EASC3420: Machine Learning for Earth and Planetary Sciences

Machine learning is transforming how Earth and planetary scientists analyse complex observations and simulations. This course provides hands-on experience applying modern methods to real datasets from climate science, seismology, remote sensing, and planetary exploration.

Prerequisite

EASC2410 or demonstrated fluency in Python through equivalent coursework or research experience. Students who have not completed EASC2410 should seek the instructor’s approval.

Learning objectives

Students completing the course should be able to:

• explain fundamental machine-learning concepts and their scientific relevance;
• prepare and explore geoscience datasets using Python;
• apply classification, regression, clustering, and introductory deep-learning methods;
• evaluate model performance, uncertainty, limitations, and scientific credibility; and
• design and communicate a reproducible machine-learning investigation.

Course modules

1. Scientific questions, data preparation, and exploratory analysis
2. Supervised learning: classification and regression
3. Model selection, validation, interpretation, and uncertainty
4. Unsupervised learning and clustering
5. Introduction to neural networks and deep learning
6. Research applications in climate, remote sensing, seismology, and planetary science
7. Independent project design, analysis, and presentation

The course uses current Python 3 together with NumPy, pandas, scikit-learn, Matplotlib, and selected deep-learning and geospatial tools. Current schedules and assessment details are provided through HKU’s learning platform.

Responsible use of AI

AI tools may support learning and coding when permitted for a task. Students must disclose the nature and extent of AI assistance, verify generated results, and remain responsible for the scientific reasoning and work they submit.