start by reading threads here, they may refer you to useful information on these topics.

patiam Posted at 1-27 09:25
RTK GNSS-based elevations in the exif are in Height Above Ellipsoid (HAE).

While you had RTK "fixed" during your flight, your accuracy is compromised due to the fact that the base didn't know it's position with high confidence. If you did not enter precise coordinates in the D-RTK 2, it was in autonomous mode so your uncertainty is likely in decimeters (at least) or possibly as bad as 1-2 m horizontal, and worse vertical.

appreciate the help! wasn't really expecting survey grade, especially just jumping in, but didn't expect the base station to be (what i thought was) that far off all by itself. What you're saying is I need to establish a GCP and then set the base over that?

So over a really small area with the base station as the center, the measurements can be off several meters? I guess what I'm asking isn't will i get very exact long/lat/elev for every point, but would the resulting dtm show me anything close to correct elevation change in points a couple hundred feet away?

RTK GNSS is a type of Differentially Corrected GNSS. What that means is that a base station on a precisely known location generates corrections by comparing the position solution it gets from satellite ranging to the known position, and these are provided to the nearby (< 10 km preferably) rover, which uses them to correct it's own position. The "differential" is the difference between the computed and the real, correct position.
If the base station's location is not precisely and accurately known, the corrections are very likely not accurate. Inaccurate corrections yield inaccurate positions. Using the position calculated by the uncorrected D-RTK 2 as the base station position for generating RTK corrections is exactly this situation.

You need to either put the D-RTK 2 on a benchmark with precisiely known coordinates (and enter them into the app), or provide it with RTK corrections from another source so that it can accurately determine its own position, or use PPK.

This is basic GNSS stuff. It behooves you to educate yourself on it if you endeavor to use RTK and expect accurate results.

patiam Posted at 1-26 14:59
Hi bac123-

Glad to help figure out what's going on if we can, but much more info is needed:

Hello , can you help please?
I have noticed that for a month now, I have differences in altitude in my Phantom 4- rtk mode surveys.
I have a difference of 40 to 60cm and I want to solve my problem.
The IMU calibration has been done, the height compensation at 1.802m has been done, the number of satellites is 27-28 in rtk mode. After processing with Metashape and comparison with previous data, there is a difference beyond the tolerance of -/+ 5cm.

Thanks
M

Marcellin.o Posted at 7-26 08:05
Hello , can you help please?
I have noticed that for a month now, I have differences in altitude in my Phantom 4- rtk mode surveys.
I have a difference of 40 to 60cm and I want to solve my problem.

What is the altitude, and are you using "elevation optimization"?

Without a few oblique images, the Z accuracy of P4R photogrammetry is inversely related to altitude above 40 m or so AGL. Using the elevation optimization option during mapping helps mitigate this.