The first months have been mostly dedicated to reading articles about modelling submarine landslides and to understanding why it is not as "easy" as it seems. As we learned during the workshop in Aussois, one key issue of submarine landslides models is that the landslide is best described in "local" coordinates because it moves parallel to the slope, while the water close to the free surface is best modeled using cartesian coordinates. Most of the models I read about separate the two materials with two layers and use different coordinates in each layer. This means the change of coordinates is occuring at the interface, which is moving in time, and this can be the source of a few problems (notably that the wave generated by the landslide is not propagated in the right direction).
However, it is possible to consider a "transitional" layer of water in local coordinates, between the avalanche and the "free" water layers, so that there is a "fictive" interface between water and water, where the change of coordinates happens and that is unmoving, hopefully removing some of the problems. I used my newfound knowledge to write the model based on this idea, and recently started writing a Fortran code implementing it.