Scores API
The scores of the Scores page are served as files, for every published version, region and track. On the page, use the Export row next to the Region and Track pickers to download the displayed scores, or to open them in the EDITO Data Explorer.
Endpoints
Replace <version>, <region> and <track> with an OceanBench version (for example 0.6.0), a region identifier (for example global) and a track (high_resolution or one_degree).
| Endpoint | Content |
|---|---|
/api/scores/index.json |
The available versions, regions and tracks, with the path of each file |
/api/scores/<version>/<region>/<track>.csv |
Scores as CSV |
/api/scores/<version>/<region>/<track>.parquet |
Scores as Parquet |
/api/scores/<version>/<region>/<track>.zarr |
Scores as a multi-dimensional Zarr store |
Only the combinations listed in index.json exist.
Columns of the CSV and Parquet files
Each row is the score of one challenger, for one metric, depth, variable and lead day.
| Column | Description |
|---|---|
version |
OceanBench version |
region |
Region identifier |
track |
high_resolution or one_degree |
challenger |
Challenger identifier |
comparison |
observations, reanalysis or analysis |
metric |
Metric title |
depth |
Depth range, empty for metrics without depth |
variable |
Variable name |
standard_name |
CF standard name of the variable |
unit |
Unit of the score |
lead_day |
Forecast lead day |
value |
Score, empty when unavailable |
Zarr structure
The Zarr store is a dataset with the dimensions challenger, comparison, metric, depth, variable and lead_day. The value array holds the scores, empty where a combination does not exist. The unit and standard_name arrays give the unit and CF standard name of each comparison, metric, depth and variable.
Examples
curl -O https://oceanbench.edito.eu/api/scores/index.json
curl -O https://oceanbench.edito.eu/api/scores/<version>/global/high_resolution.csvimport pandas
import xarray
scores = pandas.read_parquet("https://oceanbench.edito.eu/api/scores/<version>/global/high_resolution.parquet")
dataset = xarray.open_zarr("https://oceanbench.edito.eu/api/scores/<version>/global/high_resolution.zarr", consolidated=True)
glonet_scores = dataset["value"].sel(challenger="glonet", comparison="reanalysis", metric="Forecasted variables")The Export popup on the Scores page shows these snippets for the displayed version, region and track.
To open the scores in the EDITO Data Explorer, pass the CSV or Parquet URL as the source parameter:
https://datalab.dive.edito.eu/data-explorer?source=<url-encoded CSV or Parquet URL>