You may have experienced this before. You are defining a new station and all points with known coordinates are more or less in the same direction. Be careful in this situation.
See following example:
http://geomtours.blogspot.com.au/2014/02/bad-resection.html
Forgetting point 3 for the moment - the azimuth error between points 2 and 4 will be transferred to the derived co-ordinate of point 5. (i have assumed the three points are 2 to 4, read from left to right.
So 10mm out in opposite directions on the co-ordinates of points 2 and 4 will give approx 60mm azimuth error at 5. (distance 2 to 5 is approx 3 times the distance 2 to 4)
The point I want to make though is that the result from a free-station (angles and distances) will be no better or worse than if you traversed from 2 and 4. The addition of point 3 in a free station should hopefully make it give a better solution.
The fundamental error in the given example is "coming from a short to a long". It's not much more complicated than that as far as I'm aware.
The effect of geometry on an angles only resection is great but this is often (incorrectly) repeated and said to affect a free-station (angles and distances).
Free stationing rocks. So does strength of figure.
This is a case of extrapolation vs interpolation. Interpolation is always going to be more accurate than extrapolation because errors are scaled down in an interpolated scenario, whereas errors are scaled up in an extrapolated scenario.
I'm glad you've brought up resections though as I believe this topic is not discussed enough. We use resections a great deal. Resections are incredibly precise and in some circumstances are better than direct observation. Many people have poor opinions of resections that I believe are unfounded, so long as care is taken to obtain good results. Geometry is one of those important caveats.
My Father never talked to me about the facts of resections when I was young. He covered another subject, but not resections :-(.
He also didn't cover the iPad helpfully changing things for me either.
>
> The effect of geometry on an angles only resection is great but this is often (incorrectly) repeated and said to affect a free-station (angles and distances).
This has been proven time and time again, yet people keep regurgitating the same dogma. In a free-station geometry means nothing.
In fact, the closer you are to co-linearity in a 2 point free station the better.
I realize this guy is trying to pimp his app by stirring up discussion. But I'm not biting.
You are referring to my traverse blog http://geomtours.blogspot.com.au/2014/02/analyse-traverse-example.html
I agree with you. There are a lot of different rules which you can apply. For example your fundamental error "coming from a short to a long". All of them are very good. What happens when you have combinations or not very clear situation or you simply don’t know all rules? Just calculate it with my android tool. It is very quick. An uncertainty of known point (accuracy, standard deviation) is included in my calculation (least square adjustment) . They are not error free! Sometimes this can affect your results.