SciGroveBeta
Climate

A Deep Learning Earth System Model Simulation of Indian Monsoon Intraseasonal and Interannual Variability

Bijit Kumar Banerjee, Devabrat Sharma, R. I. Sujith, Chandrashekar Lakshminarayanan, Manikandan Narayanan, Subodh K. Saha, Anurag Dipankar, Utpal Sarma, B. N. Goswami

Featured July 10, 2026

Original Paper

This analysis was generated by SciGrove. Upload your own PDFs or enter a DOI — and get the same AI breakdown on any paper.

Get started

AI-generated analysis — This is SciGrove's AI interpretation of the paper, not peer-reviewed content. Always refer to the original paper.

Simply

A new AI model called SamudrACE can quickly simulate Earth's climate, but when tested on the Indian monsoon, it showed important differences from real-world observations, especially in how it handles big weather events like El Niño.

In depth
The paper rigorously evaluates SamudrACE, a novel deep learning 3D ocean-atmosphere coupled model, for its ability to simulate Indian monsoon intraseasonal and interannual variability. It systematically documents significant biases in SamudrACE's representation of key phenomena like equatorial waves, MISO propagation, ENSO amplitude, and the ENSO-monsoon teleconnection, providing a crucial benchmark for future improvements in AI-driven Earth system emulators.

Key Takeaways

  • 1
    SamudrACE, a deep learning Earth system emulator, offers a computationally efficient alternative to traditional climate models for sub-seasonal to seasonal (S2S) prediction.
  • 2
    The study identifies significant biases in SamudrACE's simulation of Indian monsoon intraseasonal oscillations (MISO) and interannual variability, including weaker ENSO amplitude and incorrect lead-lag relationships.
  • 3
    The documented biases, often inherited from its training data (GFDL-CM4), provide a critical benchmark for retraining and redesigning AI-coupled models to improve S2S prediction skill.

Conceptual Flow

HIGH LEVEL
1
Methodology: How SamudrACE Works

The AI model learns how the ocean and atmosphere interact separately, then links them together to simulate Earth's climate much faster than traditional methods.

Ocean Data
Atmosphere Data
AI Learns Climate Rules
Ocean AI Model
Atmosphere AI Model
Coupled Earth System
2
Results: Identifying Model Biases

The study compared the AI model's monsoon simulations to real observations and found specific areas where the AI model didn't quite match, like weaker El Niño events.

Real Monsoon
AI Model Monsoon
Compare Differences
Identified Biases
Areas for Improvement