Ground-Based Solar Research: An Assessment and Strategy for the Future (1998)

NOTE: The material in this appendix is reprinted from material distributed to the Task Group on Ground-based Solar Research on April 28, 1997, and later revised by the scientific staff of the National Solar Observatory.

Scientific Staff

National Solar Observatory

National Optical Astronomy Observatories*

July 25, 1997;

minor revisions June 19, 1998

This report discusses some of the scientific rationale and closely related technical issues that justify a US investment in a national, large-aperture, ground-based optical telescope in order to advance studies and understanding of the Sun by the solar research community. This is not an in-depth study, nor has it been reviewed by the community of users of existing major facilities. Potential international collaborators have not been consulted. This should be regarded as a preliminary working paper.

The Sun has been the subject of human wonder and scrutiny since the dawn of civilization. There are at least three major reasons for this intense, continuing interest:

• The Sun sustains life on Earth; it controls our environment and impacts our technological civilization.

• The Sun is the nearest and most readily studied example of many physical processes that form the foundations of our current understanding of the Universe.

• The Sun presents us with many important unsolved mysteries and unexplored domains that challenge science.

The key goals for solar physics have been compiled by numerous study groups in recent years. A space-based perspective is found in the “Sun-Earth Connection Roadmap” (NASA 1997). A recent NRC decadal study, “A Science Strategy for Space Physics” (NRC 1995), embraces both space- and ground-based activities, as does the solar contribution to the last astronomy decadal survey “Working Papers -Astronomy and Astrophysics Panel Reports” (NRC 1991). The most recent ground-based perspective is found in “The Field of Solar Physics: Review and Recommendations for Ground-Based Solar Research” (NRC 1989). The NSO strategy document “Outline of a Program for the Understanding and Prediction of Solar Variability” (1995) also lists key goals. Research goals having practical applications are found in “The National Space Weather Program - The Implementation Plan” (OFCM, 1997). These reports feature several recurring major themes:

• Measurement of the interior structure and dynamics.

• Measurement of the spectrum of solar neutrinos.

• The origins of solar variability and the solar activity cycle.

• Physics at small spatial scales.

• Physics of transient eruptions.

• Origin and evolution of the corona and solar wind.

• Links between the changes in the Sun and effects on Earth.

• How understanding the Sun is necessary to understand other stars and vice versa.

• The synergy between solar physics, stellar physics, and astrophysics.

These themes in turn produce more specific research opportunities for ground-based optical investigations that include improving our understanding of:

• interior

• the operation of the solar dynamo.

• how interior changes cause solar cycles.

• the nature of convection.

• the structure of the base of the convection zone.

• the role of magnetic buoyancy in the convection zone.

• how differential rotation and meridional flow are maintained.

  • the structure of the Sun's internal magnetic and velocity fields and how they are related to the surface fields.

• surface and atmosphere

  • what causes the strong intermittency of magnetic fields.

  • how magnetic flux erupts, evolves and disappears.

  • how magnetic activity and solar output variations are coupled.

  • the precursors to solar activity – the short-term process of active region development and the long-term buildup of polar fields.

  • how magnetic energy is stored and released in the solar atmosphere.

  • what magnetic configurations and evolutionary paths lead to flares and coronal mass ejections

  • the mechanisms responsible for variations in the spectral and total irradiance of the Sun and solar-type stars.

  • the variability of the Sun in the short wavelengths and effects on the ionosphere and thermosphere.

  • what heats and cools the chromosphere, prominences, and low corona.

  • the three-dimensional structure of coronal magnetic fields and its relation to the corona's thermal and dynam*ic evolution.

  • how to predict coronal mass ejections.

  • what heats and accelerates the solar wind and causes the nonuniform characteristics of its flow and composition.

• outer corona and beyond

  • how particles are accelerated to high energies (MeV and GeV) and how they propagate through the interplanetary medium.

  • the related physical processes that modulate the flux at 1 AU of galactic cosmic rays.

  • how long-term changes at the Sun are coupled to changes in the heliosphere.

  • the activity cycles and internal properties of other stars.

These examples are just some of the problems and opportunities awaiting solution by current and future studies of the Sun. The fact that so many of these basic problems are unsolved is largely due to a lack of facilities to make critical observations. Models of many of these phenomena exist but cannot now be verified or advanced. Some of these problems are most likely to be solved by space observations while others require larger, ground-based facilities – especially a facility that can make high signal-to-noise ratio measurements at small spatial scales using robust observational diagnostics.

Science advances most rapidly when theory and observation progress in tandem. During much of its history, solar physics has been paced by observations. Innovative telescopes at good sites and a sequence of space observatories uncovered a wealth of puzzling phenomena, and theory often lagged behind.

In the past decade this situation has reversed. Powered by fast new computers and armed with the principles of physics, theory has surged ahead. Numerical simulations now are able to model the physics of solar processes in great detail. However, detail is insufficient if it is not relevant, and to an increasing degree, theoreticians have lacked the guidance of critical observations. As a result, in many cases we have competing theories that cannot be observationally distinguished or tested. The consequence is stagnation caused by a lack of crucial observations that cannot be made with existing facilities.

It is sobering to realize how many long-standing beliefs about the Sun have been found to be incorrect, sometimes spectacularly so, when critical observations are made. Two recent examples are the surprising rotation profile inside the Sun and the unexpected need to dismiss current models of the chromosphere. We believe that a breakthrough in our ability to understand the Sun is necessary and possible if a new, large-aperture solar telescope is built having the following observational capabilities:

• high angular resolution

• high photon flux

• a wide wavelength range including access to unique physical diagnostics in the infrared

• low scattered light to allow observations of the chromosphere and corona.

The need for a large-aperture facility has long been recognized and it is useful to try to draw lessons from recent history. We note that almost exactly 10 years ago, studies were done that parallel today's activities to a remarkable degree. A decade ago there was an AURA-sponsored study of a large-aperture solar telescope (MacQueen report, published in slightly modified form as LEST Report No. 24, 1987). The HAO prepared a study of LEST (“Report on the Scientific Potential of the LEST”, 1988) and NSO conducted a similar study (High Resolution Committee, 1987). These were steps that led to US participation in the LEST project as the only solar initiative recommended by the last Astronomy and Astrophysics Survey Committee (NRC, 1991) and a subsequent proposal to the NSF. There was also an NRC study of ground-based solar physics (Rosner Committee, 1989) that recommended development of a large-aperture solar telescope. The analog today is the present NRC/SSB Task Group on Ground-Based Solar Research. A decade ago, the efforts failed.

Today the situation is different. The need for a new large-aperture facility is even more acute. The main reasons are a wealth of new observations from existing ground-based facilities, new space observations from Yohkoh, SOHO, and TRACE, and especially the development of sophisticated models of many solar phenomena that cannot be tested observationally with existing facilities. Stunning technical developments in detectors and adaptive optics during the last decade also make a large-aperture facility more attractive and feasible. The budget situation is better now, with the major facility needs of the nighttime optical and radio communities satisfied by very large capital investments during the last decade through the present, and with positive words and actions by Congress to increase the NSF budget.

The studies cited above, and several others not mentioned, have presented comprehensive lists of scientific goals for high-resolution solar observations. Rather than simply repeat those discussions here we attempt here to highlight some key areas and indicate some more specific broad goals. The reason for this menu approach is not to slight the importance of the science but simply to give a fast overview. In keeping with the theme that the major purpose of a new facility is to catch up with modeling, the emphasis is on testing models. In the rest of the report, we select some of the specific goals for more extensive discussion.

• magnetic fields and mass motions

  • test models of magnetoconvection (e.g. Brummell et al. 1995)

  • test models of the eruption of magnetic flux (e.g. Moreno-Insertis and Emonet 1996)

  • test models of the production of acoustic oscillations (e.g. Rast 1997)

  • measure the characteristics of weak photospheric flux features (chromosphere, prominences, corona)

  • quantify the processes by which magnetic flux is dissipated

• heating and cooling

  • test models of shock wave heating of the chromosphere (e.g. Carlsson and Stein 1997)

  • test models of radiative cooling of the COmosphere (e.g. Ayres 1981)

  • test models of wave and topological heating of the corona (e.g. Vekstein 1996)

  • test models of the formation of prominences (e.g. Priest et al. 1996)

• activity and the solar cycle

  • test models of flares, jets and other eruptions (e.g. Hori et al. 1997)

  • test models of irradiance variations that arise from small features (e.g. Topka et al. 1997)

  • measure the contribution of weak flux features to the solar cycle

• measure the energy budget of active regions and compare with explosive activity

A large aperture provides two unique functions: high angular resolution and high photon flux. Why do we need these capabilities now? Because mass motions and magnetic fields involve a hierarchy of structures, and crucial parts of their distributions fall below the the capabilities of current instrumentation. Furthermore, convection and magnetic fields can have strong influences on one another that lead to rapid temporal evolution at the smallest spatial scales. This produces a complimentary need to make high-sensitivity measurements quickly. Thanks to sophisticated image processing techniques and advanced detector systems, we have recently reached the limits of the largest present-day telescopes. These limits are too low to solve many of the problems listed above.

As has been suggested frequently in the past (e.g. Parker 1996), what is needed is a “solar microscope” to place elusive solar phenomena under the intense scrutiny required to unravel the complex physical process involved with mass motions and magnetism. The analogy to microbiology is compelling. Right now we are glimpsing some of the ‘cells' of the Sun, but their internal machinery is hidden from view.

Although the focus here is on the Sun, all stars eventually pass through the cool half of the H-R diagram. Convection and magnetism are common threads that run through this half of the diagram. Many key stellar phenomena – chromospheres, coronae, winds, etc. – can be traced to interactions between the two. To understand stars, one must understand both convection and magnetic fields. Since starlight is what conveys most of our knowledge of the Universe, to understand distant stars we must first understand the Sun, specifically convection and magnetism. Currently, that last step is stymied by a lack of resolution and photon flux in today 's solar facilities.

Here we present just three of many possible goals for a new facility. These highlight the need for high resolution and high photon flux. The current state of the art allows exciting, nearly diffraction-limited snapshots to be made under good seeing conditions with the 76 cm NSO/SP Vacuum Tower Telescope (e.g. Rimmele and Lin 1997) and the 48 cm Swedish Solar Telescope (e.g. Berger et al. 1995). These snapshots require short exposure times and have low spectral resolution. Scales as small as 0.2 arcsec can be glimpsed in the blue end of the spectrum. Application of image restoration techniques such as differential speckle and phase diversity speckle have allowed high-angular-resolution time series observations of granulation and photospheric magnetic fields to be obtained (e.g. Paxman et al. 1996). While important new characteristics of the Sun have been discovered, even with these techniques, many solar processes remain unresolved with present telescopes.

One of the early discoveries about photospheric magnetic fields was that they do not smoothly vary across the surface as in the case of the Earth, but are strongly concentrated into remarkably small patches. A close association of these concentrations with the boundaries of supergranules led Parker (1963) to suggest that they resulted from the field being swept to small downflow regions in the photospheric flow field. In this scenario one might expect the magnetic pressure in such a concentration to be roughly equal to the surrounding gas and kinetic pressure, that is, a field strength of several hundred gauss. Measurements showed instead that much of the magnetic flux is more strongly concentrated to kilogauss strength. Parker (1978) proposed that the amplification resulted from thermal isolation of the magnetic structure from its surroundings. This allows an initial downflow to be accelerated, leading to a partial evacuation of material from within the magnetic region and a resulting increase of magnetic pressure.

Recent observations have indicated that the magnetic concentrations also tend to be located on the downflowing boundaries of granules as well as supergranules (e.g. Berger et al. 1995). Numerical models

(e.g. Steiner et al. 1997) suggest that the concentrations are very dynamic as a result of the violent motions in the photosphere. It is likely that wave motions and shocks of various types are produced, magnetic reconnection takes place, and the uniform outward flow of solar radiation is strongly perturbed at these concentrations. If so, these concentrations play a major role in the flow of solar energy upward into the atmosphere and space. However, this is speculation based on suggestions from models and hints from indirect observations. What is needed to test the scenario and move to more advanced models are definitive observations with enough angular resolution to resolve the concentrations and enough signal-to-noise ratio to make meaningful measurements. Specifically, a resolution of 0.1 arcsec and an ability to make accurate vector magnetic field measurements at this scale Are required.

Since their discovery in 1974, ubiquitous weak, bipolar flux regions have resisted obser-vational scrutiny. The relationship of these fields to the more obvious stronger fields is not clear. It has been speculated that they may simply be the small scale end of a wide spectrum of flux element sizes and indicate flux that is temporarily not concentrated by the processes mentioned above. On the other hand, the highly random orientation of the fields and the observation that they appear to be constantly erupting suggests that another process may be at work. This might be a turbulent dynamo that amplifies fields locally in contrast to the dynamo that is thought to produce the fields of active regions at the base of the convection zone. In this case, the weak flux features may be the large-scale end of a spectrum of flux sizes. High resolution, high sensitivity observations are required to sort out these possibilities.

Solar flares represent the most dramatic examples of magnetic field instabilities. According to current ideas, the shearing and twisting of coronal magnetic fields by photospheric convective motions stores energy in coronal electrical currents. When some unknown, critical threshold is crossed, the current system develops instabilities and the energy is released catastrophically, accompanied by charged particle acceleration.

A vast literature of models, theory and scenarios explores these complex processes. Numerical modeling of coronal force-free fields has progressed considerably, but is by its nature inadequate to represent the slowly varying, non-force free fields that actually occur. Recent observations of the buildup of magnetic shear are highly suggestive but do not offer as yet any clear insights into the critical thresholds. Thus both modeling and observations are inadequate for a full understanding of the pre-flare buildup of energy.

The answers must lie in the details of the interaction between the subarcsecond field concentrations and convective motions. Vector magnetic field and velocity measurements, with 0.1 arc-sec resolution and high signal-to-noise ratio, in combination with X-ray coronal observations from future satellite missions, offer the best approach to crack the flare buildup problem. A large aperture telescope provides the resolution and photon flux needed for such a program.

In cosmology and other parts of astrophysics, structure has become a major research topic. The Sun is no exception. The chromosphere and corona of the Sun are threaded with an incredible amount of small-scale, highly dynamic structure far different from what might be expected based on simple one-dimensional models. This structure has a profound effect on local dynamics, energy flow and radiative transfer, and indicates the presence of dynamic processes that are not understood. Since we expect the atmosphere to be completely filled with magnetic flux expanding from the small concentrations in the photosphere, a major puzzle is why are coronal and chromospheric features so spatially structured in a presumably smoothly varying magnetic field?

One suggestion is that there is some intermittent process acting at the base of the structures that loads certain field lines with enough mass to become visible. If so, what is the process and what does the selection? Pressure-driven siphon flows have long been considered as one possibility (e.g. Thomas 1988) and observational support has been reported (e.g. Rüedi, et al. 1992; Zirker, 1994). But another possibility is that energy dissipated in the corona causes thermal effects that affect the visibility of mass on selected

field lines. A developing scenario is that certain parts of the atmospheric magnetic field are topological boundaries (loosely, ‘cracks') prone to energy dissipation (e.g. Parnell et al. 1994) and might organize the apparent selective structuring of the atmosphere.

To learn more about this problem, several space observations are planned. Indeed, much has already been learned about the linkage between coronal loops and a sunspot (Sams et al. 1992), and the recently launched TRACE mission is giving us remarkable views of the threedimensional structure of atmospheric structures. An expected result is that most features will be traced to roots in strong network magnetic fields. However, this expectation may be unrealistic. For example, Engvold (1997) reported that the barbs of quiescent filaments connect to the weak intranetwork magnetic fields instead of the strong flux concentrations at the network boundaries. Given the small scale of structures in the chromosphere and photosphere, it is a major observational challenge to try to connect one layer to another even with excellent angular resolution. Inadequate resolution might well lead to incorrect conclusions. This study of connectivity is an ideal problem for a large-aperture ground based telescope.

Scattered light degrades solar observations by adding noise from photons that arise from locations other than the one being observed. In other words, it adds noise but no signal. For photospheric observations it is generally a nuisance rather than crippling in the sense that small degradations and small corrections are the usual situation. A major exception is measurements of sunspots. In the chromosphere and corona, scattered light can completely swamp tiny signals. To study these important regions with a new telescope, it must have low scattered light characteristics. We concentrate on coronal objectives because that is the most difficult observational challenge and also because there are a number of outstanding scientific problems that can be addressed with the new telescope.

It is difficult to overestimate the importance of coronal magnetic fields for the physics of the corona. The field is responsible for the highly inhomogeneous structure, with scales ranging from sub arcsec (November and Koutchmy 1996, Golub et al. 1990) to many radii (streamers). More important, magnetic fields are thought to be crucial for the heating of the solar corona (see reviews by Priest 1994, Brown 1991, Cargill 1994, Zirker 1993, Vekstein 1996).

To date, direct observations of coronal magnetic fields have been virtually impossible (except for measurements of strong fields over sunspots by radio techniques – e.g. Gary and Hurford 1994). Since the Zeeman splitting is a tiny fraction of the Doppler width of coronal spectrum lines, very high sensitivity is required in faint targets. As a result we have been forced to rely on potential or force-free extrapolations of the line-of-sight or vector photospheric field in order to estimate coronal fields. This technique is improving, with the availability of vector field measurements in the photosphere at higher spatial resolution, but the extrapolations still leave considerable uncertainty. The fields are measured on a surface where they are not force-free. Then the observations must be adjusted to fit into a force-free model and the extrapolations are themselves subject to numerical difficulties. Until we have a run of measurements up into the corona, even if fraught with measurement problems, we have no check on models. So far the emphasis in this work has been on matching the X-ray shapes of loops, a relatively weak constraint, since potential field extrapolations are almost as good. As yet there is no check available on the magnitude of the field. Independent measurements of the same loop system with a coronagraph could provide an empirical test of the extrapolations.

How well could a large-aperture coronagraph measure coronal fields? We have two benchmarks. Harvey (1969) obtained 3σ, detections with about 0.6 gauss noise level using the 530.3 nm line, the 40 cm

Climax coronagraph, a large sampling aperture, and integrations of 30 to 60 minutes. These results with 1960s technology are equivalent to roughly a 700 gauss noise level low in the corona using a one-meter aperture telescope with a one arc-sec sample exposed for one second. Kuhn (1995) has obtained an upper limit of 40 gauss in unresolved corona, using the Evans 40 cm coronagraph, and a 300 second exposure to record the Zeeman splitting of the Fe XIII lines at one µm. The normalized noise level from this modern measurement is fairly close the old result despite the different detectors and spectrum lines used. He has scaled this measurement for larger apertures. It appears that a four-meter (two-meter) aperture could detect a limiting field of three (six) gauss at one radius above the limb. This sensitivity could be sufficient to probe the dome of a helmet streamer, for example, and to test a variety of heating mechanisms.

An independent estimate is given in the attached graph (Figure 1) for isolated loops of varying brightness in Fe XIII 1074.8 nm. It appears that a four (two) meter telescope could detect a limiting field (S/N = 3) of 5 gauss in a loop as faint as 3 (10) millionths within the lifetime of the loop (1000 s). The interpretation of such measurements, accounting for line-of-sight integration through the corona, will be challenging; but the basic sensitivity appears to be attainable.

Several mechanisms have been proposed for heating the corona. They fall into two categories, depending on the typical time scales of the photospheric drivers. There are field-aligned currents, produced by slow persistent braiding (Parker 1981) or twisting (Sturrock and Uchida 1981) of coronal field lines by photospheric motions of their footpoints. Or conversely, there is dissipation of MHD waves that are generated by more rapid photospheric motions of the corona (Hollweg 1991, Davila 1991). Parker's mechanism involves a slow buildup of free energy, followed by an impulsive release in tiny “nanoflares ”.

Porter and Klimchuk (1995) have proposed a statistical method to decide among these alternatives in loops. The method depends ultimately on determining the correlation of field strength with loop length. The exponent of the assumed power-law relation ranges from −2.0 (Alfv én waves) to −0.5 (nanoflares) to 0.0 (twisting by vortices) for the three different mechanisms. Porter and Klimchuk propose the usual field extrapolation from the photosphere, we propose direct Zeeman measurements in loops.

One of the more promising methods proposed for heating coronal loops is the resonance absorption of Alfvén waves that are generated in the photosphere and propagate as tube waves. A loop acts as a resonant cavity, and selects a narrow band of frequencies from a broad-band spectrum. The resonance frequency excites a global mode (Davila 1995) which in turn excites a surface wave at the interface between the loop and its low-density surroundings. Although it is the surface wave that heats the loop, it is the global wave that could be detected.

Yet observations of wave motions in the corona are rare, unreproducible and difficult to interpret (Tsubaki 1988). Line broadening, with non-thermal speeds up to 60 km/s have been reported (e.g. Hassler et al. 1990; Hassler and Moran 1993) in coronal spectrum lines, but periodic Doppler shifts have not. Presumably line-of-sight integration has averaged out discrete Doppler shifts. A search for periodic or episodic Doppler shifts, accompanied by intensity variations in optical/IR coronal lines, within isolated structures, would be feasible with a three-meter coronagraph.

If such signs of wave motion are detected, the global resonance frequency (Alfvén speed divided by loop length) could be determined directly from observations of the field strength and plasma density in a loop. One could therefore test the proposals that Alfvén waves are present in loops and that resonance absorption may function there. The pair of Fe XIII lines (1074.7 and 1079.8 nm) are ideal for this project. Their ratio is density sensitive and their Zeeman splitting is within reach.

Smartt et al. (1993) have reported interactions between post-flare loops that resemble Parker's proposed “nanoflares”. Where two loops cross in the line of sight an impulsive heating event occurs, followed by expansion along the loops and cooling. The observations were made sequentially in the cororial green and red lines and in Hα, with the NSO/SP 20 cm refractor coronagraph. The total energy released is typically 10²⁸ ergs, in the microflare range. It is natural to expect much smaller and more frequent interactions at lower energy, especially in the quiet corona. A search for such interactions with a large aperture coronagraph would help to establish the nanoflare mechanism.

Simultaneous images of the quiet corona in Fe X 637.4 nm (IMK) and Fe XIV 530.3 nm (2MK) often show entirely different structures. Is this an evolutionary effect in which hot loops cool from two to one million Kelvin? Or are several stable temperature configurations possible, each with a different rate of energy deposition? How is the equilibrium temperature related to the loop's magnetic field strength? A large aperture coronagraph could isolate individual structures and investigate the relationship between field strength and equilibrium temperature.

Low (1996) has offered a scenario for the eruption of prominences and the accompanying CME. In his picture the prominence hangs in a horizontal flux rope which coincides with the coronal “cavity”. An arcade of coronal loops encloses this cavity. When the eruption begins, the arcade is assumed to reconfigure through reconnection, leaving a low arcade under the rising prominence. The line-of-sight field in this low arcade should have the opposite sign to that in the prominence. If this reversed sign could be detected, the scenario would be closer to confirmation.

Except for their role in coronal mass ejections, quiescent prominences have not attracted nearly the same level of interest as say, flares or the heating of the corona. And yet they present a fascinating array of challenging physical problems, for which many solutions have been proposed, but none proven (see reviews by Priest 1989, Zirker 1989, Tandberg-Hanssen 1995, etc). Among these issues are:

• the source of radiative energy

• the three-dimensional magnetic structure and its stability against deformation

• the causes of ultimate eruption

• the origin of sub-arcsec vertical filamentation

Prominences radiate at volumetric rates orders of magnitude larger than the corona. How is this energy loss supplied? Heat conduction, X-ray absorption, wave dissipation, joule dissipation are all candidates, but each of them has flaws and none has been positively identified.

The answers may lie in the corona-prominence interface (Orrall and Schmahl 1981). The interface is thought to be a thin sheath with steep temperature and magnetic field gradients. Spectroscopy and polarimetry at 0.1 arc-sec resolution (e.g. in Ca 11 854.2 nm or He 1 1083 nm) could demonstrate whether heat conduction from the corona or absorption of coronal and chromospheric radiation accounts better for the temperature gradients and energy supply in the interface. Alternatively, a search for wave motions in the prominence fine structure, at sub arc-sec resolution, could set hard limits on wave amplitudes.

A French group headed by J-L Leroy (1984) has established some basic features of the internal magnetic fields of quiescent prominences, using the Hanle effect (see also Lin et al. 1996). Their results indicate a horizontal component of 3-20 gauss. One of their more surprising results is that the field show neither filamentary structure nor vertical components associated with the Hα threads, at least at their resolution of five arcsec.

Engvold (1976) has measured the Hα brightness of arcsec threads in Dunn's prominence observations. Using his measurements we estimate that a three-meter coronagraph could measure a line-of-sight field of 10 gauss, (S/N = 10), with a spatial resolution of 0.1 arcsec, in a one-minute integration. Or measure a transverse field component (along the axis of a vertical thread) of 10 gauss in ten minutes, with 0.3 arcsec resolution and S/N = 10. Such observations could help to unravel the complicated internal fields within quiescent prominences.

Infrared observations beyond 2.5 µm are expected to contribute uniquely to solving two key problems of solar physics: the role of weak surface magnetic fields in the solar cycle and the origin of the chromosphere. More generally, infrared measurements of magnetic field strength and temperature, density, and chemical composition are the most direct and sensitive of any known techniques.

Infrared observations from space with useful angular resolution are impractical, and only one ground-based telescope in the world currently accesses the full infrared spectrum. This unique capability, exploited using modern detectors, has given us a new window on solar physics. A larger aperture is required to further develop the powerful diagnostic possibilities available in the infrared.

Since the invention of the solar magnetograph, our understanding of photospheric magnetic fields has undergone one revolution and may well be in the process of another. The first revolution established that most of the magnetic flux in plages and the magnetic network is in the form of kilogauss-strength, sub arcsec field concentrations rather than the much apparent weaker field that corresponds to the measured flux. The second revolution concerns the importance of the “magnetic carpet” that covers the rest of the Sun outside active regions. Several indirect but independent estimates (Stenflo 1994) indicate that the flux in this component is comparable to the strong-field flux, even at solar maximum. The weak-field component continually renews itself on a time scale of a few days at most (Schrijver et al. 1997).

How do strong fields and weak fields interact? Does the weak-field component have largescale structure? How is it generated? How does it disappear? Stenflo (1994) emphasized the importance of understanding the role of weak fields in the solar cycle: “A tiny, non-random component would however suffice for the IN [internetwork] emergence to be the dominating source of the large-scale pattern, since the IN flux emergence rate is 10⁴ times larger than that of AR [active regions].”

Our ambitious but attainable goal should be to measure the magnetic flux at every point on the solar surface and to measure the field strength at as many points as possible. The “12-µm” emission lines (e.g., Mg I 12.32 µm) will play a key role in reaching this goal because they alone permit a model-independent Zeeman measurement of magnetic field strength down to ~100 gauss (Solanki 1994). The well-known pair of Fe I lines near 1.56 µm is sensitive down to ~250 gauss and will complement the 12-µm lines because the Fe I lines are formed near the base of the photosphere while the 12-µm lines are formed about two pressure scale heights above. Kilogauss flux concentrations with low filling factor will spread and weaken significantly between these two heights, whereas space-filling weak fields may change only slightly.

Another advantage of the 12-µm lines, connected with their height of formation, concerns the extrapolation of surface magnetic field measurements into the outer atmosphere, an application that has become increasingly important with the advent of high-quality coronal images from Yohkoh, SOHO and TRACE. The extrapolation techniques usually assume that the field is force-free, but this is a poor assumption for visible magnetograph lines such as Fe 1 630.2 nm and even poorer for the 1.56-µm lines: gas pressure is comparable to magnetic pressure at these levels. For the 12-µm lines, however, the ratio of magnetic pressure to gas pressure is an order of magnitude greater. Thus, the 12-µm lines provide a consistent “platform” for inferring the structure of the magnetic field at higher levels.

The mystery of outer stellar atmospheres begins, not in the corona, but at the top of the photosphere in the so-called temperature minimum region. In the absence of non-radiative heating, there would be no temperature minimum: the temperature would decline steadily with height until it reached equilibrium

with the interstellar medium.

Classical chromospheric models are one-dimensional, semi-empirical, hydrostatic atmospheres that assume whatever distribution of non-radiative heating is necessary to produce the “desired” run of temperature from photosphere to corona. Nearly every textbook presents a similar figure. Various sources for non-radiative heating have been proposed; none has been proven.

Infrared observations of the carbon monoxide molecule provide some of the strongest evidence to date that the textbook picture – of the Sun, and late-type stars in general – is wrong.

Models (Ayres and Rabin 1996 and references therein) founded on spectra of the CO vibration-rotation band system near 4.8 µm show that, below a height of about 1000 km, over most of the solar surface, most of the time, there is no chromosphere. At any one time, only about 20% of the solar surface produces the radiation from what we have traditionally called the chromospheric “layer.” Most of the volume below the magnetic canopy is occupied by a “COmosphere” that is cooler than the temperature minimum of standard models. In this respect, one-dimensional models have little physical meaning because the mean properties represented by the model may be encountered almost nowhere in the actual atmosphere. Dynamical simulations by Carlsson and Stein (1995) demonstrate that this apparently extreme characterization has a physical basis. Carlsson and Stein (1997) reinforce from a modeling perspective the startling conclusion implied by CO observations: “Despite long held beliefs, the Sun may not have a classical chromosphere in magnetic field free internetwork regions at heights below 1 Mm.”

The unique advantages of CO in elucidating this new picture are that

• CO dissociates strongly above 4000 K;

• the vibration-rotation lines are formed closely in LTE;

• they have narrow contribution functions that dissect the mean vertical structure of the temperature minimum region.

CO is not the only molecule that can probe the transition from the photosphere to the outer atmosphere. Deming et al. (1984) used LTE rotational lines of the OH radical near 11 µm to show that the thermal inhomogeneity seen strongly in CO is present even lower in the atmosphere, at (500 nm) optical depths ~10^-2.

The molecular observations to date have made progress in directly imaging the “unchromosphere,” in establishing the range of its physical properties, and in demonstrating substantial time variability. However, the measurements are acquired too slowly (due to inadequate flux and under-instrumentation at the focal plane) to resolve the time variations (including pmode oscillations) over a useful field of view (several supergranules). In other words, we have gone some way toward answering “what” and “where” but have barely scratched the surface of “when” and “why.”

In highlighting two important research areas in which thermal infrared observations play a central role, we short change a broad range of applications. The scope and rapid growth of infrared solar physics can be judged from the two recent volumes, “Infrared Solar Physics ” (Rabin et al. 1994) and “Infrared Tools for Solar Astrophysics: What's Next” (Kuhn and Penn 1995). We confine ourselves here to the briefest reminder that the infrared spectrum offers proven advantages for measuring the “bread-and-butter” quantities of stellar physics: temperature, density, and chemical composition, as well as magnetic field strength.

At photospheric temperatures, the intensity of the thermal infrared continuum closely approximates a Rayleigh-Jeans distribution. It is easier to measure temperature in the infrared (including the effects of spatially unresolved fluctuations) because the intensity is linear in temperature. Moreover, a true continuum is often accessible (unlike the heavily-blanketed visible spectrum), and the opacity is dominated by a single, well-understood source (H− free-free absorption). The infrared continuum at unit

optical depth spans about 250 km in height, compared to 40 km in the visible.

Kopp and Rabin (1992) used infrared intensities to demonstrate a relationship between temperature and magnetic field strength in sunspots (the lower contrast and stray-light contamination of sunspots in the IR was an additional advantage). Moran et al. (1992) verified that faculae are dark at disk center at wavelengths beyond 1.6 µ m, a property that may also extend to visible wavelengths (as micropores) at high angular resolution (Topka et al. 1992).

As discussed above, the CO lines are a sensitive thermometer for cool gas.

Like the Zeeman effect, the linear Stark effect is quadratic in wavelength. For hydrogen lines in the mid-inftared region (8-20 µm), the Stark broadening in prominences is so strong that the Stark and Doppler components of the line profile can be separated. Chang and Deming (1997) have used this property to derive order-of-magnitude more accurate temperatures and densities in both active and quiet prominences. Casini and Foukal (1996) have shown that the H 1 15-9 transition is sensitive to directed electric fields as small as 0.5 V cm^-1 an order of magnitude better than the upper limits from visible-light measurements. This gain in sensitivity is particularly important because it opens the door to electric fields generated by processes that are thought to occur continually – such as MHD wave heating – as opposed to the larger fields that occur only transiently in small regions (Foukal and Hinata 1991).

Accurate values of the solar abundances of C, N, and 0 are fundamental anchor points for the calculation of stellar opacities. Grevesse et al. (1994) stress that the most secure indicators, both atomic and molecular, occur in the infrared spectrum beyond 2 µm; this is even more true for isotopic ratios. In deriving abundances, it is vital to use spatially resolved observations so that each spectrum can be assigned a unique temperature and density. The combination of reasonable (arcsec) angular resolution with the very high signal-to-noise ratios (~10³) needed for accurate analysis demands a large-aperture infrared-capable telescope.

Science is a balance between understanding what we have seen and opening windows onto the unknown. This discussion has stressed the known and tangible applications of solar infrared measurements. Still, it is important to remember how incompletely explored the infrared spectrum still is compared to the visible. For example, almost nothing is known about the infrared spectrum of a solar flare, or about the spectrum of the corona beyond 2.5 µm. The first mid-infrared (8-21 µm) atlas of a sunspot was produced at the McMath-Pierce only three years ago (Wallace et al. 1994) and quickly led to the identification of the water molecule on the Sun (Wallace et al. 1995). Followup analysis published in July 1997 in Science (Polyansky et al. 1997; Oka 1997) required a new approach to the solution of the vibration-rotation Schr6dinger equation to reproduce the solar spectrum; the authors expect that the solar results will result in a fundamental change in the chemical analysis of other hot polyatomic molecules.

Still, the diagnostic potential of molecules such as H₂0, CN, and OH has barely been scratched. Why? Primarily because imaging technology – so crucial to understanding the pervasively inhomogeneous solar atmosphere – has only become available in the infrared during this decade.

Where should a large aperture solar telescope be located? Space would seem to be the ideal venue because of its freedom from atmospheric degradations and the possibility of continuous viewing of the Sun. Indeed, there have been efforts to place a big solar telescope in space for decades. For various reasons

(see “A Space Physics Paradox”, NRC 1994 for a history) past efforts have been unsuccessful. Finally, the Japanese-led project, Solar-B, which includes a 50 cm visible light telescope, appears to be on track for a three-year mission starting in 2004. A one-meter aperture, space-qualified telescope (SolarLite) is under construction to take advantage of possible future flight opportunities (Title 1996).

Given the difficulties experienced in placing even moderate aperture solar telescopes in space, at present, the ground is the only option for a large-aperture telescope. Here are some of the advantages:

• aperture size is not limited by weight or volume constraints

• data flow is not limited by telemetry bottlenecks

• focal plane instrumentation can readily be updated to take advantage of new technical developments

• bulky, power-consuming, or cryogenic instruments can easily be accommodated

• operational and scheduling flexibility

• adaptable to new and unforeseen science investigations

• long useful lifetime

Comprehensive studies of the Sun as a complete physical system benefit from both space and the ground observations. Each venue should emphasize its strengths and each should work to compliment the other to maximize scientific advances.

A major physical scale in the solar photosphere is 0.1 arcsec or about 70 km. This is the scale of intergranular lanes, magnetic flux tubes and the pressure scale height. There is strong evidence for shock waves and other violent activity on this scale. The best numerical models of convection in the photosphere show significant variations of all physical quantities on this scale. To resolve this scale requires a telescope having a diffraction limit of about 0.05 arcsec. Even then, better resolution would be desirable since the contrast of 0.1 arcsec features will be attenuated by diffraction more than a factor of two, requiring significant corrections. Furthermore, for measurements of electric currents, it is important to be able to measure angular gradients of observable quantities across scales as small as 0.1 arcsec. This scale is also important in other features such as sunspot penumbras, umbral dots, chromospheric structures and the corona. For example, it has been suggested that coronal loops are myriads of currently unresolvable individual threads rather than just diffuse, fuzzy structures.

Advances in adaptive optics are revolutionizing nighttime astronomy. Application of these advances to ground-based solar observations is now mainly a matter of money with no known technical impediments. The remaining significant technical issue is wavefront distortion detection because the solar surface is more complex and has lower contrast than targets used in nighttime adaptive optics. On the other hand, the Sun provides a large photon flux. It appears likely that a correlating Shack-Hartmann technique will do the job nicely in moderately good seeing conditions.

We envisage using a hybrid approach to obtain very high angular resolution. This idea is not new (e.g. Roggemann 1991). An adaptive optics system would provide partial correction for atmospheric and telescope aberrations, and post-processing methods would complete the job. In the infrared, adaptive optics alone should often allow the diffraction limit to be approached. Moving toward shorter wavelengths, staying close to the diffraction limit will usually require additional processing. There are several proven techniques to do this. These include speckle image reconstruction, and phase diversity speckle image reconstruction (Paxman et al. 1996). The latter technique is especially powerful but currently requires extensive calculations. Practical techniques exist to deal with the relatively small area of high-quality restoration, situations with low photon fluxes (Keller and von der Liihe 1992), and slit spectroscopy (Keller and Johannesson 1995).

We can thus expect to produce nearly diffraction-limited images for significant periods of time at a good seeing site. This will be accomplished first in the near infrared and later in the visible spectrum. It is

recognized that the point-spread function will be a noisy, changing function of time and that this will limit observations that require extremely stable angular resolution and/or high photometric stability. Another limitation will be the small size of the high-resolution image. For these cases, space is the obvious solution. However, even with these limitations, the high angular resolution provided by a large aperture ground-based solar telescope equipped with adaptive optics will be a stunning breakthrough for solar research.

The relation between telescope diameter D, and its diffraction-limited Airy disk radius Δθ, is a well-known function of wavelength λ:

Δθ = 1.22λ/D.

(1)

The table below shows the Airy disk radius in arcsec for a series of wavelengths in nm and aperture diameter in cm. The first four rows represent existing telescopes. It is worth noting that for an aberration-free unobstructed optical system the values in the table correspond to spatial frequency scales that have 7% of their true contrast. Values should be decreased by 1.22 to get the diffraction-limited cutoff scales. We also note that a central obstruction raises the observed contrast of small-scale features at the expense of the contrast of larger scale objects and additional general scattered light.

An obvious conclusion is that to achieve a diffraction limit of 0.05 arcsec (and a corresponding resolution of 0.1 arcsec) with a practical sized telescope will require working at the

shorter wavelength range. At shorter wavelengths it is harder for adaptive optics to reach the diffraction limit and the ability to measure magnetic fields accurately is reduced. These tradeoffs suggest that a telescope with an aperture of about 3 meters is the minimum required to make measurements at the scientifically-driven scale of 0.1 arcsec.

Angular resolution has to be accompanied by adequate photon flux to enable useful measurements. Unlike the vast majority of nighttime objects, the Sun changes on short time scales and we do not have the luxury of being able to make long exposures to build up signal and reduce noise. An obvious example is a tiny flare event.

A major decision for a solar observer is how to divide the available photons into appropriate spectral, spatial, polarization, and temporal slices. This strongly depends on the problem being addressed. However, a few general statements can be made. Motions of several km/sec are common on the Sun and motions of tens of km/sec are not rare. In addition, most features on the Sun are transient; some last for just a few seconds. Even for longer-lived features, significant evolution often occurs in a few seconds. These effects mean that the spectrum line profile of a single point changes with time as does an image at a single wavelength. How long an exposure can be made without smearing depends on the speed at which the studied objects are changing.

For example, suppose we naively use a narrow-band filter to take two sequential images in opposite states of polarization in the wing of a spectrum line to estimate magnetic fields. In this case the difference between the two images is interpreted as a Zeeman splitting. Due to the five-minute oscillation and also to the evolution and motion of granules, the intensity at a well-resolved spatial pixel will change at the rate of at least 0.2% per second. This will produce a false magnetic field signal roughly equivalent to 10 gauss per second. Of course, this well-known problem is avoided by using fast modulation in practice, but serves to illustrate the need to obtain measurements quickly.

Similarly, suppose we take a series of spectrograms across a granule in order to map its characteristics using spectrum lines formed at different heights. Evolution and motion of the granule at 2 km/s will blur its structure by 0.1 arcsec in about 30 sec. Thus we need to make spectra at the rate of about one per second to map the entire granule before it changes at the scale of the observational resolution. Many chromospheric features require an order-of-magnitude faster scanning.

We conclude that the dynamic nature of the Sun requires that high-resolution observations have to be done quickly to avoid serious distortion of the measurements.

A related issue is the maximum obtainable signal-to-noise ratio. With modern CCD cameras and very careful attention to calibration and noise reduction, it is possible to obtain signal-to-noise performance at a level of about 10000 at a single pixel. Calibration at this level between different pixels is more difficult but may be possible with great care. Cameras exist that can read out at a rate equivalent to a detected photon flux of 6 × 10⁷ per second, so two seconds of such a readout could provide a photon shot noise of 10000 per pixel. The present and foreseeable state of the art does not promise much higher signal-to-noise ratios.

A basic measurement may be characterized by signal-to-noise ratio S/N, spectral resolution R, angular square pixel size Δθ, exposure time Δt, and telescope diameter D. If we let P be the number of photons per steradian-cm²-Å-s multiplied by the wavelength, let e be the total system efficiency for collecting photons at the focal plane of the telescope, and assume that photon noise is the only source of noise, then we find that

S/N = DΔθ/2 (Peπ/R)^1/2

(2)

Within a factor of two, the value of P for the line-smoothed photospheric continuum in the wavelength range of 400 to 4000 nm is 2.5x 10²¹ if we specify D in cm, Δθ in radians, and Δt in seconds. The strongest chromospheric spectrum lines are a factor of ten less intense and the strongest coronal emission lines are reduced by a factor of 10⁸. A total system efficiency, e, of 0.05 is an ambitious but feasible value.

Three different spectral resolutions tend to be frequently used for solar observations: spectrograph ~300,000, narrow-band filter ~30,000, interference filter ~1000.

The discussion above indicates that an exposure much longer than a few seconds is likely to lead to degraded high-resolution (0.05 arcsec pixel) observations. Some observations would demand considerably shorter exposures. We use 5 seconds as an exposure time equivalent to the time for a Mach I motion to traverse 0.05 arcsec in the photosphere.

For morphological image studies a signal-to-noise ratio of 100 to 300 is often adequate. Line profile measurements are often done using S/N between 500 and 1000. For example, a Doppler shift measurement of a spectrum line having a width of 8 km/sec using a spectral resolution of 300,000 would produce a velocity noise of about 3 m/s for S/N of 1000. To measure the line-of-sight component of a uniform magnetic field to I gauss requires about 5000, and the transverse component to 30 gauss about 10,000. The exact values depend on the spectrum lines and techniques used to fit polarized line profiles. For example, Skumanich et al. (1997) indicate that if a magnetic feature is resolved it may be possible to measure true vector fields as weak as 100 gauss with errors of 25 gauss and 6 degrees of field direction with S/N equal to 2000.

Suppose we want to measure polarized spectrum line profiles at a resolution of 300,000, using 5 sec exposures and 0.1 arcsec pixels. The value of S/N as a function of aperture is shown in the table here.

We see that even under these compromised observing conditions, only the largest apertures can provide vector field measurements of uniform fields at the 100 gauss level. One could compromise spectral resolution by a factor of 10 and double the exposure time in order to approach the signal-to-noise ratio limit of detector systems of 10000, but only for the largest apertures. The situation for chromospheric lines is about a factor of three worse and for coronal lines a factor of 10⁴.

We conclude that the combination of short time scales and small spatial scales demands large photon fluxes if the physics is to be studied with precision. Progress has been made with existing facilities by improving the efficiencies of focal plane instrumentation and detectors. Even so, compromises are often necessary such as sacrificing temporal, spectral or spatial resolution because of a lack of photons. Thus to advance, solar physics requires a telescope with a larger aperture than any now in existence and one capable of reaching the relevant sub arc-second scales for a significant fraction of the possible observing time.

An obvious question is “Why consider a classical coronagraph, which is limited to limb observations, when X-ray telescopes observe the corona so much better on the disk? ” Indeed after one has seen the exquisite soft X-ray images obtained by NIXT, Yohkoh, and TRACE, this question comes highly relevant. Current and future space missions, such as TRACE and Solar-B will undoubtedly make great strides toward the resolution of several important issues in coronal physics.

We would point out, however, that a three-meter all-reflecting coronagraph has important advantages that satellite instruments do not. Foremost is the prospect of measuring coronal magnetic fields. Second, the large flux available in a large aperture coronagraph offers unprecedented sensitivity. And with adaptive optics, the coronagraph would have diffractionlimited performance, sufficient to match the observed diameters of X-ray coronal loops in active regions. At wavelengths of one or two µm the sky is impressively dark and, at a good site, can approach a Rayleigh sky. With advanced detectors like ZIMPOL, a large coronagraph could observe structures an order of magnitude fainter than ever before, even in relatively poor sky conditions.

A drawback to all coronal observations, especially at the limb, is long integration along the line-of-sight. This problem is real but not insuperable. Even with poor resolution in one spatial dimension, good resolution in the other two and in time has allowed definitive studies of many coronal phenomena. If disk observations (in Hα, or X-rays) of a region are made preceding the region's passage to the limb, the large-scale magnetic geometry at the limb can be recovered.

Similarly, force-free field extrapolations of photospheric magnetic fields (e.g. Jiao, McClymont and Mikic 1996) can help in defining the coronal geometry at the limb. With such aids it may be possible to isolate a single loop with known orientation and to carry out unique observations at arcsec resolution.

In general we should consider ground-based and space coronal observations as complementary. Neither has all the answers: together they may extend our knowledge.

The decade of wavelength between 1 and 22 µm is comparable in size and richness to the entire visible spectrum, so it is not surprising that it cannot be reduced to a single objective or figure of merit – comparable, say, to resolving magnetic flux tubes. Indeed, it will not be practical to achieve very high (0.1 arcsec) angular resolution in the thermal infrared through direct imaging.

However, some general statements can be made.

First and foremost, the thermal infrared is useless by definition if it cannot be observed. Currently, only one telescope in the world can observe the infrared beyond 2.5 µm with angular resolution sufficient (~1 arcsec at 5 µm) to relate directly to visible-light measurements. A larger aperture would greatly improve the quality and value of wide spectral range observations; to preclude future access to the thermal infrared altogether would be akin to completely eliminating infrared or radio wavelengths as tools of nighttime astronomy – which is not even a remote possibility.

The forefront of angular resolution is ~0.1 arcsec. A three-meter class telescope has a diffraction limit at this scale in the near infrared. The infrared offers the advantage of a larger isoplanatic patch size as one moves toward longer wavelengths. This means that reduced angular resolution is offset by an increasing high-quality field of view. For those investigations that do not need the highest possible angular resolution, but benefit from a big field, the infrared is very attractive. The value of infrared measurements will be maximized when they can be compared directly with visible measurements. The following apertures are required to reach various diffraction limits at three infrared wavelengths (a highly Zeemansensitive Fe I line is near 1.56 µm; the fundamental CO vibration-rotation bands are centered around 4.8 µm; the extremely Zeeman-sensitive Mg I lines are near 12.2 µm):

An aperture of about 3 m or larger will access the full range 1-5 µm at 0.5 arcsec or better; moreover, the diffraction limit should be routinely attainable with low-order adaptive optics (even just tip-tilt at the longer wavelengths). Angular resolution much below 1 arcsec is not practical at 12 µm through direct imaging. However, the real power of the 12-µm lines lies in their ability to measure weak fields that are not expected to be concentrated like kilogauss flux tubes in the low photosphere. For studying the role of weak fields in the solar cycle, a 12-µm magnetograph with 1-2 arcsec resolution would ideally complement the next generation of synoptic vector magnetographs represented by SOLIS (which will have 1 arcsec pixels).

Polarimetry of the 1.6-µm lines from a meter-class telescope in space would be scientifically valuable in combination with visible measurements from the same spacecraft. At longer wavelengths, solar infrared measurements with useful angular resolution are logically done from the ground.

The photon flux in the infrared is reasonably well matched to the attainable apertures. For example, the cadence of current CO measurements is too slow by a factor of ~5 to study the dynamical properties of the COmosphere over a reasonable area (~2 arc min). However, currently about 80% of the time is occupied by inefficient image scanning and data transfer. The warm spectrograph is inefficient (~1%) and contributes high thermal background. A 3-m telescope with precise pointing and scanning, coupled to a high (~8%) efficiency cryogenic spectrograph, would provide more than enough flux to increase the cadence by an order of magnitude using 0.25 arcsec pixels. In terms of thermal background, the emissivity of the telescope is not very important for non-coronal observations, but a well-baffled cryogenic spectrograph or high-resolution filtergraph is essential.

The relative importance of focal-plane instrumentation in the infrared is even greater than it is in the visible because the technology is not as mature and is improving rapidly. This is one reason why existing telescopes, despite their inadequacies, have by no means exhausted their research potential in the IR. The next generation of 1-5 µm IR arrays (1024 × lO24 framing at 15 Hz) should be in place within the next three years; comparable arrays for the 8-20 µm range may be 3-5 years farther off.

We are in a new era in solar research where the major problems need to be addressed by a wide range of observational tools working together. We conclude that a large-aperture solar telescope is an essential part of a program that studies the Sun and its processes as a complex, coupled physical system. High angular resolution is required to study the small scale end of the broad spectrum of phenomena that need to be understood. Reaching the important scale of 0.1 arcsec requires nearly diffraction-limited performance by an aperture no smaller than 3 meters, from the near infrared to as far blueward as possible. Such performance is attainable at a good site on the ground by using a combination of adaptive optics and image reconstruction techniques. The large aperture is also required in order to obtain high signal to-noise ratio measurements in times that are short compared with the evolution of solar features. The new telescope should allow access to a wide wavelength range, especially the robust diagnostics available in the thermal infrared, and should have low scatter characteristics to allow measurements of the chromosphere and corona.

Ayres, T. R. 1981, ApJ 244, 1064

Ayres, T. R., and Rabin, D. 1996, ApJ 460, 1042

Berger, T. E., Schrijver, C. J., Shine, R. A., Tarbell, T. D., Title, A. M., Scharmer, G. 1995, ApJ 454, 531.

Brown, P. K. 1991, Plasma Phys. Control. Fusion 33, 842

Brummell, N., Cattaneo, F., Toomre, J. 1995, Science 269, 1370

Cargill, P. 1994, in Solar System Plasma Physics, eds, J. Burch and J. Waite, (Geophys. Mono. 84), ( Washington: AGU), 31

Carlsson, M., and Stein, R. F. 1995, ApJ 440, L29

Carlsson, M., and Stein, R. F. 1997, ApJ 481, 500

Carlsson, M., and Stein, R. F. 1997, in Solar Magnetic Fields, ed. V. H. Hansteen Oslo:(U. Norway), 59

Casini, R., and Foukal, P. 1996, Solar Phys. 163, 65

Chang, E. S., and Deming, D. 1997, BAAS 29, 906

Davila, J. M. 1991 in Mechanisms of Chromospheric and Coronal Heating, eds. P. Ulmschneider, E. Priest, R. Rosner, ( Berlin: Springer), 464

Deming, D., Hillman, J. J., Kostiuk, T., and Mumma, M. 1984, Solar Phys. 94, 57

Engvold, 0. 1976, Solar Phys. 49, 283

Engvold, 0. 1997, in Solar Magnetic Fields, ed. V. H. Hansteen ( Oslo: U. Norway), 103

Falchi, A. Falciani, R., and Mauas, P. 1994, inInfrared Solar Physics (IAU Symp. 154), eds. D. Rabin et al. ( Dordrecht: Kluwer), 113

Foukal, P., and Hinata, S. 1991, Solar Phys. 132, 307

Gary, D. E., Hurford, G. J. 1994, ApJ 420, 903.

Golub, L. et al. 1990, Nature, 344, 842

Grevesse, N., Sauval, A. J., and Blomme, R. 1994,in Infrared Solar Physics (IAU Symp. 154), eds. D. Rabin et al., ( Dordrecht: Kluwer), 539

Harvey, J. W. 1969, Ph.D. Thesis, U. Colo.

Hassler, D., Moran, T. 1993, BAAS 25, 1200

Hassler, D., Rottman, G., Shoub, E., Holzer, T. 1990, ApJ 348, L77

Hollweg, J. V. 1991 in Mechanisms of Chromospheric and Coronal Heating, eds. P. Ulmschneider, E. Priest, R. Rosner, ( Berlin: Springer), 423

Hori, K, Yokoyama, T., Kosugi, T., Shibata, K. 1997, ApJ 489, 426

Jiao, L. A., McClymont, A. N., Mikic, Z. 1997, Solar Phys. 174, 311

Keller, C. U., Johannesson, A. 1995, AASup. 110, 565

Keller, C. U., von der Lühe, 0. 1992, AA 261,321

Kopp, G., and Rabin, D. 1992, Solar Phys. 141, 253

Kuhn, J. R. 1995, inInfrared Tools for Solar Astrophysics, eds. J. Kuhn and M. Penn ( New Jersey: World Scientific), 89

Kuhn, J. R., and Penn, M. J. 1995, Infrared Tools for Solar Astrophysics ( New Jersey: World Scientific)

Leroy, J-L., Bommier, V., Sahal-Brechot, S. 1984, AA 131, 33

Lin, H., Kuhn, J. R., Penn, M. 1996, BAAS 28, 956

Longcope, D. W. 1996, Solar Phys. 169, 91

Low, B. C. 1996, Solar Phys. 167, 217

Moran, T., Foukal, P., and Rabin, D. 1992, Solar Phys. 142, 35

Moreno-Insertis, F., Emonet, T. 1996, ApJ 472, L53

November, L., Koutchmy 1996, ApJ 466, 512

Oka, T. 1997,Science 277, 328

Orrall, F. Q., Schmahl, E. 1981, Solar Phys. 50, 365

Parker, E. N. 1963, ApJ 138, 552

Parker, E. N. 1978, ApJ 221, 368

Parker, E. N. 1981, ApJ 244, 644

Parker, E. N. 1996, inMHD Phenomena in the Solar Atmosphere, eds. Y. Uchida, T. Kosugi, H. Hudson,

( Dordrecht: Kluwer), 399

Parnell, C. E., Priest, E. R., Golub, L. 1994, Solar Phys. 151, 57

Paxman, R. G., Seldin, J. H., Loefdahl, M. G., Scharmer, G. B., Keller, C. U. 1996, ApJ 466, 1087.

Polyansky, 0. L., Zobov, N. F., Viti, S., Tennyson, J., Bernath, P., and Wallace, L. 1997, Science 277, 346

Porter, L., Klimchuk, J. 1995, ApJ 454, 499

Priest, E. R. 1994 inPhysics of Solar and Stellar Coronae, eds. J. Linsky and S. Serio, ( Dordrecht: Kluwer), 515

Priest, E. R., van Ballegooijen, A. A., MacKay, D. 19967 ApJ 460, 530

Rabin, D., Jefferies, J. and Lindsey, C., eds. 1994, Infrared Solar Physics (IAU Symp. 154) ( Dordrecht: Kluwer)

Rast, M ' P. 1997, in Solar Convection and Oscillations and their Relationship, eds. F. Pijpers, J. Christensen-Dalsgaard, C. Rosenthal ( Dordrecht: Kluwer), 135

Rüedi, I., Solanki, S. K., Rabin, D. 1992, AA 261, L21

Rimmele, T., Lin, H. 1997, BAAS 29, 901

Roggemann, M. C. 1991, Appl. Opt. 30, 4227

Sams, B. J., Golub, L., Weiss, N. 0. 1992, ApJ 3997 313

Schrijver, C. J., Title, A. M., van Ballegooijen, A. A., Hagenaar, H. J., Shine, R. A. 1997, ApJ 487, 424

Skumanich, A., Lites, B. W., Martinez Pillet, V., Seagraves, P. 19977 APJS 110, 357

Smartt, R. N., Zhang, Z., Smutko, M. F. 1993, Solar Phys. 148, 139

Solanki, S. 1994, inInfrared Solar Physics (IAU Symp. 154), eds. D. Rabin et al. ( Dordrecht: Kluwer), 393

Steiner, O., Knölker, M., Schfissler, M. 1997, in Solar Magnetic Fields, ed. V. H. Hansteen ( Oslo: U. Norway), 31

Stenflo, J. 0. 1994, Solar Magnetic Fields ( Dordrecht: Kluwer), 20

Sturrock, P., Uchida, Y. 1981, ApJ 246, 331.

Thomas, J. H. 1988, ApJ 333, 407

Title, A. 1996 BAAS 28, 933

Topka, K. P., Tarbell, T. D., and Title, A. M. 1992, ApJ 396, 351

Topka, K. P., Tarbell, T. D., and Title, A. M. 1997, ApJ 484, 479

Tsubaki, T. inSolar and Stellar Coronal Structure and Dynamics, ed. R. Altrock ( Sunspot: NSO), 14

Vekstein, G. E. 1996, in MHD Phenomena in the Solar Atmosphere, eds. Y. Uchida, T.

Kosugi H., Hudson, ( Dordrecht: Kluwer), 29

Wallace, L., Bernath, P., Livingston, W., Hinkle, K., Busler, J.,Guo, B., and Zhang, K. 1995, Science 268, 1155

Wallace, L., Livingston, W., and Bernath, P. 1994, An Atlas of the Sunspot Spectrum from 1233 cm^-1 470 t (8.1 to 21 µm) and the Photospheric Spectrum from 460 to 630 cm^-1 (16 to 22 µm) ( Tucson: NSO)

Zirker, J. B. 1993, Solar Phys. 148, 43

Zirker, J. B. 1994, BAAS 26, 1378

Figure 1. The time required for coronal 3σ measurements of two different line-of-sight magnetic field strengths as a function of target brightness. Values are shown for two different field strengths and two aperture sizes. The spectrum line is Fe XIII 1074.8 nm and the sampling aperture is one arc sec square. A sky background of 10 × 10^-6 and an overall efficiency of 0.1 are assumed. A loop lifetime of 1000 seconds enforces a need to use large apertures to reach expected magnetic field strengths. [NOTE: Figure not included with this reprint.]