Introduction to Data Assimilation (intensive)
This webpage will be updated as needed.
- Time of class: 9/11-13, 2017 10:25am-4:40pm
- Instructor: Takemasa Miyoshi
Catalog description: This course is an introduction to data assimilation for both theory and applications, and aims to provide practical knowledge to apply data assimilation to actual problems. In numerical weather prediction, data assimilation plays an equally essential role as the simulation model. Data assimilation is now being applied to even broader fields.
This course is based on the speed-learning course on introductory practical data assimilation first held in Numerical Prediction Division, Japan Meteorological Agency in Fall 2008. The course contents are all original by the instructor. The course was conveyed repeatedly at different places including Tohoku University Spring School 2010 (February 8-10, 2010), semester-long courses at Kyoto University in Spring 2016 and 2017, RIKEN data assimilation camp (November 14-18, 2016) and University of Tsukuba Atmospheric Science intensive lecture (January 7-9, 2017). Every time the course was very successful with positive feedback from the participants.
We will follow the materials of the most recent semester-long course “Data Assimilation A” in Kyoto University in Spring 2017. However, this is a 3-day intensive lecture course like the one in University of Tsukuba in January 2017. Therefore, we cannot do the project part of this course, although the project is truly essential. For your reference, I will hand out the project problems. I would strongly encourage to work on the project problems by yourself afterwards.
Through the coursework, students will be able to
- understand the general concept of data assimilation,
- implement a dynamical model and understand the deterministic chaos,
- understand and implement the Kalman filter,
- understand the difference between the Kalman filter and 3D-Var,
- understand and implement the ensemble Kalman filter,
- understand and implement the 4D-Var,
- choose and implement appropriate data assimilation methods for a given problem, and
- understand the concept of cutting-edge data assimilation research.
- Kalnay, E.: Atmospheric Modeling, Data Assimilation and Predictability. Cambridge University Press, ISBN: 0521796296
ScheduleThis section will be updated as needed.
|9/11||2. Chaotic dynamical systems|
|9/11||3. Observation model|
|9/11||4. Kalman filter|
|9/12||5. Tangent linear model and covariance inflation|
|9/12||6. Optimal interpolation and 3D-Var|
|9/12||7. Ensemble Kalman filter|
|9/12||8. Perturbed observation method and square root filter|
|9/12||9. Covariance localization|
|9/12||10. Variance inflation and innovation statistics|
|9/13||11. 4D-EnKF and smoother|
|9/13||12. 4D-Var and the variational solver|
|9/13||13. Model parameter estimation|
Grading is based on the examination (100%).
Grading criteria: A+: 95.00% -- 100.00% A : 85.00% -- 94.99% A-: 80:00% -- 84.99% B+: 77.50% -- 79.99% B : 72.50% -- 77.49% B-: 70.00% -- 72.49% C+: 67.50% -- 69.99% C : 62.50% -- 67.49% C-: 60.00% -- 62.49% D+: 57.50% -- 59.99% D : 52.50% -- 57.49% D-: 50.00% -- 52.49% F : 0.00% -- 49.99%