The limitations to astrometric accuracy from ground interferometers for
application to planet detection are surprisingly small. One exploits the
narrow-angle nature of the problem to make the position measurement relative
to an angularly nearby reference. For references within 20 to 30 arcseconds of
the target and observations at K, cophasing can be used to increase sensitivity
in order to use very faint references. With 1.5 to 2.0 m telescopes at a good
site, near full-sky coverage is available. Accuracies at the tens of
microarcsecond level in an hour of integration time should be possible,
enabled by long interferometer baselines as well as the small star separation.
Development of interferometry at Keck Observatory has commenced, with the
initial phase to be the combination of the two 10-m telescopes. Among its
initial objectives is characterization of the exozodiacal environment of
nearby stars in preparation for future space missions for direct planet
detection. The proposed second phase would add outrigger telescopes to
enable synthesis imaging as well as narrow-angle astrometry for an
extensive planet search. Technology development for Keck has been ongoing
with the Palomar Testbed Interferometer. The interferometer uses a 110 m
baseline with 40 cm apertures at K, and achieved first fringes last year.
The interferometer architecture is similar to that proposed for Keck,
with active delay lines and fringe tracking, and a dual-star configuration.
Current engineering work is directed toward validating the narrow-angle
astrometry technique and investigating the achievable astrometric accuracy.
