Skip to main content

Wherobots

Wherobots provides a cloud-native environment for geospatial data analysis, powered by Apache Sedona, allowing you to work with massive geospatial datasets at scale. Wherobots, an Overture Maps Foundation member, maintains a mirror of the Overture Maps data catalog on its platform.

In this example, we'll show you two ways to work with Overture Maps data in the Wherobots platform: loading and querying the Overture Maps data tables that already exist in the Wherobots cloud; and accessing and querying GeoParquet files directly from Overture Map's own Amazon S3 buckets.

info

Before you get started you need to create a Wherobots account. You can create a free Wherobots Community Edition Organization or you can sign up for a paid Wherobots Professional Organization on AWS marketplace.

Part 1: Working with Overture Maps data tables in the Wherobots platform​

Setting up your notebook​

In your Wherobots account, go to the notebooks section and create a “Tiny” instance. This will spin up a Sedona Spark cluster with a Jupyter notebook environment. You'll notice that you have several example notebooks available to you. For this exercise, you'll need to create a new notebook with a Python 3 kernel. Run the following code in a notebook cell to import the necessary dependencies and create a SedonaContext that provides access to powerful spatial operations.

from sedona.spark import *
from pyspark.sql import functions as f
from pyspark.sql.functions import col

config = (SedonaContext.builder().getOrCreate())
sedona = SedonaContext.create(config)

Available datasets​

In your notebook, you can use the following line of code to list the available Overture Maps data tables.

# List all available Overture datasets
sedona.sql("SHOW TABLES IN wherobots_open_data.overture_maps_foundation").show(truncate=False)

When you run that query you’ll get a list of all the available datasets under tableName:

+------------------------+---------------------------+-----------+
|namespace               |tableName                  |isTemporary|
+------------------------+---------------------------+-----------+
|overture_maps_foundation|geocodes                   |false      |
|overture_maps_foundation|places_place               |false      |
|overture_maps_foundation|transportation_segment     |false      |
|overture_maps_foundation|transportation_connector   |false      |
|overture_maps_foundation|buildings_building         |false      |
|overture_maps_foundation|divisions_division_area    |false      |
|overture_maps_foundation|buildings_building_part    |false      |
|overture_maps_foundation|base_land_cover            |false      |
|overture_maps_foundation|base_land_use              |false      |
|overture_maps_foundation|base_infrastructure        |false      |
|overture_maps_foundation|divisions_division_boundary|false      |
|overture_maps_foundation|base_land                  |false      |
|overture_maps_foundation|base_water                 |false      |
|overture_maps_foundation|divisions_division         |false      |
|overture_maps_foundation|addresses_address          |false      |
|overture_maps_foundation|base_bathymetry            |false      |
+------------------------+---------------------------+-----------+

To call any of these datasets, use the following syntax for the table name: wherobots_open_data.overture_maps_foundation.{tableName}.

Loading Overture data​

You can load Overture Maps data into your notebook using either the DataFrame API or SQL API:

Here's how to create a Wherobots dataframe with of the Overture Places dataset:

# Load Places dataset
places_df = sedona.table("wherobots_open_data.overture_maps_foundation.places_place")

# Display sample data
places_df.show(5)

Example query: filtering by geography​

Let's run a query to get a count of all the Overture Maps places in San Francisco.

# Define an area of interest (e.g., San Francisco bounding box)
sf_bbox = "POLYGON((-122.512 37.708, -122.512 37.810, -122.357 37.810, -122.357 37.708, -122.512 37.708))"

# Find places within San Francisco
sf_places = sedona.sql(f"""
    SELECT *
    FROM places
    WHERE ST_Intersects(geometry, ST_GeomFromWKT('{sf_bbox}'))
""")

# Show result count
print(f"Found {sf_places.count()} places in the specified area")
sf_places.show()

Part 2: Access and query Overture Maps data directly from Amazon S3​

While Wherobots provides its own hosted version of the Overture dataset, the next part of this example explains how to perform Spatial SQL queries directly on the Overture Maps data stored on Amazon S3. We'll demonstrate a simple ETL pipeline: extract GeoParquet data from S3, transform it through geographic filtering and schema refinement, and load it into a clean dataframe in your Jupyter notebook.

Get the Amazon S3 path for the Overture Maps data you want to query​

Review the getting started page in the Overture Maps documentation to determine the Amazon S3 Path of the data that you wish to query. This example will query the building type from the buildings theme and the segment type from the transportation theme.

Setting up your notebook environment​

On the Wherobots Cloud landing page, select an available runtime to launch a new notebook. Once the runtime has loaded, open the notebook. In a notebook cell, import the necessary dependencies and initialize the Sedona Context.

from sedona.spark import *
from pyspark.sql import functions as f
from pyspark.sql.functions import col

config = (SedonaContext.builder().getOrCreate())
sedona = SedonaContext.create(config)

Querying Overture Maps data in Wherobots​

Once you have the correct S3 path, you can query the Overture Maps data. The code example below takes two separate, siloed datasets (buildings and transportation) and merges them into a single, unified table.

Define your area of interest (in this case, Golden Gate Park) and a reusable helper function to load, filter, and combine the buildings and transportation data. Click Run > Run Selected Cell to execute the query and see the results.

# Define the region of interest as San Francisco's Golden Gate Park
region_wkt = "POLYGON ((-122.5106 37.7736, -122.4549 37.7788, -122.4491 37.7656, -122.5103 37.7606, -122.5106 37.7736))"

# Define the release version as a variable for configurability
RELEASE_VERSION = "2025-06-25.0"  # Update this value when a new release is available

def process_overture_layer(theme, type, region_wkt):
    """
    Loads an Overture theme from S3, filters it by a geographic region,
    and selects the required columns.
    """
    s3_path = f"s3://overturemaps-us-west-2/release/{RELEASE_VERSION}/theme={theme}/type={type}"
    print(f"Processing {theme} in Golden Gate Park...")

    df = (
        sedona.read.parquet(s3_path)
        .withColumn("geometry", f.expr("ST_GeomFromWKB(geometry)"))
        .where(f.expr(f"ST_Intersects(geometry, ST_GeomFromText('{region_wkt}'))"))
        .select(
            "geometry",
            f.lit(theme).alias("layer"),
            f.element_at(f.col("sources"), 1)["dataset"].alias("source")
        )
    )
    return df

# Process each layer using the function
buildings_df = process_overture_layer("buildings", "building", region_wkt)
roads_df = process_overture_layer("transportation", "segment", region_wkt)

# Union the processed dataframes and get the final count
combined_df = buildings_df.unionByName(roads_df)

print("\n--- Analysis Complete ---")
print(f"Found {combined_df.count()} total features in Golden Gate Park.")
combined_df.show(10)

Resources​

For more on using the Wherobots platform, see the official Wherobots documentation.

Last updated on