I am using two rtk surveyors in rover and base station modes. My goal is to geolocate specific spots around a test site so cm error accuracy is very important. I have the surveyor connected to a Linux machine running GPSD to interpret the NMEA or ublox data. What I am finding is that while a rover is in 1 cm accuracy mode the horizontal error data reports arbitrary date ranging from ~ 10-20. From the GPSD docs it explains this data is calculated by the GPS and is in meters which is far from the cm accuracy for my needs.
My questions are: Am I interpreting this data correctly. and if so how can I get more accurate error measurements for lat, long, and alt from the surveyor.