Lectures

Statistics for UQ and Time Series

Adam Sykulski, Imperial College London | July 16, 2024

Learn how to apply the Bootstrap method and the Central Limit Theorem (CLT) to quantify uncertainty in ocean data. Discover practical approaches to estimating confidence intervals that can help improve ocean modeling and forecasting research. Delve into more advanced techniques for analyzing time series and conducting spectral analysis, and learn best practices in sampling to address uncertainty. Listen to the end for a bonus section demonstrating how to bootstrap a time series.

Click to view the recording and lecture slides.

Deriving Uncertainties for the Global Drifter Program: Case Studies

Shane Elipot, University of Miami | July 16, 2024

Explore three insightful case studies focused on the characterization and derivation of uncertainty estimates within the Global Drifter Program (GDP). Delve into techniques for identifying and modeling errors that affect parameter estimation, and discover how to use the bootstrap method to derive uncertainty estimates. Learn the steps involved in this technique and how to interpret results. Explore uncertainties associated with drifter position errors and sea surface temperature readings from drifting buoys.

Click to view the recording and lecture slides.

Spatio-temporal Statistics for Mapping Oceanographic Observations

Mikael Kuusela, Carnegie Mellon University | July 17, 2024

Discover Gaussian Processes (GP) for spatio-temporal mapping of oceanographic observations. This lecture will explore the theory and practice of spatio-temporal interpolation, with a focus on GP regression. Understand the foundational concepts that make GP regression a standard method widely used in oceanography. Strengthen knowledge of mean functions, covariance functions and parameter estimation, and GP for regression for interpolating oceanographic data.

Click to view the recording and lecture slides.