Climate Prediction Challenges with Machine Learning

Spring 2025 (Syllabus)

A climate data science course from LEAP STC

A project-based learning course where teams of climate science and data science students collaborate to create machine learning predictive models for challenges inspired by LEAP's research.

Shortcuts: Project 1 | Project 2 | Project 3

---

Project cycle 1: Hurricanes, Climate, Clustering (Exploratory Data Analysis and Visualization)

Following the work of

• Nakamura et al. (2009). Classifying North Atlantic Tropical Cyclone Tracks by Mass Moments. Journal of Climate, 22(20), 5481–5494. doi:10.1175/2009jcli2828.1

(starter codes)

Week 1 (Jan 21)

• [Introduction to LEAP CPC] (McKinley)
• [Introduction to Earth Systems and Climate Change] (McKinley)
• Tutorial on LEAP Pangeo
• Project 1 description Hurricanes, Climate, Clustering starts
• Team activities
  • Introduction and a fun fact
  • Review and discuss project 1 materials as a group

Week 2 (Jan 28)

• [A deep dive into Project 1] (McKinley)
• Project 1 starter code review (TA Xinyi Ke)
• Discussion / Q&A
• Round robin - teams share project plans

Week 3 (Feb 4)

• Presentation and submission instruction
• Discussion and Q&A

Week 4 (Feb 11)

• Project 1 presentations
• Discussion and Q&A

---

Shortcuts: Project 1 | Project 3

Project cycle 2: Parameterizing Earth System Models with Machine Learning

Following the work of

• Sane, A. et al. (2023). Parameterizing Vertical Mixing Coefficients in the Ocean Surface Boundary Layer Using Neural Networks. Journal of Advances in Modeling Earth Systems, 15(10). doi:10.1029/2023ms003890

([starter codes])

Week 5 (Feb 18)

• [Project 2] starts.
• Introduction to Project 2 and the challenge of parameterization (McKinley)
• Tutorial on neural networks (Zheng)
• Project 2 [starter codes]
• Discussion and Q&A

Week 6 (Feb 25)

• [Tutorial] [Ocean mixing] (McKinley)
• Discussion and Q&A
• Round robin - teams share project plans

Week 7 (Mar 4)

• Visit by study lead author Dr. Sane
• Group work
• Discussion and Q&A

Week 8 (Mar 11)

• Group work
• Discussion and Q&A

Week 9 (Mar 25)

• Project 2 presentations
• Discussion and Q&A

---

Shortcuts: Project 1 | Project 2

Project cycle 3: Machine Learning to Extrapolate from Sparse Data

Following the work of

• Gloege, L. et al. (2021) Quantifying errors in observationally-based estimates of ocean carbon sink variability, Global Biogeochem. Cycles doi:10.1029/2020GB006788.
• Heimdal, T.H. and G.A. McKinley (2024) Using observing system simulation experiments to assess impacts of observational uncertainties in surface ocean pCO2 machine learning reconstructions, Scientific Rep.doi:10.1038/s41598-024-70617-x.
• Heimdal, et al. (2024) Assessing improvements in global ocean pCO2 machine learning reconstructions with Southern Ocean autonomous sampling, Biogeosciences doi:10.5194/bg-21-2159-2024.
• and other papers from the McKinley group

([starter codes])

Week 10 (April 1)

• [Project 3] starts.
• [Science Tutorial on "Air-Sea Flux of CO2"] (McKinley)
• Review of starter codes
• Discussion and Q&A

Week 11 (Apr 8)

• [Tutorial on decision tree, random forests and xgboost] (Zheng)
• Discussion and Q&A
• Round robin - teams share project plans

Week 12 (Apr 15)

• Group work
• Discussion and Q&A

Week 13 (Apr 22)

• Group work
• Discussion and Q&A

Week 14 (Apr 29)

• Project 3 presentations
• Discussion and Q&A
• Celebrate a great semester!

Shortcuts: Project 1 | Project 3