The rasters/ directory is intentionally at the repository root. It contains
the aligned uint8 GeoTIFF products used by the pixel-space dataloader. The
compact argo/argo_profiles_on_grid.zarr store is the grid-indexed ARGO input
used by that dataloader.
The package intentionally contains two ARGO Zarr stores with different roles.
argo/argo_profiles_on_grid.zarr is the compact grid-indexed store meant to be
used together with the GeoTIFF raster dataset. data/argo_glors_ostia_ssh.zarr
is the full enriched profile-level store and holds the complete ARGO
collocation dataset, including the sampled GLORYS, OSTIA, sea-level, and
sea-surface-salinity context.
Raster Products
All GeoTIFF rasters are exported on the GLORYS 0.1 degree global grid
(EPSG:4326, 3600 x 1800 pixels, west-to-east longitudes from -180 to 180 and
north-to-south latitudes from 90 to -90). The current package contains 761
weekly target dates per raster product, from 2010-01-01 through 2024-07-26.
Files are named <variable>_YYYYMMDD.tif.
The GLORYS variables are depth-resolved 50-band GeoTIFFs:
rasters/glorys/thetao/: potential temperature, encoded as Kelvin.
rasters/glorys/so/: salinity, encoded as PSU.
The surface products are single-band GeoTIFFs aggregated to the same weekly
target dates with a centered 7-day mean window:
rasters/ostia/analysed_sst/: OSTIA analysed sea-surface temperature in Kelvin.
rasters/sealevel/adt/: absolute dynamic topography in meters.
rasters/sss/sos/: sea-surface salinity in PSU.
rasters/sss/dos/: sea-surface density in kg/m3.
Raster pixels are stored as uint8 with 255 reserved for nodata. Valid codes
0..254 are linearly decoded using the stretch ranges in manifest.yaml;
per-file statistics, source filenames, compression, target dates, and the full
depth axis are also recorded there.
ARGO Alignment Examples
ARGO profiles are projected onto the fixed 50-level GLORYS depth axis before
spatial rasterization. The examples below show the grid-indexed ARGO
representation and profile-level alignment quality.
The full enriched profile-level ARGO collocation dataset is available at:
python
import xarray as xr
ds = xr.open_zarr("data/argo_glors_ostia_ssh.zarr", consolidated=None)
The lightweight Parquet indices are included for preview and filtering:
python
import pandas as pd
profiles = pd.read_parquet("indices/profiles.parquet")
variables = pd.read_parquet("indices/variables.parquet")
Coverage:
Raster target dates: 2010-01-01 to 2024-07-26
Enriched ARGO profile dates: 2010-01-01 to 2024-07-31
GLORYS depth levels: 50
Upstream product licenses and citation requirements for EN4/ARGO, GLORYS,
OSTIA, sea-level, and sea-surface-salinity products still apply.