Sane et al 2023 JAMES GOTM for NN training

[galenmckinley]

Dataset Name

Data for NN training from Sane et al. 2023

Dataset URL

https://zenodo.org/records/8293998

Description

For our Spring 2025 course, I need only

Data for training the NN under the Data/
training_data_raw_time_index_49_149.nc
training_data_for_SF_hbl_gaps_filled.nc
data_testing_4_paper.txt

Notebooks for this are in development at https://github.com/leap-stc/LEAPCourse-Climate-Pred-Challenges/tree/main/Project-StarterCodes/Project2-Parameterization

Size

3 files, about 1 GB

License

Creative Commons Attribution 4.0 International

Data Format

Other (describe below)

Data Format (other)

both NC and TXT

Access protocol

HTTP(S)

Source File Organization

data are filtered output from a Generalized Ocean Turbulence Model (GOTM) this is used in Sane et al. 2023 to train a NN parameterization

Example URLs

No response

Authorization

No; data are fully public

Transformation / Processing

The raw data are a greater number of profiles, not filtered and adjusted as discussed in Sane et al. 2023. These are plotted in a first figure.

The SF file is the one used in training.

Target Format

Zarr

Comments

The whole dataset from Sane et al. 2023 includes the training output, as well as the inputs to the GOTM model used to produce this training set. I do not need these, and think it would be confusing to the students to have all this in the dataset they are looking at. I am open to putting this on leap-persistent if that is a better choice. I discussed with Julius briefly on 12/19/24. Thank you! Galen