This repository is both an installable package, and a collection of issues/notebooks that document the work done in the ocean-transport group and beyond. The goal of this project is to quantify the rectified effect of small scale heterogeneity in the atmophere and ocean on the air-sea fluxes computed via bulk formulae. See below for relevant publications.
You can install this package by cloning it locally, navigating to the repository and runnning
pip install .
The science published from the repository is using a custom Docker image (installing additional dependencies on top of the pangeo docker image).
You can find the image on quay.io and refer to specific tags used in the notebooks in ./pipeline.
To work with the custom image on the LEAP-Pangeo Jupyterhub, just follow the instructions to use a custom image, and enter quay.io/jbusecke/scale-aware-air-sea:<tag>, where you replace tag with the relevant version tag.
E.g. for tag 68b654d76dce:
Follow the above instructions but install the package via
pip install -e ".[dev]"
The Overlooked Sub-Grid Air-Sea Flux in Climate Models (Preprint)
.
├── paper [Code used to produce the GRL Paper plots]
├── pipeline [Raw processing code (smoothing + offline flux computation)]
├── scale_aware_air_sea [installable software package]
Ommited folders are either part of the installable package, or testing notebooks and other non relevant code.
The raw data is available as a mix of egress-free and requesters pays data (see scale-aware-air-sea/stages.py for detailed paths).
All aggregated data used for plotting is available on the M²LInES funded Open Storage Network Pod in egress free object storage at https://nyu1.osn.mghpcc.org/leap-pubs/busecke_balwada_et_al_GRL/v1.0.1/plotting.
To reproduce the results from the 2025 GRL Paper exactly, check out the code at the v1.1.0 tag (which corresponds to the final code+figures after revision), and run the paper/full_plots-submission.ipynb notebook within the quay.io/jbusecke/scale-aware-air-sea:68b654d76dce docker image (find some guidance how to do this locally here or follow the advice for a 2i2c JupyterHub below).
Warning
Even the picture processing is fairly slow. Be prepared to take a coffee break if you are running this on a smaller laptop.