WIP Notes on STAC Collection temporal extent and datetime summaries

[anayeaye]

What

This WIP tracks how to update the temporal extent and collection summaries for collections of items that use start/end datetimes. This provides guidance for manually updating collections and includes some notes about what is currently automated across the board--some implementations may already correct this, but here are methods for manually fixing a single collection.

Currently the functions built in to the veda-stac-ingestor might use the nominal datetime instant which does not capture items that represent a date range. The latest pgstac versions will update collection extents when a trigger is enabled but we need to confirm that the full datetime range is included.

Collection Temporal Extent

For now, a function that generates the temporal extent using the max end_datetime

-- make a re-usable function for the collection max temporal extent
CREATE OR REPLACE FUNCTION pgstac.collection_max_temporal_extent(id text)
 RETURNS jsonb
 LANGUAGE sql
 IMMUTABLE PARALLEL SAFE
 SET search_path TO 'pgstac', 'public'
 SET timezone to 'UTC'
AS $function$
    SELECT to_jsonb(
	            array[array[
			        to_char(min(items.datetime) at time zone 'Z', 'YYYY-MM-DD"T"HH24:MI:SS"Z"'),
            			to_char(max(items.end_datetime) at time zone 'Z', 'YYYY-MM-DD"T"HH24:MI:SS"Z"')
		            ]]
	            )
    FROM items WHERE collection=$1;;
$function$;

Now the above function can be used to set a collections temporal extent

-- update collection extents
SET timezone to 'UTC';
UPDATE collections SET
    "content" = "content" ||
    jsonb_build_object(
        'extent', jsonb_build_object(
            'spatial', jsonb_build_object(
                'bbox', collection_bbox(collections.id)
            ),
            'temporal', jsonb_build_object(
	            'interval', collection_max_temporal_extent(collections.id)
        	)	
    	)
   )
WHERE collections.id=:_collection_id;

Collection datetime summaries

We have already corrected the collection summaries here.
This function will eventually be updated to remove the cog_default raster logic, but today it works as designed for datetimes.

For dashboard:is_periodic collections the method is currently designed to return the full range of dates covered by the collection. For non-periodic collections every distinct nominal datetime instant will be returned. The purpose is to provide the minimum amount of information needed for a time selector in a dashboard explore view.

Update the summaries for a single collection

SELECT update_collection_default_summaries(collections.id)
FROM collections
WHERE collections.id = :_collection_id;

[anayeaye]

#344