I performed a localization in Emlid Flow w/ Survey on a rectangular piece of ground that covers about 100 acres and is about 3,700 feet long by 350 feet wide. Five years ago I set fifteen ground control points across the property and was able to relocate four of them last week for the initial localization. The control points are in a local coordinate system relative to a flat, level horizontal plane established using a total station. Those four points are located roughly toward one corner of the 100 acre rectangle. The localization using the four “Measured” points had good accuracy with very small errors (0.041’ horizontal and 0.039’ vertical). After applying the localization, I used Stakeout to locate and “Measure” another four points near the other end of the property. What I noticed was that, during Stakeout, I could precisely locate each stakeout point horizontally, but vertically there was increasing vertical error the farther I got from the four points used for localization (e.g. about 1.5’ vertical error at the stakeout point 2,000’ away from the localization area). I assume this is due to the Localization creating an inclined plane. However, the Localization Preview screen (in the Inclined Plane section) didn’t seem to indicate that the Localization contained any slope (Preview Results showed 0.0 ppm for both East and North). What am I missing here? When I located the four additional control points (using Stakeout) and “Measured” them to include them in the Localization, the sloping plane seemed to flatten out very nicely and I was able to go to any of the 15 control points (or hundreds of survey points from five years) with ago very little vertical difference from the old coordinates. Perhaps I don’t fully understand what’s going on in Localization, but it appears to me that Localization calculates and applies a (potentially sloping) plane based on whatever points are used in the Localization and, using only three or four Control/Measure pairs over a small area runs the risk of creating a sloping plane that could translate to very large vertical errors at points distant from the Measure group of points. Is this correct? Is there a setting in Localization that forces it to create flat, level, horizontal plane similar to what a total station will create? Thanks.
I dont really use Flow, but other survey software… but from what i understand, to avoid an inclined plane, ONLY hold to 1 vertical, not ALL of them when localizing. Typically you can still measure them all, evaluate your verticals and only use the BEST one to hold to, unchecking or not using the rest of the verticals with high residuals.
That makes way too much sense. Such a simple solution. I certainly over-thought the problem. Thank you!!
We tend to ALL do. 
I’ll have to look again… but I think there is a warning in Flow about selecting / holding to more than 1 vertical will cause a inclined plane? Cannot remember… and if not, probably should have this warning in red.
Using 4 points horizontal and vertical on one side of the site isn’t an accurate way to relocate points 3700’ on the other side of a project and to expect great accuracy recovery. If I did find them, I would take their stated accuracy in the software with a grain of salt.
As @timd1971 said, you are constraining your localization to a few points with supposed 0.04’ accuracy horizontal/vertical then transferring that accuracy 3700’ on the other side of the site. Imagine using a terrestrial traverse (which you used to begin with) to the other side of the site expecting the same error accuracy… it won’t happen. Was the original traverse net a closed loop and balanced or just an open traverse ?
Using your control marks as “fixed” for the rest of the site is not very wise. Use at least 3-4 points scattered throughout the site with holding 1 vertical component and then compare. You’re magnifying your found points error exponentially across the site.