down the rabbit hole of Overture's admins dataset, 2024-02-15 data release

Raw

This query lets us grab the locality (point geometry and attributes) that represents the United States and its associated area, in this case a multipolygon geometry.

CREATE VIEW admins_view AS (
    SELECT
        *
    FROM
        read_parquet('s3://overturemaps-us-west-2/release/2024-02-15-alpha.0/theme=admins/type=*/*', filename=true, hive_partitioning=1)COPY (
COPY (
SELECT
            admins.id,
            admins.subType,
            admins.isoCountryCodeAlpha2,
            names.primary AS primary_name,
            sources[1].dataset AS primary_source,
            areas.areaId,
            ST_GeomFromWKB(areas.areaGeometry) as geometry
    FROM admins_view AS admins
    INNER JOIN (
        SELECT
            id as areaId,
            localityId,
            geometry AS areaGeometry
        FROM admins_view
    ) AS areas ON areas.localityId == admins.id
    WHERE admins.adminLevel = 1
    AND admins.isoCountryCodeAlpha2 = 'US'
    ) TO 'us-boundary.geojson'
WITH (FORMAT GDAL, DRIVER 'GeoJSON');

Would something like this work to get all the buildings within that multipolygon area for the United States?

CREATE VIEW buildings_view AS (
    SELECT
        *
    FROM
        read_parquet('s3://overturemaps-us-west-2/release/2024-02-15-alpha.0/theme=buildings/type=*/*', filename=true, hive_partitioning=1)
);

SELECT 
    admins.id, 
    admins.subType, 
    admins.isoCountryCodeAlpha2,
    admins.names.primary,
    sources[1].dataset AS primary_source,
    --areas.areaId,
    buildings.id,
    buildings.names.primary AS building_name,
    ST_GeomFromWKB(areas.areaGeometry) as geometry
FROM admins_view as admins
    
INNER JOIN (
    buildings_view as buildings ON ST_Intersects(ST_GeomFromWKB(areas.areaGeometry), ST_GeomFromWKB(buildings.geometry))

WHERE
    admins.adminLevel = 1
    AND admins.names.primary LIKE 'United States'

LIMIT 10;

Raw

locality-type-count.md

Get a list and count of each locality type in the Overture admins dataset.

SELECT 
    adminLevel, localityType, COUNT(1) n
FROM 
    read_parquet('s3://overturemaps-us-west-2/release/2024-02-15-alpha.0/theme=admins/type=*/*', filename=true, hive_partitioning=1)
WHERE 
    localityType IS NOT NULL
GROUP BY adminLevel, localityType
ORDER BY n DESC;

See the output table below. Why are so many locality types missing an admin level? Also seems like there are lots of incorrect assignments. Create task to investigate.

adminLevel	localityType	n
	hamlet	1634526
	village	1423009
	neighborhood	790512
	suburb	153593
	town	110569
3	county	34763
	city	15825
4	region	8255
4	village	6338
3	town	4618
	borough	4089
2	state	3357
	municipality	2455
4	district	1652
3	city	1144
3	municipality	1038
4	county	1021
4	town	733
	county	731
4	city	444
	district	414
3	district	412
3	region	373
1	country	281
3	province	201
2	province	143
2	county	142
3	state	87
3	borough	67
2	region	61
2	district	56
2	municipality	25
4	suburb	12
4	hamlet	6
4	municipality	3
	state	2
	region	2
4	neighborhood	2

Raw

overture-admins.md

Find the features at adminlevel = 2 (state level) that match the name Pennsylvania. Grab the GERS ID for PA.

SELECT id, isoSubCountryCode, contextId, names.primary
  FROM 
    read_parquet('s3://overturemaps-us-west-2/release/2024-02-15-alpha.0/theme=admins/type=*/*', filename=true, hive_partitioning=1)
   WHERE type='locality'
   AND subType='administrativeLocality'
   AND adminLevel=2
   AND names.primary LIKE 'Pennsylvania';

Now we have the GERS ID for Pennsylvania and its context ID (aka its parent ID), which is also the GERS ID for the United States.

id	isoSubCountryCode	contextId	primary
varchar	varchar	varchar	varchar
0857b2b73fffffff01914c596abbfacd	US-PA	0850c861bfffffff01aeb407d56d3441	Pennsylvania

Get all the children associated with Pennsylvania's GERS ID.

CREATE VIEW admins_view AS (
    SELECT
        *
    FROM
        read_parquet('s3://overturemaps-us-west-2/release/2024-02-15-alpha.0/theme=admins/type=*/*', filename=true, hive_partitioning=1)
);

COPY (
  WITH RECURSIVE hierarchy(id, adminLevel, localityType, contextId, names) AS (
      SELECT id, adminLevel, localityType, contextId, names.primary as names
      FROM 
        admins_view
      WHERE 
        type = 'locality'
      AND id = '0857b2b73fffffff01914c596abbfacd'
     
 
      UNION ALL
 
      SELECT admins.id, admins.adminLevel, admins.localityType, admins.contextId, admins.names.primary as names
      FROM admins_view AS admins
      JOIN hierarchy
      ON admins.contextId = hierarchy.id
      WHERE 
        type = 'locality'
      )
  
  SELECT *
  FROM hierarchy
  ORDER BY adminLevel, localityType, id
) TO 'beautiful_pennsylvania.geojson'
WITH (FORMAT GDAL, DRIVER 'GeoJSON');

From that resulting GeoJSON, I pulled out the GERS ID for Fishtown, a neighborhood in Philadelphia: 08522a98bfffffff01ebb35b247d2da3. Let's walk it up to the root.

   CREATE VIEW admins_view AS (
     SELECT
        *
     FROM
        read_parquet('s3://overturemaps-us-west-2/release/2024-02-15-alpha.0/theme=admins/type=*/*', filename=true, hive_partitioning=1)
);

COPY (
   WITH RECURSIVE hierarchy(id, adminLevel, localityType,
                         contextId, names, geometry,
                         isoCountryCodeAlpha2, isoSubCountryCode, sources
   ) AS (
  SELECT id, adminLevel, localityType, contextId, names, geometry, isoCountryCodeAlpha2, isoSubCountryCode, sources
    FROM admins_view
   WHERE 
     type = 'locality'
     AND id = '08522a98bfffffff01ebb35b247d2da3'
 
   UNION ALL
 
  SELECT admins.id, admins.adminLevel, admins.localityType,
         admins.contextId, admins.names, admins.geometry,
         admins.isoCountryCodeAlpha2, admins.isoSubCountryCode, admins.sources
    FROM admins_view AS admins
    JOIN hierarchy
      ON admins.id = hierarchy.contextId
   WHERE 
     type = 'locality'
)
  SELECT 
         id, 
         adminLevel, 
         localityType,
         names.primary AS name,
         isoCountryCodeAlpha2 AS country, 
         isoSubCountryCode AS subCountry,
         ST_GeomFromWKB(geometry) as geometry,
         contextId AS parentId,
         sources[1].dataset AS source
    FROM hierarchy
ORDER BY adminLevel, localityType, id
) TO 'fishtown-to-us.geojson'
WITH (FORMAT GDAL, DRIVER 'GeoJSON');

Here's a table of each administrative unit from neighborhood to country, with attributes. The GeoJSON file contains the geometries.

gers_id	locality_type	name	parent_id	source
0850c861bfffffff01aeb407d56d3441	country	United States		OpenStreetMap
0857b2b73fffffff01914c596abbfacd	state	Pennsylvania	0850c861bfffffff01aeb407d56d3441	OpenStreetMap
08522114ffffffff01a739d1d7602762	county	Philadelphia County	0857b2b73fffffff01914c596abbfacd	OpenStreetMap
08522f093fffffff013db4adcc60888e	city	Philadelphia	08522114ffffffff01a739d1d7602762	OpenStreetMap
08522a98bfffffff01ebb35b247d2da3	neighborhood	Fishtown	08522f093fffffff013db4adcc60888e	OpenStreetMap

danabauer/boundaries-buildings.md