In the published paper, I believe I have identified an error in Eq 2.4.
In the denominator, where it says [f cos D – xL sinD cosTL – yL sinD sinTL], it should be [f cos D – xL sinD cosTL + yL sinD sinTL] (i.e. wrong sign for yL term). Since both D and TL are usually small, this in practice makes little difference anyway, but the error does seem to have been in the printed text not the code. I still believe the code used to generate the distributions, and made available on this website, was correct. Please let me know if you find otherwise.

23/2/2014