Mem. S.A.It. Vol. 76, 719c© SAIt 2005 Memorie della

Solar variability: a brief review

M. P. Rast

High Altitude Observatory, NCAR, Boulder, CO 80307-3000, USAe-mail: mprast@hao.ucar.edu

Abstract. This paper provides a brief review of solar particulate and radiative variabilityfocusing on four topics: solar energetic particle events, cosmic ray modulation, total solarirradiance, and solar spectral irradiance. Magnetized plasma variability is discussed onlyin the context of energetic particle fluxes. Emphasis throughout is on the current under-standing of the physical mechanisms responsible for the observed variability. Referencesare representative, not comprehensive.

Key words. Sun: variability – Sun: total solar irradiance – Sun: solar spectral irradiance –Heliosphere: cosmic ray modulation – Heliosphere: solar energetic particles

1. Introduction

The Sun’s particulate and radiative outputs dis-play variability on a wide range of spatial andtemporal scales. The observed radiative vari-ability is largely, perhaps completely, due tothe temporal evolution of the magnetic fieldsthreading the solar surface and the motion ofthese nonuniform fields with respect to a fixedobserver as a consequence of the solar rotation.The observed particulate variability results in-directly from changes in the heliospheric mag-netic field, which modulates the galactic cos-mic ray flux, and directly from changes in thesolar plasma outflow depending on the levelof solar magnetic activity. In this paper webriefly discuss four topics in solar particulateand radiative variability: solar energetic parti-cle events, cosmic ray modulation, total solarirradiance variation, and solar spectral irradi-ance variation. Magnetized plasma variability(that for example due to coronal mass ejec-tions and solar wind interactions) is discussedonly in the context of energetic particle fluxes.

We focus on current understanding of the phys-ical mechanisms governing the observed be-haviors. The discussion is necessarily brief, in-complete, and without full historical develop-ment. Consequently, references are mainly torecent reviews, within which the reader mayfind more complete discussions and more com-prehensive citations.

2. Solar particulate variability

Energetic particles observed at the Earth havetwo main sources: distant supernovae and thenearby Sun. Galactic cosmic rays are acceler-ated in supernovae shocks and have energiesto > 30GeV. Solar energetic particles are ac-celerated to energies of ∼10KeV to ∼300MeVduring magnetic reconnection in solar flares orin shocks produced by the interaction betweencoronal mass ejections and the solar wind. Theflux of both galactic cosmic rays and solar en-ergetic particles are modulated by the solarmagnetic activity cycle.

720 Rast: Solar variability

2.1. Solar energetic particle events

Solar energetic particles can plunge deep intoEarth’s polar mesosphere and upper strato-sphere, causing increases in NOx concentra-tions and consequent catalytic destruction ofO3. Energetic particle events, during whichparticle fluxes can rise by many orders of mag-nitude for periods of days, can have effectson stratospheric ozone concentrations lastingmonths to years (Rusch et al. 1981, Reid etal. 1991, Jackman et al. 1999, 2001). Theseevents appear to be of two types (sometimesmixed) associated with distinct sources and ac-celeration processes (Reames 1999, Figure 1).Impulsive events are believed to originate withparticle acceleration by resonant wave-particleinteractions at solar flare sites. The presenceof high ionization states and the enhancementof the 3He/4He and Fe/O ratios over coronalvalues suggest that the particles originate froma small region of hot (> 107K) plasma mag-netically connected to the observer. Gradualevents, on the other hand, originate with fastcoronal mass ejections which form shocks inthe solar wind. These events display elemen-tal abundances and ionization states character-istic of the ambient 1 − 2 × 106K solar coronaand wind, with high particle intensities occur-ring over a wide range of solar longitudes forseveral days. Most large solar energetic parti-cle events are of this second type.

Shock accelerated energetic particle spec-tra depend, not only on the interaction betweenthe coronal mass ejection and the solar windand thus the temporal evolution and spatialstructure of the shock itself, but also on themagnetic connectivity of the shock region withthe observer (Reames 1999, Figure 2). Themost significant particle acceleration occurs inthe strong shock region just ahead of the coro-nal mass ejection where the flow speeds arehighest. Due to the spiral structure of the he-liospheric magnetic field (§2.2), observers onthe eastern flank of the shock (left in Figure 2)are magnetically connected to the strong shockregion well before shock passage. The oppositeis true of observers on the western flank. Thuspeak energetic particle intensity can occur be-fore, after, or nearly simultaneous with shock

Fig. 1. Solar energetic particle events are of twotypes. In (a), energetic particle fluxes during a coro-nal mass ejection without significant flaring. In (b),particle fluxes during a flare unaccompanied bya coronal mass ejection. Sketches below the timetraces illustrate the extent of the source regions forthe two acceleration processes. All images fromReames (1999).

Fig. 2. The temporal evolution of the solar ener-getic particle intensity depends on the positioningof the observer with respect to coronal mass ejec-tion shock and the magnetic connectivity betweenthe two. From Reames (1999).

passage depending on the solar longitude atwhich the coronal mass ejection is observed.The intensity-time profile of any individual so-lar energetic particle event largely traces thetemporal evolution of the observers magneticconnectivity with the shock region.

Rast: Solar variability 721

Fig. 3. The number and speed of coronal mass ejec-tions (CMEs) depends on the phase of solar cycle(progressing in this figure, top to bottom, from so-lar minimum to solar maximum). Only a fraction ofall coronal mass ejections have sufficient speed toshock the solar wind. Dashed and solid lines indi-cate typical slow and fast solar wind speeds. Figurefrom Yashiro et al. (2004).

Since the slow and fast solar wind compo-nents have speeds of ∼ 400km/s and ∼ 600 −800km/s respectively, only the fastest coronalmass ejections have speeds sufficient to shockin the solar wind and cause solar energetic par-ticle events. Since both the number and thespeed of coronal mass ejections increase withincreasing solar activity (Yashiro et al. 2004,Figure 3), so to do the number of solar en-ergetic particle events observed. Interestinglyhowever, their number peaks not at solar max-imum, as might be expected, but during therising and declining phases of the solar cycle(Simunac & Armstrong 2004, Figure 4). This

Fig. 4. Sunspot number (dark) and integrated(0.394 − 440MeV) proton flux (grey) as a func-tion of time, as measured by the ChargedParticle Measurement Experiment onboard theInterplanetary Monitoring Platform 8. Figure fromSimunac & Armstrong (2004).

double peak is also seen in ice-core nitrate con-centrations (Ogurtsov et al. 2004), which addi-tionally reflect strong individual particle events(some associated with known historic whitelight flares) and long term trends well back intime (McCracken et al. 2001a,b). The cause ofthe double peak may be related to the latitu-dinal distribution of coronal mass ejections onthe Sun. They are confined to the equatorialstreamer belt during solar minimum but extendto high latitudes in the multipole configurationof solar maximum (Yashiro et al. 2004). Thiscould lead to a reduction in the number of geo-effective particle events during solar maximumdespite an absolute increase in the number andspeed of coronal mass ejections. Preliminaryestimates from the published latitudinal distri-bution histograms (Yashiro et al. 2004) how-ever suggest that this is not the case. Anotherpossibility is that both the measured chargedparticle flux and the ice-core nitrate concentra-tion are sensitive to low energy galactic cos-mic rays. As will be discussed below, galacticcosmic rays are modulated by the heliosphericmagnetic field is a way that reduces their fluxduring solar maximum (Figure 5). If some ofthe presumed solar energetic particle signal isactually due to galactic cosmic rays, the re-duced galactic cosmic ray flux at solar maxi-mum, combined with an actual increase in so-

722 Rast: Solar variability

lar energetic particle events, may explain theobserved double peak.

2.2. Cosmic ray modulation

High energy galactic cosmic rays (above thegeomagnetic shielding limit; above 1− 15GeVpole to equator) can penetrate deep into Earth’satmosphere altering isotopic concentrations inthe stratosphere and troposphere and leavingbehind a trail of ionized particles. The cosmo-genic isotopes are incorporated into ice, ocean,and biomass deposits leaving a record of thegalactic cosmic ray flux, while the ionized par-ticles may directly or indirectly modify cloudnucleation processes providing a link betweensolar activity and terrestrial weather and cli-mate (Dickinson 1975). The possibility of sucha link has lately received considerable atten-tion (eg. Svensmark & Friis-Christensen 1997,Tinsley 2000, Marsch & Svensmark 2000a, b,Sun & Bradley 2002, Kazil & Lovejoy 2004),and debate continues about both its opera-tion and importance. Notwithstanding, isotopicmeasures of cosmic ray modulation provide themost direct measure of solar variability into thepre-scientific past.

Galactic cosmic rays are thought to beaccelerated primarily in supernovae shocks.Their apparently isotropic flux into the helio-sphere suggests that the interstellar mediumis filled with highly structured magnetic fieldswhich scramble the source direction by impos-ing a convoluted path on the charged particlesas they travel from the shock acceleration siteto the heliopause. Since massive rapidly agingstars are the primary source of supernovae ex-plosions, the galactic cosmic ray flux into theheliosphere is expected to be modulated by afactor of ∼0.25−1.35 on a time scale reflectingthe interval between galactic spiral arm cross-ings (∼140× 106yr) during which compressionof the interstellar medium triggers increasedstar formation in the local solar neighborhood(Shaviv 2002). A true constancy of the galac-tic cosmic ray flux into the heliosphere, or adetailed understanding of the causes of any in-consistency, is of course needed if one is to em-ploy cosmic ray modulation (and surrogate iso-topic concentrations) to infer solar variability.

Fig. 5. Galactic cosmic ray intensity spectra ob-served at Earth around the time of solar minimum(filled symbols) and solar maximum (open sym-bols). Dashed curve is an assumed interstellar spec-trum. Figure from Caballero-Lopez et al. (2004).

Once produced, galactic cosmic raysmust make their way into the inner helio-sphere along heliospheric magnetic field lines.Collisions of cosmic rays with the ambientplasma are negligible, so the heliospheric fielddetermines their motion. Similarly, the mag-netic field itself is frozen into the highly con-ductive solar wind plasma and so it is car-ried outward by the wind. Under these condi-tions four main processes contribute to cosmicray transport: diffusion of the cosmic rays in-ward along magnetic field lines at a rate lim-ited by scattering of the charged particles fromsmall scale magnetic irregularities, advectionof the magnetic field, and therefore the cos-mic rays, outward by the solar wind, gradientand curvature drifts imparting global scale mo-tions depending on the large-scale heliosphericfield configuration, and adiabatic work by ra-dial expansion or shock compression result-ing in energy transfer between the cosmic rayparticles and solar wind plasma. These pro-cesses are summarized by Parker’s transport

Rast: Solar variability 723

Fig. 6. Cosmic ray intensity is anti-correlated withsolar activity. Alternate cycles show more or lesssharply peaked intensity maxima. The box delin-eates the time series used by Marsch and Svensmark(2000a) to support a strong correlation betweengalactic cosmic ray flux and low cloud cover. Itwould be interesting to know whether this corre-lation persists over adjacent cycles. Figure fromhttp://ulysses.sr.unh.edu .

equation governing the quasi-isotropic distri-bution function (locally isotropic velocity dis-tribution), f (x, p, t), for cosmic rays of mo-mentum p at position x and time t (Parker1965, Jokipii 1991):

∂ f∂t

=∂

∂xi

(κi j∂ f∂x j

)− Ui

∂ f∂xi− Vdi

∂ f∂xi

+13∂Ui

∂xi

(∂ f∂lnp

)+ S (xi, p, t) , (1)

where S represents the outer heliosphericboundary source.

The time scales for cosmic ray equilibra-tion in and solar wind transit of the heliosphereare on the order of 1 − 3 years. For this rea-son a first approximation to cosmic ray mod-ulation solves Equation 1 in the static limitand looks at solutions for differing magneticfield configurations depending on the phase ofthe solar cycle. If cosmic ray diffusion inwardis balanced by solar wind advection outward,an analytic spherically symmetric solution, theforce – field solution, can be obtained (Gleeson& Axford 1968). In this case the observed re-duction in cosmic ray intensity at solar maxi-mum (Figure 5) is either due to an increase inthe average solar wind speed or a decrease inthe diffusion coefficient (increased small scale

Fig. 7. (a) Solar rotation combined with solar windadvection draws the heliospheric magnetic field outinto Archimedean spiral (equatorial plane shown).Figure from Parker (1958). (b) Open field lines meetin an equatorial current sheet (axial plane shown).Figure from Low (1996). (c) The equatorial currentsheet becomes increasingly wavy before magneticfield reversal at solar maximum. Figure from Kelley(1989).

scattering). While these processes may be im-portant and are likely variable over solar cy-cle, such a simplified model does not explicitlytake into account the heliospheric field config-uration and can not therefore explain observa-tions that depend on it. One of the most intrigu-ing of these is that alternate solar cycles display

724 Rast: Solar variability

Fig. 8. Combined curvature and gradient drift di-rections for a positively charged cosmic ray particlein a heliospheric field configuration idealizing thatof solar minimum with the dipolar field oriented sothat the north polar fields point outward from theSun. Particles drift in from the heliospheric polesand are ejected rapidly outward along the equato-rial current sheet. Alternate solar cycles induce op-positely signed drift motions. Figure from Jokipii(1991).

characteristically different cosmic ray intensityprofiles (Figure 6), reflecting cycle dependentlarge-scale cosmic ray drift motions.

The heliospheric magnetic field configu-ration is determined by the inner magneticboundary condition at the Sun and the so-lar wind. Solar rotation combined with radialadvection of the magnetic field by the solarwind configure the heliospheric field in anArchimedean spiral (Parker 1958, Figure 7a),with the field pointing inward in one hemi-sphere and outward in the other, defining a cur-rent sheet in the equatorial plane (Figure 7b).The current sheet is not completely flat nor az-imuthally symmetric due to continually evolv-ing irregularities in the distribution of surfacemagnetic fields, and the waviness (Figure 7c)of the current sheet increases towards solarmaximum as it rotates to accommodate thechange in sign of the global field.

The global heliospheric field configurationleads to large scale cosmic ray drifts of magni-tude comparable to or greater than the motionsdue to solar wind advection and inward dif-fusion (Jokipii 1991). The drift motions carrycosmic ray particles inward or outward alongthe equatorial current sheet depending on thedirection of the field and sign of the charged

particles (Jokipii 1991, Figure 8). Since theglobal solar field reverses every 11 years, theheliospheric boundary origin of the galacticcosmic rays and the drift directions also doso. This is the only physical mechanism thatcan produce the distinctive alternate-cycle ef-fects in the observed cosmic ray modulation(Lockwood & Webber 2005).

3. Solar radiative variability

The presence of magnetic fields in the solarphotosphere, chromosphere, and corona affectsthe thermodynamic structure and stratificationof the solar atmosphere. As a consequence ofthe horizontal inhomogeneities in the field, animage of the Sun at any given wavelength isa composite of the solar atmosphere at differ-ing heights depending on the emissivity or ab-sorptivity of the horizontally inhomogeneousplasma at the wavelength of observation. Thisresults in distinctly different appearances of theSun at differing wavelengths (Figure 9), andin combination with with magnetic field evo-lution and solar rotation, yields wavelength de-pendent variability.

3.1. Total solar irradiance

Total solar irradiance is a measure of the spec-trally integrated incident solar energy per unitarea at one astronomical unit. It is dominatedby the solar visible, near ultraviolet, and nearinfrared radiation. For more than 25 years pre-cise measurements of the total solar irradiancehave been made from space. These observa-tions can be combined to produce a singlecomposite time series (Frohlich & Lean 2004,Figure 10), but since multiple spacecraft datasets are involved, such composites are sensitiveto modeled instrumental degradation and datamatching techniques. As a result, small but im-portant differences (eg. secular trends) exist be-tween results obtained by different groups. Allof them, however, show peak to peak total so-lar irradiance changes of ∼0.06 − 0.07 % overthe last three solar cycles. This variation canbe largely accounted for by a combination ofstrong downward excursions (reduced total so-lar irradiance) due to disk passage of sunspots

Rast: Solar variability 725

Fig. 9. The appearance of solar magnetic features depends critically on the wavelength at which they areobserved. Solar variability, due to the evolution of the magnetic fields with time and the rotation of the Sunwith respect to the observer, is thus wavelength dependent.

726 Rast: Solar variability

Fig. 10. Composite daily averaged total solar ir-radiance time series (grey curve in lower threepanels) and their 81 day running averages (blackcurves in lower three panels) as derived from spacecraft radiometer data (top panel) by separate re-search groups (see Frohlich & Lean (2004) for de-tails). Small but significant differences between thecomposite time series are apparent. Figure fromhttp://http://www.pmodwrc.ch .

and significant upward excursions (increasedtotal solar irradiance) due to the presence ofbright facular/plage regions on the solar disk.During solar activity maximum the average in-crease in facular brightening more than com-pensates for sunspot darkening, resulting in anet increase in total solar irradiance.

Why do sunspots appear dark and facu-lae bright at the near infrared, visible, andnear infrared wavelengths that dominate totalsolar irradiance measurements? Two compet-ing effects contribute to the observed intensityof solar magnetic structures: reduced radiativeopacity and suppressed convective transport.Since the magnetic field exerts a supportingforce on the ionizing plasma within a magneticflux concentration, the gas pressure and there-fore the continuum radiative opacity there is re-duced. Consequently one sees slightly deeperinto the solar photosphere in regions where

Fig. 11. Schematic of a magnetic flux tube in thesolar photosphere, showing the depressed opticaldepth surface (τ = 2/3), the suppressed convectiveflux (Fi < Fe), and the lateral heating (Fr) within thestructure. Figure from Schrijver & Zwaan (2000).

the magnetic flux density is higher, causingthe magnetic structure to appear brighter thanits surroundings. Moreover, radiation entersthe low opacity interior of the flux concentra-tion laterally from the walls. When the widthof the object exceeds the photon mean freepath (∼ 100km) an internal radial tempera-ture/intensity gradient developes, causing thestructure to look brighter when viewed from anangle than when viewed directly (Spruit 1976,Figure 11, the ’hot-wall’ model). Thus facu-lae appear brighter than their surroundings anddisplay a center-to-limb intensity variation thedetails of which which depend on the observa-tion direction and their interior structure.

Since the photon mean free path also gov-erns the depth of the photospheric bound-ary layer, and therefore the convective down-flow plume width, at about the same size atwhich lateral radiative heating of the interiorof a magnetic flux concentration becomes in-efficient the object also begins to interferewith vertical transport of heat by convection.While small magnetic flux concentrations areadvected by the turbulent flows in the photo-sphere, ultimately ending up in the convergentboundaries and vertices between supergran-ules (the solar magnetic network), large mag-netic structures inhibit the convective motionslocally, reducing the transport of heat from

Rast: Solar variability 727

Fig. 12. Passage of a spot dominated (a) and a facu-lae dominated (b) active region across the solar disk,showing the sunspot area (top), facular area (mid-dle), and total solar irradiance perturbation (bottom)for each. Figure from Lockwood (2005).

the solar interior with a consequent decreasein their internal temperature. Thus pores andsunspots appear darker than their surroundings.

Figure 12 plots the change in total so-lar irradiance during disk passage of two ac-tive regions, one (Figure 12a) dominated by asunspot and the other (Figure 12b) by faculae.Also shown for each are the disk areas occu-pied by the sunspot or faculae. The decrease intotal solar irradiance due to sunspot disk pas-sage occurs in phase with its area. The largerthe projected area of the spot on the disk thedeeper the irradiance perturbation. The totalsolar irradiance increase due to facular bright-ening (Figure 12b) is not as tightly correlatedwith facular area. Because of their center-to-limb variation, faculae near disk center andvery close to the limb have extremely low con-

Fig. 13. Empirical reconstruction (dark curves) ofthe total solar irradiance variability (light curves) forlow and high solar activity periods using sunspot ob-servations and a facular proxy. Bottom panel showssmoothed measurements and model results for a pe-riod spanning two solar cycles. Figure from Frohlich& Lean (2004).

trast with the surrounding quiet Sun and aretherefore very difficult to identify in the visi-ble continuum. As a consequence the measuredfacular area falls to zero as the active regioncrosses disk center and also quickly towardzero when it is close to either limb. Despite thisapparent decrease in disk coverage, the facu-lar contribution to the measured total solar irra-diance perturbation remains significant (lowerpanel Figure 12b).

This inability to identify the presence oflow contrast features on the Sun, despite theirsignificant contribution to total solar irradi-ance, is critical for models that attempt to de-duce irradiance variations by summing overmagnetic structure contributions. The prob-lem can be largely avoided by identifying themagnetic structures, either with proxy indicesor in images, in wavelengths at which their

728 Rast: Solar variability

contrast is large. This generally means ob-serving in a wavelength at which chromo-spheric emission is high to deduce the fac-ular contribution, and empirical models us-ing sunspot observations and a chromosphericfacular proxy (eg. the MgII index) can ac-count for 77% (or more) of the observed to-tal solar irradiance variance simply by regress-ing the sunspot and facular areas (or indices) against the observed irradiance (Frohlich &Lean 2004, Figure 13). This remarkable factsuggests that perhaps all solar radiative vari-ability results from photospheric modulationof radiative losses by the presence of sur-face magnetic structures with no accompany-ing deep rooted structural changes within theSun. The upcoming Centre National d’EtudesSpatiales micro-satellite mission Picard, to belaunched in 2008, will attempt to address thisissue directly with a combination of irradiance,helioseismic, and solar diameter measurements(Thuillier et al. 2003).

3.2. Solar spectral irradiance

Although the total solar irradiance is domi-nated by near infrared, visible, and near ultra-violet radiation from the Sun, the amplitude ofthe variability is greater at shorter wavelengths(Figure 14). This increase at short wavelengthsis due to the temperature and density structureof the solar atmosphere. Figure 15a plots thetemperature and density of a model solar at-mosphere as a function of height, along withthe formation height of lines often used for so-lar observations. Visible radiation largely orig-inates in the solar photosphere, with ultravioletand x-ray radiation having their sources in thechromosphere, transition region, and corona.Within magnetic structures, both the increasedtemperature and decreased density require asteeper temperature increase with height sothat the downward conduction of heat can bal-ance radiative loss. The larger the magneticflux density the smaller the atmospheric tem-perature scale height, resulting in an increase inthe temperature/intensity contrast between thestructures with increasing height (Figure 15b).Even sunspots become bright compared totheir surroundings in the upper chromosphere,

Fig. 14. Solar spectral irradiance (middle panel) atthe top of the Earth’s atmosphere (dark curve) andat the Earth’s surface (light curve). Absolute (bot-tom panel) and relative (top panel) irradiance changeover a solar activity cycle as a function of wave-length. Figures from Frohlich & Lean (2004).

since convective suppression no longer playsa role in determining their temperature struc-ture. As the magnetic fields evolve or the Sunrotates, the increased contrast of the magneticfeatures with height yields increased radiativevariability in wavelengths at which the radia-tion has a chromospheric or coronal origin.

Two component empirical models basedon sunspot and facular indices, derived ei-ther from chromospheric proxies or images,reproduce solar ultraviolet variability better atshorter wavelengths than longer ones (Frohlich& Lean 2004). They do so, however with lessof the total variance explained than when totalsolar irradiance variations are similarly mod-eled. This is because the details of the spec-tral irradiance variability amplitude and phas-ing depend critically on the observed struc-tures, their position, and the wavelengths at

Rast: Solar variability 729

Fig. 15. In (a), temperature and density profilesin a model solar atmosphere and the approximateformation heights of selected lines and continua.Figure from Avrett (1992). In (b), magnetize atmo-sphere models of Fontenla et al. (1999) indicatingthe increased temperature contrast between mag-netic structures with height. Figure from Frohlich &Lean (2004).

which they are observed. These are not easilycaptured by simple proxy models. As an ex-ample, consider the observed full and half so-lar rotation period variability observed by theSolar EUV Experiment as a function of wave-length (Woods et al. 2005, Figure 16). Shortwavelength coronal emissions have an elevated13.5day periodicity because prominent struc-tures off the limb rotate limb to limb in one-half rotation period. By contrast, the elevated13.5 day periodicity in the far ultraviolet isgreatest at those wavelengths with the largestcenter to limb darkening. Here the half-periodpower is due to the appearance and disappear-ance of active regions as they cross the disk.In order to empirically reproduce the solarspectral variability the atmospheric structure of

Fig. 16. Solar variability at full and half rotationalperiods as a function of observation wavelength.Half-period (13.5 day) variability is enhance atwavelengths that are strongly limb brightened (coro-nal emissions off limb) and at those that are stronglylimb darkened (ultraviolet emission at disk center).Figure from Woods et al. (2005).

all components must be known well enoughto synthesize their spectral contributions, as afunction of viewing angle, over broad and wellresolved wavelength bands.

The focus of solar radiative variability re-search depends sensitively on one’s primaryinterest. Interest in the coupling between so-lar output and terrestrial climate motivatesthe question of what amplitude variation oc-curs on what time scales at what frequencies.This question is best addressed by observa-tions aimed at the spectral mapping of so-lar magnetic structures over broad but wellresolved wavelength bands to understand theSun’s spectral output as a function of themagnetic flux density and structure size dis-tribution on its surface. The frequencies ob-served should be informed by climate sensi-

730 Rast: Solar variability

tivity. Interest in the underlying physical ori-gin of solar radiative variability itself motivatesthe question of whether all the observed vari-ability is due to surface modulation of radia-tive loss by magnetic structures or also reflectssome additional as yet unknown thermal mod-ification of the interior. This question is bestaddress by observations that aim at disentan-gling surface effects from deeper rooted causesby simultaneously measuring the surface mag-netic fields and photometric intensity in highspatial resolution images. For this purpose in-frared observations which penetrate the deepphotosphere at the opacity minimum shouldaccompany those made in the visible.

Acknowledgements. The National Center forAtmospheric Research is sponsored by the NationalScience Foundation. Special thanks to I. Ermolliand T. Holzer.

References

Avrett, E.H. 1992, SOLER22 Workshop, p. 20Caballero-Lopez, R.A., et al. 2004, JGR, 109,

A12102Dickinson, R.E. 1975, BAMS, 56, 1240Fontenla, J., et al. 1999, ApJ, 518, 480Frohlich, C., & Lean, J. 2004, A&A Rev, 12,

273Gleeson, L.J., & Axford, W.I. 1968, ApJ, 154,

1011Jackman, C.H., et al. 1999, Adv. Space Res.,

24, 625Jackman, C.H., et al. 2001, GRL, 28, 2883Jokipii, J.R. 1991, in The Sun in Time, p. 205Kazil, J., & Lovejoy, E.R. 2004, JGR, 109,

D19206

Kelley, M.C. 1989, The Earth’s Ionosphere,San Diego: Academic Press

Lockwood, J.A., & Webber, W.R. 2005, JGR,110, A04102

Lockwood, M. 2005, in The Sun, SolarAnalogs and the Climate, p. 109

Low, B.C. 1996, Solar Phys., 167, 217Marsch, N.D., & Svensmark, H. 2000a, Phys.

Rev. Lett. 85, 5004Marsch, N., & Svensmark, H. 2000b, SSRv 94,

215McCracken, K.G., et al. 2001a, JGR, 106,

21585McCracken, K.G., et al. 2001b, JGR, 106,

21599Ogurtsov, M.G., et al. 2004, Solar Phys., 222,

177Parker, E.N. 1958, ApJ, 128, 664Parker, E.N. 1965, P&SS, 13, 9Reames, D.V. 1999, SSRv, 90, 413Reid, G.C., Solomon, S., & Garcia, R.R. 1991,

GRL, 18, 1019Rusch, D.W., et al. 1981, P&SS, 29, 767Schrijver, C.J., & Zwaan, C. 2000, Solar

and Stellar Magnetic Activity, Cambridge:Cambridge Univ Press

Shaviv, N.J. 2002, Phys. Rev. Lett., 89, 051102Simunac, K.D.C., & Armstrong, T.P. 2004,

JGR, 109, A10101Spruit, H.C. 1976, Solar Phys., 50, 269Sun, B., & Bradley, R.S. 2002, JGR, 107, 4211Svensmark, H., & Friis-Christensen, E. 1997,

J. Atmos. Sol. Terr. Phys., 59, 1225Thuillier, G., Joukoff, A., &, Schmutz W. 2003,

ISCS Symposium 23, ESA SP-535, 251Tinsley, B.A. 2000, SSRv, 94, 231Woods, T.N., et al. 2005, JGR, 110, A01312Yashiro, S., et al. 2004, JGR, 109, A07105