CNN based all-season ENSO forecast model

• Main training with CMIP5 data (csh/1.main_training.sh)

• Fine Tuning with SODA data (reanalysis) (csh/2.fine_tuning.sh)

• Heatmap analysis (csh/3.heatmap.sh)

• you can download data set here: https://drive.google.com/file/d/1JxSbxu1bNHqw-FFcLnY-jXWeGxo1t45z/view?usp=sharing

• The data set consists of the following:

  (1) Training set for main training (CMIP5): 
  
      Input: [CMIP5_tr.input.1861_2001.nc]
      Label: [CMIP5_tr.label.1863_2003.nc]
  
  (2) validation set for main training (CMIP5):
  
      Input: [CMIP5_val.input.1861_2001.nc]
      Label: [CMIP5_val.label.1861_2001.nc]
  
  (3) Training set for fine tuning (SODA):
  
      Input: [soda.input.1871_1970.nc]
      Label: [soda.label.1871_1970.nc]
  
  (4) Test set (GODAS):
  
      Input: [godas.input.1980_2017.nc]
      Label: [godas.label.1980_2017.nc]
  

• Ham, Y. G., Kim, J. H., Kim, E. S., & On, K. W. Unified deep learning model for El Niño/Southern Oscillation forecasts by incorporating seasonality in climate data. Science Bulletin 66, https://doi.org/10.1016/j.scib.2021.03.009 (2021).
• Ham, Y. G., Kim, J. H. & Luo, J.-J. Deep learning for multi-year ENSO forecasts. Nature 573, https://doi.org/10.1029/2010JC006695 (2019).

• Tensowflow (> version 2.0, https://www.tensorflow.org/install/)
• netCDF4
• numpy