ok so this has been a bad week. did a topographic survey for a small construction site, maybe 8 acres, and the client needs it for grading calculations. i placed 6 GCPs using my emlid reach RS2 and thought i had good accuracy on them, base was running for like 45 minutes before i started collecting points. flew the mission, processed in pix4d, and the vertical accuracy on the check points is off by like 12-15cm which is way outside what they need for this.

went back and looked at my logs and i think the base might not have converged properly before i started logging the GCP coordinates, or maybe i screwed up the antenna height on one of them. i dont really know where to start with this. re-flying isnt a huge deal since its a small site but id rather figure out what actually went wrong first so i dont just repeat the mistake. anyone dealt with emlid accuracy issues like this or have a checklist for diagnosing bad GCP data

antenna height is the one that gets everyone at least once. even being off by 2cm on the ARP to ground measurement compounds across your whole control network. the RS2 has that little rubber bump on the bottom and people forget to account for it or measure to the wrong reference point on the pole. check your Emlid docs for the exact ARP offset for your setup because its not always obvious.

also 45 min base convergence is borderline depending on your baseline length and satellite conditions that day. id really want to see at least an hour for anything beyond like 10km baseline, and if you were doing RTK rather than PPK on the GCPs theres always the question of whether you had enough fix quality at the moment you logged each point. PPK processing the base and rover logs after the fact with RINEX from a nearby CORS station will tell you a lot more about what your actual positional accuracy was during collection.

had something similiar happen to me last year, turned out i had the coordinate system set wrong in pix4d when i imported the GCPs. like the points were in the right physical location but i had a geoid model mismatch between what the emlid was outputting and what pix4d was using for the vertical datum. vertical errors were all over the place until i sorted that out. might be worth double checking that before you assume the field collection was the problem