NinoLearn - A research framework for statistical ENSO prediction

NinoLearn is a open source research framework for statistical ENSO prediction that is initiated to facilitate collaboration, speed up the start up of research and make realized research more transparent, comparable and reproducible.

The current framework was developed in 2019 during the Master Thesis of Paul Petersik and is available on GitHub (see here). It lays the groundwork for a further development and the integration of other postprocessing routines and statistical models.

Contents

• Introduction
  • The El Niño Southern Oscillation
  • ENSO forecasts
  • Existing statistical models
  • The aim of this package
  • How does NinoLearn work?
  • Cited literature
• Installation
• NinoLearn package
  • Download
  • Input/Output
  • Preprocess
  • Learn
• Tutorials
  • Download and read raw data
  • Data preparation and read postprocessed data
  • Postprocess data
  • Standardized Research
  • Deep ensemble for ENSO-forecasting
  • Encoder-Decoder model for ENSO-forecasting
• Some forecasts