how to extract all countries geometry from Openstreet map dataset in BigQuery

Question

I am using this query to extract the geometry of all countries using OSM, it works ok, but I am sure, it is creating a duplicated as I am using flag as a reference, some places

Answer 1

According to https://wiki.openstreetmap.org/wiki/Tag:boundary%3Dadministrative#National admin_level = 2 describes countries.

So I tidied your query and added admin_level = 2 filter so that it only includes countries.

SELECT 
  feature_type, osm_id, osm_timestamp, geometry,
  (SELECT value FROM UNNEST(all_tags) WHERE key = 'flag') as flag,
  (SELECT value FROM UNNEST(all_tags) WHERE key = 'name') as name,
  st_area(geometry) as area
FROM `bigquery-public-data.geo_openstreetmap.planet_features`
WHERE 
  feature_type in ("polygon", "multipolygon")
  AND EXISTS (SELECT 1 FROM UNNEST(all_tags) WHERE key = 'boundary' AND value = 'administrative')
  AND EXISTS (SELECT 1 FROM UNNEST(all_tags) WHERE key = 'flag') 
  AND EXISTS (SELECT 1 FROM UNNEST(all_tags) WHERE key = 'admin_level' AND value = '2') 
ORDER BY area desc

For USA, I found that link that explains everything about USA.. https://wiki.openstreetmap.org/wiki/United_States/Boundaries#National_boundary

You can see openstreetmap record for the USA here https://www.openstreetmap.org/relation/148838#map=1/41/0 On the left side, there are all the features.

Also, you can find USA record in BigQuery with this:

SELECT *
FROM `bigquery-public-data.geo_openstreetmap.planet_features`
where osm_id = '148838'

Even there is admin_level in openstreetmap record, it doesn't exist in BigQuery record. I don't know why, it may just be an old version.

So you can optimize your filters using the query above to include the USA.

Answer 2

We can a list of all countries and their geometries by merging 2 tables:

SELECT features.feature_type, features.osm_id
  , ARRAY(
     SELECT DISTINCT AS STRUCT * FROM UNNEST(features.all_tags||relations.all_tags)
     WHERE key IN('int_name', 'name')
     ORDER BY 1 LIMIT 1 
  ) features
 , ROUND(ST_AREA(geometry)/1e6,1) area
FROM
  `bigquery-public-data.geo_openstreetmap.planet_features` AS features,
  `bigquery-public-data.geo_openstreetmap.planet_relations` AS relations
WHERE ('boundary','administrative') IN (SELECT (key,value) FROM UNNEST(features.all_tags))
AND ('admin_level','2') IN (SELECT (key,value) FROM UNNEST(relations.all_tags))
AND feature_type = 'multipolygon'
AND relations.id=SAFE_CAST(features.osm_id AS INT64)
ORDER BY area

(ref)