Hi,
I used the same dataset captured with the same drone (DJI Mini 3 Pro). The datasets were taken three months apart. The same settings were used for both model creation and export. No GCPs were used - only the drone’s built-in GPS. i made two 3d models representing each dataset.
The horizontal coordinates of both models match within 70 cm, but the vertical values are off by 140 meters.
Any idea why this might be happening?
Hi,
drone’s GPS is not so precise and it basically measured only the relative height (mostly), so there could be a difference. But 140 m seems to be too much.
Can you check the EXIF GPS values for your images?
In such cases the best option is to use the measured GCPs. As they weren’t use, you can “prepare” some from the first dataset and use them in your second measurements (you will need to use stable points between the datasets). Then they should be in the same system.
You are right - the prior pose absolute accuracy is nearly the same for horizontal coordinates, but for altitude, there’s a significant difference: dataset 1 shows 772 m, while dataset 2 shows 632 m. This is odd because both missions had the same home point, and the second mission’s flight altitude was only 10–15 meters higher than that of the first dataset.
I will get some coordinates from first model and use it as GCPs, but would like to find out why is there such difference in altitude.
Vertical will always be greater than horizontal. The way GNSS works, the calculations that provide the positioning information have a clearer view of the sky to provide horizontal. In fact, while GCPs are great to tie horizontally, they’re really meant to specify the elevation relative to the referenced vertical datum.
If you’re only using GPS (the US GNSS constellation) you’re going to be off by 3-4m horizontally, and 5+ meters vertically. You may see better values for precision, but absolute accuracy will be off.