Accelerating Radiative Transfer for Planetary Atmospheres by Orders of Magnitude with a Transformer-Based Machine Learning Model

2025-10-30

AI Summary

Radiative transfer calculations in planetary atmospheres are computationally expensive, limiting both the accuracy and efficiency of general circulation models (GCMs). To address this, we propose the first Transformer-based surrogate model specifically designed for hot Jupiter atmospheres. Our encoder-only architecture takes one-dimensional atmospheric profiles—comprising static parameters such as temperature, pressure, and chemical composition—as input and directly predicts two-stream radiative fluxes. Trained via supervised learning, the model achieves a mean relative error of approximately 1% on held-out test data while accelerating computations by roughly two orders of magnitude compared to conventional numerical radiative transfer solvers. This work marks the first successful application of self-attention mechanisms to exoplanetary thermal radiative transfer modeling, achieving both high fidelity and real-time inference capability. It provides a scalable, physically consistent machine learning alternative for embedding radiation schemes into GCMs.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Planning, Routing, and Scheduling: Planning with Language Models
• Computer Vision: Diffusion Models for Vision

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Radiative transfer calculations are essential for modeling planetary atmospheres. However, standard methods are computationally demanding and impose accuracy-speed trade-offs. High computational costs force numerical simplifications in large models (e.g., General Circulation Models) that degrade the accuracy of the simulation. Radiative transfer calculations are an ideal candidate for machine learning emulation: fundamentally, it is a well-defined physical mapping from a static atmospheric profile to the resulting fluxes, and high-fidelity training data can be created from first principles calculations. We developed a radiative transfer emulator using an encoder-only transformer neural network architecture, trained on 1D profiles representative of solar-composition hot Jupiter atmospheres. Our emulator reproduced bolometric two-stream layer fluxes with mean test set errors of ~1% compared to the traditional method and achieved speedups of 100x. Emulating radiative transfer with machine learning opens up the possibility for faster and more accurate routines within planetary atmospheric models such as GCMs.

Problem

• Accelerating radiative transfer calculations for planetary atmospheres
• Overcoming computational demands and accuracy-speed trade-offs
• Enabling faster and more accurate atmospheric modeling routines

Innovation

• Transformer-based neural network emulates radiative transfer
• Achieves 100x speedup with 1% mean error
• Enables faster, more accurate planetary atmospheric modeling