Inconsistency between DSMs generated with images acquired at different altitudes.

It is expected that DSMs generated with images acquired at different altitudes of the same area should be generally consistent with each other. However, we found there are great variations between these DSMs.

Fig. 1(a) and (b) are DSMs generated with images obtained with altitudes of 30m and 60m, respectively. They are visually consistent with each other. However, when subtracting (b) from (a), we found that for a large proportion of the area the difference is over 5 cm ((c) with red color). We further identified that the difference between (a) and (b) is generally within 10 cm ((d) with blue and green color).

My question is why there are such huge variations between DSMs with different flying altitudes, and is it possible to reduce the vertical difference from 10 cm to 5 cm in this case, and how?

Any comments will be greatly appreciated!!

Our images were collected with DJI PHANTOM 4 at the altitude of 30m and 60m above ground, and the DSMs are generated using Pix4Dmapper Pro 2.1. The same GCPs were used for generation of both DSMs, which were measured with a surveyor grade dual frequency RTK GNSS receiver (Fig 2).

Errors of GCPs and Check points of altitude of 30 m and 60 m are given in table1 and 2, respectively.


Please send us the Quality Reports and the log files of the two projects to:

Best Regards, 

Dear support, the Quality Reports and the log files of the two projects were emailed to:

Cheers and looking forwards to your reply!

What is the character of the subject being heighted? There are a lot of variables at play that could influence results. But imagery from 30m is going to identify some detail not seen at 60m. Is this bare earth?


thanks for your comment, Steven.

it’s a newly plowed agricultural field without any vegetation so far, it’s expected that the DSMs generated by images acquired at different altitudes should be consistent under a certain accuracy level (e.g. 5 cm or less), so that the UAV can be employed to monitor crop growth.

However, it turned out to be 10 cm in our experiment. we’r trying to figure out wht’s contribute to the large discrepancy in Z direction.

S what was the end result of this? any workarounds?

Hello everyone and thank you for sharing your feedback with us. We are continuously working on improving our software. 

We would recommend you to download the latest version 3.1.23 where there are many improvements and various bug fixes.  

As a workaround, you could use the option “All Prior” which will force the optimal internal parameters to be close to the initial values.
You can check here for more information.

Best Regards,