NavList:

Dave,

Thanks for an interesting puzzle. Here's my approach:

Given assumed position and assumed time, then for any pair of bodies it is possible to compute apparent lunar distance by reversing the steps in the usual cosine-rule lunar clearing. The necessary apparent altitudes can also be computed in reverse.

This gives three equations in three unknowns, which can be solved numerically -- of course not always, but in this case there were no problems. I used directly lat, lon, and time for the unknowns (assumed all three LDs were measured at the same time).

The "reverse lunar clearing" function I already had in my toolbox, and a suitable solver was found in the "Scipy" library: "fsolve", some form of modified Powell's method according to it's docs.

The solution:

38°43.4'N 077°01.6'W
time 3:27:22

But it seems to be very sensitive to the original lunar distances. To get some idea of this, I solved the problem again for +-1 arc second changes in each of the lunar distances. The results vary a lot: latitude from 38°28'N to 39°02'N, longitude from 075°41'W to 078°31'W, time from 3:24 to 3:30. And all of this is done with a simple refraction model (the same formula as in NA).

So solving the puzzle is doable in principle, but the results are probably not that useful...

Cheers,
Harri

----------------------------------------------------------------
NavList message boards and member settings: www.fer3.com/NavList
Members may optionally receive posts by email.
To cancel email delivery, send a message to NoMail[at]fer3.com
----------------------------------------------------------------