examples

[philippemiron]

Hi all,

Would it be possible to include at least ~1-2 examples where all the input datasets are publicly available?

More precisely, after installing plasticparcels, I can't reproduce any of the examples, as e.g. example_Italy_coast requires the model settings:

# Load the model settings
settings_file = 'docs/examples/example_Italy_coast_settings.json'
settings = pp.utils.load_settings(settings_file)

which are not part of the path.

Also, looking at those example configuration files in the repo, they all rely on private storage /storage/shared/oceanparcels/input_data/.

openjournals/joss-reviews#7094

[michaeldenes]

Hi @philippemiron,

You've raised a good point, to run plasticparcels requires some hydrodynamic model data, biogeochemical model data, as well as wind and wave data. The example notebooks all use this model data, hence the storage filepath /storage/shared/oceanparcels/input_data/. We'll make this more clear in the plasticparcels documentation. We will, however, provide an example using an idealised fields that can be run independent of the required datasets.

In essence, plasticparcels can be run on any dataset that parcels can be run on, however we have configured it for Copernicus Marine Service data for now. Our near-future goal is to release plasticparcels on a cloud-platform, ideally in an environment with a variety of available hydrodynamic and biogeochemical datasets.

Expect to see an idealised example to try out soon!

[philippemiron]

I think that will be very helpful to test out the kernels, and will also aligned a bit more with the rest of the parcel documentation.

[michaeldenes]

Hi @philippemiron, we've now created an idealised example that requires no additional data (see #55), and have updated the documentation for all the other examples. We'll roll these out in a new release soon, but they are all in the main branch now. Let me know if you have any issues, feel free to close this issue if all works! Cheers!

[philippemiron]

I just reviewed the example and it is great, but I had a problem with the conda-forge package on two machines (intel osx and arm64 osx) running the first cell:

---------------------------------------------------------------------------
ModuleNotFoundError                       Traceback (most recent call last)
Cell In[1], line 7
      4 import numpy as np
      6 # parcels and plasticparcels imports
----> 7 import plasticparcels as pp
      8 import parcels
     10 # Plotting imports

File [~/micromamba/envs/test/lib/python3.12/site-packages/plasticparcels/__init__.py:5](http://localhost:8888/lab/tree/Downloads/plasticparcels/docs/examples/~/micromamba/envs/test/lib/python3.12/site-packages/plasticparcels/__init__.py#line=4)
      2 from plasticparcels.particlefile import *  # noqa
      3 from plasticparcels.utils import *  # noqa
----> 5 from plasticparcels._version_setup import __version__, __version_tuple__  # noqa

ModuleNotFoundError: No module named 'plasticparcels._version_setup'

It worked once I installed the current main branch of plasticparcels using pip... I will only recommend we use the next version of plasticparcels alongside the JOSS paper so the example is included.

[philippemiron]

Small comment, in the last cell where you are comparing the 5 experiments, I think ax.set_ylabel('Depth [m]') should be ax.set_ylabel('Meridional distance [m]').

[michaeldenes]

Thanks for raising the versioning issue @philippemiron, this should be fixed by #59, and I've updated the y-axis labelling in #58. Cheers!