CEFI is NOAA's Changing Ecosystems and Fisheries Initiative, and the part of it this package reads is the regional MOM6-COBALT output published by the Physical Sciences Laboratory. What is built in here is the Northwest Atlantic domain (NWA12) — the US east coast, the Gulf of Mexico and the Caribbean, at a twelfth of a degree.
Value
a named list, one entry per archive, each with path, domain,
experiment, frequency, grid, release, kind, start, end,
variables, label and reference
Why reach for it
Every other model in this package is global, run once for the whole ocean and sampled wherever you happen to work. NWA12 is the opposite: a regional configuration built for this shelf, with a coupled biogeochemistry (COBALT) that carries nutrients, oxygen, carbonate chemistry and plankton biomass on the same grid as the physics. The Copernicus biogeochemical reanalysis is a quarter degree; this is a twelfth, and it resolves the Gulf of Maine and the Scotian Shelf rather than smoothing across them.
It also publishes sea-floor salinity directly as sob, where GLORYS12V1
does not — so BOTS from CEFI is the model's own diagnostic rather than the
deepest-wet-level derivation accessCopernicus() has to fall back on.
The regridded product is what is read
MOM6 runs on a curvilinear grid, and CEFI publishes both that (raw) and a
regular latitude-longitude interpolation of it (regrid). This package reads
the regridded files, because a regular grid is what everything downstream
here expects and what makes an NWA12 field overlay a Copernicus one.
That is a real choice and not a free one: the regridding is bilinear, so
values near the coast and near the ice edge have been interpolated across
cells the model treated separately. Where that matters, read the raw files
yourself rather than through this.
The velocities are the exception worth naming. CEFI publishes uo_rotate and
vo_rotate, already rotated onto true east and north, so UO and VO here
are in the same frame as every other source rather than in the model's own
grid directions.
Hindcast and forecast are different products
hindcast is a reconstruction forced by reanalysis and is the entry to reach
for. decadal_forecast is a ten-year prediction initialised each January,
with ten ensemble members, and is a fundamentally different kind of number —
see the Forecasts are experimental section of accessCEFI().
See also
accessCEFI(), cefi_archive() for any other CEFI path
Examples
names(cefi_archives())
#> [1] "NWA12" "NWA12_DECADAL"
cefi_archives()$NWA12$path
#> [1] "Projects/CEFI/regional_mom6/cefi_portal/northwest_atlantic/full_domain/hindcast"