22 September 97. Solar observations in the radio part of the electromagnetic spectrum provide a unique perspective

Transcription

1 TOWARDS A FREQUENCY-AGILE SOLAR RADIOTELESCOPE D.E. Gary (NJIT), T.S. Bastian (NRAO), S.M. White (UMd) 22 September 97 1 Introduction Solar observations in the radio part of the electromagnetic spectrum provide a unique perspective on virtually all phenomena in the solar atmosphere. Over the last twenty years the radio community has developed a detailed understanding of solar radio emission and its interpretation. However, the solar-specic capabilities of the radio facilities available in the US have not kept pace with our progress in understanding radio emission. The radio technology, image processing techniques, and understanding of radio diagnostics are now mature. A consensus exists among the international solar radio community, as expressed at an NSF and NASA supported Solar Radio Telescope Workshop held in San Juan Capistrano in 1995, concerning the future directions that must taken. It is therefore possible and timely to take advantage of the unique diagnostic potential oered by the radio regime through the design and construction of an advanced solar-dedicated radiotelescope. In these notes we highlight the more important capabilities of a proposed Frequency-Agile Solar Radiotelescope (FASR) and outline the outstanding science problems it will address. These include: Determining the quantitative structure of the coronal magnetic eld by performing coronal magnetography in and near active regions. Determining the thermal structure of the quiet solar atmosphere from low-chromospheric to coronal heights with arcsecond angular resolution. Elucidating the electron energy distribution, resolved in space and time, of both thermal and nonthermal electrons associated with energetic transient phenomena ares, CMEs, and lament eruptions through imaging and spectroscopy with a high temporal cadence. Probing the very smallest energy release events and their relationship to coronal heating and solar wind production and acceleration. 2 Radio Diagnostics Radio emission in the wavelength regime spanned by the FASR { roughly cm { is particularly rich in diagnostic potential. Both coherent and incoherent emission mechanisms occur 1

2 in this wavelength range, as do both thermal and nonthermal emissions. Some radio emissions transition from optically thick to optically thin over this wavelength range, thereby increasing their diagnostic value. The key to tapping this diagnostic potential is to obtain high resolution radio brightness temperature spectra. Brightness temperature spectra can be obtained only by imaging at a large number of frequencies with sucient spatial resolution to resolve the sources. While a number of emission mechanisms play a role in various solar phenomena, the dominant emission mechanisms and their diagnostic potential are briey as follows: Free-free radiation: Dominant in most of the quiescent solar atmosphere, free-free radiation at radio wavelengths (where the Rayleigh-Jeans approximation applies) provides a simple linear thermometer for probing the thermal structure of the solar atmosphere. The relevant temperature range is 8000 K to 4 MK or higher. In the presence of magnetic elds, optically thin free-free radiation is weakly to moderately circularly polarized and constraints may be placed on the longitudinal component of the magnetic eld in the upper chromosphere and corona. Gyroresonance radiation: Dominant in solar active regions, optically thick gyroresonance radiation provides a clean measure of temperature as a function of coronal magnetic eld strength. As its name suggests, gyroresonance radiation is due to the cyclotron resonance at the rst few harmonics of the gyrofrequency, Be = eb=2m e c 2:8B MHz, with B in gauss. The emission may be strongly circular polarized, and spatially resolved measurements of the spectral and polarization characteristics of the emission form the basis for a true coronal magnetograph, as discussed below. Gyrosynchrotron radiation: Dominant during ares and other activity, gyrosynchrotron radiation provides numerous diagnostics of energetic electrons and the ambient medium. During ares, electrons are promply heated to energies of 10s of kev, and/or accelerated to much higher energies, in some cases to 100 MeV. Energetic electrons produce continuum emission at higher harmonics of the electron gyrofrequency, typically 10s to 100s of times Be. The emission is moderately circularly polarized, and again spatially resolved measurements of the spectral and polarization characteristics form the basis for quantitative measurements of coronal magnetic eld strength and the energy distribution function of both thermal and nonthermal electrons. Plasma radiation: Dominant at decimetric ( 10, 100 cm) and longer wavelengths, plasma radiation provides a sensitive means of detecting the presence of electron beams (type III bursts and varients) in the corona. Because the emission is coherent, its intensity is not simply related to temperature or numbers of electrons, however. Plasma emission provides good density and density-gradient diagnostics through its generation p at the fundamental or second harmonic of the electron plasma frequency, pe = n e e 2 =m e 9000 n 1=2 e Hz. 2

3 60 Magnetic field lines of force Height [arc sec] Optical image 0 Radio emission layers: 5 GHz 8 GHz 11 GHz Horizontal distance [arc sec] Figure 1: This gure shows optical and magnetic eld data for an active region observed on 1991 May 07. A perspective view of a complex sunspot group in optical continuum is shown with eld lines extrapolated into the corona using a nonlinear force-free extrapolation by Z. Mikic. The three surfaces are the calculated gyroresonant surfaces in the corona that will dominate the radio opacity ateach of three radio frequencies: 5 GHz (B = 600 G), 8 GHz (B = 950 G) and 11 GHz (B = 1300 G). The optically thick layers are only 100 km thick. By observing at a range of progressively greater frequencies, which are sensitive to increasingly larger magnetic eld strengths, we view the corona in thin, progressively deeper layers. 3 Science Objectives Given the diagnostic potential of the decimetric and microwave regimes, an extremely broad program of science studies in both the corona and chromosphere will be supported by the FASR, due to its ground-breaking combination of superb image quality and spectral coverage. Here we highlight several key topics. 3.1 Coronal Magnetography Quantitative knowledge of coronal magnetic elds is crucial for virtually all solar physics above the chromosphere, including the structure and evolution of active regions, ares, laments, coronal heating, and coronal mass ejections. Considerable resources are devoted to the measurement of photospheric magnetic elds and extrapolation of the measured source surface into the corona. 3

4 However, photospheric magnetograms are not perfect (saturation in sunspot umbrae, limitations on temporal, spatial or spectral resolution, ambiguities due to the physics of line-formation, etc.) and when extrapolated into the corona, such imperfections are magnied to an extent that is largely unknown, since there is currently no way to quantitatively check the results. It will help to have chromospheric eld measurements using new infra-red (IR) techniques, but clearly it is important to have a technique for direct measurement of magnetic eld strength in the corona that can be used to complement these measurements from lower heights to guide and give a check on extrapolations, and to uncover new physics. An important goal of the FASR design is to fulll the potential of radio observations, demonstrated by a large body of previous research (e.g., see White and Kundu 1997), to carry out genuine coronal magnetography of active regions. There are two approaches to such measurements: (1) a well-established method (e.g. Hurford and Gary 1986; Gary and Hurford 1994; Lee et al. 1997a) which gives the 2D eld strength at a single surface at the base of the corona; and (2) a still unproven method (e.g., Lee et al. 1997b) which exploits the 3D information present in the radio data to determine as much information about the 3D structure of B, n, and T in the active region corona as possible D Coronal Magnetography On at least one surface in the corona, at its base, magnetography of active regions with FASR will be straightforward. The outer boundary of an optically thick gyroresonant source (the outer edge of one of the surfaces shown in Fig. 1) corresponds to the line where the surface drops into the chromosphere and accordingly the observed temperature drops sharply. Thus the outer edge of the optically thick radio source at a given frequency denes a constant-eld-strength contour on the base of the corona. As the frequency changes, so does the corresponding value of the eld strength whose boundary is delineated. This technique is already routinely carried out with the VLA, but at limited numbers of frequencies corresponding to a few xed values of B. FASR will measure magnetic strengths at the base of the corona running continuously from 200 to > 2000 G. As might be expected, the outer boundary of a source at a given frequency can be aected by other complications such as marginal optical depth due to unusual structure in temperature, density, or inclination of B in the active region. However, such unusual structure can easily be distinguished and corrected or avoided when the data are viewed not as a sequence of images at dierent frequencies, but rather as a spatially resolved set of gyroresonance spectra. This is the essential factor in FASR's vast improvement over a limited-frequency instrument like the VLA, and it will allow FASR to provide routine, accurate 2D coronal magnetic eld measurements at the base of the corona. The ease with which this measurement can be made will allow near-realtime coronal magnetograms to be produced throughout each observing day, allowing the observer to not only measure the magnetic eld at the base of the corona, but to track its evolution through a period of hours and from day today D Coronal Magnetography The radio measurements provide not just the outer boundaries of the surfaces shown in Fig. 1, but also the brightness temperature distribution all along the surface within those boundaries. 4

5 This represents additional information on the distribution of coronal temperature, density, magnetic eld strength and orientation (e.g., Vourlidas, Bastian and Aschwanden 1997). We are aided in the interpretation of this information by the fact that the emission arises on clearly de- ned, nested surfaces (the variously colored, hatched surfaces shown in Fig. 1) which uniformly and monotonically range from low to high coronal heights as frequency decreases. This property of gyroresonance emission provides a very dierent perspective compared to an optically thin diagnostic such as X{rays, where all the emission along a given line of sight isintegrated and projected onto the sky plane regardless of its height. The challenge in interpreting gyroresonance emission is to sort out the inuences of the various parameters, which will require sophisticated modelling of the spatially resolved spectra. Obviously, we will be aided greatly in this goal by using complementary information from (i) magnetic eld extrapolations (perhaps now based on three heights at the photosphere [optical], chromosphere [IR], and base of the corona [microwaves]), and (ii) temperature and density constraints from soft X-rays, EUV, and other means. The essential fact in this work is that the microwave spectral intensity and polarization information provide a wealth of highly specic constraints to guide the modelling. Although the height corresponding to each frequency is not known a priori, itisknown that the emission is on thin layers ( 100 km thick) of known eld strength arranged in rigorous height order. The dependence of gyroresonance opacity onb, n and T is well understood. The ultimate goal would be to obtain unique, 3D models of B, n and T in the corona over active regions which satisfy the FASR images (at every frequency and each oftwo polarizations) and the complementary data. Prescriptions for doing so have not yet been developed, but we emphasize that, because of their unique optically{thick perspective, radio observations of gyroresonance (GR) emission are likely to be an essential ingredient inanyalgorithm for the derivation of complete 3D coronal models. They complement soft X{ray (SXR) observations in several ways: (i) SXR emission is proportional to n 2 and thus emphasizes high{density structures, whereas GR opacity is directly due to the magnetic eld and relatively weakly dependent on density, thereby emphasizing isogauss surfaces; (ii) SXR emission is optically thin and therefore contains information on the whole volume, but suers from line{of{sight projection. Even if one could identify a suitable coronal line in which the magnetically-sensitive components can be measured, and if a stereoscopic space mission (at considerable expense) is launched to provide a second perspective, one would still have to solve a formally ill-posed inversion problem in order to obtain 3D information, whereas the GR data, being optically thick, provide data on a sequence of surfaces stepping through the coronal volume as frequency changes. Whether the ultimate goal of true 3D magnetography can be achieved cannot be answered at present, but if it is to be achieved it will require an instrument with the imaging quality and frequency agility offasr; it cannot be accomplished with current radio facilities. 3.2 Thermal Structure of the Solar Atmosphere In weak eld regions where thermal gyroresonance emission is absent, radio emission remains sensitive to thermal structure through thermal free-free emission. As is the case for observations in CO, microwave radiation is formed under conditions of LTE; the source function is therefore Planckian. For microwave observations, h k B T so the Rayleigh-Jeans approximation is valid, 5

6 Figure 2: 17 GHz map ( =1:76 cm; right) and a He 304A map (left) of the quiet Sun observed on 6 July 1996 by the Nobeyama radioheliograph and the SOHO EIT, respectively. and the observed intensity is linearly proportional to the kinetic temperature of the emitting material for optically thick sources. The optical depth is ff /,2 n 2 T,3=2. By varying the frequency from 1{26 GHz, one samples the thermal state of optically thick plasma from the mid-chromosphere to the low-corona. The broadband imaging capability of the FASR will be exploited to probe the thermal structure of the solar atmosphere in active regions, the quiet Sun, and coronal holes, as well as in laments and prominences. We cannot discuss all possibilities here { however, by way of example we briey discuss the capabilities of FASR in elucidating the structure of the quiet Sun atmosphere. The chromosphere has been the subject of lively debate in recent years. It has become increasingly apparent that the prevailing semi-empirical chromospheric models, largely based on non-lte UV/EUV line and IR/submm/mm continuum observations and computed under the assumption of hydrostatic equilibrium, are in stark disagreement with CO and microwave observations. In particular, observations of the CO molecule near 4.7m show that the lowchromosphere contains a substantial amount of cool (3800 K) material leading to the view that the chromosphere is fundametally bifurcated between cool and hot material (e.g., Ayres and Rabin 1996). Accurate broadband microwave (1{18 GHz) spectroscopy of the quiet Sun (Zirin, Baumert, and Hurford 1991) convincingly demonstrate that the prevailing semi-empirical models include an over-abundance of warm chromospheric material (Bastian, Dulk, and Leblanc 1996). These developments have caused the solar community to begin to re-think the solar chromosphere. Schematic multi-component models have been proposed which emphasize the pervasive cool component in the solar atmosphere (e.g., Avrett 1995; Ayres and Rabin 1996). Another 6

7 approach has recognized that chromospheric dynamics play a critical role in understanding the structure of the chromosphere. Numerical models (Carlsson and Stein 1995), in which the chromosphere is permeated with a ux of dissipative waves, demonstrate that the mean physical conditions within the chromosphere may dier substantially from those derived from the same models by conventional non-lte UV/EUV line diagnostics. Modern chromospheric models require spatially and temporally resolved observations of the thermal state of the chromosphere on the relevant spatial and temporal scales. A key advantage of the CO lines is their narrow contribution function which allows ne vertical discrimination in the structure of the low chromosphere. The microwave contribution function (thermal free-free) is much broader and is therefore a cruder discriminant. However, CO dissociates at a temperature of 4000 K. It therefore provides the thermal \footprint" of the chromosphere. Spatially resolved microwave observations spanning the 1{26 GHz will map the thermal structure of the solar atmosphere from chromospheric heights on up into the corona. Microwaves are therefore a valuable complement to CO and other observations of the solar chromosphere. The FASR design will allow us to sample the thermal structure of the chromosphere down to the height where T e 8000 K with a frequency-dependent angular resolution of The sensitivity of the FASR, as presently conceived, will allow us to study the time variability of the thermal structure of the solar chromosphere in a single frequency band on a timescale 5 min (T B < 100K). Over a period of several hours, the FASR will provide high quality maps of the mean thermal state of the chromosphere over its entire frequency range, corresponding to a range of heights spanning the mid-chromosphere to the low-corona. FASR observations will therefore provide a comprehensive specication of the thermal structure of the chromosphere in coronal holes, quiet regions, enhanced network, plages as an input for modern models of the inhomogeneous and dynamic chromosphere. 3.3 Flares and Coronal Mass Ejections The FASR will, for the rst time, allow full exploitation of the microwave/decimetric emission for are studies. The possibilities are numerous and exciting: Microwave emission in ares is due to incoherent gyrosynchrotron emission from electrons with energies of several 10s of kev to a few MeV. The spectrum of emission typically peaks at frequencies of 5-15 GHz, although some events peak at even higher frequencies. The peak frequency pk at a given location depends sensitively on the local magnetic eld strength and the angle between the line of sight and the magnetic eld vector. Joint observations of pk and the source polarization will allow the magnetic eld strength and orientation to be inferred for the aring source as a function of time. No other technique is available for this purpose. Magnetic eld in the aring volume: Electron acceleration and transport: The microwave spectrum is a powerful diagnostic of the details of the emitting distribution of energetic electrons. The optically thick part of the spectrum establishes the mean energy of the emitting electrons at each location. The optically thin part of the spectrum establishes whether the emitting distribution is thermal or nonthermal, and if the latter, what the spectral index of the spectrum is. The optically thin part of the spectrum is also sensitive to high energy cutos in the spectrum 7

8 Figure 3: Gyrosynchrotron emission from a model coronal magnetic loop. a) A representation of the magnetic eld lines emanating from two solenoids: one has a eld strength of G; the other a eld strength of +500 G. The black lines of force demarcate the FWHM of the electron number density of nonthermal electrons; b) Brightness temperature spectra of the resulting gyrosynchrotron emission at both magnetic footpoints and at the looptop. The frequency in GHz is indicated along the x-axis; c) The brightness distribution of the Stokes I parameter (contours) and the degree of polarization (greyscale) at 2.9 GHz. The two have been oset from each other for clarity; d) same for 4.7 GHz; e) same 8.6 GHz; f) same for 17.7 GHz. From Bastian, Benz, and Gary

9 and to anisotropies in the distribution function. It is also worth pointing out that the relative timing of temporal features at dierent frequencies oers an additional diagnostic of acceleration and transport. Electron trapping is believed to play a role in most ares. The details of the relative delays and the emission decay time at each frequency allow one to constrain the ambient density (in cases where electron trapping is mediated by Coulomb collisions) or whether wave-particle interactions are relevant, with ne spatial resolution. Multitudes of bi-directional type III bursts those which result from bidirectional electron beams accompany the impulsive phase of many ares (Aschwanden et al. 1995). The frequency that demarcates positive and negative drifts in the type III also demarcates the plasma frequency (and therefore the density) where the electrons were initially accelerated. While spectroscopic observations of bi-directional type IIIs during ares have been performed, none have been imaged. The FASR will identify the location of these bursts, presumably intimately related to the primary energy release, and trace their trajectories both upward and downward in the aring volume. In so doing, it will also trace the density structure of the energy release site and its immediate surroundings. Location and properties of the energy release site: In addition to diagnosing the magnetic eld and the details of the energetic electron population, radio observations oer a means of probing the density and evolution (due, for example, to chromospheric ablation) of the ambient plasma. The optically thick side of the microwave spectrum is modied by Razin suppression, due to gyroemission in the presence of a background plasma. The specics of the suppression depend on the density of the ambient plasma and the local magnetic eld strength. The Razin eect oers a means of probing the ambient density and its evolution in time throughout the aring volume. Chromospheric ablation: The FASR will also be used for CME detection. Extensive simulation studies have been performed by Bastian and Gary (1997) and show that the decimeter range is optimum for CME detection at radio wavelengths. Bastian et al. (1998) have initiated a pilot program to demonstrate the \proof of concept" using new receiver systems at the VLA in conjunction with the LASCO instrument on board SOHO. CME detection relies on dierential techniques { that is, dierencing data from dierent times to remove background emission. The advantages of CME detection at radio wavelengths are that i) there is no occulting disk, so earth-directed CMEs will be detected; ii) CMEs will be detected in their nascent stages of development; iii) CMEs can be directly associated with structures such as active regions or lament channel arcades; iv) unlike SXR and whitelight observations, observations at radio wavelengths are sensitive to both thermal free-free emission from CMEs and possible nonthermal constituents. Gyrosynchrotron emission would allow constraints to be placed on the strengths of magnetic elds entrained in the CME. Another of the important strengths of the FASR, however, is that owing to its frequency agility it could provide a comprehensive observational picture of CMEs and associated phenomena over the entire decimetric and microwave band of frequencies. For example, in addition to imaging the radio counterpart of a CME through the thermal bremsstrahlung radiation it emits, the FASR could also observe, simultaneously and with CME detection: 9

10 Figure 4: The GOES M8.7 are of 17 June 1989 during the rise phase, as observed by the VLA at a wavelength of 6 cm (contours) and in H (greyscale). The time evolution of the radio emission shows the strongly magnetized footpoint(s) emitting rst (upper-right), followed by the conjugate footpoint(s) (upper-left). The emission then bridges the magnetic neutral line (lower-right). Near the time of maximum an arcade of magnetic loops bridging the neutral line are bright at 6 cm(lower-right). perfect coregistration, microwave emission from slow-moving prominence material, from an associated are and, in some cases, the high frequency extensions of associated bursts of type II or type IV. As such, the FASR will serve as a powerful tool for disentangling the complex relationship between CMEs, ares, lament eruptions, and radio bursts. 3.4 Microares and Coronal Heating One of the fundamental questions in solar physics is how the solar corona maintains its high temperature, in excess of 1 MK, above a surface with temperature 6000 K. The power needed 10

11 to maintain the corona above an active region against radiative and conductive losses is > erg s,1 (Shimizu 1995). It is possible that \ares" (that is, impulsive releases of energy arising from magnetic reconnection) may supply the needed energy, but Shimizu (1995) has shown that ares of all observable sizes down to erg, when added together, still fall short by almost an order of magnitude. However, as Parker (1988) points out, much of the energy released by magnetic reconnection may take place in numerous, extremely small events (nanoares, roughly erg) that are too small to observe in the corona with current instrumentation. As noted by Hudson (1991), a key parameter is the occurrence rate of such events as a function of energy. Numerous studies (e.g. Crosby, Aschwanden and Dennis 1993; Lee, Petrosian, and McTiernan 1993; Shimizu 1995) have shown that events ranging over as much as ve orders of magnitude in energy, from to erg, form a single powerlaw with slope Smaller events cannot be energetically signicant relative to the larger events unless the rate distribution at lower energies becomes signicantly steeper. To test this hypothesis, we must extend our sensitivity down to events even weaker than the erg brightenings reported by Shimizu. If coronal heating is indeed due to nanoares, we expect to see an upturn in the rate distribution as we approach the nanoare energy range. Even if coronal heating is found not to be entirely due to such impulsive events, we already know that such events are numerous enough to play a role in coronal conditions, and are therefore important to characterize and understand in either case. Radio data are sensitive to small energy releases because energetic electrons are so ecient at radiating in the radio regime. Thus Gary, Hartl and Shimizu (1997) established that the erg events of Shimizu (1995) are accompanied by nonthermal electrons, which searches in hard X{rays had been unable to detect (White et al. 1995). Even events that are near the limit of visibility for SXT typically have radio ux densities which are easily detectable in total power by small non{imaging radio telescopes, and high-quality imaging can lower the ux limit one achieves in microwaves by orders of magnitude. Gopalswamy et al. (1994, 1997) have demonstrated this with the VLA by detecting events over the penumbra of a large sunspot which are two orders of magnitude weaker than those of Shimizu: the inferred energy release rates are as low as ergs s,1, and total energies are not much above Parker's (1988) canonical nanoare value of ergs. Soft X-ray transient brightenings occur in active regions, but the quiet Sun corona must be heated too. Here again microwave observations with sucient spatial resolution have indicated that small brightenings are common. Gary and Zirin (1988) observed radio brightenings in the network with radio uxes similar to those detected by Gopalswamy et al and suggested that the quiet Sun network radio brightness may be dominated by small brightenings too numerous to be imaged with the VLA. This supports the idea proposed by Porter et al. (1984) based on a study of C IV network enhancements, and similar results are being obtained in ongoing work with SOHO EIT (Schrijver et al. 1997). Recently, Krucker et al. (1997) imaged four weak network events in a limited area of the Sun and estimated that they occur once every3sover the whole Sun, each releasing erg. Although this occurrence rate is not sucient to suggest an upturn in the rate distribution, the statistics of small numbers leaves a great deal of uncertainty and the question remains open whether these and smaller events are numerous enough to supply the missing energy needed to heat the corona. In conclusion, tiny events in both active regions and the quiet Sun network are commonly seen 11

12 in microwaves with current instruments. FASR will improve on the state-of-the-art, however, by providing better spatial resolution, frequency coverage and sensitivity than Nobeyama; better frequency coverage and more solar observing time than the VLA. It is a signicant challenge to observe microares down to brightness levels that are interesting for coronal heating, but our current estimate is that FASR will do an excellent job on full-sun imaging and spectra of events down to erg in low-eld regions and to yet lower energies in regions of high magnetic eld, with the chief uncertainty being how the brightness and spatial size of the events scale with total energy. The study of microares with FASR will be aided by the instrument's exibility, which will allow the data to be treated in an optimum way for a given goal. For example, spatial, spectral, and/or time resolution can be traded o for sensitivity at the data analysis stage. This and the instrument's full-sun capability should allow FASR to obtain accurate counting statistics on the occurrence rate of these events, and to determine whether that rate increases greatly enough at low energies to heat the corona. 4 Instrumental Requirements The primary goal is to design and construct an instrument which fully exploits solar microwave emission as a diagnostic of physical processes on the Sun. To this end, a number of instrumental requirements have been identied: 1. Imaging: The sources of radio emission on the Sun must be imaged with high dynamic range, delity, and angular resolution, with good sensitivity to both compact and extended sources of emission. A dynamic range of order 1000:1 and angular resolution of 1 00 at a frequency of 20 GHz are considered reasonable goals. 2. Broadband spectroscopy: The brightness temperature spectrum as a function of position is required for both decimetric and microwave phenomena. Hence, spectroscopic coverage over the frequency range of 300 MHz to > 26:5 GHz with a spectral resolution of < 5% is needed. 3. Polarimetry: Observations in both the right- and left-hand senses of circular polarization are needed to form the Stokes I and V polarization parameters. 4. High time resolution: Brightness temperature spectra must be acquired at a rate sucient to resolve the timescale on which the phenomenon of interest evolves. The most demanding requirement is imposed by the impulsive phase of ares, which will require a time resolution of < 1 sec. 5. Large eld of view: In the interest of maximizing observing eciency and in matching the capabilities of many full disk spectrographs and imagers, a full disk imaging capability is desired at most frequencies. 6. Good absolute positional accuracy: Instruments in most wavelength bands now possess an angular resolution ranging from less than 1 arcsec to several arcsec. Quantitative crosscomparisons between various wavelength regimes will require absolute source positions to a similar accuracy. 12

13 Figure 5: Cartoon of the \strawman" FASR array 7. Upgradability: The instrument should be designed with future enhancments in mind. Examples include support of very high-time- and high-frequency- resolution observations of decimetric phenomena and the extension of frequency coverage to millimeter wavelengths. 8. Easy access by the solar community: The instrument should not place the burden of reduction on the user. All reductions should be performed on-site and a wide variety of data products should be made available for immediate and open use by the community at large. 5 The FASR Design These requirements lead us to propose the following strawman design. High angular resolution imaging requires an instrument which employs Fourier synthesis imaging using an interferometric array of antennas. The requirements of high dynamic range and image delity, and of high 13

14 sensitivity to a wide range of angular scales require a large number of antennas to properly sample the Fourier (or u-v) plane. Preliminary simulations suggest 40 antenna elements, providing measurements of 800 Fourier components instantaneously. An angular resolution of 1 00 at a frequency of 20 GHz requires a maximum antenna baseline of 3 km while good sensitivity to extended emission will require adequate numbers of short antenna baselines. It is anticipated that frequency synthesis techniques will also be exploited (e.g., Bastian 1989), perhaps in the form of spatial/spectral image reconstruction (Komm, Hurford and Gary 1997). Logarithmic antenna spacings will accomodate frequency synthesis, and allow identical u-v coverage and angular resolution at selected frequencies due to the scaling of the coverage with frequency. For a factor 1.5 spacing of antennas along each arm, for example, each factor of 1.5 in frequency will have identical u-v coverage, e.g. frequencies 3.3, 5, 7.5, 11.25, 16.9, and 25.3 GHz will all have identical u-v coverage on a subset of antennas. The requirement of full disk imaging at most frequencies suggests the use of small antenna elements. The strawman design calls for apertures of 2 m. On the other hand, the need for good absolute calibration implies a need for astronomical calibration; i.e., sucient sensitivity is needed to compare the source position with that of known cosmic reference sources. This, in turn, implies a need for large, sensitive antennas. Hence, in addition to many small antennas, the strawman design calls for at least one large antenna. The construction of one or more new 25 m antennas would be prohibitively expensive. An aordable alternative is to refurbish existing antennas e.g., one or more of the two27mantennas at Owens Valley or the three 25 m antennas at Green Bank. These will be used to calibrate the instrument against cosmic standards and could also support high-resolution decimetric spectroscopy, and perhaps nighttime observing. The requirement of broadband spectroscopy in the decimetric and microwave bands suggests a frequency-agile design. The technology for the detection, transmission and amplication of broadband microwave signals is mature, and o-the-shelf components are available at low cost in several bands of interest (e.g., 1-8 GHz, 8-18 GHz, GHz, etc.). A match to commercially available microwave bands is therefore desirable, and an initial core operating frequency of GHz is called for in the strawman design. Both high- and low-frequency extensions are feasible: to 300 MHz at low frequencies, and up to 40 GHz and/or the GHz band at high frequencies. The need for measurements of the total and circularly polarized radiation requires the use of dual-polarization feeds, receivers, and data transmission elements. High time resolution may appear problematic in view of the large number of baselines and the broadband frequency coverage required to process a bandwidth of GHz for an array of 40 antennas (or 780 baselines) instantaneously would require a very large correlator. However, the entire spectrum need not be sampled instantaneously and the frequency resolution and rate at which spectra are acquired depends on the phenomenon of interest. The high ux levels from the Sun allow an extremely fast sampling rate (10 msec) with good signal to noise. Thus, the strawman design assumes that 500 MHz wide sections of the total bandwidth will be sampled sequentially; i.e., that frequency multiplexing will be performed. The 500 MHz IF bandwidth would itself be coarsely divided into 16 channels of 32 MHz bandwidth to provide the necessary spectral resolution and to prevent bandwidth smearing of the correlated signals. With the wide instantaneous IF bandwidths (> 500 MHz) now available, the entire 1-26 GHz band would be covered with 50 samples in <1 sec. In other words, during ares, a set of 50 maps could be 14

15 produced in less than 1 s. It would be wise to design the IF/LO system and correlator with sucient exibility that the specic frequencies sampled, their number, and the rate at which they are sampled, are selectable parameters. In this way the instrument can be used most eectively for imaging spectroscopy of a wide variety of transient phenomena. Finally, we come to operational considerations. It is anticipated that the instrument will produce a large number of data products of great value to the research and space environment forecasting communities. These will be available rain or shine, and include: 1. Real-time precision ux spectra of the sun as a star. 2. Near real-time data cubes which display the brightness temperature in two dimensions from chromospheric to coronal heights. 3. Near real-time coronal magnetograms which would provide the magnetic eld strength at the base of the corona in active regions. 4. A real-time catalog of ares - their locations, morphology, and a number of indices characterizing their radio spectrum and its evolution. 5. A real-time list of prominence eruptions and CMEs. Of course, many research projects will require complete and detailed analyses of the data. For example, are studies will require deconvolved maps for every frequency and integration time. These reductions will be largely automated, but will likely be performed oine and then archived. Users will be able to access the archive in a exible way. Sophisticated users may wish to work with raw data. The means will be provided to do so. While a broad program of science will be supported by routine operation of the facility, some users may wish to perform specialized observations (e.g., extremely high-time- or high-spectralresolution studies). Every eort will be made to design the instrument and its operation for maximum exibility. 6 Summary It is now possible to design a powerful solar-dedicated radio telescope to perform broadband imaging spectroscopy in the decimeter and centimeter wavelength bands. Such a telescope would attack a broad range of scientic problems, from the ground, at modest cost. The instrument concept outlined above represents a distillation of many conversations and ideas from members of both the radio community and the broader solar community. These ideas were studied in some detail at an NSF and NASA supported Solar Radio Telescope Workshop, held in San Juan Capistrano, CA in 1995, and attended by about 40 US and international experts in solar radio physics. The Frequency-Agile Solar Radiotelescope concept thus represents a consensus opinion on the question of which direction groundbased solar radio astronomy can and must go in the near future. The proposed instrument represents a major improvement on current radio facilities by providing the following capabilities: 15

16 It will be solar-dedicated. It will provide imaging far superior to that of any current instrument. It will perform broadband spectroscopy a spectrum will be available for each resolution element in the full-sun eld of view. It will provide \ready to use" data products to the community at large for both basic research and forecasting purposes. It will be a unique instrument that will provide an entirely new view of the Sun, with the potential for many new discoveries. 16

17 REFERENCES Aschwanden, M.J., Benz, A.O., Dennis, B.R., and Schwartz, R.A. (1995): ApJ, 455, 347 Avrett, E.H. (1995): in Infrared Tools for Solar Astrophysics: What's Next?, Proceedings of the 15th NSO/Sacramento Peak Workshop, J. R. Kuhn and M. J. Penn (eds.), p Ayres, T.R., and Rabin, D. (1996): ApJ, 460, 1042 Bastian, T.S. (1989): in Synthesis Imaging in Radio Astronomy, ASP Conf. Series, v. 6, R.A. Perley, F.R. Schwab, & A.H. Bridle (eds.), Provo: Brigham Young University Print Services Bastian, T.S., Dulk, G.A., and Leblanc, Y. (1996): ApJ, 473, 539 Bastian, T.S., and Gary, D.E. (1997): JGR, 102, Bastian, T.S., Benz, A.O., and Gary, D.E. 1998, ARAA Carlsson, M., and Stein, R.F. (1995): ApJ, 440, L29 Crosby, N., Aschwanden, M., and Dennis, B.R. (1993): Solar Phys, 143, 275 Gary, D.E., and Zirin, H. (1988): ApJ, 329, 991 Gary, D.E., and Hurford, G. J. (1994): ApJ, 420, 903 Gary, D.E., Hartl, M., and Shimizu, T. (1997): ApJ, 477, 958 Gopalswamy, N., et al. (1994): ApJ, 437, 522 Gopalswamy, N., Zhang, J., Kundu, M. R., Schmahl, E. J., & Lemen, J. R. (1997): ApJL, submitted Hudson, H. (1991): Solar Phys, 133, 357 Hurford, G. J., and Gary, D.E. (1986): in Coronal and Prominence Plasmas, ed. A.I. Poland (NASA CP-2442), 319 Komm, R., Gary, D.E., and Hurford, G. J. (1997): A&A, 122, 181 Krucker, S., Benz, A.O., Bastian, T.S., and Acton, L.W. (1997): ApJ, in press. Lee, T.T., Petrosian, V., and McTiernan, J.M. (1993), ApJ, 412, 401 Lee, J.W., White, S.M., Gopalswamy, N., and Kundu, M.R. (1997a): Solar Physics, in press. Lee, J.W., McClymont, A.N., Mikic, Z., White, S.M., and Kundu, M.R. (1997b): Solar Physics, submitted. Parker, E.N. (1988): ApJ, 330, 474 Porter, J.G., Toomre, J., and Gebbie, K.B. (1984): ApJ, 283, 879 Shimizu, T. (1995): PASJ, 47, 251 C.J. Schrijver, R.A. Shine, N.E. Hurlburt, T.D. Tarbell, and J.R. Lemen, 1997, \The Dynamic Quiet Solar Corona: 4 Days of Joint Observing with MDI and EIT", in Conference Proceedings ESA SP-404, in press. Vourlidas, A.V., Bastian, T.S., and Aschwanden, M.J. (1997): ApJ, Nov. 1 issue White, S.M., Kundu, M.R., Shimizu, T., Shibasaki, K., and Enome, S. (1995): ApJ, 450, 435 White, S.M., and Kundu, M.R. (1997): Solar Phys., in press Zirin, H., Baumert, B.M., and Hurford, G.J. (1991): ApJ, 370,