Data for "Global Precipitation Correction Across a Range of Climates Using CycleGAN"

# Data for "Global Precipitation Correction Across a Range of Climates Using CycleGAN"

This repository contains the data used in the paper "Global Precipitation Correction Across a Range of Climates Using CycleGAN" by J. McGibbon et al. (2023, *in review*).

`train_val` contains the model outputs for the training and validation sets used in the paper. Model spinup data is not included. The C384 runs are stored as a single series along a concatenated time axes. All C384 data has been coarsened to C384 resolution.

`ramping_data` contains the model outputs from the 4-year, 3-month ramping simulation, including spinup data.

`predicted` is included for convenience, and contains the model outputs from the "best" model, as was used to create figures shown in the paper. This model output was created from the CycleGAN using the validation dataset (stored in `train_val`) as input, and years 2 and 3 of the ramping simulation in `ramping_data` (following the 3-month spinup period). Ramping predictions are stored as separate datasets for the C48 (backwards) prediction and the C384 prediction. Validation data is stored alongside its respective target data in a single netCDF file, labelled "processed-agg". These filenames are intentionally left unchanged so that they correspond with the names of files used in the `process_combined_aggregate.py` script in `projects/cyclegan` of the code DOI for this paper, which was used to create its figures.