Page History

2. Course code

ATM316

Aikaisemmat leikkaavat opintojaksot 53654 Numeerinen meteorologia II, 5 op.

3. Course status: compulsory/optional

-Which degree programme is responsible for the course?
Master's Programme in Atmospheric Sciences

-Which module does the course belong to?
ATM300 Advanced Studies in Atmospheric Sciences
compulsory for

• Study Track in Meteorology (if student chooses Dynamic meteorology III -course package)

optional for

• Study Track in Aerosol Physics

TCM300 Advanced Studies in Theoretical and Computational Methods

-Is the course available to students from other degree programmes?
Yes

4. Course level (first-, second-, third-cycle/EQF levels 6, 7 and 8)

Master’s level, degree programmes in medicine, dentistry and veterinary medicine = secondcycle
degree/EQF level 7
Doctoral level = third-cycle (doctoral) degree/EQF level 8

-Does the course belong to basic, intermediate or advanced studies (cf. Government Decree
on University Degrees)?
Advanced studies

5. Recommended time/stage of studies for completion

-The recommended time for completion may be, e.g., after certain relevant courses have
been completed.

6. Term/teaching period when the course will be offered

The course will be lectured every other year (odd years) in the II period.

7. Scope of the course in credits

5 cr

8. Teacher coordinating the course

Prof. Heikki Järvinen

9. Course learning outcomes

At end of the course, students should know the principles of data assimilation applied in numerical weather prediction (or, numerical prediction of the ocean state), and can at practical level write Extended Kalman Filter in the context of simple linear and nonlinear prediction models (Lorenz 3-parameter model in the non-linear case). Students can make numerical experiments with the prediction system, and visualize and interpret the results.

10. Course completion methods

The are three meetings every week with short (15-20 min) lectures at the beginning of the lectures, and hands-on practical supervised sessions to progressively develop the Kalman filter code. All exercises are distributed and returned via the Moodle -page.

11. Prerequisites

Some mathematical and statistical aptitude is important, but only sufficient coding skills are really necessary. The coding language is a student's choice: Fortran, Matlab, Python, or such like will do well. For understanding the connection of the topic with numerical weather (ocean) prediction, basics of either meteorology or oceanography are needed. This course is however possible to complete with any background, as long as coding skills and mathematical background are there.

12. Recommended optional studies

For meteorology, oceanography, and hydrospheric geophysics students it is recommended to take Numerical meteorology I first. Laboratory course in numerical meteorology requires somewhat deeper knowledge of atmospheric thermodynamics, general circulation, synoptic meteorology, and physical processes that this course.

13. Course content

1. Least squares estimation, 2. The maximum likelihood approach and the Bayes theorem, 3. Gaussian probability densities in the Bayes formula, 4. Sequential estimation, 5. Multivariate estimation, 6. Observation operator, 7. Kalman filter in multi-variate case, 8. Kalman filtering exercise with linear/non-linear models.

14. Recommended and required literature

All lecture materials are provided for students.

Interested students can read

• Chapter 5 from "Atmospheric modeling, data assimilation and predictability" by Eugenia Kalnay (2003; Cambridge University Press).
• Chapter 6 from "Numerical weather and climate prediction" by Thomas Tomkins Warner (2011; Cambridge University Press).

15. Activities and teaching methods in support of learning

There are three weekly meetings, and each of them is a practical session where students develop their own codes. Lap-top computers are thus needed in the lectures.

16. Assessment practices and criteria, grading scale

There is a course exam based on the coding exercises (modify an existing code, or make a new type of experiment - mainly to show that the student masters the code he/she has written).

The grading is based on 1. Overall progress made in coding the Kalman filter, and 2. Course exam.

17. Teaching language

English (Finnish if all present are Finnish speaking)