Portolan makes geospatial data
easy to share and use.
Portolan is an open-source specification and toolkit that combines cloud-native file formats with clear metadata and documentation so that people and agents can publish and use data at any scale.
A better way to work with geospatial data
Start with real questions, not a script. Watch an agent find relevant catalogs in the Portolan registry and answer complex, natural-language questions about flood risk in Moldova.
A terminal-only recording shows an agent find data about flood risk in Chisinau through the live Portolan registry. It inspects the Moldova geodata catalog and related catalogs for flood extents, flood depth, buildings, population, land cover, hydrography, roads, utilities, and boundaries. It reports Moldova flood scenarios for 1%, 0.5%, and 0.1% annual probabilities, dam-breach extents, JRC GLOFAS depth layers, and GHSL population grids. It then calculates screening estimates for exposed buildings and population, and ranks several planning corridors. The result notes that the estimates lack local depth, velocity, arrival time, and building-vulnerability data.
Why Portolan
Open and interoperable
Portolan builds on open standards like GeoParquet, PMTiles, COG, and STAC so that data can be dropped right into tools like QGIS, ArcGIS, Pandas, or BigQuery.
Simple, cheap, scalable, sovereign
With Portolan, your data are just files in a cloud bucket. Store them with any S3-compatible provider, in any jurisdiction you choose. Your only costs are storage and egress. Scale from megabytes to petabytes with no servers, databases, or custom APIs.
Built for people and agents
Portolan catalogs are well-structured, well-formatted, and well-documented, so people and agents can find, understand, and use them. Browse the registry yourself to discover new data, or point an agent at the registry and see what it can do.
Who it's for
Large publishers
Large archives, such as satellite imagery or national-scale geoportals, are expensive to store and hard to maintain.
Portolan makes it straightforward to publish that data as cheaply as possible, without servers, databases, or APIs, while making it easier to discover and use.
Small publishers
Many city governments, NGOs, and other smaller data producers can't afford the cost or complexity of publishing data with tools like GeoServer or Esri.
Portolan lowers the barrier to entry by letting you share your data with nothing more than a cloud storage bucket.
Data users
Most geospatial data is scattered, poorly documented, and stored in ways that make it hard to use.
Portolan requires clear metadata and documentation, uses formats designed for direct access, and provides a central registry of public catalogs. You spend less time finding, downloading, and wrangling data, and more time putting it to work.
How it works
Use the Portolan CLI or agent skills to transform your existing data into a Portolan catalog.
A terminal-only recording shows an agent create and publish a Portolan catalog from ten Open Data Philly datasets. The request names property parcels, land use, zoning base districts and overlays, zoning descriptions, vacant-property indicators, City Council districts, building footprints, and affordable housing production. The agent loads the Portolan bootstrap skill, researches and downloads the sources, enriches metadata, creates thumbnails and map styles, validates the catalog, and publishes it. The terminal reports about 1.78 million features and that all catalog gates pass. It also records unresolved source fields, license details, and upstream tool issues before it syncs the published catalog files and agent guides.
- 01
Convert
Whether you're working with Shapefiles and GeoPackages, a WFS endpoint, or an Esri server, your data end up as GeoParquet and PMTiles for vector and COG for raster.
- 02
Catalog
Your cloud-optimized data are organized as a STAC catalog with clear metadata and docs, making the catalog easy to find, understand, and use for people and agents.
- 03
Publish
Your catalog is published to any S3-compatible storage, such as AWS, GCS, Hetzner, or MinIO, with no servers or APIs to set up.
- 04
Use
Your data can now be dropped directly into tools like QGIS, ArcGIS, or Python, or you can pass the catalog URL to an agent and let it get to work.
Explore catalogs
See independently hosted geospatial catalogs from publishers around the world.
Browse the registry
Ecosystem
The Portolan specification can be implemented in a number of ways. Here is a list of core, community-contributed, and commercial tools to make publishing and working with Portolan data easy.
The specification that defines a Portolan catalog.
The validator for checking whether a catalog conforms to the Portolan specification.
A command-line tool for creating, checking, and publishing Portolan catalogs.
A central registry of public Portolan catalogs.
Agent skills for creating and working with Portolan catalogs.
A spatial data infrastructure for organizations, built on Portolan.
Commercial products on the open core
Portolan standardizes the published catalog, not the software around it, so companies can sell capabilities the open tools do not cover. CARTO SDI extends Portolan for organizations with private datasets, governance and audits, dataset commercialization, and writable datasets. If a product goes away, the catalog keeps working and another tool can take over.
Get involved
Portolan is an open-source community project. We welcome new users and contributors. Join the #portolan channel on Slack and follow the Portolan Google Group for details on our weekly meetings.
The simplest way to get started right now is to publish your data as a Portolan catalog and let us know how it goes. Feedback on what works, what doesn't, and how Portolan can be improved is always helpful, especially when your implementation differs from existing catalogs.
We also welcome contributions to the Portolan spec and tooling in the form of bug reports, feature requests, and PRs (and especially PRs). Check out our open issues to get started.