Skip to main content

Accessing the Data

Official Overture sources​

The Overture Maps Foundation provides free access to its data on Amazon S3 and Microsoft Azure Blob Storage, with no account registration or credentials required.

ProviderLocation
Amazon S3s3://overturemaps-us-west-2/release/<RELEASE>
Microsoft Azure Blob Storagehttps://overturemapswestus2.blob.core.windows.net/release/<RELEASE>

The latest Overture data <RELEASE> is: 

2025-04-23.0/

Getting only the data you want​

The base paths above point to all of the data in a particular Overture release, about 500 GigaBytes (GB) worth. Overture partitions the data by theme and type; specifying those directories in your path will significantly narrow the universe of data you are trying to access. For example, here's the S3 path to the infrastructure feature type in the base theme:

s3://overturemaps-us-west-2/release/2025-04-23.0/theme=base/type=infrastructure/*.parquet

Let's break down the components in that path:

  • <RELEASE>: Overture releases data monthly — following a date-based versioning scheme in the format yyyy-mm-dd.x —and maintains an archive of releases on Amazon S3 and Azure Blob Storage
  • <theme>: one of Overture's six data themes — addresses, base, buildings, divisions, places, and transportation — and the first level of partitioning of the data
  • <type>: a feature type within a theme, e.g. infrastructure within base, and the second level of partitioning of the data |
  • *.parquet: the file type Overture uses to store and deliver the data, the * indicates you want all of the Parquet files in a particular directory

Parquet files​

Overture distributes its data as GeoParquet files, a column-oriented spatial data format that is a backwards-compatible extension of Apache Parquet. Parquet (and GeoParquet) is optimized for cloud-native queries, which means you can efficiently fetch chunks of the data from cloud storage. Instead of extracting one huge data file containing a planet's worth of data, you can scan across the many GeoParquet files in each type directory and pull only the data you need. To learn more about GeoParquet, you can consult this guide from the Cloud Native Geospatial Forum.

Quickstart​

Let's go back to the example above with the infrastructure map data. Now that you know the correct path, how do you actually query and download the data? Below are a few examples to get you started quickly.

AWS CLI​

You can install the AWS CLI to download the Parquet files from S3 at the command line. Set <DESTINATION> to a local directory path to download the files, or to an s3:// path you control to copy them into your S3 bucket.

In this example, you can use the AWS CLI to download all nine Parquet files, about one GB of data, in the infrastructure feature type to the current working directory on your local drive.

aws s3 cp --no-sign-request --recursive s3://overturemaps-us-west-2/release/2025-04-23.0/theme=base/type=infrastructure/ .

Here's another example with the AWS CLI, this time extracting all of the data for the building feature type:

aws s3 cp --no-sign-request --recursive s3://overturemaps-us-west-2/release/2025-04-23.0/theme=buildings/type=building/ .

This will download 200 GeoParquet files containing all of the building feature type data around the world — that's more than 230 GB of data!

tip

If you'd like to query the data directly on the AWS platform, please read our documentation about setting up an Overture data catalog on Athena. You'll need an AWS account secured with a credit card.

AzCopy​

You can download the files from Azure Blob Storage directly using Azure Storage Explorer or by installing Microsoft's AzCopy command-line tool. Here's how you would use theazcopy command to grab all of the place data:

azcopy copy "https://overturemapswestus2.dfs.core.windows.net/release/2025-04-23.0/theme=places/type=place/" "<<local directory path>>"  --recursive
tip

You can read more about querying with the data directly in Azure using Synapse. You'll need an account backed by a credit card to do this.

DuckDB​

DuckDB is one of our favorite tools for querying the data and downloading only what you need. We have a full page of example queries here; we'll give you one quick example below.

Let's say you don't want the entire buildings dataset for the planet. With DuckDB, you can scan the Parquet files in the S3 bucket and extract only the data for a particular area of interest:

install spatial;
load spatial;
COPY(                             
      SELECT
         id, -- GERS ID
         names.primary as name,
         confidence,
         CAST(socials AS JSON) as socials, -- serialize property to JSON
         geometry -- DuckDB understands the geometry type
      FROM read_parquet('s3://overturemaps-us-west-2/release/2025-04-23.0/theme=places/type=place/*', filename=true, hive_partitioning=1)
      WHERE categories.primary = 'pizza_restaurant'
      AND bbox.xmin BETWEEN -75 AND -73  AND bbox.ymin BETWEEN 40 AND 41 -- with point geometries you only need the bbox min values
  
  ) TO 'nyc_pizza.geojson' WITH (FORMAT GDAL, DRIVER 'GeoJSON');

Alternatively, you can put your query in a SQL file and run duckdb -f query.sql at the command line.

Python CLI​

Overture also offers its own Python CLI for downloading data at the command line.

$ overturemaps download --bbox=-75,40,-73,41 -f geojson --type=building -o nyc_buildings.geojson

Data mirrors​

In addition to the official sources maintained by the Overture Maps Foundation, several friends of Overture are maintaining mirrors of the data on these platforms:

These are community-maintained resources. These mirrors may offer different access patterns, additional tooling, or other platform-specific benefits. Please consult the documentation for details.

Last updated on