Occasionally we use an ATV with a Topcon hyper V to topo large open areas. Our QC procedures have caught a few shots out by up to 0.7 feet vertical. That's a few shots out of tens of thousands.
The other day the crew did a large flat pad. All of the check shots and redundant checks worked perfectly (for RTK). The QC numbers showed no appreciable fluctuations. Smack dab in the middle of this flat pad a 300 by 200 foot hump appeared in the data. Slightly southeast a 250 foot depression area appeared. Both are about 3 foot vertical.
One is a string of shots, the other is a group of shots from three different 'strings'. There are no obstructions and these are separated in time. The time between shots in the strings shows the ATV was traveling in a straight line at a consistent speed.
To add to the mix we have seen a severe spike in bad inits. Our procedures catch them, but I am losing sleep.
I am a QC fanatic. I like to find answers but this has me stumped...
Look at the constellation configuration in some planning software for that time, also check space weather for that day.
Auto topo is very rapid, on-the-fly, low quality. Latency will affect it big time, so radio interference is also a major component. This will just happen from time to time.
I only use it when I can do a checkered grid, gather one direction, check another. Having the separate grids usually makes bad shots stick out like a pyramid on the DTM.
Already looked at all the standard stuff. The only apparent commonality is location. One area all shots are high the other they are all low. Very strange...
I learned years ago to hesitate about one second before taking an RTK shot, more if it is a quality point. That one second makes a big difference in the accuracy.
Run your lines in PPK at the same time as you gather RTK data if your equipment allows. PPK removes the radio component / latency out of the equation. My guess is the radio component is causing your sporadic problems.
Start and end your survey with a 20 minutes static observation (equipment on quad, stationnary), both static baselines used in the PPK postprocessing.
A good method to check the quality of RTK data in continuous topo mode.
Why rely on RTK? Did you also record raw data? If yes, then post-process using your GPS software. There is really no advantage using RTK if you are going to use the data for topo processing. Depending on your software, set the base to "static" & the rover to "kinematic" & fire away. From the PP data you can then remove points with variations greater than 0.xx depending on your judgement.