The singlecolumn model is in src_sam


Step-by-step technical guide


A single parameter setting is used to run 1D simulations of three cases. Each simulation advances height profiles in time. The result depends on the specific setting. This document describes how to run these cases for one specific setting and how to get the rmse of that setting.

I'm running everything on Lawrencium or Edison. Instructions are for Edison specifically: 

Get the code
git clone -b master git@bitbucket.org:wlanghans/singlecolumn.git singlecolumn
This creates the directory singlecolumn
compile the src code of the singlecolumn model
cd singlecolumn/src_sam
see a list of my loaded modules below; on Edison load cray-netcdf and ncl
cp Makefile_edison Makefile
make clean;make
this should generate (after about 3 min) the executable sc_sam_wl in directory singlecolumn
Go to src_single and modify the following things in runsingle_juli
rundir: change to your directory, eg on scratch file system
srcdir: change to your directory that holds the executable sc_sam_wl
ncldir: change to .../singlecolumn/src_ensemble
Run the simulations (1 case)
./runsingle_juli "runname" 0.0 0.0 0.0 0.4 500. 0.6 0
This runs the BOMEX and CPBL2 case with a decent parameter setting
first argument is the name that identifies the specific parameter setting
second to seventh argument identifies the six parameters (you can find min/maxvalues for each parameter in runsingle_juli (look for pmin and pmax vectors) )
8th argument: 0: delete netcdf output from simulations after rmse has been evaluated; 1: keep netcdf files
This will run the BOMEX and CPBL2 case (both described below) and will compute rmse for each case by comparing against a LES benchmark. An overall rmse is computed as the average of the two.
Find the overall rmse in text file $rundir/rmse_tot_thvd, and the rmse of each individual case in each case directory: the files are named rmse_thvd_BOMEX and rmse_thvd_CPBL2. In each of these files, the rmse error will be the first number in the text file and is written with a "13.8f" (8 digits after the decimal point).


The definition of the rmse is described below for each case. The overall rmse is a linear combination of those.