Learning Repository / Tutorial Fork
This repository is my personal working fork of the Climate Change AI Summer School tutorial on statistical downscaling of climate projections using deep learning. It is preserved as a record of my hands-on work while completing the tutorial and is not an original project authored by me.
Through this tutorial, I worked with a deep-learning-based statistical downscaling workflow for climate projections, including:
- Preparing climate data for machine learning
- Understanding large-scale climate predictors and local climate variables
- Applying deep learning to statistical downscaling
- Evaluating downscaled precipitation predictions
- Comparing performance on mean precipitation and extreme events
- Understanding limitations of statistical downscaling under future climate conditions
- Exploring the role of machine learning in climate adaptation
- Tracking computational energy use and emissions
Some of the concepts explored through the notebook include:
- Statistical climate downscaling
- Perfect-prognosis downscaling
- General Circulation Models (GCMs)
- Deep learning for climate science
- Precipitation prediction
- Extreme-event evaluation
- Climate-model bias
- Model generalization under climate change
- Responsible AI for climate applications
Python · Deep Learning · Jupyter Notebook · Climate Data · Scientific Computing
The primary work is contained in:
Statistical_Downscaling_of_Climate_Projections_with_Deep_Learning.ipynb
The notebook contains the tutorial workflow and my executed experimental work while completing the exercise.
This repository is a fork of the Climate Change AI Tutorials project.
Original tutorial:
Statistical Downscaling of Climate Projections with Deep Learning
Original author:
José González-Abad Instituto de Física de Cantabria (IFCA), CSIC–Universidad de Cantabria
The original tutorial was developed for the Climate Change AI Summer School.
Please refer to the upstream Climate Change AI repository for the original tutorial, documentation, attribution, and citation information.
Archived learning repository
This repository is preserved to document my learning and experimentation with machine-learning-based climate downscaling. It is not being maintained as an independent software project.