FedRAIN-Lite: Federated Reinforcement Algorithms for Improving Idealised Numerical Weather and Climate Models

2025-08-19

AI Summary

Climate model subgrid parameterizations traditionally rely on static, offline tuning, limiting adaptability to dynamically evolving climate states. To address this, we propose FedRAIN-Lite—the first federated reinforcement learning framework tailored for idealized climate models. It deploys agents across latitude bands to enable geographically localized parameter learning and globally coordinated aggregation. By integrating federated learning with deep deterministic policy gradient (DDPG), FedRAIN-Lite constructs a lightweight multi-agent architecture on energy balance models (EBM-v1/v2/v3), balancing physical interpretability and computational efficiency. Experiments demonstrate substantial improvements over static and single-agent baselines: weighted RMSE reductions in tropical and mid-latitude regions, accelerated convergence, and strong cross-hyperparameter transferability. FedRAIN-Lite establishes a novel, evolvable, and geographically adaptive paradigm for online optimization in climate modeling.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Multiagent Systems: Multiagent Learning
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Sub-grid parameterisations in climate models are traditionally static and tuned offline, limiting adaptability to evolving states. This work introduces FedRAIN-Lite, a federated reinforcement learning (FedRL) framework that mirrors the spatial decomposition used in general circulation models (GCMs) by assigning agents to latitude bands, enabling local parameter learning with periodic global aggregation. Using a hierarchy of simplified energy-balance climate models, from a single-agent baseline (ebm-v1) to multi-agent ensemble (ebm-v2) and GCM-like (ebm-v3) setups, we benchmark three RL algorithms under different FedRL configurations. Results show that Deep Deterministic Policy Gradient (DDPG) consistently outperforms both static and single-agent baselines, with faster convergence and lower area-weighted RMSE in tropical and mid-latitude zones across both ebm-v2 and ebm-v3 setups. DDPG's ability to transfer across hyperparameters and low computational cost make it well-suited for geographically adaptive parameter learning. This capability offers a scalable pathway towards high-complexity GCMs and provides a prototype for physically aligned, online-learning climate models that can evolve with a changing climate. Code accessible at https://github.com/p3jitnath/climate-rl-fedrl.

Problem

• Improving static sub-grid parameterizations in climate models
• Enabling adaptive learning with federated reinforcement algorithms
• Scaling parameter learning for high-complexity general circulation models

Innovation

• Federated reinforcement learning for climate models
• DDPG algorithm outperforms static baselines
• Geographically adaptive parameter learning approach