Introduction

A number of different approaches have been used to overcome the blurring effects of the Earth's turbulent atmosphere, and so obtain high resolution images from large ground-based telescopes. The most commonly used methods have been speckle interferometry (Labeyrie 1970) and Non-Redundant Masking (NRM; eg Wilson et al. 1992). Bedding et al. (1994) describe the MAPPIT interferometer at the Anglo-Australian Telescope (AAT), which has been used for NRM observations. At present a number of adaptive optics (AO) systems are being developed and brought into use. AO differs from interferometric methods in that wavefront distortions imposed by the atmosphere are sensed and removed in real time, allowing a high-resolution image to be formed on a long-exposure detector or to be placed on a spectrograph entrance aperture.

Results to date from speckle interferometry and NRM have concentrated on binary and multiple star separations and orbits, and measurements of stellar diameters and surface features for cool giants with discs resolvable by 4m telescopes. The restriction of current optical interferometric methods to relatively simple objects, showing at most a small number of non-zero resolution elements along each axis, stems from the nature of the available data. Because turbulence in the Earth's atmosphere produces decorrelation of phases over spatial scales larger than about 10 cm (the Fried length ), and over time scales longer than about 10 ms, it is not possible to measure directly the interference fringe visibilities and phases for the various baselines, as is done in a phase-stable radio synthesis telescope. Instead, methods equivalent to autocorrelation (or calculation of the power spectrum) are used to extract fringe visibility amplitudes, resulting in a response proportional to , where V is the true object fringe visibility. For a well-resolved (ie complex) object, many of the intermediate and long-baseline fringes will have visibilities of order 1%, and these become too weak to be measured satisfactorily in the presence of noise when the system response is proportional to .

Phase information is also required to construct a true image, and is normally extracted in the form of the closure phase, which is the argument of the complex bispectrum (eg Haniff 1989). In the case of speckle interferometry the use of the bispectrum is termed speckle masking (Weigelt et al. 1986). The bispectrum depends on the triple correlation around a triangle of baselines; thus it is proportional to (or closer to if one short baseline is included). While the resulting closure phase has the desirable property of independence from atmospherically-induced phase corruption, the dependence on results in the signal becoming difficult or impossible to detect in the presence of noise for low visibility baselines. For example, the surface features on red giant stars were difficult observations for interferometry even though the stellar discs were only resolved into about beamwidths (FWHM), and the target stars were bright.

Although they also work through the turbulent atmosphere, AO systems are not subject to the same limitations, because they incorporate a wavefront sensor. Within certain constraints, the wavefront data enables the imaging to be done in an effectively phase-stable manner. AO systems, however, are complex and expensive to set up. They generally aim for full diffraction-limited resolution only in the infrared, and/or on intermediate-sized telescopes, because working at longer wavelengths or with a smaller primary aperture reduces the number of sensors and actuators needed. It is possible to reach fainter objects in the infrared, thus extending the sky coverage (eg Beckers 1993).

© Copyright Astronomical Society of Australia 1997