·
17 commits
to main
since this release
This is the version of the code used to train the emulator used in Doury et al 2022.
This repository contains the main functions used to build the Emul-UNet introduced in Doury et al (2022, in review).
It is organised as follow :
INPUT_MAKER.py creates the emulator inputs
make_unet.py designs the neural network architecture behind the emulator
TRAINER_EXAMPLE.py organises and call the functions below to train the emulator.
Doury, A., Somot, S., Gadat, S. et al. Regional climate model emulator based on deep learning: concept and first evaluation of a novel hybrid downscaling approach. Clim Dyn (2022). https://doi.org/10.1007/s00382-022-06343-9