Large-amplitude Variability Driven by Giant Dust Storms on a Planetary-mass Companion

TL;DR

This work addresses extreme variability in VHS 1256B by developing a self-consistent 3D cloudy atmosphere model in which giant dust storms organize into tropical waves. Cloud radiative feedback drives a Matsuno-Gill–type circulation that sustains planetary-scale storms, reproducing both the large amplitude (up to ~40%) near-infrared variability and the red JWST spectrum with a pronounced 10 μm silicate feature. The results propose a universal weather mechanism for dusty substellar atmospheres, linking spectral reddening, color-magnitude trends, and L/T-like transitions to cloud dynamics and planetary rotation, with broad implications for interpreting directly imaged planets and brown dwarfs.

Abstract

Large-amplitude variations are commonly observed in the atmospheres of directly imaged exoplanets and brown dwarfs. VHS 1256B, the most variable known planet-mass object, exhibits a near-infrared flux change of nearly 40%, with red color and silicate features revealed in recent JWST spectra, challenging current theories. Using a general circulation model, we demonstrate that VHS 1256B's atmosphere is dominated by planetary-scale dust storms persisting for tens of days, with large patchy clouds propagating with equatorial waves. This weather pattern, distinct from the banded structures seen on solar system giants, simultaneously explains the observed spectra and critical features in the rotational light curves, including the large amplitude, irregular evolution, and wavelength dependence, as well as the variability trends observed in near-infrared color-magnitude diagrams of dusty substellar atmospheres.

Figures (18)

• Figure 1: Bolometric thermal flux maps, their time evolution near the equator, and the resulting long-term light-curve evolution from the GCM.Panel A, B, C: Top-of-atmosphere (TOA) bolometric flux maps from the GCM at different epochs, showing global-scale dust storms (the dark regions) at the equator. Panel D: the Hovmoller diagram (longitude-time sequence) of the TOA bolometric flux near the equator. The flux is averaged over latitudes within $\pm 10$ degrees to cover a fraction of the meridionally extended waves. The dark features evolving towards the upper right mean the eastward propagation of the dust storms. Panel E: the long-term white light curve of the GCM (black line). The red line is the light curve with periodic signals less than 4 rotation periods filtered out. The much smaller amplitude of the filtered light curve suggests that the variability (black line) is mostly rotational rather than from an intrinsic variation of the mean atmospheric state. Epochs 1, 2, and 3 enclosed in the panel correspond to a few different characteristic states of the equatorial waves. Panel F: a zoom-in of the light curve between elapsed time of 4400 to 6500 hours.
• Figure 2: A sketch of the wave generation mechanism and wave diagnoses based on GCM outputs.Panel A: schematic of the horizontal structure of idealized tropical waves relevant to our GCM results. Red shading shows the cloud formation and radiatively heated region at the equator; initially westward-propagating Rossby waves (with wiggle arrows representing propagation direction) are off-equator, and initially eastward-propagating Kelvin waves are east of the heating region. Wave velocities are shown as black arrows. The sketch is similar to that shown in Vallis 2021 vallis2021. Panel B: diagnosis of cloud structure (color), geopotential anomalies (solid lines representing positive values and dotted lines representing negative values), and horizontal velocity fields (red arrows). Panel C: top-of-atmosphere thermal flux corresponding to the cloud structure shown in panel B.
• Figure 3: Model spectroscopic light curves, spectrum, and comparisons to the JWST data.Panels A, B, and C: synthetic light curves from the GCM in three epochs of outputs corresponding to three characteristic evolution stages of the equatorial dust storm. Blue lines are light curves running through the HST/WFC3 G141 filter and the red lines are those running through the Spitzer 4.5 $\mu$m filter. Panel D: model spectrum of VHS 1256B based on the GCM outputs at the time when the light curve reaches the minimum at epoch 1, with a spectral resolution of $R\sim 3000$ (black line), and the observed spectrum of VHS 1256B obtained by JWST (colored lines) miles2023. In the radiative transfer post-processing, we have slightly altered the gas compositions of H$_2$O, CO and CH$_4$ taking into account the possible chemical disequilibrium caused by dynamical mixing miles2023. The JWST spectrum is scaled for a direct comparison with the model spectrum, and different colors are those measured by different instruments, and the grey lines represent error bars. Important molecular gas and silicate cloud absorbers are highlighted. Panels E, F, G, and H: zoom-in of the spectra in a few interesting wavelength regions: panel E contains H$_2$O and alkali metal absorption; panel F shows primarily the CH$_4$ absorption; panel G highlights the CO absorption; and panel H exhibits the silicate cloud absorption.
• Figure 4: Modeled variability in the near infrared $J$ vs $J-H$ CMD using 2MASS filters and comparisons to observed data. The modeled variation trend represented by red points, based on time-resolved spectra at epoch 1 (Fig. \ref{['fig.summary']}), agrees well with that observed for VHS 1256B (blue points) using HST data of two epochs bowler2020zhou2020zhou2022. Among other dusty brown dwarfs' variability observed with HST (magenta points), their variability trends in the CMD also qualitatively agree with our model outputs, indicating the same mechanism operates on a wider population of brown dwarfs and directly imaged giant planets. The magenta points include HST variability observations compiled in lew2020, in which dots are observed time-resolved data and the lines represent linear fits to the data (see details in lew2020). Note that in all colored points, the $J$ and $H$ band wavelength ranges are limited to 1.18 to 1.33 $\mu$m and 1.50 to 1.65 $\mu$m, respectively, as the original data observed by HST/WFC3 do not cover the full wavelength range of the 2MASS filters lew2020. The grey dots are data obtained from the UltracoolSheet best_2024_10573247dupuy2012dupuy2013liu2016bdbest2018best2021sanghi2023schneider2023 whose wavelength ranges cover the full 2MASS filters.
• Figure S1: Previously observed HST/WFC3 and Spitzer 4.5 $\mu$m light curves of VHS 1256B presented in bowler2020zhou2020zhou2022. The light curve of HST/WFC3 is normalized by the median value of the data obtained in both epochs. Black dots in the HST/WFC3 panels A and B are fluxes integrated through the bandwidth, red dots are fluxes at 1.30 $\mu$m (peak flux at the water opacity window), and blue dots are fluxes at 1.38 $\mu$m (low flux at the water absorption band). The light curve of Spitzer 4.5 $\mu$m is normalized by its median value. Grey lines are error bars of the band-integrated fluxes.