A. Quirrenbach et al., PHASE-REFERENCED VISIBILITY AVERAGING IN OPTICAL LONG-BASE-LINE INTERFEROMETRY, Astronomy and astrophysics, 286(3), 1994, pp. 1019-1027
In the photon-starved regime, the signal-to-noise ratio of interferome
tric data depends on the coherent integration time. To extend the inte
gration time beyond the limit imposed by atmospheric fluctuations, the
phases taken in a narrow ''signal'' channel can be corrected by refer
encing them to those taken in a wider ''tracking'' channel. A number o
f instrumental and atmospheric effects decorrelate the phases in the t
wo channels and thus constrain the range of conditions under which the
phase-referencing technique can be used. In the case of the MkIII ste
llar interferometer, differential refraction at intermediate to large
zenith angles is the most important limitation. Tests with MkIII data
in the photon-rich regime demonstrate that the phase-referencing techn
ique works well at moderate zenith angles. In the photon-starved regim
e, the expected improvement of the signal-to-noise ratio is readily ob
served. We use phase-referenced data taken on the bright star a Boo cl
ose to the first null of the visibility function to show that the MkII
I data are free from additive bias at the V2 less-than-or-equal-to 10(
-4) level. The absence of any bias larger than this value is an import
ant requirement for future imaging interferometers.