Minor update to depth- and time-dependent unit testing #65
Conversation
…o parcels. The testing data uses only 5 levels, down to 4.7m depth. Particles would cause an out of bounds error in such a case, and be deleted. Updating the density of the particles lets them still sink, but not as quickly as before.
|
Additionally, there was some missing wind data (updated in 2e48261, which includes a few more days of data) causing a time-interpolation error. I'm not sure how these passed in the past, unless time_extrapolation was set to true by default, and no longer? |
…, and our tests only run on two days of data, I have named the second week data 2020-01-05, even though the data is defined for 2020-01-11. Parcels will correctly interpolate the data temporally, but this is just to ensure the data is read by the constructors.
michaeldenes
changed the title
…xing kernel to be free to mix particles beyond 4m depth!
…ure particles aren't kicked beyond the depth field!
The change we made in v3.0.5 that could have impacted this is that the |
|
@erikvansebille, this makes a lot of sense, and why I didn't pickup any of these data related issues earlier! Thanks for linking this! |
…errors when particles are kicked below the test dataset.
for more information, see https://pre-commit.ci
…els/PlasticParcels into fix_unit_test_failures
…ce the VerticalMixing kernel uses a 0.5m distance in the finite-differencing scheme. Also only running the test for 1d rather than 2.
|
@VeckoTheGecko would you mind helping me look at why the readthedocs build fails? |
|
Its to do with Python 3.13. |
|
Thank you! Glad I asked, I wouldn't have picked up on that! |
|
Done. Can you check the site is as you expect once its rendered? (i.e., no visual differences). It looked fine when I did it, so should be fine |
|
@VeckoTheGecko looks good, thanks mate! |
There was an issue with the unit testing, caused by a recent update to parcels. The testing data uses only 5 levels, down to 4.7m depth. Particles would cause an out of bounds error in such a case, and be deleted. Updating the density of the particles lets them still sink, but not as quickly as before.