This hands-on tutorial introduces statistical downscaling of climate projections using deep learning. Participants will explore why coarse-resolution climate models may not adequately represent local climate conditions and how machine learning can help produce higher-resolution climate information.
Using climate data from New Zealand and the DeepESD workflow, participants will prepare data, train a deep-learning model, evaluate its performance, and interpret the resulting projections. The tutorial also discusses model limitations, responsible use, energy and emissions tracking, and pathways from research to climate adaptation and decision-making.
Author:
- Jose González-Abad, Instituto de Física de Cantabria (IFCA), CSIC-Universidad de Cantabria, gonzabad@ifca.es
We recommend executing this notebook in a Colab environment to gain access to GPUs and to manage all necessary dependencies.
Please refer to these GitHub instructions to open a pull request via the "fork and pull request" workflow.
Pull requests will be reviewed by members of the Climate Change AI Tutorials team for relevance, accuracy, and conciseness.
Check out the tutorials page on our website for a full list of tutorials demonstrating how AI can be used to tackle problems related to climate change.
Usage of this tutorial is subject to the MIT License.
González-Abad, J. (2026). Statistical Downscaling of Climate Projections with Deep Learning [Tutorial]. In Climate Change AI Summer School. Climate Change AI. https://doi.org/10.5281/zenodo.21446887
@misc{gonzalez2026downscaling,
title={Statistical Downscaling of Climate Projections with Deep Learning},
author={González-Abad, Jose},
year={2026},
organization={Climate Change AI},
type={Tutorial},
doi={https://doi.org/10.5281/zenodo.21446887},
booktitle={Climate Change AI Summer School},
howpublished={\url{https://github.com/climatechange-ai-tutorials/downscaling-climate-projections}}
}