Solar Spectroheliogram Spectral Atlas

• Spectral atlas of solar spectroheliograms is an organized collection that represents solar structures across wavelengths, merging historical imagery with modern 3D datacubes.
• It integrates diverse resources including broad-band 1D spectra, slit-spectroscopic atlases, and full-disk imaging to annotate spectral lines and magnetic features.
• The atlas provides actionable insights into center-to-limb variations using quantitative metrics to guide the selection of lines for magnetic diagnostics and morphology analysis.

Searching arXiv for papers on solar spectral atlases and spectroheliograms. The spectral atlas of solar spectroheliograms is the organized representation of solar structure as a function of wavelength, line position, and observing geometry, spanning both monochromatic full-disk imagery and spectrally resolved reference datasets. In the literature, the term does not refer to a single data product. It encompasses several related classes of resources: broad-band compilations of 1D solar spectra used for line identification and magnetic diagnostics; slit-spectroscopic atlases of disk center and limb spectra; synoptic products derived from historical H$\alpha$ spectroheliograms; full-disk spectroheliogram atlases indexed by wavelength; and, more recently, full-disk $(x,y,\lambda)$ datacubes that make it possible to reconstruct spectroheliograms at many sampled wavelengths. Taken together, these resources define the observational basis for interpreting how solar morphology changes from continuum to line wing to line core, from photosphere to chromosphere, and from disk center to limb (Malherbe, 2024).

1. Historical emergence of the spectroheliographic atlas

The modern concept derives from the spectroheliograph itself, designed for monochromatic imagery of the Sun. Henri Deslandres initiated imaging spectroscopy of the solar atmosphere in 1892 at Paris Observatory and invented, independently of George Hale, the photographic spectroheliograph for monochromatic solar imaging. Deslandres developed two instrument classes: the “spectrohéliographe des formes,” a narrow-bandpass imaging instrument used to reveal filaments, prominences, plages, and active regions; and the “spectrohéliographe des vitesses,” or “section” spectroheliograph, which recorded the line profiles of cross-sections of the Sun so that Doppler shifts of dynamic structures could be measured (Malherbe, 2023).

At Meudon, this program became systematic. Ca II K observations began in 1908 and H$\alpha$ in 1909, eventually producing one of the world’s longest solar image archives, with more than 100,000 observations over 10 solar cycles (Malherbe, 2023). The archive was built around the idea that different parts of a line profile image different atmospheric layers. In the Meudon framework, Ca II K3 and H$\alpha$ were used for the chromosphere, while Ca II $K1_v$ was used for the photosphere; long-exposure Ca II K3 observations with a disk attenuator were used for prominences (Malherbe, 2023).

This historical distinction between line core and line wing is foundational for any spectral atlas of spectroheliograms. The Meudon studies concluded that the standard choices of Ca II K3, Ca II $K1_v$, and H$\alpha$ were the most effective basis for imaging spectroscopy of the chromosphere and photosphere, even after tests in Balmer lines, He I, Ca I, Fe I, Sr II, Mg I, Na I, and Ca II infrared lines (Malherbe, 2023). A plausible implication is that the canonical solar spectroheliogram atlas is not merely a collection of images but a height-structured spectral system whose interpretive power depends on where the passband is placed within each line.

2. Spectral reference atlases and line-based annotation

A second major component of the field is the 1D solar spectral atlas used to identify lines, compare center and limb profiles, and evaluate magnetic sensitivity. A recent compilation brings together the atlases of Delbouille, Kurucz, Stenflo, and Gandorfer into a single presentation spanning $\lambda=3000$–$8800\,\mathrm{\AA}$, with Moore-table identifications and equivalent Landé factors $g^*$ (Malherbe, 2024). Its purpose is explicitly practical: it is a visually cross-referenced comparative atlas, not a replacement for the original sources.

The observing geometries in that compilation are central. The disk-center spectra are at $(x,y,\lambda)$0, while the limb references are at $(x,y,\lambda)$1 and $(x,y,\lambda)$2, with

$(x,y,\lambda)$3

Here $(x,y,\lambda)$4 corresponds to the line of sight normal to the surface, whereas the limb spectra correspond to strongly oblique viewing (Malherbe, 2024). The atlas restricts its effective coverage to $(x,y,\lambda)$5–$(x,y,\lambda)$6 because that is the range for which line identifications from Moore, Minnaert, and Houtgast are available; $(x,y,\lambda)$7 annotations are available only from $(x,y,\lambda)$8 to $(x,y,\lambda)$9 (Malherbe, 2024).

A distinctive feature of this compilation is that all spectra are continuum normalized, written in the paper as $\alpha$0, with the intent that each spectrum is normalized to the adjacent continuum (Malherbe, 2024). This facilitates direct visual comparison across instruments and sites. The atlas also encodes equivalent width using violet dashed vertical bars of length

$\alpha$1

with $\alpha$2 in m\AA. The paper gives the Ca II K example at $\alpha$3, where $\alpha$4 yields $\alpha$5 (Malherbe, 2024). For spectroheliogram planning, this matters because line choice depends not only on identity but on strength, blending, and magnetic sensitivity.

The magnetic annotation is especially relevant to active-region spectroheliography. The compilation integrates equivalent Landé factors $\alpha$6 as blue numerical annotations and emphasizes that Stokes $\alpha$7 and the circular polarization rate are proportional to $\alpha$8 (Malherbe, 2024). This suggests that the atlas is not only descriptive but operational: it can guide the selection of lines for magnetic spectroheliograms, filtergrams, and Fabry–Pérot observations.

A different type of reference atlas, "HelioSpectrotron 5000" (Pietrow, 23 Feb 2026), addresses instrument matching rather than solar line formation itself. It is based on the Hamburg FTS atlas and provides absolute-intensity and continuum-normalized spectra at arbitrary wavelength ranges and resolutions, with curated line identifications and optional telluric contamination. Its current implementation is disk-center only, so it is not a spatially resolved atlas or a center-to-limb atlas, but it is explicitly designed to support spectroheliograph users by enabling wavelength calibration, blend recognition, and simulated 2D slit-like context displays (Pietrow, 23 Feb 2026).

3. Center-to-limb atlases as the spectroscopic basis of spectroheliograms

A third pillar of the subject is the center-to-limb atlas. Spectroheliograms are often interpreted as if a given line has a fixed morphology across the disk, but several atlases show that line depth, width, wings, and even contrast signatures change strongly with $\alpha$9.

One influential reference is the FTS atlas of the Sun’s spectrally resolved center-to-limb variation, built from a disk-center atlas and a limb atlas recorded 10 arcsec inside the solar limb, corresponding to an adopted $\alpha$0 (Stenflo, 2014). The key quantity is the ratio spectrum

$\alpha$1

where $\alpha$2 is continuum-normalized intensity. The full center-to-limb variation factorizes as

$\alpha$3

The paper identifies this ratio spectrum as SS3, distinct from SS1, the ordinary intensity spectrum, and SS2, the Second Solar Spectrum (Stenflo, 2014).

The empirical result is that SS3 is as richly structured as the ordinary intensity spectrum but differs strongly in shape and amplitude. Weak lines can disappear, moderate LTE-type lines can be enhanced, and strong non-LTE lines such as H$\alpha$4 6563 Å and Ca II 8542 Å behave differently (Stenflo, 2014). For weak to medium-strong LTE-type lines, the paper proposes the mapping

$\alpha$5

with

$\alpha$6

and wavelength-dependent parameters

$\alpha$7

with $\alpha$8 in Å (Stenflo, 2014). In atlas terms, this makes center-to-limb line behavior a quantitative constraint on model atmospheres, not merely a visual curiosity.

The same theme appears in the IRSOL atlas of the solar intensity spectrum and its center-to-limb variation, which samples 10 heliocentric angles from $\alpha$9 to $K1_v$0 in steps of 0.1 over $K1_v$1–$K1_v$2 nm (Ramelli et al., 2017). That atlas uses the ratio

$K1_v$3

and recommends reconstructing the intensity spectrum at a given $K1_v$4 as

$K1_v$5

combining the observed ratio with Neckel continuum CLV data and the Kurucz FTS disk-center atlas (Ramelli et al., 2017). The data are interpreted as information about the anisotropy of the emergent radiation field and therefore as observational input for modeling the Second Solar Spectrum.

A higher-resolution resolved atlas from Göttingen extends this logic. The IAG center-to-limb atlas of the spatially resolved quiet Sun covers $K1_v$6–$K1_v$7 at $K1_v$8, corresponding to $K1_v$9 at $K1_v$0, and samples 14 heliocentric positions from $K1_v$1 to $K1_v$2 (Ellwarth et al., 2023). Its wavelength correction uses telluric O$K1_v$3 lines, with

$K1_v$4

and it corrects for solar differential rotation using

$K1_v$5

The atlas shows that line depth decreases, lines broaden, and line centers shift to longer wavelengths toward the limb; Fe I 6175 Å bisectors are generally consistent with previous observations but differ from model line profiles, especially close to the solar limb (Ellwarth et al., 2023). This is directly relevant to spectroheliograms because a narrowband image at fixed wavelength samples a different local spectral context at different $K1_v$6.

A complementary slit-spectroscopic atlas of the Sun from 3980 to 7100 Å compares $K1_v$7 and $K1_v$8 at $K1_v$9 and SNR 400–600 (Fathivavsari et al., 2014). Its graphical presentation includes the difference spectrum

$\alpha$0

which makes center-to-limb profile changes and telluric signatures easy to identify (Fathivavsari et al., 2014). The paper highlights that the Balmer-line wings of H$\alpha$1 and H$\alpha$2 are narrower at the limb than at disk center, interpreting this as weaker Stark broadening in cooler, higher atmospheric layers (Fathivavsari et al., 2014).

4. Imaging atlases and wavelength-indexed full-disk morphology

The most literal realization of a spectral atlas of solar spectroheliograms is a full-disk imaging atlas indexed by wavelength. A recent example is "An Atlas of Spectroheliograms from 3641 to 6600 Å" (Nagy et al., 17 Jul 2025), which presents a large spectral atlas of full-disk solar spectroheliograms spanning $\alpha$3–$\alpha$4, with continuous coverage from $\alpha$5 to $\alpha$6 and sparser coverage outside that interval. The spectral resolution varies between $\alpha$7 and $\alpha$8, the spectral step is $\alpha$9–$\lambda=3000$0 mÅ, the spatial resolution averages around $\lambda=3000$1 arcseconds, and the atlas contains approximately 50,000 spectroheliograms (Nagy et al., 17 Jul 2025).

The data were acquired between 2025-04-20 and 2025-07-06 using amateur spectroheliographs, Sol'Ex and ML Astro SHG 700, both with 2400 lines mm$\lambda=3000$2 gratings, and both using a ZWO ASI 678MM camera (Nagy et al., 17 Jul 2025). The atlas is explicitly designed to fill a gap left by classic solar spectral atlases, which provide 1D spectra and line lists but not full-disk morphology at each wavelength. Its online interface links 1D and 2D spectral context to individual monochromatic full-disk images, enabling direct browsing from line wing to line core (Nagy et al., 17 Jul 2025).

The reduction strategy is procedural rather than radiometric. JSol'Ex finds a polynomial describing the line curvature (“smile”), straightens the spectrum, geometrically circularizes and centers the disk, and rotates the result to align with the solar equator (Nagy et al., 17 Jul 2025). Calibration uses Moore et al. (1966) and the BASS2000 atlas derived from the Liège atlas. The authors explicitly avoided cosmetic corrections such as anti-jagging and CLAHE (Nagy et al., 17 Jul 2025). Because adjacent spectral rasters were usually not recorded on the same day, the atlas is spectrally continuous but not temporally simultaneous across the full wavelength range (Nagy et al., 17 Jul 2025). This is a major interpretive caveat: apparent morphological differences between distant wavelengths may partly reflect solar evolution.

A different but related imaging approach appears in the 2025 VTT drift-scan demonstration of multi-wavelength center-to-limb observations (Verma et al., 7 Nov 2025). That work is not yet a mature atlas release, but it shows that the VTT echelle spectrograph and FaMuLUS cameras can produce high-spectral-resolution drift scans suitable for spectroheliograms and parameter maps. At H$\lambda=3000$3, the nominal resolving power is $\lambda=3000$4; each spectrum contains 1856 wavelength points over $\lambda=3000$5, with reciprocal dispersion $\lambda=3000$6, and each scan records approximately 800,000 individual spectra per scan and per spectral line (Verma et al., 7 Nov 2025). The paper explicitly states that data products include spectroheliograms and maps of line-of-sight velocity, line width, and line-core intensity, while quiet-Sun CLV spectra are averaged in $\lambda=3000$7 bins from 0.2 to 1.0 in steps of 0.1 (Verma et al., 7 Nov 2025). This suggests a transition from atlas-as-image to atlas-as-physical-parameter field.

5. Synoptic and temporal atlases from historical H$\lambda=3000$8 spectroheliograms

The spectral atlas of spectroheliograms also has a temporal, synoptic form. The Kodaikanal H$\lambda=3000$9 archive spans 1914–2007 and is described as the oldest uniform digitized dataset with daily images available today in H$8800\,\mathrm{\AA}$0 (Chatterjee et al., 2017). The observations were taken with a spectroheliograph using a 30 cm objective at $8800\,\mathrm{\AA}$1, recorded first on photographic plates and later on film. After digitization with a $8800\,\mathrm{\AA}$2 CCD cooled to $8800\,\mathrm{\AA}$3C, the full-disk images achieve a plate scale of about $8800\,\mathrm{\AA}$4 arcsec pixel$8800\,\mathrm{\AA}$5, with effective resolution about $8800\,\mathrm{\AA}$6 arcsec (Chatterjee et al., 2017).

The Kodaikanal pipeline inverts grayscale density images, detects the solar disk using edge operators and a circle Hough transform, centers the image, derives an asymmetric limb-darkening profile by median filtering, and divides the original image by that profile to produce limb-darkening-corrected images (Chatterjee et al., 2017). The paper explicitly states that no density-to-intensity calibration has been performed yet, so the maps are suitable for morphology and relative contrast structure, not for absolute photometric use (Chatterjee et al., 2017).

The synoptic-map construction is highly specific. For each limb-darkening-corrected H$8800\,\mathrm{\AA}$7 image, the pipeline extracts longitude bands from $8800\,\mathrm{\AA}$8 to $8800\,\mathrm{\AA}$9 around the central meridian, applies $g^*$0-angle correction, weights heliographic longitude by a cosine fourth-power function, and accumulates strips over 27.2753 days to build Carrington maps of size $g^*$1, with the vertical axis in sine latitude from $g^*$2 to $g^*$3 (Chatterjee et al., 2017). Filaments are then segmented semi-automatically with histogram equalization and 8-neighbourhood region growing, and fragmentation effects are tested with morphological closing using disk kernels of radii 22 and 30 pixels (Chatterjee et al., 2017).

This synoptic atlas yields scientifically rich outputs: filament butterfly diagrams, poleward migration patterns, and a correlation of $g^*$4 between the delay of polar-filament-number maxima relative to sunspot maxima and the delay of polar-field reversal relative to sunspot maxima, with adjusted coefficient 0.98 using

$g^*$5

Because only three cycles enter that correlation, the paper treats it cautiously (Chatterjee et al., 2017). In the context of a spectral atlas of spectroheliograms, the significance lies less in the correlation itself than in the demonstration that historical H$g^*$6 spectroheliograms can be transformed into a longitudinally indexed atlas of chromospheric magnetic structure.

6. Full-disk spectral datacubes and the transition to 3D solar atlases

The most advanced realization of the subject is the full-disk spectral datacube. Since 2017, Meudon has recorded daily spectroscopic datacubes in H$g^*$7, Ca II K, Ca II H, and H$g^*$8, stored as 3D FITS files $g^*$9 (Malherbe, 18 May 2026). This is a major transition from traditional monochromatic spectroheliograms to a genuine full-disk spectral atlas.

The datacubes have the following properties. H$(x,y,\lambda)$00 is recorded at $(x,y,\lambda)$01, Ca II K at $(x,y,\lambda)$02, Ca II H at $(x,y,\lambda)$03, and H$(x,y,\lambda)$04 at $(x,y,\lambda)$05, the last lying in the wing of Ca II H (Malherbe, 18 May 2026). The spectral resolution is $(x,y,\lambda)$06 for Calcium and H$(x,y,\lambda)$07, with spectral sampling $(x,y,\lambda)$08, and $(x,y,\lambda)$09 for H$(x,y,\lambda)$10, with spectral sampling $(x,y,\lambda)$11. The spatial sampling is about 1 arc sec, while the usual seeing is 2 arcsec (Malherbe, 18 May 2026). The archive provides 100 spectral points for Ca II K, Ca II H, and H$(x,y,\lambda)$12, and 40 spectral points for H$(x,y,\lambda)$13 (Malherbe, 18 May 2026).

The data model distinguishes level 0 and level 1. Raw level 0 files are 3D 16-bit unsigned integer TIF cubes in $(x,y,\lambda)$14, with no correction for dark current, line curvature, line inclination, solar $(x,y,\lambda)$15 angle, or coelostat rotation; they preserve line-profile fidelity because no interpolation is applied (Malherbe, 18 May 2026). Processed level 1 FITS cubes are 3D 16-bit unsigned integer FITS files in $(x,y,\lambda)$16, corrected for dark current, line inclination, line curvature via parabolic adjustment, coelostat rotation, and solar $(x,y,\lambda)$17-angle rotation so that solar north is up; rotation uses bicubic interpolation (Malherbe, 18 May 2026). The paper explicitly notes that processed FITS cubes are directly usable for science but slightly degrade spectral resolution, spatial resolution, and photometric accuracy of line profiles, and that level 1 files lose about 10% of spectral pixels because of geometric correction (Malherbe, 18 May 2026).

This archive already begins to function as a feature-resolved spectral atlas. The paper presents representative profiles for quiet Sun, bright points, flare kernels, sunspots, faculae, filaments, and prominences (Malherbe, 18 May 2026). In bright points and flare kernels, Ca II H3 and K3 show intense core emission, while H$(x,y,\lambda)$18 becomes emissive; for the flaring active region of 6 November 2017, the paper states that H$(x,y,\lambda)$19 is not so reactive by comparison (Malherbe, 18 May 2026). In sunspots, the profiles are dark relative to disk center, with small emission in Ca II H3 and K3, while H$(x,y,\lambda)$20 remains in absorption (Malherbe, 18 May 2026). In faculae, there is small facular emission in Ca II H3 and K3, H$(x,y,\lambda)$21 is in emission, and K2 and H2 peaks appear (Malherbe, 18 May 2026). Filaments are visible in absorption in Ca II H3 and H$(x,y,\lambda)$22, though only their densest parts appear in H$(x,y,\lambda)$23; prominences are bright in Ca II H3 and H$(x,y,\lambda)$24, while H$(x,y,\lambda)$25 is very faint and traces only the densest parts (Malherbe, 18 May 2026).

This suggests an important redefinition of the atlas concept. In earlier spectroheliograph practice, the atlas indexed wavelength-selected images. In the datacube era, the atlas can instead index feature classes by their local line profiles and derive line-center, wing, continuum, contrast, and parameter maps from the same underlying cube.

7. Scope, limitations, and conceptual boundaries

A recurring issue in the literature is terminological. Many papers highly relevant to the spectral atlas of solar spectroheliograms are not spectroheliogram atlases in the strict 2D imaging sense. The Delbouille–Kurucz–Stenflo–Gandorfer compilation is a 1D comparative spectral atlas (Malherbe, 2024). The IRSOL and FTS CLV atlases are slit- or aperture-based center-to-limb atlases, not full-disk image atlases [(Ramelli et al., 2017); (Stenflo, 2014)]. The IAG center-to-limb atlas is a resolved quiet-Sun atlas across 14 $(x,y,\lambda)$26 values, but not a spatially continuous 2D map (Ellwarth et al., 2023). Precision atlases such as the LFC-calibrated solar atlas are disk-integrated, not spatially resolved (Molaro et al., 2013). Telluric-corrected flux atlases are similarly Sun-as-a-star references (Baker et al., 2020). Yet all of these are indispensable because spectroheliogram interpretation depends on line identification, absolute wavelength standards, telluric discrimination, and center-to-limb spectral behavior.

The literal imaging atlases also have strong constraints. The 2025 wavelength-indexed full-disk atlas is broad and large, but only $(x,y,\lambda)$27–$(x,y,\lambda)$28 is continuous, quality varies with seeing and instrument stability, adjacent cubes were usually not taken on the same day, and public products are downsampled JPEGs rather than the original FITS reconstructions (Nagy et al., 17 Jul 2025). Historical synoptic atlases such as Kodaikanal are extraordinarily long but are not density-to-intensity calibrated, contain scratches and defects, and require semi-manual intervention in segmentation (Chatterjee et al., 2017). Modern datacubes such as the Meudon archive provide full $(x,y,\lambda)$29 information, but the processed products introduce interpolation, are not absolutely radiometrically calibrated, and their moderate spectral resolution remains lower than large-telescope spectroscopy by about a factor of 10 in sampling (Malherbe, 18 May 2026).

A common misconception is that a spectroheliogram atlas and a spectral atlas are interchangeable. The literature does not support that equivalence. Instead, it supports a layered view. A true spectral atlas of solar spectroheliograms requires at least four linked ingredients: full-disk or drift-scan morphology indexed by wavelength; line-identification and magnetic-sensitivity annotation; center-to-limb reference spectra; and, increasingly, full-disk datacubes that preserve the local line profile. This suggests that the mature form of the field is converging toward a hybrid atlas architecture: historical monochromatic archives for temporal depth, full-disk wavelength-indexed atlases for spectral browsing, center-to-limb atlases for radiative-transfer interpretation, and daily datacubes for feature-resolved spectral classification.