Fixing a station's height only, or x,y only

[Olly Betts]

On Sun, Aug 15, 2021 at 07:01:22PM +0100, Tarquin Wilton-Jones wrote:
> On 15/08/2021 18:32, Martin Green wrote:
> > Do you have two legs between the benchmark and the circle?  Thus making
> > it hard to determine the vertical offset?
> 
> In my real case, I have 3. Because trigpoints are made from reinforced
> concrete, and the iron causes a Disto (or compass) to give wildly
> erroneous readings, so I needed to get quite some distance away, and
> didn't have a useful place to put a tripod where I could see everything.
> Had to do it in 3 hops.
> 
> (For anyone planning to do this in future, a plumb is a whole lot
> easier, as long as the top of the trigpoint is level.)
> 
> Like I said, that's OK in this specific case, I could add the vertical
> components together, get the top height, and then fix just the top. And
> then leave a comment saying why the top has the "wrong" height (not the
> one you see in OS records), and why the benchmark doesn't have one at all.

This is the solution I'd recommend.  It can be a bit fiddly, but has the
huge advantage of actually working.

> > You could try x, y, z fixes with rough estimates for the unknown axes.
> > Then specify zero standard error for the known axes and a large standard
> > error for the unknown axes.
> 
> Yes, I thought of this as a possibility. It made me feel like I needed
> to take a shower ;)
> 
> It works, but it has subtle effects when other fixes are involved
> elsewhere in the survey. It's certainly not something that is intuitive
> for anyone else maintaining the data.

I'd avoid relying on large standard deviations being effectively
ignored.  There's likely to be a small residual effect, and if
there's no other connection then the dummy value will quietly
get used.

> > This is not something I have done, and it is a shame that you need very
> > roughly estimate the parts of the fix you would rather leave blank.
> 
> Exactly that :)
> 
> You know how clino readings can be left blank with a - character. That
> would be lovely for the *fix command.

What that actually does is to set the reading to 0 and the standard
deviation to a value proportional to the tape reading (as the manual
documents), so it's really the second approach (but for a clino reading
it's less bad as the vertical change for a leg is bounded by the tape
measurement +/- tape error).

> Would that actually be possible to add?

It should be possible to allow height-only or no-height fixes (or legs
which only connect horizontally or only connect vertically) with the
current loop closure code, but I think it would be a lot of work.

Cheers,
    Olly