Advanced long-term earth system forecasting by learning the small-scale nature

2025-05-26

AI Summary

AI models suffer from spectral bias in long-horizon autoregressive Earth system forecasting, impairing their ability to resolve high-frequency, small-scale processes and leading to error accumulation and dynamical instability. To address this, we propose Triton—a novel multi-resolution hierarchical architecture that introduces explicit cross-scale dynamical modeling. Triton integrates multi-scale feature extraction, hierarchical spatiotemporal representation, and an unforced self-evolution training paradigm, fundamentally mitigating long-range instability. Experiments demonstrate stable 1-year global surface temperature prediction, 120-day Kuroshio eddy trajectory forecasting, and high-fidelity 3D turbulence simulation. Triton achieves significantly superior long-term accuracy and dynamical consistency compared to state-of-the-art AI methods, breaking through both the spectral bias bottleneck and the autoregressive stability limit in AI-based meteorological modeling.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Reliable long-term forecast of Earth system dynamics is heavily hampered by instabilities in current AI models during extended autoregressive simulations. These failures often originate from inherent spectral bias, leading to inadequate representation of critical high-frequency, small-scale processes and subsequent uncontrolled error amplification. We present Triton, an AI framework designed to address this fundamental challenge. Inspired by increasing grids to explicitly resolve small scales in numerical models, Triton employs a hierarchical architecture processing information across multiple resolutions to mitigate spectral bias and explicitly model cross-scale dynamics. We demonstrate Triton's superior performance on challenging forecast tasks, achieving stable year-long global temperature forecasts, skillful Kuroshio eddy predictions till 120 days, and high-fidelity turbulence simulations preserving fine-scale structures all without external forcing, with significantly surpassing baseline AI models in long-term stability and accuracy. By effectively suppressing high-frequency error accumulation, Triton offers a promising pathway towards trustworthy AI-driven simulation for climate and earth system science.

Problem

• Addresses AI model instabilities in long-term Earth system forecasting
• Mitigates spectral bias to improve high-frequency small-scale process representation
• Enables stable, accurate year-long climate and turbulence simulations

Innovation

• Hierarchical architecture for multi-resolution processing
• Explicit modeling of cross-scale dynamics
• Suppressing high-frequency error accumulation effectively