The code on which this work is focused is a coupled physical-biogeochemical model
which solves both the transport equation and the biogeochemical reactions in the
seawater. Up to the current version, the only domain in which the model is supposed
to run is the Mediterranean Sea. Its boundary conditions are thought to be invariant,
both in their geometry, in their numerical schemes and in their parameterization. In
the current version, therefore, any change in the boundary conditions is possible
only for specialized developers and is increasing a lot the time to solution anytime
any new configuration is required. Referring to fig. 1.1, the most of the time in this
case is spent in adding new subroutines on anyway modifying the existing ones,
changing some hard-coded parameters, changing input files, scripts etc. and the
run-time is only a small fraction of the overall time to solution. The final goal to which this project contributes to is to unlock the HPC potential
of the model by making it adaptable to different domains and to ensemble simulations
scenarios. The first step in this direction, and actually the real aim of this thesis,
is to build a versatile interface and infrastructure to assign the boundary conditions.
Configuring the existing boundaries, but also changing the boundaries themselves,
have to be fully supported operations. In this way the model is able to be run easily
multiple times, potentially in any domain and with different boundary conditions.
Furthermore, this operations should be carried on by the user itself, without even
touching the source code, in order to dramatically reduce the time to solution and to
enable the model to be run by a wider community.
