DESI Legacy Imaging Surveys Overview

• DESI Legacy Imaging Surveys is a deep, multi-band optical resource that quantitatively estimates galaxy stellar masses using a robust two-parameter model calibrated against S⁴G measurements.
• The methodology employs GALFIT-based photometry and regression on g- and r-band magnitudes, achieving a scatter of approximately 25% compared to high-fidelity Spitzer data.
• An automated Python pipeline, photomass_ls.py, streamlines data processing and mass estimation, enabling large-scale, reproducible extragalactic analyses and efficient target selection.

The DESI Legacy Imaging Surveys (DESI-LS) provide a quantitative framework for estimating stellar masses of galaxies based on deep, wide-field optical imaging in the $g$, $r$, and $z$ bands. Leveraging calibration against robust mid-infrared stellar mass measurements from the S$^4$G survey, a simple, linear prescription has been derived for rapid and automated stellar mass estimates from DESI-LS data. This methodology supports multiple scientific applications—from statistical galaxy evolution studies to construction of large, homogeneous galaxy catalogs—while emphasizing scalability and reproducibility suitable for modern extragalactic research.

1. Formula Derivation and Calibration

The core of the method is a two-parameter model that links the DESI-LS $g$ and $r$ absolute magnitudes to stellar mass. Integrated optical magnitudes are obtained by fitting DESI-LS images with GALFIT, producing extinction-corrected, distance-calibrated photometry. Comparison against the high-fidelity S$^4$G stellar masses (derived from Spitzer 3.6–4.5 μm imaging) yields the regression:

$$\log(M_*/\mathrm{M}_\odot) = 0.673\,M_g - 1.108\,M_r + 0.996$$

where $M_g$ and $M_r$ are the absolute magnitudes in the respective DESI-LS bands. This statistically optimal form achieves a scatter of approximately 25% (in logarithmic units) against the S$^4$G reference. Systematic offsets due to alternative S$^4$G mass calibrations (e.g., mass-to-light ratios, assumed IMF) are typically of order 0.07–0.09 dex, underscoring intrinsic uncertainties in any photometrically derived stellar mass.

2. Role and Contribution of Photometric Bands

DESI-LS provides imaging in at least three optical bands: $g$, $r$, and $z$. Comprehensive testing demonstrates that $g$ and $r$ alone capture the bulk of the diagnostic power for stellar mass estimation. The $r$ band is the single most informative photometric channel. Adding $z$ band, although redder and closer to S$^4$G wavelengths, confers negligible improvement in the scatter; the $z$ band’s lower SNR and subtle systematics likely limit its utility in this context. Neither higher-order polynomial functions nor incorporates extrinsic structural parameters (e.g., Sèrsic index, ellipticity) materially reduce the uncertainty below the $\sim$25% floor set by $g$ and $r$ magnitudes alone.

3. Tools, Automation, and Data Pipeline

The mass estimation workflow is encapsulated in the Python package photomass_ls.py, engineered for reproducibility and high throughput. Key components include:

• Automated data retrieval: FITS images in $g$, $r$, (optionally $z$) bands are programmatically downloaded from the Legacy Surveys DR10 archive.
• Foreground and contaminant masking: Using SEP (SExtractor in Python), the script auto-generates object masks to exclude stars and neighbors from the GALFIT modeling.
• Input parameter estimation: The pipeline computes initial guesses for crucial fit parameters—effective radius, axis ratio, position angle—based on the segmentation output, embedding these into the GALFIT configuration.
• Photometric modeling and extinction correction: GALFIT extracts integrated model magnitudes, which are then corrected for Galactic foreground extinction (e.g., via NED).
• Absolute magnitude and mass calculation: With supplied redshifts or distances, the code converts apparent to absolute magnitudes, applies the calibrated mass formula, and outputs $\log(M_*)$ estimates.

Optional features include the application of K-corrections for higher-redshift systems and selection among alternative S$^4$G mass calibrations.

4. Alternative Calibrations and Generality

In addition to the default formula based on direct GALFIT measurements, an alternative calibration using photometry from the Siena Galaxy Atlas 2020 (SGA-2020) is provided. The corresponding scatter is $\sim$29%, reflecting differences in aperture definition, photometric depth, and data quality. Analyses demonstrate that even with different extraction methodologies or input data, the general $g$/$r$-based approach remains effective and robust, although caution should be exercised when comparing mass catalogs constructed from disparate pipelines. Furthermore, coefficients are available for several S$^4$G-based calibrations (e.g., after Meidt et al. 2014, Querejeta et al. 2015), allowing users to match assumptions as needed for specific scientific contexts.

5. Applications and Implications

The DESI-LS stellar mass estimator serves as a practical backbone for large-scale studies of galaxy populations:

• Statistical extragalactic studies: The ability to derive homogeneous stellar masses for millions of galaxies enables refined measurements of the stellar mass function, mass–metallicity relation, and stellar-to-halo mass scaling across cosmic environments.
• Comparison with dynamical and lensing masses: Automated mass catalogs facilitate cross-matching with dynamical and lensing studies, permitting critical comparisons of stellar and total mass measurement techniques.
• Target selection and ancillary science: Rapid mass estimation benefits legacy survey target selection for follow-up campaigns, including deep spectroscopy or high-resolution imaging.
• Systematic uncertainties: The accuracy of individual mass estimates is limited by the S$^4$G calibration and the universality of the underlying stellar population assumptions. The method performs optimally for isolated galaxies with high-quality DESI-LS imaging and limited morphological complexity. Complex objects—such as those with tidal features or significant non-stellar contributions—may require domain-specific caution or future methodological refinements.

6. Limitations and Prospects for Improvement

Limiting factors include: (1) a systematic calibration floor ($\gtrsim$0.1 dex) from both S$^4$G reference uncertainties and propagation of errors in distance and extinction; (2) relatively diminished utility for galaxies with atypical morphologies or strong nebular emission; and (3) the lack of improvement when additional photometric bands or structural parameters are introduced. Ongoing and future wide-area surveys may explore the integration of advanced machine learning techniques, multi-wavelength SED fitting, or more sophisticated modeling of stellar population varaibility to further reduce the intrinsic scatter and tailor mass estimates for diverse populations.

Conclusion

The calibrated mass estimator for DESI Legacy Imaging Surveys, built upon rigorous regression against S$^4$G stellar masses and automated with an open-source Python workflow, provides a practical bridge between large-area optical imaging and robust extragalactic stellar mass inference. Its demonstrated precision, simplicity, and scalability ensure wide applicability within both standalone and cross-survey galaxy evolution analyses (Ebrová et al., 2 Oct 2025).