ecmwf-lab/ai-models

Run AI-based weather forecasting models with ECMWF data

What it solves

This project provides a unified command-line interface to run various AI-based weather forecasting models, simplifying the process of managing assets, fetching input data from meteorological archives, and generating forecasts.

How it works

The ai-models tool acts as a wrapper for several AI weather models. It handles the downloading of trained weights (assets) and the retrieval of initial conditions (input data) from sources like the ECMWF MARS archive, the Copernicus Climate Data Store (CDS), or local GRIB files. Users can specify the forecast lead time and output format, with the tool managing the execution on either CPU or GPU (via ONNX runtime).

Who it’s for

Meteorologists, climate scientists, and researchers who want to run AI-based weather forecasts without needing to manually handle the complex data pipelines for each individual model.

Highlights

  • Multi-model support: Supports plugins for models such as Pangu-Weather, FourCastNet, FourCastNetV2, and GraphCast.
  • Flexible data sourcing: Automatically fetches input data from MARS, CDS, or accepts local GRIB files.
  • Customizable forecasts: Allows users to define lead times (e.g., 10 or 15 days) and and output file paths using metadata placeholders.
  • GPU acceleration: Optimized for GPU performance to reduce forecast time from several hours to approximately one minute.

Related

  • Project
    NVIDIA/earth2studio

    A Python-based AI inference pipeline toolkit for weather and climate science that provides a unified API to integrate various AI models and data sources.

  • Project
    DashAISoftware/dashAI

    A graphical toolbox for training, evaluating, and deploying state-of-the-art AI models with support for various hardware accelerators.

  • Project
    galilai-group/stable-worldmodel

    A platform for reproducible world model research and evaluation that provides a unified interface for data collection, training, and model-predictive control evaluation.

  • Project
    reyamira/models

    A TUI and CLI tool for browsing AI models, comparing performance benchmarks, tracking AI coding agents, and monitoring provider service statuses.

  • Project
    openclimatefix/graph_weather

    A PyTorch library implementing various graph neural networks for weather forecasting and data assimilation, including models like GenCast and Aurora.