Watching the first run¶
As soon as the run starts it creates time-dependent NetCDF files ts_g20km_10ka.nc
and
ex_g20km_10ka.nc
. The latter file, which has spatially-dependent fields at each time,
is created after the first 100 model years, a few wall clock seconds in this case. The
command -extra_file ex_g20km_10ka.nc -extra_times -10000:100:0
adds a
spatially-dependent “frame” at model times -9900, -9800, …, 0.
To look at the spatial-fields output graphically, do:
ncview ex_g20km_10ka.nc
We see that ex_g20km_10ka.nc
contains growing “movies” of the fields chosen by the
-extra_vars
option. A frame of the ice thickness field thk
is shown in
Fig. 2 (left).
The time-series file ts_g20km_10ka.nc
is also growing. It contains spatially-averaged
“scalar” diagnostics like the total ice volume or the ice-sheet-wide maximum velocity
(variable ice_volume_glacierized
and max_hor_vel
, respectively). It can be viewed by
running
ncview ts_g20km_10ka.nc
The growing time series for ice_volume_glacierized
is shown in Fig. 2
(right). Recall that our intention was to generate a minimal model of the Greenland ice
sheet in approximate steady-state with a steady (constant-in-time) climate. The measurable
steadiness of the ice_volume_glacierized
time series is a possible standard for steady
state (see [12], for exampe).
At the end of the run the output file g20km_10ka.nc
is generated.
Fig. 3 shows some fields from this file. In the next subsections we
consider their “quality” as model results. To see a report on computational performance,
we do:
ncdump -h g20km_10ka.nc | grep history
which prints
:history = "user@machine 2017-10-04 19:16:08 AKDT: PISM done. Performance stats: 0.1784 wall clock hours, 0.7136 proc.-hours, 14005.0054 model years per proc.-hour.\n",
Previous | Up | Next |