Examples
Worked reads of published datasets
Each example is one query against one published dataset. Every reader that answers it has its own tab, and the page records who ran each reader, where, when, and what came back, so equal answers across readers are on record.
Adding an example
Copy docs/examples/_template/ to docs/examples/<slug>/ and fill in:
- front matter:
title,description,date(last run),categories(dataset family, forms, languages); - the Query block: dataset, selection, and the expected result (shape, first and last values);
- one tab per reader: code that runs as pasted, plus a status line
[ok],[no]or[not run]with who, where and when; - the Runs table.
Code on these pages is not executed when the site is built (no data access in CI); the values shown are pasted from real runs.
| Last run | Title | Description | Categories |
|---|---|---|---|
| Nov 2, 2026 | Bluelink salinity: point, profile, 30-day series | Icechunk with request concurrency capped, and the mdim VRT through gdalxarray, south of Tasmania. | bluelink, icechunk, mdim-vrt, python |
| Nov 2, 2026 | Bluelink temperature: chunk index over time and depth, in DuckDB | A table query, not an array read: scan the flat refs store with DuckDB and summarise chunks, bytes and file spans for each (day, depth) pair. | bluelink, flat, duckdb, sql, python, r |
| Nov 2, 2026 | Bluelink temperature: one latitude line, last day | Same values from the mdim VRT and kerchunk via gdalxarray and from the VRT via terra; GDAL rejects the Icechunk form. | bluelink, mdim-vrt, kerchunk, icechunk, python, gdal, r, terra |
No matching items