Pangeo Forge#

Pangeo Forge is an open source framework for data Extraction, Transformation, and Loading (ETL) of scientific data. The goal of Pangeo Forge is to make it easy to extract data from traditional data archives and deposit in cloud object storage in analysis-ready, cloud-optimized (ARCO) formats.

Pangeo Forge is made of three official components:

  • pangeo-forge-recipes Python SDK - A Python toolkit, built on Apache Beam, that provides components for concisely and expressively composing parallelizable ETL pipelines (“recipes”) that can be run on your local machine or cloud infrastructure. See Recipe Composition for more on composing recipes, and Deployment for ways to run them.

  • pangeo-forge-runner Command Line Interface (CLI) - A utility for managing the configuration and deployment of version-controlled recipes. See Command Line Interface for details.

  • Deploy Recipe Action - A Github Action that wraps the CLI, providing an interface for configuration and deployment of recipes in response to to GitHub Event triggers. See GitHub Action for details.

A growing Ecosystem of third-party extensions provide additional reusable components for customization of recipes.

Pangeo Forge is inspired directly by Conda Forge, a community-led collection of recipes for building conda packages. We hope that Pangeo Forge can play a similar role for datasets.

The Pangeo Forge Paper#

We wrote a long academic paper about Pangeo Forge. This is the best thing to read if you want a deep dive into why we created Pangeo Forge. If you use Pangeo Forge in academic work, please cite this paper.


As noted in the paper, “Pangeo Forge follows an agile development model, characterized by rapid iteration, frequent releases, and continuous feedback from users.” Readers should therefore be mindful that while core principles including reproducibility, collaboration, and provenance transparency remain consistent over time, implementation details and project scope have evolved since publication.

Frequently Asked Questions#

Is Pangeo Forge the right tool for my dataset?#

Pangeo Forge is currently focused primarily on multidimensional array data; i.e., data that fit the NetCDF data model and can be opened by Xarray. This includes NetCDF, GRIB, HDF4/5, Zarr, OPeNDAP, and many other obscure formats. Pangeo Forge is not currently focused on tabular data; i.e., data that can be opened by Pandas, CSV, SQL, Parquet, Arrow, etc. This data model and these formats are very well supported by other ETL tools in the modern data stack.

So if you want to do ETL on multidimensional array data, Pangeo Forge is for you!

How is Pangeo Forge funded?#

Right now, the development of Pangeo Forge is funded exclusively by the US National Science Foundation (NSF) EarthCube program, via award 2026932 to Lamont Doherty Earth Observatory. The lead PI of the award is Ryan Abernathey.

Going forward, we hope to establish a multi-stakeholder coalition to support the operation of Pangeo Forge as a stable, reliable service for the research community. Please reach out if you’re interested in helping with this effort.

How can I help with development?#

Check out Contributing for how to get started!

Connecting with the Community#

Pangeo Forge is a community run effort with a variety of roles:

  • Recipe developers — Data engineers and enthusiasts who write recipes to define the data conversions. This can be anyone with a desire to create analysis ready cloud-optimized (ARCO) data. Explore Recipe Composition for more on this role.

  • Data users - Analysts, scientists, and domain experts who use the ARCO data produced by Pangeo Forge in their work and research.

  • Tooling developers - Scientists and software developers who maintain and enhance the open-source code base which makes Pangeo Forge run. See Contributing for more.

Site Contents#