Numerical Weather Forecast Models
The simulation model solves the equations of motion for the atmosphere and the ocean numerically. Small-scale processes that cannot be resolved explicitly, for example turbulence and solar radiation transport, are described by sub-grid parameterizations. The initial condition for the simulations is obtained by data assimilation, which combines the numerical model, recent observations and the previous forecast to obtain the best estimate of the present state, the “analysis” state. A similar technique applied to historical observations is used to produce “reanalysis” fields that describe the evolution of daily weather over many decades. Data bases with such reanalysis data are essential for research on climate and weather.
This course describes how forecast models are constructed, and explores all the steps of weather forecasting, including different techniques for data assimilation. The difference between regular weather forecasts and seasonal forecasts, in which the ocean dynamics is crucial, is explored. The students perform computer labs with both the idealized 3-dimensional Lorenz model, and with OpenIFS, the numerical weather forecast model used by the European Centre for Medium Range Weather Forecast (ECMWF).
The course thus enables the student use numerical weather forecast models. It also opens the possibility to do research on numerical weather forecasting and data assimilation in universities or national weather services, or to work with organizations and companies dealing with weather forecast products, such as insurance agencies.
Lectures cover the following topics:
- Predictability and forecast verification
- Forecast model structure
- Sub-grid parameterizations
- Observational data for assimilation
- Techniques for data assimilation: optimal interpolation, 3- and 4-dimensional variational assimilation, ensemble Kalman filter
- Reanalysis
- Ensemble prediction
- Seasonal prediction
There are two computer labs. In the first one, 4-D variational assimilation is applied to the 3-dimensional Lorenz model. In the second lab the OpenIFS model from ECMWF is used to forecast a real storm that hit Western Europe, tweaking a number of parameters of the model.
Teaching Format
Lectures, exercises and computer labs.
Course material
Grading criteria, course literature and other material and correspondence related to the course will be available on the course Athena-site once you have registered for the course.
Assessment
The examination is done through the written reports of both labs, and an oral presentation of the lab with OpenIFS.
