NVIDIA/earth2studio

Open-source deep-learning framework for exploring, building and deploying AI weather/climate workflows.

What it solves

Earth2Studio simplifies the process of building and researching AI-driven weather and climate science. It removes the friction of setting up complex AI Earth system models by providing a unified interface to various third-party models, datasets, and frameworks, allowing researchers to run forecasts quickly without managing disparate APIs.

How it works

It functions as an AI inference pipeline toolkit. It provides a unified API that allows users to compose pipelines by chaining together different components: data sources (such as GFS or IFS), AI models (such as FourCastNet3, AIFS, or GraphCast), and output backends (like Zarr). This modular architecture allows users to easily swap out models or data sources to compare results or refine their workflows.

Who it’s for

This toolkit is designed for researchers and developers in the field of weather and climate science who want to leverage AI for forecasting and Earth system modeling.

Highlights

  • Extensive Model Zoo: Access to a wide range of weather and climate AI models, including NVIDIA's open models and third-party operational models.
  • Composable Pipelines: Ability to chain multiple data sources, AI models, and other modules into complex inference workflows.
  • Unified API: A single interface that works across different AI frameworks and model architectures.
  • Optimized Data Access: Includes tools for efficient access to cloud data stores and statistical operations.
  • Agent-Assisted Setup: Integration with coding agents to automate environment configuration and workflow creation.

Related

  • Project
    ecmwf-lab/ai-models

    A command-line tool for running AI-based weather forecasting models, providing a unified interface for data retrieval and model execution.

  • Project
    opengeos/geoai

    A Python package that integrates AI frameworks with geospatial data analysis, providing tools for satellite imagery processing, model training, and inference.

  • Project
    GT4SD/gt4sd-core

    An open-source platform that provides a unified API and toolkit for using state-of-the-art generative AI models to accelerate hypothesis generation in scientific discovery.

  • Project
    microsoft/farmvibes-ai

    A multi-modal geospatial machine learning framework for agriculture and sustainability that fuses satellite imagery, weather data, and elevation maps to generate robust agricultural insights.

  • Project
    google-gemini/genai-processors

    A lightweight Python library for building modular, asynchronous, and composable AI pipelines that unify multimodal content processing and streaming.