gilesfearon / aquacosm1d

Aquacosm 1D ((C) 2019-2021 Francesco Paparella)

The actual code is found in aquacosm1D_lib. It is mostly a pure python code, but it links to some C code that implements diffusion. Because of this, it can't be used out of the box: one has to compile the C code into a shared library, first. On a linux box, with the gcc compiler, this is as easy as issuing the following command in the aquacosm1D_lib folder:

gcc -std=c99 -pedantic -Wall -O3 -fPIC aqc1D_diffusion.c -shared -lm -o aqc1D_diffusion.so

On other platforms (Winzzozz, Mac), I have no idea what to do. But the C code strictly adheres to the c99 standard, and it is linked to python via the ctypes interface, which is supported on all platforms, thus there's got to be a way to run the code just about anywhere python can run.

The /configs/ directory contains various configurations for running the code Once the compilation is done, go to the /configs/PAPA_example folder and run the script PAPA_simulation_p1e-7.py. That performs a 4-years simulation using the eddy diffusivity data in PAPA_6h_1Y_L75_DN.nc (it repeats 4 times the same year and will last several hours on a recent computer).

Most of the action occurs in the "wc" object, which is an instance of the "water_column_netcdf" class, defined in aquacosm1D_lib/aquacosm1D_watercolumn.py

The biology is found in aquacosm1D_lib/aquacosm1D_reactions.py. Several examples are present. It should be straightforward to construct others by analogy.

The file aquacosm1D_lib/aquacosm1D_transport.py implements the aquacosms' Brownian motion by means of a stochastic integrator. There are several integrators available. Use Milstein's. Upon testing, I found no appreciable advantages with the stiff version by Yin and Gan. The other integrators are provably inferior.

IMPORTANT NOTE: the stochastic integrators have very strong times step constraints (see e.g. https://doi.org/10.4319/lom.2004.2.289 or Visser (1997) Marine Ecology Progress Series 158, 275-281). A time step of 5 seconds should be adequate for open ocean situations. Too large a time step will not crash the code, but it will introduce subtle inhomogeneities in the distribution of the particles, thus biasing the results. If you're careless, YOU'LL SHOOT YOURSELF IN THE FOOT.

The /comms/ dir contains various communications (presentations) for facilitating collaboration