Mapping atmospheric features of the planetary-mass brown dwarf SIMP 0136 with JWST NIRISS

TL;DR

This paper uses JWST NIRISS time-series spectroscopy of the planetary-mass brown dwarf SIMP J01365662+093347 to dissect atmospheric variability across depth. Through PCA, model admixtures, and spherical-harmonic mapping, the authors identify at least three distinct spectral regions and infer a multi-layer atmospheric structure featuring a deep forsterite cloud deck atop an iron cloud layer, with vertical coupling between layers reflected in the lightcurves and spectra. Atmospheric retrievals support a patchy, adiabatic-like T–P profile and reveal North–South brightness asymmetry, while Doppler tomography demonstrates km s$^{-1}$-level RV signatures that could constrain brightness maps with higher-resolution data. The work highlights the potential of combining regionally resolved spectra, mapping techniques, and Doppler information to break degeneracies in brown dwarf weather mapping, and points to future observations across multiple rotations and higher spectral resolution to refine the 3D atmospheric picture.

Abstract

In this paper, we analyze James Webb Space Telescope Near Infrared Imager and Slitless Spectrograph time-series spectroscopy data to characterize the atmosphere of the planetary-mass brown dwarf SIMP J01365662+093347. Principal component analysis reveals that 81\% of spectral variations can be described by two components, implying that variability within a single rotational phase is induced by at least three distinct spectral regions. By comparing our data to a grid of Sonora Diamondback atmospheric models, we confirm that the time-averaged spectrum cannot be explained by a single model but require a linear combination of at least three regions. Projecting these models onto the principal component plane shows that the overall variability is highly correlated with changes in temperature, cloud coverage, and possibly effective metallicity. We also extract brightness maps from the lightcurve and establish North-South asymmetry in the atmosphere. A combined multidimensional analysis of spectro-photometric variability links the three spectral regions to three atmospheric layers. Forsterite cloud and water abundance at each level form unique harmonics of atmospheric variability observed in different spectral bands. Atmospheric retrievals on the time-averaged spectrum are consistent with an optically thick iron cloud deck beneath a patchy forsterite cloud layer and with the overall adiabatic curve. We also demonstrate two new analysis methods: a regionally-resolved spectra retrieval that relies on multi-wavelength spherical harmonics maps, and a method to constrain brightness maps using Doppler information present in the spectra. Future observations of variable brown dwarfs of higher spectral resolution or spanning multiple rotations should help break mapping degeneracy.

Figures (12)

• Figure 1: Left: Time-averaged spectrum of SIMP 0136 over an entire rotation with NIRISS/SOSS. The hashed red region was not considered in our analysis due to contamination of the SOSS trace from field stars. Coloured bands highlight 0.2 $\mu$m spectral bins. Right: lightcurve of each bin with best fit Imber models Plummer2023Plummer2024a. The bins from $2.2-2.8 \mu$m (yellow) are best fit by a single peak per rotation. The intermediate bins from $1.8-2.2 \mu$m (magenta) exhibit both 2 and 3 peaks per rotation. The shortest wavelength bins from $1.0-1.8 \mu$m (purple) are best fit by $3^{\rm rd}$ harmonic model. Fourier fitting (black lines) was performed with Imber.
• Figure 2: Projection of the time varying spectrum onto the principal component plane. Percentage indicates the fraction of explained variance by the components. Low-contrast and high-contrast shaded curves respectively show the time series at the original time sampling and in 9-min bins. Colour coding of both curves gives the rotational phase. The two components are orthogonal, but do not directly correspond to physical parameters. We projected Sonora Diamondback models to see how changing f$_{\rm sed}$, T and effective [Fe/H] impact the spectra. Starting at the bottom right of the plot, we can infer sequential change in T and f$_{\rm sed}$ as SIMP 0136 completes a rotation.
• Figure 3: NIRISS time-average spectrum over an entire rotation (black line) compared to various Diamondback grid models and their combinations. A single Diamondback model cannot accurately reproduce data. The interpolation of parameters allows to achieve higher accuracy with green line showing closer match with data. However, only linear combination of grid models can substantially explain data. The red line shows such combination of 3 spectra of different physical properties, and the blue line is an admixture of 6 models, which is favoured by BIC. We allowed gravitational acceleration to vary because the consensus is that it's between 100 and 316 cm s$^{-2}$. The lower part of the figure shows residuals normalized to data error. The error was scaled up by 7.8 to efficiently utilize BIC.
• Figure 4: Comparison of lightcurve morphology at 3 distinct wavelength bins. Black error bars show NIRISS data, similarly to Figure \ref{['fig:niriss data']}; Green triangles show lightcurves extracted from NIRISS data that was projected on the first two principal components. Red squares are lightcurves of an admixture model consisting of 3 Diamondback spectra. Observed lightcurve morphology can be explained by 2 components, confirming that it is caused by 3 or more spectral regions rotating in and out of view. The consistency between the observed and reconstructed light curves suggests that the bulk properties of the underlying cloud structure are correctly reproduced.
• Figure 5: Retrieved Brewster model thermal profile (brown line, with coloured shading for 1-$\sigma$ confidence interval) for SIMP 0136 . Previous retrieval result from Vos_2023 is plotted as a black curve for comparison. Self-consistent model profiles from the Sonora Diamondback grid are plotted as solid coloured lines. Phase-equilibrium condensation curves for plausible cloud species are plotted as coloured dashed lines. The clouds pressures are indicated in bars to the right. Purple bar indicates the median cloud location for the forsterite slab and the optically thick extent of the iron deck, with grey shading indicating the $1-\sigma$ range. Graduated shading shows the range over which the deck cloud optical depth drops to $\rm 1/e$. Forsterite and enstatite are fairly challenging to distinguish in retrievals of the spectra, meaning high altitude clouds could actually be enstatite, or a mixture of both.