The output files¶
Two trees under the output path (Output by default):
Output/
├── nc/AIS/VUW/PISM1/CORE/C007/ one NetCDF file per scalar; mirrors the model tree
└── csv/ one CSV per run with all enabled SLC rows; flat
The NetCDF tree mirrors the model tree so a submission’s output sits where its input does. The CSVs are flat because they are meant to be joined into one community-wide table.
Filenames¶
A NetCDF file’s name is the scalar, then the same nine fields as the model files it came from, then the years actually written:
slvaf_AIS_VUW_PISM1_m001_CESM2-WACCM_f001_ssp585_C007_1850-2300.nc
The year range depends on --histout; here all of the historical run was
prepended.
The SLC CSV is named sl_ followed by those run fields and years, then the
mask selection it holds. It contains all enabled SLC methods, GIC variants, and
selected masks for that run:
Mask selection |
CSV name |
|---|---|
default |
sl_AIS_…_mm.csv |
|
sl_AIS_…_mm-basins.csv |
|
sl_AIS_…_basins.csv |
The selection is part of the name so that two runs of the same experiment with different masks do not overwrite each other.
With --basins the mask name is added after the scalar in NetCDF filenames,
and the whole-sheet file gains one too so that every file in the directory is
named the same way:
Whole sheet |
Basin r01 |
|
|---|---|---|
default |
slvaf_AIS_… |
not written |
|
slvaf_ais_AIS_… |
slvaf_r01_AIS_… |
|
not written |
slvaf_r01_AIS_… |
A -gic suffix on the scalar marks the variant with glaciers and ice caps left out of the integral: slvaf-gic_r01_AIS_… .
Sea-level contribution¶
Three scalars, in metres of global mean sea level, positive for sea-level rise. Every series is zero at the reference timestep.
Scalar |
Method |
|---|---|
slvaf |
volume above flotation |
slg20 |
Goelzer et al. (2020) |
sla20 |
Adhikari et al. (2020) |
--base-variants adds two more, slvafbase and slg20base, written the same
way. The sea-level methods describes them all. Each is computed twice, with and without
glaciers and ice caps, and written like this:
NetCDF |
CSV |
|
|---|---|---|
including glaciers and ice caps |
yes |
yes |
excluding them (-gic) |
yes |
--csv and --no-csv override CSV output for both SLC variants. --netcdf
and --no-netcdf override NetCDF output for all scalar classes; explicitly
enabling NetCDF also enables the GIC-masked SLC NetCDF. When omitted, each
format keeps the defaults shown above and in the state/flux sections below.
Disabling NetCDF skips the state and flux scalars entirely, because they have
no CSV form.
The CSV form¶
One header row and one data row per SLC method, GIC mode, and selected mask. The first ten columns identify the run and scalar:
Column |
Holds |
|---|---|
ice_source |
AIS or GrIS |
region |
the geographic mask name (ais, gris or a basin) |
scalar |
|
group, model, model_variant, scenario, GCM, forcingid, configid |
the run’s fields; model_variant is the ISM member ID, scenario the experiment |
Then one column per nominal year, y1850 to y2300. Years the run does not cover
are NA. A run that reaches outside that window has those years dropped, with a
warning naming them. With AIS --basins, the CSV has 132 data rows when the
whole-sheet mask is included, or 126 with --no-mm.
State scalars¶
One NetCDF file each, integrated over the mask:
Scalar |
What it is |
Units |
|---|---|---|
lim |
ice mass |
kg |
limnsw |
ice mass not displacing sea water |
kg |
iareagr |
grounded ice area |
m² |
iareafl |
floating ice area |
m² |
These need sftgrf and sftflf in the model output. Without them the four are skipped with a warning and the rest of the run still succeeds.
Flux scalars¶
The gridded mass fluxes integrated over the mask, one NetCDF file each, all in kg/s:
Scalar |
From |
Flux |
|---|---|---|
tendacabf |
acabf |
surface mass balance |
tendlibmassbfgr |
libmassbfgr |
basal mass balance, grounded |
tendlibmassbffl |
libmassbffl |
basal mass balance, floating |
tendlicalvf |
licalvf |
calving |
tendlifmassbf |
lifmassbf |
ice front melting |
tendligroundf |
ligroundf |
grounding line migration |
Each is skipped on its own if its input file is absent, and the run lists the ones it skipped at the end. Flux files have their own time axis (Jul 1 of each year rather than Jan 1 of the next), so their year range can differ from the other files of the same run.
The flux files also give the period each value covers, in a time_bounds
variable copied from the model’s flux file (and from the historical run’s
file, for the historical years). If either file has no time bounds, the run
prints a warning and writes the scalar file without them, and the compliance
checker will reject it. The state and sea-level files have no time bounds,
because their model files have none.
Attributes¶
Every NetCDF file holds a time coordinate and one variable, both single
precision (f4) with the netCDF default fill value recorded as _FillValue,
on an unlimited time dimension. The time coordinate copies its units,
calendar and long name from the model file. The state and flux scalars
carry the long name, units and standard name given in the ISMIP7 data
request, read from the isschecker package (see Where the conventions come from); the
data request gives no standard name for tendlifmassbf and tendligroundf, so
those two carry none. The sea-level scalars are this package’s own and have
no standard name either.
Each file also copies five global attributes from the model file its scalar
is computed from: group, model, contact_name, contact_email and
crs. The state and sea-level scalars take them from lithk, the grounded and
floating ice areas from sftgrf and sftflf, and each flux scalar from its flux
file. The compliance checker requires all five. If a model file is missing
one, the run prints a warning and the scalar file is missing it too.