Draft - April 12, 2000
The high leverage of employing the world's largest telscopes with the largest interferometric baselines motivates the OHANA consortium to undertake a three phase project to prepare, demonstrate and implement optical interferometry among the Mauna Kea ohana of telescopes. In Phase 1, a number of technical tests and demonstrations will be carried out to show clearly the feasibility and cost of OHANA. In Phase 2, a demonstration of fiber-linked interferometry between two existing Mauna Kea telescopes will be prepared, and one or more science demonstration projects will be carried out. In Phase 3, an OHANA facility suitable for substantial science programs will be defined and implemented.
The concept embodied in OHANA has been discussed in various places for many years, most extensively in Mariotti et al. Over time, three major critical concerns have been recognized. These are: fiber availability and related dispersion and efficiency issues; a cost-effective and feasible delay line concept; and the need to coordinate facilities under completely independent ownership and scientific management. The first two of these are technical problems, for which solutions are offered here. The third concern is addressed directly by the coordination and cooperation of four major Mauna Kea organizations in the OHANA project.
The following pages present the scientific and technical cases for OHANA,
Phases 1 and 2, and the strategy for studying and preparing the case for
Phase 3. The technical issues are addressed in sufficient detail
to show that cost-effective solutions, already in hand, ensure achievement
of OHANA's scientific potential.
Optical interferometry has advanced well beyond the prototype stage, with array projects like IOTA, NPOI and PTI churning out both technical demonstrations and scientific results. The construction of the first generation of observatory user facilities (some derived from former prototype arrays) is well under way. GI2T, ISI, CHARA, Keck and VLTI are bringing on-line a growing observational capability, with more and larger telescopes, larger selection of baselines, and increasingly sensitive and capable instruments.
The future of optical interferometry can be reasonably foreseen to evolve in the direction of more capable arrays in the future, some general purpose and some specialized. Some will develop from current facilities, and some will be undertaken as new initiatives.
Current programs will execute increasingly varied and sophisticated observations. Initially, these will be almost exclusively observations of stars and circumstellar phenomena, but increasingly, optical interferometry will be extended to extragalactic targets. The extension to increasingly faint and complex sources will push the capability of existing arrays to their limit, and steadily improve our understanding of the motivation and requirements for next-generation array facilities.
The Keck and VLTI will provide a critical capability with large aperture telescopes operating on baselines up to about 200 meters. It is likely, however, that this baseline will be insufficient to even partially resolve many important types of sources. Examples are accretion disks around interacting binary stars, young stellar objects, and AGN's.
The existing selection of telescopes on Mauna Kea offers a unique scientific
and technical opportunity, with multiple telescopes of 3.5 to 10 meter
aperture, and a variety of baselines from 50 to 1000 meters. This
selection of telescopes and baselines could not reasonably be proposed
as a dedicated interferometer test facility, did it not exist already.
But given its existence, is is possible to exploit it for its unique capabilities,
not currently planned for any other interferometer.
Models for the interior structure and physics of Seyfert nuclei are varied, but they share several features. There is a central continuum source (possibly an accretion disk), of typical dimension less than
0.1 pc. There is a surrounding, clumpy broad line region (BLR) up to about 1 pc in size. On a still larger scale, there is an obscuring region usually called the dusty torus, which may in fact be an extension of the accretion disk structure. According to the unified AGN model, viewing geometry is a major factor in the appearance and empirical classification of AGNs. For type 1 Seyfert galaxies, the continuum region is viewed from high latitude above the torus, with access to direct emission from the BLR. In cases with no evidence for obscuration by the torus, the source may be described as a ``bare AGN''. Some models require distributed or multiple component continuum sources (Blandford, 1990) which might be resolved directly by OHANA.. The central regions of QSOs are understood to resemble Seyfert nuclei, but on a larger, more powerful scale. The BLR in QSOs is believed to have a dimension of order 10 pc. At the distance of nearby QSOs, OHANA should resolve the BLR.
Accretion in the protostellar and early YSO stages may be inaccessible to direct optical observations due to high extinction in the surrounding molecular clouds, and perhaps in the disk itself. However,
it may be possible to catch accretion disks under special circumstances, after the extinction has been reduced by evolution or transient processes. The YSO FU Ori is visible through negligible
extinction, and may have such an accretion disk (Bell, 1994). The OHANA angular resolution will be approximately equal to the stellar diameter. At this time it is not clear to what extent details of the
structure and kinematics of an optically thick accretion disk may in some cases be discerned from an out-of-plane angle. Models (Bouvier, 1994) clearly show that high angular resolution will be the key to
distinguishing centrally localized viscous heating and reprocessing emissions (inside 1-10 AU) from the broadly diffused scattering fluxes (100 AU). OHANA resolution of 0.04 AU (< R*) is well suited for studying the innermost accretion zone.
The many types of interacting binaries offer almost unlimited opportunities for interferometric study. The most important to astronomy are probably those which demonstrate processes of wide generality. For example bipolar outflow and/or jets are observed in red giants, protoplanetary nebulae and planetary nebulae, symbiotic stars, cataclysmic variables and novae (Cohen, 1986). Cataclysmic variables have broad, single peaked emission lines similar to those of AGNs (Chiang, 1997) and the study of wind driving mechanisms in CVs may advance the understanding of winds in AGNs. For example, the prototypical CV U Gem has Roche lobe, disk and orbit diameters on the order of the solar radius (Cherepachchuk, 1996). At its distance of about 100 pc, OHANA could resolve the binary orbit, and partially resolve the diameters of the components mentioned, thus allowing model dependent measurement of the systems physical dimensions. (Baselines much longer than 1000 meters would be required to resolve accretion flow details.) The binary SS433 is well known for extreme kinematic and energetic behavior associated with precessing jets which are believed to originate in a relativistic accretion disk. This perhaps unique galactic object offers a rare view of physical processes in a relatively nearby, highly observable, relativistic source. VLBI observations (Vermeulen, 1993) reach to the 10 milliarcsec (50 AU) level, which is insufficient to resolve the Roche lobe, the accretion disk, or the binary separation. According to both neutron star and black hole models (Fukue, 1992) all of these should be detected and partially resolved with OHANA.
|Number of 3-10 meter apertures||2||4||6|
|Number of 1.8 meter apertures||4||3|
|Number of baselines||15||Moveable ATs||15|
|Maximum large-aperture baseline||75 m||130 m||800 m|
With respect to a dedicated very large optical array, the Keck and VLTI have several significant short-comings. These are: the small number of telescopes and the limited snapshot imaging capability; the small maximim baselines; and the limited access to the large aperture telescopes due to competition for general astronomy access by large user communities. In the case of the Keck, the expected NASA priority of interferometer use for focussed mission science is also an important limitation.
OHANA will address one of these limitations - the limitation to short baselines. OHANA will offer baselines up to 4X the VLTI maximum, and up to 6X the longest Keck baseline. As described above, this resolution range extends interferometry into an important parameter space for astrophysical studies in both galactic and extragalactic science.
OHANA will provide a unique augmentation to both Keck and VLTI interferometric operation. By offering a selection of larger baselines, with the highest senstivity, OHANA will bear a relationship to these arrays analagous to that of the early ad hoc VLBI to the VLA, by augmenting the imaging capability with partial coverage of higher spatial frequencies. The requirement for coordinated allocation of major telescopes ensures that OHANA will only be utilized for the highest priority, most unique science programs. The participation of astronomers from both the Keck and the VLTI communities ensures close communication on OHANA issues of mutual interest.
Concurrently with actual science operation, OHANA will prepare the ground
for planning of future interferometric arrays. Reconnaissance in
key science araeas, such as those mentioned above, is needed to determine
characteristic dimensions, complexity, and variability, in order to adequately
define the science driven requirements for a very large optical array.
Preliminary visibility measurements will also provide critical information
pertaining to the phasing of an array on various types of sources - a problem
which does not arise in radio interferometry.
The technical readiness for OHANA is good, and Phase 1 ane 2 can be undertaken immediately with full confidence of success. The feasibility of Phase 3 is also not in doubt, though a number of trade studies are still needed in the conceptual design stage.
The most demanding materials requirement for OHANA is the availability of suitable phase-preserving fibers. The most demanding instrumental requirement is the injection of light from an astronomical source into these fibers. As will be seen below, these problems have been resolved.
Optical fibers for interferometry have profited from more than a decade
of development. Silica fibers, drawn from a substantial communication industry
heritage, have been studied at the University
of Limoges, and the basic problems of injection and control have been
mastered. Fluoride fibers, which do not have a similar large commercial
market, have been developed in France within an industry-university collaboration,
and implemented successfully at the IOTA interferometer in Arizona.
Both types of fiber have been extensively characterized, and are now available
with material properties that permit them to be used for low-loss, phase-preserving
light transmission over hundreds of meters.
The operation of AO with fiber injection in not difficult to describe
and understand. Furthermore, on-telescope and on-sky tests of AO
fiber injection have been carried out at ESO. Nevertheless, this
is a critical, telescope-specific part of OHANA. Therefore, fiber
injection at each OHANA member telescope AO system will be part of the
technical demonstration program planned for Phase 1. On-sky measurements
will be made of injection efficiency and telescope + AO photometric noise
power spectrum. In phase 2, the fraction of coherent energy will
Test unit planned for Phase 1
OHANA requires a laboratory space of order 10 m in size or larger, in order to allow for the required optical delay. It would be unrealistic to construct such a facility on Mauna Kea for Phase 2 tests. Fortunately, a nearly ideal laboratory is available at the CFHT facility. The location is a coude area. Its floor is supported by the telescope pier. It is thermally and acoustically quiet. The CFHT is not centrally located with respect to the other telescopes, but with the low fiber losses, there is no significant penalty for the location. Further, the location is almost ideally placed with respect to the existing network of underground utility conduits which would be the natural path for OHANA fibers.
Infrastructure improvements needed
Optical fiber offers great simplification in transporting light from a telescope to a fixed instrument, as it eliminates the requirement for mechanically and optically complex coude systems. However, in the case of single mode fibers, the polarization characteristics can be changed by bending and twisting stresses, thus the routing of the fiber from telescope to pier must be planned with care. Existing cable wraps may be satisfactory, but must be examined and possibly tested in each case. Hence cable-wrap tests are part of the OHANA Phase 1 program.
CFHT - from telescope to pier to lab Gemini - from telescope to pier out of the building Keck - from telescope to pier out of the building
Cable-ways - ownership, access, physical conditions
Into the beam combination lab
An alternative to stretching is the use of temperature control to stabilize fiber lengths. This has been demonstrated in silica fibers (ref) with techniques which could be extended to fluoride fibers.
However, a full implementation of fiber delay matching can await Phase
3. For Phase 2 tests, limited fiber lengths and passive compensation will
A technical report with additional details about physical configuration and phase delay control is available (Ridgway, 2000).
A full segmented vacuum delay would only be required in Phase 3.
For Phase 2, a small number of fixed air segments could be employed, with
limited sky coverage.
For Phase 3, a dedicated OHANA implementation would be expected. For Phase 2, there are several simpler possible appraoches, depending on resources. It may be possible to borrow moderate travel delay lines (eg, from the old MkIII).
Finally, it would be possible to dispense with continuously controlled
delay, and implement a step and hold operation, as in IOTA
and the early FLUOR, allowing
earth rotation to scan through the OPD.
Variable light travel speed in media, whether air or glass, introduces differential phase delay which varies with wavelength. As long as this is identical in the feeds from all telescopes, the zero OPD will still coincide for all wavelengths. If the optical paths are not matched in each medium, the interference pattern will be "chirped", with a wavelength dependent OPD. The most satisfactory solution is to match the OPD in each medium, hence the plan, above, for: matched fiber lengths, including active tuning of the lengths for balance; matched air paths; and matched vacuum paths.
However, for various reasons, the ideal of equal material paths may not be achieved. Additional dispersion compensation can be achieved by introducing a variable thickness of a glass, chosen for its dispersion characteristics. This is a standard technique in optical interferometry, and is currently in use at NPOI and SUSI. However, under some circumstances it may be dispensed with. With increasing spectral resolution, and decreased bandwidth per detector pixel, the tolerance to dispersion increases approximately as the resolution. The dispersion effect is systematic, and modest dispersion can be removed numerically, provided the loss of fringe modulation is not so severe as to reduce detectability of the source.
For Phase 2, no automatic or continuously updated dispersion compensation
will be provided. for Phase 3, the requirement will be determined
based on experience in Phase 2 and at other interferometers.
Even in an ideal, homogeneous fiber, polarization rotation and mixing
may occur due to inhomogeneous stresses of curvature and twist. The polarization
characteristics specific to silica and fluoride fibers are discussed above.
With moderate fiber lengths, the polarization axes may be rotated by mechanically
adjusting the fiber, and mixing can be suppressed (with some loss of efficiency)
with polarizing filters. Laboratory tests are required to determine
the adequacy of these techniques for the long fiber segments needed for
Alternatively, polarization preserving fibers may be employed. These are more expensive. In order to avoid 50% loss of light, it would be necessary to introduce polarization splitting and separate light channels, which would add complexity and cost.
Thus there are several possible solutions of increasing performance
and cost. For Phase 2, standard non-polarization perserving fibers
should be adequate, perhaps with polarization filtering and some sensitivity
loss. The polarization changes within fibers constrained to run through
existing cable-wraps and conduits will be studied. For Phase 3, an approach
will be designed based on experience with the required long fiber segments.
In addition, the relative positions of the telescope vertices will be needed, in three dimensions. Due to the varying construction dates, it may be difficult to confirm consistency to a common grid. In that case, an excellent survey technique for nearby facilities on an irregular terrain is the use of differential satellite ranging, which provides measurements to the few mm level.
Finally, the optical paths in the laboratory must be determined, including the path through the delay line. This can be measured by conventional means, or readily with a laser ranging meter, to a few mm. Considering the various error sources, it should be easy to place the optical delay withing 25 cm of the ZPD.
With AO stabilized starlight, we can be confident that the ZPD fringes
will be detected in a single scan of the optical delay, of probable length
no greater than 50 cm. With a typical 2 micron fringe rate of 500
Hz, this will require no more than 10 minutes. This will determine
the "magic constant", reducing the uncertainty in ZPD position to a small
range. A few additional fringe detections around the sky will quickly
determine the baseline vector in three dimensions. An existing software
package (Wallace, 2000) can be used to determine any telescope dependence
of the effective baseline calculation. This process will be repeated for
each telescope added to the array, but should not be needed for every possible
baseline provided by the new telescope.
The OHANA team has extensive experience in fiber and integrated optics,
in which pupil and image plane are not distinguished. The beam combination
means already well established for single-mode interferometry are well
matched to the OHANA requirements.
One option, already developed and demonstrated for the two-telescope case, will be a FLUOR-style fiber beam combiner, utilizing three X-couplers for two-telescope interferometry with photometric calibration (ref).
Multiwavelength operation (???)
With single fibers and multiple fibers.
Several OPD scanning techniques are available. The simplest, employed at the IRMA interferometer and early FLUOR, consists of a step and pause delay line, with earth rotation scanned OPD. This is a technique of limited flexibiilty and low efficiency, but allows the use of a very inexpensive continuous delay line.
A more satisfactory technique is the continuous tracking delay line
with sawtooth scanning of the OPD around the ZPD. This scheme is
used currently at, for example, IOTA FLUOR, CHARA. It may be generalized
to encode several baselines simultaneously, as at COAST. This scheme
requires a continuous, accurate and smoothly tracking delay line, whose
availability is well established as described earlier.
Optical interferometers natural profit from the most sensitive available detectors. Today, the most sensitive infrared detectors are array detectors with 105 or more pixels. Ironically, an interferometer typically needs only a few or a few tens of these pixels. In many detector types, however, this allows the rapid readout (hundreds of Hertz) of the pixels utilized.
There is no reason for the OHANA detector to be highly integrated into the beam combination scheme, and in fact it is natural for the beam combiner to form an image outside the combiner envelope where it can be detected with an independent camera. It is thus reasonable to consider part-time utilization of a detector from another instrument - perhaps one which would be excluded from operation while OHANA was in use.
OHANA would also be an opportunity to exploit a developmental detector system in a detector-friendly, laboratory environment. Discussions are underway with D. Hall of the IFA, concerning new detectors currently under development at Rockwell, with very low (few e- noise) in the near-IR. These detectors are optimized for wavefront sensing, and would be well-suited for OHANA as fringe-tracking and/or science detectors.
For use from the visible to approximately 1 micron, CCD detectors can
be employed as described above for infrared detectors. An alternative
is the use of low noise APD's, which offer a well-developed, single-pixel
technology in a convenient instrumental package.
In Phase 2, OHANA will be phased by reference to the source itself, hence limited by the number of photons in a coherence volume (ref). In Phase 3, dual beam capability for offset phasing may be added, allowing interferometric observation of much fainter sources, but with reduced sky coverage.
In an initial, single beam implementation, the limiting sensitivity of OHANA will be determined exactly as is the limiting sensitivity of most existing optical interferometers - this is by the requirement that the number of detected photons in a coherence volume should give a sufficient signal-to-noise to provide the feedback information needed to keep the OPD within the coherence envelope.
The OHANA sensitivity computation may be conveniently separated into two parts. The first is the delivered Strehl for the telescope-AO system, and the second is the efficiency for the telescope-AO-fiber-interferometry system.
The AO performance is estimated based on an assumed Strehl at 2.2 microns of S = 0.5. This performance is then scaled to other wavelengths using the Marechal approximation, which is a good predictor for coherent energy, which deviates from the Strehl ratio for small values.
Two approaches have been taken to estimating the telescope-AO-fiber-interferometry efficiency. First, an empirical approach has been used, based on the measured sensitivity of the IOTA FLUOR interferometer, which is very similar in functionality to OHANA Phase 2. A model has been developed for the IOTA sensitivity, including explicitly the atmosphere and telescope dependence, and adjusting an empirical "efficiency" factor until the model agrees with the FLUOR experience. In fact, this gave an "efficiency" for FLUOR of about 4%. The parameters of the model were then adjusted to the OHANA case, but with an additional 2X reduction in efficiency, in order ensure a conservative estimate.
Second, a bottoms-up efficiency budget was created for OHANA. This is based on ideal fiber injection efficiency and standard handbook efficiencies for clean coatings, hence should represent an upper limit to the efficiency estimate.
The actual performance is likely to be somewhere between the two estimates.
Throughput budget - good, average, poor seeing
(Following is the calculation presented in Waimea, pending replacement
with an update.)
A sensitivity model was first devised for FLUOR/IOTA, computing the number of detected photons per time constant per bandpass. The actual observing limits at K were used to adjust the efficiency so that the model was an accurate predictor of performance at K. (The derived efficiency was 0.04.) The assumed Strehl for IOTA at K was 1.0.
The appropriate parameters were then scaled to an 8 meter aperture. The collecting area was increased, scaling from 0.4 to 8 meters. The time constant was increased, by the ratio 8/0.4, to account for the slower evolution of the piston term and the longer allowed integration time. The Strehl factor for K was set to 0.6.
The efficiency was reduced by an additional factor of 2X, to 0.02.
This is believed to give a very conservative estimate of sensitivity. The IOTA/FLUOR beam combiner has a very low efficiency of 0.25, which can be improved when replaced, and the detectors described below have approximately 10X lower noise than the IOTA/FLUOR detectors.
For the thermal IR, the IOTA sensitivity limits were assumed to be L = -1 and M = -4. These were scaled by the aperture area (i.e. assuming that the thermal background on the detector is independent of aperture), and by the square root of the integration time (since the observation is background-limited).
The assumed coherent energy fraction:
The derived limiting OHANA sensitivities for 8m apertures are:
Note the interesting result that, although OHANA is conceived as a near-infrared array, it would appear to offer the most sensitive optical interferometer in the world, from V to M bands, if available today. Of course, other better optimized facilities should eventually take the lead in sensitivity at short and long wavelengths.
The calculation above presents estimates of the limiting magnitude for
OHANA observations, but this does not imply that observations of sources
at the limiting magnitude will necessarily have low signal-to-noise.
Due to the required short time constant in a coherence volume, continued
integration will very quickly build up signal-to-noise. For example,
if the S/N is 5 in 0.01 second (typical in FLUOR faint-source observations),
it would increase to 500 in 100 seconds. It is in this sense that
all interferometry employing on-source phasing is bright-source interferometry.
A typical, unreddened K0 star of magnitude K=0 will have an angular diameter of approximately 4.8 milliarcsec (Dyck et al, van Belle). At K=12, such a star will have an apparent diameter of 20 microarcsec, 15X smaller than the OHANA direct resolution limit at 1 micron. Thus there will be an abundance of potential reference sources. Care will be necessary to avoid binaries.
The initial institutional participation in the OHANA committee
Canada-France-Hawaii Telescope Corportation Department Spatial, Observatoire de Paris-Meudon Gemini Observatory Institute for Astronomy, University of Hawaii W.M. Keck Observatory National Optical Astronomy Observatories
The initial committee members are:
Olivier Lai, chairman (CFHT) Peter Wizinowich (Keck) Steve Ridgway (NOAO) Francois Rigaut (Gemini) Pierre Lena, Guy Perrin (DESPA) Francois Roddier (IfA)
P.O. Box 1597,
Kamuela, HI 96743,
tel (808) 885 7944
fax (808) 885 7288
The OHANA committee contact for information is:
The construction of a prototype beam extractor and its test on the sky to measure injection efficiency and stability with AO. The evaluation and measurement of routing, bending and twisting impacts on fiber behaviour for the various telescope configurations (Cassegrain, equatorial and alt-az, and Nasmyth on alt-az). Design of phase 2: science programs, detailed instrumental plan and costing.
Fiber layout from telescope to laboratory with suitable protection (about 300 meter fibers), Construction of delay lines, possibly of a simple type and limited range to begin with, Installation of beam combination, detection and control, Test on the sky, validation of performances and demonstration of significant science on a few objects.
Enhanced science capability, More telescope pairs, Extension toward shorter wavelengths, Fringe tracking, Imaging capability.
Dual beam extraction and faint objects science.
|Spring 2000||initiate PhD thesis on Phase 1-2;
coordinate Phase 1 funding;
|Summer 2000||preliminary site visits;
Phase 1 test unit design;
|Fall 2000||initial Mauna Kea on-telescope AO injection tests;
fiber wrap-up tests;
|Spring 2001||complete AO on-sky injection tests;
laboratory long-fiber tests;
conceptual design for Phase 2 instrumentation;
|Summer 2001||prepare proposals for Phase 2;|
|Fall 2001||submit Phase 2 proposals;|
|Spring 2002||initiate Phase 2 implementation;|
|Spring 2003||Phase 2 operational.|
As the planning is still at a very early stage, there are numerous "risk" items that are expected to be secured before the project gets very far underway, but these are listed here for completeness.
The cost of polarization-preserving fibers may be high. Interferometry with non-polarization preserving fibers in lengths of a few tens of meters, works well. In fact, it has been preferred because use of both polarizations has been achieved, maximizing efficiency. Laboratory tests can be carried out with longer fiber pairs, as soon as they are available, to investigate the possibility of using non-polarization preserving fibers for OHANA.
Starlight fiber injection efficiencies achieved in practice have been low. This is probably consistent with wavefront quality through the atmosphere with tilt correction only and without AO. Injection efficiency tests will be carried out at Keck, Gemini and CFHT in Phase 1. Phase stability will be confirmed in Phase 2.
The infrared detector is a potentially high-cost item if it is necessary
to procure a dedicated camera. During Phase 1, a commitment will
be sought to provide the IR detection capability by loan of a camera from
one of the participant organizations. As the planning is still at a very
early stage, there are numerous
"risk" items that are expected to be secured before the project gets very far underway, but these are listed here for completeness.
Telescope access may be problematic if subject to independent TAC processes. The OHANA consortium members will subscribe to a memorandum of understanding, which will commit the members to providing agreed telescope access to accomplish the minimum OHANA goals, subject to appropriate conditions of demonstrated readiness and performance.