Training course: Predictability and ensemble forecast systems

ECMWF | Reading | 4-8 November 2024

This five-day course will consider the challenge of making predictions from days to seasons ahead. 

The predictability of the atmosphere in the medium and extended range will be considered, including different aspects of the Earth system which give rise to predictability on longer time scales. Elements of a probabilistic forecast system as well as its calibration and verification will be presented.

The course is a combination of lectures, hands-on practicals and group discussion sessions and will include some elements of pre-course study. Please consult last year's timetable on the left-hand side menu to get an idea of possible modules and timings. The final timetable will be shared in due time.

Please note that this course does not cover methodologies for downscaling/post-processing, probabilistic products or discuss decadal prediction.

Main topics:

  • Theoretical ideas - chaos, predictability limits
  • Ensemble methods - taking account of uncertainty in initial conditions and in the model
  • Predictability in the extended range - ocean, sea-ice, stratosphere, land
  • Probabilistic forecast initialisation, modelling, evaluation and verification
  • ECMWF medium-range, monthly and seasonal systems


Participants should have a good meteorological and mathematical background, and are expected to be familiar with the contents of standard meteorological and mathematical textbooks.

Introductory material not covered by the course can be found in our lecture note series.

Some practical experience in numerical weather prediction is an advantage.

All lectures will be given in English.

4 November (09:00) - 8 November (13:00) 2024

Location: Reading (UK)

Course format: In-person only

Course code: NWP-PR

Course fee: £980

A course fee is payable by participants who do not reside in an ECMWF Member or Co-operating State.

More information about our fees

Application for this course closes 31 July 2024.