Mars climate model offers faster weather forecasting with machine learning
AEGIS: Differentiable Mars Climate Model with Neural Closures
Artificial Intelligence
Summary
Mars has a thin atmosphere and strange weather, making it hard to predict and understand with traditional tools. The authors built a new Mars climate model called AEGIS that combines detailed physics with modern neural network techniques. Unlike older models, AEGIS can learn and improve by adjusting itself using gradients. It can run stable long-term simulations of Mars weather and surface conditions in a more efficient and flexible way.
What this means in practice
- •For space mission engineers: Provide faster and tunable Mars weather forecasts critical for landing and operational decisions by using a differentiable climate model with neural components.
- •For weather model developers: Incorporate differentiability and neural closure modules into atmospheric models to accelerate calibration and physical parameter learning.
Authors
Sameera S Kashyap, Victor Cruz, Angel Yepez, Razvan Marinescu
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{} 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{} cycle while conserving the total \co{} 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.