Quickstart¶
Dependencies¶
Before starting, make sure an appropriate fortran compiler is installed. The code has been tested and developed mainly with gfortran.
In addition, make sure the netcdf-fortran library is installed:
sudo apt install netcdf-bin libnetcdf-dev libnetcdff-dev
The VOM will by default search for the netcdf installation in usr/include.
Compiling and testing the model¶
To compile the model on a unix machine with the gfortran compiler, type :
make
This will create an executable file called “model.x”. Now test the model by typing:
make check
When the test is passed, the model is ready to use. The VOM_namelist contains all settings, and the model should always be run with the directory of this namelist as working directory. Input and output directories can be defined in the namelist with absolute or relative paths from the working directory.
Running the model¶
The executable “model.x” can be run :
./model.x
To run successfully, these input-files are needed:
- vom_namelist
- Contains all settings to run the VOM.
- pars.txt
- Contains the (optimized) vegetation parameters (only needed for single run).
- dailyweather.prn
- Contains the meteorological forcing.
By default, the executable looks for the vom_namelist in the current workdirectory. The default directory for the other files is /input, relative to the workdirectory. This can be changed in vom_namelist, or on the command line:
-i Inputpath to directory with dailyweather.prn, and optionally pars.txt.
-o Outputpath for all outputfiles.
-n The VOM_namelist (filename can be different)
To run successfully, these input-files are needed:
- vom_namelist
- Contains all settings to run the VOM.
- pars.txt
- Contains the (optimized) vegetation parameters (only needed for single run).
- dailyweather.prn
- Contains the meteorological forcing.
By default, the executable looks for the vom_namelist in the current workdirectory. The default directory for the other files is /input, relative to the workdirectory. This can be changed in vom_namelist, or on the command line:
-i Inputpath to directory with dailyweather.prn, and optionally pars.txt.
-o Outputpath for all outputfiles.
-n The VOM_namelist (filename can be different)
Scripts for running the model¶
In the folder VOM-scripts, two shell-scripts can be found for running the VOM:
vom_parallel.job
run_vom.sh
The model is compiled and executed by the script run_vom.sh. It can be used in the following way:
bash src_sh/run_vom.sh src/VOM/VOM_Fortran/ <inputdir> <dailyweather.prn> <soilprofile.par> <vom_namelist> <outputdir>
vom_parallel.job is a job script for the LIST HPC using SLURM. Note that the script needs to be modified for own applications of the VOM on an HPC and serves here as an example. In this example, submitting a job with SLURM functions as follows:
sbatch -o <output_logfile> -e <error_logfile> -n <number of cpu's> -J <jobname> vom_parallel.job <VOM-src> <input-directory> <dailyweather.prn> <soilprofile.par> <vom_namelist> <output-directory>
Model modes¶
The model can be run in 4 different modes, defined by VOM_command in the VOM_namelist:
1 Optimize the model with the Shuffled Complex Evolution algorithm.
2 Run without optimization, based on the parameters in pars.txt.
3 Run without optimization, based on the parameters in pars.txt. Returns only NCP values as output.
4 Run the model with a set of random parameters.
Other options¶
Initially, the VOM schematized the vegetation as two big leaves, for the perennial and seasonal vegetation. In the newest VOM version, leaf area dynamics can be included in a dynamic way as well, by setting the parameter i_lai_function in the VOM_namelist:
1 No LAI dynamics are included.
2 LAI is dynamically modelled, but there is no distinction between shaded and sunlit leaves, as well as different radiation components.
3 LAI is dynamically modelled. Shaded and sunlit fractions are determined and a distinction between direct and diffuse radiation is made, but leaves are still treated as a big leave with a single photosynthetic capacity.
4 LAI is dynamically modelled. Shaded and sunlit fractions are determined and a distinction between direct and diffuse radiation is made. The model considers shaded and sunlit leaves with different photosynthetic capacities. Eletron transport capacities are separately optimized for shaded and sunlit leaves.