Data assimilation (Meteo 597, Penn State Dept. Meteorology and Atmospheric Science, Spring 2018)
Data assimilation seeks to find the best estimate of the state of a dynamical system and its uncertainty by combining information from model forecasts and observations. It has been adopted for state and parameter estimation for a wide range of dynamical systems across many disciplines such as weather, ocean, land, water, air quality, climate, ecosystem and astrophysics. This course covers various data assimilation approaches such as variational, ensemble-based, hybrid and nonlinear methods.
Co-instructors: Steve Greybush, Fuqing Zhang
More information can be found in the Course Syllabus.
Lecture 1: Overview of data assimilation.
Lecture 2: Least squares approach.
Notes
Lecture 3: Observation operators.
Notes
Lecture 4: Optimal interpolation.
Notes
Lecture 5: Bayesian approach.
Notes
Lecture 6: 3DVar method.
Notes
Lecture 7: Minimization algorithms.
Notes
Lecture 8: Preconditioning.
Notes
Lecture 9: Nonlinear test models.
Notes
Lecture 10: Tangent linear model.
Notes
Lecture 11: Adjoint model.
Notes
Lecture 12: Characterizing dynamical error growth.
Notes
Lecture 13: Kalman filter.
Notes
Lecture 14: Extended Kalman filter.
Notes
Lecture 15: Ensemble Kalman filter (EnKF).
Notes
Lecture 16: A serial variant of EnKF.
Notes
Lecture 17: Square root filters.
Notes
Lecture 18: Practical Modifications to EnKF: inflation and localization.
Lecture 19: Innovation statistics.
Notes
Lecture 20: Adaptive localization and inflation.
Notes
Lecture 21: Ensemble transform Kalman filter (ETKF).
Notes
Lecture 22: LETKF.
Lecture 23: Parameter estimation.
Notes
Lecture 24: 4DVar method.
Notes
Lectures 25,26: Hybrid methods.
Notes
Lecture 27: Assimilation system design.
Lecture 28: Adjoint and ensemble sensitivity.
Notes
Lecture 29: Observation impact.
Notes
Lecture 30: Operational data assimilation systems.
Lecture 31: Radar data assimilation.
Lecture 32: Regional-scale ensemble data assimilation.
Lecture 33: Representation errors.
Notes
Lecture 34: Particle filters.
Notes
Lectures 35,36: Non-Gaussian methods.
Notes