Ocean acidification indicators, North American margins — Icechunk

🌐 View data in browser · 💻 Data access (code) · 📦 Data access (NCEI) · 📄 DOI 10.25921/g8pb-zy76

An Icechunk store of the NCEI coastal climatology of ocean acidification indicators on the North American ocean margins — twelve indicators at 14 standard depth levels from the surface to 500 m, on a 1° × 1° grid, gridded from observations collected 2003-12-06 to 2018-11-22 and adjusted to the year 2010.

https://data.source.coop/ocean-icechunks/oa-indicators/climatology

This store is virtual: it holds Zarr metadata and byte-range references, and no array data at all. The whole store is 84 objects and 35 kB, pointing at 82 MB of NetCDF that stays at NCEI. Every data read goes back to www.ncei.noaa.gov, so the store is only as available as NCEI is.

What it adds over the source archive is that the twelve files become one dataset. NCEI ships one NetCDF per indicator with no usable coordinates; here they are merged into 72 variables on a shared, CF-compliant (depth, lat, lon) grid you can sel() into.

The source accession is a finished, static product, so this store needs no update pipeline.

View it in a browser

No install, no account, no download — the viewer streams chunks straight from the store, and the science arrays straight from NCEI:

Those are the _an (objectively analysed mean) fields; the viewer's own dropdown reaches all 72 variables, and the Depth slider moves through the 14 levels.

The links carry an opening view — centred on the middle of the grid, at the surface — and the viewer's dataset picker offers this store rather than gridlook's 70 demo datasets, because the published copy ships its own static/catalog-extended.json.

The viewer is gridlook, a WebGL globe for cloud-hosted Zarr and Icechunk stores, published alongside the data at oa-indicators/viewer/. The store to open is in the URL fragment after #, so any store can be swapped into the same link — nothing about the viewer is specific to this dataset.

Unlike the CoastWatch stores in this repository, this one draws in an ordinary browser. A virtual store needs CORS on two hosts, because metadata and data come from different places. Source Cooperative is wide open, and www.ncei.noaa.gov also sends Access-Control-Allow-Origin: * on ranged GETs — where coastwatch.noaa.gov sends no CORS headers at all and is therefore blocked. A Range: bytes=a-b header is CORS-safelisted, so no preflight is involved.

How to open it

Needs icechunk >= 2.2 and xarray. No credentials and no account.

icechunk 1.x will not work. icechunk.http_storage does not exist in icechunk 1.x (it arrived in 2.0), and icechunk.credentials.HttpAccess arrived in 2.1. Every icechunk 2.x release needs Python 3.12 or newer, so on an older Python pip install icechunk quietly installs 1.1.x, and the code below fails with AttributeError: module 'icechunk' has no attribute 'http_storage'. Check what you have: python -c "import icechunk; print(icechunk.__version__)".

Measured anonymously over the public endpoint: opening the store and reading its metadata takes about 1.5 s, and any single field about 1 s.

Use chunks=None, not chunks={}, unless you want dask. Each variable is a single chunk covering the whole array, because the source arrays are contiguous and uncompressed. There is nothing to parallelise within a variable, and any read of any part of a variable fetches all 1.2 MB of it.

Land and unsampled cells are NaN. The source uses NaN directly and declares no _FillValue or missing_value, so nothing has to be unmasked.

About the data

These are World Ocean Atlas–style objective analyses of carbonate-system observations from CODAP-NA (Version 2021) and GLODAPv2.2021, adjusted to the year 2010 with ESPER (Carter et al. 2021) before gridding.

• Grid: 1° latitude × 1° longitude, 14 depth levels — 0, 10, 20, 30, 50, 75, 100, 125, 150, 200, 250, 300, 400, 500 m
• Extent: lat 10.5 → 85.5 °N, lon −179.5 → −39.5 °E. Coverage is coastal, not global: about 63 % of the surface grid has data
• No time dimension. It is a climatology; the contributing observations span 2003-12-06 to 2018-11-22
• Conventions: CF-1.12, ACDD-1.3 (added here — see "What was changed from the source")
• Archive: NCEI accession 0270962, DOI 10.25921/g8pb-zy76

Variables

Twelve indicators × six fields = 72 variables, all float64 on (depth, lat, lon).

A dash means the CF standard name table has no entry for that quantity — it was checked, and none was invented.

Each prefix carries six fields:

So OmegaA_an is the analysed aragonite saturation state, and OmegaA_dd the number of observations behind it — useful for deciding where the analysis is actually constrained.

How this was built

The notebook that built it sits next to this README: oa-indicators-icechunk-sc.ipynb (also in the GitHub repository https://github.com/ocean-icechunks/icechunks, directory oa-indicators/). It opens each of the twelve source files with VirtualiZarr, merges them, rewrites the metadata, writes the store, and then validates the published store from the anonymous read path — including reading all 72 arrays back and comparing them element by element against the source files at NCEI.

It runs end to end without credentials and without writing anything: RUN_WRITE defaults to False, so it builds the virtual dataset and the metadata, skips the write, and validates the live store. The committed outputs are from exactly that run. requirements.txt beside it lists the package floors; Python 3.12 or newer is required, because icechunk 2.x is published requires_python = ">=3.12".

The source is twelve NetCDF-4 files, one per indicator, at

https://www.ncei.noaa.gov/data/oceans/ncei/ocads/data/0270962/

NCEI serves them over HTTPS with Accept-Ranges: bytes, which is what makes virtualizing possible, and — unlike NOAA CoastWatch — with Access-Control-Allow-Origin: *.

What was changed from the source

Metadata only. No data value was altered — which is the only kind of change a virtual store can make, since the bytes stay in NCEI's files.

• The coordinates were unusable as delivered. The files carry phony HDF5 dimension scales dep, lat and lon — arrays of zeros marked "This is a netCDF dimension but not a netCDF variable" — while the real values sit in three separate variables depth, latitude and longitude. xarray hides the phony scales and reports Dimensions without coordinates, so sel(lat=...) did not work on the source at all. The real variables are promoted to index coordinates named depth, lat and lon.
• units was not UDUNITS. Every dimensionless field said "N/A", and temperature said "degrees Celsius" — with a space, which UDUNITS will not parse. They are now "1" and "degree_Celsius".
• standard_name held free text, not CF standard names — "Aragonite saturation state", "Number_of_observations for pH on Total Scale". That text moved to long_name (and the source's own wrapped long_name was kept verbatim as comment), and a real standard_name was supplied only where the CF table has one.
• There were no global attributes beyond a per-file title, abstract and citation. The store has a full CF/ACDD set: Conventions, title, summary, keywords, source, history, creator and publisher fields, the DOI, the citation, the geospatial bounds and the observation time coverage.
• Coordinates gained standard_name, units, axis and — for depth — positive: down.
• Each variable gained a source attribute naming the NetCDF file it references.

The viewer is published by publish_viewer.py in the same repository (python publish_viewer.py --product oa-indicators --build ~/gridlook); it is a plain static build of gridlook and holds no data of its own.

Provenance

Every code block on this page was executed against the live store before it was published.

Reuse and citation

Code. The notebook and helpers are released under Apache-2.0 and are free to use, copy, adapt and redistribute, commercially or not — no attribution required.

Data. The data is not ours, and this store contains none of it — only references to NCEI's files. NCEI states no use restrictions for accession 0270962. Please cite it:

Jiang, Li-Qing; Boyer, Tim P.; Paver, Christopher R.; Reagan, James R.; Alin, Simone R.; Barbero, Leticia; Carter, Brendan R.; Feely, Richard A.; Wanninkhof, Rik (2022). Climatological distribution of ocean acidification indicators from surface to 500 meters water depth on the North American ocean margins from 2003-12-06 to 2018-11-22 (NCEI Accession 0270962). NOAA National Centers for Environmental Information. Dataset. https://doi.org/10.25921/g8pb-zy76

and, for the observations the analysis is built on:

Jiang, L.-Q., Feely, R. A., Wanninkhof, R., Greeley, D., Barbero, L., Alin, S., et al. (2021). Coastal Ocean Data Analysis Product in North America (CODAP-NA) — an internally consistent data product for discrete inorganic carbon, oxygen, and nutrients on the North American ocean margins. Earth System Science Data, 13(6), 2777–2799. https://doi.org/10.5194/essd-13-2777-2021

Credits

• Data: Li-Qing Jiang, Tim P. Boyer, Christopher R. Paver, Hyelim Yoo, James R. Reagan, Simone R. Alin, Leticia Barbero, Brendan R. Carter, Richard A. Feely and Rik Wanninkhof — CISESS/University of Maryland, NOAA NCEI, NOAA PMEL, NOAA AOML, CIMAS and CICOES. Funded by the NOAA Ocean Acidification Program and archived by NOAA NCEI.
• Icechunk packaging: built with Icechunk, VirtualiZarr and Xarray, hosted on Source Cooperative.