given an accurate, known point, is there any difference in accuracy between setting the base up on that point or setting the base elsewhere and base shifting off that point?
If you can’t set up the base at the known point, just place it nearby and do the base shift. It’ll give you accurate results. But the longer the radius from the calibration point is, the bigger the error can be. So placing the base directly above the known point is the best option.
would there ever be a need to use base shift when setting base up on a known point?
Usually, you don’t need a base shift when you use a known point to set up the base. The only rare case I can think of is if you need your base station to be in another datum. My colleague Svetlana described how the base shift could help here in this thread.
I often use base shift rather than leave my base unattended on the side of the road in the country, for security. (Where the local known point is.)
My process is to place the base on the site, and setup Emlid caster correction.
Then drive 3-4km to the nearest known point by the road and base shift to that known point.
Then drive back to the site and start the survey. This process has worked well for me in theory.
But now I’m worried base shift will have a dramatic error because of the 3-4km distance? Is this a concern?
What would the expected error be? Under 1cm or more?
A couple of km from the calibration point usually works well. But it’s pretty hard to estimate the error of the final solution without local tests. There are different factors involved, like the datum of the known point, the baseline, and local environmental conditions.