Lecturer

Bedartha Goswami

Schedule

21 April 2021 – 28 July 2021
Wednesdays, 10:00 – 12:00
14 lectures of 2 hours each
No lecture on 26 May 2021 (Pentecost break).

Location

Online via Zoom (link after registration via email)

Credits and Workload

ECTS: 3
Workload: 90 h
Contact hours: 30 h
Self study: 60 h
Duration: 1 semester
Medium of instruction: English

Evaluation

Paper presentation, project presentation, and written project report

About the course

The course aims to discuss the state-of-the-art in machine learning approaches that are used to address problems in climate research. It will first introduce the fundamentals of climate dynamics and the principles of atmospheric and oceanic circulation. The basics of climate data analytical approaches for both current and paleoclimatic time scales will also be covered. Following this, classical statistical learning approaches such as unsupervised clustering, component analysis, and trend tests will be covered. Next, the course will delve into artificial neural network based approaches that have been used to address predictive problems in climate. Additional topics include the use of complex network approaches in studying climate systems will also be introduced and the use of artificial neural networks in general circulation models.

Teaching methods

The lectures are designed to be interactive and dialogic, and will involve lectures as well as the reading of current journal articles. The final evaluation will be on the basis of of an oral presentation and a written report.

Learning target

At the end of the course, students will posses a basic understanding of how the climate works, and of how we measure our current and past climates. They will be able to review the current literature on relevant problems in climate science and be able to relate those problems to relevant machine learning approaches that can be used to tackle them.

Keywords

machine learning, climate science, statistics, climate models, paleoclimatology, meteorology

Enrollment information

Simply send an email to bedartha.goswami@uni-tuebingen.de expressing your interest to register in the course. You will be contacted later with more details (Zoom link, etc.).

Update 19 April 2021:
Registration for the course is now closed. A maximum of 30 participants is already reached.

Tentative Lecture plan

L1. 21 April 2021: Introduction and Preview
Highlights of course contents; evaluation criteria; general information
slides

L2. 28 April 2021: An introduction to the earth’s climate
What is climate; what are the components of the climate system; a detailed look at the atmosphere; additional topics: climate forcings, climate sensitivity
slides

L3. 5 May 2021: Measurement of weather and climate
how do we measure weather; station based measurements; satellite based measurements; paleoclimate proxy records
slides

L4. 12 May 2021: Models of the weather and climate
energy balance revisited; general circulation models; conceptual models; numerical weather models; statistical weather generators

L5. 19 May 2021: Important climatic systems
El Niño Southern Oscillation; North Atlantic Oscillation; Pacific Decadal Oscillation; Indian Ocean Dipole; monsoon systems

26 May 2021: Pfingstpause (Pentecost break)
No lecture

31 May 2021: Last date for project topic finalisation
Students have to hand in their project topics by this date.

L6. 2 June 2021: Classical learning approaches for weather and climate
empirical orthogonal functions; clustering approaches; Markov models; statistical models for trend, seasonality, and residuals

L7. 9 June 2021: Deep learning approaches for weather and climate
weather forecasts; forecasting climate indices; improving climate models; finding low dimensional representations

L8. 16 June 2021: Climate network approaches for weather and climate
what are climate networks; hidden spatial patterns; teleconnections; early warning systems; feature detection for deep learning

L9. 23 June 2021: Machine learning approaches in paleoclimate
climate field reconstructions; detecting abrupt transitions; paleoclimate proxy parametrizations

L10. 30 June 2021: Tutorial lecture
discuss points that need further clarification; students have to come up with a lost of topics from previous lectures that were difficult to understand; further discussions possible on ongoing project work

L11. 7 July 2021: Introduction to scientific communication
how to write a scientific paper; how to give a scientific presentation; how to search for scientific state-of-the-art on the internet

L12. 14 July 2021: Guest lecture by Peter Dueben, ECMWF, UK
Peter Dueben conducts research on high-resolute weather and climate simulations at the European Center for Medium-range Weather Forecasting (ECMWF) at Reading in the UK. He will talk to us about his views on machine learning for weather and climate predictions.

L13. 21 July 2021: Presentations I
Project presentations by the students.

L14. 28 July 2021: Presentations II
Project presentations by the students.