Sonora Bobcat Substellar Models

• Sonora Bobcat is a suite of self-consistent atmospheric and evolutionary models that convert observed brown dwarf luminosities and spectra into physical parameters.
• It predicts key quantities such as bolometric luminosity, effective temperature, radius, and gravity across varied metallicities and ages.
• The framework is used for benchmarking empirical systems, interpreting spectral variability, and generating synthetic data for machine-learning classification.

Sonora Bobcat, often written Sonora–Bobcat, is a set of substellar atmosphere and evolutionary models used to connect observables of brown dwarfs and giant planets to physical quantities such as bolometric luminosity, effective temperature, radius, surface gravity, mass, and age. In recent work it functions as an inferential framework rather than merely a spectral library: authors use it to estimate masses and radii in benchmark binaries, derive radius and $T_{\rm eff}$ distributions for JWST late-T and Y dwarfs, interpret pressure-dependent variability, construct synthetic Solar-Neighborhood populations, and generate training data for machine-learning classification of ultracool dwarfs (French et al., 2023, Beiler et al., 2024).

1. Definition within the Sonora model family

Several recent studies describe Sonora Bobcat as part of the broader Sonora project of self-consistent atmospheric and evolutionary models with updated opacities and chemistry. One JWST luminosity study characterizes the adopted tracks as “the Sonora Bobcat solar metallicity evolutionary model, part of new generation of self-consistent atmospheric and evolutionary models that include updated opacities and atmospheric chemistry,” while a retrieval study treats the relevant implementation as the “cloudless Sonora-Bobcat evolutionary models” used in the $(T_{\rm eff}, \log g)$ plane (Beiler et al., 2024, Kothari et al., 6 Apr 2026).

Across the literature in the data block, Bobcat appears in two closely related roles. First, it is an evolutionary grid that provides $R$, $L_{\rm bol}$, $T_{\rm eff}$, and $\log g$ as functions of mass, age, and composition. Second, it is an atmospheric grid supplying pressure–temperature structures, emergent spectra, and broad-band photometry. The same family is therefore used both to infer bulk structure from luminosity and age constraints and to interpret wavelength-dependent atmospheric behavior such as contribution functions, molecular bands, and cloud-sensitive variability (Frost et al., 2024, McCarthy et al., 2024).

The practical scope of the grid is broad. In one intercomparison study, Sonora Bobcat is listed with wavelength coverage from $0.4$ to $50\,\mu\mathrm{m}$, $T_{\rm eff}=200$–2400 K, $\log g = 3.25$–5.5, and metallicities $(T_{\rm eff}, \log g)$0; a later photometric-classification study similarly treats it as having three metallicity branches at $(T_{\rm eff}, \log g)$1, $(T_{\rm eff}, \log g)$2, and $(T_{\rm eff}, \log g)$3, over $(T_{\rm eff}, \log g)$4 and $(T_{\rm eff}, \log g)$5 (Lueber et al., 2023, Biswas, 1 Jul 2025).

2. Physical assumptions and parameterization

The papers in the data block converge on a common description of the adopted Bobcat grids as cloudless or cloud-free models in radiative–convective equilibrium with rainout chemical equilibrium. One model-intercomparison paper describes Sonora Bobcat as “cloud-free models computed for radiative–convective equilibrium atmospheres utilizing a layer-by-layer convective adjustment approach, allowing for solutions with detached convective zones,” with condensation from the gas phase included for rainout chemistry but “no explicit cloud opacity” (Lueber et al., 2023). A retrieval study of cool JWST dwarfs likewise identifies the comparison grid as “cloudless, solar-metallicity brown dwarf evolutionary tracks” under rainout chemical equilibrium, while the LHS 6343 C analysis uses Sonora-Bobcat as cloudless atmosphere and evolution models with rainout chemical equilibrium and discrete metallicities $(T_{\rm eff}, \log g)$6 (Kothari et al., 6 Apr 2026, Frost et al., 2024).

The parameterization used in applications depends on the problem. For atmospheric fitting, Sonora-Bobcat spectra are indexed by $(T_{\rm eff}, \log g)$7, $(T_{\rm eff}, \log g)$8, and metallicity, with fixed solar C/O in the LHS 6343 C study because the grid is too sparse to vary C/O freely (Frost et al., 2024). For variability work on SIMP J013656.5+093347, a specific Bobcat atmosphere is selected at $(T_{\rm eff}, \log g)$9 K, $R$0, $R$1, and solar C/O $R$2, and is combined with Sonora correlated-$R$3 coefficients to compute wavelength- and pressure-dependent flux contribution functions (McCarthy et al., 2024).

A recurrent practical distinction is that many papers use solar-metallicity evolutionary tracks even when the science target may not be exactly solar. The SDSS J222551.65+001637.7AB study explicitly compares the L4 companion to Sonora–Bobcat “assuming solar metallicity,” justifying that as a reasonable baseline because the system is likely thick-disc but not an extremely metal-poor halo object and because there is no evidence of metal pollution in the white dwarf atmosphere (French et al., 2023). Other studies instead exploit the multi-metallicity branches directly, especially in synthetic-photometry and machine-learning contexts (Biswas, 1 Jul 2025).

3. Inference workflows and governing relations

In the applications summarized here, Sonora Bobcat is typically queried or interpolated after an observational pipeline has reduced spectra, photometry, parallaxes, or dynamical constraints to a small set of physical observables. The governing relations are the standard gravity law,

$R$4

and the Stefan–Boltzmann law,

$R$5

which appear explicitly in multiple studies and are treated as embodied in the evolutionary tracks (French et al., 2023, Frost et al., 2024).

One common workflow is age-anchored inference. In the resolved white-dwarf–brown-dwarf binary SDSS J222551.65+001637.7AB, the white dwarf provides a total system age of $R$6 Gyr through wdwarfdate, Montreal White Dwarf Group cooling models, a Cummings et al. (2018) initial–final mass relation, and MIST isochrones. The brown-dwarf companion is then characterized by comparing $R$7 K and $R$8, together with $R$9-band constraints, to solar-metallicity Sonora–Bobcat tracks (French et al., 2023).

A second workflow is luminosity-to-radius-to-temperature inference. In the JWST study of 23 late-T and Y dwarfs, the authors first construct spectral energy distributions from NIRSpec PRISM, MIRI LRS, and MIRI photometry, integrate to obtain $L_{\rm bol}$0, convert to $L_{\rm bol}$1 using parallaxes, and then draw $L_{\rm bol}$2 age–luminosity pairs. For each pair they linearly interpolate within the Bobcat luminosity–age–radius grid to obtain a radius distribution, which is then converted to $L_{\rm bol}$3 through the Stefan–Boltzmann law (Beiler et al., 2024).

A third workflow is post-retrieval evolutionary placement. In the 22-object JWST atmospheric-retrieval analysis, masses and radii are retrieved directly from spectra; $L_{\rm bol}$4 is then derived through $L_{\rm bol}$5, and $L_{\rm bol}$6 through luminosity and radius. Those $L_{\rm bol}$7 points are overlaid on Sonora Bobcat cooling tracks and isochrones to infer an age range of 0.4 to 10 Gyr across the sample (Kothari et al., 6 Apr 2026).

A fourth workflow is benchmark interpolation in mass–luminosity space. For LHS 6343 C, whose mass and radius are empirically constrained by RV and eclipse geometry, Sonora-Bobcat evolutionary models are sampled in Monte Carlo fashion using the measured mass and $L_{\rm bol}$8. The interpolation returns an age, predicted radius, $L_{\rm bol}$9, and $T_{\rm eff}$0, allowing a direct comparison between the evolutionary grid and a benchmark brown dwarf with negligible irradiation (Frost et al., 2024).

4. Benchmark systems and empirical calibration

Benchmark systems are central to how Sonora Bobcat is evaluated. Wide white-dwarf–brown-dwarf binaries and eclipsing brown dwarfs reduce the usual age–mass–luminosity degeneracy because age, mass, or radius can be constrained independently and then propagated through the model grid (French et al., 2023, Frost et al., 2024).

System	Sonora Bobcat use	Representative result
SDSS J222551.65+001637.7B	Mass and radius from $T_{\rm eff}$1, $T_{\rm eff}$2, $T_{\rm eff}$3, and white-dwarf age	$T_{\rm eff}$4–$T_{\rm eff}$5; $T_{\rm eff}$6–$T_{\rm eff}$7
LHS 6343 C	Age and predicted structure from measured mass and luminosity	$T_{\rm eff}$8 Gyr; $T_{\rm eff}$9; $\log g$0 K
JWST late-T/Y sample	Approximate ages from retrieved $\log g$1 placement	$\log g$2 to $\log g$3 Gyr across the sample

In SDSS J222551.65+001637.7AB, Sonora–Bobcat yields a companion mass of $\log g$4–$\log g$5 and a radius of $\log g$6–$\log g$7 at $\log g$8. The same study states that “the most appropriate model for our bolometric luminosity and effective temperature provides an age estimate and a $\log g$9-band magnitude that are consistent with our results,” so the grid reproduces the observed $0.4$0 brightness, empirically inferred temperature, luminosity, and white-dwarf-derived age simultaneously (French et al., 2023).

LHS 6343 C provides a more stringent benchmark because its mass and radius are not inferred from Sonora Bobcat. The empirical values are $0.4$1, $0.4$2, $0.4$3, and $0.4$4 K from the measured radius and luminosity. Interpolation in the Sonora-Bobcat evolutionary grid gives an age of $0.4$5 Gyr, $0.4$6, $0.4$7 K, and $0.4$8, all within $0.4$9 of the directly measured quantities (Frost et al., 2024).

These benchmark applications establish two distinct empirical roles. First, they show that Sonora Bobcat can act as a consistency check when age is supplied externally, as in white-dwarf binaries. Second, they show that the evolutionary tracks can be directly tested against measured substellar mass–radius–luminosity combinations, as in LHS 6343 C (French et al., 2023, Frost et al., 2024).

5. Spectral, atmospheric, and variability applications

The most extensive modern use of Sonora Bobcat in the data block is the conversion of high-fidelity JWST luminosities into radii and temperatures for very cool brown dwarfs. For 23 late-T and Y dwarfs, Bobcat-based radius distributions yield $50\,\mu\mathrm{m}$0 estimates spanning $50\,\mu\mathrm{m}$1–350 K, with a median uncertainty of $50\,\mu\mathrm{m}$2 K. The study emphasizes that the radius inference, rather than photometry or spectroscopy, has become the limiting factor in $50\,\mu\mathrm{m}$3 precision, and that current evolutionary models and age information cap the achievable precision at $50\,\mu\mathrm{m}$4 even if $50\,\mu\mathrm{m}$5 were known exactly (Beiler et al., 2024).

Those temperatures provide a physical ordering of spectra that does not perfectly follow near-infrared spectral type. In the JWST sample, the region below $50\,\mu\mathrm{m}$6 and the region above $50\,\mu\mathrm{m}$7 show relatively clear sequences with decreasing $50\,\mu\mathrm{m}$8, whereas the $50\,\mu\mathrm{m}$9 region shows no clear monotonic sequence. The same study uses Bobcat chiefly to provide radii and the physical axes $T_{\rm eff}=200$0 and $T_{\rm eff}=200$1, while the spectral diversity at $T_{\rm eff}=200$2 is interpreted in terms of metallicity, C/O, disequilibrium chemistry, and gravity (Beiler et al., 2024).

Sonora Bobcat is also used as a depth-mapping tool. In SIMP J013656.5+093347, a cloud-free Bobcat atmosphere and correlated-$T_{\rm eff}=200$3 opacities are used to calculate where the central $T_{\rm eff}=200$4 of the received flux in each band originates: 12.6–26.2 bar for J, 3.6–20.0 bar for H, and 1.9–14.1 bar for $T_{\rm eff}=200$5. Coupled to the observed $T_{\rm eff}=200$6 phase shift between J and $T_{\rm eff}=200$7, these contribution functions support the conclusion that at least two different patchy cloud layers must be present (McCarthy et al., 2024).

In the 22-object retrieval analysis, Sonora Bobcat is not the retrieval engine itself but the evolutionary comparison space. The retrieved thermal profiles are compared to Sonora Elf-Owl forward-model thermal profiles, and the paper reports a systematic difference between the two “likely arising due to the difference in chemistry treatment.” Because Bobcat and Elf-Owl share the Sonora evolutionary framework, this comparison bears directly on how securely Bobcat-based $T_{\rm eff}=200$8 interpretations can be extended into the coldest late-T and Y regimes (Kothari et al., 6 Apr 2026).

6. Population synthesis and machine-learning applications

In population studies, Sonora Bobcat is used as a pure-cooling baseline. The Solar-Neighborhood simulation of brown dwarfs above and below the Galactic Plane constructs a synthetic population from Gaia-derived star-formation histories, an IMF,

$T_{\rm eff}=200$9

and an age–metallicity relation,

$\log g = 3.25$0

and then interpolates Sonora Bobcat in $\log g = 3.25$1, $\log g = 3.25$2, and $\log g = 3.25$3 to obtain $\log g = 3.25$4, $\log g = 3.25$5, $\log g = 3.25$6, and $\log g = 3.25$7 (Honaker et al., 2024).

Within that framework, Bobcat differs systematically from hybrid cloudy–clear model sets. Because it is cloudless, it does not produce the L–T transition pileups seen in SM08 and Sonora Diamondback; its luminosity and temperature functions are smooth. The study uses this contrast to argue that global brown-dwarf population statistics depend jointly on evolutionary model choice, the IMF, the star-formation history, and kinematic heating, and that sub-populations farther from the Galactic Plane are older and occupy a different region of parameter space than younger sub-populations closer to the Plane (Honaker et al., 2024).

A distinct use is synthetic training-data generation for classification. The ultracool TY-dwarf selection study combines ATMO 2020 and Sonora Bobcat photometry, filters the models to $\log g = 3.25$8 K, and linearly interpolates Sonora Bobcat in $\log g = 3.25$9, $(T_{\rm eff}, \log g)$00, mass, and $(T_{\rm eff}, \log g)$01 to create 8000 additional synthetic models with YJHKW1W2W3 magnitudes. Spectral types are then assigned through empirical color–spectral-type polynomials using a reduced $(T_{\rm eff}, \log g)$02 statistic, and the resulting synthetic set trains an ensemble of classifiers (Biswas, 1 Jul 2025).

That machine-learning framework reports object-classification metrics $(T_{\rm eff}, \log g)$03 and an average spectral-type precision within $(T_{\rm eff}, \log g)$04 subtypes on empirical validation, and its application to a $(T_{\rm eff}, \log g)$05 region around Pisces and the UKIDSS UDS field produces one previously uncatalogued T8.2 candidate. In this setting, Sonora Bobcat is valuable less as a direct physical inference tool than as a way to populate late-T and Y color space beyond the sparse empirical sample (Biswas, 1 Jul 2025).

7. Uncertainties, model dependence, and open issues

The data block presents Sonora Bobcat as useful but not definitive. The white-dwarf–brown-dwarf benchmark paper notes that brown-dwarf parameter estimation remains model-dependent and that properties are “degenerate” and “sensitive to atmospheric processes that remain poorly understood,” especially non-equilibrium chemistry and atmospheric clouds. It further identifies age systematics, metallicity assumptions, and atmospheric-model dependence as caveats that can shift inferred masses and radii even when quoted statistical errors are modest (French et al., 2023).

The strongest model-dependence in the supplied studies concerns surface gravity. In the machine-learning intercomparison of 14 model grids, effective temperature is found to be robustly predicted independent of model grid, whereas $(T_{\rm eff}, \log g)$06 is model-dependent; BT-Settl, Sonora Bobcat, and Sonora Cholla tend to predict $(T_{\rm eff}, \log g)$07–4 even after wavelengths blueward of $(T_{\rm eff}, \log g)$08 are excluded. The same paper emphasizes two longstanding problems: clouds cannot yet be modeled from first principles, and cloud models cannot yet be robustly validated (Lueber et al., 2023).

A second recurrent limitation is that Sonora-Bobcat atmospheric fits need not be the best atmospheric fits even when the evolutionary tracks remain successful. For LHS 6343 C, ATMO-2020 models with strong non-equilibrium chemistry provide the best fit to the combined Kepler, HST/WFC3, and Spitzer data, whereas Sonora-Bobcat equilibrium atmospheres perform similarly to other equilibrium grids and significantly worse than ATMO non-equilibrium models. Yet the Sonora-Bobcat evolutionary tracks still return an age and structural parameters consistent with the empirical benchmark (Frost et al., 2024).

The 22-object JWST retrieval study introduces a further caution for the coolest dwarfs. It reports retrieved masses and radii spanning approximately $(T_{\rm eff}, \log g)$09–$(T_{\rm eff}, \log g)$10 and $(T_{\rm eff}, \log g)$11–$(T_{\rm eff}, \log g)$12, retrieved $(T_{\rm eff}, \log g)$13 values of about 4–5.5 and $(T_{\rm eff}, \log g)$14 values of about 350–1100 K, and an inferred age range of 0.4 to 10 Gyr when those parameters are compared to Sonora Bobcat. However, the same paper notes that many objects have $(T_{\rm eff}, \log g)$15, that masses at $(T_{\rm eff}, \log g)$16 K are systematically higher than expected from evolutionary models, and that the cloudless, solar-metallicity Bobcat grid may not fully capture the effects of metallicity, clouds, and disequilibrium chemistry in very cool late-T and Y atmospheres (Kothari et al., 6 Apr 2026).

A common misconception would be to treat Sonora Bobcat as a single, context-free answer to brown-dwarf characterization. The papers summarized here instead support a narrower formulation: Sonora Bobcat is a modern baseline for cloud-free, rainout-equilibrium substellar evolution and atmospheres, highly effective for converting luminosities, ages, and benchmark constraints into physically interpretable parameters, but still subject to systematic uncertainty in gravity, chemistry, metallicity, and cloud treatment (Beiler et al., 2024, Lueber et al., 2023).