Searcharxiv⌕ Search

arXiv · 2610.04081

AEGIS: Differentiable Mars Climate Model with Neural Closures

Abstract

General circulation models (GCMs) are the primary tool for simulating planetary atmospheres. They play a vital role in understanding Mars's atmosphere, as forecasting its unique weather is mission-critical for operations such as entry, descent, and landing. Mars poses unusual challenges for these models, as observations are sparse compared to Earth. In addition, a thin $CO_2$ atmosphere alongside a radiatively active dust cycle creates a volatile atmosphere with large diurnal temperature swings and no true terrestrial analog for validation. Existing Mars GCMs, including the LMD PCM, the NASA Ames Mars GCM, and PlanetWRF, are mature and physically detailed but are implemented in legacy Fortran with finite-difference or finite-volume solvers, and they do not expose gradients for calibration or machine learning. Here we present AEGIS, a modular differentiable Mars climate model that couples Mars's unique atmospheric physics to the Dinosaur dynamical core, with interfaces for neural closures. We showcase stable ten-Mars-year simulations that reproduce the seasonal $CO_2$ cycle while conserving the total $CO_2$ inventory, capture realistic large-scale surface-temperature structure, and produce surface pressure that follows Mars Orbiter Laser Altimeter (MOLA) topography. Gradients through coupled trajectories agree with finite differences and support physical calibration and neural training. We compare with conventional GCMs, highlighting the framework's computational efficiency and differentiability.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Sameera S Kashyap, Victor Cruz, Angel Yepez, Razvan Marinescu. 2026-11-17. AEGIS: Differentiable Mars Climate Model with Neural Closures. https://searcharxiv.com/abs/2610.04081

Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite Searcharxiv by Indiciae: BibTeX · RIS.

KEEP EXPLORING

Related papers

Testing the prevalence of hydrogen-silicate miscibility in young sub-Neptunes

Hydrogen-silicate miscibility can significantly alter the interior structure and thermal evolution of sub-Neptunes. We consider the interplay between this miscibility and stellar-driven atmospheric escape. We find that, for the first $\sim 100$ Myrs, sub-Neptunes store most of their hydrogen content within their miscible interiors, protecting it from escape. As hydrogen is removed from the top of the atmosphere, more hydrogen is exsolved from the miscible interior, resupplying the envelope mass and delaying the planet's contraction when compared with models that do not account for miscibility. We construct a population-level Bayesian hierarchical inference model which tests for the prevalence of fully miscible sub-Neptunes. With conservative measurement uncertainty and no completeness corrections, we predict that $\sim 100-200$ observed young sub-Neptunes with measured radii, masses and ages $\lesssim 1$ Gyr are required to differentiate between fully miscible and immiscible models. This survey size requirement decreases with improved measurement uncertainty. Furthermore, one can reliably infer the underlying distributions of planetary initial conditions if completeness corrections are included in the analysis.

astro-ph.EP↗

Identifying deformed exoplanets with JWST and Ariel: A target selection metric for phase curve observations

The shape of an exoplanet encodes fundamental information about its internal structure. For close-in giant planets, strong stellar tidal forces induce measurable deformations that can be probed through high-precision photometry. These deformations are quantified by the fluid Love numbers, which characterise a planet's response to external perturbations and are directly linked to its internal density distribution. In this work, we assess the potential of JWST and the future Ariel mission to constrain the Love numbers of hot Jupiters through full-orbit phase curve observations. To identify the most promising targets for this science case, we introduce a new detection metric, the deformation metric (D), which combines the expected deformation amplitude with the anticipated photometric signal-to-noise ratio. We apply this metric to the Ariel target list and identify WASP-33 b, WASP-76 b, and HAT-P-32 b as the three most favourable targets that currently lack Love number measurements. Using WASP-76 b as a benchmark case, we perform end-to-end retrieval simulations and show that a single continuous JWST phase curve observation can constrain the Love number $h_{2f}$ to a precision of $σ_{h_2f} \approx 0.09$, while a single Ariel visit could achieve $σ_{h_{2f}} \approx 0.18$. Such precision is likely sufficient for Love number measurements to provide meaningful constraints on planetary interior models and to distinguish between competing interior-structure scenarios, opening a new observational avenue for probing the deep interiors of exoplanets.

astro-ph.EP↗

Are habitable-zone TESS Objects of Interest really there? Period resolution from the full archive

TESS Objects of Interest (TOIs) seen as single or duo transits carry a catalogue period that is undefined or equal to the separation of the two events, and their insolation and any habitable-zone label inherit it. We reanalyse all 187 TOIs catalogued with P = 0 or P >= 200 d using all public TESS data through Sector 107, with a flux-based method that weights each alias by per-epoch transit likelihoods, the transit duration and a geometric prior, and tests for coherence, data edges, contamination by other planets and instrumental artefacts. It recovers 12 published periods blindly and, in injection-recovery tests, 71 per cent of periods constrained by a new transit, with no wrong period. It resolves 28 periods and fixes 8 more by exclusion. Are habitable-zone TOIs really there? Mostly not: of the 28 duo-transits that a catalogue-period criterion places in the conservative habitable zone, about two are expected to lie there, both giant planets, and none of the 12 small (Rp < 4 Earth radii) ones; those resolved are aliases of hotter planets. Conversely, habitable-zone planets hide behind missing or wrong periods: we recover three published M-dwarf examples and validate TOI-2298 b, a 3.0 Earth-radius sub-Neptune at 146.58 d in the conservative habitable zone of a K dwarf. Of three further planets validated with new periods, TOI-6075 b (1.7 Earth radii, 416.5 d, 833 d not excluded) lies beyond the outer edge and is the second validated transiting planet smaller than 2 Earth radii with P > 1 yr. We release alias-marginalised habitable-zone probabilities for the whole sample.

astro-ph.EP↗