NOAO hosted a two-day workshop on 26-27 February 2001 concerning the analysis of images obtained with the Gemini North Hokupa'a adaptive optics camera. The primary goal of the meeting was to share understanding of the diverse issues, problems, tactics, algorithms, and so on, required to produce research-quality results from Hokupa'a images. Approximately 40 people attended, with heavy representation from the NOAO staff, the Gemini Observatory, and the NSF's Center for Adaptive Optics. A portion of the attendees included those who had worked with other adaptive optics systems besides Hokupa'a, itself; most of the problems faced by observers using Hokupa'a are generic to natural guide-star adaptive optics systems. All but a few of the talks have been converted to web documents that can be found on the presentations page.
Briefly, Hokupa'a is producing cutting-edge results in a diverse set of research areas. The ease of extracting quantitative results varies widely from problem to problem, and successful observing with Hokupa'a requires a sophisticated understanding of its limitations. Robert Blum (NOAO), for example, is conducting crowded-field stellar photometry on the Gemini Galactic Center data set, and obtains significantly smaller scatter in an H-K CMD than that shown by HST NICMOS observations of areas in common. Tim Davidge (HIA) and collaborators at Gemini have recovered the AGB tip in highly crowded Hokupa'a images of M32. Joe Jensen (Gemini), in some sense observing "ultra" crowded fields, has obtained reliable SBF observations of nearby galaxies with Hokupa'a.
The most difficult observations with Hokupa'a are those that require accurate knowledge of the PSF. François Rigaut (Gemini) presented a sobering assessment of the form of the Hokupa'a PSF and its variability with time and detector position. In short:
The PSF resolution and Strehl ratio decrease steadily with angular distance from the AO guide star.
The PSF is highly time variable-10% variations in FWHM are likely to occur on time scales longer than a few minutes.
PSF Strehl is a function of the AO guide star apparent magnitude.
The PSF has a broad halo.
At high-contrast ratios, the PSF halo is highly non-uniform. Quasi-stationary "speckles," or hot spots, occur in the halo that only change over a time scale of several minutes or with selection of different stars.
Observations that are particularly challenging are detection of faint companions to the AO guide star and the detection of faint host galaxies associated with QSOs. For observations of these sorts there is no simple path to success. A key minimal requirement does appear to be repeated observations over a span of time and conditions to discriminate between artifacts and real source structure, as well as obtaining PSF calibrations over a span of time with attention to characterizing the range of PSF variability and care to matching PSF observations to the properties of the source AO guide star.
Many of the talks presented at the workshop, in fact, focused on characterization of the PSF, given its central role in the analysis of Hokupa'a observations. Mark Chun (Gemini), Eric Steinbring (CfAO), and Tod Lauer (NOAO) emphasized programs to characterize the field variability of the PSF. Joe Jensen (Gemini) and Tod Lauer also explored recovery of the PSF from the SBF power spectrum, itself. In short, it appears possible to characterize PSF spatial variations in crowded fields from a relatively sparse sampling of the PSFs over the field. The most challenging observations are those where only the AO guide star PSF is available, or the guide star itself is the target.
Clever image analysis algorithms are also central to success with AO observations. Keith Hege (Steward) and Julian Christou (CfAO), in particular, discussed the method of "blind deconvolution," which is particularly useful in an AO context because it is explicitly designed to leverage multiple observations obtained under varying conditions to separate the PSF and source structure.
In conclusion, the workshop proved to be a successful sharing of AO experience, observing, analysis, calibration, and reduction issues. A quick skimming of the Web pages of the presentations should serve as an excellent point of departure for anyone interested in the problems of analyzing AO imagery.