Multi-scale Spectral Index Maps

• Multi-scale spectral index maps are spatially resolved representations that capture how the emission from astronomical sources changes with frequency across various spatial scales.
• They utilize methodologies such as common-beam pixel alignment, wideband deconvolution, and adaptive binning to extract spatially distinct spectral information.
• They enable detailed component separation in radio jets, analysis of dust evolution in protoplanetary discs, and calibration of diffuse Galactic synchrotron emissions.

Searching arXiv for recent and foundational papers on multi-scale spectral index maps. Multi-scale spectral index maps are spatially resolved representations of how measured emission changes with frequency across an extended source, a sky region, or an adaptively defined set of subregions. In the current literature, the same basic construct appears in several forms: pixel-by-pixel radio maps of active galactic nuclei, wavelength-pair maps of protoplanetary discs, photon-index maps of pulsar wind nebulae, patch-based synchrotron index maps of the Galaxy, and event-horizon-scale maps of black-hole accretion flows (Kim et al., 2014, Pavlyuchenkov et al., 2019, Järvelä et al., 2021, Guest et al., 2020, Ricarte et al., 2022, Nasirudin et al., 15 Sep 2025). The qualifier “multi-scale” refers not to a single formalism but to a family of practices in which spectral slopes are tracked across multiple spatial resolutions, wavelength intervals, adaptive bins, or hierarchical sky partitions.

1. Definitions and notation

The most common radio and millimetre convention writes flux density or specific intensity as a power law, $S_\nu \propto \nu^\alpha$ or $I_\nu \propto \nu^\alpha$, with the two-band estimate

$\alpha = \frac{\log(S_1/S_2)}{\log(\nu_1/\nu_2)}$

or its intensity equivalent (Järvelä et al., 2021, Pavlyuchenkov et al., 2019, Kim et al., 2014). In large-scale Galactic work, the temperature spectral index is often എഴുത as

$$\beta_{\nu_1-\nu_2}=-\frac{\log\left(T_{\nu_1}/T_{\nu_2}\right)}{\log\left(\nu_1/\nu_2\right)},$$

with the Rayleigh–Jeans relation $\beta=\alpha+2$ connecting temperature and intensity conventions (Guzmán et al., 2010). In X-ray studies of pulsar wind nebulae, the mapped quantity is usually the photon index $\Gamma$ of an absorbed power law,

$F(E) \propto e^{-N_{\rm H}\sigma(E)} E^{-\Gamma},$

rather than $\alpha$ directly, although the work still functions as a spatially resolved spectral-index analysis (Guest et al., 2020).

A recurrent caution in the literature is that a measured spectral index is not a direct one-to-one tracer of a single physical property. In protoplanetary discs, $\alpha$ depends not only on dust opacity but also on optical depth $\tau_\nu$ and temperature $I_\nu \propto \nu^\alpha$0, through

$I_\nu \propto \nu^\alpha$1

so the same observed $I_\nu \propto \nu^\alpha$2 can arise either from optically thick emission with ordinary dust or from optically thin emission dominated by large grains (Pavlyuchenkov et al., 2019). In diffuse Galactic synchrotron inference, small shifts in $I_\nu \propto \nu^\alpha$3 can propagate into large absolute temperature residuals at low frequency, even when the reference-frequency map is well reproduced (Nasirudin et al., 15 Sep 2025). These results show that a spectral index map is best understood as a diagnostic field whose interpretation is model-dependent.

2. Construction methodologies

The simplest construction uses two aligned images with matched resolution. “VIMAP” implements this approach for multi-frequency VLBI maps of AGN. Because self-calibration removes absolute positional information and the core position itself shifts with frequency, the maps must first be convolved to a common beam, placed on the same pixel grid, and aligned by masking the core and maximizing a two-dimensional cross-correlation coefficient over the optically thin jet emission. After registration, the code computes both a spectral-index map and an error map, with cutoffs on minimum intensity and maximum allowed index error (Kim et al., 2014).

Wideband radio imaging generalizes this idea by estimating spectral behavior during deconvolution rather than after separate imaging. In the JVLA study of absorbed-jet narrow-line Seyfert 1 galaxies, CASA’s multi-term multi-frequency synthesis (mtmfs) models the image with Taylor terms around a reference frequency $I_\nu \propto \nu^\alpha$4, with TT0 as the intensity map and TT1 yielding the first-order spectral behavior through $I_\nu \propto \nu^\alpha$5. The workflow includes wideband primary-beam correction, masking below $I_\nu \propto \nu^\alpha$6, removal of regions with $I_\nu \propto \nu^\alpha$7, and smoothing of the $I_\nu \propto \nu^\alpha$8 and $I_\nu \propto \nu^\alpha$9 maps to the clean-beam resolution (Järvelä et al., 2021).

Other domains replace fixed pixels with statistically controlled regions. In Chandra studies of pulsar wind nebulae, adaptive binning with CIAO/CONTBIN produces “puzzle-piece” regions that follow surface-brightness structure while meeting a minimum counts criterion; spectra are extracted for each region and epoch and then fitted in XSPEC. Significance maps, error maps, and reduced-$\alpha = \frac{\log(S_1/S_2)}{\log(\nu_1/\nu_2)}$0 maps are used alongside the photon-index map, so the output is a spatially resolved statistical comparison rather than a purely visual product (Guest et al., 2020).

Large-area synchrotron analyses often adopt hierarchical or Bayesian constructions. A Bayesian recalibration framework for the Haslam 408 MHz map jointly infers the true sky brightness field $\alpha = \frac{\log(S_1/S_2)}{\log(\nu_1/\nu_2)}$1, spatially varying flux-scale factors $\alpha = \frac{\log(S_1/S_2)}{\log(\nu_1/\nu_2)}$2, and optionally regional synchrotron spectral indices $\alpha = \frac{\log(S_1/S_2)}{\log(\nu_1/\nu_2)}$3, combining a $\alpha = \frac{\log(S_1/S_2)}{\log(\nu_1/\nu_2)}$4 408 MHz map with absolutely calibrated $\alpha = \frac{\log(S_1/S_2)}{\log(\nu_1/\nu_2)}$5 maps between 50 and 150 MHz. The inference uses a Gibbs sampler, Gaussian constrained realizations for Gaussian conditional posteriors, and MCMC for the nonlinear spectral-index parameters (Nasirudin et al., 15 Sep 2025). A complementary semi-blind approach reconstructs the first two synchrotron moments with constrained ILC and then estimates $\alpha = \frac{\log(S_1/S_2)}{\log(\nu_1/\nu_2)}$6 by T–T regression between the recovered moment maps on sky patches, yielding a patch-based spectral-index map at $\alpha = \frac{\log(S_1/S_2)}{\log(\nu_1/\nu_2)}$7 (Adak et al., 20 Oct 2025).

3. What “multi-scale” denotes

Across the literature, “multi-scale” has several non-equivalent meanings.

Domain	Scale construction	Representative outcome
VLBI and JVLA AGN	common-beam pixel maps across frequency	core, jet, and absorbed components separated
Protoplanetary discs	multiple wavelength intervals	$\alpha = \frac{\log(S_1/S_2)}{\log(\nu_1/\nu_2)}$8-maxima trace different grain-size windows
PWNe	adaptive spatial bins and multi-epoch comparison	localized spectral evolution revealed
Diffuse synchrotron sky	mixed angular resolutions and regional partitions	coarse $\alpha = \frac{\log(S_1/S_2)}{\log(\nu_1/\nu_2)}$9 zones and recalibrated all-sky models
EHT accretion flows	event-horizon, photon-ring, and outer-flow scales	radial and photon-ring spectral-index structure

In AGN and PWNe, the emphasis is on preserving small structures without sacrificing spectral reliability. In the NLS1 JVLA analysis, 1.6 GHz is best for diffuse steep-spectrum emission, 5.2 GHz is intermediate, and 9.0 GHz is more sensitive to compact cores; together they span almost a factor of 6 in frequency and allow spectral curvature to be identified (Järvelä et al., 2021). In PWNe, the “multi-scale” character is practical rather than formal: bins must be small enough to retain arcsecond-scale variability, or the changes are smoothed away (Guest et al., 2020).

In protoplanetary discs, multi-scale mapping is tied to wavelength-dependent grain-size sensitivity. Maps formed from 0.45–0.90 mm, 1.3–3.0 mm, and 3.0–7.0 mm intervals probe progressively larger grains, and the high-$$\beta_{\nu_1-\nu_2}=-\frac{\log\left(T_{\nu_1}/T_{\nu_2}\right)}{\log\left(\nu_1/\nu_2\right)},$$0 regions move inward as the wavelength interval shifts to longer wavelengths because the larger grains are concentrated closer to the star in the model (Pavlyuchenkov et al., 2019). In event-horizon imaging, the relevant scales are the photon ring, the inner accretion flow, and the outskirts of the image; after blurring, the sharp photon-ring spike weakens but the radial trend remains (Ricarte et al., 2022).

Diffuse-sky work introduces another meaning: the combination of heterogeneous angular resolutions. The Bayesian Haslam recalibration explicitly couples $$\beta_{\nu_1-\nu_2}=-\frac{\log\left(T_{\nu_1}/T_{\nu_2}\right)}{\log\left(\nu_1/\nu_2\right)},$$1 and $$\beta_{\nu_1-\nu_2}=-\frac{\log\left(T_{\nu_1}/T_{\nu_2}\right)}{\log\left(\nu_1/\nu_2\right)},$$2 data, while the neural-network simulation paper generates a 56 arcmin proxy full-sky synchrotron spectral-index map from a 5° template to test the significance of added small-scale structure (Nasirudin et al., 15 Sep 2025, Irfan, 2023).

4. Physical interpretation

The diagnostic content of spectral-index maps depends on the emitting process. In radio AGN, steep negative values, roughly $$\beta_{\nu_1-\nu_2}=-\frac{\log\left(T_{\nu_1}/T_{\nu_2}\right)}{\log\left(\nu_1/\nu_2\right)},$$3, usually indicate optically thin synchrotron emission from extended jets, lobes, aged diffuse nonthermal emission, or sometimes star-formation-related cosmic-ray emission; flatter values around $$\beta_{\nu_1-\nu_2}=-\frac{\log\left(T_{\nu_1}/T_{\nu_2}\right)}{\log\left(\nu_1/\nu_2\right)},$$4 to $$\beta_{\nu_1-\nu_2}=-\frac{\log\left(T_{\nu_1}/T_{\nu_2}\right)}{\log\left(\nu_1/\nu_2\right)},$$5 often indicate a compact core or the base of a jet; and inverted values $$\beta_{\nu_1-\nu_2}=-\frac{\log\left(T_{\nu_1}/T_{\nu_2}\right)}{\log\left(\nu_1/\nu_2\right)},$$6 suggest synchrotron self-absorption, free-free absorption, or a spectral turnover near or above the observed band (Järvelä et al., 2021). A key corrective point follows: a source can have a steep total spectrum while still containing a flat core.

In protoplanetary discs, the central result is that regions where $$\beta_{\nu_1-\nu_2}=-\frac{\log\left(T_{\nu_1}/T_{\nu_2}\right)}{\log\left(\nu_1/\nu_2\right)},$$7 is maximal for a given wavelength interval correspond to regions where grains with $$\beta_{\nu_1-\nu_2}=-\frac{\log\left(T_{\nu_1}/T_{\nu_2}\right)}{\log\left(\nu_1/\nu_2\right)},$$8 are most prevalent. The method works because the opacity slope changes most strongly when the grain-size distribution contains many grains comparable to the observing wavelength. At the same time, optically thick regions drive $$\beta_{\nu_1-\nu_2}=-\frac{\log\left(T_{\nu_1}/T_{\nu_2}\right)}{\log\left(\nu_1/\nu_2\right)},$$9 in the Rayleigh–Jeans limit, and at $\beta=\alpha+2$0 K the minimum $\beta=\alpha+2$1 can drop to about 1.8 because the Rayleigh–Jeans approximation fails (Pavlyuchenkov et al., 2019).

In black-hole accretion-flow models, the spectral index increases with increasing magnetic field strength $\beta=\alpha+2$2, electron temperature $\beta=\alpha+2$3, and optical depth $\beta=\alpha+2$4, and therefore becomes more negative with increasing radius in almost all models. Photon-ring geodesics exhibit more positive spectral indices because they sample the innermost plasma and have longer path lengths. The maps also discriminate magnetic states: Standard and Normal Evolution flows tend to exhibit more negative spectral indices than Magnetically Arrested Disk flows (Ricarte et al., 2022).

Diffuse Galactic work highlights environmental effects. The all-sky 45–408 MHz temperature spectral-index map ranges from about 2.1 to 2.7, with flatter values at low latitude attributed to thermal free-free absorption and a steep maximum adjacent to the Northern Spur (Guzmán et al., 2010). In polarized synchrotron, the semi-blind QUIJOTE analysis finds $\beta=\alpha+2$5 in the Galactic plane and $\beta=\alpha+2$6 at high latitudes, indicating moderate steepening away from the plane (Adak et al., 20 Oct 2025).

5. Representative scientific uses

One established use is component disentangling in radio galaxies and jetted Seyfert systems. In the absorbed-jet NLS1 sample, spectral-index maps separate steep extended structures from flatter central components and thereby distinguish star formation, optically thin jet emission, compact self-absorbed cores, and absorption effects. J1522+3934 is identified as the strongest case for a dominant AGN/jet origin; J1228+5017 is interpreted as a mixed or restarted system; and J1641+3454 appears star-formation-like at low frequency even though a gamma-ray detection implies a jet that is likely strongly absorbed below 9 GHz (Järvelä et al., 2021).

A second use is revealing variability that brightness maps do not show. In PWNe, the maps expose localized hardening and softening in G21.5–0.9, Kes 75, and 3C 58 that are not recovered by simple radial profiles or by total-intensity inspection alone. The principal claim is that flux variability and spectral variability are not the same thing, and that spatially resolved spectral comparisons over time uncover previously hidden evolution in the emitting particle population (Guest et al., 2020).

A third use is building or testing foreground models. The 45–408 MHz all-sky index map produces a calibrated large-scale Galactic diagnostic after explicit correction for the CMB, extragalactic background, and zero-level offsets (Guzmán et al., 2010). The Bayesian Haslam recalibration extends this logic by inferring regional calibration factors and coarse spectral-index maps jointly, recovering flux-scale factors to about $\beta=\alpha+2$7 in most regions in a fiducial simulation and rectifying the sky map to within $\beta=\alpha+2$8 K of the truth in all cases (Nasirudin et al., 15 Sep 2025). The neural-network simulation paper then asks how much small-scale spectral-index structure matters for 21 cm intensity-mapping foreground tests; for a cosine aperture taper beam with FWHM between 1.1° and 1.6° and PCA cleaning, it concludes that no greater resolution than 5° is required (Irfan, 2023).

A fourth use is model discrimination in horizon-scale accretion physics. Even when unresolved, the net spectral index around 230 GHz can distinguish families of GRMHD+GRRT models, although the paper emphasizes that interpretation remains sensitive to uncertain plasma heating prescriptions and the electron distribution function (Ricarte et al., 2022).

6. Limitations, misconceptions, and current directions

A persistent misconception is that integrated spectra, brightness images, or single-frequency morphologies are sufficient. The recent literature consistently argues otherwise. A steep total spectrum can conceal a flat-spectrum core (Järvelä et al., 2021); brightness-stable PWNe can still exhibit localized spectral evolution (Guest et al., 2020); and an apparently good fit at 408 MHz does not guarantee accuracy at 50–150 MHz (Nasirudin et al., 15 Sep 2025).

Methodological sensitivity is equally central. VLBI maps require accurate alignment because self-calibration removes absolute coordinates and core shift introduces real frequency-dependent displacements (Kim et al., 2014). Disc applications require well-calibrated fluxes, sufficient signal-to-noise, and adequate angular resolution, and the sharp $\beta=\alpha+2$9 peak may weaken or disappear for porous aggregates or different grain compositions (Pavlyuchenkov et al., 2019). In full-sky synchrotron inference, the beam, region boundaries, and spectral model couple scales nontrivially, and the posterior exhibits strong anti-correlation between sky brightness and flux-scale factors as well as significant degeneracy between $\Gamma$0 and $\Gamma$1 (Nasirudin et al., 15 Sep 2025).

Current directions are correspondingly explicit. For absorbed-jet NLS1s, the immediate requirement is higher-frequency radio imaging, higher angular resolution, simultaneous multi-frequency observations, and spatially resolved spectroscopy of the host galaxies (Järvelä et al., 2021). For polarized Galactic foregrounds, the moment-based formalism is positioned as extensible to higher-order moments and hence to synchrotron spectral curvature when better data become available (Adak et al., 20 Oct 2025). For event-horizon-scale black holes, improved technology and analysis pipelines are expected to make resolved spectral-index maps increasingly practical, with direct implications for plasma microphysics (Ricarte et al., 2022).

Taken together, these studies define multi-scale spectral index maps less as a single instrument-specific product than as a general inferential framework. Whether the target is a jet, a dust disc, a pulsar wind nebula, the Galactic synchrotron sky, or an accretion flow at the event horizon, the common objective is to recover spatially varying spectral slopes in a form that preserves physically meaningful structure across frequency and scale.