Parquet Data Analysis

Query remote Parquet files in-browser with DuckDB WebAssembly — no server required.

Data Source
Accepts any publicly accessible Parquet URL. Use {SOURCE} in queries.
SQL Query
Use {SOURCE} as a placeholder for the Parquet URL above.
DuckDB is ready. Enter a query and click Execute.
Common functions
  • SELECT * FROM 'url' — read directly from Parquet URL
  • COUNT(*) — total record count
  • DESCRIBE SELECT * FROM 'url' — column information
  • WHERE col = value — filter records
  • ORDER BY col DESC — sort results
  • GROUP BY col — aggregate groups
  • LIMIT n — cap row count
Hive partitioning tips
  • Partition pruning: WHERE model_name = 'WRF-NARR_HIS'
  • Cross-model: GROUP BY model_name
  • URL pattern: model_name%3DWRF-NARR_HIS encodes =
Query Results
Run a query to see results here.