Yue (Michael) Ying

Data Assimilation Interactive Tutorials

Tutorials outline

A short series of tutorials on the core ideas of data assimilation, taught through small interactive widgets — web ports of sections of the NEDAS tutorial notebooks. One concept per chapter, one widget per concept: drag, click, and watch the analysis respond. The arc follows Evensen, Vossepoel & van Leeuwen, Data Assimilation Fundamentals (Springer, 2022): Bayesian foundations first, then the analysis schemes built on them, then what lies beyond linear-Gaussian.

6 of 7 chapters live — the rest are planned.

• 01
  Bayes' rule for Gaussians

  The analysis as posterior ∝ prior × likelihood, in one variable then two: covariance, correlation, marginals. (book Ch 2–3, 6)
  Live
• 02
  Observation–state correlation

  One observation informs a whole neighbourhood: correlation maps and ensemble scatter from a 2-D field. (book Ch 8)
  Live
• 03
  The ensemble Kalman filter

  The analysis as a precision-weighted average — scalar to vector — then as a linear combination of ensemble perturbations. (book Ch 6, 8)
  Live
• 04
  Localization and inflation

  The two knobs that make ensemble DA work in high dimensions: localizing the sample covariance, and inflating the ensemble spread. (book Ch 10)
  Live
• 05
  Challenges from dynamical systems

  Chaos, predictability and non-Gaussianity from two toy models: Lorenz-96 error growth, and the vort3d hurricane — rapid intensification and the non-Gaussian error distributions that position uncertainty produces. (book Ch 1)
  Live
• 06
  Nonlinear filters

  Particle filters, and when they beat the EnKF. (book Ch 9)
  Live
• 07
  Parameter estimation

  Extending the state vector to uncertain model parameters; applications in multiscale systems. (book Ch 19–23)
  Planned