The directory is structured as follows:

• code folder with python routines that include:
  • particle_advection.py contains the functions that execute the particle advection on a given field with given settings. This file has three subfiles: _settings.py, _kernels.py, and _start_conditions.py
    • _settings.py contains the settings for the particle advection. It ensures consistent settings for the considered experiments, and simultaneously allows for easy adaptions
    • _kernels.py contains the kernels that ensure smooth advection and tracking of variables along the particle trajectories
    • _start_conditions.py sets the initial particle positions and release time. While only from_csv() is used and shown, particle start positions can be freely defined.
  • zarr_routines.py are used to prepare the raw .zarroutput for analysis
  • metadata_production.py contains routines to extract "metadata" such as subduction time and location, as well as mapped trajectories from the .zarrfiles of the Lagrangian advection output (adapted with the help of zarr_routines.py). These data are stored in easy to handle .csvfiles to be used for the analysis.
  • plotter.py contains routines to visualise and quantify the results.
    • _plotter_helper.pycontains support functions for plotting.

Files only available in zenodo

• metadata folder contains .csv subduction files of each experiment, where time, loaction, and properties at subduction for each particle is stored. It also contains .zarr pathway files, that contain the track of each particle mapped onto a 2x2° grid
• trajectory folder contains the original trajectories (.zarr files) with a time step of 5 days. Here, only lon, lat, depth, and mixed layer depth (HMXL) are presented due to storage limitations.