General introduction

Data assimilation is the meeting point of physical modelling, uncertainty analysis and mathematical algorithm. Given measurements with error and considering that modelling cannot be exact, the purpose of data assimilation is to mix these different sources of information with the proper weighting. Engineer applications based on modelling are making a grow use of data assimilation techniques with the increase of quantitative informations on the physical system.

Fondamental

The present course provides a basic introduction to data assimilation with the point of view of practical applications. It is organized with the following chapters:

  1. From weather forecast to engineer applications: basic ideas on data assimilation are presented on several example, starting with weather forecast

  2. Three simple examples of data assimilation: estimations of clock time, hydraulic jump velocity or tank filling parameters are used for hands on

  3. Some classical data assimilation methods: opening towards real applications methods are hinted

MéthodeLecture notes and slides

FondamentalJupyter notebooks: Variational methods

What time is it ? (clock)

How will the bore propagate? (bore)

Will the water overflow? (water)

FondamentalJupyter notebooks: Ensemble methods

Ensemble method for the bore propagation (ensbore)

Artificial fonction R30R10 (ensR30R10)

FondamentalJupyter notebooks: Comparaison of methods

Several methods for the Lorenz model