Spectral Energy Distribution (SED) Analysis

• SED Analysis is the quantitative modeling of broadband photometry and spectroscopy used to infer stellar, dust, and galaxy properties from UV to FIR wavelengths.
• It employs energy-balanced methods, such as CIGALE’s delayed SFH and detailed dust templates, to accurately estimate galaxy mass, SFR, and attenuation parameters.
• The framework integrates extensive multi-band data, advanced Bayesian parameter estimation, and upper limit treatments to support robust analyses for local and high-redshift systems.

Spectral Energy Distribution (SED) Analysis is the quantitative modeling and interpretation of the flux of astrophysical objects across the electromagnetic spectrum, enabling physical inference of stellar, dust, and galaxy properties. It provides the fundamental link from observed broadband/narrowband photometry and spectroscopy to galaxy mass, star formation rate (SFR), dust characteristics, and recent star-formation history, using physical models constrained by empirically calibrated energy balance and carefully parameterized star formation histories. SED analysis is also essential for connecting observed diagnostics to theoretical predictions, supporting population synthesis, galaxy evolution context, and accurate measurements across cosmic epochs.

1. Observational Inputs and Data Handling

SED analysis relies on assembling flux measurements covering a wide spectral baseline. In the Local Volume Legacy (LVL) sample, SEDs include up to 26 bands per galaxy:

• Ultraviolet: GALEX FUV (0.154 μm), NUV (0.231 μm)
• Optical: Ground-based U, B, V, R_C, I_C; SDSS u, g, r, i, z
• Near-IR: 2MASS J, H, K_s
• Mid-IR: Spitzer IRAC (3.6–8.0 μm)
• Far-IR: Spitzer MIPS (24, 70, 160 μm), IRAS (12, 25, 60, 100 μm)
• Narrowband: Hα (0.656 μm, [N II] corrected)

Upper limits are encoded according to the Sawicki (2012) method as formulated by Boquien et al. (2019): in non-detection bands, model fluxes exceeding the upper limit are penalized during χ² computation, otherwise excluded.

Wavelength coverage from 0.15 to 160 μm ensures sensitivity to both direct starlight (young and old), nebular emission, PAHs, and reprocessed dust components. The approach is fully compatible with incorporating new bands (e.g., JWST/future facilities) through modular filter libraries.

2. Physical Parameterization and SED Modeling

LVL SED fitting employs the CIGALE code’s “sfhdelayedbq” module, parameterizing star formation history (SFH) with a delayed exponential main episode and an optional late-time burst/quenching:

$$\mathrm{SFR}(t)\propto \begin{cases} t\, e^{-t/\tau_\mathrm{main}} & t\le t_\mathrm{bq} \ r_\mathrm{SFR}\, \mathrm{SFR}(t_\mathrm{bq}) & t>t_\mathrm{bq} \end{cases}$$

Key SFH parameters:

• τ_main: e-folding timescale (grid: 1–10 Gyr)
• age_main: fixed to 13 Gyr
• age_bq: burst/quench duration (10–100 Myr grid)
• r_SFR: amplitude ratio post-burst/quench (0–10)

Dust attenuation follows a modified starburst law (Calzetti et al. 2000 plus Leitherer et al. 2002), allowing for:

• power-law slope δ (–1.5 to 0.2)
• differential E(B–V) factors between nebular and continuum
• no UV bump term

Dust emission templates adopt Draine et al. (2014) models, with PAH fraction q_PAH as a free parameter and distributed radiation field density U_min, a power-law dM_dust/dU∝U^–α, and fraction γ of dust in PDR-like heated regime.

A strict energy balance is enforced: $$\int L_{\rm UV/opt}(\lambda)\,d\lambda = \int L_{\rm IR}(\lambda)\,d\lambda$$ —this ensures every erg absorbed at short wavelengths is reradiated in the FIR, allowing direct inference of dust mass and PAH content.

3. Model Grid, Computational Strategy, and Fit Metrics

SED modeling uses a grid approach: ~3.4×10⁹ models are generated spanning all parameter combinations (see Table below). Each galaxy’s observed flux vector is compared to every model by χ² minimization. Both best-fit and Bayesian (likelihood-weighted) parameter estimates are recorded to mitigate discretization artifacts due to finite grid steps.

Parameter	Grid values
Stellar Z	0.004, 0.008, 0.02, 0.05
τ_main (Gyr)	1, 2, 3, 4, 5, 7.5, 10
age_bq (Myr)	10, 25, 50, 75, 100
r_SFR	0–10 (14 steps)
E(B–V)_lines (mag)	0.01–1.0 (18 steps)
E(B–V)_factor	0.25, 0.50, 0.75
δ	–1.5 to 0.2 (Δ0.1)
q_PAH (%)	0.47 ... 6.63
U_min	0.1–50 (9 steps)
α	1.5, 2.0, 2.5
γ	10^–3 to 10^–0.5 (Δ0.25 dex)

Quality of fit is scored by the reduced χ²:

• 94% of LVL fits yield χ²_red < 3 (median = 0.81 ± 0.41)
• 98% have χ²_red < 4

Bayesian posteriors are preferred for estimating physical parameters to reduce grid-edge artifacts and sample multimodal likelihoods.

4. Physical Quantity Extraction and Validation

Key properties are inferred directly from the best-fitting or Bayesian-sampled SEDs:

• Stellar Mass $M_*$: integration of the SFH weighted by the Bruzual & Charlot (2003) SSPs and Chabrier IMF.
• SFR_100: SFH-averaged star formation rate over the last 100 Myr.
• Dust luminosity $L_{\rm dust}$: direct integral over the DL2014 templates.

Validation is performed via empirical calibrations:

• $M_*$ derived from SEDs agrees with 3.6 μm mass-to-light based estimates (scatter: 0.07 dex; median ratio: 1.00).
• SFR_100 is ~88% of hybrid indicator (FUV+TIR or Hα+24 μm) values (scatter: 0.13 dex).

No systematic mass or SFR trend is seen in the cross-validations: SED-derived quantities robustly track standard observational proxies for normal and low-mass dwarfs.

5. Population Trends, SFH Diagnostics, and Implications

Analysis reveals strong systematics as functions of galaxy mass and SFH:

• Bursty SFH: For r_SFR > 1, the mass fraction in the late burst is inversely correlated with total M_*. Low-mass dwarfs exhibit high burstiness, consistent with resolved-star-population work (cf. Weisz et al. 2012).
• 75% of the LVL sample are low-metallicity (M_* < 10⁹ M_⊙): analysis demonstrates the necessity of burst/quench-capable SFH and full UV–FIR coverage to avoid biasing mass and SFR.
• Dust emission diagnostics (e.g., PAH fraction q_PAH) and radiation-field trends (U_min) are reliably extracted through energy balance, revealing that q_PAH decreases as U_min increases—evidence for the impact of diffuse ISM radiation on grain survival.

This parameterization and fitting protocol sets the stage for robust interpretation of metal-poor galaxies, providing templates and methodology transferable to JWST-era deep field and high-z dwarf studies.

6. Methodological Significance and Forward Prospects

The LVL SED analysis demonstrates the integration of physically motivated, energy-balanced SED modeling with full panchromatic photometry and modular, grid-based parameter estimation. Reliable (χ²_red < 3) fits for 94% of a heterogeneous local sample affirm the viability of the CIGALE “delayed+bq” SFH model coupled with detailed dust emission templates.

Key methodological features:

• Bayesian posterior sampling to mitigate grid artifacts and characterize uncertainties.
• Explicit treatment of photometric upper limits and bandpass integration.
• Cross-validation against hybrid SFR proxies and mass-to-light recipes.

The SED analysis framework is a template for future population synthesis and high-redshift galaxy studies, especially in contexts requiring unbiased masses/SFRs for galaxies with non-trivial SFH evolution and significant dust attenuation. The paradigm is extensible to JWST rest-frame mid-IR/near-IR coverage and adaptable for low-metallicity, burst-prone systems at z > 1 (Dale et al., 2023).

---

References:

Spectral Energy Distributions for 258 Local Volume Galaxies (Dale et al., 2023)