Skip to content

Latest commit

 

History

History
84 lines (63 loc) · 2.52 KB

cosine.md

File metadata and controls

84 lines (63 loc) · 2.52 KB

First, I synched my schism fork with newest schism, then checked if cgem-box still worked the same.

Compiling schism-cosine

Set the environment. Clone my repo, since CGEM is there, and I also have the Cmake file for cosine.
For the first compilation, you need to build makefiles with CMAKE. After that, just make. Actually, I guess you won't change the cosine code, so probably never have to care about that again.

Environment

Set the environment

module load cpu/0.15.4
module load intel/19.1.1.217
module load intel-mpi/2019.8.254
module load netcdf-c/4.7.4
module load netcdf-fortran/4.5.3
module load cmake

First compilation

Get my forked schism repo:

git clone https://github.com/oybcst/schism.git

Make a build directory, cmake, then make.

cd ~/schism
mkdir build_cosine
cd build_cosine
cmake -C ../cmake/SCHISM.cosine.build -C ../cmake/SCHISM.local.comet ../src/
make

Further compilations

After modifying, to recompile, set the environment, then:

cd ~/schism/build
make

Get the box model

I copied my cgem-box to cosine-box and cleaned out cgem related files.

Then, get sample cosine.nml file from the repo:

cp ~/schism/sample_inputs/cosine.nml .

Modify param.nml to include cosine related stuff. I cleaned my param.nml of non-GEN parameters. Here, I looked at param.in from schism_verification_tests, COSINE, SF Bay. The only thing was iof_cos.

Created init_cos.nml for initial values. Get initial values from SFBay...they are all zero! SFBay uses hotstart I think. I opened each of the variables from SFBay, plotted depth average, and tried to pick a middle ground point. (Image below.)

Modify write_initial_conditions_cosine.py. Submit job, since environment gets messed up otherwise:

sbatch submit.python.sh

Check:

rm *.bak
head -3 COS*

Get executable.

cp ~/schism/build_cosine/bin/pschism_COSINE_TVD-VL .

Tried to run, but things were missing in param.nml. I copied the one from SFBay, and figure out what to turn off so it wouldn't look for non-box stuff.

Moving output files to local computer. Better to remove individual files, so I have rm.sh, then source rm.sh. to remove junk from outputs.

Box output, quick check...

  • DSi goes negative (DN gets small, but not negative)
  • S1, S2, Z1,Z2 all go negative

We are not providing solar radiation, so it makes sense they would tank. (and the zeroes didn't seem to blow anything up.)

Initial value look sketchy? Got it from their plots... How I picked initial conditions...