A Strategy for Ground-Based Optical and Infrared Astronomy

As described in Section II, in "Current Resources for OIR Astronomy," the independent observatories control more than three-fourths of the major telescope assets available to U.S. astronomers, and this situation will prevail for the foreseeable future. Thanks to efforts by visionary astronomers and to the generosity of individuals, foundations, and state governments, U.S. astronomers have the capacity to carry out far more research in OIR astronomy than can be supported by NSF funds alone.

New technologies offer opportunities to increase the performance of all telescopes by huge factors at relatively modest cost compared to that of the telescopes themselves. The cost of these new instruments is not trivial, however. A major facility-class instrument, such as a multiobject spectrograph, can cost several million dollars. As adaptive optics technology becomes more mature, the panel foresees a widespread demand to implement this technology to improve the performance of many major telescopes.

Many independent observatories lack the financial resources to equip their telescopes with instrumentation that will enable the telescopes to perform at their full potential. In many instances, NSF investment in instrumentation for independent observatories will be the most cost-effective way to achieve specific goals of OIR astronomy. A modest increment in NSF's astronomy instrumentation budget is reasonable given the $300 M of state and private capitalization for the new large telescopes.

To estimate the net cost of providing modern instrumentation for telescopes at the independent observatories, one can assume conservatively that every such telescope listed in Table 1 should be equipped with one new facility-class instrument every five years, and that the average cost per instrument will be $2 M for telescopes of aperture 2 to 5 meters and $5 M for telescopes of aperture greater than 5 meters. (This estimate is consistent with one made by a group of observatory directors at a recent meeting.) The calculation yields a net funding rate of $12.4 M/year. Assuming that the independent observatories share roughly 30% of the costs on average, a very strong scientific case exists for NSF to support the development of such instruments at a level of about $9 M per year. Such a funding level would vastly increase the scientific productivity of the nation's telescopes.

In a constrained funding environment, it is unrealistic for the NSF Division of Astronomical Sciences to provide such a funding level, especially in view of the need for funding Gemini operations and modernizing the telescopes at NOAO, and for NOAO to provide a broad spectrum of observing options to the nation's astronomers. Indeed, the panel cannot realistically expect NOAO to meet these demands in any case. As has been discussed, to maintain scientific leadership within a constrained budget, NOAO must narrow its focus to those activities it can do best. If it does so, the panel must then ask: Is there another way to provide some of the observing options that NOAO must curtail?

For the above reasons, the panel recommends that the NSF Division of Astronomical Sciences establish a new program to provide instruments at independent observatories that agree to provide national peer-reviewed access to their facilities in proportion to the funds provided.

The proposed facility (including possibly an instrument, a mirror, and/or a telescope) must leverage substantial nonfederal investment, which may be in the form of existing telescopes built with nonfederal funds and/or cost sharing with nonfederal funds.

NSF funds must be used only to provide capital equipment that will directly augment the scientific performance of the telescope. The panel does not recommend that NSF provide funds for operations or maintenance of independent observatories. That would only create a dangerous incentive for independent observatories to begin counting on the NSF to make up for inadequate fiscal planning.

This program should be distinguished from the ATI program. In such a program, it is often impossible to predict that a given effort will yield a working device, which would probably not be suitable for general use in any case. In contrast, the instruments to be funded under the program the panel recommends should have a reasonable expectation of providing important and reliable observational capability, based on prior successful experience with similar instruments. Of course, there is a continuum between ATI and the development of facility instruments. It would be inappropriate to rule out some level of innovation and risk in the latter. Therefore, ongoing judgments will be required to determine whether the proposed facility instrument meets the "reasonable expectation" criterion. A mechanism to make such judgments is suggested in the subsection "Review of Proposals for Instrument Development" below.

This program should be regarded as experimental, and its growth or termination should depend on scientific performance. Appropriate indicators of performance are (1) the quality of the science produced, by astronomers at the host institution and by external users, as a result of the program; (2) the number and quality of proposals to build new instruments; and (3) the intensity and quality of the competition for national access to the facilities.

The panel believes that an appropriate level of NSF support for this program is about $7 M/year. In fact, the NSF already supports the development of OIR instrumentation through its grants program, at a current level of about $7 M/year (Section II). Most (about 75%) of the NSF funding for OIR instrumentation has been devoted to the development of advanced technologies, such as adaptive optics and interferometry. Funding of these activities was highly recommended by the AASC report, and this panel recommends that NSF continue to fund such programs aggressively with no strings attached. However, some (about 25%) of the ATI funding of OIR astronomy has been used to build facility instruments and telescopes at independent observatories. The panel recommends that this fraction, about $2 M/year in 1993, be removed from the ATI program and augmented by approximately $5 M/year of new funds in the NSF Division of Astronomical Sciences budget to meet the recommended funding level of the new facility instrumentation program.

The panel suggests that NSF implement this program immediately, beginning with a portion of existing funding in the present ATI program and augmenting the program as rapidly as the availability of new funds permits. Since the scheme for national access is untried, we need to gain some experience to know whether it will in fact deliver excellent science at low cost. If the program can provide a broad and growing range of observing options to all astronomers through its provision for national access, the need for NOAO to provide such a range of options on its own facilities will diminish. This scheme might create an environment in which all observatories can realize cost savings by specializing their facilities. The need for immediacy arises from the fact that there is a window of a few years before NSF must provide its full share of Gemini operations costs. At that time, NSF and NOAO may have to make hard choices regarding priorities for facilities. These choices might be more optimal if they could be made on the basis of some experience with the new instrumentation program.

The goals of the national access provision are (1) to ensure that the program yields the best science, (2) to provide national access to a broad range of observing facilities, and (3) to realize the cost savings that may accrue from efficient modes of operation of independent observatories. To achieve these goals, the panel proposes the following guidelines.

First, the conditions for national access must be flexible and responsive to the operating constraints of each participating observatory. Any provisions requiring substantial changes in operations will drive costs up and will be a deterrent for that observatory to participate in the program. Therefore, in the first instance the participating observatory should propose its own provisions for national access so as to minimize the impact on costs. In an optimum system, the possible modes of national access might vary widely from one observatory to another. For example, one observatory might elect to provide only "data on demand," through queue observing by its own staff. An observatory equipped for remote observing might provide that option. Another observatory might elect to provide hands-on training of students by its own staff, and another might support long-term projects by experienced astronomers. Of course, observatories could also choose to provide any combination of the services listed above (or others not listed).

A mechanism is needed to ensure that the aggregate of participating observatories will meet national needs for a variety of observing facilities and modes. It is important for each participating observatory to understand whether its provisions for national access are responsive to unfulfilled needs. Since NOAO already has responsibility to provide national access, and much experience in doing so, NOAO might undertake the responsibilities to provide this information and work with proposing observatories toward an optimal balance of options to the national community. The proposing observatories would be able to discuss their provisions for national access with NOAO before submitting their proposal, and perhaps modify these provisions to be more responsive to unmet needs as appropriate.

The primary goal of the program is to enable excellent science, for both the astronomers at independent observatories and those without access to their own facilities. Therefore, the fraction of telescope time provided for national access by participating observatories should be proportionate to the NSF funds provided for new instrumentation, as a fraction of the amortized capital cost of the nonfederal facilities. If the fraction were greater, that would remove the incentive for many of the best independent observatories to participate and the program would not yield the best science. If less, the program would not meet the national need for access to a broad spectrum of observing options.

The principle of proportionate access and how much national access time the program might deliver can be illustrated by two hypothetical examples. First, suppose that an independent observatory has built a modern 3.5- meter telescope, at a net capital cost of $15 M, and submits a proposal to NSF for funding to build an instrument costing $3 M. Suppose that the observatory wishes to discharge its obligation to provide national access over a period of six years. Assuming that the telescope value decreases exponentially with a mean life of 20 years, the net depreciation of the telescope during the first six years would be $3.9 M. Suppose further that the annual operating costs are $1.5 M. Then, the net cost to the observatory for the first six years would be $12.9 M. Then, a reasonable fraction of telescope time to provide for national access would be ($3 M)/($12.9 M) = 0.23, or about 85 nights per year for six years.

As a second example, suppose that the Keck Observatory submits a proposal to NSF for $4 M to support in part the construction of a new instrument and wishes to discharge its obligation for national access in four years. Then, assuming a capital investment of $80 M, a mean life of 20 years, and operating costs of $6 M/year, a similar calculation yields 38 nights per year of national access to Keck for four years. However, in this case a further correction is warranted because the OIR astronomers at the California Institute of Technology and the University of California system already represent a significant fraction (about 15%) of the active OIR astronomers nationwide. Since it would be awkward for these astronomers to apply for national access time on their own telescope, it would be appropriate to reduce the national access time by a factor of 0.85, giving a final result of 32 nights per year for four years.

Additional examples of sharing are the Sloan Digital Sky Survey and 2MASS. These are projects currently being undertaken by university consortia to produce large-scale photometric and spectroscopic surveys using special-purpose telescopes and instruments. In these examples, the national benefit is open access to extraordinarily powerful and unique databases. Such arrangements would be attractive to the owners of the telescopes if they were to individually reap more, not less, high-quality data by participating in such a program. That may be true in many cases, because modern instruments with wide fields can often provide major gains in telescope efficiency. For such a scheme to be acceptable to the community, it is also essential that new funding, not repackaged funding, be used to initiate this program.

An instrumentation program funded steadily at the recommended level of $7 M/year would provide for national access the equivalent of 85 nights per year of Keck time plus two modern 3.5-meter telescopes full-time. (In fact, the aggregate program might yield a richer mix of observing options.) This program could significantly alleviate the current shortage of access time to well-equipped telescopes. The national access time provided would not be the only scientific benefit of the program, however. Additional scientific benefit would result from the increased observing power that would accrue to the independent observatories.

These hypothetical examples are intended not to serve as specific guidelines, but rather to illustrate the principles by which a reasonable amount of national access might be calculated. The program will probably work best if participating observatories are free to propose any provisions that they see fit. For example, an observatory may wish to propose a mix of observing options on a variety of its telescopes. The panel believes that the review process would provide sufficient incentive for participating observatories to offer a reasonable amount of national access time on their facilities.

Since each national access arrangement would carry administrative and other costs, there would be a threshold instrument cost below which the benefits of national access are not sufficient to justify incurring these costs. This threshold might be in the range of $0.5 M to $1 M, depending on circumstances. A flexible mechanism for determining this threshold is suggested in the next subsection, "Review of Proposals for Instrument Development."

The proposed terms for national access, including a plan for user support, would then become part of the proposal to NSF for the new instrument. The proposal should then be judged on overall scientific merit, with criteria including (1) the value of the science enabled for both the host observatory and national users, (2) the scientific leverage provided by nonfederal cost sharing, and (3) the extent to which the proposed instrument meets an unfulfilled scientific requirement.

Finally, the panel recommends that the national access time provided by the participating independent observatories be distributed through a national time allocation committee (TAC). Of course, before the TAC meets, the independent observatories should screen proposals to use their facilities, just as NOAO does. The TAC will need to know whether the proposals are suitable for that facility and the reasons, technical and otherwise. A national TAC would have the following advantages: (1) it would ensure that the national access time is granted on the basis of scientific merit alone, as determined by competitive peer review; (2) astronomers could propose to a single agency, according to a standard format; (3) if a proposal were found to be scientifically excellent but unsuitable for a given observatory, the TAC could attempt to identify an alternative facility; and (4) a single TAC would probably be the most efficient procedure.

The proposed program would have at least one significant new advantage for the science that could be carried out: it would greatly simplify and streamline programs of coordinated or synoptic observations. Many such programs arise in OIR astronomy, from studies of time-variable phenomena and periodicity searches, and in particular observations coordinated with spaceborne observatories (such as HST, ROSAT) for which increased longitude coverage is often crucial. A national TAC could consider proposals for near-simultaneous or sequential use of several large telescopes that would otherwise be unlikely to be scheduled separately for a single program. This qualitatively new observing capability might also justify modest NASA support for such programs, or the instruments to carry them out, as suggested in Section VI.

The panel also recommends that proposals to NSF for grants for instrumentation development (both facility instruments and advanced technology instruments) should be selected on the basis of an annual review by an NSF jury committee. The panel believes that such a process, details of which are described below, would be an effective mechanism for optimizing the scientific benefits of NSF funding of instrument development. The jury committee would review and seek to coordinate instrumentation plans at NOAO, the independent observatories, and the collaborations of outsiders with NOAO.

There are two major advantages to a jury review. First, it solves a problem raised by a number of correspondents--namely, that they cannot make informed judgments when reviewing a proposal for instrumentation because they do not have a clear understanding of the global context. A given program may be fine technically, but it is difficult to assess whether it is the scientifically most valuable one relative to other alternatives. Second, and perhaps most important, a jury review provides a powerful educational forum for all participants, which could accelerate technology development and encourage cooperation where appropriate. Indeed, the jury committee should search for economies of scale and opportunities to avoid duplication of effort, especially in instrument subsystems (e.g., controllers, detector arrays). Perhaps the greatest benefit of such a review committee would come not in the judgment of the proposals at hand, but in the guidance provided for future instrument development.

In addition to determining the scientific merits of the proposed instruments, the jury committee might be able to advise NSF whether a given instrument proposal should be regarded as an ATI program item or a facility-class instrument. If the latter, the committee must further decide whether the proposed instrument meets the cost threshold for national access, and whether the national access provisions are equitable, according to the principles described in "Guidelines for National Access" above.

The National Academies Home Page | NAP Home Page | Report Home Page