a problem between real world and DTM.

Hi friends.
First of all, thank you very much for the help from the many articles here.

I have a problem.
I aerially photographed an area of ​​about 1.3 square kilometers with a Matrice 210 RTK V2 and X5S.
The purpose of the flight was to extract the DTM.
The target is a quite up and down mountainous area with an elevation difference of about 200m.
The RTK was a network RTK and the drone took images while flying at an altitude of 3.8 cm GSD. The overlap was 75% and all images were processed.
The output was generated in a coordinate system suitable for Korea, and the geoid was EGM2008.

The result made me discouraged.
This is because the difference between the DTM generated by PIX4D and the actual measured altitude was too large. There was a large difference of more than 10m, and in some open areas, there was a difference of several tens of centimeters from the surveyed altitude.
Even when comparing DTM and DSM, there was a reasonable difference, but the difference from the actual measured altitude is so big that it is difficult for me.

Is there something I haven’t considered?

Could you share the quality report?

Hi Kim,
Jaakko is correct. Can you share your quality report? This will shed light on what might be going on with your model. Without it we can not determine the problem. Thanks.