One of the topics we covered at a recent training event was using Variety to demonstrate a good solution. I have collected many thousands of points and found that a variety of 1.3 is very reliable. I'm always careful to never say that any technique can "never fail" or is "always correct", but we deal in terms of risk and production. For example, "How likely is a particular method or statistic to failure?". It's an important question for pushing RTK into difficult environments, and at our trainings we try to always setup courses for our field exercises that are in bad environments.
Point 3 was one such point. It was only 2 or 3 feet from the face of a large oak tree (probably more than 24" diameter) with several smaller trees around it and a fairly large cedar tree (about 10" in diameter) about 10 feet away. It was an ugly point for RTK. The red arrow in the Google Street view image roughly points to where the point was.
I located the point twice with the following screen captures:
The first observation met my required value for Variety and had an RTPK observation that agreed with RTK.
The second observation did not have agreement between RTK and RTPK, but the Variety was 2.06. I've been very pleased with Variety of 1.3 or higher, so I felt confident in the RTK observation without paying too much attention to the RTPK result. That was a mistake on my part. The RTPK had two observations that were in agreement. At 60 second processing interval, it is possible for RTPK to have two bad fixes that agree. (It is unlikely that two RTPK results will agree and be wrong with 2 minutes of data each, but at 1 minute each it is possible, even if very improbable).
So what is the take away? For me, it is that RTK with Variety and RTPK agreement remains a very good test for verification. Variety is still a very good test for verification, but nothing is guaranteed. Many attendees observed this point the next day when our training event started and Variety and RTPK were excellent at preventing bad fixes from being stored and were quicker than previous verification procedures.
I share this because I want our users to have the best tools available to them for making reliable measurements efficiently in bad environments. Will this cause me to change my routine? Probably not much. RTK with Variety and RTPK did not agree, so I had a warning sign. I will likely continue to use the profile with Variety of 0.75 Plus RTPK agreement for my difficult shots. If I choose to accept a point with RTK/Variety only or with RTPK/RTPK agreement only, and it does not have RTK/Variety plus RTPK agreement, I will likely test that point with a distance to last, two-point offset, to be certain that the fix was good.
One other thing not obvious from the screen captures, because I changed the coordinate system between the first observation and the second, is that the RTPK results of the second observation were in great agreement with the RTK/RTPK results from the first observation (H: 0.104', V: 0.012').
Point 3 was one such point. It was only 2 or 3 feet from the face of a large oak tree (probably more than 24" diameter) with several smaller trees around it and a fairly large cedar tree (about 10" in diameter) about 10 feet away. It was an ugly point for RTK. The red arrow in the Google Street view image roughly points to where the point was.
I located the point twice with the following screen captures:
The first observation met my required value for Variety and had an RTPK observation that agreed with RTK.
The second observation did not have agreement between RTK and RTPK, but the Variety was 2.06. I've been very pleased with Variety of 1.3 or higher, so I felt confident in the RTK observation without paying too much attention to the RTPK result. That was a mistake on my part. The RTPK had two observations that were in agreement. At 60 second processing interval, it is possible for RTPK to have two bad fixes that agree. (It is unlikely that two RTPK results will agree and be wrong with 2 minutes of data each, but at 1 minute each it is possible, even if very improbable).
So what is the take away? For me, it is that RTK with Variety and RTPK agreement remains a very good test for verification. Variety is still a very good test for verification, but nothing is guaranteed. Many attendees observed this point the next day when our training event started and Variety and RTPK were excellent at preventing bad fixes from being stored and were quicker than previous verification procedures.
I share this because I want our users to have the best tools available to them for making reliable measurements efficiently in bad environments. Will this cause me to change my routine? Probably not much. RTK with Variety and RTPK did not agree, so I had a warning sign. I will likely continue to use the profile with Variety of 0.75 Plus RTPK agreement for my difficult shots. If I choose to accept a point with RTK/Variety only or with RTPK/RTPK agreement only, and it does not have RTK/Variety plus RTPK agreement, I will likely test that point with a distance to last, two-point offset, to be certain that the fix was good.
One other thing not obvious from the screen captures, because I changed the coordinate system between the first observation and the second, is that the RTPK results of the second observation were in great agreement with the RTK/RTPK results from the first observation (H: 0.104', V: 0.012').