The Solar Corona at Totality

Home , Coronal hole, Helmet streamer, Wind (spacecraft)

THE SOLAR CORONA AT TOTALITY

Airborne experiments probe the dynamic conditions of the sun’s atmosphere during a total solar eclipse.

n February 16, 1980, the had an unobstructed overhead view of measurements are needed to resolve two moon’s shadow raced across the eclipse for 7 minutes 7 seconds while major questions in coronal physics. How o our planet from the Atlantic observers at the main ground site in is the corona heated? And how is the Ocean to China at about 1600 kilo- India saw the event between clouds at a solar wind generated? meters per hour (Fig. 1). Thousands of low angle in the afternoon sky for only 2 We will look at what some of these spectators and scientists scattered along minutes 9 seconds. measurements are and how they tit into the shadow’s path lifted their heads to This exceptional high-altitude plat- study of the corona. We will also de- glimpse the pearly glow of the solar form outfitted with state-of-the-art in- scribe the technology developed at Los corona, the sun’s outermost atmosphere. struments for tracking control and data Alamos and how it has been used in Near the midpoint of this path where acquisition (many of which derived from carrying out these missions. sun, moon, and earth were aligned for and contributed to the weapons testing the longest time, a high-altitude United program) has enabled Los Alamos to Early Coronal Studies States Air Force jet made its carefully achieve many firsts in observation of programmed rendezvous with the eclipse coronal temperatures and densities. Such Total solar eclipses have been ob- shadow. Aboard the aircraft were 18 Los Alamos scientists, observers from Indiana University and Kitt Peak Na- tional Observatory, a 12-man Air Force flight and support crew, a Kenyan observer, and two media representatives. We were intent on recording the conditions in the solar corona during the second highest peak in solar activity in the past century. Five experiments, planned months ahead, were mounted inside the aircraft to make high- resolution measurements of coronal morphology, electron densities, electron temperatures, ion temperatures, and cir- cumsolar dust rings. Our expedition was the seventh in a series of Los Alamos airborne missions that began in 1965 as an outgrowth of the national Test Read- iness Program.* During a total eclipse the aircraft provides near ideal conditions for observing the solar corona. At an altitude of 11 kilometers (36,000 feet) we enjoy freedom from cloud cover and reduced Fig. I. The path of the moon’s shadow first touched the earth in the Atlantic Ocean, sky brightness and, by flying along the then crossed Africa, the Indian Ocean, and India, and finally left the earth over eclipse path, we increase the time of China. Los Alamos observers launched rockets near Malindi, Kenya. The Los Alamos totality by 50% or more. In 1980, we airborne expedition was based in Nairobi and flew to the point of longest totality duration just south of the equator over the Indian Ocean. The main ground-based *See “In Flight: The Story of Los Alamos Eclipse Missions” in this issue. scientific site was near Hyderabad, India.

LOS ALAMOS SCIENCE 5 oronal morphology (depicted densities. Polar plumes extend from the streamer. here out to 3 solar radii) is most coronal holes at the sun’s north and Sunspots, prominences, filaments, and c likely shaped by the sun’s mag- south poles. flares also form in active regions of the netic field, which changes over an During a period of maximum solar sun’s surface. These regions seem to be 1 l-year cycle of activity. activity, as complex local magnetic fields confined to well-defined belts that move During a period of quiescence, coro- develop over much of the sun’s surface, from high latitudes toward the equator nal structure suggests that open lines of the open dipole field lines and, with as solar activity declines. the sun’s magnetic dipole field spread them, coronal holes appear to be restrict- Coronal enhancements and condensa- out from almost all solar regions except ed primarily to the polar regions. Many tions, bright areas near the sun’s limb, those near the equator, where the corona distinctive coronal features are formed are also associated with active regions. is brightest and fairly uniform. These as a result of this increased activity. Their brightness is due to increased regions of open field lines tend to be less Perhaps the most striking feature is particle densities. An enhancement may bright and may qualify as coronal holes, the helmet structure that extends out- persist for months, whereas a condensa- features characterized by drastically re- ward into a long radial streamer at about tion may appear and disappear over a duced x-ray emission, low particle densi- 2 solar radii. The helmet is probably period of hours or days. ty, and possibly low temperature. It is formed by magnetic field loops that rise Changes in coronal morphology dur- known that, during periods of solar from active regions of the photosphere. ing the solar cycle are best portrayed by inactivity, coronal holes extend from the Charged particles are trapped along our image-enhanced photographs of the poles toward the equator and cover these field lines, but near 2 solar radii corona in the note “The Changing Coro- much of the sun’s surface. Presumably magnetic field strengths decrease suffi- na” ❑ solar wind flows strongly from coronal ciently so that particles can drag the field holes, thus depleting particle and energy lines radially outward to form a

6 LOS ALAMOS SCIENCE THE SOLAR CORONA AT TOTALITY

served for all of recorded time and duce such highly ionized states. Also, for as the solar wind. Its existence was probably since humans began to look at atoms to remain highly ionized long confirmed in the early 1960s when satel- the sky, but the origin and nature of the enough to deexcite through these forbid- lites first began to penetrate beyond the sun’s magnificent halo has bathed even den transitions, the density must be very earth’s magnetic field and detected parti- the most astute observers. In 1605 low. Thus the corona is a very hot cles traveling away from the sun at Kepler attributed the display to a lunar plasma of ionized hydrogen, helium, and velocities of 400 kilometers per second. atmosphere. Others associated it with heavier elements with a temperature of a This variable stream of charged particles the atmosphere of the earth or the sun. (mostly electrons and protons) is respon- Not until the 1842 solar eclipse, when a and a density one-trillionth the density of sible for the sharp discontinuity of comet spectacular display of corona and prom- the earth’s atmosphere. That the corona tails as they pass near the sun, a inences was observed by astronomers in should be so hot was quite unexpected phenomenon first noted by L. Biermann several European countries, was there because it overlies the relatively cool in 1951. The solar wind also causes a general agreement on its solar origin. surface of the sun (photosphere), which variety of phenomena as it “blows” past The corona presented many puzzles, is about 6000 K. the earth—from geomagnetic storms the most important being the absence of What are the dynamics of this hot, that interfere with radio communications emission lines from hydrogen and helium rarefied atmosphere? How is it heated? to possible subtle effects on our weather and the presence of the coronal green How does it sustain itself when it is causing climate variations that appear to line, a mysterious emission line that bounded below by the cool photosphere be synchronized with the sun’s magnetic could not be identified with any known and above by the cold vacuum of inactivity cycles. element on the sun. The green line was terstellar space? The answers to these It is interesting that until now meas- observed with a spectroscope by Hark- questions still remain quite tentative. urements of solar wind flow, interpreted ness and Young in 1869. Originally this through solar wind theory, have proba- and other unidentified spectral lines were The Solar Wind bly given as good an idea of the average attributed to a new element named cor- conditions in the inner corona as direct onium. But there seemed to be no place observations. To match solar wind fluxes In the late 1950s the gross dynamics for coronium in the periodic table, so the measured near the earth, these models of the outer corona began to be under- mystery of the spectral lines remained. predict that temperatures in the corona stood. In 1957 Chapman noted that a Then in the late 1930s the lines were simple model of the corona in hy- identified in the spectrum of Nova Pic- decrease very slowly thereafter. In par- drostatic equilibrium leads to a con- toris, an exploding star whose at- ticular, if the temperature T falls off with tradiction. According to such a model, mosphere was known to be very hot. the corona must extend to the earth’s Finally in 1942, the great Swedish spec- orbit and beyond. Its pressure must fall troscopist B. Edlen made laboratory flow begins at a critical distance from the so slowly that at infinity the coronal measurements of hot plasmas and identi- solar surface. Values of b may be esti- pressure would be much too high to be fied nearly all coronal emission lines as mated from observed particle fluxes. balanced by the pressure of interstellar due to “forbidden” atomic transitions Beyond a few solar radii, collisions gas. from highly ionized elements. The green among particles and photons are so rare In 1958 E. Parker solved this problem line was produced by a transition in Fe that ionization levels are “frozen in.” In by introducing the revolutionary and XIV, an iron atom with 13 of its 26 this collisionless plasma, temperatures initially unpopular idea that the corona electrons stripped away.* Other lines refer to an average velocity distribution is not static at all, but instead is in a were attributed to other ionization states and do not imply local thermodynamic constant state of expansion. His model of iron, calcium, and nickel. predicted a supersonic flow of particles Tremendous energy is required to pro- out from the corona beginning at a few *Distances from the center of the sun are com- ● Physicists refer to unionized iron as Fe I. solar radii. This flow came to be known

LOS ALAMOS SCIENCE 7

THE SOLAR CORONA AT TOTALITY

equilibrium. cists since the discovery of the hot magnetic waves are generated all over Temperature measurements needed to corona. By what mechanism is energy the sun’s surface. confirm simple fluid models of solar from the photosphere deposited in the The details of these phenomena are wind flow have been in a state of con- coronal plasma? too complicated to be modeled at pres- fusion. Not only are these measurements Radiation heating by the enormous ent. But there are some questions we can very difficult to make but the corona is a flux of visible light emanating from the hope to answer through observations. complex, inhomogeneous structure that photosphere is ruled out because the First, how far up in the corona is heat changes with time. coronal plasma is essentially transparent deposited? Is it all deposited in the During 1973 and 1974, observations to visible wavelengths. We must look transition zone between the chro- of the x-ray and extreme ultraviolet instead to mechanisms that could be mosphere and the corona where the coronal spectrum from the Apollo Tele- produced by the interaction of mass temperature rises from 0.01 to 1 MK in scope Mount aboard Skylab revealed motions and magnetic fields in the tur- less than 1000 km? Or does the heating many facts about the inhomogeneous bulent convection layers of the pho- mechanism extend out to several solar nature of the solar corona. One of the tosphere (Fig. 2). radii, as suggested by solar wind theory? most important for solar wind theory If so, how is energy transported through was the existence of coronal holes, re- Large volumes of gas constantly the corona? gions of low density, perhaps low tem- emerge from the surface, spread out and perature, and open magnetic field lines cool, and finally sink back into deeper Is energy distributed equally among that are almost devoid of x-ray emission layers of the photosphere. This convec- electrons, or are protons heated prefer- at the solar surface. At the time of the tive energy could generate acoustic entially by nonlinear processes, thus ex- Skylab observations, when the sun was waves that propagate into the chro- plaining why solar wind protons ob- in a quiet phase, the hole regions ex- mosphere and the extreme lower corona. served near the earth have higher tem- tended down toward the equator. The Alternatively, magnetic waves generated peratures? Finally, is solar wind flow a very high velocity component of the by the interaction of convective cells result of temperature gradients alone or solar wind observed during that time with magnetic fields might provide a are there mechanisms that accelerate the evidently emanated from the equatorial heating mechanism that extends far out wind without heating it? into the corona. (Magnetic fields extend coronal hole region where particles If acoustic waves are the major heat outward from the photosphere through stream out at high velocities along open source, the lower coronal temperature field lines. However, the mechanism for the chromosphere and outward with the would be maximum at approximately accelerating these particles as well as the corona.) 1.1 R@ because acoustic waves would contributions to the solar wind from This process might take place in very be damped in the chromosphere or ex- other regions of the corona are still not active regions of the solar surface. In treme lower corona. If magnetic waves well known. these regions of high magnetic field are the major source, the temperature High-resolution data from our 1980 strengths, we see eruptive prominences, maximum would move farther out in the mission will help to determine the tem- filaments, and sun spots. Coronal con- perature structure in coronal holes and densations overlie these active regions. magnetic waves would lose their energy “ other coronal features and thus help to Perhaps as tangled loops of magnetic much more gradually, possibly out as far explain how the solar wind is generated fields move around, currents flow that Direct observations of this in each of these areas. untangle field lines by reconnecting them temperature maximum would obviously in simpler configurations, thereby releas- help to determine the mechanism of Coronal Heating ing large amounts of magnetic field coronal heating. Also, direct observa- energy in the form of magnetic waves. A tions of periodic variations in brightness Our 1980 data may also shed light on similar but more extreme process is would help to establish the existence of a problem that has puzzled solar physi- proposed for solar flares. Or perhaps extended wave heating.

LOS ALAMOS SCIENCE 9 where P is pressure, r is radial distance from the center of the sun in solar radii,

surface, and the material density p is directly proportional to the particle density n. Since coronal densities are low, we assume that the ionized gases obey the ideal gas law,

P = nkT , (2)

which relates the pressure to the particle number n and temperature T. If we assume that the temperature is constant, we can derive the density variation with radial distance by differentiating Eq. (2) and substituting the result into Eq. (l). We obtain

Fig. 3. Approximate variation of particle density and temperature as a function of (3) radial distance. (Note radial distance scale changes.) The steep density decrease above the photosphere and the hundredfold increase in temperature from the base of the Integrating Eq. (3), we find that a plot of chromosphere to the corona are fairly well established but temperatures and gradients log n versus l/r is a straight line with a in the chromosphere and above are still the subject of research. The exact location of slope proportional to l/T. Since the the steep temperature rise in the chromosphere-corona transition region is uncertain, corona is nearly all hydrogen we may although this region is known to be thin. Temperature and density values in the corona replace particle density with electron may vary from one feature to another and may be modified by the presence of n. Consequently, if the prominences, flares, and active regions in the photosphere below. Temperature and experimental values of log n lie on a density of the solar wind at the earth are shown for comparison. e straight line, then the so-called hydrostatic temperature may be deduced from its slope,

Actual measurements of log ne, versus

The Problem of the methods are complementary; if done with a slope corresponding to 1.4 to 1.6 Coronal Temperatures in a coordinated fashion, the separate MK. However in equatorial regions of measurements can be compared to yield the sun the slope changes between 3 and more information than any single one The key parameters for modeling 5 R@ and then beyond 5 R @ it has a energy transport in the corona are the can yield alone. constant value corresponding to slightly distribution of electrons and ions in the HYDROSTATIC TEMPERATURES. up- less than 1 MK. This change in slope corona and their temperature, or, more per and lower limits on average coronal indicates that the corona does not have a precisely, their velocity distribution. temperatures can be obtained from elec- constant temperature. Instead the tem- The average values of the particle tron density measurements. The deriva- perature decreases with distance from distribution are fairly well known, but tions for both limits assume that elec- the sun, a result consistent with an the temperature distribution is not (Fig. trons and ions are in local thermo- expanding rather than a static corona. 3). We would like to know not only gross dynamic equilibrium and electron densi- Nevertheless, hydrostatic temperatures averages but details of temperature dis- ty equals ion density at each point in the are useful as lower limits on average tributions in various structural features corona. The lower limit estimate is based coronal temperature. of the corona since the energy transport on a model of the corona in hydrostatic TEMPERATURES FOR AN EXPANDING will vary from one to another. At present equilibrium (neither expanding nor con- CORONA. To determine temperatures for the results from various methods tracting). The pressure gradient exactly an expanding corona in, for example, a gathered at various times are in disagree- balances the acceleration of gravity, coronal hole, we add a velocity depen- ment. Although these discrepancies are dent term to Eq. (l). understandable, given the fact that the various methods for measuring tem- (1) perature involve different assumptions,

10 LOS ALAMOS SCIENCE THE SOLAR CORONA AT TOTALITY where v(r) is the radial expansion veloc- infrequent that local thermodynamic ity. If we assume that the average solar equilibrium cannot be assumed. A con- wind mass flux K is constant and equal sequence would be that electron and ion to the value observed near the earth, we temperatures are not the same. Second, can obtain a rough estimate for the the method assumes that pressure gra- expansion velocity v from the relation dients, and thus the acceleration of the solar wind, are due to temperature gra- ne,(r) v(r) A(r) = K , (5) dients created by wave heating of the coronal plasma. However magnetic where A is the surface area. We then use waves moving radially outward may the expansion velocity to determine the accelerate the solar wind without heating emission line measurements by two pressure gradient, dP/dr, from Eq. (4). it. Since Eq. (4) does not explicitly methods. One method is based on com- Assuming that the pressure gradient is separate this nonthermal contribution to paring measured and calculated in- due to wave heating of the corona, we the velocity v(r), it yields effective pres- tensities of resonance lines from allowed determine the temperature gradient from sures, and through Eq. (2), effective atomic transitions. These resonance lines Eq. (2). Then, with a good value for the temperatures that are higher than the are at extreme ultraviolet (EUV) wave- temperature from an independent meas- kinetic pressures and temperatures we lengths (Fig. 4). EUV data from 1 to 1.4 urement, we can use these temperature are seeking. Temperatures derived from R@ have been measured from satellites. gradients to determine the variation of Eqs. (2), (4), and (5) must therefore be Calculations assume an equilibrium temperature with radius. When applied interpreted as upper limits. By consider- model of coronal temperatures and den- to the 1973 Skylab electron density data, ing these limits in conjunction with ion sities and use known atomic parameters this method yielded a steeply rising tem- temperatures and velocity distributions to determine the probability for reso- perature curve for the coronal hole determined from emission line measure- nance line emission. The assumed tem- above the solar north pole. ments, it may be possible to separate peratures are adjusted to match meas- OTHER CONSIDERATION. This result thermal and nonthermal sources of solar ured and calculated line intensities. Er- must be questioned for two reasons. wind acceleration. rors are introduced by the crudeness of First, since observed densities in coronal EUV ION TEMPERATURES. Ion tem- the assumed coronal model and by un- holes are very low, collisions may be so peratures have been obtained from certainties in atomic parameters.

EXTREME VISIBLE ULTRAVIOLET INFRARED ULTRAVIOLET WHITE LIGHT

EUV LINES LINES OBSERVED BY THE USED TO CONTINUUM RADIATION EMISSION LINE CAMERA DETERMINE (K CORONA) FOR T e , EXPERIMENT \ . Fe XIV GREENLlNE

LINE USED TO DETERMINE T PROTON I

H&K OF Ca II ABSORPTION LINES

Fig. 4. Relative positions in the electromagnetic spectrum of merely indicate the wavelengths at which observations are observations made to determine coronal conditions. We do not made. attempt here to reproduce the actual coronal spectrum but

LOS ALAMOS SCIENCE 11 THE CHANGING CORONA

ortraits of the corona during five June 1976 1978 1980 1982 1984 1986 eclipses display its changing ap- P pearance as the sun passes 160 through its cycle of magnetic activity. Cycle21 The eclipse dates are given on the ac- companying graph of sunspot activity, which covers the present (red) and previous (blue) solar cycles. Each portrait is a composite of 20 to 30 photographic images taken from the NC- 135 jet aircraft with the Laboratory-designed camera-polarimeter. These portraits of the corona extending from the sun’s limb to 12 solar radii are unparalleled. Photographs taken from the ground record only the inner corona to about 4 solar radii. Even October 1964 1966 1968 1970 1972 1974 Skylab’s superb coronagraph can view

12 LOS ALAMOS SCIENCE EMISSION LINE PROFILE ION TEM- PERATURES. A somewhat more direct method involves the analysis of measured intensity versus wavelength profiles of emission lines from highly ionized iron and calcium. (Several emission lines have wavelengths in the visible spectrum.) The width of these Doppler- broadened emission lines is a direct measurement of ion kinetic temperatures, provided all the broadening is due to thermal motions. However mass motions and turbulence can also contribute to the broadening. The Los Alamos line profile data out

to 3 R@ are the most extensive available. Our analysis suggests that ion temperatures deduced directly from the widths of heavy ion emission lines may be larger than kinetic temperatures by as much as 30% because of nonthermal contributions to the line broadening. Kinetic temperatures may be deduced directly from the width of hydrogen’s Lyman-a emission line at 1216 A be- only the region between 1.5 and 6 solar ods of high activity (1970, 1979, and cause nonthermal contributions to the radii. 1980), streamers approach and often line broadening are proportionally much From this unique set of portraits, we cover the polar regions and the dipole less than for heavier mass ions. Since can follow the evolution of streamers. field is bent only slightly away from the Lyman-a radiation cannot penetrate our The marked variation of the streamer poles. atmosphere, such measurements must be pattern over the sun’s 1 l-year activity Individual portraits bear careful made at altitudes of 100 km or greater. cycle gives us a clear picture of the effect study. Close to the sun, streamers are Data between 1.5 and 3 R@ are now of solar activity on coronal structure and obviously dominated by magnetic fields. available from rocket-borne experiments solar magnetic field. We see them twisted and curved in all performed by scientists from Harvard The most striking structural change in directions. At larger distances, however, University and the High Altitude Ob- the corona is the lack of bright streamers many streamers are seen to bend gradu- servatory of the National Center for over polar regions during periods of low ally back toward a radial direction, pos- Atmospheric Research. solar activity. At such times ( 1972 and sibly under the influence of the expand- The results of recent temperature 1973), polar coronal holes extend to ing solar wind, Others remain nonradial measurements shown in Fig. 5 are in lower solar latitudes. Solar wind flows even at the farthest distances. The in- obvious disagreement. In coronal holes from coronal holes and apparently bends formation on streamer dynamics con- (Fig. 5a), we see upper and lower limits the sun’s magnetic dipole field strongly tained in these images awaits theoretical on kinetic temperatures based on elec- toward equatorial regions. During peri- study c tron density data. The lower limit is -1

LOS ALAMOS SCIENCE 13 MK based on the assumption of a static, isothermal corona. The Skylab electron 4 density data, interpreted through a mod- (a) Coronal Holes el that accounts for expansion velocities as described above, results in a steeply rising curve that reaches 3.5 MK by 3 The other three observations fall within these bounds but are too sparse to allow determination of the position of the temperature maximum. The single EUV temperature and the single Lyman-a proton temperature suggest extended magnetic heating. When the Los Alamos iron ion temperatures are included, the temperature maximum moves below 1.3

Figure 5b shows similar results for relatively quiet regions of the corona (regions that are devoid of obvious streamers but are not coronal holes). Neglecting the iron ion temperature puts the temperature maximum beyond 1.4

R @; including it moves the maximum Preliminary iron ion temperatures from the Los Alamos 1973 eclipse observations show ever increas- electron density and the EUV results in the lower corona can be believed, then the iron ion results must include a large turbulent contribution to line broadening. Interpretation of the results shown in Fig. 5 is further complicated by the fact that the data were obtained at different times and at different places in the corona. To resolve the discrepancies, we need simultaneous observations of ion temperatures, electron temperatures, and Fig. 5. Plots of coronal temperature versus radial distance for (a) coronal holes and electron densities. One of the primary (b) a quiet coronal region. As discussed in the text, data are sparse and, when goals of the Laboratory’s 1980 solar compared, confusing. The main purpose of the Los Alamos expedition was to attempt eclipse expedition was to do just this. to reduce some of the sources of disagreement by making cotemporal and cospatial measurements. The 1980 Experiments The National Aeronautics and Space ments, including one from Los Alamos, The sun’s magnetic activity follows a Administration (NASA) planned to to measure proton temperatures from well-known 1 l-year cycle becoming very launch the Solar Maximum Mission, a the Lyman-a line. The Naval Research active, waning to near total inactivity, satellite similar to Skylab’s observatory. Laboratory’s orbiting coronagraph was and then repeating. The amplitude of this Aboard it would be experiments to de- to make hourly recordings of the white variation is never the same, but by 1979 termine electron densities and to obtain a it was obvious that the 1980 maximum crude estimate of ion temperatures from The National Science Foundation (NSF) would be one of the highest in history Fe XIV emission intensities. (Un- planned to send to India a large con- and many experiments were planned to fortunately this experiment was not op- tingent of ground-based observers, who observe the solar corona during the erating by the time of the eclipse.) would make a variety of measurements. eclipse in February, 1980. NASA also funded two rocket experi- In planning the Los Alamos airborne

14 LOS ALAMOS SCIENCE THE SOLAR CORONA AT TOTALITY expedition, we were in communication tempted by our expedition. All were with scientists from NASA projects and designed to provide good two- some NSF projects. It was apparent that dimensional spatial resolution. We ex- cross-calibration and data comparison pected to be able to determine all the would assure maximum confidence in quantities not only in the relatively uni- the results. form portions of the corona, but also As discussed above, to answer signifi- within major features such as streamers, cant questions about the nature of coro- condensations, and coronal holes. Table nal heating and solar wind acceleration I summarizes the experiments and the requires simultaneous measurements of information sought. the following: To determine ion temperatures we ● electron density, ne measured the Doppler-broadened

• ion temperature, TiOn emission lines from two heavy ions—Fe ● electron temperature, Te. XIV and Ca XV. We will compare our All these measurements were at- results with proton temperatures de-

1980 Solar Eclipse Experiments

Principal Information Experiment Investigators’ Obtained or Sought Resolution

Emission Line Ion Temperature D. H. Liebenberg Temperatures of Fe XIV and 0.005 R@ (3500 km) Measurement of coronal Fe XIV E. A. Brown Ca XV, variation of emission and Ca XV emission line profiles R. N. Kennedy line intensity with radial with Fabry-Perot interferometer H. S. Murray distance, and nonthermal W. M. Sanders contributions to ion tem-

perature from 1 to 3 R@

Electron Temperature M. T. Sandford Electron temperature and 0.067 R@ (46,000 km) Measurement of K coronal F. J. Honey coronal spectra at 1.2, 1.6, spectral intensity with R. K. Honeycutt, and 2.0 R@ silicon photodiode Indiana University detector array

Electron Density C. F. Keller Electron density, K + F corona 0,067 R@ (46,000 km) Measurement of intensity and J. A. Montoya intensity, and K corona polarization of coronal light B. G. Strait polarization from 1.1 to 5 R@ with camera-polarimeter in polar regions and from 1.1

to 12 R@ in equatorial regions and image-enhanced photographs

of streamers from 1 to 20 R@

W. H. Regan Detailed coronal structure 0.033 R@ (23,000 km)

Coronal photography with C. G. Lilliequist from 1 to 6 R@ radially graded filter and internal occulting disc

Infrared Emission of Dust Rings J. P. Mutschlecner Radial location of dust rings 0.25 R@ (175,000 km) coronal infrared R. R. Brownlee and infrared emission emission intensity with InSb D. N. Hall, Kitt Peak intensity from 2 to detector and charge-injection- National Observatory 50 R@ device television camera aUnless otherwise indicated, the investigator is affiliated with Los Alamos, Other Laboratory personnel contributing to the 1980 solar eclipse expedition were W. H. Roach (scientific commander), R. A. Jeffries (in charge of liaison with the USAF), D. H. Collins (logistics officer), C. T. Barnett

LOS ALAMOS SCIENCE 15 Fig. 6. The white light we observe during a total solar eclipse has two components, o light from the photosphere scattered to solar Disk us by electrons in the corona (straight –1 line) and light from the photosphere scattered to us by interplanetary dust -2 I (wavy line). I –3 Fig. 7. Relative intensity of components of coronal light as a function of radial -4 distance. The K corona is continuous light from electron scattering. The F corona is light from dust scattering. E is the combined light of emission lines. Relative background intensities under various observation conditions are also shown. Note the tenfold reduction in sky background intensity for high-altitude observations over ground-based observations.

r/R@

16 LOS ALAMOS SCIENCE THE SOLAR CORONA AT TOTALITY

Fig. 8. Eclipse observations of the corona are made along a line of sight, but in many cases single coronal features can be identified reliably. This is because features closer to the limb are more intense and therefore contribute more light to the measured signal For example, the line of sight shown here in- tercepts a streamer in the plane of the

sky at 1.68 Ra and an identical streamer 30° out of the plane of the sky

at 1.94 Ra. The streamer in the plane of the sky contributes about 2.5 times as much white light and about 5.5 times as much emission line intensity to the observed signal as does the streamer out of the plane of the sky. duced from rocket measurements of spatial structures. these spectral continua are emission lines Lyman-a emission at several points in Data reduction and comparison is a from highly ionized heavy ions. Figure 7 the corona beyond 1.5 R@ to distinguish long and diflicult process that is only displays the relative intensities of the thermal from nonthermal contributions beginning. Before we discuss each ex- various components of coronal light; to line broadening. We are particularly periment and give an early estimate of these intensities are compared with the interested in determining whether the the data it acquired, let us consider some brightness of the sky under various ob- extremely high ion temperatures at the of the general conditions of coronal servational conditions. Note the sharp base of the corona deduced from our observation. reduction in sky brightness during a 1973 results are correct or should be solar eclipse. Although emission line in- attributed to the effects of turbulence, Observing the Corona tensities are much less than white light but we must await data from these low intensities, they are clearly visible above altitudes. We also measured electron The angular sizes of the moon and the the white light background when ob- densities, recorded high-resolution im- sun, as seen by an observer on (or near) servations are limited to very narrow ages of the corona, and attempted to the earth are approximately equal. Dur- wavelength bands. However, the rapid observe electron temperatures by a new. ing a total eclipse, this fortuitous circum- decrease of emission line intensity with untried method. Finally, we attempted to stance enables us to view the solar radial distance from the sun makes determine the location of hot dust rings corona, a region otherwise obscured by measurements impossible beyond 3 R@ around the sun from their infrared the intensity of the photosphere. Coronal with present instrumentation. emissions. light comes from several sources, the Apart from the signal intensity, the Except for the data loss from a me- most intense being light from the other unalterable limitation on observa- chanical failure in the electron tem- photosphere that has been scattered by tions is geometry—we can only record perature experiment, we obtained ex- particles in the corona. This white light two-dimensional information about cellent data in all areas. And photo- consists of two components, the K coro- three-dimensional objects. As shown in graphs of the corona taken from the na—sunlight scattered by electrons— Fig. 8, every observed signal is inte- aircraft will enable us to associate our and the F corona—sunlight scattered by grated along a line of sight through the detailed observations with particular dust grains (Fig. 6). Superimposed on corona. If the corona were spherically

LOS ALAMOS SCIENCE 17

symmetric, this would pose no problem. But in fact its visible brightness varies he color photograph of a huge about 15 minutes earlier; from the dif- markedly from one region to another. bubble-like structure extending ference we infer an expansion velocity of Thus interpretation of all measurements T from the sun’s limb to 7 solar about 500 km/s. This value is in agree- requires assumptions, based on detailed radii was a most exciting result of our ment with the eruption’s absence from photographs, about the three- 1980 expedition. The structure is formed photographs taken from the main scien- dimensional structure of the corona. by a hydrodynamic eruptive disturbance tific site in India, where the eclipse that ejects large quantities of mass into occurred about 90 minutes later. During Direct Photography — interplanetary space. The eruption is that time, the eruption would have Imaging the Corona very clearly visible on the computer- moved beyond range of observation enhanced image processed from 32 from the ground. If the velocity of ex- Good photographic prints of the solar digitized photographs. These large and pansion is nearly constant during its corona are extremely difficult to make frequent mass eruptions have been re- lifetime, the disturbance must have because coronal brightness varies by

corded since the early 1970s by Naval begun just as the eclipse was arriving on factors of 1000 from 1-4 R@ and 10,000 Research Laboratory satellites and by Africa’s west coast, about 90 minutes from 1-10 R@, whereas prints can Skylab, but the bases of the eruptions before we saw it. We are at present display only a factor of 10. We use three were always obscured by the oversized trying to obtain good photographs taken methods to reduce this radial brightness occulting discs used in satellite coro- from Zaire and Tanzania to study the gradient. nagraphs. Thus ours is the first complete earlier phases of the event. record of the phenomenon. Such eruptions are apparently quite 1. A radially graded filter placed just in At first glance the structure looks like common. They are estimated to have front of the film transmits light in a tennis racket with its handle in the sun. occurred at least once a day in 1973 precisely the desired amount as a Closer examination reveals that it is not during low solar activity and perhaps function of distance from the sun. entirely symmetric about an axis extend- three times more frequently during this 2. An internal occulting disc in the con- ing radially from the sun. Its polar side is past year’s maximum solar activity. verging light path within the camera markedly flattened and a density en- Each one typically ejects a mass of also uniformly reduces the brightness hancement, which is possibly due to a about 10]3 kg and an energy of 1024 J. gradient. shock wave ahead of the eruption, is The apparent rate of occurrence would 3. And computer processing reduces the very prominent above and on the equa- make them responsible for at least 10% gradient in the digitized photographs torial side, but is nearly absent on the of the entire solar mass efflux! from the electron density experiment. polar side. Nearby major streamers that We also obtained camera-polarimeter usually extend in a radial direction from images and Fe XIV emission line profiles All three methods were used during the sun are bent toward the disturbance, of the eruption. Its Thompson-scattered the 1980 expedition. Because of a slight more so on the equatorial than on the light is highly polarized so we will be misalignment of the mirror tracking sys- polar side. In addition, the eruption may able to determine electron densities and tem aboard the aircraft, the first two possibly be bent slightly toward the solar its base appears to have the most intense gave good photographic records of only equator. Characteristics similar to these green line (Fe XIV) emission in the two-thirds of the corona. Nevertheless have been reported for eruptions ob- corona. We anticipate that these meas- these pictures give us a remarkably de- served from Skylab. urements and our photographs, which tailed view of the corona during the sun’s The position of this eruptive dis- together comprise a wealth of intercon- maximum activity and a particularly fine turbance recorded from the aircraft dif- nected information, will answer several look at a large mass ejection extending fers markedly from that recorded by the questions about these important sources Laboratory’s rocket team in Kenya of solar wind ■ One of these photographs, together

LOS ALAMOS SCIENCE 19 ties. Since we can measure only total white light intensities (K + F corona), we need a method to subtract the unwanted F coronal light. Much of the K coronal light we observe during an eclipse has been scattered through angles near 90°. The K corona is therefore highly polarized (Fig. 9), whereas the F corona is not. The two EARTHBOUND OBSERVER components can be separated by measuring both the absolute intensity of the K Fig. 9. Schematic of light scattered at + F corona (K + F) and the fraction and 90° by an electron in the plane of the sky direction of polarization at each point on to an earthbound observer. This light is digitized images. The measured frac- 100% linearly polarized along the direc- tional polarization can be written tion tangent to the sun’s radial vector, because its electric field vector (indicated by the arrows under the wavy lines) is always perpendicular to its direction of motion. However, not all the light we where Pk is the fractional polarization of observe in our camera-polarimeter has the K corona, and K and (K + F) are the Fig. IO. Los Alamos-designed telescope been scattered in the plane of the sky. K coronal intensity and total white light for photographic polarimetry and elec- Each point on our digitized photograph intensity, respectively. Data reduction to tronic control equipment (lower left) is a sum of contributions that have been determine K requires a model of the mounted in aircraft. This instrument corona. We assume that the corona has scattered from many angles into the line photographs the corona beyond 12 Ra of sight. Consequently, the maximum cylindrical symmetry and that its density and has provided data for determining observable polarization of the K corona varies smoothly between polar and equa- electron densities and image-enhanced from a spherically symmetric corona is torial regions. From this model, we cal- photographs of the outer corona. (See only 670/o. This theoretical maximum is culate Pk and K and compare them with note “Unique Record of the Changing reduced even further because the corona the measured quantities Ptotal and (K + Corona.") The blue cylinder (upper is not spherically symmetric and, in the F). Finally we vary electron densities in right) is one of three orthogonally inner corona, photospheric light is trav- the model to obtain consistency between mounted gyroscopes obtained from eling in nonradial directions before the two measured and two calculated NASA Apollo program. These provide being scattered. Thus the expected quantities. the inertial references that allow this polarization as we move toward the Photographs of the K + F corona are instrument to point at high accuracy O sun limb decreases from 67°/0 to O /O. taken with a camera-polarimeter through even when the aircraft is in motion. plane polarizing filters oriented at three different angles. A fourth photograph taken without any polarizing filter com- pletes a set. During the 1980 eclipse, we with a computer-enhanced image proc- made 10 sets of photographs using greater distances than have been essed from 32 digitized exposures taken high-resolution film for the bright inner achieved from ground observations dur- with our camera-polarimeter, is shown in corona and very fast low-resolution film ing an eclipse or from Skylab observa- the note “A Mass Eruption From the for the outer corona. This is the first time tions outside of eclipse. In the 1979 and Sun.” we have taken special care to get 1980 eclipses we were able to make two high-resolution data from the inner coro- Electron Density Experiment na. We have made similar measurements The camera and its center-of-mass, using the same instrument at five times The scattering process that produces three-axis gyrostabilized tracking system during the sun’s 1 l-year magnetic activi- the K corona (see Figs. 6 and 7), name- were designed and built at Los Alamos ty cycle (1970, 1972, 1973, 1979, and ly, Thompson scattering of photospheric (Fig. 10). With this tracking system, 1980). We thus have a very uniform set light by electrons, is independent of motion during a 3-second exposure was of data with which to compare variations wavelength and depends only on elec- less than 20 arc seconds. Our polariza- in coronal structures, brightness, electron density. K coronal intensities are tron density, and material distribution as thus a direct measure of electron densi- equatorial region of the corona—much a function of solar activity. (See note

20 LOS ALAMOS SCIENCE THE SOLAR CORONA AT TOTALITY

“The Changing Corona.”) For instance, corona (K + F) as a function of radial standard observational models of coro- distance in the inner and outer corona, nal brightness indicate a marked vari- respectively. Also included in these plots ation between maximum and minimum is the consensus model K + F brightness solar activity. We are unique in being based on pre-1965 eclipse data. Figure able to verify this variation for the inner 1 la is a comparison of results from five independent observations of the 1973 eclipse: two from the ground, two from space coronagraphs, and one from the Los Alamos airborne expedition. Only Results of 1973 Electron the Los Alamos data extend over the Density Experiments entire range shown and beyond. In comparing our results for the 1973 In recent years the 1973 eclipse was eclipse with those of Skylab’s Apollo the most widely observed both from the Telescope Mount, we were puzzled to ground and in space, Figures 1 la and b find the Skylab values of coronal bright- show plots of absolute intensity of the ness disturbingly higher than ours. Upon

10.0

1963 8.0 1963

6.0 1973

4.0

Equatorial Region

2.0

1.5

1.0

0.8

Newly observed by Los 0.6 Alamos 1979 and 1980 ~\

0.4 (b) OUTER CORONA

4 68 10 12 15 20

r/R@

Fig. 11. Various observations of coronal brightness versus cospatial, cotemperal observations from two ground-based, one intensity for (a) the inner corona and (b) the outer corona. airborne, and two space experiments. Only the Los Alamos Also shown in both cases is the brightness caleulated from a data cover the entire range of comparison, and the Los Alamos consensus model [D. Blackwell, D. Denhirst, and M. Ingham, 1980 data, when reduced, will extend observations to 20 Ra “The Zodiacal Light,” Advances in Astromony and for the first time since 1963. Astrophysics ,1 (1976)]. Figure 11a is a unique comparison of

LOS ALAMOS SCIENCE 21 careful examination of Skylab reduction procedure. we discovered an error in absolute calibration that resulted in a 90/0 reduction of all Skylab values of coronal brightness. This transfers roughly to a similar reduction in all published results prior to 1978. The intensities plotted in Fig. 1 la have been corrected for this error, Figure 1 lb compares Los Alamos data for the outer corona with the two best recent observations of that region. Figure 12 shows our 1973 results compared with those from K. Saito’s ground-based observations.* Because of sky background brightness. ground- based observations seldom are reliable beyond 5 R@ in equatorial regions and 3

R@ in polar regions, but the Los Alamos intensity data appear excellent beyond Polarization measurements in the equatorial regions are also good to this distance, but over the sun’s poles the K corona is so faint that polarization falls below the limit of photographic fainter corona in polar regions than Saito does and therefore conclude that electron densities in polar coronal holes are lower than previously thought. Fig. 12. Electron density as a function of radial distance determined from In equatorial regions the situation is photographic polarization data gathered during the Los Alamos airborne expedition more complicated. We find coronal in- of 1973. The results of K. Saito, a respected ground-based observer, are shown for tensity to be lower, but we measure comparison. Photographic polarization data for the polar regions are reliable only to measurements suggest that in equatorial the reduction in atmospheric contamination realized by our airborne expedition are higher and, more significantly for solar wind models, do not fall off as fast as we proceed outward from the sun. We also observed this significant result at the Emission Line Experiment pendicular to the line of sight) easily 1970 and 1972 eclipses. contribute the major portion of our We look forward to comparing elec- We consider the coronal emission line measured signal along a line of sight tron densities from the 1973 eclipse with experiment to be the most important because emission line intensities fall off those of the 1980 eclipse as soon as our aboard the aircraft because the meas- very rapidly with distance from the sun. 1980 data have been reduced and eval- ured line profiles contain a wealth of Therefore we can attribute measured uated. information on the state of the coronal signals to specific coronal features with a Preliminary analysis of our 1980 in- plasma. The shape of the line profiles fair degree of confidence. tensity data shows that the corona was can be analyzed to yield ion tem- Since 1965, we have made airborne about three times brighter than it was in peratures and nonthermal components measurements of the Fe XIV green line 1973, and that polar regions are as of the velocity distribution. The vari- intensity and wavelength broadening us- ations of line intensity with position, or ing a Fabry-Perot interferometer to ob- where the F corona begins to dominate. with time, contain information about the tain spectral (wavelength) resolution. We mechanisms that excite the emitting ions. chose an interferometer for several rea-

*K. Saito, Annals of the Tokyo Astronomicai Bright features in or near the plane of the sons. It is a relatively small instrument Observatory 13, 93 (1972). sky (plane through the sun’s center per- well suited to the space constraints

22 LOS ALAMOS SCIENCE I THE SOLAR CORONA AT TOTALITY

Fig. 13. Isodensity tracing of the interferometer image of the Fe XIV emission line obtained during the 1%5 total solar eclipse. The heliocentric coor- dinates have been overlaid and the lunar limb position is indicated.

aboard the aircraft. It also has a high spatial resolution, in part because it preserves a two-dimensional image rather than the one-dimensional images obtained from grating spectrographs and other instruments that pass the incoming light through a slit. Perhaps most important, spectral distortions introduced by the interferometer can be calculated quite accurately and subtracted from the measured line profiles, enabling us to achieve very high spectral resolution.. In 1965 we obtained the first Fe XIV

line profiles out to 2 R @ from a photographic interferometer image of the corona off the sun’s west limb (Fig. 13). A radial trace from the center of the fringe system through each fringe produces an intensity versus wavelength profile emitted by the corresponding region of the corona. Our spatial resolution was limited by tracking to the order of minutes of arc. As shown in Fig. 14, these fringes passed through a helmet streamer, a coronal hole region, and an enhancement. From these data we estimated ion temperature, temperature gradients, and

Fig. 14. An artist’s sketch of the white light corona in the northwest quadrant during made the first tentative determinations of the May 30, 1965, total solar eclipse. The locations of the interferometer fringes are ion temperatures in helmet streamers shown together with the position angles of two coronal features.

LOS ALAMOS SCIENCE 23 Fig. B.

Wavelength THE FABRY-PEROT Fig. D. INTERFEROMETER

he heart of the emission line reflecting surfaces. When this path camera is a Fabry-Perot in- length is exactly equal to an integral T terferometer. It resolves the light number m of wavelengths, that is, when from a particular atomic transition oc- curring in the corona into a high- (A) resolution profile (emission line profile) of intensity versus wavelength. then the reflected and incident light The basic instrument consists of two waves are exactly in phase and construc- accurately plane quartz plates main- tive interference (reinforcement) takes tained strictly parallel at a constant place. Successive values of m correspond distance D. The opposing plate surfaces to successive orders of interference. If are coated to be highly reflecting. The the incident angle deviates even slightly plates are enclosed in an airtight chamber and the space between them is reduces the intensity because of the large filled with gas under pressure. The index number of reflections in a highly re- of refraction n of the gas-filled space can flective Fabry-Perot interferometer. be varied by changing the gas pressure. Thus monochromatic light from an ex- A beam of monochromatic light with tended source produces a series of con- centric bright rings corresponding to successive orders of interference. between the plates many times (Fig. A). The coronal spectrum includes many The beam’s optical path length is 2nD widely different wavelengths that can interfere constructively for a given value

24 LOS ALAMOS SCIENCE ring to extract intensity versus 9, which printed our 1980 photographs, the read- a prefilter to keep the long wavelength is easily converted by using Eq. (A) to er can observe the movement of the rings end of one order from overlapping the caused by the pressure scan. From this short wavelength end of the next order. these profiles are deduced for an ex- series of images corresponding to dif- The prefilter is centered on the coronal tended region of the corona correspond- ferent values of n, we can plot the emission line being measured and trans- ing to less than the width of the broad intensity change at a single spatial point interferometer rings. as the gas pressure is varied with time. Coronal emission lines are Dop- The partial rings produced during the This plot of intensity versus n is con- 1965 eclipse by a segment of the corona full width at half-maximum intensity. If are shown in Fig. 13 of the main text. In Our video recording system stored an 1965 we had time to take data only at a array of 250,000 spatial data points fixed value of 2nD, so the only regions of from each image; during a pressure scan then another wavelength in the profile, the corona that could be studied were of one order, 120 images were recorded. those corresponding to the bright rings Each point can produce such a profile.

slightly different incident angle 02 will in Fig. 13. One such profile is shown schematically also interfere constructively (Fig. B). During the 1980 eclipse, our rapid in Fig. D. The line profile is repeated as Thus successive wavelengths in the line data-acquisition system gave us time to the pressure scan moves the ring corre- profile are imaged by the interferometer vary the gas pressure inside the in- sponding to the succeeding order across as adjacent rings at successive values of terferometer, and thus to change n and the point being plotted. The spatial reso- cause the rings to traverse the entire lution of a line profile deduced from a coronal image by changing the angle for single point is 4-5 arc seconds, which is constructive interference. By thumbing an order of magnitude better than can be carefully makes a trace radially across a the corner of this journal, on which are obtained from a radial trace ■

LOS ALAMOS SCIENCE 25 TRACKING CONTROL

In 1973, we scanned the entire corona

15-20 arc seconds. We measured very weak polarization of the green line along radial vectors from the sun. Weak polarization implies that green line MIRROR emission is excited by electron collisions rather than absorption of photons, This result confirms expectations that electron collisions are the dominant mechanism by which energy is distributed in the inner corona. We also observed that Fig. 15. Main components of the tracking control for the coronal emission line Ca XV yellow line emission occurred camera. Tracking along two orthogonal axes is controlled directly by an operator or over a much larger region of the lower by feedback from gyroscopes mounted on the telescope. Only one of these gyroscopes corona than previously expected. The (L gyro) and its associated control mechanisms are shown here. Gyroscope drift in the presence of the yellow line was thought two orthogonal axes is compensated by error signals from a set of photodiodes that to signify very hot regions of the corona, views an image of the eclipsed sun. The tracking plane controller is preprogmmmed corresponding to the 4-5 MK ionization with the various tracking patterns to be followed during totality. In addition, rotation potential maximum of the Ca XV ioniza- of the telescope about its optic axis is required to correct not only for aircraft motion tion state. but the 1973 line widths but also for the apparent rotation of the sun during the duration of totality, which indicate that the yellow line is produced amounted to nearly 2° during the 1980 eclipse. This rotation is controlled by feedback in cooler regions as well, and thus that from a gyroscope (R gyro) and by preset gyroscope precession in the R controller. coronal temperatures may be much Error signals, the difference between a gyroscope axis and the required axis, are about more uniform on a scale of arc minutes 4-5 arc seconds for the telescope in flight. Similar performance accuracy has been than had been guessed from other determined from photographs and video recordings of lunar limb motion. emission line measurements. Our 1980 data is much more extensive and has a much higher spatial profiles from the base of the corona and by offsetting the entire 210-kg emission resolution than that from previous line camera (telescope, interferometer, eclipses. detector was designed specifically for and detector) with large hydraulic ac- Light from the 1980 experiment was astronomical measurements and for tuators that are controlled by a sophisti- collected by the "Rube Goldberg,” a compatibility with standard computer cated tracking system accurate to 4-5 massive 10-inch telescope with an image analysis equipment by M. T. arc seconds (Fig. 15). The hydraulic 80-inch focal length that has been used Sandford. Its progenitor was developed actuators are able to move the massive aboard the aircraft since our 1965 ex- at Los Alamos for diagnostics in the instrument at frequencies of up to 10 Hz pedition. Emission line signals, imaged weapons program. to correct for higher-frequency aircraft by the telescope and interferometer, were The line resolution and diameter of the motions. amplified and recorded on videotape at detector, combined with the relatively Data from the 1980 measurements 16-ins intervals from an image- large image size (we view the corona in are much easier to reduce than those intensified vidicon detector. This rapid segments), determine a spatial resolution from previous eclipses. Preliminary re- data acquisition system, a dramatic im- of 4-5 arc seconds. But to realize this sults show wide variations in line shapes provement over the 30-second exposure resolution, we must keep the image cen- and are suggestive of the turbulent con- times required by photographic tech- tered in the interferometer and on the ditions that may be present during a niques, collected well over 20,000 line axis of the telescope, We accomplish this solar maximum. We also see many de-

26 LOS ALAMOS SCIENCE THE SOLAR CORONA AT TOTALITY

r- 8 70 J is the mass of the emitting ion. Note that the kinetic temperature contribution depends on the mass of the ion, whereas the contribution from macroscopic turbulence is mass independent. Therefore it is possible to separate thermal and turbulent contributions to the line broadening by measuring emission line profiles of two ions with widely differing masses. But these measurements must be made at the same time and at the same place in Column the corona so that we can assume that the two ions have the same kinetic Fig. 16. These preliminary 1980 Fe XIV emission line profiles were obtained by temperature and average turbulent veloc- digitizing a sequence of video frames and reordering the data to give a signal versus ity V. frame number (or wavelength) profile of the intensity at a single pixel (corresponding t Our Ca XV data from the 1980 to a single spatial location in the corona). In (a) we have fitted both a Gaussian and a eclipse may be appropriate for com- Lorentzian profile to the emission line such that the full width at half-maximum parison with our Fe XIV measurements, intensity and the peak intensity agree with observations. The wings of the observed line but the emission was weaker than ex- are not fitted by either profile. In (b) an example of a more complicated emission line pected and so the profiles may not be is shown. This profile could represent coronal material with significantly different accurate enough. However, the Lyman-a velocities along the line of sight. A relative velocity of about 1.5 km/s would fit this data recorded by rocket experiments at profile. Some further corrections for intensity and wavelength calibration will be several places in the corona will certainly applied to these data before our analysis is complete. be useful in determining nonthermal con- tails within small regions of high activity. perature of the emitting ions. However, tributions to our Fe XIV line profiles. However, data reduction must be com- ion temperatures deduced directly from Variations in line shape are also in- pleted before we can make definitive observed Gaussian line profiles (for ex- dications of nonthermal velocities. For interpretations. ample, the iron temperatures shown in example, calculations show that large Fig, 5) may be too high because non- expansion velocities from solar wind Interpreting Emission Line Profiles thermal effects, such as macroscopic flow tend to flatten and extend what turbulence and magnetic wave accelera- would have been a Gaussian line shape. Both line shapes and line intensities tion of the solar wind, may add to the Although these departures are not signif- are analyzed to learn about the condi- velocities of the ions and, in turn, to the icant until expansion velocities are 40 tions in the coronal plasma. Since the line broadening. km/s or greater, such velocities are pre- intrinsic line widths (from ions at rest) An approximate expression for the dicted by some models of solar wind are extremely narrow, the observed Dop- line width in terms of the kinetic tem- pler-broadened line profiles are identical perature T and the average turbulent profile shapes are also altered by

in shape to the velocity distribution of macroscopic velocity Vt is given by large-scale turbulent motion or signifi- the emitting ions. cant magnetic wave acceleration of the For ions in thermal equilibrium, both plasma. the velocity distribution and the emission Figure 16 illustrates the variety of line line profile would be Gaussian in shape. shapes that were actually measured in The width of the line profile would then 1980. We see asymmetries, as well as be proportional to the kinetic tem- profile at half-maximum intensity and m departures from Gaussian shapes. In-

LOS ALAMOS SCIENCE 27 terpretation of these line shapes depends HELMET STREAMER. One inter- magnetic field lines, the helmet should on theoretical assumptions and support- ferometer fringe (see Figs. 13 and 14) have above-average temperatures. In ing data, except perhaps for the case of crossed a helmet streamer near the top particular, along a line through the top split lines, which clearly indicate relative of the helmet we should see a tem- bulk motion of the emitting material. begin to change from a closed loop to an perature increase in the helmet and a Analysis of line intensities can also be open-line configuration extending radi- temperature decrease on either side. quite revealing. Since emission intensity ally from the sun. Since particles in the However, our 1965 data (Fig. 18) show depends on density and temperature, helmet are presumably trapped along that temperatures along this line fluc- comparison with independent electron density measurements is helpful in de- ciding whether bright regions are due to high temperatures or, alternatively, to high densities. We can also analyze intensity variations as a function of radial distance to infer excitation mechanisms. If ions are excited by electron collisions, intensities are proportional to 2 the square of the electron density (ne) and thus should decrease as l/r12. On the other hand if photon excitation is the 0.8 dominant mechanism, then line in-

tensities are proportional to ne, and thus decrease as l/r6. Finally we can try to find periodic variations in intensity and line shape to 0.4 identify wave heating of the corona. We have recorded tentative evidence of periodic variations in some of our 1973 data taken aboard the Concorde during a

record-setting 74 minutes of totality 0 (Fig. 17). 0 5 10 15

Emission Line Data Time (rein) Results and Questions Fig. 17. Peak intensity versus time for two 15-minute segments of Fe XIV emission line Our 1965 results (see Fig. 5 for aver- data obtained aboard the Concorde 001 at the June 30, 1973, total eclipse. The data age ion temperature values) showed ion temperatures decreasing outward from variations with a periodicity of about 5 minutes corresponding perhaps to a coronal the base of the corona with little vari- response to the 5-minute oscillations of the photosphere. Tracking problems in- ation in temperature from one coronal troduced the errors shown by the bars. The data in (b) are from an equatorial bright feature to another. A more detailed look at the data in each feature is illuminat- suggests the presence of a shock disturbance. The variations have a periodicity of ing. about 6 minutes.

28 LOS ALAMOS SCIENCE THE SOLAR CORONA AT TOTALITY

peratures in this region. Coronal holes are expected to have lower temperatures and densities than other regions, and since they are known to be a source of Fig. 18. Temperatures across a helmet structure deduced from Fe XIV emission line solar wind flow, we expect temperature data obtained during the May 30, 1%5, total solar eclipse. gradients consistent with solar wind theory. One surprise was the high tem-

much higher than those determined for the base of the coronal holes by the EUV measurements. Our temperatures, which we estimated directly from line broadening, would be slightly less but still much higher than those determined from other experiments, if turbulent velocities contribute to the line broadening. In fact, we have invoked turbulent velocities in some parts of this region to deduce temperature gradients consistent with solar wind flow. If the temperature at the

fixed at 2.5 MK, then successive tem-

peratures at 1.62 R@ and 1.74 R@ are 2.43 MK and 2.39 MK, respectively, provided the turbulent velocities at these three altitudes are 35, 10, and 5 km/s, respectively. The temperature variation Fig. 19. Intensity versus wavelength profile of the coronal Fe XIV emission line obtained during the May 30, 1965, total solar eclipse. The maximum intensity I is at Q We also determined the variation of line intensity with radius in this coronal tuate from about 2.5-3.0 MK without where the field lines open to form a hole region. Figure 20 shows that in- much systematic change. The Gaussian streamer. We are particularly interested tensities fluctuate rapidly near the base line shapes from the helmet (Fig. 19) in knowing whether temperature gra- of the corona, but at greater heights the 12 2 suggest that, as expected, turbulent ve- dients above the helmet are consistent intensity is proportional to l/r , or (ne) . locities are small, perhaps less than 25 with solar wind flow. At present the This rapid falloff is consistent with a km/s in this long-lived magnetically con- contribution of streamer regions to the collisional excitation mechanism—a fined region. solar wind is poorly known. rather surprising result given the very Our 1980 data will provide a more CORONAL HOLES. Several fringes of low density in this region determined by detailed look at the temperature struc- our 1965 data crossed a coronal hole independent electron density measure- ture across a helmet and will extend our region, so we were able to estimate ments. However, if magnetic field lines view of the region above the helmet temperature gradients as well as tem- are clustered, rather than uniformly dis-

LOS ALAMOS SCIENCE 29 tributed, and if electrons are trapped along these clustered lines, then the average electron density would remain low, as measured along a line of sight, but local densities would be high enough to produce ion excitations by collisional processes. Evidence for density inhomogeneities is seen on white light photographs of the corona, but whether these inhomogeneities correspond to clustering of field lines is not at all certain. More detailed intensity measurements from the 1980 eclipse may clarify the situation. CORONAL ENHANCEMENT. A coronal enhancement is a long-lived region of higher density that overlies an active region in the photosphere. Consequently, we expect higher turbulent velocities in such a region than in helmet streamers. From our 1965 data we estimated temperatures of -3.0 MK after subtracting turbulent velocity contributions to the broadening. Turbulent velocities in the Fig. 20. Emission line intensity versus radial distance for (a) a coronal enhancement enhancement are about 25 kmls and and (b) a coronal hole region. Data were obtained during the May 30, 1965, total decrease with altitude. Temperature gra- solar eclipse. Letters label lines through data taken at the same position angle. Curves dients. when corrected for turbulent velocity broadening, suggest that solar of I/r” (n = IO and 12) are shown for comparison. wind flow is probably very weak in this bright region on the west limb (Fig. 21) lation maximum. This region has the region. whose position angle is near that de- right size and density enhancement to We also determined intensity variation termined for the eruptive disturbance qualify as a coronal condensation. If photographed by the coronal camera. other observations taken a day or two Electron density decreases by a factor of We recorded this region a second time later show that this condensation is 10 over this region. As seen in Fig. 20 when we were looking for the Ca XV sporadic or short-lived, we might 12 2 intensity falls off as l/r or as (ne) ; yellow line. The bright knot appears speculate that perhaps many such con- collisional excitation, as expected, domi- again, this time in the K coronal light densations are related to the develop- nates in this relatively dense region. from Thompson scattering that is ac- ment of hydrodynamic eruptive dis- THE 1980 CONDENSATION. One of the cepted by the interferometer along with turbances. Then previous observations most exciting observations during the the yellow line (Fig. 22). The K coronal of coronal condensations might be used 1980 eclipse was the hydrodynamic brightness in this region indicates high to determine the frequency of such mass eruptive disturbance from the base of the electron densities and so the Fe XIV eruptions throughout the solar activity brightness is probably related to in- cycle. a first look at our Fe XIV emission line creased density and not just to a tem- NEW CORONAL IMAGING TECH- data we have been able to identify a very perature close to the Fe XIV ion popu- NIQUE. The emission line data from the

30 LOS ALAMOS SCIENCE

scanning interferometer can be used to obtain a picture of the corona in the light of an emission line. Although such photographs have been obtained with expensive Lyot filters specially designed for operation at one wavelength, the new technique we present here is cheaper and quite adaptable to a variety of wavelengths. A pressure scan over one order of the interferometer is integrated onto a single photograph. As shown in Fig. 23, details of the base of two helmet streamers, the bright knot, and other features on the west limb can be seen on such an image.

Electron Temperature Experiment

Coronal electron temperatures as a function of radial distance are badly needed to determine whether electrons and ions are in thermal equilibrium. In 1976 L. E. Cram* suggested a method to determine these temperatures from spectral intensities of the K corona. Our 1980 attempt to perform the measurement used a much more sensitive technique than has been tried before and, although we failed to record data, we did confirm the feasibility of the technique. The principle behind the experiment is based on Cram’s analysis of the K coronal spectrum. This spectrum is formed as light coming from the two helmet streamers and a bright knot. photosphere (the Fraunhofer spectrum) other absorption features of the Fraun- and K absorption lines of CA II in is scattered by energetic coronal elec- hofer spectrum. Second, the spectra un- photographic records of absorption trons. The Doppler shifts suffered by dulate about these nodes with an spectra from the 1936 total eclipse). these scattered photons broaden the amplitude that is related to the electron To achieve the required sensitivity, we absorption lines in the original Fraun- temperature. Thus, the ratio of two built our instrument around the only hofer spectrum to produce the diffuse measured intensities, one at a nodal type of detector suitable for the task— spectrum known as the K corona. wavelength and one at a wavelength an array of silicon photodiodes. These Cram’s calculated spectra for a nearby, should determine the electron solid-state photosensitive devices have spherically symmetric, isothermal coro- temperature uniquely. very high quantum efficiency in the na at four electron temperatures are Although such an experiment is very spectral range of interest. shown in Fig. 24. His results display two simple in principle, in practice it is quite A commercially available detector, very interesting features. First, there ex- difficult because the intensities must be consisting of a linear array of512 silicon ist spectral “nodes,” that is, wavelengths measured with great precision. An error photodiodes on a single integrated at which the intensity is independent of circuit, served as the main component of the assumed electron temperature. These our K corona spectrograph (Fig. 25). nodes occur on either side of the H and electron temperature. The 5- 1OO/o errors This large array enabled us to look not K absorption lines of Ca II and near common in standard photometric tech- just at two wavelengths but at niques would obliterate the information *D. H. Menzel and J. M. Pasachoff. “On the *L. E. Cram, “Determination of the Temperature sought (as learned by Menzel and Obliteration of Strong Fraunhofer Lines by Elec- of the Solar Corona From the Spectrum of the tron Scattering in the Solar Corona, ” Publica- Electron-Scattering Continuum, ” Solar Physics Pasachoff* who searched without suc- tions of the Astronomical Society of the Pacific 48, 3-19 ([976). cess for residual depressions from the H 80, 458 (1968).

32 LOS ALAMOS SCIENCE THE SOLAR CORONA AT TOTALITY

Fig. 25. The K corona spectrograph for the electron temperature experiment consists of a telescope, entrance slit as- Fig. 24. Absolute intensity of the K coronal spectrum for four coronal electron sembly, a transmission grating spectro- temperatures, as calculated by Cram. The absolute intensity of the photospheric graph, a silicon photodiode detector ar- spectrum from the center of the sun's disc is shown for comparison. ray, and electronics for data acquisition and storage. The f/2.0, 400-mm focal length telescope maximizes the light at the detector. Three entrance slits, corresponding to radial distances from the

ENTRANCE by photolithographic techniques on a SLITS glass disc. A computer-controlled lever-arm and motor-drive assembly positions the selected entrance slit in the telescope focal plane. The entrance slit length is magnified by the spectrograph to match and thereby utilize the entire photodiode length. The spectrograph

per channel. Exposure times range from

be obtained only during eclipses of rela- CASSEGRAIN tively long duration or from aircraft TELESCOPE whose flight path lengthens the duration of totality.

LOS ALAMOS SCIENCE 33 wavelengths of the K coronal spectrum

The experiment was in some sense a fishing expedition. We set out to observe the entire spectrum with high precision and search for Cram’s predicted undula- tion patterns at three radial distances from the sun. During the eclipse, the spectrograph appeared to function well, and we were able to confirm that its sensitivity is adequate to measure the shape of the K coronal spectrum and, in principle, to determine electron temperatures. The data that appeared on our monitoring screen were not recorded by our com- puterized data-acquisition system (Fig. 26) because both the computer disk drive and the back-up drive failed to function. Indeed the equipment failure was a tremendous disappointment, but we did prove that the experiment was feasible, and are now redesigning some of the Fig. 26. Block diagram of computer-controlled data-acquisition and storage system electronics with hopes of flying again for electron temperature experiment. The detector array was read out with Reticon during the 1983 solar eclipse. Corporation’s evaluation circuitry. Although not ideal for this application, the circuitry provided a signal-to-noise ratio of greater than 100. Data were to be stored Infrared Observations of Dust Rings in near real time on floppy disks. Computer programs were written in FORTH, a language developed at the University of Rochester. The F corona and the zodiacal light, a zone of scattered light symmetric about dynamics of the dust particles is in- flectivity and evaporation temperature. the approximate plane of the planetary fluenced by four factors: the sun’s grav- Likely compositions include obsidian, orbits, give clear evidence of the pres- ity, radiation pressure, the Poynting- silicate-type rock, iron, and perhaps wa- ence of dust throughout much of the Robertson effect, and particle evapora- ter-ice. solar system. These observations also tion. Calculations including these factors The presence of glowing dust rings at indicate that the dust particles are very indicate that the dust will spiral in 4, 9, and 20 R@ and elsewhere has been fine—only a few micrometers in toward the sun but may ultimately settle deduced from observations of infrared diameter. Their presence may be a rem- at particular distances and form broad emission features (Fig. 27). These nant of the formation of the solar system rings about the sun. Evaporation of the emission features are presumably pro- but probably wandering comets have dust may also lead to its being blown duced as dust particles heat up and also left a contribution. outward once again. By determining the evaporate, emitting radiation corre- Whatever the source, theorists have radial location of these rings we learn sponding to black-body temperatures be- modeled the fate of the dust once it is something about the dust’s composition tween 500 and 2000 K. These observa- deposited in the solar system. The because its location depends on its re- tions are difficult to make from the

34 LOS ALAMOS SCIENCE THE SOLAR CORONA AT TOTALITY

Fig. 27. Approximate radial location and extent of infrared emission features reported

ground and, with the exception of the confirm the existence of the feature at 4 indicate that the maximum might be closer to the limb, although some of the features is in dispute. errors have less effect. Fe XIV emission line broadening can From aboard the aircraft during the Our analyses to date strongly suggest always be attributed to large-scale tur- 1980 eclipse, we made a new attempt to bulence or to expansion velocities. How- determine the spatial distribution of in- indicate no other features at greater ever, comparison with proton tem- frared emission features. distances from the sun. Thus we find no perature values from the rocket-borne Infrared intensity as a function of coronagraph at a few positions in the distance from the sun was measured at location, which corresponds to a tem- corona will allow us to determine such 2.2 µm with an InSb detector, a standard velocities if they exist. We can then technique, and at -0.7 µm with a by E. P. Nye on the basis of ground-level interpret the rest of our iron emission charge-injection-device television cam- observations at the February 26, 1979, line data in light of these findings. The era. and February 16, 1980, eclipses; how- fact that many of our 1980 profiles are The corona was scanned along the ever, at both eclipses his observations actually double-peaked indicates large ecliptic (plane of the earth’s orbit) out to may have been contaminated by the differential mass motions that are proba-

50 R@ east and west of the sun and at effects of clouds. Existence of such a bly due to the extremely complex nature several angles to the ecliptic. The scan feature would be of interest because dust of coronal features during maximum pattern was designed to determine the shells with temperatures of -800 K have solar activity, radial location and approximate extent been found about some other stars. Ion temperature gradients measured of any dust rings. in 1965 indicated that coronal holes Both detection systems worked well Conclusions could support solar wind flow. We now and yielded high-quality data. However, have high-resolution data from our 1980 tracking errors during totality will se- Most scientists studying the solar co- measurements of the residual coronal verely compromise interpretation of data rona believe that a temperature max- hole above the sun’s south pole to com- close to the sun, so we will not be able to pare with the 1965 results. We will also

LOS ALAMOS SCIENCE 35 be able to analyze temperature gradients and velocity distributions in streamer tion to the surface was quite broad. structures, to determine whether these Evidently the energy source for some structures also contribute to solar wind mass eruptions is localized and concen- flow. trated. We may learn more about the Our emission line intensity data from dynamics at the base of this disturbance several previous solar eclipse observa- once our electron density and Fe XIV tions suggested that collisional mecha- data have been reduced. nisms dominate energy transport in the Our photographic record also shows that nearby streamers respond to the resolution 1980 data from several eruption by bending around it, an effect streamer structures and one coronal con- also seen on Skylab photographs but densation, when combined with electron densities determined from camera- long streamers above the sun’s south polarimeter measurements, will be a bet- pole suggests that a second eruptive ter test of this model. disturbance may also have been in The possibility of extended wave heat- progress at the same time. ing in the corona has become increasing- Electron densities resulting from our ly significant. Our emission line intensity observations agree with other studies in data from the 1973 Concorde flight, which gave 74 minutes of totality, pro- equatorial regions the density decrease vided tentative evidence for periodic tem- with radial distance is much more gradu- perature variations, and in 1980 we al than indicated previously. This could looked for temporal changes on the sun’s have a marked effect on theoretical mod- west limb by taking a sequence of in- els of the solar wind. tensity measurements at 30-second in- In summary, we now have collected tervals over the 7 minutes of totality. But ion temperature and electron density a more extensive search will be required data at all phases of solar activity except to establish the dynamics and the extent during a deep minimum. Our 1980 data of plasma heating. combined with proton temperatures Comparison of the computer- from rocket experiments will provide a enhanced coronal photographs from five focal point for all future analysis. For the eclipses shows that the corona at the first time, we will have an accurate 1980 eclipse exhibited features never determination of material density, ion before seen, including the umbilical con- temperatures, and nonthermal velocities nection of the hydrodynamic eruptive at several places in the corona. We hope disturbance to the solar surface. This these results will be a basis for more observation calls into question previous meaningful interpretation of our data speculation, based on Skylab photo- throughout the rest of the inner corona.

36 LOS ALAMOS SCIENCE Donald H. Liebenberg earned his Bachelor of Science in 1954 and his Ph.D. in physics in 1971, both from the University of Wisconsin. He has been a Staff Member at LoS Alamos since 1961. His interest in solar physics, which dates to his undergraduate days, led in 1963 to the Laboratory’s involvement in observations of total solar eclipses. This activity has resulted in detailed studies over a solar cycle of the physical properties and energy transfer mechanisms of the coronal plasma. During 1967 and 1968 he was Program Director for Solar Terrestrial Research at the National Science Foundation. In 1973 he received a National Science Foundation Grant for observation of a solar eclipse from the French-British Concorde 001. His other research interests include properties of gases at high pressures and laser optics and technology.

Charles F. Keller, Jr., is well known for his involvement in airborne eclipse expeditions, having served as principal investigator of polarization experiments on missions in 1970, 1972, 1973, 1977, and 1980; he was scientific coordinator for missions mounted in 1979 and 1980. After receiving his Bachelor of Arts from St. Vincent College in 1961, he went on to earn a Bachelor of Science in physics from Pennsylvania State University and a Master of Science and Ph.D. in astronomy from Indiana University in 1967 and 1969, respectively. His thesis, “Computer Models of Pulsating Stars Including Radiative Transfer,” was produced jointly at Los Alamos and Indiana University. He joined the Laboratory in 1969 and has served as a Staff Member, Assistant Group Leader, and Group Leader of the Diagnostics Design Group of the Field Testing Division. At present he is with the Laboratory’s Computer Modeling Group of tbe Geosciences Division. Keller served as the Field Testing Division’s representative on the Director’s Computer Advisory Committee, and as the Laboratory’s coordinator of computer modeling for the Department of Energy’s Atmospheric Studies over Complex Terrain (ASCOT) Program. He is a member of the American Astronomical Society and its Solar Physics Division.