First try compiling the contour advection code. In directory cas
, run gfortran -o cas cas.f uvw.f
. Now see if it will run: try ./cas < run_params_will.in
. This should produce a file output.cas
-
cas/
: code for running contour advection (CA)cas_output/
: output from CA calculations goes here (contents is ignored by git)input contours/
: initial contours for CA calculations, calculated bypython/make_contour.py
winds/
: input winds for CA scriptcas
: this is the executable for the CA calculation, which can be compiled using gfortran with the following commandgfortran -o cas cas.f uvw.f
. CA code is comprised of the following (seecas/README
for more details):cas.f
: bulk of the CA codeuvw.f
: subroutines to initialize wind grids and to read in wind datacas.h
: "include" common blocks
output.cas
: 'default' output file for CA output, then copied to a more descriptive filename incas_output/
output.log
: logfile for CA outputpv.in
: 'default' input contour file (copied over frominput_contours/
)run_multi_contour.py
: This script runs the CA code over a series of contours frominput_contours/
, and using winds from../model_output/
, then puts the output incas_output/
. It is currently a python script but could be easily changed to a shell script.run_params_will.in
: input for thecas
executable, which is run with the following command:./cas < run_params_will.in
. The only thing you need to change here is the number on the final line which is the start day for the calculation.
-
model_output/
: Binary model output, netCDF files made from model output, and wind files for input into CA code. Too large for github, but can dowload here: https://www.dropbox.com/s/1lhhv52l1cc56z6/model_output.zip?dl=0 (about 2GB). -
plots/
: currently shows results of CA calculations for two runs with same wave 2 topography but one with zero relaxation and one with relaxation giving an annular vortex. -
python/
: python scripts for making inputs for CA code and plotting outputsinterpolate_winds.py
: produces the wind input files for the CA code (found in../model_output/winds_*
)make_contour.py
: makes input contours for CA code. Note the length of the CA calculation is determined by the number in the contour file header (currently set to 50 time steps (i.e. 25 days)).make_netCDF.py
: produces a netCDF file of q, u, and v from swbob binary outputplot_cas_multicontour.py
: plots results of CA code (as in../plots/
directory)plot_pv.py
: produces quick plots of PV for checking output of swbob runsread_BOB.py
: reads swbob binary output and converts to a (lat,lon) array
-
swbob/
: code for running the shallow water model. Can add more info here if needed.
A typical filename from swbob output (and other files derived from this output) might look like the following:
res_test57.-70.-nu4-urlx-kt0.0.c-0020sat200.0.T85
Here's what it all means:
res_test
: name of the type of run. I called it this when I was testing different resolutions, should probably update to something more descriptive.57.-70.
: inner and outer latitudes of the initial annular vortexnu4-urlx
: doesn't change among runskt0.0
: relaxation parameter. In this case, 0 relaxation. A relaxation parameter 4.0 would mean a timescale of 0.25 days (i.e. kt = 1/t_relax).c-0020
: angular frequency of the tropographysat200.0
: maximum topography amplitudeT85
: spectral resolution