Searth Transformer: A Transformer Architecture Incorporating Earth's Geospheric Physical Priors for Global Mid-Range Weather Forecasting

2026-01-14

AI Summary

This work addresses critical limitations in existing Transformer-based weather forecasting models, which neglect Earth’s spherical geometry, zonal periodicity, and meridional boundary conditions, while also suffering from high computational costs and severe error accumulation under conventional autoregressive training. To overcome these challenges, the authors propose the Searth Transformer architecture, which uniquely integrates geophysical priors into a windowed self-attention mechanism, alongside a low-overhead relay autoregressive (RAR) fine-tuning strategy. The resulting YanTian model, evaluated at 1° resolution, outperforms the ECMWF high-resolution system, achieving a Z500 anomaly correlation skill horizon of 10.3 days. Moreover, it reduces computational cost by approximately 200-fold compared to standard autoregressive approaches, substantially enhancing both efficiency and accuracy in medium-range global weather prediction.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Computer Vision: Diffusion Models for Vision
• Planning, Routing, and Scheduling: Planning with Language Models

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
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Accurate global medium-range weather forecasting is fundamental to Earth system science. Most existing Transformer-based forecasting models adopt vision-centric architectures that neglect the Earth's spherical geometry and zonal periodicity. In addition, conventional autoregressive training is computationally expensive and limits forecast horizons due to error accumulation. To address these challenges, we propose the Shifted Earth Transformer (Searth Transformer), a physics-informed architecture that incorporates zonal periodicity and meridional boundaries into window-based self-attention for physically consistent global information exchange. We further introduce a Relay Autoregressive (RAR) fine-tuning strategy that enables learning long-range atmospheric evolution under constrained memory and computational budgets. Based on these methods, we develop YanTian, a global medium-range weather forecasting model. YanTian achieves higher accuracy than the high-resolution forecast of the European Centre for Medium-Range Weather Forecasts and performs competitively with state-of-the-art AI models at one-degree resolution, while requiring roughly 200 times lower computational cost than standard autoregressive fine-tuning. Furthermore, YanTian attains a longer skillful forecast lead time for Z500 (10.3 days) than HRES (9 days). Beyond weather forecasting, this work establishes a robust algorithmic foundation for predictive modeling of complex global-scale geophysical circulation systems, offering new pathways for Earth system science.

Problem

• weather forecasting
• Transformer architecture
• spherical geometry
• zonal periodicity
• autoregressive training

Innovation

• Searth Transformer
• zonal periodicity
• Relay Autoregressive fine-tuning
• physics-informed architecture
• global weather forecasting