Accurate medium-range global weather forecasting with 3D neural networks
2023/07/05 by Kaifeng Bi, Lingxi Xie, Hengheng Zhang +3 · 152 citations
Earth and Planetary Sciences · Environmental Science · #Climate variability and models #Meteorological Phenomena and Simulations #Tropical and Extratropical Cyclones Research
paper · pdf · doi:10.1038/s41586-023-06185-3
openalex publication_date 2023/07/05 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/31
Abstract
Abstract Weather forecasting is important for science and society. At present, the most accurate forecast system is the numerical weather prediction (NWP) method, which represents atmospheric states as discretized grids and numerically solves partial differential equations that describe the transition between those states 1 . However, this procedure is computationally expensive. Recently, artificial-intelligence-based methods 2 have shown potential in accelerating weather forecasting by orders of magnitude, but the forecast accuracy is still significantly lower than that of NWP methods. Here we introduce an artificial-intelligence-based method for accurate, medium-range global weather forecasting. We show that three-dimensional deep networks equipped with Earth-specific priors are effective at dealing with complex patterns in weather data, and that a hierarchical temporal aggregation strategy reduces accumulation errors in medium-range forecasting. Trained on 39 years of global data, our program, Pangu-Weather, obtains stronger deterministic forecast results on reanalysis data in all tested variables when compared with the world’s best NWP system, the operational integrated forecasting system of the European Centre for Medium-Range Weather Forecasts (ECMWF) 3 . Our method also works well with extreme weather forecasts and ensemble forecasts. When initialized with reanalysis data, the accuracy of tracking tropical cyclones is also higher than that of ECMWF-HRES.
Citations
Cited by
- Artificial intelligence-driven design of inorganic materials for biomedical applications
- Applications of Machine Learning and Artificial Intelligence in Tropospheric Ozone Research
- An Earth-System-Oriented View of the S2S Predictability of North American Weather Regimes
- Postprocessing East African Rainfall Forecasts Using a Generative Machine Learning Model
- The Case Against Climate Doom
- Land surface information from satellites boost near-surface temperature forecast skill
- Multi-scale Autoregressive Models are Laplacian, Discrete, and Latent Diffusion Models in Disguise
- Practical Quantum Advantage before Fault Tolerance via Quantum-Informed Machine Learning
- Hierarchical Testing of a Hybrid Machine Learning‐Physics Global Atmosphere Model
- Extreme Weather Bench: A framework and benchmark for evaluation of high-impact weather
- The Rise of AI in Weather and Climate Information and its Impact on Global Inequality
- AI-Based Regional Emulation for Kilometer-Scale Dynamical Downscaling
- FreIE: Low-Frequency Spectral Bias in Neural Networks for Time-Series Tasks
- Evaluating Extreme Precipitation Forecasts: A Threshold-Weighted, Spatial Verification Approach for Comparing an AI Weather Prediction Model Against a High-Resolution NWP Model
- Quadratic Direct Forecast for Training Multi-Step Time-Series Forecast Models
- Learning from History: A Retrieval-Augmented Framework for Spatiotemporal Prediction
- A PDE-Informed Latent Diffusion Model for 2-m Temperature Downscaling
- Unlocking the benefits of transparent and reusable science for climate risk management
- A Review of End-to-End Precipitation Prediction Using Remote Sensing Data: from Divination to Machine Learning
- A Framework for Hybrid Physics-AI Coupled Ocean Models
- HPC-Driven Modeling with ML-Based Surrogates for Magnon-Photon Dynamics in Hybrid Quantum Systems
- Hierarchical Graph Networks for Accurate Weather Forecasting via Lightweight Training
- Learning Coupled Earth System Dynamics with GraphDOP
- Signature Kernel Scoring Rule: A Spatio-Temporal Diagnostic for Probabilistic Weather Forecasting
- Sensing Without Colocation: Operator-Based Virtual Instrumentation for Domains Beyond Physical Reach
- Cortical-SSM: A Deep State Space Model for EEG and ECoG Motor Imagery Decoding
- Assessing Monotone Dependence: Area Under the Curve Meets Rank Correlation
- OmniCast: A Masked Latent Diffusion Model for Weather Forecasting Across Time Scales
- Monitoring and Modeling the Soil‐Plant System Toward Understanding Soil Health
- Foundation Models for Scientific Discovery: From Paradigm Enhancement to Paradigm Transition
- AB-UPT for Automotive and Aerospace Applications
- RainDiff: End-to-end Precipitation Nowcasting Via Token-wise Attention Diffusion
- An Operational Deep Learning System for Satellite-Based High-Resolution Global Nowcasting
- Assessing the Geographic Generalization and Physical Consistency of Generative Models for Climate Downscaling
- Deploying Atmospheric and Oceanic AI Models on Chinese Hardware and Framework: Migration Strategies, Performance Optimization and Analysis
- PAINT: Parallel-in-time Neural Twins for Dynamical System Reconstruction
- DAWP: A framework for global observation forecasting via Data Assimilation and Weather Prediction in satellite observation space
- Learning Operators through Coefficient Mappings in Fixed Basis Spaces
- Adversarial Attacks on Downstream Weather Forecasting Models: Application to Tropical Cyclone Trajectory Prediction
- Evaluating the Shanghai Typhoon Model against State-of-the-Art Machine-Learning Weather Prediction Models: A Case Study for Typhoon Danas (2025)
- The Equilibrium Response of Atmospheric Machine-Learning Models to Uniform Sea Surface Temperature Warming
- ARROW: An Adaptive Rollout and Routing Method for Global Weather Forecasting
- Simple and Robust Forecasting of Spatiotemporally Correlated Small Earth Data with A Tabular Foundation Model
- Quantifying Very Extreme Precipitation and Temperature Using Huge Ensembles Generated by Machine Learning-based Climate Model Emulators
- GyroSwin: 5D Surrogates for Gyrokinetic Plasma Turbulence Simulations
- Zeeman: A Deep Learning Regional Atmospheric Chemistry Transport Model
- Deep Learning Reconstruction of Tropical Cyclogenesis in the Western North Pacific from Climate Reanalysis Dataset
- Benchmarking atmospheric circulation variability in an AI emulator, ACE2, and a hybrid model, NeuralGCM
- Spatiotemporal Forecasting as Planning: A Model-Based Reinforcement Learning Approach with Generative World Models
- Zephyrus: An Agentic Framework for Weather Science
- Learning more physically realistic dynamics in machine-learning based weather forecasting with latent-space constraints
- Accuracy Law for the Future of Deep Time Series Forecasting
- Probability calibration for precipitation nowcasting
- Typhoon Tracks Regulated by Feedbacks of Fine-Scale Clouds to Environment
- A Synergistic Approach: Dynamics–AI Ensemble in Tropical Cyclone Forecasting
- Predictability Study of Weather and Climate Events Related to Artificial Intelligence Models
- A Weather Foundation Model for the Power Grid
- Using deep learning to generate key variables in global mitigation scenarios
- TimeExpert: Boosting Long Time Series Forecasting with Temporal Mix of Experts
- PHASE: Physics-Integrated, Heterogeneity-Aware Surrogates for Scientific Simulations
- Model Training, Data Assimilation, and Forecast Experiments with a Hybrid Atmospheric Model that Incorporates Machine Learning
- Forecasting the Future with Yesterday's Climate: Temperature Bias in AI Weather and Climate Models
- Task-Adaptive Parameter-Efficient Fine-Tuning for Weather Foundation Models
- VISION: Prompting Ocean Vertical Velocity Reconstruction from Incomplete Observations
- Differential-Integral Neural Operator for Long-Term Turbulence Forecasting
- Mesh Interpolation Graph Network for Dynamic and Spatially Irregular Global Weather Forecasting
- Graph-based Neural Space Weather Forecasting
- Can AI Weather Models Predict Beyond Two Weeks? A Quantitative Benchmark and Analysis of Long Rollouts
- A Big Chill: Examining an Extremely Sharp Cold Front from Three Perspectives
- Simultaneous Emulation and Downscaling With Physically Consistent Deep Learning‐Based Regional Ocean Emulators
- Global Forecasting of Tropical Cyclone Intensity Using Neural Weather Models
- The Impact of Tropical Convective Heating on the Amundsen Sea Low in a Deep-Learning Weather Forecasting Model
- Climate-Adaptive and Cascade-Constrained Machine Learning Prediction for Sea Surface Height under Greenhouse Warming
- Technical overview and architecture of the FastNet Machine Learning weather prediction model, version 1.0
- FastNet: Improving the physical consistency of machine-learning weather prediction models through loss function design
- Key role of the Madden-Julian Oscillation on tropical and subtropical humid heat and heatwaves
- An AutoML Framework using AutoGluonTS for Forecasting Seasonal Extreme Temperatures
- Communications to Circulations: Real-Time 3D Wind Field Prediction Using 5G GNSS Signals and Deep Learning
- STCast: Adaptive Boundary Alignment for Global and Regional Weather Forecasting
- A data-driven global ocean forecasting model with sub-daily and eddy-resolving resolution
- DPSformer: A long-tail-aware model for improving heavy rainfall prediction
- Accurate typhoon intensity forecasts using a non-iterative spatiotemporal transformer model
- FlowCast-ODE: Continuous Hourly Weather Forecasting with Dynamic Flow Matching and ODE Solver
- AERIS: Argonne Earth Systems Model for Reliable and Skillful Predictions
- FusionMAE: large-scale pretrained model to optimize and simplify diagnostic and control of fusion plasma
- SamudrACE: Fast and Accurate Coupled Climate Modeling with 3D Ocean and Atmosphere Emulators
- Data-Efficient Ensemble Weather Forecasting with Diffusion Models
- Knowledge-Guided Adaptive Mixture of Experts for Precipitation Prediction
- Two ways to knowledge?
- How does an AI Weather Model Learn to Forecast Extreme Weather?
- Breaking the Statistical Similarity Trap in Extreme Convection Detection
- Intelligent Shanghai Typhoon Model (ISTM): A generative probabilistic emulator for typhoon hybrid modeling
- S2Transformer: Scalable Structured Transformers for Global Station Weather Forecasting
- Seasonal forecasting using the GenCast probabilistic machine learning model
- Data-driven solar forecasting enables near-optimal economic decisions
- Select, then Balance: Exploring Exogenous Variable Modeling of Spatio-Temporal Forecasting
- Finetuning AI Foundation Models to Develop Subgrid-Scale Parameterizations: A Case Study on Atmospheric Gravity Waves
- LUCIE-3D: A three-dimensional climate emulator for forced responses
- MAUSAM: An Observations-focused assessment of Global AI Weather Prediction Models During the South Asian Monsoon
- FuXi-TC: A generative framework integrating deep learning and physics-based models for improved tropical cyclone forecasts
- From Noise to Precision: A Diffusion-Driven Approach to Zero-Inflated Precipitation Prediction
- DeepSeasons: a Deep Learning scale-selecting approach to Seasonal Forecasts
- Post-processing of ensemble photovoltaic power forecasts with distributional and quantile regression methods
- Automated discovery of finite volume schemes using Graph Neural Networks
- Validating Deep Learning Weather Forecast Models on Recent High-Impact Extreme Events
- Huracan: A skillful end-to-end data-driven system for ensemble data assimilation and weather prediction
- Learning from nature: insights into GraphDOP's representations of the Earth System
- GEN2: A Generative Prediction-Correction Framework for Long-time Emulations of Spatially-Resolved Climate Extremes
- MedFormer: a data-driven model for forecasting the Mediterranean Sea
- A Retrieval Augmented Spatio-Temporal Framework for Traffic Prediction
- Incoherent Light-Driven Nonlinear Optical Extreme Learner via Data Reverberation
- Lightning Prediction under Uncertainty: DeepLight with Hazy Loss
- Nonparametric Reaction Coordinate Optimization with Histories: A Framework for Rare Event Dynamics
- Large Model Driven Solar Activity AI Forecaster: A Scalable Dual Data-Model Framework
- MeteorPred: A Meteorological Multimodal Large Model and Dataset for Severe Weather Event Prediction
- Discovery Learning accelerates battery design evaluation
- Diffeomorphic Neural Operator Learning
- Enhanced climate reproducibility testing with false discovery rate correction
- FeDaL: Federated Dataset Learning for Time Series Foundation Models
- SolarSeer: Ultrafast and accurate 24-hour solar irradiance forecasts outperforming numerical weather prediction across the USA
- JustDense: Just using Dense instead of Sequence Mixer for Time Series analysis
- DMSC: Dynamic Multi-Scale Coordination Framework for Time Series Forecasting
- Flow Matching for Probabilistic Learning of Dynamical Systems from Missing or Noisy Data
- PnP-DA: Towards Principled Plug-and-Play Integration of Variational Data Assimilation and Generative Models
- A Deep Learning Earth System Model for Efficient Simulation of the Observed Climate
- OKG-LLM: Aligning Ocean Knowledge Graph with Observation Data via LLMs for Global Sea Surface Temperature Prediction
- Data-driven global ocean model resolving ocean-atmosphere coupling dynamics
- Physics-constrained generative machine learning-based high-resolution downscaling of Greenland's surface mass balance and surface temperature
- DualSG: A Dual-Stream Explicit Semantic-Guided Multivariate Time Series Forecasting Framework
- Unlocking Interpretability for RF Sensing: A Complex-Valued White-Box Transformer
- Extreme Solar Storm Reveals Causal Interactions in Space Weather
- Quantum-Informed Machine Learning for Predicting Spatiotemporal Chaos
- Low-rank adaptive physics-informed HyperDeepONets for solving differential equations
- A comparison of stretched-grid and limited-area modelling for data-driven regional weather forecasting
- Global weather forecasting through deep learning: applications and challenges
- FourCastNet 3: A geometric approach to probabilistic machine-learning weather forecasting at scale
- MVAR: MultiVariate AutoRegressive Air Pollutants Forecasting Model
- Role of Artificial Intelligence in Scientific Research: Classification Framework and Empirical Insights
- EPT-2 Technical Report
- Controllable Patching for Compute-Adaptive Surrogate Modeling of Partial Differential Equations
- XiChen: An observation-scalable fully AI-driven global weather forecasting system with 4D variational knowledge
- BioAnalyst: A Foundation Model for Biodiversity
- Intraseasonal Equatorial Kelvin and Rossby Waves in Modern AI-ML Models
- Mathematical artificial data for operator learning
- From General to Specialized: The Need for Foundational Models in Agriculture
- SciVid: Cross-Domain Evaluation of Video Models in Scientific Applications
- Deep Learning Atmospheric Models Reliably Simulate Out-of-Sample Land Heat and Cold Wave Frequencies
- Transparent Machine Learning: Training and Refining an Explainable Boosting Machine to Identify Overshooting Tops in Satellite Imagery
- Data and Code for Efficient Kilometer-Scale Precipitation Downscaling with Conditional Wavelet Diffusion
- Accurate Mediterranean Sea forecasting via graph-based deep learning
- Frequency-Aligned Knowledge Distillation for Lightweight Spatiotemporal Forecasting
- Probing forced responses and causality in data-driven climate emulators: conceptual limitations and the role of reduced-order models
Related