Deep Learning in Earth and Environmental Science
Machine learning (ML) has become an increasingly important tool for improving predictive and diagnostic models. It is also being adopted across a multitude of physical disciplines. However, in most physical science applications (e.g., seismology and geophysics, climate and atmospheric science, environmental and ecosystem science), researchers are interested in doing more than merely improving their models of real-world phenomena: we are interested in using the broader family of AI tools to deepen our understanding of the dynamic, physical or ecological processes that are fundamental to these fields. This course will explore deep learning, how machine learning relates to first principles physical models in a wide range of geological and geophysical sciences, atmospheric sciences, environmental sciences, and hydrogeology/hydrology.
Topics
- Design and implement novel deep learning approaches for analysis of large datasets in geoscience, climate science, and atmospheric sciences.
- Demonstrate ability to deploy and utilize cloud computing resources in conjunction with statistical and machine learning algorithms.
- Compare and contrast physics-based inversions, statistical learning, and deep-learning approaches, and recognize limitations and benefits of each approach.
- Integrate natural science principles and deep learning models to extract new insights into naturally occurring processes in earth systems.
- Draw statistically significant inferences from data and produce statistically-sound predictive models for future states or events in weather, climate, and hazards.
Official roster learning outcomes
Prerequisites & course information
Linear algebra (e.g., MATH 2210 or similar), Python programming (e.g., CS 1133 or similar) or multi-variate statistics (e.g., BEE 4310 or similar), or permission of instructor. An introductory machine learning course is recommended.
Sources
- Course catalogue ↗2026-2027