/Producing-realistic-climate-data-with-GANs

Associated repository to the article Producing realistic climate data (10.5281/zenodo.4436274)

Primary LanguageJupyter Notebook

Producing-realistic-climate-data-with-GANs

Associated repository to the article Producing realistic climate data (10.5281/zenodo.4436274)

Installation

  • Clone the repository in your working folder
  • Create a virtual environment and install the requirements.txt file
    • python -m venv env_folder
    • source env_folder/bin/activate
    • pip install -r requirements.txt
  • /!\ The code is in version tensorflow==1.14.0 and keras==2.2.4 as a consequence the virtual environment needs to be created using python 3.7.X .
  • /!\ Cartopy package requires some preinstallation (see documentation)
  • Download the data (a sample is available here) and place it in the ./data/raw/ folder.
  • Finally run python training.py .

Notebooks

In the notebooks the code used to create the figures in the article is available. These notebooks run on the reduced dataset, consequently the figures and the results will be less accurate due to the small sample size.

However, the saved model of the GAN Generator used for the article can be found in the folder model.

Interpolation example

Radial interpolation :

Radial interpolation

Linear interpolation in latent space :

Linear interpolation in latent space.

Linear interpolation in image space :

Linear interpolation in image space.

Spherical interpolation :

Spherical interpolation